Raman Spectroscopy and Confocal Raman Imaging in Mineralogy and Petrography

Abstract

Raman spectroscopy has long been used in geosciences and a wealth of data and publications are available. The majority of
this information originates from point measurements using micro-Raman setups. With the application of confocal Raman imaging,
additional analytical possibilities arise with respect to analyzing the three-dimensional spatial distribution of inorganic
as well as organic phases on the centimeter to sub-micrometer scale. This chapter will highlight some of the key aspects experimenters
should take into consideration when performing confocal Raman measurements as well as experimental results showing the insight
gained into geological samples by the use of confocal Raman imaging.

Full text

1
Raman Spectroscopy and Confocal Raman
Imaging in Mineralogy and Petrography
Marc Fries
1
and Andrew Steele
2
1
NASA Jet Propulsion Laboratory
4800 Oak Grove Drive
Pasadena, CA 91001
USA
marc.d.fries@jpl.nasa.gov
2
Geophysical Laboratory of the Carnegie Institution of Washington
5251 Broad Branch Rd. NW
Washington, DC 20015
asteele@ciw.edu
1.1 Introduction
Raman sp e ctroscopy is especially well suited for use in mineralogy and pet-
rographic studies, as it provides non-destructive mineral identificatio n very
quickly and with excellent specificity. Additionally, it is sensitive to the pres-
ence and structure of carbonaceous phases that are difficult to characterize
by optical microscopy or electron beam methods. Raman spec troscopy has
the advantage over electron beam methods that Raman spectra can reveal
the presence o f a wide variety of mineral phases in a single spectrum o f sub-
micrometer resolution, as opposed to an electr on beam that performs an el-
emental analysis that can include any or a ll phases within the excitation
volume. The practical result of this is that a Raman spectrum can identify in-
dividual mineral phases in fine-grained clusters, while electron probe analysis
only produce the bulk composition of the intermixed phases. With the rela-
tively recent developement of confoca l Raman imaging, definitive petrography
(the study of rocks in thin section) is now possible in fields of view ranging
from tens of micrometers to ce ntimeters across. Data gleaned in this way are
used in turn for petrolo gy s tudies (the study of the formation and alteration
of rocks primarily through p e trographic analysis). An additional benefit of the
use of a visible-light excitation laser is the interrogation of features that are
buried in the interior o f mineral grains, such as opaque and fluid inclusions.
All of this can be do ne on either prepared thin sections or unprepared mineral
surfaces, which places Raman spectroscopic techniques very prominently in a
mineralogist’s “tool kit” of a nalytical techniques.

2 Marc Fries and Andrew Steele
1.2 Raman Spectroscopy and Imaging as a
Mineralogy/Petrography Tool
Mineralogy and petrography techniques have traditionally relied heavily on
optical micr oscopy, utilizing reflected, transmitted, and cross-polarized mi-
croscopy to identify minerals in thin section a nd des c ribe their layout. The
identities and textures of mineral phases are used to interpret the formation
and subsequent alteration history of those minerals and their parent rock for-
mation (Fig. 1.1). Fro m there the investigator can draw conclusions about
the history of the entire rock formation and its r e lationship to larger geologic
units, thereby working with observations made under a microscope to draw
insights on the creation, destruction, and movement of entire continents. In
the case of fossilized materials, petrography is used to make fundamental de-
scriptions of the evolution of life over geolo gic time as well as to discuss the
very origin of life itself.
Fig. 1.1. Example of mineral phase mapping using Raman imaging in a thin section
from the Strelley Pool formation. A) Transmitted light image showing dark, filamentous
features. A Ra ma n image was collected in the red box and the blue scale bar is 80
microns in length. B) Image of quartz within the red box in A) showing the mineral
matrix. Scale bar is 10 microns long. C) Image of carbon, which somewhat surprisingly
appears as a diffuse blebby phase and is not t he primary constituent of the filaments.
D) Small pyrite grains (yellow spots to right and left of the image center). E) Sulfate
minerals that make up the filaments in this image, presumably formed by oxidation of
pyrite grains in D. Sample provided by M. Brasier and D. Wacey, of the D ept. of Earth
Sciences, Oxford U., UK.
Raman spectroscopy is a very useful tool in petrographic analy sis like
these, as it can be used not only for mineral identification but also for investi-
gation of carbonaceous matter in rocks. The study of carbonac e ous material as
a mineral phase in general is somewhat problematic. Under a microscop e , con-
densed carbonaceous phases simply appear opaque. E lec tron beam techniques

1 Raman Imaging in Mineralogy and Petrography 3
such as SEM or EDX, which are commonly used in mineralogical studies, are
less sensitive to their presence and structure than phases compo sed of more
massive elements. Some other techniques such as bulk chemical measurements
require crushing and extraction, which produces useful information but de-
stroys the fine mineralogical context needed to completely describe the orig in
and alteration history of the materia l. Raman spectra are sensitive to the
identity and structure of all known condensed carbon phases and are very
sp e cific in the identification of other mineral pha ses (Fig. 1.2). In addition,
Raman imaging provides mapping of carbo naceous phases within the miner-
alogical context of their parent rock. For example, it allows an investigator
to discern carbonac e ous matter within primar y fluid inclusions that formed
as a rock first cooled from its parent melt from other carbonaceous matter
found in late-phase alteration veins within the s ame rock. This distinction is
especially important in the study of meteorites, where many features of the
rock could be either an original feature of the meteorite, or could possibly be
the result of a lteration after the meteorite fell to Earth.
Fig. 1.2. Simplified diagram showing the general positions of prominent Raman peaks
for a variety of common rock-forming minerals and compounds. Note t hat different
families of minerals tend to have separate “fingerprint regions”, and more importantly
that many of th ese“fingerprint regions” can exist in a single spectrum without substantial
overlap. This allows definitive mineral phase identification even in complex mineralogical
mixtures.
The implementation of Raman spectroscopy is very straightforward, as
very little sample preparation is necessary for a Raman measurement. Mea-
surements ca n be performed on whole rocks placed under a microscope in
the lab, or on rock formations in situ by portable Raman instruments in
field studies. Rocks are usually either ground to produce a flat surface for
simple petrographic examination and Raman point measurement (such as in

4 Marc Fries and Andrew Steele
field studies), prepared as thin section for Raman imaging of fine details, or
ground to a powder to perfo rm point Raman analysis of a sta tistically rel-
evant number of grains as a represe ntative analysis of the host rock (but
without the contextual information). Practically speaking, however, the most
useful mineralogical/petrographic Raman measurements are done using imag-
ing techniques on ground, otherwise smoothed, o r polished surfaces both to
maximize the signal quality and to remove surface weathering products from
the rock. Unless, of course , the goa l of the measurement is to examine surface
weathering products!
1.2.1 Working with Thin Sections
A typical petrographic thin s e ction consists of a thin wafer of a rock sample
that has been polished on both sides until transparent and is thin enough to
reveal mineral birefringence in polar iz e d light. The rock slice is affixed to a
glass slide on one side, and either sealed with a glass cover slip on the opposing
face, or left ex posed. Confocal Raman spectroscopy is well suited for analysis
of thin layers of material (i.e. Macdonald e t al. 2005), as the excitation volume
can be constrained in the z-axis to control signal input from polymer and glass
layers, as well as the x- and y-axes to control and optimize image resolution.
This allows collection of Rama n spectra from selec ted mineral pha ses while
rejecting spectral contributions from the glass slide, cover slip (if pr e sent), or
bonding polymer. However, a number of cons iderations must be accounted for
when working with thin sections. Namely:
1. Thin section surface irregularities and contamination:
Depending on the minerals present in the thin section and the quality of
preparation, some surface pits and irregularities may occur in the thin
section sur fa ce. These typically occur in anisotropic phases such as py-
roxenes and softer minerals such as clays. Surface roughness within these
flaws can scatter the excitation beam somewhat, but the more important
consideration is that polishing materials such as diamond paste and abra-
sive grains can become lodged in them. When collecting Raman spectra
or imag e s from near the surface of the thin section, these contaminants
can be misinterpreted as materials native to the thin sectio n rock. To
avoid this error , examine the thin section in reflected light befor e analysis
to familiarize yourself with the lo c ation and size of flaws. It can also be
useful to create an image of the Rayleigh-scattered las er line itself, which
serves as a monochromatic reflectivity image at the same resolution as
calculated Raman images.
2. Mounting polymer permeation:
Since mounting polymer started out as a liquid, expect to find it in every
crack in the thin section mineral slab. This is especially true of thin sec-
tions made of friable materials, as these thin sections ar e often prepared

