Research article

Received: 24 April 2012 Revised: 3 July 2012 Accepted: 8 July 2012 Published online in Wiley Online Library: 2 October 2012

(wileyonlinelibrary.com) DOI 10.1002/jrs.4161

Imaging of internal stress around a mineral

inclusion in a sapphire crystal: application
of micro-Raman and photoluminescence
spectroscopy
Naoki Noguchi,a* Ahmadjan Abduriyim,b Ichiko Shimizu,c
Nanako Kamegata,a Shoko Odakea and Hiroyuki Kagia

We developed a micro-Raman and photoluminescence imaging technique for visualizing the internal stress fields in a sapphire
crystal. The technique was applied to an Australian sapphire gemstone with a zircon inclusion. Considering piezospectroscopic
effects on Raman and photoluminescence spectra, the Raman shifts of sapphire around the zircon inclusion were converted to
hydrostatic pressure and deviatoric components of stress tensor. The internal stress was highly concentrated at the tips of the
zircon crystal, where the deviatoric stress and the hydrostatic pressure component reached 700 and 470 MPa, respectively.
Generation of compressive stress on the crystal surface of zircon can be explained by the difference in thermal expansion
coefficients and elastic constants between sapphire and zircon. In general, internal stress fields induced by mineral inclusions
reflect the pressure and temperature conditions at which the host sapphire gemstones were crystallized. Thus, the present
technique can be utilized to identify the origin of gemstones. Copyright © 2012 John Wiley & Sons, Ltd.
Supporting information may be found in the online version of this article.

Keywords: micro-Raman spectroscopy; photoluminescence spectroscopy; sapphire; inclusion; internal stress

Introduction Here, our attention is focused on the internal stress retained in

Sapphire and ruby are precious and popular colored gemstone
varieties. The main chemical component of these gemstones is
Al2O3; sapphire also contains Ti, Fe, and Cr as impurities. The
location where these gemstones are mined is an important factor
in determining their value.[1] At present, the location can be
identified by quasi-destructive geochemical analyses, such as
laser ablation–inductively coupled plasma mass spectroscopy
and particle-induced X-ray emission (PIXE).[2–4] The development
of a nondestructive method to identify the geological origin of a
sapphire gemstone is a current objective in gemology.
Gemstones of sapphire often contain other mineral species
as inclusions.[5,6] Such inclusions become trapped inside the
sapphire, which is referred to as the ‘host’, during its formation.
Because the mineral inclusion species depend on the geophysical
and geochemical conditions where the inclusion was formed, the
identification of inclusions in sapphire can provide information
regarding its origin. Micro-Raman spectroscopy is a promising
method for the identification of inclusions in sapphire. Previous
studies have undoubtedly characterized sapphires by the identi-
fication of their inclusions using micro-Raman spectroscopy.[7,8]
However, if geological conditions at different sapphire mines
are similar, the mineral inclusion species overlap. For example,
the inclusion species in sapphires that originate from the alkali
basalt regions of Australia, Thailand, China, Cambodia, Laos,
Vietnam, and Nigeria overlap.[9–13] Thus, for a more precise classi-
fication, the inclusions need to be characterized by additional
the sapphire gemstone.
Internal stress is induced around an inclusion in a sapphire
after its ascent to the surface from the deep crust where it
originally formed. The internal stress around an inclusion is
caused by differences in the elasticity and thermal expansivity
between the host and inclusion minerals.[14] Because both these
physical constants are anisotropic, the internal stress field around
an inclusion depends on the relative crystallographic orientations
of the host and the inclusion mineral and the shape of the
inclusion. Therefore, the internal stress field is expected to be
isotropic. The internal stress field around the inclusion reflects
the pressure–temperature (P–T) conditions at the locus of the
formation of the inclusion during the ascent of the sapphire.
Thus, the determination of the stress field around an inclusion
in a sapphire may help to locate the depth of its point even for
identical mineral inclusion species. Previous studies reported
* Correspondence to: Naoki Noguchi, Geochemical Research Center, Graduate
School of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033,
Japan. E-mail: noguchi@eqchem.s.u-tokyo.ac.jp
a Geochemical Research Center, Graduate School of Science, The University of
Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
b Gemological Institute of America (GIA) Tokyo Laboratory, Taito 4-19-9, Taito-Ku,
Tokyo, 113-0016, Japan
c Department of Earth and Planetary Science, Graduate School of Science, The
147

factors that reflect the geological conditions of the source region. University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan

