microprobe techniques in the earth sciences || micro-raman spectroscopy in the earth sciences

CHAPTER TEN

Micro-Raman spectroscopy in the Earth Sciences

Stephen Roberts and Ian Beattie

10.1 Introduction

When a monochromatic (i.e. single-frequency) beam of light traverses a medium (gas, liquid or solid) the majority of the scattered light will remain at the incident frequency. However, a small proportion of the scattered light will be at changed frequencies, above and below the incident frequency, and this is referred to as the Raman effect. The Raman effect was first observed by Raman and Krishnan (1928) using focused sunlight and filters and relied on the visual observation of colour changes in the scattered light. However, it was not until the advent of continuous wave visible lasers, during the 1960s. that the importance of Raman spectroscopy as a routine analytical technique was realized. Furthermore, the availability of this highly intense monochromatic light source, which could be focused to a narrow waist, allowed the analysis of small volumes of gas, liquid or solid.

Today several instrument manufacturers produce Raman instruments, with microscope attachments, which enable the routine application of Raman spectroscopy to small samples such as fibres or dust particles. Yet. within the Earth Sciences, micro-Raman spectroscopy represents a relatively new analytical technique, which to date is only established in a few geological laboratories. Nevertheless. the use of light to probe the vibrational behaviour of molecular systems is undoubtedly a 'growth area'. Particularly attractive aspects of the Raman technique include the small spot size, relative lack of sample preparation and the normally non-destructive nature of the analysis. These attributes have resulted in a variety of geological applications of which examples include the characterization and structural study of minerals, and the discrimination of the calcium pol)'morphs involved in the construction of foraminiferal tests.

10.2 Principles

If a gas, liquid or solid is illuminated with a monochromatic light source, most of the light will pass through the sample without undergoing any

Microprobe Techniques in the Earth Sciences. Edited by P.J. Potts, J.F.W. Bowles, S.J.B. Reed and M.R. Cave. Published in 1995 by Chapman & Hall, London. ISBN 0 412 55100 4

MICRO-RAMAN SPECTROSCOPY IN THE EARTH SCIENCES

change (Figure 10.1). However, a small proportion will be scattered by the sample. Measurement of this scattered light by a spectrometer reveals that about 10-3 of the incident intensity has been scattered with the same frequency (va) as that of the incident light source. The process by which light is scattered at the same frequency as that of the incident beam is often called elastic or Rayleigh scattering.

In addition to the Rayleigh scattering, about 10-6 of the incident intensity is scattered at new frquencies above (va + i1v) and below (va - i1v) the incident frequency. This is referred to as Raman scattering or the Raman effect. The shifts in frequency (±i1v) from that of the incident radiation are independent of the exciting radiation Va and are characteristic of the species which gives rise to the scattering.

Another straightforward way to consider the Raman effect is through an energy level diagram (Figure 10.2). Here an incident photon (va), interacts with a vibrating molecule and is annihilated. As a result of this process a new photon at a lower frequency (v r ) than that of the incident photon is created, and the molecule undergoes a Raman active vibrational (rotational or, unusually, electronic) transition VR to a higher energy level (Figure 10.2(a». The difference in frequency (va - Vr = VR) is a vibrational frequency of the molecule under study. Alternatively, if the molecule is in an excited state (Figure 1O.2(b» and undergoes a transition to the ground state during the Raman process then (vr - Va = VR)' This time the scattered photon (v r ) is shifted to high frequency, that is the photon is blue shifted relative to the incident beam.

Raman transitions to lower frequency, i.e. red shifted bands (Figure 1O.2(a», are referred to as Stokes lines, whereas transitions to higher frequency, blue shifted (Figure 1O.2(b», are referred to as anti-Stokes lines. Stokes lines are normally much more intense than anti-Stokes lines, as the population of the ground state is usually very much greater than that of the excited state of the molecule. However, this is not true if experiments are carried out at high temperatures or where Raman frequencies are very close

Source

Monochromatic light

r-":',:;:':":;f·';=,!..-~::~~ Sample r-~~;"""'''''''

Frequency Vo Intensity 10

h f Scattered light

Spectrometer I Frequency Vo. intensity ca. 10-3/0.

Rayleigh scattering plus Frequency Vo + Av. intensity ca. 10-610. Raman scattering

Figure 10.1 Schematic outline of a simple Raman experiment. The monochromatic light source is normally a laser operating in the visible region. typically at 488 or 514 nm.

388

PRINCIPLES

Vo Vo l' r

Stokes lines Anti-Stokes lines

(a) (b)

Figure 10.2 Energy-level diagram of the Raman effect. See text for explanation of the notation used.

to the incident frequency so that VR is very small. The contrasting intensity of Stokes and anti-Stokes lines is clearly illustrated in Figure 10.3. which shows a Raman spectrum of CCl4 .

