Silicon vacancy related defect in 4H and 6H SiC
E. So
¨
rman, N. T. Son, W. M. Chen, O. Kordina, C. Hallin, and E. Janze
´
n
Department of Physics and Measurement Technology, Linko ¨ping University, S-581 83 Linko ¨ping, Sweden
Received 18 August 1999
We report on an irradiation-induced photoluminescence PL band in 4H and 6H SiC and the corresponding
optically detected magnetic resonance ODMR signals from this band. The deep PL band has the same
number of no-phonon lines as there are inequivalent sites in the respective polytype. These lines are at 1352
and 1438 meV in the case of 4H and at 1366, 1398, and 1433 meV in the case of 6H. The intensity of the PL
lines is reduced after a short anneal at 750 °C. ODMR measurements with above-band-gap excitation show
that two spin-triplet ( S =1) states with a weak axial character are detected via each PL line in these bands. One
of these two triplet states can be selectively excited with the excitation energy of the corresponding PL line.
These triplet signals can therefore be detected separately and only then can the well documented and charac-
teristic hyperfine interaction of the silicon vacancy in SiC be resolved. Considering the correlation between the
irradiation dose and the signal strength, the well established annealing temperature and the characteristic
hyperfine pattern, we suggest that this PL band is related to the isolated silicon vacancy in 4H and 6H SiC. The
spin state ( S =1) implies a charge state of the vacancy with an even number of electrons. By combining the
knowledge from complementary electron-spin resonance measurements and theoretical calculations we hold
the neutral charge state for the strongest candidate.
I. INTRODUCTION
SiC is a very promising semiconductor for devices that
have to work under extreme conditions. Material properties,
such as a wide band gap, good thermal conductivity, a rela-
tively high mobility, and high breakdown voltage, make it
appropriate for many demanding applications. SiC is the
classical example of polytypism, i.e., the material can crys-
tallize in many different but closely related crystal structures.
The unit cells of the 4H and 6H polytypes contains 4 * 2
=8 and 6 * 2 =12 atoms, respectively. The various atomic
sites in such large unit cells have different configurations of
their neighbors. The nearest neighbors are always tetra-
hedally oriented but the next-nearest neighbors can either be
in a cubic k or a hexagonal h configuration. In 4H SiC
there is one site of each type h and k while in 6H SiC there
are one hexagonal and two cubic sites h, k
1
and k
2
. The
two cubic sites differ in the third shell of neighbors.
This paper will dwell on a primary defect in SiC that in
this case has been produced by high-energy particle bom-
bardment. If the energy of these particles is high enough, the
atoms in the lattice can actually be pushed out of their
places. The primary defects created in this way are generally
vacancies and interstitials, but in a binary compound like SiC
also the antisites can be formed. There is a reason why a
good knowledge about primary defects is especially impor-
tant for SiC. The poor diffusivity of dopants in SiC makes
ion implantation a more attractive doping method. The bom-
bardment of the material with heavy ions creates a lot of
defects in the material. In contrast to silicon these primary
defects in SiC seem to be stable at room temperature
1–4
and
some secondary defects produced during a post-irradiation
thermal annealing are stable at least up to 2000 °C,
5
i.e.,
they are virtually impossible to get rid of. This group of
defects are therefore not only of a great scientific but also of
great technological interest, since they have to be dealt with
in any device process that involves ion implantation.
The primary defects are of course also fundamentally in-
teresting. At quite an early stage in semiconductor science
late 1960s and early 1970s electron-spin resonance ESR
measurements revealed some of the fascinating microscopic
properties of the vacancy in silicon. Less is still known about
the vacancy in diamond, but the properties of the negative
charge state has been thoroughly analyzed by ESR. The re-
sults from silicon and diamond are in one respect strikingly
different. While the different charge states of the vacancy in
silicon exhibit a strong lattice relaxation, i.e., a dominating
Jahn-Teller effect, the vacancy in diamond has the high-spin
ground-state characteristic for a dominating spin-spin inter-
action. The extreme relaxation in silicon even results in a
reversing of the expected ordering of the different charge
states in the band gap. It is therefore not strange that these
two fundamental point defects have served as interesting
model cases in the development of defect theories.
6
Since
SiC in a way is a mixture of the two, it is now interesting if
some of the properties of its vacancies can be investigated.
Despite all the results reported from investigations of pri-
mary defects in SiC by methods like ESR, photolumines-
cence PL and deep-level transient spectroscopy DLTS,
the knowledge about them is far from as developed as sili-
con. Some experimental knowledge about the carbon and
silicon vacancy has, however, been gained. In this paper we
will mainly refer back to the extensive ESR investigations on
3C,
1,4
6H,
3,4
and 4H Ref. 7 SiC, of what is now accepted to
be the ESR signal from the negatively charged silicon va-
cancy. The deep PL band, that we here will claim to be
related to the silicon vacancy, can be observed from as-
grown Lely crystals and was thoroughly investigated already
during the 1970s. The characteristics of the band was then
reported for the 6H,
8,9
15R,
8,9
and 33R Ref. 10 polytype.
Historical remark: It was then usually referred to as the abc
band. The most thorough work is the PL study by Hagen
PHYSICAL REVIEW B 15 JANUARY 2000-II VOLUME 61, NUMBER 4
PRB 61 0163-1829/2000/614/26138/$15.00 2613 ©2000 The American Physical Society