4.6 Raman Spectroscopy

R.H. ATALLA, U.P. AGARWAL, and J.S. BOND

4.6.1 Introduction

Raman scattering was first observed in 1928 and was used to investigate the vibrational states of many molecules in the 1930s. Initially, spectroscopic methods based on the phenomenon were used in research on the structure of relatively simple molecules. Over the past 20 years, however, the development of laser sources and new generations of monochromators and detectors has made possible the application of Raman spectroscopy to the solution of many problems of technological interest. In many industrial laboratories, Raman spectroscopy is routinely used, together with infrared spectroscopy, for acquisition of vibrational spectra. Raman spectrometer systems for routine analytical applications are com- mercially available. An important expansion of the potential of the technique has arisen from the use of the microprobe, which permits acquisition of spectra from domains as small as one micron. Application of Raman spectroscopy to lignin analytical chemistry is rela- tively new, and only limited information has been obtained. However, the technique offers several potential advantages and, though it is complementary to infrared spectroscopy, it gives information that is not accessible with the latter alone.

4.6.2 Principle

The phenomena underlying Raman spectroscopy can be described by com- parison with infrared spectroscopy as shown schematically in Fig. 4.6.1. The primary event in infrared absorption is the transition of a molecule from a ground state (M) to a vibrationally excited state (M*) by absorption of an infrared photon with energy equal to the difference between the energies of the ground and the excited states. The reverse process, infrared emission, occurs when a molecule in the excited state (M*) emits a photon during the transition to a ground state (M). In infrared spectroscopy, one derives information by measuring the frequencies of infrared photons that a molecule absorbs and interpreting these frequencies in terms of the characteristic vibrational motions of the molecule. In complex molecules, some of the frequencies are associated with functional groups that have characteristic localized modes of vibration.

Springer Series in Wood Science Methods in Lignin Chemistry (Edited by S.Y. Lin and C.W. Dence) Springer-Verlag Berlin Heidelberg 1992 Raman Spectroscopy 163

As is also shown in Fig.4.6.1, the same transitions between molecular vibrational states (M) and (M*) can result in Raman scattering. A key difference between the Raman and infrared processes is that, in the former process, the photons involved are not absorbed or emitted but rather shifted in frequency by an amount corresponding to the energy of the particular vibrational transition. In the Stokes process, which is the parallel of absorption, the scattered photons are shifted to lower frequencies as the molecules abstract energy from the exciting photons; in the anti-Stokes process,which is parallel to emission, tte scattered photons are shifted to higher frequencies as they pick up the energy released by the molecules in the course of transitions to the ground state. In addition, a substantial number of the scattered photons are not shifted in frequency. The process which gives rise to these photons is known as Rayleigh scattering. This scattering arises from density variations and optical hetero- genieties and is many orders of magnitude more intense than Raman scattering. Figure 4.6.1 depicts another major difference between Raman scattering and infrared processes. To be active in the infrared spectra, transitions must have a change in the molecular dipole associated with them. For Raman activity, in contrast, the change has to be in the polarizability of the molecule. These two molecular characteristics are qualitatively inversely related. A Raman spectrum is obtained by exposure of a sample to a monochromatic source of exciting photons and measurement of the frequencies of the scattered light. Because the intensity of the Raman scattered component is much lower than the Rayleigh scattered component, a highly selective monochromator and a very sensitive detector are required. 164 R.H. ATALLA et al.

The exciting photons are typically of much higher energies than those of the fundamental vibrations of most chemical bonds or systems of bonds, usually by a factor ranging from about 6 for O—H and C—H bonds to about 200 for bonds

between very heavy atoms, as for example in 12. The 514.5 and 488 nm lines from an argon ion laser are often used as exciting frequencies.