Not all molecular vibrations are Raman active. since the condition for obtaining a Raman spectrum for a given molecule depends on a change in the polarizability of the molecule during a vibration. When an oscillating electric field is applied to a molecule. the electrons migrate to follow the field. An incident beam of light has an oscillating electric field perpendicular to the direction in which the beam is propagating. Classically. the induced oscillating dipole of the molecule, or induced polarization (P) (i.e. the ease with which the electron cloud is deformed) is proportional to the applied field (E):

P = aE (10.1)

where a is the polarizability. This polarizability must change during a molecular vibration if a band is to be Raman active. The polarizability of a molecule can be represented by a triaxial ellipsoid. Consider the linear molecule CO2 , with its three fundamental vibrations illustrated in Figure lO.4(a). For the symmetrical stretching vibration of CO2 (Figure lO.4(a)(l)) the molecule clearly changes shape and so does its polarizability from (I) to (II) to (III) (Figure lO.4(b)). If we plot a graph of polarizability of the molecule (a) against some vibrational coordinate (Q) (Figure lO.4(c)) we can see that the change in shape results in a change in polarizability where the curve cuts the axis. Thus (8aI8Q) =1= 0, and the vibration is therefore Raman active. If however, we consider the antisymmetric stretch (Figure 1O.4(a)(2)), although the molecule changes shape during the vibration the two 'end-member' vibrational configurations give identical polarizability

389


Absolute wavelength (A) Ao 540 530 520 514.5 510 500 490 nm ______ ~I ________ ~I ______ ~IL_ __ ~I~ __ LI ________ LI ________ -LI __ ___

Absolute ,wavenumber ('IT) I ~o

18500 19000 1 I t I I I

20000 I I I

20500 cm- 1 I !

19436

u,

Stokes Anti-Stokes

Raman wavenumber (~ij) , , 1000 600

, 200

I

o I

200 600 1000 cm- 1 I I I I I I I

Figure 10.3 Raman spectrum of CCl4 showing the more intense Stokes lines. Various units of light measurement are shown for comparison; relative wave numbers (cm-l) are most commonly employed as these are independent of the exciting radiation. The most intense band in the spectrum (ud is where the carbon remains stationary and the four chlorines move in and out in sinusoidal motion simultaneously in phase. This is sometimes called the 'breathing frequency' for obvious reasons.

ellipsoids (Figure lO.4(b) (IV and VI)). Thus, in this instance a plot of pol ariz ability against the vibrational coordinate results in (oalaQ) = 0 and the vibration is Raman inactive (Figure lO.4(c)). A similar argument would apply for the bending vibration (Figure 1O.4(a)(3)). Although the bending vibration is Raman inactive, it is infrared active as the bending involves a change in the dipole moment of the molecule. Similarly, the anti symmetric stretching vibration is also infrared active. For infrared activity it is the change in the dipole that is important as the molecule need not have a permanent dipole. This contrasting infrared and Raman active nature of the vibrations serves to demonstrate the complementary nature of the two techniques. Thus, in the above example, only the symmetrical CO2 vibration causes a change in the pol ariz ability of the molecule and is Raman active (but is infrared inactive). It is generally found that symmetric vibrations tend to give rise to intense Raman lines whereas non-symmetric vibrations are usually weaker.

Nishimura, Hirawaka and Tsuboi (1978) have summarized some common observations about Raman spectral intensities into three main generalizations.

390

(a)

(b)

(c)

PRINCIPLES

- ~ 0--- c--O (1) Symmetric

~ ~ ~

O--c--O (2) Antisymmetric

1 (3) Bending

0-- C --0 <£J e <£J

/---- .... ,O-C-O> -----'" One extreme

(I)

,.,.-------...,. ~O-C-O) -- - - - --

(IV)

ex

o

"..---------\. 0 -C--O' ----------""

Undisplaced (II)

---------- ..... (O-C--O, ----------

(V)

-E O~+Q

Raman active Symmetric stretch

Other extreme of vibration (III)

(-0-=-=;:'=0" " / ------(VI)

ex

o

Raman inactive Antisymmetric stretch

Figure 10.4 (a) Fundamental vibrations of the CO 2 molecule. (1) symmetric stretch. (2) antisymmetric stretch, (3) bending vibration. (b) Schematic outline of polarizability ellipsoids associated with the antisymmetric and symmetric stretch of the CO 2 molecule. (c) Polarizability (a) against a displacement vector (Q), for the Raman active symmetric stretch and the Raman inactive anti symmetric stretching vibration.

1. Stretching vibrations associated with chemical bonds should be more intense than deformational vibrations.

2. Multiple chemical bonds should give rise to more intense stretching modes, e.g. Raman lines due to a C=C vibration should be more intense than a C-C vibration.

391


3. Bonds involving atoms of large atomic mass are expected to give rise to stretching vibrations of high Raman intensity.

A vibrational spectrum is characteristic of a given sample and as individual peaks may be associated with the presence of particular structural groups within the sample, the spectra may be used to infer the presence of a particular phase or certain molecular groups. Further information can sometimes be obtained during Raman experiments if the degree of depolarization of the scattered light can be determined. The ability of minerals, e.g. calcite, to polarize normal light is familiar to geologists. The interaction of highly polarized laser radiation with molecules during Raman spectroscopy gives spectra with lines which are found to be polarized to differing extents: for example, the breathing frequency of CCl4 (VI) gives a line which is polarized exclusively in the direction of polarization of the incident laser beam. This information can be important in assigning bands to particular vibration modes. The degree of depolarization can be estimated by observing intensity variations when a piece of polarizer is rotated in the Raman beam. It also provides a method for 'removing' polarized bands from a spectrum which may contain overlapping bands.

10.3 Instrumentation

To complete a Raman experiment, the essential components required are a light source, a sample point, collection optic or optics, a dispersing optical element and a detection system system (Figure 10.5). The principal design criteria behind this instrumentation are the inherent weakness of the scattering phenomenon and the necessity to maximize the number of detected photon events.

