Content uploaded by C.A. Londos
Author content
All content in this area was uploaded by C.A. Londos on Mar 19, 2014
Content may be subject to copyright.
Divacancy-oxygen and trivacancy-oxygen complexes in silicon:
Local Vibrational Mode studies
L.I. Murin1,a, B.G. Svensson2,b,J.L. Lindström3,c
V.P. Markevich4,d and C.A. Londos5,e
1Scientific-Practical Materials Research Centre of NAS of Belarus, BY-220072 Minsk, Belarus
2Oslo University, Physics Department/Centre for Materials Science and Nanotechnology,
N-0318 Oslo, Norway
3Lund University, Division of Solid State Physics, SE-22100 Lund, Sweden
4University of Manchester, School of Electrical and Electronic Engineering,
Manchester M60 1QD, UK
5Physics Department, Athens University, 15784 Athens, Greece
amurin@ifttp.bas-net.by, bb.g.svensson@fys.uio.no, cLennart.Lindstrom@ftf.lth.se
dv.markevich@manchester.ac.uk, ehlontos@phys.uoa.gr
Key Words: Silicon, vacancy-oxygen complexes, vibrational modes.
Abstract. Fourier transform infrared absorption spectroscopy was used to study the evolution of
multivacancy-oxygen-related defects in the temperature range 200-300 °C in Czochralski-grown Si
samples irradiated with MeV electrons or neutrons. A clear correlation between disappearance of
the divacancy (V2) related absorption band at 2767 cm-1 and appearance of two absorption bands
positioned at 833.4 and 842.4 cm-1 at 20 K (at 825.7 and 839.1 cm-1 at room temperature) has been
found. Both these two emerging bands have previously been assigned to a divacancy-oxygen defect
formed via interaction of mobile V2 with interstitial oxygen (Oi) atoms. The present study shows,
however, that the two bands arise from different defects since the ratio of their intensities depends
on the type of irradiation. The 842.4 cm-1 band is much more pronounced in neutron irradiated
samples and we argue that it is related to a trivacancy-oxygen defect (V3O) formed via interaction
of mobile V3 with Oi atoms or/and interaction of mobile V2 with VO defects.
Introduction
Local Vibrational Mode (LVM) spectroscopy has appeared to be a very powerful tool in studies of
the oxygen-related defects of different type in Si [1-13], including small oxygen clusters [9], self-
interstitial- [10] and vacancy-oxygen aggregates [11]. Among the latter defects are the well known
vacancy-oxygen (VO) complex or A-center [2], VO2, VO3 and VO4 defects [3-8]. More recently,
LVM signatures of more complicated defects, VO5 and VO6, have been found [12]. However, there
is another group of vacancy-oxygen aggregates, the so-called multivacancy-oxygen (VnO, n≥2)
defects, for which the previous LVM studies have not led to a clear and self-consistent picture.
One of the main reasons is that all members of the VnO family contain a Si-O-Si bonding
structure like that for VO, and as it was already noted in Ref. 13 the oxygen-related vibrational
bands of VnO in some cases could hardly be resolved from the more intensive 836 cm-1 band due to
the A-center. This is in agreement with the ab-initio calculations [14, 15], which have also predicted
that the frequencies of LVMs of such defects as VO and V2O should be very close. Since
concentrations of the VnO defects are lower than that of VO the vibrational bands of VnO are
expected to appear as satellites to the main VO band. In 1964 Ramdas and Rao [4] reported a
number of lines appearing around the main VO band, upon annealing in the temperature range 200-
400 °C of neutron-irradiated Cz-Si. The most pronounced satellites were positioned at about 829,
833 and 842 cm-1 at low temperature and were labeled as S1, S2 and S3, respectively. The centers
Solid State Phenomena Vols. 156-158 (2010) pp 129-134
online at http://www.scientific.net
© (2010) Trans Tech Publications, Switzerland
All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of the
publisher: Trans Tech Publications Ltd, Switzerland, www.ttp.net. (ID: 195.134.94.101-13/07/09,10:04:36)
responsible for the latter two bands were generated without any expense of the VO defect, and were
suggested to arise from multivacancy-oxygen complexes with a V2O center as a very likely
candidate.
