Full text of DOI:10.1557/JMR.2002.0391

Synchrotron radiography and x-ray topography studies of
hexagonal habitus SiC bulk crystals
T.S. Argunova
Ioffe Physical Technical Institute, Russian Academy of Sciences, St. Petersburg, Russia, and
Department of Materials Science and Engineering, Pohang University of Science and Technology,
Pohang, Korea
M.Yu. Gutkin
Institute of Problems of Mechanical Engineering, Russian Academy of Sciences, St.
Petersburg, Russia
a)
Jung Ho Je and H.S. Kang
Department of Materials Science and Engineering, Pohang University of Science and Technology,
Pohang, Korea
Y. Hwu and W-L. Tsai
Institute of Physics, Academia Sinica, Nankang, Taipei, Taiwan, Republic of China
G. Margaritondo
Institute de physique appliqu e´ e, Ecole Polytechnique F e´ d e´ rale de Lausanne, CH-1015
Lausanne, Switzerland
(Received 16 March 2002; accepted 30 July 2002)
Phase-sensitive synchrotron radiation (SR) radiography was combined with x-ray
diffraction topography to study structural defects of SiC crystals. The particular bulk
SiC crystals examined had a low micropipe density and a hexagonal habitus composed
of prismatic, pyramidal, and basal faces well developed. X-ray diffraction topography
images of the sliced (0001) wafers, which were formed due to the complex lattice
distortions associated with defective boundaries, demonstrated the existence of
two-dimensional defective boundaries in the radial direction, normal to the (0001)
¯
planes. In particular, those parallel to the 〈1120〉 directions extended rather far from the
seed. On the other hand, by phase-sensitive SR radiography the effect of micropipe
collection was detected. Micropipes grouped mostly in the vicinities of the defective
boundaries but rarely appeared between groups. Some general remarks about possible
reasons for the development of such peculiar defect structures were made.
I. INTRODUCTION
hollow core dislocations, or micropipes (MPs), and
1
,2
the formation of other polytype inclusions. Low diver-
gence of the synchrotron beam at the Stony Brook Syn-
chrotron Topography Facility made it possible to
establish an accurate method to simulate the pure orien-
Silicon carbide (SiC) attracts persistent research inter-
est due to its unique electronic, thermal, mechanical, and
other properties. The problem of manufacture of large-
size single crystals free from structural defects, however,
remains unsolved. The solution requires detailed inves-
tigations of the SiC structural quality, which is sensitive
to growth conditions. For nondestructive purposes, these
are typically done by x-ray methods. X-ray imaging tech-
niques allow the visualization of lattice defects within a
crystal interior. Synchrotron radiation (SR) x-ray imag-
ing has advantages caused by very high intensities and
good collimations of synchrotron beams. SR diffraction
topography has already allowed to show, in SiC wafers
of relatively high structural quality, the generation of
3
tation contrast from super screw dislocations. High
flux and high energy of the x-ray beams available at
the European Synchrotron Radiation Facility (ESRF),
together with the large beam size, permitted studying
2
the whole volume of the bulky SiC ingots. Moreover,
special properties of the third-generation synchrotron
sources, like the ESRF, opened the way to novel imaging
4
,5
techniques like phase contrast radiography.
Being
insensitive to strong deformations specific to sublimated
grown SiC crystals of large area, this method allowed
one to image MPs in samples of any crystalline perfec-
6
,7
tion. Besides the mapping and the determination of
the size of the micropipes, phase-sensitive radiography
provided information about their shape and spatial
a)
Address all correspondence to this author.
e-mail: jhje@postech.ac.kr
J. Mater. Res., Vol. 17, No. 10, Oct 2002
© 2002 Materials Research Society
2705
http://journals.cambridge.org
Downloaded: 18 Mar 2015
IP address: 169.230.243.252

T.S. Argunova et al.: Synchrotron radiography and x-ray topography studies of hexagonal habitus SiC bulk crystals
distribution. Principal differences in image formation
mechanisms make x-ray topography and phase-sensitive
radiography complementary techniques, and investiga-
tions of SiC structural quality will benefit by the use of
both methods.
X-ray diffraction topography investigation was done
on a commercial source in one crystal set up with the
Lang method in Bragg and Laue geometries by using Cu
K and Mo K radiations. In the plane of scattering,
␣
␣
angular divergence of the beam at the specimen was of
In this work we present such an approach to study SiC
bulk crystals that were grown by the sublimation sand-
wich method described in Ref. 8. A peculiar feature in
the studied crystals was a well-defined hexagonal habitus
composed of prismatic, pyramidal, and basal faces. The
other specific attribute was a very low micropipe density
observed over a large area located at a boule periphery.