1 Raman Imaging in Mineralogy and Petrography 5
using a vacuum impregnation method. Acquiring a reference spec trum of
the polymer used to pre pare the thin section under examination is a good
idea. This spectrum can be used to differentiate mounting po lymer from
any organic compounds prese nt in the thin section, usually by identifying
sp e ctral features in the C-H stretch mode region which is well removed
from Raman modes of rock-forming minerals.
3. Minor polymer contribution through confocal “leaking”:
Confocal spectroscopy constrains light collection to within a narrow ba nd
very close to the instr ument’s fo c al plane, because light arising from ei-
ther a bove or below the focal plane is rejected at the confoc al aperture
or pinhole. A small amount of light arising from directly along the axis
of the ex citation beam, however, may pass through the confocal aperture .
Three conditions must be met for this to occur. O ne, the Raman signal of
the interfering medium must be relatively strong, such as the C-H band
stretch region of a thin section mounting polymer. Secondly, the confo-
cal aperture is relatively large (depending on the instrument and sample,
roughly ≥ 50µm). Finally, the excitation b e am power is ty pically quite
high, as the excited volume of material responsible for this light leak is
smaller than the diameter of the excitation beam, and so the intensity of
the interfering Raman signal is intrinsically small. This effect is usually a
problem only for spectra collected fairly deep into a thin section such that
the excitation volume is close to the lower mounting polymer layer. The
result is that a minor C-H stretch mode Raman spectrum can appear in
some spectra collected from a thin section and us ually see n through large
clear grains, which may lead to the mistaken conclusion that an aliphatic
compound is pr e sent within the mineral phase.
4. Focus on the distal mineral surface:
When viewing a thin section in transmitted light, it ca n be possible to
fo c us on the mineral slab surface adjacent to the glass slide instead of
the thin section surface, especia lly if the illumination s ource is intense.
This is a minor problem, but can be rather embarrassing if yo u happen
to be showing off your shiny new Raman spectrometer to a visitor at the
time. The easy way around this is to focus on the thin section surface in
reflected lig ht w hen you s tart your work.
5. Previous carbon coating:
Some thin sections, and espec ially those pr epared out of rare materia ls
such as meteorites, are examined with multiple methods in multiple lab-
orator ie s. If the thin section has been examined using a vacuum electron
probe technique such as scanning electron microscopy (SEM) or elec-
tron probe analysis, then it has probably been coated with an electri-
cally conductive coating such as Au-Pd o r carbon. These coatings can
be “removed”, but Raman sp e ctroscopy is extraordinarily sensitive to the

6 Marc Fries and Andrew Steele
presence of car bo n in particular. Previous Au-Pd coatings will appear as
laser-reflective materials in cracks, low spots, and surface imperfections
in the thin section. A carbon coating that has been “ removed” will show
up as an opaque or mildly transparent coating in the same s urface imper-
fections but will yield a Ra man spectrum consistent with that of glassy
carbon. That is, the carbo n G and D bands around 1590 and 1350cm
-1
,
respectively (Tuinstra and Koenig 1970), will be broadened to the point
that they overlap considerably. The optimal situation is to utilize Raman
microscopy before applying techniques that may require elaborate sample
preparation.
6. Surface Alteration of Soft Carbonaceous Phases During Thin
Section Preparation:
Raman spectrosco py is extraordinarily sensitive to the structure of c arbon.
The actual interaction volume of the Raman excitation laser is very shal-
low in opaque phases to the point that it is comparable to the thickness
of the surface layer disrupted by mechanical polis hing. In other words, a
Raman spectrum of a soft, opaque carbonaceous phase in a thin section is
very likely only measuring the surface that has bee n mechanically abraded
during polishing. Several researchers have noted this effect [i.e. [1]] and
pointed out the imp ortance of taking it into acco unt w hen describing the
structure of carbonaceous phases in thin section. To put it simply, do
not accept measurements of carbon phase structure in thin section if the
Raman measurement is made at the thin section surface, as that measure-
ment is made only of the thin surface layer that has been disrupted by
polishing. Some researchers [1,2] state that Raman measurements are best
made on the contact betwe e n carbonaceous materials and any transpar-
ent mineral grain, on a surface that is embedded within the thin sectio n
and removed from the ac tua l thin section surface. There is some room for
argument as to whether or not this technique measures car bo n structure
that is truly representative of the carbonaceous material bulk str uctur e ,
however. Whenever possible, it is advisable to work with both a thin sec-
tion and fresh fracture surface of the same material. Raman images of the
thin section can produce in situ maps of carbonace ous material to pro-
vide mineralog ic al context, and multiple spot Raman spectra of the same
carbonaceous material on the fractured surface provide a statistically re l-
evant set of spectra for interpretation of bulk microstructure of that s ame
carbonaceous material in order to ascertain the extent of alteration of the
carbon measured in thin se c tion.
As long as these considerations are acknowledged and accounted for, none of
the ab ove should be a real obstacle to producing high-quality Raman images
of samples in thin section.

1 Raman Imaging in Mineralogy and Petrography 7
1.2.2 Control of Laser Power
The signal intensity of Raman scattered photons is commonly quoted to be
around one one-millionth o f the excitation laser strength, a lthough in real-
ity this value varies from miner al to mineral and from one Raman mode to
another, as well as whether or not a particular mode undergoes resonant en-
hancement, etc. It is genera lly true, however, that the Raman effect is weak
relative to the laser power used to observe it, and this leads to the very real
possibility that a sample can be damaged by the excitation laser (Fig. 1.3 and
Fig. 1.4).
Fig. 1.3. Laser beam damage in a fullerene soot. The square artifact in the center
of the image is a hole that perfectly preserves the shape of a Raman image collected
with an accidental“overdose” of laser power. A misread dial allowed ap proximately fifty
times as much power as was intended, evaporating the carbonaceous material to leave a
square hole and small, surviving silicate contaminants. The scale bar is 80 micrometers
in length.
While many early pa pers described the laser power used in terms o f output
power at the las e r itself, recent literature generally acknowledges the many
factors involved a nd describes power in terms of power density (laser power
per unit area) as measured at the laser spot focal plane. This is espec ially
impo rtant considering that all Raman instruments feature some variability in

