Isotopic heterogeneity in synthetic and natural silicon carbide

Article abstract of DOI:10.1016/j.jpcs.2008.05.005

The distribution of both carbon and silicon isotopes in synthetic sublimation growth SiC wafers and in natural SiC grains was studied using secondary ion mass-spectrometry (SIMS). Significant variations in both isotopic ratios were observed which were broadly correlated with the crystalline perfection as documented by Raman microspectroscopy. Domains consisting of 15R (or with its admixture) are, on average, enriched in ¹²C isotope relative to 6H domains, and they also show larger scatter in their observed silicon isotope ratios. We ascribe such heterogeneity to fluctuations of Si/C ratio in the growth medium and it is possible to model the spatial extent of such fluctuations. For the natural SiC grains the isotopic data suggest that they grew under relatively stable conditions, although some of them show significant isotopic zoning.

Full text of DOI:10.1016/j.jpcs.2008.05.005

ARTICLE IN PRESS
Journal of Physics and Chemistry of Solids 69 (2008) 2492– 2498
Contents lists available at ScienceDirect
Journal of Physics and Chemistry of Solids
journal homepage: www.elsevier.com/locate/jpcs
Isotopic heterogeneity in synthetic and natural silicon carbide
a,
Ã
b
c
d
A.A. Shiryaev , M. Wiedenbeck , V. Reutsky , V.B. Polyakov ,
e
f
g
N.N. Mel’nik , A.A. Lebedev , R. Yakimova
a
Shubnikov Institute of Crystallography RAS, Leninsky pr. 59, 119333 Moscow, Russia
GeoForschungsZentrum, Telegrafenberg, 14473 Potsdam, Germany
Institute of Geology and Mineralogy, SB RAS, pr. Ak.Koptyuga 3, 630090 Novossibirsk, Russia
Institute of Experimental Mineralogy RAS, Chernogolovka, Russia
Lebedev Physics Institute RAS, Leninsky pr. 53, Moscow, Russia
b
c
d
e
f
Ioffe Physico-Technical Institute RAS, 194021 Sankt Petersburg, Russia
Linkoping University, 58183 Linkoping, Sweden
g
a r t i c l e i n f o
a b s t r a c t
Article history:
The distribution of both carbon and silicon isotopes in synthetic sublimation growth SiC wafers and in
natural SiC grains was studied using secondary ion mass-spectrometry (SIMS). Significant variations in
both isotopic ratios were observed which were broadly correlated with the crystalline perfection as
documented by Raman microspectroscopy. Domains consisting of 15R (or with its admixture) are, on
Received 20 March 2007
Received in revised form
16 April 2008
Accepted 3 May 2008
12
average, enriched in C isotope relative to 6H domains, and they also show larger scatter in their
observed silicon isotope ratios. We ascribe such heterogeneity to fluctuations of Si/C ratio in the growth
medium and it is possible to model the spatial extent of such fluctuations. For the natural SiC grains the
isotopic data suggest that they grew under relatively stable conditions, although some of them show
significant isotopic zoning.
Keywords:
A. Inorganic compounds
B. Crystal growth
C. Raman spectroscopy
&
2008 Elsevier Ltd. All rights reserved.
1
. Introduction
presence of several polytypes in a single SiC wafer is usually
detrimental for technological applications. Pressure and tempera-
ture during crystal growth both play important roles determining
which polytype is produced [2]. It has also been suggested that
impurities and intrinsic defects may influence the stability of
polytypes ([3–5] and references therein). Additional information
contributing to the understanding of polytypes formation is of
considerable applied and basic interest.
Besides its significance in the field of solid state physics,
studies of SiC are valuable for geosciences and astrophysics.
Silicon carbide occurs in nature as the rather rare mineral
moissanite. Numerous finds of moissanite grains in heavy
fractions of kimberlites, high-grade metamorphic and mantle
derived rocks (for a review see [6]) as well as in meteorites and
interstellar dust [7] have already been reported. Since the
formation of SiC requires extremely low oxygen fugacity, some
Silicon carbide is a material of significant interest for various
fundamental and applied fields: from high-power electronics to
crystallography and problems of the Earth mantle oxidation state.