We report the observation of SiC defects using both
strain-sensitive (topography) and density gradient sensi-
tive (radiography) x-ray imaging techniques and discuss
their nature and peculiarities of distribution in the crystal
interior.
the order of angular separation between ␣ and ␣ x-ray
1
2
lines. When the specimen to film distance was kept
5–10 mm, a few micrometers spatial resolution normal to
the scattering plane could be achieved.
B. Samples under study
Bulk SiC crystals were grown by a sublimation sand-
8
wich method in a tantalum container. Features of the
boules that attracted investigation are evident in the op-
tical photographs presented in Figs. 1(a) and 1(b). In pro-
jection on basal plane the crystals were hexagonal in
II. EXPERIMENTAL
A. Instrumental
Phase-sensitive SR radiography experiments were per-
formed at the Pohang Light Source (PLS Pohang, Ko-
rea). We used the 5C1 beamline, whose 1.32-T bending
magnet gives an effective source size (measured at
5.5 keV operated at 2.5 GeV) of the order of 60 ␮m in
the vertical direction and 160 ␮m horizontally, at a dis-
tance of 30 m from the studied sample. The experiments
were performed in the edge detection regime with a
sample-to-detector distance in the range from few centi-
meters to 1 m. A model suggested by Margaritondo and
9
10
Tromba and verified by Hwu et al. showed that, in this
regime, image sharpening could be achieved under less
stringent conditions than those used in pioneering SR
4
,5
phase radiography experiments. Namely, on one hand,
the longitudinal coherence length is not a rigid require-
ment and, without too much loss of resolution, a poly-
chromatic beam may be utilized. On the other hand, the
requirements concerning the lateral coherence length are
rather forgiving and can be fulfilled for SR beam param-
eters considerably less extreme than those achievable at
“
long” beamlines. We used unmonochromatized
“white”) light with no optical elements except beryllium
windows.
The x-ray images were detected by a slow scan
(
charge-coupled device (CCD) camera after converting
the x-rays to visible light with a thin CdWO scintillator
4
crystal. Different microscope objectives were used to
2
change the field of view between 4 × 3 mm and 0.6 ×
2
0
.4 mm . The 12-bit CCD camera had a 1392 × 1040
pixels matrix size. Resolution of a few micrometers was
routinely obtained. Each sequence of measurements
was accomplished by a background subtraction. All im-
ages displayed here are background-subtracted with no
further processing.
FIG. 1. Photographs of the top (a) and side (b) views of SiC boule of
hexagonal habitus.
2
706
J. Mater. Res., Vol. 17, No. 10, Oct 2002
http://journals.cambridge.org
Downloaded: 18 Mar 2015
IP address: 169.230.243.252

T.S. Argunova et al.: Synchrotron radiography and x-ray topography studies of hexagonal habitus SiC bulk crystals
shape with obtuse corners. The periphery of the boules
exhibited remarkable prismatic faces crossed by sections
of pyramidal faces, as illustrated in Fig. 2. A central
part of the boule which was close to seed in the horizon-
tal direction showed the lack of hexagonality (l in Fig. 2).
A circular site (O in Fig. 2) parallel to the basal plane was
placed just above the seed. Crystals grown by this
method can be as large as 40 mm in diameter with the
height up to approximately 20 mm. Note that the sample
shown in Fig. 2 has an approximately 20-mm diameter.
A boule near surface area (l and O in Fig. 2) was studied
by x-ray diffraction topography in reflection geometry.
For a further x-ray investigation the boules were sliced
wafer periphery in radial directions are visible, although
their origin is not clear from the topographs alone. By
correlating the positions of these lines in the wafers
sliced at different distances from the beginning of
growth, we suggested that they were basal plane traces
¯
into wafers of a (0001) basal plane as well as (1010) and
¯
1120) prism planes.
(
III. RESULTS
A. X-ray diffraction topography
A typical example of an x-ray diffraction topograph
of the (0001)-oriented wafer is shown in Fig. 3. Lines of
bright contrast (high local intensity) extending out to the
FIG. 3. (a) X-ray diffraction topograph of major area of (0001) SiC
¯
FIG. 2. Scheme of SiC crystal boule of a hexagonal habitus composed
by the faces: (0001) (O); {1010} (m); {1011} (R). l is a part of the
wafer: Lang method in Braff geometry, 101.10, Cu K , Bragg angle
␣
¯
¯
35.8° (5×). (b) Higher magnification image: Lang method in Bragg
geometry, 101.10, Cu K (12×). The contrast is inversed to the original
¯
boule close to seed in the horizontal direction. Boules were sliced
␣
¯
¯
parallel to the planes (0001), (1010), and (1120) (n).
image.