Studies employing high resolution Fourier transform infrared (FTIR) absorption
spectroscopy have confirmed the results of Ref. 4 and show that the picture is even more
complicated with a very rich satellite spectrum of VO [16]. A number of satellite lines were also
observed in electron irradiated Cz-Si upon annealing in the temperature range 200-400 °C [17, 18].
However, no attempts were undertaken in Refs. 16-18 to elucidate the origin of the defects giving
rise to the satellite bands. A more straightforward and rather extensive study of VnO defects in
neutron-irradiated Cz-Si has been done by the Athens group [7, 19, 20]. Based on annealing studies
and semi-empirical modeling, Londos et al. [7, 19, 20] assigned the bands positioned at about 839
and 884 cm-1 at room temperature to V2O and V3O complexes, respectively. The 839 cm-1 band was
produced during anneals at temperatures above 230 °C, when divacancies are known to become
mobile and the appearance of the band was linked to the trapping of V2 by Oi. However, V2
annealing was not monitored in Ref. 19. Further, the identification of the V3O band is also rather
tentative. A side-band at 884 cm-1 was found only as a shoulder of the more intense VO2 band at
887 cm-1 via a fitting procedure [20]. In general, appearance of such shoulders can occur due to the
presence of 29Si and 30Si isotopes in natural silicon, and in this context it can be noted that earlier [1,
13] the 887 cm-1 band itself was erroneously assigned to V3O.
There is no consensus on identification of the V2O bands either. Low-temperature FTIR studies
by Lindström et al. [8] showed a correlation between the disappearance of V2, monitored through its
electronic transition observed at 2767 cm-1, and the growth of the S2 shoulder at 833.4 cm-1 in 2.5
MeV electron-irradiated Cz-Si samples. Accordingly, the 833.4 cm-1 band was assigned to V2O.
In the present work an attempt is made to obtain more solid identifications of LVMs due to V2O
and V3O defects via comparative FTIR studies of the VO satellite lines in electron- and neutron
irradiated Czochralski-grown (Cz) Si materials with measurements carried out at both, low and
room temperatures.
Experimental details
The samples used in this investigation were prepared from phosphorus doped n-type Cz-Si crystals
(ρ = 1-50 Ω-cm),). The concentrations of interstitial oxygen ([Oi] = (0.8-1.3)×1018 cm-3) and
substitutional carbon ([Cs] = (1-50)×1015 cm-3) were determined from measurements of intensities
of absorption bands at 1107 and 605 cm-1 using the calibration coefficients 3.14×1017 and 0.94×1017
cm-2, respectively [11]. The samples were polished to an optical surface on two sides and the
dimensions were 10×6×3 mm3 or 10×6×5 mm3.
Irradiations with 2.5 MeV electrons and fast neutrons (5 MeV) were performed at nominal room
temperature (≤ 350 K) with fluencies in the range 1⋅1016-1⋅1018 cm-2 and the samples were kept at
RT at least for several weeks before measurements. Isochronal annealing studies have been carried
out in the temperature range 75-400 °C with 25 °C increments for 30 min at each temperature.
IR absorption analysis was carried out using a Bruker IFS 113v spectrometer. A spectral
resolution of 0.5 or 1.0 cm-1 was used and the samples were measured at about 20 K (low
temperature - LT), and at room temperature (RT).