From the scientific points of view one of the most interesting
features of SiC is its ability to form numerous polytypes, i.e.
structural modifications, each of which may be regarded as built
through the stacking layers of (nearly) identical structure and
composition, wherein different crystallographic types differ only
in their stacking sequence [1]. The most common polytypes are
hexagonal 4H and 6H; rhombohedral 15R and the cubic 3C. In the
Ramsdell notation used in this paper to indicate the polytype, the
figure indicates number of non-equivalent layers and the letter
type of lattice packing: cubic, hexagonal or rhombohedral. Though
many physical properties (hardness, thermal conductivity, ther-
mal capacity, etc.) of the polytypes are fairly similar, some impor-
tant differences do exist, e.g., the band gap varies from 2.4 eV in 3C
to 3.23 eV in 2H. Such variations in properties are partly due to
various coordinations of Si and C atoms in polytypes, which
induce differences in the electronic structure. The simultaneous
geologists consider moissanite as
a ‘‘geological aberration’’.
However, a growing body of data confirms the rather widespread
existence of small SiC grains in various geological settings.
Interestingly, carbon isotopic analysis of natural SiC grains always
12
shows that the SiC is enriched in C carbon [6,8]. The genesis of
natural SiC remains a matter for debate [6,8,9].
This papers deals with isotopic fractionation related to growth
of SiC. Several processes responsible for isotopic fractionation are
known [10,11]. Being a thermodynamic property, equilibrium
Ã
Corresponding author. Tel.: +74991351011; fax: +74991355450.
E-mail address: shiryaev@ns.crys.ras.ru (A.A. Shiryaev).
0
022-3697/$ - see front matter & 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jpcs.2008.05.005

ARTICLE IN PRESS
A.A. Shiryaev et al. / Journal of Physics and Chemistry of Solids 69 (2008) 2492–2498
2493
isotopic fractionation can be expressed in terms of vibrational
densities of states (DOS) of substances under consideration via
appropriate equations from statistical physics [12,13]. There are
some indications that impurities and intrinsic defects may also
influence the degree of the isotope fractionation [14]. Polytypes
are interesting for assessing the influence of various lattice
imperfections on isotopic fractionation, since polytypes could be
represented as stacking faults in regular hexagonal or cubic
lattices (e.g., stacking faults in the 3C-SiC are hexagonal
polytypes). Importantly, disorder in SiC is not completely random
and disordered regions consist of mixture of randomly disordered
simple polytype domains and those containing stacking faults
distributed periodically or almost periodically. Only few studies
have addressed the relationship between isotopic composition
and polytypism, one of which documented the dependence of
oxygen isotope composition of biotite on relative fraction of
polytypes in mica [15].
In the current work, we have used secondary ions mass
spectrometry (SIMS) to investigate both carbon and silicon
12
13
28
29
30
isotopic heterogeneity (ratios C/ C and Si/ Si/ Si) in SiC
wafers grown using the sublimation method at very high
temperatures as well as in grains of natural moissanite. The main
task was to establish the relationship between crystalline quality
(polytype and/or disorder) and isotopic composition of SiC.
2
. Experimental
2.1. Samples
In this work samples of natural and synthetic SiC were studied.
Natural SiC was represented by grains handpicked from heavy
fraction of several Yakutian kimberlites. The grains were
0
.3–0.5 mm in size, and represented by flattened hexagons or
fragments thereof. The colors of the natural grains varied from
transparent to dark violet. Raman microspectroscopy showed that
the grains were either hexagonal (6H and 4H) or rhombohedral
Fig. 1. General view of synthetic SiC wafer and example of SIMS measurement
area. Length of the scratch is 0.6 mm.
(
15R) polytypes.
in sample holders using Ta masks. The three other samples, two
pieces from one synthetic wafer and a grain mount consisting of
five individual grains of natural SiC, were cast in epoxy and
polished flat prior to cleaning and gold coating.
Synthetic SiC wafers were grown at Linkoping University using
the sublimation method [16] at a temperature of circa 2400 1C;
temperature gradients over the 2 in wafer diameter did not
exceeded 10–201. During the growth process some instabilities
have occurred, leading to the development of macroscopic (few
Due to the destructive nature of SIMS, questions may arise
concerning analytical artifacts of SiC polytypes that might be
induced by differences in sputter rate (so-called ‘‘matrix effect’’).