8 Marc Fries and Andrew Steele
total laser throughput due to sc attering losses in the focusing optics as well
as in the spot size at the focal plane. It is sufficient to des c rib e the power
density for a measurement simply by measuring the total laser power at the
measurement focus a nd dividing by the calculated spot area, but there is room
for improving upon this method. Laser power distribution at the measurement
sp ot is not uniformly distributed, and there is a need at the time of this
writing for establishment o f a standard method of r e po rting laser power values
that takes into considera tion this and other parameters to include pe ak laser
power fo r a given measurement s po t. Beyond these considerations, for the
application of “conventional” Raman spectroscopy (a s opposed to resonantly
enhanced methods described elsewhere), there are two primary considerations
in managing laser powe r - the nature of the target material and the capabilities
of the Raman device used.
Nature of the Sample:
Understanding the nature of the sample is an impo rtant part of managing
laser power. For example, a materia l that contains fine sulfides that are
susceptible to oxidation on heating must be handled much more “lightly”
than a sample composed entirely of transparent silicates. A mineral such
as quartz or olivine, which exhibits very low absorption in visible lig ht
can tolerate very hig h excitation beam flux without damage. Excitation in
such a case occurs over a sample volume with a size and shape dependent
on beam geometry, and thus on the optics of the instrument used. For
opaque materials, such as chromite and reduced carbon, however, the
excitation “volume” is practically a two-dimensional surface due to strong
absorption of the beam at the mineral surface. Power density for such a
case can become extraordinar ily high since the power density formula’s
denominator (excitation area) becomes very small, thereby increasing the
likelihood of damaging an opaque phase with what is a benign amount of
laser power for transparent gains. Fig. 1.4 illustrates this effect, s howing
small, opaque chromite grains that have been damaged by an excitation
beam while surrounding, transparent olivine grains are unaffected.
The message here is that applied power density vs. acquisition time per
pixel is one of the trade-offs applied to a measurement and that this factor
should be applied with an understanding o f the nature of the sample. This
also serves to reinforce the notion that Raman imaging is best utilized in
conjunction with optical p etrography techniques, that can provide the
general information on sample character needed to select an appropriate
amount of e xcitation la ser power .
Other examples of minerals which are especially prone to laser da mage
include fine-grained minerals, sulfides, or samples embedded in an insu-
lating medium such as aerogel. Fine-grained minera ls will include some
proportion of por osity and/or poorly-ordered materials in its intergranular
volume. Both of these impede thermal conduction, resulting in a material
prone to laser heating damage. Many sulfides oxidize readily upon even

1 Raman Imaging in Mineralogy and Petrography 9
Fig. 1.4. Another one from the “What Not to Do” files, an SEM Image showing laser
damage in a chromite grain. A Raman image was collected using excessive laser power
density with the raster direction proceeding along the axis indicated by the white arrow.
The striation in the chromite grain is the result of damage incurred during rep eated,
parallel passes by the Raman excitation laser. Note that the opaque chromite grain
absorbed sufficient laser power to damage the grain’s surface, while the surrounding,
transparent olivine was undamaged. This illustrates the dependence of laser power used
to collect an image or spectrum on the nature of the material un der investigation. Image
collected by Dr. Edward Vicenzi, S mithsonian Institution, Washington DC, USA.
mild heating, necessitating a Raman micros cope with very high photo n
throughput combined with a very sensitive detector and low laser power
density to produce a usable Raman spectrum without inflicting sample
damage. A by-product of this fact is that Raman spectral standards of
sulfide minerals currently are not as prevalent as spectra of more robust
phases.

10 Marc Fries and Andrew Steele
It is worth noting that cometary particles returned to Earth by the Star-
dust spacecr aft e xhibit all three of the example features described in the
previous paragraph. Most of them ar e fine-grained, they contain fine sul-
fide grains [3] and in their unprocessed for m they are contained in an aero-
gel insulating medium. Raman measurements of Stardust particles focus
on analysis of carbonaceous phases as a result, as these phases typically
show strong Raman s pectral features and are reasonably resistant to heat-
ing da mage [4–6], although a limited number of measurements have been
performed on other phases [7]. This author determined that fine sulfides
fired into aerogel using a light gas gun converted readily to oxides under
laser power density of only 33µW/cm
2
[Fries, unpublished data], which
made this phase extraordinar ily difficult to analyze safely. Stardust sam-
ples are generally examined following removal from the aero gel capture
medium [8] followed by curatorial processing such as gold foil mounting
or ultramicrotoming. T he authors also utilized an empirical method for
determining the upper threshold of allowable power density for Stardust-
like samples using fine, meteoritic sulfide pa rticles. Troilite (FeS) particles
were scraped off a slab of the Toluca iron meteorite and pressed onto gold
foil just as the Stardust particles were. Sub-micrometer particles were ex-
amined over long integration times under incre asing laser power to identify
the power level where ma gnetite first started to appear due to heating in
air, as seen in the appearance of the strong 670cm
-1
Raman peak for mag-
netite.
Raman Instrument Parameters:
Due to variation in instrument design and laser spot geometry, it is ad-
visable not to rely on power settings reported in literature but rather to
perform this type of preparation using the actua l instrument to be used in
the analy sis. As far as Raman instruments are concerned, there are a few
impo rtant avenues available for control of sample damage. These are opti-
mizing instrument efficiency/detector sensitivity and the effective control
of applied p ower. For laboratory instruments, modern commercial devices
include high-throughput instruments and options for high-sensitivity de-
tectors. Beyond the choice of instrument and detector, the primary fa ctor
in instrument design is arguably the choice between fiber-based and fiber-
less Raman s ystems. A system that uses fiber optic cables as the confocal
pinhole and for light transmission to the detector is capable of very high
resolution imaging with good fluorescence suppression capability, but at
the expense of Raman signal throughput due to the small core diameter
of the fiber (typically 25-100µm). The absorption w ithin the fiber itself
is measured in dB/km and thus negligible. Convers e ly, instruments with
large pinho les generally allow more light to reach the detector, but at the
exp ense o f depth resolution and fluorescence rejection. In any event, laser
power applied to a sample must be carefully applied to prevent damage,
and the best means to address it is by a careful se t of heating trials us -

1 Raman Imaging in Mineralogy and Petrography 11
ing minerals that each investigator intends to analyze, using the same
instrument to be used in the analysis. Differences in signal throughput,
sensitivity, excitation beam shape and other factors between instruments
demand that a set of tr ials is done on the actual instrument in use, and
that any such measurements done on similar instruments should be used
only as br oad guidelines a nd only if necessa ry.
1.3 “Raman Mineralogy” Using Imaging Raman
Techniques
Petrography, at its core, is the study of minerals in thin sections. This means
that a petrographer interprets the identity of each mineral in a rock and
draws conclusions on the formation and alteration history of the rock and its
parent formation from the setting and identity of the suite of minerals present.
Imaging Raman spectroscopy is especially well suited to this task, as it not
only c an be used to identify the minerals but also to map their appearance in
a thin sectio n (as in Fig. 1.2). Fine details such as crystallographic orientation,
overall water content, presence of aqueous alteration products, compositional
variation and others can also be identified and separately mapped (i.e. Fig. 1.5)
to highlight and interpret them.
1.3.1 Mineral Phase Imaging
Raman spectroscopy of minerals produces information on the identification of
condensed phases with the exceptions of metals and of phases that fluoresce
strongly under the excitation beam. A very wide variety of phases produce rea-
sonable Raman spectra , to include transparent, opaque, and metallic phase s
(to include sulfides). To illustrate the broad range of mineral species that ca n
be identified using Raman spectroscopy, note that the freely-available RRUFF
online Raman database contains over 2000 distinct mineral spectra [9]. Raman
sp e ctra c an also contain information on a mineral phase’s chemical compo-
sition [10, 11], latent strain (i.e. [12, 13], crystalline orientation [11, 14], grain
size [15, 16], thermal history in some [17, 18], and are very sensitive to the
presence and structur e of carbon [19–21].
With the advent of Raman imaging , that information can be expanded
from a point measurement to images of petrographic thin sections or even un-
prepared rock, presented in terms of mineral phase maps or images of mineral
properties (Fig. 1.6).
Where electron probe or SEM/EDS imaging produces maps of mineral
elemental compo sition that can identify, for example, a n SiO
2
phase, Raman
imaging of the sa me phase c an generate maps of quartz, c oesite, tridymite,
cristobalite, lechatlierite or silicate glass and place that phase in a minera logic
context. Each pixel in such a Raman image can be as small as hundreds of
nanometers across, depending on the instrument, which is a finer resolution