J. Mater. Res., Vol. 17, No. 10, Oct 2002
2707
http://journals.cambridge.org
Downloaded: 18 Mar 2015
IP address: 169.230.243.252

T.S. Argunova et al.: Synchrotron radiography and x-ray topography studies of hexagonal habitus SiC bulk crystals
of two-dimensional (2D) defects, or defective boundaries.
These traces were produced by diffraction contrast in
both basal plane and prism plane reflections due to lattice
distortions associated with the defective boundaries.
The spotlike contrasts, as can be seen from the reflec-
tion topograph in Fig. 3, are another type of defect, which
can be correlated with hollow-core screw dislocations or
micropipe defects. Their distribution primarily concen-
trated in the central, close-to-seed area. A large area over
the wafer periphery, on the other hand, was almost free
from such diffraction spots [Fig. 3(b)]. With scanning
electron microscopy etch pits associated with hollow-
core screw dislocations in molten KOH treated wafers
(data not shown) were observed in a very similar distri-
bution. The dislocation density was found high in the
central area but decreased with distance from the center.
The density was also high close to the traces
¯
In particular, the traces parallel to the 〈1120〉 direc-
tions, as magnified in Fig. 3(b), were visible continu-
ously through several basal plane slices, indicating that
the corresponding defective boundaries extended rather
far from the seed. On the other hand, those defects rep-
resented by curved lines in the topograph were mostly
lost across the depth of one wafer. Comparing x-ray to-
pographs obtained in the reflection and transmission ge-
ometries, such as shown in Fig. 4(a) and 4(b), from the
part of the sample displayed in Fig. 3, reveals that these
two types of defects indeed differ from each other. The
marked traces [Fig. 4(b)] clearly visible in the transmis-
sion topograph while invisible in the reflection mode
imply that the related boundaries are local.
¯
of the 〈1120〉 defective boundaries with little in between.
The resemblance therefore strongly suggested that spot-
like contrasts we observed in the topograph were those
micropipe defects.
The central part of the image [Fig. 3(a)] shows no
contrast due to other polytype inclusions disoriented rela-
tive to the main lattice.
B. Phase-sensitive SR radiography
Direct evidence for the distribution of tubular-shaped
defects in crystal interiors was obtained by SR radiogra-
phy. By this technique the wafers sliced at different dis-
¯
¯
tances from the seed parallel to (1120) and (1010) crystal
planes were examined. Figure 5 shows the map of de-
¯
fects in the (1120)-oriented wafer cut, i.e., parallel to the
micropipes and the direction normal to the basal plane
traces of the far-reaching 2D defective boundaries, from
FIG. 4. (a) Reflection and (b) transmission topographs of the sample
¯
shown in Fig. 3: (a) Lang method in Bragg geometry, 101.10, Cu K
␣
¯
¯
(
6×); (b) Lang technique in Laue geometry, 1120, Mo K , Bragg angle
FIG. 5. SR microradiographs of (1120)-oriented SiC wafer. The inset
␣
␪
ס 13.3° (6×).
shows a high-magnification image of bundled micropipes.
2
708
J. Mater. Res., Vol. 17, No. 10, Oct 2002
http://journals.cambridge.org
Downloaded: 18 Mar 2015
IP address: 169.230.243.252

T.S. Argunova et al.: Synchrotron radiography and x-ray topography studies of hexagonal habitus SiC bulk crystals
a central part of the boule. The most striking features are
the defects vertically aligned in the central part of the
sample (it should be noted that the radiograph presents
only half of the wafer whose central region corresponds
to the left-hand section of the image). As demonstrated in
the left three figures that are the representative images
magnified from the white square on top, this feature
shown in the phase-sensitive radiograph resulted from
closely spaced or interweaving tubular-shaped defects of
approximately 10 ␮m in diameter. We call such bundles
of micropipes here “superpipes.” Increase or decrease in
the image intensity of this feature was attributed to the
depth variation of their sites under the wafer surface.