However, it has been shown both experimentally and theoretically
that properties of different SiC polytypes are rather similar in
respect of their stability against radiation damage [17]. Therefore,
from the point of view of SIMS as an analytical tool, different SiC
polytypes would be predicted to behave virtually identically.
2
mm ) domains of different polytype compositions and/or differ-
ent concentrations of charge carriers. The boundaries between the
domains were clearly visible due to different coloration (Fig. 1).
Raman microspectroscopy revealed that the synthetic samples
studied in this work are represented by hexagonal (6H) and
rhombohedral (15R) polytypes with domains of disordered 6H SiC
with 15R-like regions. In some samples neighboring domains of
different color consisted of the same polytype (6H). These
variations in optical properties are due to variable concentration
of charge compensating impurities.
133
+
Our SIMS analyses employed a 500 pA, 17.5 kV
Cs primary
beam focused to ꢀ5 mm on the polished sample surface. A
secondary extraction voltage of ꢁ7.5 kV was used to which no
energy offset was applied to a 50 eV energy bandpass. For C
isotopic analyses the instrument was operated at elevated mass
resolving power of M/DME5100, which was adequate to resolve
2
.2. SIMS analytical technique
13
1
12
C from the nearby isobaric interference H C. A 20 mm contrast
To assess the isotopic homogeneity of the natural and synthetic
aperture was used in conjunction with a 750 mm field aperture
(equivalent to a 60 mm field-of-view). A single analysis, consisting
of a 5 min preburn followed by 100 cycles of the peak stepping
SiC samples the Cameca ims 6f SIMS instrument at Geo-
ForschungsZentrum was employed to make mass-spectrometric
analyses along straight lines, intersecting visually observed
boundaries between polytypes. Prior to SIMS analysis the SiC
samples were ultrasonically cleaned in high-purity ethanol, dried
at 60 1C in atmosphere and then sputter coated with a high purity,
12
13
sequence—mass 11.9 (background, 0.1 s per cycle), C (2 s),
C
(10 s)—lasted 26 min. Due to the relatively high mass resolving
power employed, an enhanced magnet control (310,000 bits at the
12
C mass station) was used and this was recalibrated prior to each
3
5 nm thick conductive gold coat. Two different synthetic samples
analysis. As the samples have an intrinsic electrical conductivity
electron flooding was not necessary, thus permitting the electron
were placed in the SIMS instrument as polished slices of SiC held

ARTICLE IN PRESS
2
494
A.A. Shiryaev et al. / Journal of Physics and Chemistry of Solids 69 (2008) 2492–2498
deflecting magnets to be turned off. Under these conditions a
12
typical C count rate started at ꢀ300 kHz and grew to ꢀ400 kHz
by the end of the single analysis. A dead time correction for
discreet dynode ion counting system of 16 ns was applied to all
13
12
data. Ion counting based precision values for
( C/ C)meas
of 70.4% to 70.5% were typical for such measurements,
whereas the observed internal precisions were typically around
70.6% (1s).
Silicon isotopic measurements were also conducted on
selected grains from the grain mount as well as on one of the
synthetic sample. Si analyses were run under conditions similar to
those used for C, the exception being a reduced mass resolving
power of M/DME4900, which is more than sufficient to resolve all
Si species from the isobaric SiH peaks. A single analysis, consisting
of a 5 min preburn followed by 100 cycles of the peak stepping
Fig. 2. Representative Raman spectra of several spots on one of the SiC wafer.
28
29
30
Spectra show variations in crystallographic quality of the different spots.
sequence—mass 27.9 (background, 0.1 s), Si (2 s), Si (8 s), Si
28
(
10 s), lasted for a total of 40 min. Typical Si count rates of
initially 600 kHz and increasing up to 1 MHz were observed
during such analyses. Ion counting-based precision values of
29
28
30
28
7
0.20% were typical for both the Si/ Si and Si/ Si measured
ratios, whereas the observed internal precision were typically
around 70.25% and 70.45%, respectively. This difference
between the Poisson-based and observed internal precisions
resulted from a reproducible trend towards lighter isotopic values
3
0
28
(
ꢀ2% for Si/ Si) seen over the course of single measurements.