J. Raman Spectrosc. 2013, 44, 147–154 Copyright © 2012 John Wiley & Sons, Ltd.

internal pressure data for various host minerals (except sapphire)
or their inclusions using micro-Raman spectroscopy.[15–18] Raman
mapping was also applied to image the internal pressure field
around an inclusion.[19–22] However, no spectroscopic studies
on the determination of the components of a stress tensor
around an inclusion in a host mineral have been published.
In this study, we report a new spectroscopic method for
measuring an internal stress field around an inclusion in a sapphire
by two-dimensional (2D) Raman and three-dimensional (3D)
photoluminescence (PL) mapping measurements. This study is
the first step toward the establishment of a new method for iden-
tifying the location in which a sapphire gemstone originated.
Sample characterization
The sapphire gemstone examined in this study was obtained from
the Swamp Oak mine, Inverell district, New South Wales, Australia
(Fig. 1). With geochemical and petrological observations, several
models have been proposed for the formation process of
sapphire-bearing rocks in this region.[5,6,23,24] Two representative
models suggest that the sapphire gemstones were formed either
in the middle crust by a hybrid reaction between carbonatite melts
and silicic magma[5] or in the lower crust by a contact metamorphic
process of aluminous rocks.[6] The P–T conditions for the formation
Figure 1. Left: photograph of the slab of an Australian sapphire. Right
(top): close-up of the rectangular area where mapping analysis was
performed. A zircon inclusion is embedded in the sample. Right (bottom):
crystallographic orientations of the host sapphire grain and the zircon
inclusion.
Table 1. Comparative crystal chemistry of sapphire and inclusions
Crystal system Bulk modulus/GPa Linear thermal expansion coeff./106/K Volume thermal expansion coeff./106/K
Sapphire (a-Al2O3) Trigonal 257
Zircon (ZrSiO4) Tetragonal 227
Rutile (TiO2) Tetragonal 216
Albite (NaAlSi3O8) Triclinic 70 a: 15.3 b: 5.7 c: 2.0
148
wileyonlinelibrary.com/journal/jrs Copyright © 2012 John Wiley & Sons, Ltd.
N. Noguchi et al.
of sapphire suggested in these models are ~0.5 GPa and ~400
C
and 0.7–1.1 GPa and 700
C–900
C, respectively.
The trace element contents in the sample were measured by
laser ablation–inductively coupled plasma mass spectroscopy as
follows: Fe, 4198.7  97.1 ppm; Ti, 226.9  41.7 ppm; Ga, 89.3  1.4
ppm; Mg, 17.7  3.2 ppm; V, 8.6  0.3 ppm; and Cr, 4.1  1.3 ppm.
The sapphire grain contains several zircon (ZrSiO4), rutile (TiO2),
and albite (NaAlSiO4) inclusions. Most of the zircon and albite
inclusions are accompanied by radial cracks in the host sapphire
crystal. Table 1 lists the thermoelastic constants of these minerals.
Among them, the difference in thermoelastic constants between
host and inclusion minerals becomes largest for zircon inclusions.
Internal pressure estimates based on the isotropic elastic inclusion
model[33] (Fig. S1) demonstrated that the large difference in
the thermoelastic constants could induce high internal stress
around the zircon inclusions. Thus, we selected a zircon inclusion
with the dimensions of 30  30  65 mm3 for the spectroscopic
analysis (Fig. 1). A sapphire slab in the thickness of 2.2 mm was
prepared so that the zircon inclusion was not exposed to the
surface. Considering the selection rule for sapphire with the D3d
space group,[34] the slab was cut almost parallel to the (2 –1 –1 0)
plane so that sufficient signals can be obtained for the six
Raman bands, i.e. two A1g modes (418.5 and 645.2 cm1) and
four Eg modes (378.4, 430.6, 576.4, and 750.6 cm1). The precise
crystallographic orientation was determined by the electron
back-scattering diffraction technique using a scanning electron
microscope (JSM-7000F, JEOL Co. Ltd.) equipped with an
electron back-scattering diffraction detector. The difference in
crystallographic orientations between the host sapphire grain
and the zircon inclusion was determined by a micro-X-ray
diffractometer equipped with a multilayer confocal mirror
(Micro Max 007HF X-ray generator with an R-AXIS IV imaging
plate, Rigaku Co. Ltd.). The crystallographic orientations of these
minerals are illustrated in Fig. 1 by using the VESTA program.[35]
The c axis of zircon is almost parallel to that of sapphire; the
angle between them is 3.0
.
Experimental methods
Outlines of stress measurements
The internal stress, sij (i, j = x, y, z), of sapphire can be divided
into a hydrostatic pressure component, P, and deviatoric stress
components, sij0 (i, j = x, y, z), as
sij ¼ sij 0  Pdij
where dij is the Kronecker’s delta. Positive values of the axial stres-
ses (sxx, syy, and szz) represent tensile stresses. In high-pressure
experiments using diamond anvil cells, photoluminescence
of sapphire with wavelengths of 694.3 (R1) and 692.9 nm (R2),
which are excited at Cr3+ impurities, has been widely used for
Ref.
[25,26]
a: 7.3 c: 8.3 23.0
[27,28]
a: 4.4 c: 7.6 9.1
[29,30]
a:8.9 c: 11.1 29.1
[31,32]
28.9
J. Raman Spectrosc. 2013, 44, 147–154
Imaging of internal stress around an inclusion in a sapphire
evaluation of P.[36] The R2-line is especially useful because it is