To produce a detectable number of Raman scattered photons an intense light source is required. Continuous wave lasers offer sufficient power levels in the visible region of the spectrum. This property is important since the Raman scattering cross-section varies as the fourth power of the frequency. The laser source (having a typical output power of between 0.01 and 1 W) can be focused on the sample, either within a 'macro-chamber' or by a conventional microscope. With micro-Raman spectoscopy, the laser beam is directed into the microscope and onto a semi-reflecting mirror (beam splitter) (Figure 10.5) and a portion of the incident radiation is reflected down towards the sample. This reflected beam then passes through the microscope objective lens which focuses the beam onto the sample. It should be noted that when a laser has been focused through a microscope objective, 1 W of laser power focused to a 2,um spot size gives an intensity of the order of 108 W cm-2 which in certain instances can result in damage to the sample. The scattered beam containing the Raman signal is then collected back

392

INSTRUMENTATION

Workstation

o

Figure 10.5 Schematic outline and optical diagram of the Jobin Yvon 53000 Raman microscope.

through the objective and is transmitted through the beam splitter and directed to the coupling optics at the entrance of the spectrometer. The coupling optics generally consist of an achromatic lens which refocuses the light onto the entrance slit of the monochromator.

The monochromator is required to disperse collected light across the exit slit for sequential presentation to a detector, or to disperse it across an array detector. In the instrument illustrated in Figure 10.5, a non-dispersive double monochromator rejects stray light prior to dispersion within the spectrometer. Since the scattered light also contains the elastically scattered (Rayleigh) component, the monochromator must be able to discriminate effectively against this large signal. The major problem associated with these systems is transmission loss owing to the many optical elements. In the system mllustrated, the spectrometer disperses the collected light across a multichannel detector (Figure 10.5).

Detector systems can be categorized as single and multichannel systems. Single-channel detection normally involves photon counting using photomultiplier tubes and is frequently applied where high resolution is required. Detector technology has developed rapidly during the past few years with multichannel detectors including diode array systems and, more recently still, multichannel charge coupled device (CCD) detectors. CCDs are integrated circuits within which discrete potential wells are capable of storing electrons created by incident photons. Multichannel detectors allow the simultaneous collection of data over a broad bandpass (typically >400 cm ~ I) and the extremely low noise of a cooled CCD allows for long integration times, which is a particularly attractive feature when measuring weak Raman signals.

393


10.4 Sample handling and routine operation

Sample preparation for micro-Raman spectroscopy can be extremely simple, the most convenient sample support being an ordinary glass microscope slide upon which the sample is placed, so that individual particles can be selected and brought into focus. By appropriate choice of high magnification and/or long working distance objectives, a range of sample sizes and shapes can be accommodated and analysed successfully.

Frequency calibration of the instrument is most readily achieved by checking the frequency of plasma emission lines from the laser source. During analysis, these lines are removed either by a filter or by the use of a laser pre-monochromator. Frequency response calibrations are less straightforward. with each of the various component parts of the instrument. optical elements (notably gratings), detector chip efficiency and linearity all contributing to the final spectrum. Clearly, standards need to be run which permit data from different laboratories to be compared. Typically, a standard sample such as sulphur or indene, both of which are good Raman scatterers. provides information on the relative intensity of the peak positions monitored. Alternatively. light from a quartz-tungsten halogen lamp may be introduced into the system for calibration.

The resolution of the instrument is dependent on the particular arrangement employed and is affected by the choice and combination of slit widths. gratings and use of mono- or multichannel detectors. In a multichannel system with a microscope. the instrument resolution (typically 1.0cm- 1) is largely determined by the groove density of the grating employed and the separation of individual pixels within the detector chip.

In addition to 'routine' operation, Raman spectrometers are well suited to the attachment of variable temperature and/or pressure cells. For example, a heatingifreezing stage can be readily attached which allows for the in situ measurement of substances through a range of temperatures. Similarly. diamond anvil cells may be fitted, allowing the effects of pressure to be monitored.

10.5 Advantages and disadvantages of the Raman technique

The principal advantages of the technique include the following.

1. Each scattering species gives its own characteristic vibrational spectrum, which can be used as a fingerprint for qualitative identification. In ideal circumstances, the intensity of the Raman scattering is proportional to the concentration of the scattering species. Both structural and compositional information can be obtained for the analysed material.

2. Relative lack of interference, with the vibrational bands of many molecules well separated and often narrower than the corresponding infrared absorption bands.

394

SOME APPLICATIONS OF MICRO-RAMAN SPECTROSCOPY IN THE EARTH SCIENCES

3. Spatial resolution. By using a microscope with high-power objectives (routinely xlOO) spot sizes of down to 2,um are attainable.

4. No sophisticated sample preparation techniques are required and all phases - gas, liquid or solid, large or small - can be analysed, providing the host matrix transmits the incident and Raman radiation.

By contrast, the principal disadvantages of the technique include the following.

1. Inherent weakness of the Raman effect. 2. Fluorescence. As a phenomenon, fluorescence is 106-lOR times stronger

than the Raman radiation. Thus, in the presence of fluorescence. the much weaker Raman signal may prove impossible to detect.

3. Coloured samples may absorb the laser beam resulting in the heating of the sample which may consequently decompose.

4. In the region of an absorption band of the sample, the intensity of certain vibrations may be enhanced by orders of magnitude. Further, additional frequencies at 2VR, 3VR and 4VR may be observed. This is known as the resonance Raman effect which, although a disadvantage for quantitative work, has been used to detect Si and S3 in ultramarine (Clark and Franks, 1975). Whenever a sample is coloured, resonance effects may be observed and to obtain quantitative data it is essential to obtain spectra with several excitation lines separated as widely as possible.