12 Marc Fries and Andrew Steele
Fig. 1.5. Raman imaging analysis of a portion of the QUE 94366 carbonaceous chon-
drite, a meteorite found in Antarctica. A) shows a transmitted light microscopy image
with t he red box highlighting the area examined using Raman imaging. B) shows two
Raman spectra of olivine within the scan area, showing the variation in peak intensity
arising from changes in crystalline orientation relative to the polarization plane of the
excitation beam. C) is an image of intensity of a Gaussian fit to the ≈850cm
-1
Raman
peak of olivine, showing th e location of olivine within the scan area. D) shows olivine iron
content in terms of Fo (Fosterite) number, showing sub-micron variation in iron content
in th is phase. E) is an image of the ratio between the ≈820 and ≈850cm
-1
Raman peak
intensities, ind icating crystalline orientation of the various grains. Note th at the small
grains in the lower center of the image are revealed to be polycrystalline or separate from
the larger grains. Also notice that variations in iron content exist within single grains,
as shown by comparing D) and E).
than that featured by many analytical techniques commonly employed in min-
eralogy/petrography. Additionally, each Raman s pectrum can c ontain Raman
sp e ctral lines of several mineral phases and by (a happy) coincidence most
common rock types generate lines that do not interfere spectrally with those
of other phases. For example, have a look at Fig. 1.7.
The mineralogy at the rim of a meteorite chondrule shown there is to o
fine-grained to completely characterize the minerals present us ing optical pet-
rography techniques. Raman imaging reveals the various phases present and

1 Raman Imaging in Mineralogy and Petrography 13
Fig. 1.6. Alteration vein in the MIL 03346 martian meteorite. Left: Reflected light
image of a polished thin section. Right: Raman image with mineral phases highlighted.
Red is jarosite, green is goethite, an d blue are clay minerals. Isotopic analysis of these
phases show they are martian in origin, which provides researchers with samples of
martian a queous alteration products for study [from Vicenzi et al 2007].
is used to produce very finely detailed images of each of them, regardless
of whether they are transparent silicates or opaque carbonaceous material.
This capability allows not only mineralogy by Raman spectrosc opy but a lso
petrography by Raman imaging. In other words, a bulk rock can be identi-
fied through identification and morphology characterization of its constituent
minerals through Raman imaging.
1.3.2 Crystallographic Orientation Imaging
Fundamentally, Raman spectroscopy works by exciting vibrations in the crys-
tal matrix of a target material using an excitation laser. Some of these vi-
brations propagate along a particular crystallo graphic direction or directions
such that they propagate asymmetrically through the crystallo graphic unit
cell. In this case, changing the vector of the excitation beam relative to the
vibration propagation vector will diminish the likelihood of exciting that par-
ticular v ibrational mode. The practical effect of this feature is that the same
mineral phase in different crystallographic orientations can exhibit a peak or
peaks with variable intensity. Raman imaging can reveal the orientation of
these grains by generating a calculated imag e of the ratio of the intensity of
an orientation-dependent Raman peak to another p e ak in the same mineral
that is relatively invariant in intensity Fig. 1.8. This calculation has the added
benefit of no rmalizing the image to remove relative phase spectrum intensity
from consideration. The re sulting image reveals the orientation of all the g rains
of a particular phase in a sca n field, which allows measurement of grain size
as well as providing more detail in the determination of petrogr aphic phase
relationships.

14 Marc Fries and Andrew Steele
Fig. 1.7. Identification of individual mineral phases in a complex, fine-grained mixture.
A) Transmitted light microscopy image of the fine-grained rim on a chondrule in t he QUE
94366 CV-type carbonaceous chondrite. A Raman image was collected approximately
within the red square, and the scale bar is 10µm long. B) Image of the olivine within th e
scanned area, as seen from the intensity of the 820cm
-1
olivine peak. C) Pyroxenes within
the scan, as measured using the intensity of the ≈1010cm
-1
peak. D) Macromolecular
carbon as seen from the intensity of the ≈1580cm
-1
G band. E) Whitlockite, an anhydrous
phosphate mineral. These phases are distinguished by Raman spectral features on a scale
that is impractical for standard petrographic analysis.
The example shown in Fig. 1.8 shows quartz grains in the early Devonian
Rhynie Chert ima ged using the ra tio of the 128cm
-1
peak intensity to that
of the 465cm
-1
Raman peak. The peak at 128cm
-1
is sensitive to cr ystalline
orientation, allowing the calculation of this image. Many minerals exhibit
Raman peaks with orientation-sensitive peak intensity and so this technique
can be widely applied.
1.3.3 Phase Composition Imaging
Variations in the crystalline unit cell dimensions of a phase are the basis fo r
measurement of phase composition variation, as well as for latent strain. In the
case of pha se composition, variations in unit cell parameters arise from sub-
stitution of one atom fo r another. A good e xample of this phenomenon is the
replacement of iron with magnesium (or vice versa, depending on the mech-
anism at work) in the case of the fayalite (Fe
2
SiO
4
) / forsterite (Mg
2
SiO
4
)
solid solution (Fig. 1.5D). The Raman spectra of these phases are similar
and are dominated by symmetric and anti-symmetric SiO
4
tetrathedra vibra-
tional modes, with a continuous shift to higher wavenumbers concurrent with
replacement of iron with magnesium (or vice versa, if that’s the way you’d

1 Raman Imaging in Mineralogy and Petrography 15
Fig. 1.8. Left image: Image of quartz from a thin section of the Rhynie chert. Right
image: Dimensionless quartz grain orientation image assembled by calculating the rel-
ative intensity variation of the 132cm
-1
quartz peak relative to the 466cm
-1
peak. The
132cm
-1
peak varies in intensity with change in orientation to the excitation beam but
the 466cm
-1
does not. Expressing the image as a ratio removes intensity variation due t o
local quartz abundance and oth er effects. Sample provided by M. Schweizer, Oxford U.,
UK.
prefer to loo k at it!). This feature is useful for spot Raman analysis, but
is especially informative in the form of calculated Raman images of mineral
composition. This analysis must be performed with a pair of cavea ts in place,
namely:
1) latent c rystal s train can also produce a Raman peak shift and must be
considered, and
2) Raman peak shift is a function of atomic replacement, but it is not neces-
sarily specific as to which atom is substituted.
As an example of the latter, s ome olivine grains in extraterrestrial samples are
known to feature enhanced manganese up to several weight per c ent (LIME,
or Low-Iron Manganese Enriched olivine; [22]). An elemental analysis tech-
nique such as electron probe analysis is necessary to discern this effect from
iron/magnesium variation, but in any event the shift of Raman peaks in olivine
can be used to produce images of Fe-c ontent variation. In the case of olivine
composition, LIME olivines are uncommon and are only known to occur in un-
equilibrated, ancient meteoritic materials, so are not usually a consideration.
Other minerals such as the pyroxene family, however, feature cation substitu-
tion by a range of elements to include iron, c alcium, aluminum, magnesium,
chromium, and other s and so elemental substitution is much more difficult to
deconvolute from Raman spectra alone. Some researchers have addressed this
effect, however, and produced systematic tr e nds using natural samples [10,23].