insensitive to differential stress.[37–39] Pressure dependence of
Raman shifts was also quantified in diamond anvil cells at
quasi-hydrostatic conditions (Table 2[40]). However, full compo-
nents of stress tensor have not been determined.
Recently, Kimachi et al.[41] conducted bending tests of a syn-
thetic sapphire crystal at atmospheric pressure and determined
the relationship between Raman shifts and strain tensor, eij
(i, j = x, y, z), on the basis of the deformation potential theory
(see the Appendix). Given the elastic stiffness constants,[42] sij
can be derived from the strain tensor. However, Kimachi’s
equation cannot accurately describe the hydrostatic pressure
dependence of the Raman shifts given in Table 2. This fact
suggests that influence of nonhydrostatic stress on Raman shifts
is different from that of hydrostatic pressure; in the former case,
relative distances of the atoms and the bond angles are changed,
whereas only isotropic changes occur in the latter case. Hence, we
applied Kimachi’s method to determine the deviatoric
components of stress tensor. Accordingly, the sapphire crystal
under hydrostatic pressure P was chosen as a reference state of
eij. P in the aforementioned equation was independently
determined by the R2-line of sapphire. Before applying Kimachi’s
equation, the pressure-induced shift of the Raman bands
corresponding to the internal pressure P was calculated with the
coefficients given in Table 2[40] and was subtracted from the
raw data of the Raman shifts.
Micro-Raman and photoluminescence measurements
The micro-Raman measurement system is composed of an
optical microscope (BX60, Olympus Optical Co. Ltd.) equipped
with a motor-driven X–Y–Z sample stage (FC-101G, Sigma Koki
Co. Ltd.; AF/s, Flovel Co. Ltd.), an Ar+ laser oscillator (IMA, Melles
Griot Co. Ltd.), a holographic notch filter (HIPF-514.5-1.0, Kaiser
Table 2. Pressure derivatives of Raman shifts determined in a
high-pressure experiment of sapphire.[40] The values in the paren-
theses show the pressure dependencies estimated from Kimachi’s
equation[41]
o0 378.4 418.5 430.6 576.4 645.2
(Eg) (A1g) (Eg) (Eg) (A1g)
doο
dP =cm1 =GPa 1.34 1.70 2.79 2.76 3.48
(7.77) (2.31) (10.71) (1.49) (4.29)
Figure 2. Raman spectra of sapphire. (a) The Raman spectrum was acquired by the polarization of the laser beam parallel to the c axis of the sapphire
(Fig. 1). (b) The Raman spectrum was acquired by the polarization of the laser beam perpendicular to the c axis.
J. Raman Spectrosc. 2013, 44, 147–154 Copyright © 2012 John Wiley & Sons, Ltd.
Optical Systems, Inc.), and a polychromator (500IS, CHOROMEX
Co. Ltd.) with the focal length of 50 cm. The polychromator is
equipped with a back-illuminated charge-coupled device with
128  1024 pixels (DU-401-BR-DD, Andor Technology Co. Ltd).
Diffractive gratings of 2400 and 1200 grooves/mm were selected
for the Raman and PL measurements, respectively. The entrance
slit width of the spectrometer was fixed at 80 mm. Spectral resolu-
tions per pixel are 0.6 cm1 and 0.04 nm for Raman and PL
measurements, respectively. The laser emission line of 514.5 nm
was used to excite Raman scattering and PL. The laser power at
the focal point was 17 mW for the Raman measurements and
0.28 nW for the PL measurements. The theoretical spot size of
the laser beam, which was focused through an objective lens of
magnification 50 (Olympus MPlan FL N50, N.A. = 0.80) is 2 mm.
The energy axes of the Raman and PL spectra were calibrated
by neon emission lines, as described later.
The 3D PL mapping was carried out to visualize the residual
pressure distribution around the zircon inclusion and identify
the horizontal (X–Y) plane where the maximum pressure is
retained. Mapping of the PL spectra was conducted in a
140  160  110 mm3 region with the grid size of 5  5  22 mm3
by using point-by-point illumination through the sequential
translation of the sample stage. The exposure time for each
measurement point was 3 s, and the total time for acquiring all
spectra was 5 h.
Considering the selection rule of Raman bands, 2D Raman
mapping measurements were carried out in two optical
configurations. First, the c axis of the sapphire was set parallel
to the polarization plane of the laser light to detect the Eg bands
(Fig. 2(a)). Second, to detect the A1g bands, the polarization of
the laser beam was rotated 90
by inserting a half-wavelength
retardation plate (Fig. 2(b)). At each configuration, the sample
stage was moved in the X–Y direction with a step size of 5 mm,
and an area of 140  160 mm2 (Fig. 1) was scanned by the laser
beam. The exposure time was 60 s for each point, and the total
time to scan the entire area was 16 h.
Calibration of the energy axis
The stress-induced Raman shift of crystalline solids is generally
very small; for example, in sapphire, it is approximately
750.6
(Eg) 1.0 cm1 even when the elastic strain of the sapphire exceeds
103.[41] Thus, precise spectroscopic measurements are needed
4.22 to detect the stress-induced shifts of Raman bands. The long-
(10.80) term stability of the spectrometer must be considered, especially
with mapping measurements. Fukura et al.[43] reported that room
149
wileyonlinelibrary.com/journal/jrs
 N. Noguchi et al.