10.6 Some applications of micro-Raman spectroscopy in the Earth Sciences

10.6.1 Micro-Raman analysis offluid inclusions

As crystals grow, they invariably contain imperfections in the form of occluded liquids, solids and/or vapours. Within a geological context, analysis of this trapped material can provide vital information to understand the physical and chemical conditions prevalent at the time of crystal growth. Unfortunately, the inclusions are typically rather small, of the order of a few tens of micrometres, and, as a crystal may endure a long residence time within the Earth, more than a single generation of inclusions may be present. Thus. an analytical method capable of analysing individual inclusions would be highly desirable.

As Raman spectroscopy is carried out within the visible region of the spectrum, the incident laser beam can be focused by normal light optics to give spatial resolution in the region of micrometres. It therefore provides a rapid, non-destructive means of analysing the molecular species of fluid inclusions (Delhaye and Dhamelincourt, 1975; Rosasco, Roedder and Simmons, 1975; Dhamelincourt et al., 1979). This technique has proven to be particularly successful for the qualitative analysis of species such as CO2 ,

CO, CH4 , C2H 6 , N2 , H 20, H2S, HS~, 02, SO~~.

395

MKRO-RA.MAN SPECTROSCOPY IN THE EARTH SCIENCES

inclusion substrate does not constitute for measurements. Not

1280 1320 i380 ;440 ticrrf- i

1,000

0.000

information on the of

FiglIl"e 11),6 Raman spectra of carbon dioxide at ~·1280 and 1385 and methane about obtained from a two"phase lndusion within quartz from an auriferous vein, a photomicrograph which is shown in the insert.

396


index of the substrate be dependent on the wavelength of the exciting light but, in the region of an absorption band, attenuation of the incident beam or Raman radiation may also occur, together with significant changes in refractive index. Furthermore, the spectra of the constituents of the inclusions will be dependent on the physical conditions within the sample. Frequency shifts, changes in peak half-widths and a change of scattering crosssection are inevitable consequences of factors such as pressure, change of phase (including solution of gases in liquids), interaction with other species etc. Apart from the nature of the substrate, additional factors that will affect the accurate measurement of intensities include grating transmission and detector response, which are frequency (and hence excitation frequency) dependent. Factors which affect quantification, accuracy and precision of the analysis of gas mixtures of geological interest have been discussed by Wopenka and Pasteris (1986) and Pasteris, Wopenka and Seitz (1988).

Polyatomic ions such as C05-, NO;, HCO;, SO~-, pol- and HSexhibit Raman spectra and can be detected within inclusions at concentrations greater than 1000 ppm (Dubessy et al., 1982). Monatomic ions such as Na+, K+. Ca2+ and Mg2+ cannot be detected in inclusions. However, salt hydrates of various cationic species do show characteristic spectra (Dubessy et at.. 1982) and these can be used to identify the major cation species present, a technique which involves cooling the sample under the Raman microscope to cause crystallization. Another approach is to study the O-H stretching region (2800- 3800 cm -\) of aqueous solutions as this region is sensitive to changes in salinity (Georgiev et al., 1984). Systematic changes in spectral data permit the determination of the salinity of fluid inclusions at room temperature from Raman microprobe spectra (Mernagh and Wilde. 1989).

Another aspect of fluid inclusion research, facilitated by the ability to obtain spectra at differing temperatures and pressures, has involved the analysis of gas hydrates (clathrates) during microthermometric analysis. Raman analysis of clathrates enables estimates of the variable partitioning of gas species during the freezing of aqueo-carbonic inclusions (Seitz, Pasteris and Wopenka, 1987). The results of such studies are crucial to our interpretation of microthermometric data and suggest that if these data are not taken into account, significant errors can be made in the estimate of additional volatile species present.

A further application of micro-Raman in fluid inclusion research involves the identification of 'daughter' and trapped mineral phases within fluid inclusions, which cannot be measured by conventional microbeam techniques. In certain instances, these phases can be characterized by Raman spectroscopy. Carbonate and sulphate daughter minerals and a variety of silicate minerals can be identified by comparison with suitable standard spectra.

397


10.6.2 Microanalysis of carbon

The Raman analysis of carbonaceous material provides a good example of the molecular and structural information which vibrational spectra can offer. As Raman scattering of carbon material is sensitive to the structure within the sample, micro-Raman provides a useful non-destructive technique for characterization of carbon materials.

Tuinstra and Koenig (1970) demonstrated that high-quality single crystals of natural graphite show a single Raman line at 1575 cm- I . However. carbonaceous material shows an additional line at 1355 cm -I, whose intensity increases as a function of the amount of 'unorganized' carbon in the sample and a decrease in the graphite crystal size. Thus, an evaluation of the crystallinity in terms of the (001) in-plane crystallite size of carbonaceous material in natural samples is possible.

As organic matter is metamorphosed, chemical changes occur which involve the driving off of volatiles which include hydrogen, nitrogen and oxygen and cause the resulting material to become enriched in carbon. In conjunction with these changes, the basic structural units, which make up the material and which are randomly ordered at low temperatures, become increasingly ordered (Beny-Bassez and Rouzaud, 1985). Pasteris and Wopenka (1991) developed this work into a potential geothermometer by comparing carbonaceous material from a variety of metamorphic grades. Variations in the intensity ratio of the 136011572 cm -1 bands (Figure 10. 7( a)) could be correlated to a change of grade (Figure 10. 7(b)). Recent analysis of Chitinozoa (organic-walled microfossils) suggest that the poor correlation evident amongst the low rank material may be the result of maturation following a power law curve, an observation which is in agreement with spectra determined for a variety of carbonaceous materials by Beny-Bassez and Rouzaud (1985).