Some of the synthetic samples were studied using Cameca ims
1270 at the Ion Probe Facility, Edinburgh. Standardization of these
SIMS analyses was achieved by investigation of the SiC grains,
whose carbon isotopic composition was measured by conven-
tional mass spectrometry. Since SiC is a refractory material and
2
complete oxidation is hindered by rapid build-up of SiO layer,
conventional combustion does not give good results. We have
used chemical decomposition process according to reaction
SiC+4KOH +2O
easily decomposed by phosphoric acid releasing CO
2
¼ 2H
2
O+K
2
CO
3
+K
2
SiO
3
. Obtained carbonate is
for conven-
2
tional isotope ratio mass-spectrometry (IRMS). Using this ap-
proach the mean carbon isotopic composition of synthetic SiC
Fig. 3. Measured carbon isotopic ratios for grains of natural SiC. Error bars
correspond to 1s.
13
wafers (d
CPDB) is around ꢁ21%.
disordered regions contain mixture of 6H and 15R polytypes.
Representative spectra of two polytypes obtained on one of the
synthetic samples are shown in Fig. 2.
2.3. Raman microspectroscopy
The shape, symmetry and position of Raman peaks are
sensitive to polytype variety, disorder, stresses and density of
charge carriers [18–20]. In our synthetic samples the FLO mode
After the SIMS measurements the SiC polytype at each spot
locality was determined using Raman microspectroscopy using
Jobyn Yvon U1000 spectrometer using wavelength of 488 nm with
beam size of ꢀ10 mm in backscattering geometry. The measure-
ments were performed in immediate vicinity of the SIMS spots or
in the crater itself. Two spectral regions were found to be
important for the current study: ‘‘low’’ wavenumbers (acoustical
ꢁ
1
(
situated between 900 and 1000 cm ) is broadened towards the
high wavenumbers side. This indicates small degree of stress due
to extended defects [20], but points to different concentration of
electrically active dopants [19].
ꢁ
1
phonons, FTA mode, 100–250 cm ), and ‘‘high’’ wavenumbers
ꢁ1
(
optical phonons, FTO mode, 740–810 cm ).
3.2. Carbon isotopes
Carbon isotopic ratios in five mounted single grains of natural
3
. Results
SiC (Fig. 3) were analyzed. Grain #4 was the only one consisting of
5R polytype, whereas the other grains consisted of hexagonal
1
3
.1. Raman microspectroscopy
varieties. On every grain two to three SIMS measurements were
performed. The data from four of these grains (grains 1, 2, 3 and 5)
are consistent with a homogenous population. In contrast, the
Raman spectroscopy showed that the natural grains are well
13
12
ordered and that each grain consists of single polytype. Four of the
grains belong to 6H and one to 15R polytype. In contrast to the
natural samples, the selected synthetic wafers are less perfect.
Whereas some analyzed regions of the wafers consist of well-
defined polytypes, we found regions of the wafer containing
disordered SiC using a 10 mm probing laser beam. The disorder
manifests itself in a broadening and distortion of the Raman peaks
as well as the appearance of new features. In many cases
measured C/ C ratio of grain 4 (n ¼ 3) was 3.5% lighter than
that observed for the main population. As the external reprodu-
cibility of our technique is better than 1% (1s), we can conclude
with a high degree of confidence that the C isotopic composition
of grain 4 differs from that of the four other grains that were
analyzed.
Several profiles were measured on synthetic SiC wafers. Each
profile comprised between
9 and 20 points separated by

ARTICLE IN PRESS
A.A. Shiryaev et al. / Journal of Physics and Chemistry of Solids 69 (2008) 2492–2498
2495
Fig. 4. Measured carbon isotopic ratios for SiC wafers. Error bars correspond to 1s.
Fig. 5. Ratio of intensities of various peaks (modes) in Raman spectra versus C and Si isotopic ratios for several studied SiC wafers.