Figure 3. (a) Spectrum of neon emission lines. The wavenumber (Raman shift) of eight emission lines used for calibration of the energy axis is indicated
in the figure. (b) Temporal changes in the wavenumber of the neon emission lines during the mapping analysis. do: the difference between the raw
spectroscopic data and the ideal wavenumbers. (c) Representative do of the emission lines plotted against the wavenumbers at time 0, 8, and 16 h.
The dashed lines show quadric fitting curves.

temperature fluctuations induced the thermal expansion and

contraction of a spectrometer, and as a result, the deviation of
the energy axis occurred in the spectra. For instance, a room
temperature fluctuation with the amplitude of 0.8
C induced a
0.15 cm1 deviation in the Raman shift. To overcome this
problem, a real-time calibration system using a neon lamp was
constructed.[21] In this system, the Raman scattering light from
a sample and the neon emission light were simultaneously
detected at the upper and lower columns of the charge-coupled
device array, respectively. The previous calibration method was
dedicated to the measurement of a single Raman band of
diamond, which is attributable to the F2g mode (1332 cm1).
Odake et al.[21] applied a neon emission line at 918.5 cm1 or
540.1 nm for correcting the deviation of the energy axis.
In the present case of sapphire, subtle peak shifts of the six
Raman bands, observed in the 350–750 cm1 range, should be
determined within the accuracy of 0.1 cm1. Eight intense
emission lines of neon, which appeared in the 250–920 cm1
Figure 4. PL spectrum of R1-line and R2-line attributable to Cr3+.
range (Fig. 3(a)), were chosen as standard wavenumbers. These
emission lines were monitored during the Raman mapping
analysis. Figure 3(b) plots the differences (do) between the raw the wavenumber and the length of time. The deviation of each
and the ideal wavenumbers of the neon emission lines. It shows neon spectrum was expressed as a quadratic function of the
that the change in deviation over time is different for each neon wavenumber. The fitted curves, as shown in Fig. 3(c), were used
150

emission line, suggesting that the deviation depended on both to calibrate the Raman spectra.

wileyonlinelibrary.com/journal/jrs Copyright © 2012 John Wiley & Sons, Ltd. J. Raman Spectrosc. 2013, 44, 147–154
Imaging of internal stress around an inclusion in a sapphire