Experimental results
Evidently, upon room temperature irradiation the VnO defects in silicon can be generated via
sequential trapping of mobile vacancies by Oi, VO, V2O etc, i.e., via the reactions V + Oi ⇒ VO, V
+ VO ⇒ V2O, V + VnO ⇒ Vn+1O. However, in Cz-Si, where the oxygen concentration is normally
of about 1018 cm-3, the generation of VnO (n ≥ 2) may be efficient only at very high doses of
irradiation when VO concentration is comparable with [Oi]. At low fluencies when the
130 Gettering and Defect Engineering in Semiconductor Technology
XIII
concentration of radiation-induced defects is much lower than [Oi], the production of V2O appears
to be negligible even in the case of neutron irradiation. As an example, Figures 1a and 1b show the
absorption spectra around the VO band measured at 20 K and at RT for a sample irradiated with
neutrons to a fluence of 1×1017 cm-2. The band shapes are analyzed using a fitting procedure where
the effect of silicon isotopes (29Si and 30Si) has been taken into account. In the case of the LT
spectrum an excellent agreement is
observed between the calculated spectrum
obtained with the use of Lorentzians and
the measured one. The presence of an
additional peak at 834.45 cm-1, which
could be related to V2O has been found
upon fitting, but its intensity is very low
(see curve 4 in Fig. 1a). Also for the RT
measurements a reasonable agreement is
obtained between the measured and
calculated values (Fig. 1b), although the
correspondence of the spectra was worse
than that for the LT spectra. In the
following, we will concentrate mainly on
the LT spectra with the additional benefit
of monitoring the divacancy annealing via
changes in the intensity of the 2767 cm-1
band related to V2 electronic excitations
[21,22]. In all the samples studied the
2767 cm-1 band was strong and clearly
observed, e.g., in the case of neutron
irradiation its amplitude amounted up to
about 5 cm-1.
Divacancies in silicon are mobile at
temperatures above 200 °C and in Cz-Si
crystals the interstitial oxygen has been
suggested [21] to be the main trap of
mobile V2, i.e., a transformation of V2
into V2O can be expected to occur via the
reaction V2 + Oi ⇒ V
2O. It is worth
noting here that the occurrence of such a
reaction has been confirmed in detailed
DLTS studies [23]. Appearance of new
defects upon the V2 elimination has also
been clearly observed in the present
infrared absorption studies.
Fig. 2 shows a fragment of the LT
spectrum measured for the sample used for measurements shown in Fig.1 after annealing at 250 °C
for one hour. Such treatment resulted in a strong decrease (~90%) of the V2 related absorption band
at 2767 cm-1 and the appearance of a complex structure around the main VO band.
A fitting procedure using Lorentzians was used again to analyze the data. In addition to the main
absorption band related to VO, four relatively strong bands appeared in the spectra. For each band
the presence of all three Si isotopes was taken into account upon fitting. For clarity, only the fitting
sub-curves 1-5 corresponding to 28Si-O-28Si units are shown in Fig. 2, but the main fitting curve
accounts for all the contributions. For further validation the fitting results we have analyzed also
820 830 840 850
0.0
0.4
0.8
1.2
1.6
b)
4
32
1
1 - 835.78 cm-1 (28Si-O-28Si)
2 - 834.28 cm-1 (29Si-O-28Si)
3 - 832.83 cm-1 (30Si-O-28Si)
4 - 834.45 cm-1
T = 20 K
Absorption coefficient, cm-1
Wavenumber, cm-1
810 820 830 840 850
0.0
0.2
0.4
0.6
0.8
c)
3
2
1
T = 300 K
1 - 830.2 cm-1 (28Si-O-28Si)
2 - 828.7 cm-1 (29Si-O-28Si)
3 - 827.25 cm-1 (30Si-O-28Si)
Absorption coefficient, cm-1
Wavenumber, cm-1
Fig. 1 Fragments of absorption spectra measured at
20 K (a) and at room temperature (b) for a Cz-Si
sample ([Oi] = 1.3×1018, [Cs] ≤ 1×1015, [P] =
7×1013 cm-3) irradiated with 5 MeV neutrons to a
dose of 1×1017 cm-2. Solid lines are fitting curves.
Solid State Phenomena Vols. 156-158 131
difference absorption for all the bands. The corresponding difference absorption spectrum is shown
in Fig. 3.