2
0–130 mm and each profile intersected one or more visually
of domains of high-quality 6H SiC differs (usually heavier) than
observed polytype boundaries. Several representative profiles are
shown in Fig. 4. Variations in carbon isotope ratios between
domains of a given polytype (6H and 15R) are rather small: an
absolute difference in the unweighted means of the two popula-
tions does not exceed ꢀ1.5s. In line with results obtained for the
natural SiC grains, the 15R domain appears to have marginally
lighter carbon isotopic composition.
that of disordered regions where a 15R admixture is present. The
relationship between the spectroscopic and isotopic data is
addressed in Discussion section. On several wafers gradual
changes in carbon isotopic composition with increasing distance
from the interpolytype boundary are observed (the most pro-
nounced trends are shown by Fig. 4b, d).
A closer look at the available data shows that the SiC wafers are
isotopically heterogeneous on microscopic scale with consider-
3.3. Silicon isotopes
13
12
able scatter of C/ C values even between points separated by
only a few tens of microns. It is important to mention that scatter
Silicon isotopic analyses were conducted on three individual
natural grains (Fig. 6a). These included analyses of grains 2 and 3
from the dominant population identified previously on the basis
of C isotopes and Raman spectra (6H polytype) and also of grain 4
(15R) which yielded a distinctly light C isotope signature. Both
13
12
of C/ C ratios obtained on different classes of SIMS instruments
Cameca 6f and 1270) are closely comparable. As a proxy of the
(
crystalline quality we have used ratio of intensities of folded TO
modes in Raman spectra. Comparison of Raman spectra of points
with significantly different isotopic ratios shows existence of
qualitative correlation between crystalline quality of the SiC
domain and its isotopic composition (Fig. 5). Isotopic composition
2
9
28
30
28
Si/ Si and Si/ Si ratios of grain 4 were significantly heavier
than those observed for grains 2 and 3 by 1.5% and 3.1%,
respectively. No evidence for non-linear mass bias was found,

ARTICLE IN PRESS
2
496
A.A. Shiryaev et al. / Journal of Physics and Chemistry of Solids 69 (2008) 2492–2498
Fig. 6. Silicon isotope ratios for natural SiC grains with corresponding grain numbers (panel A) and for synthetic wafers (panel B). Hatched region indicates values for 15R
polytype.
though our method would be sensitive to such an effect only at
of polytypes. Since the growth process of natural SiC is presently
unknown, the corresponding part of the discussion relates only to
the synthetic samples.
the ꢀ1% level.
A
single Si isotope profile, consisting of 10 individual
measurements over a 500 mm distance, was conducted on one
synthetic sample. The profile started at 6H domain, passed
through 15R and entered another 6H domain. An absolute
differences between the unweighted means of the points belong-
Sublimation growth of synthetic SiC is a complex process [21]
where several growth-related phenomena are known to play
important roles and which could produce isotopic fractionation:
ꢁ
5
ꢁ5
ing to different polytypes of (2.571.0) ꢂ 10 and (1.271.2) ꢂ 10
(
(
G1) sublimation of the source material;
2
9
28
30
28
(
1s) were found for the Si/ Si and Si/ Si ratios, respectively,
2 2
G2) mass transport of vapor species (mostly Si C, SiC , Si) to
with the average value for the 15R polytype appearing to be
marginally lighter. Therefore, a small difference between the Si
isotopic compositions of these two SiC polytypes may be present,
and this conclusion can be asserted with a higher degree of
the growing crystal;
G3) nucleation and growth.
Finally, the fractionation may occur due to:
S1) differences in structure between polytypes.
(
(
29
28
confidence for the case of Si/ Si. Again, no evidence for non-
linear mass bias was found (Fig. 6b). As in case of carbon isotopes
a qualitative correlation between isotopic ratios and Raman
features is observed. Namely, the 6H polytype has better defined
Si isotopic composition whereas disordered domains with a
detectable 15R admixture are characterized by considerable
scatter (see Discussion below).
Since crystal growth is believed to be diffusion limited [22], the
effect of isotopic fractionation that might occur during the
sublimation process (case G1) is unlikely to be significant due to
isotopic exchange during transport of material from the source to
the seed.