16 Marc Fries and Andrew Steele
In the case of latent crystal s train, this effect usually shows a pronounced
anisotropy relative to the original shock propagation vector and is revealed
in Raman images as a deviation in a sing le direction across se veral grains.
Compositional changes , in contrast, usually vary r adially through individual
grains to produce mantle-and-rim microstructure or some other texture, which
is clearly tied to grain morphology. Latent strain effects in natural samples a re
typically rare, as most lithologies which show strain effects (i.e. ≈meta morphic
rocks which have experienced plastic flow) are e xposed to elevated tempera-
tures at the same time and small-scale strain tends to be reduced or eliminated
through annealing. Shocked material s uch as those associated with terrestrial
impact events, however, can exhibit latent strain arising from the impact shock
itself, and p otentially with associated phase chang e s, which may or may not
revert to low-pressure phases upon post-shock relaxation/annealing.
1.4 Examples of “Raman Petrography” Applications
1.4.1 Raman Analysis of Shocked Minerals
In nature, the most pronounced example of shock processing is seen in impact-
altered mineralogy. Examples of shock processing are see n in terrestrial im-
pact craters and materials ejected from craters, meteorites which have ex-
perienced impact shock prio r to falling to Earth, and terrestrial explosion
crater materials s uch as those found in and around nuclear test pits. The
shock in more materials is so intense, such as examples from the ureilite me-
teorites [24] and terrestrial impact craters such as the Popigai impact struc-
ture [25], that gra phite or other carbon pre c ursors has been transformed to
diamond. Other researchers have noted the presence of a high pressure form
of rutile (TiO
2
) [26], known as alpha- PbO
2
-structured TiO
2
in the Ries cr ater
in Germany [27], the Chesapeake impact structur e [28], and in shocked grains
associated with the Australasian tektite field [29]. These mineral phases are
relatively stable following their shock-induced phase change, as is the SiO
2
polymorph coesite [30–32], and may be signatures of impact processing. Some
shock-derived phases, howe ver, ar e stro ngly metastable at room temperature
and pressure and can revert to an ambient-stable phase upon heating with
the excitation la ser.
Collecting Raman images of shocked samples is difficult and re quires care-
ful characterization of the power output of the instrument. Even then, the
signal/noise ratio of data in the resulting images is often quite low out of
necessity, to prevent damag ing the sample with excessive laser power. Even
so, images of the petrography of shock-formed phas e s can be use ful, such
as tha t of c oesite fr om the Ries cr ater breccia shown in Fig. 1.9. In addi-
tion shock-processed samples tend to luminesce under the excitatio n beam,
as they typically contain a high defect density. Nonetheless, the s pectra in
Fig. 1.9 are sufficient to identify c oesite and the ima ge is of sufficient quality
to characterize its morphology.

1 Raman Imaging in Mineralogy and Petrography 17
Fig. 1.9. Coesite in impact suevite from th e Ries meteorite impact crater, Germany.
Left image: Silicate glass. Right image: Small bleb of coesite in the upper right hand
corner co-located with “hole” in the glass image on the left. Note the noisy image
caused by low excitation power used to preserve this metastable phase combined with
the inherent luminescence of shocked materials. Even with this limitation, however, both
the placement and identity of coesite are identifiable.
1.4.2 Contextual Imaging of Carbonaceous Materials
Raman sp e ctroscopy of condensed carbon phases such as diamond, kerogen,
polymers and macromolecular carbon (MMC, i.e. c ondensed, reduced carbon
spanning a wide range of cry stalline order) in genera l is a topic of special in-
terest when working with minerals. For one thing, the spectrum of all of these
phases is particularly intense compared to silica tes , allowing for the detection
of very small amounts of material. Additionally, other analytical methods com-
monly used in mineralogy a nd petrography, such as x-r ay diffraction (XRD),
are not especially sensitive to poorly ordered carbonaceous materials such
as kerogen and ma c romolecular carbon. Raman imaging allows placement of
these phases in a mineralogical context, which is a unique analytical capabil-
ity. Furthermore, Raman s pectra of carbonac e ous phases are very sensitive to
their degree of crystallographic order, which allows conclusions to be drawn
on the thermal history and formation conditions of these phases [17, 33, 34].
The Raman spectrum of MMC exhibits three principal features, namely
the “disordered” or D band around 1350cm
-1
, the “graphitic” or G band

18 Marc Fries and Andrew Steele
around 158 0cm
-1
[20,35,36], and a series of second-order Raman modes found
in the v ic inity of 2700cm
-1
[37] (Fig. 1.10).
Fig. 1.10. Comparison of th e Raman spectra of a few carbon phases. A) A typical thin
section mount ing polymer, with prominent C-H bands revealing the aliphatic nature of
this compound. Note that this spectrum is immediately distinct from the reduced carbon
phases h ere, which allows for definitive discrimination between mounting polymer and
native carbon species in a thin section. B) S ingle-crystal diamond spectrum showing the
distinct, very intense and sharp 1332cm
-1
Raman peak. C) Graphite from the Toluca iron
meteorite a s a n example of highly crystalline, graphitic carbon.

1 Raman Imaging in Mineralogy and Petrography 19
As visible in this figure, the primary Raman mo des of high-aromaticity
condensed ca rb on are labeled - the D (“disordered”) band, the G (“graphitic”)
band, and the various additive and multiplicative second-order bands. Crys-
talline graphite can be devoid of a D band, but this example has been pol-
ished (see text). D) Carbon from a chondrule in the QUE 94366 CV-type
carbonaceous chondrite. This example has been processed at high tempera-
ture but apparently not for as long as the Toluca example. The peaks are
narrow, which indicates relatively small distribution of graphitic crystallite
sizes, but the D band is intense relative to the G band, indicating an ove rall
disordered structure. Minor grossite (Gr) is present in this spectrum. E) An
example of kerogenous material from the Gunflint microfossil. The G band
is r e latively narr ow and intense indicating gr owth of a small size distribution
of graphitic domains under thermal metamorphism, and the complex shape
of the D band may indicate a wide range of disordered conformations arising
from heteroa tom enrichment. F) The subbituminous (C) coal DECS 1 from
the US DOE Coa l Sample Bank and Database , an e xample of a poorly ordered
coal. Broad D and G bands indicate a poorly ordered structur e overall.
In some ear lier liter ature, the G band is referred to as the “ordered” or
O band. Tuinstra and Koenig [20] we re the first to point out that the mean
crystalline domain size is proportiona l to the D/G band intensity ratio, with
refinement by Knight and White [38], followed by Matthews et al [39] that
accounts for variation due to laser excitation wavelength. Ra man spectra of
carbon phases are unusual in two aspects. For one, ther e is a measurable
shift in the D band Raman mode peak position with excitation wavelength
[39, 40]. This shift is highly unusual, as Raman modes are essentially energy
loss from an incident photon due to crystalline vibrational modes of specific
energy and so typically occur at the same Raman shift regardless of excitation
wavelength. For the other aspect, the second-order Raman modes commonly
seen in condensed c arbon are unusual in that most mineral species exhibit
first-order modes only.
As a mineralogical phase, condensed carbon covers a range of structures
ranging from diamond with its sp
3
-hybridized molecular orbitals and face-
centered cubic structure [19] to graphite, which is composed entirely of sp
2
-
hybridized carbon arranged in a hexagonal lattice. Between these end mem-
bers (and, arguably, a third end member composed of cry stalline aliphatic
polymer), condensed carbon spans an infinitely variable degree of crystalline
order, ranging from amorphous soot to poorly-ordered kerogens and mildly
heated condensed car bon, through a continuum of crystalline ordering up
to single-crystal graphite. Much of the naturally-occurring carbo n found as
a mineralogical pha se appears somewhere along this continuum. Generally
sp e aking, MMC found in rocks either precipitates fr om hydrothermal systems
[41], precipitates from mantle-derived fluids and in some pegmatites [42, 43],
or it accumulates from organic matter (many references , e.g. [44]), then un-
dergoes metamorphism along with its parent rock. Steele et al [42] notes an
abiogenic MMC formation mechanism from condensation of mantle-derived