For completeness, Fig. 4 shows the spectra measured at RT for the sample used for the spectra in
Figs. 2 and 3. Two main satellite bands positioned at about 826 and 839 cm-1 appear in the RT
spectra. Apparently, these bands correspond to the 833.4 and 842.4 cm-1 bands observed in the LT
spectra. It should be noted here that the 848.7 cm-1 band observed at LT has disappeared at RT. This
band disappears also in the LT spectra when optical excitation from the spectrometer is suppressed
by using a Ge filter and concurrently, the intensity of the 842.4 cm-1 band increases. Also
measurements with a Ge filter showed a
820 830 840 850
0.0
0.2
0.4
0.6
0.8
T = 20 K
5
4
3
2
1
1 - 833.4 cm-1
2 - 835.8 cm-1
3 - 837.0 cm-1
4 - 842.4 cm-1
5 - 848.7 cm-1
Absorption coefficient, cm-1
Wavenumber, cm-1
820 830 840 850
0.0
0.5
1.0
1.5
2.0 1 - 833.4 cm-1
2 - 835.8 cm-1
3 - 837.0 cm-1
4 - 842.4 cm-1
5 - 848.7 cm-1
5
4
3
2
1
T = 20 K
Absorption coefficient, cm-1
Wavenumber, cm-1
800 810 820 830 840 850 860
0.0
0.2
0.4
0.6
0.8
1.0 830
834
839
826
T = 300 K
2
1
Absorption coefficient, cm-1
Wavenumber, cm-1
Fig. 3. Fragment of a difference-
absorption spectrum obtained by
subtracting the spectrum measured
at 20 K after irradiation from the
spectrum measured after annealing
for 1h at 250 C for the Si sample
used for Fig. 1. Solid lines are
fittin
g
curves.
Fig. 4. Curve 1 - fragment of the
RT absorption spectrum for the
Si sample used for Fig. 1, after
annealing at 250 °C for 1 h.
Curve 2 - fragment of a
difference-absorption spectrum
obtained by subtracting the
spectrum measured at 300 K
after irradiation from the
spectrum measured after
annealing for 1h at 250 C for the
same Si sample.
Fig. 2. Fragment of the LT absorption
spectrum for the Si sample used for
Fig. 1, after annealing at 250 C for
1 h. Solid lines are fitting curves.
132 Gettering and Defect Engineering in Semiconductor Technology
XIII
significant decrease in the intensity of the band 837.0 cm-1 and a corresponding growth in intensity
of the 833.4 cm-1 band. These facts demonstrate that the 837.0 and 848.7 cm-1 bands are related to
optically excited states of the defects giving rise to the 833.4 and 842.4 cm-1 bands, respectively.
For the 837.0 cm-1 band there is also another defect which contributes since the band can be partly
observed at RT and has a higher thermal stability as compared with the 833.4 cm-1 band.
Discussion
Thus, in agreement with the previously published results, our data demonstrate that the annealing of
divacancies in neutron-irradiated Cz-Si is accompanied by the appearance of new absorption bands.
The most intense of them are located at 833.4 and 842.4 cm-1 (826 and 839 cm-1 at RT). Isochronal
(30 minutes) annealing studies have shown that not only their formation processes but also
annihilation kinetics are very similar. The bands at 833.4 and 842.4 cm-1 disappear simultaneously
in the temperature range 300-350 °C.
One can suggest that both bands arise from the same defect, namely V2O, being in different
configurations. However, there are some crucial facts that do not support such a suggestion. Firstly,
the ratio of intensities of these bands is the same at LT and at RT. It is difficult to imagine two
different configurations of the V2O defect that are equal in total energy. Besides, only one V2O
center has been observed in DLTS studies and practically full transformation of V2 into this center
occurs. The second and, probably, the most important fact is that the ratio of intensities of these
bands depends on the type and fluence of irradiation. As an example, Fig. 5 shows fragments of the
absorption spectra with these bands for electron- and neutron-irradiated samples. Evidently, the
842.4 cm-1 band is much more pronounced after neutron irradiation and so, it originates most likely
from a more complex defect than that responsible for the 833.4 cm-1 band. A possible candidate is
the V3O defect which may be generated via interaction of mobile divacancies with A-centers, i.e.,
via the reaction V2 + VO ⇒ V3O.