Case G2 is more complex. Equilibrium isotope fractionation
reaching observed magnitude of 1.5% between the vapor species
seems improbable due to the high temperatures that are involved,
though kinetic effects cannot be fully discounted.
4
. Discussion
Variations of crystal growth rate (case G3) may produce
isotopic heterogeneity due to kinetic isotopic effects. However,
since we consider macroscopically small regions of the SiC wafer
on which the surface relief was negligible, the growth rate at the
dimensions of our SIMS analysis is virtually constant.
Our SIMS isotopic study shows that on average, the rhombo-
hedral 15R SiC polytype has lighter carbon than the hexagonal 6H.
This trend is observed for both synthetic SiC wafers and for
natural SiC grains. In the case of natural SiC, four out of five
analyzed grains were of 6H polytype and their carbon was 3.5%
heavier than that determined on the single 15R grain (Fig. 3). The
Differences in structure between polytypes may also produce
some isotopic fractionation (case S1). Variations in the strength of
atomic bonding may lead to isotopic fractionation between phases
[11]. According to [23] the mean charge on Si ions depends on
hexagonality of SiC polytype. However, these differences are not
very significant and potential isotopic effect should be small. The
number of non-equivalent lattice sites is also polytype dependent:
structure of 15R contains five such nodes in comparison with
three in 6H. This fact may open room for intrastructural isotopic
fractionation similar to that observed, for example, in malachite
and phyllosilicates [24]. However, though the second and third
coordination spheres are polytype dependent, the nearest neigh-
bors are identical in all polytypes. Therefore, vibrational proper-
ties of polytypes are fairly similar, implying small magnitude of
the interpolytype isotopic fractionation. The similarity of vibra-
tional density-of-states (VDOS) of SiC polytypes is manifested in
Raman and infra-red spectra [25] and in inelastic X-ray and
neutron scattering [26]. Virtually identical VDOS of all studied
SiC polytypes and preliminary thermodynamic calculations
1
5R grain is also distinct from the 6H grains having heavier silicon
isotopic composition (Fig. 6a). Disordered regions of the SiC
wafers are characterized by the isotope ratios being noticeably
different from spots with well-defined polytype. In the analyzed
synthetic sample the most crystallographically ordered spot has
30
28
the lowest Si/ Si.
Several processes could be responsible for the observed
isotopic heterogeneity of the SiC samples. It is known that due
to quantum nature of isotopic effects their magnitude decreases
with temperature. The magnitude of isotopic effect is proportional
to the ratio of isotope masses in case of kinetic origin, and
approaches zero for equilibrium fractionation. Perhaps the most
remarkable result of the current study is the observation of
isotopic heterogeneities in crystals grown at extremely high
temperature (ꢀ2400 1C). Tentative explanation of the observed
heterogeneities follows below and we will consider several
principal stages in the SiC growth process and relevant properties

ARTICLE IN PRESS
A.A. Shiryaev et al. / Journal of Physics and Chemistry of Solids 69 (2008) 2492–2498
2497
(
Polyakov, unpublished) indicates that at the relevant tempera-
Raman microspectroscopy shows existence of a broad correlation
between perfection of crystalline lattice and its isotopic composi-
tion. Namely, domains consisting of pure 6H polytype are, on
average, enriched in C in comparison with domains with an
admixture of 15R polytype. The 6H domains also show less
variability in Si isotope ratios.
We believe that the observed heterogeneity is explained by a
kinetic effect, which is closely related to local variations in Si/C
ratio in incoming gas which gave rise to variations different
polytypes and also produce isotopic heterogeneities. We have
found that gradual changes of isotopic ratios span distances
reaching 200–400 mm from the interpolytype boundary. The
extent of such non-uniformities reflects the size of the growth
instabilities mentioned above.
tures equilibrium structure-dependent isotopic fractionation
between them should have magnitudes significantly smaller than
that observed in our experiments.