20 Marc Fries and Andrew Steele
volatiles in both terrestrial and martian rocks, implying that this is a com-
mon process in rocky bodies large enough to support both carbon-containing
volatiles at depth as well as intrusive volcanism. The MMC becomes systemat-
ically, more crys talline under heat and pressure while losing volatiles (which
is why coal seams ca n contain pockets of methane a nd/or carbon dioxide),
reaching a fully graphitized form above 400
◦
C accor ding to Landis [45] and
at eclogite fac ie s (500
◦
C and 20 kbar ) according to Beyssac et a l [46]. Exten-
sive graphitization is also noted in carbonaceous materials metamorphosed
to g ranulite [7 ] and greenschist fa c ie s [47]. During metamorphism, trends are
seen in b oth G and D band peak center position (i.e. [48]), G and D band
width [49], as well as D/G intensity and area ratios [47]. Wopenka and Pasteris
[1993] report trends in the intensity of the first overtone of the D band as well,
which they define as the “S” peak [49]. Trends are also seen with respect to the
La (i.e. crystallite length in the a direction parallel to the graphite basal plane)
grain size measurement defined by Tuinstra and Koenig [20] as well, where
greater thermal treatment generally leads to large r La grain size. Generally
sp e aking, incr easing thermal metamorphism leads to greater crystallizatio n
(graphitization), which dr ives the G band center towards ≈1582cm
-1
, G band
width towards a ins trument- limited low value, the D band towards a value that
depends on the excitation wavelength [39], and the D/G intensity and area
ratios towards zero. Unfortunately, deconvolution of the D band can produce
large fitting errors without resorting to fitting the band to multiple peaks,
especially for relatively amor phous materials thus complicating the use of D
band features a s a function of carbon metamo rphism. By contrast, use of the
G band is relatively straightforward as long as the D’ band at 161 0cm
-1
when
that band is prominent. Much of the information to be found in a Ra man
sp e ctrum of carbon can be ex pressed with a G band positio n vs. G band
width (FWHM) graph, which reveals a trend of gra phitizatio n grade as well
as a distinct portion of the graph around 1582cm
-1
and ≈20c m
-1
FWHM, re-
sp e ctively, that indicates material featuring gra phitic domains that are large
with respect to the excitatio n beam.
One form of poorly ordered ca rb on, “glassy carbon” [50,51], exhibits little
or no long-range crystallographic ordering but, according to the nomencla-
ture committee of the International Union of Pure and Applied Chemistry
(IUPAC), is not truly amorphous, as it exhibits consistent short-range or-
dering and contains few or no dangling bonds [52]. While glassy carbon is
best known as a synthetic phase, it is worthy of mention in a mineralogi-
cal context because other studies [46] show that any carbonaceous material
arising from organic c ompounds can show resistance to graphitization even
at high temperature. The mechanism at work is the same in both synthetic
and na tur ally occurring phases. Apparently, g raphitization is impeded by the
presence of a significant atomic percent of heter oatoms such as O, N, a nd
possibly S, leading to a chemically heterogenous, carbonaceo us phase which
resists metamorphism. This problematic twist means that heteroatom content
complicates the analysis of the metamorphic history of MMC. It may be pos-

1 Raman Imaging in Mineralogy and Petrography 21
sible to tightly constrain the age and thermal history of a carbonaceous phase
if bo th the crystalline structure and heteroatom content are known, and this
topic is ripe for further scientific investigation.
1.4.3 Fluid Inclusions
Fluid inclusions are e xactly what they sound like - small pockets of fluid within
a mineral grain. They hold special impor tance in the study of minerals because
they represent samples of volatile compounds in the formation environment
and subsequent metamorphism. The study of volatiles in deeply emplaced
rocks such as peridotites, for example, provides information ab out the melting
points and physical behavior of rocks in the deeper parts of the Ear th’s crust
or even the mantle (re view by Anderson [53]). Fluid inclusions include two
general types - primary and seco ndary inclusions.
Primary inclusions are essentially trapped pockets of fluid within grains
that date to the origina l crystallization of the host phase. They commonly con-
tain multiple phases, including glass, and ar e concentrated in elements that
feature poo r solubility in the pare nt mineral. Basically, primary fluid inclu-
sions are formed as the surrounding grain solidifies and rejects incompatible
elements, which concentrate as dr oplets and are frozen in place. These inclu-
sions tend to have “fluid-like” shapes with droplet-like morphology or possibly
reverse-crystal morphology that mirrors the crysta l shape o f the surrounding
mineral, a nd tend to occ ur alone or in small groups removed from the edges
of their pare nt grain (e.g. [54]). Sometimes, pr imary fluid inclusions in miner-
als that form at high temperature and pressur e will contain vapor bubbles in
them ca used by shrinkage of the parent mineral during cooling. These vapor
bubbles can sometimes contain a relative vacuum due to this shrinkage, as
improbably as that may seem.
Secondary fluid inclusions are often found lying in a plane or along a line
within a mineral grain. These inclusions form when the pare nt grain cracks
while still at relatively high pressure and/or temperature in the presence of
volatiles. The volatiles per colate into the crack and form pockets while the
parent grain “ heals” itself through grain regrowth. The resulting trains of in-
clusions include mater ial present after the rock has already crystallized from
its parent melt, and their contents provide insight into the metamorphic his-
tory of the parent rock. There is extensive literature on the analysis and
interpretation of fluid inclusions (e.g. [54, 55]) and for the purposes of this
chapter it is sufficient to simply point this out and comment that Ra man
sp e ctroscopy and confocal Raman imag ing is especially well suited to fluid in-
clusion analysis (see [56] for a review). A visible-light Raman excitation be am
can probe the contents of fluid inclusions within transparent grains such as
olivine, pyroxenes and quartz without ex po sing the inclusion at the thin sec-
tion surface [57]. This allows for direct measurement of the content of the fluid
phase, and virtually eliminates the possibility of contamination if the sample
preparation and Raman analysis are done carefully.

22 Marc Fries and Andrew Steele
Raman analysis of fluid inclusions has been performed o n meteorites in
one chrondritic setting. To da te, two chondrites - the Monahans and Zag
ordinary chondrites - have been found to contain halite/sylvite grains with
trapped brine inclusions. The parent salt grains are often a brilliant purple
color due to irradiation, indicating that these grains are very old [58 ,59 ]. This
is a remarkable find, as up to this point no fluid-filled inclusions had been
conclusively proven in any type of meteor ite. The existence of fluid inclusions
in Mona hans and Zag showed that liquids could exist at least for a shor t time
on smaller asteroidal bodies.
1.4.4 Ancient Terrestrial Carbonaceous Materials
The topic of Raman imaging of ancient carbonaceous materials, either fos-
siliferous or just purportedly so, is an expansive topic that can warrant its
own chapter. Much of the study of ancient carbonaceous materials has fixated
on as signing an age to the earliest life on Earth, perhaps to the detriment
of studies of the actual environmental conditions present when these forma-
tions - and perhaps life itself - came into being. Nonetheless, much research
into ancient car bo naceous materials features Raman spectrosc opy and Raman
mapping or imaging. Some researchers have made the claim that some par-
ticular ca rbonaceous material is biogenic based on Raman spectra [60] while
others have claimed otherwis e [61] and even demonstrated that this is not ten-
able due to lack of s ufficient differentiability between biogenic and abiogenic
carbonaceous materials, esp e c ially for ancient materials that have experienced
extensive metamorphism [41, 62]. Uncertainties arise in the interpretation of
this informatio n both in how it pertains to the genesis of the carbonaceous
material and in the p e trologic history of its host formation. This uncertainty
is magnified with increas ing age of the ma terial, as both chemical degrada-
tion and thermal metamor phism add ambiguity to both the o rigins of the
carbonaceous material and its host rock. There is no doubt that considerable
information can be gleaned o n the structure of carb onaceous materials us-
ing Ra man spectroscopy, a nd that such information tied with spatial context
through Rama n imaging gives a tremendous amount of information about the
nature of a given ca rb onaceous material and its microenvironment (Fig. 1.11).
Some work has been done in describing fossil carbonaceous material rel-
ative to coal, which is fairly well understood, with the result tha t Nestler
et al. [63] described Permian fossilized tree fern materials as having reached
anthracite rank during metamor phism.
1.5 Raman Mineralogy in Field Geology Studies
With the deve lopment of small, reasonably sensitive spectrogr aph/detector
combinations, minia tur e so lid-state lasers, and field-ruggedized computers,
portable Raman spectroscopy devices are now c ommercially available. Several