Work on carbonaceous material also reveals two practical aspects which arise during analysis of single crystals and dark materials. First, the spectroscopist should be aware of any orientation effects which may result in dramatic changes in the relative intensities of certain bands (Wang et al., 1989), as observed in the case of graphitic material (Figure 10.8). Second, dark and coloured samples can readily absorb the laser beam, resulting in rapid heating and possible decomposition of the sample. In the studies outlined above, in situ maturation of the sample under the laser beam can be achieved if the laser power is not kept to a minimum. Whenever a sample is coloured (and this includes black), resonance Raman effects may be observed.

10.6.3 Raman spectroscopy of minerals

The potential of Raman spectroscopy for the identification and structural analysis of minerals has long been recognized (Griffith, 1974; White, 1974;

398

(iv)

(iii)

(ii)

(i)

1250 1500 2600 2850 3100 3350

(a) Raman shift (cm- 1)

2

e .Q

e ~ ~ ef-iii c • OJ c

e If-

0 (i) (ii) (iii) (iv)

Metamorphic grade (b)

Figure 10.7 (a) Representative Raman analysis of carbonaceous material from four contrasting metamorphic grades. (i) chlorite zone, (ii) garnet zone, (iii) staurolite zone and (iv) sillimanite zone. (b) Plot of intensity ratio of Raman peaks I(-1360cm- I ):I(1582cm I) of graphite samples grouped according to their spectrally recognized metamorphic categories. Numerals (i)-(iv) refer to metamorphic grade, as for Figure 1O.7(a). (Both diagrams after Pasteris, and Wopenka, 1988).


Z' "iii c OJ C c Cll E Cll

CC

Figure 10.8 Evolution of the first-order Raman spectrum of a graphitic compound as a function of the angle fj between the c-axis of the crystal and the optical axis of the microscope, (From Wang et al.. 1990),

McMillan, 1985; McMillan and Hofmeister, 1988). Raman spectra of minerals can provide important information on the type of chemical bonding between atoms and on the presence of specific compounds within a sample, However, the situation for a crystal is complicated in comparison to a gas and to a lesser extent a liquid, because we are dealing with an extended array in three dimensions. The carbonates are used as the main example in this brief outline as they represent one of the most abundant minerals in the Earth's crust and tend to be studied by the whole gamut of Earth Science specialists, from micropalaeontologists, interested in the construction of calcareous tests, through to metamorphic and experimental petrologists, interested in the phase transitions/high-pressure polymorphs which calcite demonstrates.

Any crystal can be characterized by the primitive cell, this being the smallest unit which, if repeated in three dimensions, would generate the crystal. For infrared and Raman activity to be observed, all atoms related by a primitive translation must move identically. Identical atoms which are related by a translation along or parallel to one of the axes of the primitive cell, and are separated by a distance equal to that cell edge, are 'primitively related', In addition (as with discrete species), if the structure has a centre of symmetry, only vibrations which preserve that centre can be Raman

400


active. Although this may seem complicated, the value of such an approach can be seen in Figure 10.9, where the primitive cell of caesium chloride is shown together with the only possible vibration. This vibration can occur in the x, y or z direction, and obviously all have the same frequency; such a vibration is said to be triply degenerate. If the symmetry is lowered from cubic to orthorhombic then the x, y and z directions are no longer equivalent and three frequencies result instead of the one observed for cubic symmetry. These vibrations are referred to as lattice modes as they result from translational movements of the atoms/ions within the crystal lattice. Note that the vibration shown in Figure 10.9 is infrared active as translational movements generate an oscillating dipole. Thus, caesium chloride does not have a first order Raman spectrum. In general, lattice vibrations occur to low frequency except in the case of very light atoms such as Be, 0, Li and F.

If we now consider a material such as calcite (CaC03). then vibrational modes can be categorized as internal modes, here internal to the CO~- ion. and external modes based on Ca2 + and the CO~- (considered as a unit). As we have seen, the lattice or external modes tend to be at much lower frequencies than the vibrational frequencies of the carbonate ion, especially those based on stretching vibrations. Carbonates of divalent cations of intermediate size usually crystallize in the calcite structure, with two CO~ions in the primitive cell. Placing carbonate ions in the calcite crystal will clearly perturb the vibrations to some extent. The first feature that may occur in the crystal is that the crystal symmetry of the carbonate ion sites may be lower than that of the free ion. This effect can cause resolution of the degeneracy of a vibration so that V3 and V4 (the anti-symmetric stretch and in-plane bend) could appear as doublets, not single lines. In the case of CO~-, the free ion has a threefold axis of symmetry which is retained in calcite. In this case, the degeneracy is not removed and from this point of view the CO~- should display three Raman active fundamentals. of which Vj

Figure 10.9 Primitive cell and vibrations of caesium chloride.

401


(symmetric stretch) will be by far the most intense. A complication is that because there are two cOj- in the primitive cell, each vibration can couple in-phase and out of phase. This is illustrated for the VI vibration in Figure 10.10. Note that this gives one Raman active band.

Taking into account these vibrational modes, the Raman spectrum of calcite (Figure 10.11) has bands at 1088cm- 1 (the very intense breathing frequency in which the carbon atom does not move); 1443cm- 1 (clearly the antisymmetric stretching mode, the high frequency being due to the movement of the light carbon atom); and 714cm- 1 (a low-frequency mode based on angle deformation). These vibrations are thus assigned to VI, V3 and V-l

respectively, V2 being Raman inactive. Note that cOj- unlike CO2 does not have a centre of symmetry. The low-frequency vibrations at 159 and 286cm- 1 are lattice modes.