13
Here it is appropriate to recall the main reasons for polytype
transformation: Each polytype has certain field of stability on P–T
diagram [2]. In our case P–T conditions were fairly constant and
should have played only a minor role if any. Simultaneous
development of several polytypes could occur due to variations
in Si/C ratio in the growth system and due to inhomogeneous
impurity incorporation [3,4]. Si/C ratio in the growth medium is
important for at least two main reasons: (a) the composition
of polytypes shows slight variations in Si/C ratio [27]; and
(b) changes in the concentration of carbon or silicon vacancies
influences SiC hexagonality [5,28]. For example, an increase in
carbon vacancies would promote the formation of atomic bonds in
cubic nodes. Impurities may influence structure of the valence
bands and/or influence bond strengths and lengths, thus changing
the relative energies between possible stacking sequences.
The arguments given above suggest that the only process
which might have produced the observed relationship between
isotopic composition and crystalline quality involves small
variations in the isotopic composition of incoming vapor species
Investigation of natural SiC grains suggests that growth process
of silicon carbide in nature occur under relatively stable condi-
tions. Several grains show isotopic heterogeneity, which likely
reflects growth zonation. In line with results obtained for
synthetic SiC wafers, natural grains belonging to 15R polytype
were found to differ from 6H grains.
In some sense rare isotopes (e.g., C) may be considered as
isomorphic impurities. This study shows that investigation of
heterogeneities of isotope composition may complement studies
of impurity distribution in reconstruction of growth processes.
Use of isotope ratios looks especially promising for investigation
of high-purity materials with low concentration of extrinsic
defects.
13
(kinetic effect in case G2). The presence of domains containing
from multiple polytypes clearly points to some local instabilities
in vapor flow to the growing crystal. It is highly probable that such
fluctuations were accompanied by changes in the Si/C ratio and
some concomitant isotopic fractionation. Therefore, we believe
that the process leading to formation of disordered SiC lattice has
also induced heterogeneities in the isotopic composition of the
sample. It is important to mention that very high growth
temperatures failed to suppress isotopic heterogeneity.
Acknowledgments
Several of the studied natural SiC grains also show isotopic
heterogeneity. This heterogeneity might be due to growth
zonation, since Raman spectra of the grains show that their
polytypic composition is rather uniform. This suggestion requires
additional study, but it is plausible since different growth zones
could have different isotopic composition of parent material.
Intergrain variations of carbon isotopic composition of natural SiC
have already been reported [6,8].
We thank E.S. Efimova for provision of natural SiC grains and
Professors S. Nakashima and L. Falkovskii for insightful discus-
sions of Raman measurements. A.A.S. is grateful to Alexander von
Humboldt foundation and to Foundation of support of Russian
science for financial support. We thank Dr. John Craven for tuning
and maintenance of the Cameca ims 1270 at University of
Edinburgh.
Whereas we believe that the principal reason for the observed
heterogeneity is growth related, the role of lattice defects cannot
be fully discounted. As discussed above we have observed that in
several cases the 15R polytype appears to be isotopically lighter
than the 6H. This difference may be a consequence of somewhat
different P–T fields of stability of these two polytypes, but might
just as well reflect variations in concentration of intrinsic and
extrinsic defects, stabilizing respective structures.
References
[1] A.R. Verma, P. Krishna, Polymorphism and Polytypism in Crystals, Wiley,
New York, 1966.
[
[
2] W.F. Knippenberg, Philips Res. Rep. 18 (1963) 161.
3] Y.M. Tairov, V.F. Tsvetkov, in: P. Krishna (Ed.), Progress in Crystal Growth and
Characterization, Pergamon Press Oxford, New York, 1982, pp. 111–162.
[4] N.W. Jepps, T.F. Page, in: P. Krishna (Ed.), Progress in Crystal Growth and
Characterization, Pergamon Press Oxford, New York, 1982, pp. 259–308.
[5] A.A. Lebedev, Fiz. Tekh. Poluprovodn. 33 (1999) 769 (English translation:
Observation of gradual changes in carbon isotopic ratios in the
vicinity of interpolytype boundaries (Fig. 4b, d) makes it possible
to estimate the size of the wafer domain affected by mentioned
instabilities. From these graphs one can see that growth
instabilities leading to appearance of another polytype affected
regions spanning 200–400 mm from the boundary.
In line with our current results, we suggest that the observed
correlation between polytypes and oxygen isotopic composition of
natural biotite [15] likely reflects some variations in growth
conditions.
Semiconductors 33 (1999) 107).