Results section is 370 mm beneath the surface of the sample. Internal

Estimate of internal pressure
In general, a peak fitting technique can improve the wavenumber
resolution of a peak position within an accuracy of one-tenth of
the resolution per pixel.[15] Using the Igor software package
(WaveMetrix Co. Ltd.), the R1-line and R2-line in the PL spectra
(Fig. 4) were fitted by Lorentzian functions after subtracting the
background by a baseline correction.
The X–Y section of the 3D map of the R2 wavenumbers in which
the highest wavenumber was detected is shown in Fig. 5(a). The
pressures (estimated from the pressure-induced shift of the
R2-line[36]) concentrate around the zircon inclusion. The highest
pressure (470 MPa) was observed at the tip of the zircon inclusion.
The difference in wavelength between the R1-line and
R2-line (Fig. 5(b)) is sensitive to differential stress, i.e. differ-
ence between the maximum and minimum compressional
stresses of sapphire.[39] In contrast to the internal pressure
fields estimated in Fig. 5(a), Fig. 5(b) suggests that the
differential stress is concentrated at the top and bottom of
the zircon inclusion.

Figure 5. (a) Map showing the wavelength of the R2-line (lR2) around the zircon inclusion shown in Fig. 1. Scales for internal pressure and pressure-
induced shift of the Raman band at 418.5 cm1 (A1g) are also shown. The pressure-induced shift was determined from the pressure and the rate of the
shift of the 418.5 cm1 band to pressure, 1.7 cm1/GPa (Table 2). The mapped area is 140  160 mm2. (b) Map showing the difference in wavelength
between the R1-line and R2-line (lR1 – lR2).

Figure 6. Maps of the shifts of the six Raman bands – two A1g modes (a, b) and four Eg modes (c–f) – around the zircon inclusion. The peak position
of the Eg bands cannot be determined because the Raman bands of zircon overlap. Therefore, the values for the Eg field in the inclusion area have no
151

physical meanings.

J. Raman Spectrosc. 2013, 44, 147–154 Copyright © 2012 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jrs

Estimate of deviatoric stress
The peak positions of the Raman bands were determined by
fitting Lorentzian functions in a manner similar to that used in
the analysis of the PL spectra. To avoid a systematic error result-
ing from a nonlinear background curve because of luminescence
from impurities (Fig. 2(a)), each spectrum was divided into six
segments such that each segment included one of the six
Raman bands. Peak fitting was then performed on each segment.
The deviations of the Raman bands from those observed on the
sample surface were mapped on a horizontal section (Fig. 6). The
Raman bands are shifted up to 2 cm1, and the maps of the shifts
show anisotropic patterns and vary with the Raman band. For the
maps of the Eg modes (Fig. 6(c–f)), the pair comprising the 378.4
and 430.6 cm1 modes attributable to the external normal
Figure 7. Relationship between the crystallographic coordinates (a1, a2,
a3, and c) and Cartesian coordinates (x, y, and z) by which the directions
of stress and strain are described.
Figure 8. Deviatoric stress field around the zircon inclusion calculated from the Raman and PL mapping data. The spatial coordinates are shown at the
152
bottom left.
wileyonlinelibrary.com/journal/jrs Copyright © 2012 John Wiley & Sons, Ltd.
N. Noguchi et al.
vibration[44] and the pair comprising the 576.4 and 750.6 cm1
modes attributable to the internal normal vibration[44] show a
similar distribution, whereas the distributions of the positive and
negative signs in each pair were inverted. It is noteworthy that
all maps include the regions of negative shifts. This fact cannot
be explained solely by the pressure effect because the pressure-
induced shifts of these Raman bands are blue shifts.[40] Thus,
contributions of nonhydrostatic stress components induced by
the zircon inclusion should be considered.
The stress field around the zircon inclusion was estimated as
follows. First, the pressure-induced shifts of each Raman band
were calculated from the pressure determined by the R2-line
(Fig. 5(a)) and the calibration data for the pressure dependence
of the Raman shifts given in Table 2. Figure 5(a) shows an
example for the 418.5 cm1 band (A1g). The shifts induced by
the deviatoric stress were then estimated by subtracting the
pressure-induced shifts from the raw shift in the Raman spectra
(Fig. 6). Substituting these values into the left sides of Eqs 4 and
5 (Appendix), sij0 was determined at each measurement point.
Here, the spatial coordinates x and z are set parallel to one of
the a axes and the c axis of the sapphire, respectively (Fig. 7).
The stress field estimated from the spectroscopic data is shown
in Fig. 8.
Discussion
The PL spectra of sapphire revealed increased pressure around
the inclusion surface (Fig. 5 (a)). The principal deviatoric stress
parallel to the z axis is compressive (szz0 < 0) at both ends of
the zircon inclusion (Fig. 8(c)) and that in the x and y directions
is compressive (sxx0 < 0 and syy0 < 0, respectively) on both sides
of the inclusion (Fig. 8(a, b)). These 2D observations and the
J. Raman Spectrosc. 2013, 44, 147–154
Imaging of internal stress around an inclusion in a sapphire

cracking, and anisotropy in thermoelastic constants must be

taken into account in the inclusion model. Results of the present
spectroscopic method would be fully utilized to improved
inclusion models in future.