Turning now to carbonates of divalent elements of large ionic radius which crystallize in the aragonite structure (having orthorhombic symmetry). There are now four molecules in the unit cell and the site symmetry is very low - one mirror plane only passes through the carbon atoms. The loss of the threefold axis means that any degeneracy will be resolved, and it can be predicted that doublets for the degenerate vibrations (V3 and V4) will be present in the Raman spectra. Depending on the magnitude of this splitting, the doubling mayor may not be observed. Further, because the crystal structure is so different, the lattice vibrations will be radically changed. The distinction between aragonite and calcite using micro-Raman techniques will be explained in the following section.

(a) Phase transformations Although aragonite should be the stable high-pressure form of calcite (Jamieson, 1953), calcite invariably transforms to calcite-II and calcite-III at

(a) (b)

Figure 10.10 (a) In-phase and (b) out-of-phase coupling of carbonate modes within the calcite unit cell. (After White, 1974).

402


Counts S-1 x 100

1.400

1.200

1.000

0.800

0.600

0.400

0.200

400.0 800.0 R cm- 1 1200.0

Figure 10.11 Raman spectrum of a calcite crystal. See text for explanation of the bands observed.

high pressures and at temperatures below 600°C. Calcite inverts to calcite-I at around 1.5 Gpa and from calcite-II to III at around 2.0 Gpa (Fong and Nicol, 1971). Good Raman spectra of calcite-II and III have been reported by Fong and Nicol (1971) and Liu and Mernagh (1990); the results of Fong and Nicol (1971) are illustrated in Table 10.1. It can be seen that all frequencies increase at high pressures, as expected from compression of the crystal. Superimposed on the blue-shifted vibrations are other notable spectral effects, including the splitting of the V3 and V4 bands in calcite-II and VI band in calcite-III. These changes are reversible and disappear when the sample is returned to atmospheric pressure. The presence of a V2

vibration within the Raman spectra is probably due to the pressure-induced distortion of the parent calcite structure (White, 1974).

These experimental data confirm the possibility of stable carbonate minerals within the upper mantle and therefore provide a means by which carbon may be retained within this region of the Earth. These data have implications for the genesis of carbon-containing magmas and thus an understanding of the high pressure stability and compressional behaviour of these phases is of considerable geological importance.

(b) Recognition of carbonates with aragonite structure As noted earlier, carbonates of divalent elements of large ionic radius crystallize in the aragonite structure. Micro-Raman has been used to distinguish calcite from aragonite unambiguously and rapidly in samples prepared as standard petrographic thin sections. The two polymorphs show

403


Table 10.1 Raman spectra of calcite II and calcite III (cm -I)

Modes Calcite-II Calcite-III 14 kbar

18 kbar 38kbar

VI 1096 1104 1108 1087 1093

V~ 866 870 870 1471 1540 1540

U3 1445 1515 1520 721 741 745

U~ 715 733 739

Lattice modes 695

333 351 319 (p) 314 (p) 318 (p) 292 299 312 240 269 270

234 (p) 221 224 (p) 208 (p) 224 (p) 204 (n) 206 (n)

204 202 (p) 205 (p) 155 161 173 133 137 139

131 105 109

99 99 99

All data from Fong and Nicol, 1971. The notation: (p) refers to crystals oriented with the optical axis parallel to the pressure axis. (n) refers to natural orientation in which the thrust of the pistion was against a cleavage face.

contrasting low-frequency «300 cm -1) lattice vibrations: aragonite (-150, 180, 190, 206cm- 1); calcite (-153, 283cm-1). In metamorphic rocks aragonite is representative of blueschist metamorphic conditions, usually occurring with jadeite, quartz and lawsonite, which confirms its stable, highpressure, low-temperature origins. Raman analysis has been employed to confirm the presence of micrometre-sized crystals of aragonite and therefore contribute to an evaluation of the P- T evolution of a metamorphic terrane (Gillet and Goffe, 1988).

At the other end of the metamorphic spectrum, the use of micro-Raman to distinguish aragonite-calcite on a very small spot size has resulted in a detailed description and classification of foraminiferal tests. For example, Raman spectra have been used to confirm the calcite nature of Ammonia beccarii compared to the aragonite nature of Hoeglundina elegans (VenecPeyre and Jaeschke-Boyer, 1979).

404

OTHER APPLICATIONS OF MICRO-RAMAN SPECTROSCOPY IN THE EARTH SCIENCES

10.7 Other applications of micro-Raman spectroscopy in the Earth Sciences

As remarked in the introduction, Raman and micro-Raman spectroscopy is a growth area in the Earth Sciences with a burgeoning literature and a wide variety of applications. This section now very briefly describes other applications which have produced, or are in the process of producing, significant contributions to our understanding of key processes in the Earth Sciences.

Raman spectra have been used extensively in the structural studies of silicate glasses and melts. Spectra from glasses of variable composition have been assigned to various structural groups based on observed band position and symmetry character, examination of systematic changes in spectra with composition and corresponding glass and crystal spectra; studies which have been excellently reviewed by McMillan (1984(a)).