The existence of the superpipes suggested that micro-
their coalescing. Meanwhile those with Burgers vectors
of the same signs would interact in a more complicated
manner. Nevertheless, one can still expect some equilib-
rium distance exists where the repelling force due to the
long-range interaction is equal to the attractive force due
to the short-range one. This distance is rather small and
of the order of an average radius of interacting micro-
pipes. For smaller distance, the short-range attracting
force prevails and the micropipes coalesce. It is also pos-
sible that the long-range repelling force is suppressed by
an “external” (with respect to this micropipe pair) shear
stress due to residual thermal strains or other defects
(micropipes, inclusions, internal boundaries, etc.). Gath-
ering of a large amount of micropipes near the center of
the boule and the 2D boundaries, which was attributed to
higher defect density, resulted in the wafer regions be-
tween the group almost free from micropipes, as illus-
trated in Figs. 3 and 5. Meanwhile, in the wafers sliced
far from the seed, micropipes were scarce. Figure 6
shows the example of the microradiograph demonstrat-
1
1
pipes bundled into groups and interacted near the cen-
ter of the boule. Such micropipe bundling effect will be
further discussed in Sec. IV.
The lower part of the image (Fig. 5) provides the view
of macrodefects distribution over the close-to-seed area.
¯
ing the defects in a (1010)-oriented sample cut far from
IV. DISCUSSION
the central part of the boule. In this area micropipes were
practically absent. The macrodefects seen on top
were due to heavily damaged crystals containing frac-
tions of material that were recrystallized by a spontane-
ous growth in the cell between the holder and the back
side of the seed—the process scrupulously described
in Ref. 8.
Diffraction topographs and phase-sensitive radio-
¯
¯
graphs of the defects in (0001)-, (1010)-, and (1120)-
oriented slices provide complementary information on
the structure and possible nature of these defects. If the
¯
D defective boundaries that were parallel to the [1120]
2
direction and extended rather far from the seed [Fig. 3(a)]
were perpendicular to the basal plane, the projection of
¯
the defective boundaries onto the (1120) plane would
have a coincident edge at the hexagonal [0001] direction.
That was the direction in which micropipes gathered
in groups.
Due to the known distortion associated with the de-
fective boundaries, one would expect that the driving
force for the collection of micropipes is the accommo-
dation of the lattice shear. Such transformation of
micropipes to superpipes could therefore occur via coa-
lescence, which, under these conditions, could be ener-
getically favorable.
A micropipe can be considered as not only a super-
screw dislocation having a Burgers vector but also as a
1
1
tube having a free surface. When two pipes interact,
they always attract each other, on the one hand, due to
micropipe free surfaces, regardless of Burgers vector
signs. This force varies with distance as approximately
3
11
1
/r . On the other hand, the dislocation components are
repelling or attractive depending on their Burgers vector
signs, repelling (attractive) for the same (opposite) signs.
1
2
The force varies with distance as approximately 1/r. In
3
dislocation theory, the first (approximately 1/r ) type is
called a short-range interaction, while the second (ap-
1
2
proximately 1/r) type is a long-range interaction. Mi-
cropipes with Burgers vectors of opposite signs attract
each other due to both types of interaction, resulting in
¯
FIG. 6. SR microradiograph of (1010)-oriented SiC wafer.
J. Mater. Res., Vol. 17, No. 10, Oct 2002
2709
http://journals.cambridge.org
Downloaded: 18 Mar 2015
IP address: 169.230.243.252

T.S. Argunova et al.: Synchrotron radiography and x-ray topography studies of hexagonal habitus SiC bulk crystals
inside the crystal body. A scheme illustrating the possible
process of the formation of internal boundaries between
close-to-seed crystallites is shown in Fig. 7. Thick solid
lines trace the defective boundaries emerging at the joints
between individuals. Interfacial regions between the
merging crystallites were certainly imperfect, and crys-
talline defects were then generated to relax lattice strain.
Specifically, the shear component could be accommo-
dated by superscrew dislocations, or micropipes. The un-
derlying mechanism of the bundling effect of the
micropipes and their interaction with the planar defective
boundaries however remains uncertain.
The investigation of the region close to the outer sur-
face of the boule (l in Fig. 2) by diffraction topography in
reflection geometry did not reveal the defective bound-
aries. It is likely that the joints were buried inside the
crystal.
The importance of these problems is not confined by
the scope of the present study. Sublimation-grown SiC
crystals can contain internal boundaries of different
types: between other polytype inclusions; between a re-
crystallized segment and a matrix; between a matrix and
a void. The explanation of the evolution of micropipes in
the lattice disturbed by these defects would be an impor-
tant issue.