Acknowledgments
We are grateful to K. Komatsu of the University of Tokyo for
helpful discussions and technical advice for X-ray diffraction
analysis and H. Yoshida of the University of Tokyo for technical
assistance in electron back-scattering analysis. This work was
financially supported by the Grants-in-Aid for Creative Scientific
Research (19S0205) from the Japan Society for Promotion of
Science and the Global COE Programs for Chemistry Innovation
and for Deep Earth Mineralogy. We thank two anonymous
reviewers for valuable comments.

Figure 9. Schematic illustration of the strain field around the zircon Appendix
inclusion inferred from the deviatoric stress field in Fig. 8. The lattice
strain is represented by the deformation of the grid pattern. Piezospectroscopic effects on Raman bands are formulated
by the deformation potential theory.[46,47] The stress-induced
wavenumber shift and splitting of Raman bands due to a
reduction in symmetry are described by the deformation poten-
crystal symmetries indicate that the crystal surfaces of zircon are tial Vij. The effect of stress is introduced as a perturbation to a
compressed in all directions. Generation of compressive internal Hamiltonian. The change in the vibrational energy Δo can be
stress can be primarily explained by the difference in thermal calculated by solving an eigenvalue problem describing the
expansion coefficients and elastic constants between sapphire perturbing potential V, which is defined as
and zircon; in the present case, the thermal expansion coefficient
of sapphire is higher than those of zircon in all directions (Table 1), X
V¼ Vi j ei j i; j ¼ x; y; or z (1)
and the diagonal components of the stiffness tensor of the
ij
former[42] are larger than those of the latter.[45] Thus, the zircon
inclusion expands more rapidly than the sapphire crystal during where eij is the strain tensor. It is considered that the symmetry of
cooling and decompression of the host rock, resulting in extra the sapphire demands that V is invariant under the D3d symmetry
stress at the wall of the inclusion.[33] The deviatoric stress is operation, as expressed by the following equation:
very high (–szz0 ~ 700 MPa), and the hydrostatic pressure compo-  
P 1  
nent is also high (P ~ 470 MPa) at both ends of the zircon crystal V ¼ Vi j ei j ¼ Vxx þ Vyy exx þ eyy þ Vzz ezz
2 
(Figs 5(a) and 8(c)). The anisotropic stress field described here is
1  
possibly due to stress concentration at the tips of the crystal. þ Vxx  Vyy exx  eyy þ 2 Vyz eyz (2)
Whereas the deviatoric stress fields for the axial components 2 
1  
(sxx0 , syy0 , and szz0 ) are symmetric around the long axes of the þ Vxy exy þ 2ðVzx Þðezx Þ
2
inclusion (Fig. 8(a–c)), the shear components (sxy0 , syz0 , and szx0 )
show asymmetric patterns (Fig. 8(d–f)); for example, there exists
local concentrations at the top–right and bottom–left of the In Eq. 2, (Vxx + Vyy) and Vzz belong to A1g, (Vxx – Vyy) and Vyz
zircon inclusion (denoted by dashed lines in Fig. 8(d–f)). belong to Eg (x), and Vxy and Vzx belong to Eg (y). By using the
However, the axial components of deviatoric stress are almost orthogonality theorem, Δo is expressed for the A1g mode as
neutral at these sites. The stress tensor in this region could be
explained by distortion of the crystalline lattice in the diagonal  
1   
direction (Fig. 9). On the other hand, sxy0 , syz0 , and szx0 are close Δo ¼ A1g; f  Vxx þ Vyy jA1g; f i exx þ eyy þ A1g; f  Vzz jA1g; f i ezz (3)
2
to zero at the top–left and bottom–right of the inclusion. It is
possible that the large cracks observed under a microscope
where A1g,f is a vibrational wave function. The aforementioned
(Fig. 1) have relaxed the internal stress around them.
can be rewritten as
The normal stress on the crystal surface of zircon takes the
maximum value around 1.2 GPa (–szz0 + P) at the top and  
ΔoA1g ¼ c0 exx þ eyy þ c1 ezz (4)
bottom, whereas normal compression on the side surfaces of
zircon (–syy0 + P) is 0.3–0.4 GPa (Figs 5(a) and 8(b)). On the other where
hand, the thermoelastic calculation of a spherical zircon inclusion
(Fig. S1) gives internal stresses of 1.4 GPa for the lower crust 1   
origin models,[6] respectively. Relatively smaller values obtained c0 ¼ A1g; f  Vxx þ Vyy jA1g; f i and c1 ¼ A1g; f  Vzz jA1g; f i
2
in the present spectroscopic analysis may be due to the develop-
ment of the radial cracks. For accurate determination of the P–T are deformation potential constants. Similarly, the stress-induced
153