One goal of the research into silicate glasses has been to attempt to rationalize properties such as viscosity, density and heat content of magmas as a function of temperature, pressure and composition. For example. it has been noted that the isothermal viscosity of silicate liquids and glass decreases as alkali and alkali-earth oxides are added to pure silica, an observation interpreted in terms of depolymerization of the silicate melt or glass network (Matson, Sharma and Philpotts, 1983; McMillan, 1984(b)). Other workers (Mysen, Virgo and Seifert, 1982; Seifert, Mysen and Virgo, 1982) have shown that Raman spectra can be related to changes in the bulk physical properties of silicate glasses and melts, which may not be quenchable in certain instances (Hemley et al., 1986). For example, Kubicki, Hemley and Hofmeister (1992) showed that the dominant change in silicate glasses with increased pressure is a major shift in the mid-frequencies (625-650cm- 1)

which is reversible. They concluded from the Raman data that the main compression mechanism is a decrease in the average inter-tetrahedral SiO-Si angles.

Raman spectra of H20, CO2 and H2 have all been reported for silicate glasses and have enabled insights into the volatile dissolution mechnisms in melts (Mysen and Virgo, 1980, 1986; McMillan, 1984(a), 1989, for review).

Owing to the weak Raman spectrum of H20, Raman offers an ideal analytical technique with which to study solutes in hydrothermal solutions, especially given the increasing availability of P- T cells suitable for the collection of Raman spectra at pressures and temperatures well above ambient. For example, the behaviour of gold in chloride-bearing hydrothermal systems is of considerable interest to economic geologists and hydrothermal geochemists. Recently, Pan and Wood (1991) were able to provide direct spectroscopic evidence for the existence of AuCl- in hydrothermal solutions. Moreover, Peck et ai. (1991), using Raman and resonance Raman techniques, showed that with increasing pH, chloride ligands are successively replaced by hydroxide ligands, and thus reported for the first time the potential importance of mixed chloro-hydroxo species in a P- T range over which most epithermal Au deposition is believed to take place.

405


10.8 Other Raman techniques and closing remarks

In addition to Raman spectroscopy as described above, an array of Ramanbased technologies now exists which have yet to find their way into the geological laboratory to a significant extent. Amongst these are included Fourier transform (FT)-Raman, hyper-Raman and coherent anti-Stokes Raman (CARS) (see McMillan and Hofmeister, 1988, for more details). Of these, FT-Raman appears the most attractive to the geologist as it offers a potential major advantage over conventional dispersive Raman spectroscopy in its ability to give spectra which are effectively free of fluorescence interference. Commercial micro-FT-Raman systems are now available.

In closing, although Raman spectroscopy remains a relatively underemployed technique in the geological laboratory , this situation will undoubtedly change in the near future as the potential of the technique to contribute to a wide variety of problems within the Earth Sciences becomes increasingly recognized.

Acknowledgements

We extend our thanks for thoughtful and thorough reviews of the manuscript to Phil Potts, Stephen Reed, Dave Alderton, Ian Croudace, Trevor Gilson and John Murray. Barry Marsh and Anthea Dunkley are thanked for plates and cartographic work respectively. One of us (IRB) thanks the Leverhulme Trust for a fellowship.

References

Beny-Bassez. C. and Rouzaud. J.N. (l9R5) Characterisation of carbonaceous materials by correlated electron and optical microscopy and Raman microspectroscopy. Scann. Electron. :'vficrosc .. 1985. 119-32.

Clark. R.J.H. and Franks, M.L. (1975) The resonance Raman spectrum of ultramarine blue. Chell]. Ph},s. Lett., 34. 69-72.

Delhaye, M. and Dhamelincourt. P. (1975) Raman microprobe and microscope with laser excitation. 1. Raman Spectrosc .. 3. 33-43.

Dhc.melincourt. P .. Beny, J.M .. Dubessy, J. and Poty, B. (1979) Analyse d'inclusions fluides a la microsonde MOLE a effet Raman. Bull. Mineral .. 102. 600-10.

Dubcssv, 1.. Audeoud. D., Wilkins, R. and Kosztolanyi, C. (l9R2) The use of the Raman microprobe in the determination of electrolytes dissolved in the aqueous phase of fluid inclusions. Chon. CeDI., 37. 137-50.

Fong. M.Y. and Nicol, M. (1971) Raman spectrum of calcium carbonate at high pressures. 1. Chern. Phys .. 54, 579-85.

Georgiev. G.M .. Kalkanjiev. T.K .. Petrov. V.P. and Nickolov. Z. (19R4) Determination of salts in water solutions by a skewing parameter of the water Raman band. Appl. Spectrosc .. 38,593-5.

406

REFERENCES

Gillet. P. and Goffe, B. (1988) On the significance of aragonite occurrences in the Western Alps. Contrib. Mineral. Petrol., 99. 70-81.

Griffith, W.P. (1974) Raman spectroscopy of minerals, in The infrared spectra of minerals (ed. V.c. Farmer), Min. Soc. Monograph, 4, 119-35.

Hemley, R.J., Mao, H.K., Bell, P.M., Mysen, B.O. (1986) Raman spectroscopy of Si02 glass at high pressure. Phys. Rev. Lett., 57, 747-50.

Jamieson, J.J. (1953) Phase equilibria in the system calcite-aragonite. 1. Chem. Phys .. 21, 1385-90.

Kubicki, J.D., Hemley, R.J. and Hofmeister. A.M. (1992) Raman and infrared study of pressure induced structural changes in MgSi03 , CaMgSi06 , and CaSi02 glasses. Am. Mineral .. 258-69.

Liu, L.G. and Mernagh, T.P. (1990) Phase transitions and Raman spectra of calcite at high pressures and room temperature. Am. Mineral., 75, 801-6.

Matson, D.W., Sharma, S.K. and Philpotts, J.A. (1983) The structure of high sillica alkalisilicate glasses: a Raman spectroscopic investigation. J. Non-Cryst. Solids. 58. 323-52.