In summary, the combination of diffraction topogra-
phy and phase-sensitive radiography has revealed that
the micropipe defects are concentrated near internal
boundaries buried inside the crystal interior, diminishing
FIG. 7. Scheme illustrating a possible formation of internal bound-
aries (solid lines) between crystallites resulting from a multiple nu-
cleation on the seed.
their average density in SiC wafers. This information,
difficult to obtain with other techniques, is likely to pro-
vide a useful clue to the defect formation and the growth
V. CONCLUSIONS
Our x-ray imaging investigation by the combination of
topography and phase-sensitive radiography techniques
revealed that the distribution of structural defects in the
SiC crystals studied here had the following special fea-
of SiC large crystals.
ACKNOWLEDGMENTS
The authors are greatly indebted to Prof. Yu.A. Vodakov,
Dr. E.N. Mohkov, and Dr. A.D. Roenkov, from the Ioffe
Physical-Technical Institute of the Russian Academy of
Sciences (St. Petersburg, Russia), for making available
the SiC single-crystal samples. J.H.J. acknowledges the
support by the BK21 project and by the Korea Institute
of Science and Technology Evaluation and Planning
¯
tures: the 2D boundaries parallel to the directions 〈1120〉
were associated with strong lattice distortions inside
crystal interior and served as traps for micropipes. A fine
structure of such regions still remains unclear. Not know-
ing defect arrangements, one can hardly hope to find
mechanisms for their formation. Thus, in this section we
present some general considerations calling for further
investigation.
(
(
KISTEP) through the National Research Laboratory
NRL) project. T.S.A. acknowledges the support by the
Specifically, we found the existence of the planar de-
fective boundaries collecting the micropipes. We were
also able to map their distribution and orientation. Armed
with this information, it would be much easier to propose
a suitable model for the growth and therefore understand
the origin of the defects.
One could suppose that the growth of a hexagonal SiC
crystal on a cylindrical seed started from multiple nu-
cleation. With time, the nuclei appeared to be enclosed
by a set of crystallites. The crystallites would finally
come in contact and merge together leaving boundaries
KISTEP through the Korea–Russia joint R&D program.
REFERENCES
1. M. Dudley, X.R. Huang, W. Huang, A. Powell, S. Wang,
P. Neudeck, and M. Skowronski, Appl. Phys. Lett. 75, 784 (1999).
2
. E. Pernot, P. Pernot-Rejmankova, M. Anikin, B. Pelissier,
C. Moulii, and R. Madar, J. Phys. D: Appl. Phys. 34, A136 (2001).
3. M. Dudley, X.R. Huang, and W. Huang, J. Phys. D: Appl. Phys.
32, A139 (1999).
4. A. Snigirev, I. Snigireva, V. Kohn, S. Kuznetsov, and I. Schelokov,
Rev. Sci. Instrum. 66, 5486 (1995).
2
710
J. Mater. Res., Vol. 17, No. 10, Oct 2002
http://journals.cambridge.org
Downloaded: 18 Mar 2015
IP address: 169.230.243.252

T.S. Argunova et al.: Synchrotron radiography and x-ray topography studies of hexagonal habitus SiC bulk crystals
5
. P. Cloetens, R. Barrett, J. Baruchel, J-P. Guigay, and M. Schlenker,
J. Phys. D: Appl. Phys. 29, 133 (1996).
9. G. Margaritondo and G. Tromba, J. Appl. Phys. 85, 3406 (1999).
10. Y. Hwu, H.H. Hsieh, M.J. Lu, W.L. Tsai, H.M. Lin, W.C. Goh,
B. Lai, J.H. Je, C.K. Kim, D.Y. Noh, H.S. Youn, G. Tromba, and
G. Margaritondo, J. Appl. Phys. 86, 4613 (1999).
11. M.Yu. Gutkin and A.G. Sheinerman, Phys. Status Solidi (B) 231,
356 (2002).
6
. T.S. Argunova, J. Baruchel, and J. H a¨ rtwig, in IEEE 10th Inter-
national Conference Proceedings (SIMC-X), Berkeley, CA (IEEE,
Piscataway, NJ, 1998), p. 287.
7
. S. Milita, R. Madar, J. Baruchel, M. Anikin, and T. Argunova,
Mater. Sci. Eng. B61–62, 63 (1999).
. Yu.A. Vodakov, A.D. Roenkov, M.G. Ramm, E.N. Mokhov, and
12. J. Hirth and J. Lothe, Theory of Dislocations (John Wiley, New
York, 1982).
8
Yu.N. Makarov, Phys. Status Solidi (B) 202, 177 (1997).
J. Mater. Res., Vol. 17, No. 10, Oct 2002
2711
http://journals.cambridge.org
Downloaded: 18 Mar 2015
IP address: 169.230.243.252