condition of sapphire crystallization, influences of crystal shapes, shifts are written for the Eg modes as

J. Raman Spectrosc. 2013, 44, 147–154 Copyright © 2012 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jrs
 N. Noguchi et al.

Table 3. Deformation potential constants for A1g and Eg modes[41] [3] K. Zaw, F. L. Sutherland, F. Dellapasqua, C. G. Ryan, T. F. Yui, T. P. Mernagh,
D. Duncan, Mineral. Mag. 2006; 70, 669.
Mode o0/cm 1 [4] F. L. Sutherland, A. Abduriyim, J. Gemmol. 2009, 31, 203.
[5] J. F. Guo, S. Y. O’Reilly, W. L. Griffin, Contrib. Mineral. Petrol. 1996, 122,
A1g 418.5 c0 415.84 368.
c1 927.42 [6] F. L. Sutherland, P. W. O. Hoskin, C. M. Fanning, R. R. Coenraads,
645.2 c0 1311.22 Contrib. Mineral. Petrol. 1998, 133, 356.
[7] M. L. Dele, P. Dhamelincourt, J. P. Poirot, J. M. Dereppe, C. Moreaux,
c1 736.46 J. Raman Spectrosc. 1997, 28, 673.
Eg 378.4 d0 2512.97 [8] V. Palanza, D. D. Martino, A. Paleari, G. Spinolo, L. Prosperi, J. Raman
d1 1088.06 Spectrosc. 2008, 39, 1007.
d2 884.41 [9] A. Bernard, Bull. Associat. Francaise de Gemmmol. 1975, 43, 9.
[10] P. Vichit, S. Vudhichativanich, R. Hansawek, J. Geol. Soc. Thailand.
d3 48.16
1978, S3, M4–1.
430.6 d0 3506.21 [11] E. A. Jobbin, J. P. Berrange, J. Gemmol. 1981, 17, 555.
d1 1409.31 [12] M. Gunawardene, S. S. Chawla, J. Gemmol. 1984, 19, 228.
d2 1429.27 [13] T. C. Coldham, Gem. Gemmol. 1985, 21, 130.
d3 653.28 [14] J. L. Rosenfeld, A. B. Chase, Am. J. Sci. 1961, 259, 519.
[15] E. S. Izraeli, J. W. Harris, O. Navon, Earth Planet. Sci. Lett. 1999, 173,
576.4 d0 641.56 351.
d1 80.76 [16] C. D. Parkinson, I. Katayama, Geology 1999, 27, 979.
d2 545.60 [17] K. Ye, J.B. Liou, B. Cong, S. Maruyama, Am. Mineral. 2001, 86,
d3 141.57 1151.
[18] M. Enami, T. Nishiyama, T. Mouri, Am. Mineral. 2007, 92, 1303.
450.6 d0 3210.01
[19] L. Nasdala, J. W. Harris, J. Glinnemann, Am. Mineral. 2005, 90, 745.
d1 2020.55 [20] A. V. Korsakova, D. Hutsebautb, K. Theunissenc, P. Vandenabeeleb,
d2 979.30 A. S. Stepanov, Spectrochim. Acta A 2007, 68, 1046.
d3 583.04 [21] S. Odake, S. Fukura, H. Kagi, Appl. Spectrosc. 2008, 62, 1084.
[22] H. Kagi, S. Odake, S. Fukura, D. Zedgenizov, (2009) Russian Geology
Geophys. 2009, 50, 1183.
[23] G. M. Oakes, L. M. Barron, S. R. Lishmund, Aust. J. Earth Sci. 1996, 43,
289.
[24] F. L. Sutherland, Aust. J. Earth Sci. 1996, 43, 323.
[25] L. W. Finger, R. M. Hazen, J. Appl. Phys. 1978, 49, 5823.
  [26] P. Aldebert, J. P. Tranverse, High Temp. High Press. 1984, 16, 127.
ΔoEg ¼ d0 exx þ eyy þ d1 ezz  Δ (5)
[27] R. M. Hazen, L. W. Finger, J. Appl. Phys. 1979, 64, 196.
[28] E. C. Subbarao, K. V. G. K. Gokhale, J. Japan. Appl. Phys. 1968, 7, 1126.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