4.6.3 Method

4.6.3.1 Raman System

A typical Raman system consists of the following basic components: (1) an excitation source, usually a laser; (2) optics for sample illumination; (3) a double or triple monochromator; and (4) a signal processing system consisting of a detector, an amplifier, and an output device. A diagram showing various com- ponents of the Raman spectrometer is shown in Fig. 4.6.2. A number of stages are involved in the acquisition of Raman spectrum. A sample is mounted in the sample chamber and laser light is focused on it with the help of a lens. Generally, liquids and solids are sampled in a Pyrex capillary tube. The scattered light is collected using another lens and is focused at the entrance slit of the monochromator. Monochromator slit widths are set for desired spectral resolution.The monochromator effectively rejects stray light and serves as a dispersing element for incoming radiation. The light leaving the exit slit of the monochromator is collected and focused on the surface of a detector. This optical signal is converted to an electrical signal within the detector and further manipulated using detector electronics. Such a signal is stored in computer memory for each predetermined frequency interval. in a Raman Spectroscopy 165 conventional Raman system using a photomultiplier tube (PMT) detector, light intensity at various frequencies is meusured by scanning the monochromator. A plot of signal intensity against wavenumber constitutes it Raman spectrum. There are various ways in which Raman systems can be designed depending upon the nature of the information desired. They can be classified as macro- and micro-systems according to whether the information is to represent the sample at the macro level or at a particular micro-domain of the sample. For homogeneous samples, information from the two systems is identical. In con- trast, this information is different for heterogeneous samples. For exumple, a spectrum acquired over a large number of woody tritcheids indicates an aver- age over-all morphological feature and is typical of spectra acquired in the macromode. On the other hand, spectra characteristic of the middle lamella, the secondary wall, or other morphologically distinct features of woody tissue require the microsampling mode. Raman spectra obtained by using both types of systems will be considered. It is also possible to carry out time-resolved measurements of Raman spectra. This has been greatly facilitated by the avail- ability of pulsed lasers and gatable multichannel detectors. The macrumode spectra described here are acquired with an Instruments SA Jobin Yvon Ramanor HG.2S system. Sample excitation is done with either argon or krypton ion lasers. This scanning spectrometer has a thermoelectrically cooled PMT detector and is fitted with a modified Nachet 400 microscope accessory for Raman microprobe work. The microprobe is capable of providing information from domains as small as 1 µ in diameter. Recent advances in solid state detectors have led to more efficient laser Raman spectrometers. These spectrometers are based on multichannel detectors (MCD) and are at least an order of magnitude faster than the systems based on PMT detector. However, at present these systems do not match the resolution capabilities of the scanning systems. The Spex Triplemate 1877 A is a new generation instrument which can be fitted with an intensified diode array consist- ing of 1024 elements. Figure 4.6.3 shows spectra obtained with these two types of detectors.

4.6.3.2 Scattering Geometries

In a Raman experiment. light scattered from a laser excited sample is collected and analyzed. Focusing lenses are used both for sample excitation and light collection purposes. Generally, there are two geometries in which a sample can be studied (Fig. 4.6.4). In the 90° geometry. the laser beam direction and the axis of the collection lens are at 90° to each other, but in the backscattering or the 180° scattering geometry they are coincident. The 90-degree scattering geometry is frequently used in the macromode work, whereas back scattering geometry is usually the choice for microprobe experiments. Specimen sampling needs are somewhat different for these two types of geometries. 166 R.H. ATALLA et al.

4.6.3.3 Specimens and Sampling

Laser-induced fluorescence from lignin-containing samples is a major hindrance to obtaining reasonably good spectra.Certain lignin structural units and/or lignin radicals are evidently the major contributors to the problem. Two sampling procedures are effective in significantly reducing the fluorescence of lignocellulosics.

Water lmmersion Technique

The technique of water immersion involves sampling lignocellulosics in an aqueous environment. Although originally developed for Raman microprobe investigations of woody tissue,this technique has given much better quality spectra in the conventional macromode studies of lignin-containing samples. The sampling procedure for microprobe studies of woody tissue is briefly described below. Raman Spectroscopy 167

Sections of woody tissue, 30 µ in thickness, are extracted with a toluene/ ethanol mixture (2:1, v/v) and repeatedly washed with distilled water. A flat- base Pyrex glass beaker is attached to a slide in a manner that allows transmitted light to pass through. On the inner side of the bottom, another portion of a glass slide is attached. A wet cell wall section is sandwiched between the slide portion and a cover glass through the center of which a small hole has been drilled; the

cover glass holds the section immobile while D2O is slowly added to the beaker until the sample is completely immersed. The beaker is then mounted on the stage of the microprobe and the water immersion objective is lowered into the

D2O. To avoid evaporation or exchange with atmospheric moisture, a thin layer

of mineral oil is placed on the surface of the D2O after the microprobe objective is submerged in the medium. This sampling device is shown in Fig. 4.6.5 (Atalla and Agarwal 1986). Figure 4.6.6 shows improvement in the Raman signal-to-background (RS/B) ratio and elimination of Raman features of the cover glass (cf. Figs. 6.6.6b and 6.6.6c). One shortcoming associated with the immersion technique is the intensity reduction of certain lignin features with time. 168 R.H. ATALLA et al.