1 Raman Imaging in Mineralogy and Petrography 23
Fig. 1.11. Raman image of kerogenous material from the Precambrian Gunflint Ch ert.
The upper image is made from the intensity of a Gaussian fit to th e G band of car-
bonaceous material in the chert. The matrix (not shown h ere) is quartz, as is b efitting
a chert. The lower image is calculated to show the graphitic domain size in the α crys-
tallographic direction, or L
α
[Matthews et al 1999], regardless of G band intensity. The
scale bar shows L
α
in angstroms, and the lighter colored areas of the image represent rel-
atively fine-grained carbonaceous material. This effectively reveals sub-micron variation
in kerogen structure, which most likely represents metamorphosed remnants of structure
from the organic matter precursor. Sample provided by M. Schweizer, Oxford U., UK.
of these ins truments are hand-held, allowing the user to analyze the miner-
alogy a nd biological status of outcrops and other samples under their native
environmental conditions [64]. Most instruments are even specially designed
for performing field mineralogy, and many a llow the user to match spe ctra
with any da tabase to include mineral spectral libraries. Some of the pioneers
in this application brought full-sized Raman instruments designed for lab o-
ratory use into remote field locations [65] for both, mineralogy and a strobi-
ological studies. Modern field Raman instruments are capable of identifying

24 Marc Fries and Andrew Steele
most mineral species, but cur rent available models typically cannot query C-H
and O -H stretch modes at high Raman shift, and are limited to interrogat-
ing relatively robust mineral species due to their intense laser flux. The high
laser power density required fo r field Raman is a conseq uence of the un-cooled,
relatively insensitive detectors that are required for sma ll, battery powered in-
struments. Future technology improvements in the areas of battery efficiency
and/or detector sensitivity may allow for low power density laser sources that
will expand the ra nge of mineral phases identifiable using hand-held Raman
instruments. At present, the capability to easily collect Raman spectra in a
field environment is a powerful mineralogy tool. In the future, instruments will
undoubtedly improve on this capability, perhaps to include Raman imaging
capabilities.
1.5.1 Extraterrestrial Exploration
At the time of this writing, no Raman spectrometer has flown on a space ex-
ploration mission. While Raman is obviously of very little utility for flyby or
orbiter missions (a la ser powerful enough to produce a Raman signal observ -
able from orbit would produce an impressive network of glassy sidewalk s as
a by-product!), small Raman spectrometers would be extraordinarily useful
for both manned a nd unmanned landed missions [66, 67]. In pursuit of this,
a miniature Raman instrument is currently under development for the Eu-
ropean Space Agency’s (ESA) ExoMars rover mission [68]. The goal of this
mission is to search for life and/or evidence of past life on the martian surfa c e
using a variety of instruments. Such an ins trument would have to be suffi-
ciently sensitive to c ollect spectra at low laser power density values in order
to detect organic compounds, and bo th the instrumental parameters and final
instrument manifest for the mission are undecided at present. A Raman in-
strument called the Mars Multibeam Raman Spe c trometer (MMRS; [69–71]
was also one of the instruments developed for use on NASA’s MER rovers
(Mars Exploration Rovers, “Spirit” and “Opportunity”) that are actively ex-
ploring the martian surface at the time of this writing. Ultimately, however,
the MMRS was passed over for other instruments.
A Raman device in general, utilized on a r ob otic rover/lander or as a hand
tool fo r astronauts, has the advantages of producing very fast identification of
mineralogy as well as the potential to directly identify the pres e nce of some
organic compounds, or at least point out the presence of a wide variety of
carbonaceous phases that may be consistent with evidence of extinct life. As
such, Raman spectroscopy is particularly well suited to science priorities iden-
tified for exploration of Mars [66]. Rapid mineralogy investigations possible
using Raman spectroscopy methods are of particular interest in exploration of
the Moon, asteroids, and possibly comets as well. What about using a Raman
instrument on Venus, however? Since any venusian mission would be sharply
constrained in mission duration (due to the destructively high temperature),
sample handling and other factors, some researchers point to the speed and

1 Raman Imaging in Mineralogy and Petrography 25
relative simplicity of Raman analysis as advantages for mineralogy studies on
our sister planet. Significant obstacles exist in the form of atmospheric scatter
and absorption in both the excitation beam and Raman signal, the difficulty
of collecting a signal through hot optics, and other problems associated with
operating in the 92 MPa pressure and ≈4 60
◦
C conditions on the venusian
surface [72]. Raman spectroscopy may be especially well suited to this envi-
ronment, however, as its rapid acquisition speed should allow the analysis of a
large number of sa mples, and work has been done on analyzing Venus-specific
mineralogy in particular [73]. Work with pulse d Raman systems at a distance
through a simulated venusian atmosphere have als o pr oduced usable spec-
tra [74]. While this application is intriguing, significant technical challenges
must be overcome to make it practical.
A Raman device in general, utilized on a r ob otic rover/lander or as a hand
tool fo r astronauts, has the advantages of producing very fast identification of
mineralogy as well as the potential to directly identify the pres e nce of some
organic compounds, or at least point out the presence of a wide variety of
carbonaceous phas e s that may be consistent with evidence of extinct life.
1.6 Conclusion
For future applications it is reasonable to expect that technological advance-
ments will deliver improvements in detector sensitivity, power supply effi-
ciency, computing power, more compact lasers and who knows what all else
as time passes. Future generations of hand-held Raman systems can conceiv-
ably include hand-held instruments for astronauts’ use on the lunar, martian
or other planetary surfaces. Such an instrument could provide a rapid min-
eralogical assessment of rocks and outcrops and could serve as a first-pass
tool for astrobiological analysis. Rock tar gets identified as interesting by a
quick, ha nd- held Raman analysis would conceivably be returned to a labora-
tory for in-depth study, perhaps to include Raman imaging for petrographic
and petrolo gic analysis. Given present c apabilities and future possibilities, it
is easy to envision a not-so-dista nt future where field scientists wield small,
powerful Raman instruments as a matter of co urse both on this planet and
others.
References
1. J. Pasteris, Applied S pectroscopy 43, 1989 (567-570)
2. A. El Goresy, P. Gillet, S. Mostefaoui, M. Chen, V. Masaitis, Lunar and Plan-
etary Science X XXIII p. 1031 (2002)
3. M. Zolensky, . co authors, Science 314, 1735 (2006)
4. D. Fischbach, Carbon 3, 342 (1965)
5. H. Murty, E. Biederman, D.and Heintz, Carbon 7, 667 (1969)