2  2ffi [29] R. M. Hazen, L. W. Finger, J. Phys. Chem. Solids 1981, 42, 143.
Δ¼ d2 exx  eyy þ d3 eyz þ 2 d2 exy þ d3 ezx [30] K. Sugiyama, Y. Takeuchi, Z. Kristallogr. 1991, 194, 305.
[31] R. J. Angel, R. M. Hazen, T. C. McCormick, C. T. Prewitt, J. R. Smyth,
Phys. Chem. Mineral. 1988, 15, 313.
where d0, d1, d2, and d3 are deformation potential constants. The [32] M. Tribaudino, R. J. Angel, F. Ca´mara, F. Nestola, D. Pasqual,
reduction in symmetry from D3d to C2 by lattice deformation I. Margiolaki, Contrib. Mineral. Petrol. 2010, 160, 889.
induces the splitting of the Eg band into the A and B bands,[47] [33] Y. Zhang, Earth. Planet. Sci. Lett. 1998, 157, 209.
which correspond to the ‘+’ and ‘–‘ signs in Eq. 5, respectively. [34] M. C. Munisso, W. Zhu, G. Pezzotti, Phys. Status Solidi B 2009, 246,
1893.
The latter was ignored in the experimental calibration by Kimachi [35] F. Momma, F. Izumi, J. Appl. Crystallogr. 2008, 41, 653.
et al.[41] because B band is relatively insensitive to stress. The [36] H. Mao, J. Xu, P. Bell, J. Geophys. Res. 1986, 91, 4763.
deformation potential constants of the A1g and Eg bands [37] B. Okai, O. Shimomura, I. Fujishiro, Physica B 1986, 139 & 140,
determined by Kimachi et al.[41] is listed in Table 3. If the stress- 799.
induced Raman shifts are measured for these six bands, the stress [38] Y. M. Gupta, X. A. Shen, Appl. Phys. Lett. 1991, 58, 583.
[39] M. Chai, J. B. Brown, Geophys. Res. Lett. 1996, 23, 3539.
tensor can be derived from Eqs 4 and 5. [40] J. Xu, E. Huang, J. Lin, L. Y. Xu, Am. Mineral. 1995, 80, 1157.
[41] H. Kimachi, S. Yamamoto, W. Ota, K. Shirakihara, Y. Fujita, J. Soc.
Supporting information Materia. Sci. Japan. 2009, 58, 603.
[42] J. B. Wachtman Jr., W. E. Tefft, D. G. Lam Jr., R. P. Stincchfield, J. Res.
Supporting information may be found in the online version of Nat. Bureau Stand. A 1960, 64A, 213.
[43] S. Fukura, T. Mizukami, S. Odake, H. Kagi, Appl. Spectrosc. 2006, 60,
this article. 946.
[44] S. P. S. Porto, R. S. Krishnan, J. Chem. Phys. 1967, 47, 1009.
References [45] H. Ozkan, J. Appl. Phys. 1976, 47, 4472.
[46] S. Rodriguez, P. Fisher, F. Barra, Phys. Rev. 1972, B5, 2219.
[1] Abduriyim A., Kiefert L., Larson W., A. J. Molina, C. Smith, G. Roskin, [47] V. J. Tekippe, A. K. Ramdas, S. Rodriguez, Phys. Rev. 1973, B8, 706.
Gem. Gemmol. 2006, 42, 74. [48] S. Y. O’Reilly, W. L. Griffin, O. Gaul, AGSO J. Australian Geology
[2] A. Abduriyim, H. Kitawaki, J. Gemmol. 2006, 30, 23. Geophys. 1997, 17, 63.
154

wileyonlinelibrary.com/journal/jrs Copyright © 2012 John Wiley & Sons, Ltd. J. Raman Spectrosc. 2013, 44, 147–154