Oxygen Flushing Technique

Molecular oxygen is an effective quencher for lignocellulose fluorescence (Agarwal and Atalla 1986a). This effect is now routinely used to study lignin samples by conventional Raman spectroscopy, i.e., in the macromode with 90- degree scattering geometry. Lignin-containing solid samples are sandwiched between two cover slips and inserted in a specially constructed cell designed to allow gaseous flushing of the sample. Using a lens, the laser beam is focused on the sample and oxygen is flushed through the cell. The fluorescence part of the total signal declines rapidly, thereby resulting in improved RS/B ratios. A variation of the method is to modify the sample cell so that it can withstand a higher oxygen pressure. A Raman spectrum of woody tissue obtained by using the technique is shown in Fig. 4.6.7a. In this particular instance, 50 psi oxygen pressure has been used. (Care must be exercised when exposing organic compounds to focused laser radiation under high pressure oxygen. In one instance, an explosion occurred during the study of a photoyellowed handsheet specimen. ) The residual background in Fig. 4.6.7a, underlying the Raman features in the spectrum, can be removed by subtracting a mathematically constructed curve that approximates the background shape (Fig. 4.6.7b). The result of a computer subtraction is shown in Fig. 4.6.7c and d, which differ only with respect to full scale intensity values. Raman Spectroscopy 169

4.6.4 Spectral Information

4.6.4.1 Identification of Native Lignin Features

Wood and pulp samples contain, in addition to lignin, cellulose, and hemi- celluloses. As a result, lignin spectral features need to be distinguished from those of the other main components. There are a number of complementary ways to derive this information. First, there are regions of the spectrum of 170 R.H. ATALLA et al.

wood where cellulose and hemicelluloses do not contribute and only features attributable to lignin are seen. This is the case, for example, in the 1600cm-1 region where only aromatic ring stretching vibrations are observed. However, there are other regions of the Raman spectrum where all components have bands, and the interpretation is therefore somewhat complex. To identify lignin bands in these regions, the Raman spectrum of a completely delignified wood sample is subtracted from the native wood spectrum. The difference spectrum is considered to represent the native lignin.This information, along with that obtained directly from native wood spectra, is then interpreted using Raman studies of lignin models (Ehrhardt 1984). In addition, infrared assignments for various lignins and their models reported in literature (Hergert 1971) have been used in arriving at the m’ost reasonable interpretation of the lignin spectra. Raman band positions and assignments are shown in Table 4.6.1 (Ehrhardt 1984, Atalla, unpubl. 1989). Raman Spectroscopy 171

4.6.4.2 Quantitative Analysis of Lignin in Wood

In a Raman spectrum, band heights are proportional to sample concentration (Schmid and Brosa 1971). In the Raman spectroscopic analysis of lignin (Fig. 4.6.8), band height at 1595 cm-1 is used as the indicator of lignin content. The cellulose peak at 1098 cm-1 is the reference and serves as an internal standard. In Fig. 4.6.9, relative lignin peak intensity of acid chlorite-delignified southern pine wood meal is plotted against its Klason lignin content. The good correlation indicates that Raman spectroscopy is suitable for quantifying lignin in wood pulps after a correlation factor has been predetermined for such materials. In Fig. 4.6.9, the enhanced peak height of the lignin band for the untreated wood (data points in top right-hand corner) arises in part from aromatic ring- conjugated structures in lignin. Chemical compounds with such structures are known to have enhanced Raman scattering (Schmid and Brosa 1971). Experi- ments carried out with certain lignin model compounds confirm the presence of such effects (Atalla, unpubl. 1989). The intense lignin peak in Fig. 4.6.8 is also due to a pre-resonance Raman effect. The pre-resonance Raman effect observed with some chemical com- pounds (Long 1977) has also been seen when Raman studies of wood samples are carried out using a number of different excitation frequencies (Atalla. unpubl. 1989). The 1595 cm-1 band is reduced to less than half its intensity when excitation is changed from 514.4 to 647.1 nm. The pre-resonance Raman effect in wood, when it is excited at 514.5 nm, may be attributed to phenoxy radicals and other lignin structures capable of light absorption at that wavelength (Tripathi and Schuler 1984, Schmid and Brosa 1971).