[6] V.K. Marshintsev, Mineral. Zhurnal 12 (1990) 17 (in Russian).
7] T. Bernatowicz, G. Fraundorf, M. Tang, E. Anders, B. Wopenka, E. Zinner,
P. Fraundorf, Nature 330 (1987) 728.
8] E.A. Mathez, R.A. Fogel, I.D. Hutcheon, V.K. Marshintsev, Geochim. Cosmo-
chim. Acta 59 (1995) 781.
9] S. DiPierro, E. Gnos, B.H. Grobety, T. Armbruster, S.M. Bernasconi, P. Ulmer,
Am. Mineralogist 88 (2003) 1817.
10] E.M. Galimov, Stable Isotopes of Carbon in Oil and Gas Geology, Nedra Press,
1973 (English translation: Carbon Isotopes in Oil-gas Geology, NASA, 1975).
[11] J. Hoefs, Stable Isotope Geochemistry, Springer, Berlin, 2004.
[
[
[
[
[
[
[
[
[
[
[
12] H.C. Urey, J. Chem. Soc. (1947) 562.
13] J. Bigeleisen, M.G. Mayer, J. Chem. Phys. 15 (1947) 261.
14] V.B. Polyakov, S.D. Mineev, Geochim. Cosmochim. Acta 64 (2000) 849.
15] A. Wilamowski, Chem. Geol. 182 (2002) 529.
16] Y.M. Tairov, V.F. Tsvetkov, J. Cryst. Growth 43 (1978) 209.
17] S.J. Zinkle, L.L. Snead, Nucl. Instrum. Methods B116 (1996) 92.
18] S. Nakashima, H. Ohta, M. Hangyo, B. Palosz, Philos. Mag. B70 (1994) 971.
5
. Conclusions
This work reports a detailed investigation of carbon and silicon
[19] S. Nakashima, H. Harita, Phys. Stat. Sol. A162 (1997) 39.
[20] L.A. Falkovsky, Usp. Fizicheskih Nauk 174 (2004) 159 (English translation:
Physics-Uspekhi 174 (2004) 249).
isotope distributions in the vicinity of interpolytype boundaries
in sublimation growth SiC wafers. Significant fluctuations of
isotopic ratios (up to 3%) are observed at the 20–50 micron scale.
[
21] R. Yakimova, M. Syv a¨ j a¨ rvi, T. Iakimov, H. Jacobsson, A. Kakanakova-Georgieva,
P. Raback, E. Janz e´ n, Appl. Surf. Sci. 184 (2001) 27.

ARTICLE IN PRESS
2
498
A.A. Shiryaev et al. / Journal of Physics and Chemistry of Solids 69 (2008) 2492–2498
[
[
[
[
22] R. Yakimova, M. Syv a¨ j a¨ rvi, T. Iakimov, H. Jacobsson, P. Raback, A. Vehanen, E.
Janz e´ n, J. Cryst. Growth 217 (2000) 255.
23] N.D. Sorokin, Yu.M. Tairov, V.F. Tsvetkov, M.A. Chernov, Dokl. AN SSSR 267
[26] S. Bai, Y. Ke, Y. Shishkin, O. Shigiltschoff, R.P. Devaty, W.L. Choyke, D. Strauch,
B. Stojetz, B. Dorner, D. Hobgood, J. Serrano, M. Cardona, H. Nagasawa,
T. Kimoto, L.M. Porter, MRS Proc. 742 (2002) K3.1.
[27] N.D. Sorokin, Yu.M. Tairov, V.F. Tsvetkov, M.A. Chernov, Kristallografiya 28
(1983) 910 (in Russian).
(1982) 1380 (English translation: Sov. Phys. Dokl. 27(2) (1982) 170).
24] V.I. Ustinov, A.A. Ul’yanov, Geokhimiya 8 (1996) 792 (English translation:
Geochem. Int. 34 (1996) 714).
25] D.W. Feldman, J.H. Parker Jr., W.L. Choyke, L. Patrick, Phys. Rev. 173 (1968) 787.
[28] Yu.A. Vodakov, G.A. Lomakina, E.N. Mokhov, Fizika Tverdogo Tela 24 (1982)
1377 (in Russian).