Matson, D.W., Sharma, S.K. and Philpotts, J.A. (1986) The structure of high-silica alkalisilicate glasses along the orthoclase-anorthite and nepheline-anorthite joins. Am. Mineral .. 71, 694-704.

McMillan, P.F. (1984a) Structural studies of silicate glasses and melts: applications and limitations of Raman spectroscopy. Am. Mineral., 69. 622-44.

McMillan, P.F. (1984b) A Raman spectroscopic study of glasses in the system CaO-MgO-Si02' Am. Mineral., 69. 645-59.

McMillan, P.F. (1985) Vibrational spectroscopy in the mineral sciences. in Microscopic to Macroscopic (eds S.W. Kieffer and A. Navrotsky, Rev. Mineral), 14. Mineralogical Society of America. 9-63.

McMillan, P.F. (1989) Raman spectroscopy in mineralogy and geochemistry. Ann. Rev. Earth Planet. Sci., 255-83.

McMillan, P.F. and Hofmeister, A.M. (1988) Infrared and Raman spectroscopy, in Spectroscopic methods in mineralogy and geology (ed. F.C Hawthorne). Rev. Mineral .. 18. Mineralogical Society of America, 99-159.

Mernagh, T.P. and Wilde, A.R. (1989) The use of the laser Raman microprobe for the determination of salinity in fluid inclusions. Ceochim. Cosmochim. Acta. 53, 765-71.

Mysen. B.O. and Virgo, D. (1980) Solubility mechanisms of carbon dioxide in silicate melts: a Raman spectroscopic study. Am. Mineral., 65, 885-99.

Mysen. B.O. and Virgo, D. (1986) Volatiles in silicate melts at high pressure and temperature 2. Water in melts along the join NaA120 2-Si02 and a comparison of solubility mechanisms of water and fluorine. Chem. Ceol., 57, 333-58.

Mysen. B.O .. Virgo, D. and Seifert. F.A. (1982) The structure of silicate melts: implications for chemical and physical properties of natural magma. Rev. Ceophys. Space Phys .. 20. 353-83.

Nishimura. Y., Hirawaka, A. and Tsuboi, M. (1978) Resonance Raman spectroscopy of nucleic acids. Adv. Infrared Raman Spectrosc., 5, 217-75.

Pan, P. and Wood, S.A. (1991) Gold-chloride complexes in very acidic aqueous solutions at temperatures 25-300°C: a laser Raman spectroscopic study. Ceochim. Cosmochim. Acta, 55, 2365-71.

407


Pasteris, J.D. and Wopenka, B. (1991) Raman spectra of graphite as indicators of degree of metamorphism. Can. Mineral., 29,1-9.

Pasteris. J .D .. Wopenka, B. and Seitz, J.C (19RS) Practical aspects of quantitative laser Raman microprobe spectroscopy for the study of fluid inclusions. Geochim. Cnsmochim. Acta, 52, 979-88.

Peck, J .A .. Tait, CD .. Swanson, B.!. and Brown, G.E. (1991) Speciation of aqueous gold (III) chlorides from ultraviolet/visible absorption and Ramaniresonance spectroscopies. Geocililll. Cosmochim. Acta. 55, 671-76.

Raman. e.V. and Krishnan. K.S. (1928) A new type of secondary radiation. Nature. 121. SOl. Rosasco. G.J .. Roedder. E. and Simmons, J .H. (1975) Laser-excited Raman spectroscopy for

non-destructive partial analysis of individual phases in fluid inclusions in minerals. Science. 190.557-60.

Seifert. F .. Mysen. B.O. and Virgo, D. (1982) Three-dimensional network structure of quenched melts (glass) in the systems SiOc - NaAl0c- SiOc-CaAlc0 4 and SiOc-MgAl,04' Am . . \1ineral. 67. 697-717.

Seitz. J.e.. Pasteris. J.D. and Wopenka. B. (1987) Characterization of COC-CH4-HcO fluid inclusions by microthermometry and laser Raman microprobe spectroscopy: inferences for clathrate and fluid equilibria. Geochim. Cosmochim. Acta. 51. 1651-64.

Tuinstra. F. and Koenig, J.L. (1970) Raman spectrum of graphite. 1. Chem. Phrs .. 53. 1126-30.

V cnec-Pevre. M. -T. (1980) Microanalyseur ionique et microsonde moleculairc a laser mole: application a retude chimique et mineralogique du test d'ammonia becarii (Linne) Foraminifere. Bull. Centre Rech. Explor.-Prod. Elf-Aquitaine, 4, 55-79.

Vencc-Peyre, M.-T. and Jaeschke-Boyer, H. (1979) Interet de la microsonde moleculaire a laser \role en systematique: Etude du foraminifere. C. R. Acad. Sci. Paris. 288, 819-21.

Wang. A .. Dhamelincourt, P .. Dubessy, J. et al. (1989) Characterization of graphite alteration in an uranium deposit by micro-Raman spectroscopy, X-ray diffraction. transmission electron microscopy and scanning electron microscopy. Carbon. 27, 209-1S.

White. W.B. (1974) The carbonate minerals. in The infrared spectra of minerals, Mineral. Soc . . \1ol1ograph, 4. 227-79.

Wopenka. B. and Pasteris, J.D. (1986) Limitations to quantitative analysis of fluid inclusions in geological samples by laser Raman microprobe spectroscopy. Appl. Spectrosc., 40, 144-51.

408

microprobe techniques in the earth sciences || micro-raman spectroscopy in the earth sciences

Documents