4.6.4.3 Orientation and Composition Studies of Lignin in Woody Tissue

Developments in Raman spectroscopy and wood sampling techniques have made it possible to carry out in situ studies on the organization of lignin and 172 Raman Spectroscopy

other constituents in wood. The Raman microprobe can examine regions of woody tissue as small as l µ in diameter. Analyses of Raman spectra, obtained in various scattering geometries, suggest that lignin is oriented at the molecular level (Atalla and Agarwal 1985, Agarwal and Atalla 1986b). One set of such spectra. obtained on a longitudinal section. is shown in Fig. 4.6.10. The 1595 cm-1 lignin band intensity changes when the electric vector (ev) of the exciting laser radiation is rotated with respect to the cell wall orientation. This change indicates lignin organization within the cell wall. Raman Spectroscopy 173

4.6.5 Discussion

At this time, Raman spectroscopic studies of lignin have just begun and the capability of the technique has not been fully realized. Work done so far indicates a great potential for the technique. Moreover, Raman spectroscopy is by no means a fully matured branch of spectroscopy and further development in this field is expected.

4.6.5.1 Comparison With Other Techniques

The nature of the information that can be obtained from Raman experiments, namely, vibrational frequencies and band intensities, is similar but not identical to that of infrared spectroscopy. Therefore, it is appropriate to compare these two techniques. As mentioned previously, for a molecular vibration to be IR active, a change in dipole moment accompanying the vibrational transition is 174 R.H. ATALLA et al.

needed. On the other hand, a vibration is Raman active whenever a change in molecular polarizability occurs during the transition. In light of this, polar bond systems with a high dipole show up strongly in IR, whereas bond systems with highly covalent character are quite easily seen in Raman. Water, as a con- sequence, is a weak scatterer in Raman spectroscopy but strongly absorbs in IR. Indeed, water is used as a part of the sampling procedure for some of the experiments. Because of the difference in selection rules between these two types of spectroscopic transitions, the information obtained from the two types of spectra is complementary. Raman Spectroscopy 175

Ordinary Raman scattering is an inefficient process, and in general it is less sensitive than IR absorption. However, in certain lignin samples, conjugation, resonance, or pre-resonance Raman effects can arise, and for particular vibra- tional modes a higher level of sensitivity can be achieved (Long 1977, Schmid and Brosa 1971). In lignins, the latter effect can be induced by proper laser frequency selection. Another advantage of Raman spectroscopy is ease of working with heterogeneous samples. In IR spectroscopy this is difficult because of Rayleigh scatter of infrared photons. The degree of this scatter depends upon differences in the refractive indices at optical heterogeneities. Because the refractive index varies with wavelength in regions of strong IR absorption, it is difficult to separate the extinction due to molecular absorption from the extinction caused by Rayleigh scattering. For Raman spectral measurements, in contrast, dis- crimination against Rayleigh scattering is relatively simple. There are additional advantages in using Raman spectroscopy to study lignin. Since it is possible to use a number of excitation wavelengths, choice of a proper frequency allows selective lignin excitation. Time-resolved Raman studies of excited lignin and its photomodifications can be carried out using pulsed laser based Raman spectrometers. This capability opens up a new field for lignin research. IR spectroscopy, in contrast, does not have this possibility, as the frequencies involved are far removed from those which cause electronic excitation.

4.6.5.2 Future Development

Since Raman studies of lignin in heterogeneous lignocellulosic samples can now be carried out, various topics in lignin research will greatly benefit from Raman spectroscopic measurements. Using conventional instrumentation, samples under the category “difficult” are those that are highly colored. With these, the problem is of high levels of fluorescence and/or thermal degradation caused by excessive absorption of laser radiation. In the future, however, this may be remedied as 1064 nm excitation-based laser FT-Raman instruments become increasingly available. 176 R.H. ATALLA et al.: Raman Spectroscopy

In: Lin, Stephen Y.; Dence, Carlton W., eds. Methods in lignin chemistry. New York: Springer-Verlag: 162-176; 1992.

Printed on recycled paper