26 Marc Fries and Andrew Steele
6. H. Murty, D. Biederman, E. Heintz, Carbon 7, 683 (1969)
7. B. Wopenka, G. Matrajt, S. Bajt, Lu nar and Planetary Science XXXIX p. 1827
(2008)
8. B. D., et al., Science 314, 1711 (2006)
9. R. Downs, Program and Abstracts of the 19th General Meeting of the Interna-
tional Mineralogical A ssociation in Kobe O03, 13 (2006)
10. A. Wang, B. Jolliff, L. Haskin, K. Kuebler, K. Viskupic, American Mineralogists
86, 790 ( 2001)
11. M. Fries, A. Steele, Lunar and Planetary Science XXXVIII p. 2195 (2007)
12. S. Prawer, K. Nugent, D iamond and R elated Materials 7, 215 (1998)
13. L. Wong, C. Wong, J. Liu, D. Sohn, L. Chan, L. Hsia, H. Zang, Z. Ni, Z. Shen,
Jpn. J. Appl. Phys. 44, 7922 (2005)
14. H. Ishibashi, M. Arakawa, S. Ohi, J. Yamamoto, A. Miyake, H. Kagi, J. Raman
Spectrosc. 39, 1653 (2008)
15. A. Wang, Lunar and Planetary Science Conference XXX p. 1644 (1999)
16. M. Salis, P. Ricci, A. Anedda, J. Raman Spectrosc. 40, 64 (2009)
17. O. Beyssac, L. Bollinger, J. Avouac, G. B., Earth and Planetary Science Letters
225, 233 (2004)
18. G. Cody, C. Alexander, H. Yabuta, A. Kilcoyne, T. Araki, H. Ade, P. Dera,
M. Fogel, B. Militzer, B. Mysen, Earth and Planetary Science Letters 272, 446
(2008)
19. S. Solin, A. Ramdas, Phys. Rev. B 1, 1687 (1970)
20. F. Tuinstra, J. Ko enig, J. Chem. Phys. 53, 1126 (1970)
21. J. K oenig, Applied S pectroscopy Reviews 4, 233 (1971)
22. K. Kl¨ock, W. adn Thomas, D. McKay, H. Palme, Nature 339, 1989 (126-128)
23. A. Wang, L. Haskin, B. Jolliff, K. Kuebler, Lun ar and Planetary Science Con-
ference XXXI (2000)
24. C. Goodrich, Meteoritics 27, 327 (1992)
25. V. Masaitis, M. Mikhailov, T. Selivanovskaya, Meteoritics 7, 39 (1972)
26. L. Gerward, J. Olsen, J. Appl. Cryst. 30, 259 (1997)
27. A. El Goresy, M. Chen, P. Gillet, L. Dubrovinsky, G. Graup, R. Ahuja, Earth
and Planetary Science Letters 192, 485 (2001)
28. J. Jackson, J. Horton, I. Chou, H . Belkin, American Mineralogist 91, 604 (2006)
29. B. Glass, M. Fries, Meteoritics and Planetary Science 43, 1487 (2008)
30. L. Coes, Science 118, 131 (1953)
31. H. Boyer, D. Smith, C. Chopin, B. Lasnier, Physics of Chemistry and Minerals
12, 45 (1985)
32. V. St¨ahle, R. Altherr, M. Ko ch, L. Nasdala, Contributions to Mineralogy and
Petrography 155, 457 (2007)
33. J. Pasteris, B. Wopenka, The Canadian Mineralogist 29, 1 (1991)
34. T. Yui, E. H uang, J. Xu, Journal of Metamorphic Geology 14, 115 (1996)
35. C. Castiglioni, C. Mapelli, F. Negri, G. Zerbi, Journal o f Chemical Physics 114,
963 (2001)
36. S. R eich, C. Thomsen, Ph il. Trans. R. S oc. Lond. A 362, 2271 (2004)
37. R. Nemanich, S. Solin, Phys. Rev. B 20, 392 (1979)
38. D. Knight , W. W hite, Journal of Materials Research 4, 385 (1989)
39. M. Matthews, M. Pimenta, G. Dresselhaus, M. Dresselhaus, M. Endo, Physical
Review B 59, R6585 (1999)
40. A. Sood, R. Gupta, S. Asher, Journal of Applied Physics 90, 4494 (2001)

1 Raman Imaging in Mineralogy and Petrography 27
41. J. Pasteris, B. Wopenka, Astrobiology 3, 727 (2003)
42. A. Steele, M. Fries, H. Amundsen, B. Mysen, M. Fogel, M. Schweizer, N. Boctor,
Meteoritics and Planetary S cience 42, 1549 (2007)
43. M. Tsunogae, T.and Santosh, J. Dubessy, Y. Osanai, M. Owada, T. Hokada,
T. Toyoshima, Geological Society, London, Special Publications 308, 317 (2008)
44. W. Francis, Coal: its Formation and Composition (Edward Arnold (Ed.) Ltd.,
London, 1954)
45. C. Land is, Contrib. Mineral. Petrol 30, 34 (1971)
46. O. Beyssac, F. Brunet, J.P. Petitet, B. Goffe, J.N. Rouzaud, Eur. J. Mineral.
15, 2003 (937-951)
47. M. Tice, B. Bostick, D. Lowe, Geology 32, 37 (2004)
48. O. Beyssac, J.N. Rouzaud, B. Goffe, F. Brunet, C. Chopin, Contrib. Mineral.
Petrol. 143, 19 (2002)
49. B. Wopenka, J. Pasteris, American Mineralogist 78, 533 (1993)
50. R. Franklin, Proc. R. Soc. Lond. A 209, 196 (1951)
51. P. Harris, Philosophical Magazine 84(29), 3159 (2004)
52. E. Fitzer, K.H. Kochling, H. Boehm, H. Marsh, Pure. Appl. Chem. 67, 473
(1995)
53. E.R. Anderson, T.and N eumann, Lithos 55, 1 ( 2001)
54. E. Roedder, Fluid Inclusions Vol. 12. Reviews in Mineralogy (Ribbe, P. E.,
1984)
55. L. Diamond, Lithos 55, 69 (2001)
56. E. Burke, Lithos 55, 139 (2001)
57. J. Dubessey, B. Poty, C. Ramboz, European Journal of Mineralogy 1, 517 (1989)
58. M. Zolensky, R. Bodnar, D. Bogard, D. Garrison, E. Gibson, L. Nyquist,
Y. R eese, C.Y. Shih, H. Wiesmann, S cience 285, 1377 (1999)
59. A. Rubin, M. Zolensky, R. Bodnar, Meteoritics and Planetary Science 37, 125
(2002)
60. W. Schopf, A. Kudryavtsev, D. Agresti, T. Wdowiak, A. Czaja, Nature 416, 73
(2002)
61. M. Brasier, O. Green, A. Jephcoat, A. Kleppe, M. van Kranendonk, J. Lindsay,
A. St eele, N. Grassineau, Nature 416, 76 (2002)
62. J. Pasteris, B. Wopenka, Nature 420, 476 (2002)
63. K. Nester, D. Dietrich, K. Witke, R. R¨oler, G. Marx, J. of Molecular Structure
661-662, 2003 (357-362)
64. J. Jehliˇcka, P. Vtek, H. Edwards, M. Heagraves, T. Cap oun, Spectrochimica
Acta A (2008)
65. D. Wynn-Willams, H. Edwards, Icarus 144, 486 (2000)
66. A. Ellery, D. Wynn-Williams, Astrobiology 3, 565 (2003)
67. N. Tarcea, T. Frosch, P. R¨osch, M. Hilchenbach, T. Stuffler, S. Hofer, H. Thiele,
R. H ochenleitner, J. Popp, Space Sci. Rev. 135, 281 (2008)
68. I. Escudero-Sanz, B. Ahlers, G. Courreges-Lacoste, Optical Engineering 47,
033001 (2008)
69. A. Wang, L. Haskin, A. Lane, T. Wdowiak, S. Squyres, R. Wilson, L. Hovland,
K. Manatt, N. Raouf, C. Smith, Journal of Geophysical Research Planets 108,
5.1 (2003)
70. A. Wang, K. Kuebler, B. Jolliff, L. Haskin, Journal of Raman Spectroscopy 35,
504 (2004)
71. S. Sharma, A. Wang, L. Haskin, Lunar and Planetary Science Conference
XXXVI (2005)

28 Marc Fries and Andrew Steele
72. K. Lodders, B. Fegley Jr., The Planetary Scientist’s Companion (Oxford Uni-
versity Press, 1998)
73. A. Wang. Planetary raman spectroscopic study for understanding venus evo-
lution h istory. Venus Geochemistry: Progress, Prospects, and New Missions
(2009). Abstract No. 2003
74. S. Clegg, J. Barefield, R. Wiens, C. Quick, S. Sharma, A. Misra, M. D yar, M. Mc-
Canta, L. Elkins-Tanton. Venus geochemical analysis by remote raman-laser
induced breakdown sp ectroscopy (raman-libs). Venus Geo chemistry: Progress,
Prospects, and N ew Missions (2009). Abstract No. 2013