However, it appears that this reaction can not account for the observed overall generation of V3O
(the 842.4 cm-1 band), especially in samples with relatively low VO concentration. It is very likely,
that V3, produced mainly as a primary defect, has the same migration ability as V2, and V3O can be
also generated via the reaction V3 + Oi ⇒ V3O. Such a suggestion is in agreement with EPR data
[24] on the thermal stability of V3. On the other hand, according to the EPR data by Lee and Corbett
[25] the V3O defect is likely responsible for the P4 spectrum which appears upon annihilation of
VO and V2O at about 350 °C and that is at variance with our assignment of the 842.4 cm-1 band.
However, Lee and Corbett [25] have noted that a V3O2 defect, being in a certain configuration, can
also give rise to the P4 spectrum and this assignment is more consistent with our interpretation.
820 830 840 850
0.0
0.1
0.2
0.3
0.4 833.4
842.4
T = 20 K
2
1
x0.5
Absorption coefficient, cm-1
Wavenumber, cm-1
Fig. 5. Fragments of difference-
absorption spectra. Curve 1 is the
same as shown in Fig. 3, but with
intensity of the peaks scaled down
by a factor of 2. Curve 2 was
obtained by subtracting the
spectrum measured at 20 K after
irradiation from the spectrum
measured after annealing for 30
min at 280 °C for a sample
irradiated with 2.5 MeV electrons
to a dose of 1×1018 cm-2.
Solid State Phenomena Vols. 156-158 133
Conclusions
In conclusion, high resolution LVM spectroscopy have shown that two absorption bands positioned
at 833.4 and 842.4 cm-1 (at 20 K) appear simultaneously with the elimination of divacancies in
irradiated Cz-Si samples. The experimentally observed band shapes can be nicely fitted by
calculations with the use of Lorentzians, in which contributions from all the three stable Si isotopes
are taken into account. In contrast to previous studies reported in the literature, where both these
bands have been attributed to the V2O center, we have assigned only the band at 833.4 cm-1 band to
V2O, which is formed via trapping of mobile V2’s by Oi. The relative intensity of the 842.4 cm-1
band is substantially enhanced in neutron-irradiated samples compared to that in electron-irradiated
ones and hence, this band is associated with a higher order complex than that responsible for the
833.4 cm-1 band. A likely candidate is V3O center.
References
[1] B. Pajot, in Oxygen in Silicon, edited by F. Shimura, volume 42 of Semiconductors and Semimetals,
chapter 6, Academic, NY (1994).
[2] J.W. Corbett, G.D. Watkins, R.M. Chrenko and R.S. McDonald, Phys. Rev. Vol. 121(1961), p. 1015
[3] J.W. Corbett, G.D. Watkins and R.S. McDonald: Phys. Rev. Vol. 135 (1964), p. A1381
[4] A.K. Ramdas and M.G. Rao: Phys. Rev. Vol 142 (1966), p. 451
[5] B.G. Svensson and J.L. Lindström: Phys. Rev. B Vol. 34 (1986), p. 8709
[6] J.L. Lindström and B.G. Svensson: Mat. Res. Soc. Symp. Proc. Vol. 59 (1986), p. 45
[7] C.A. Londos, L.G. Fytros and G.J. Georgiou: Defect and Diffusion Forum Vol. 171-172 (1999), p. 1
[8] J.L. Lindström, L.I. Murin, V.P. Markevich, T. Hallberg and B.G. Svensson, Physica B Vol. 273-274
(1999), p. 291
[9] J Coutinho, R Jones, L.I. Murin, V.P. Markevich, J.L. Lindstrom, P.R. Briddon and S. Oberg, Phys.
Rev. Lett. Vol. 87 (2001), 235501
[10] J. Hermansson, L.I. Murin., V.P. Markevich, T.Hallberg, J.L. Lindstrom, M. Kleverman, B.G.
Svensson, Physica B Vol. 302-303 (2001), p. 188
[11] J.L. Lindström, L.I. Murin, T. Hallberg, V.P. Markevich, B.G. Svensson, M. Kleverman and J.
Hermansson: Nuclear Inst. and Methods in Physics Research B Vol. 186 (2002), p. 121
[12] L.I. Murin, J.L. Lindström, B.G. Svensson, V.P. Markevich, A.R. Peaker and C.A. Londos: Solid State
Phenomena Vol. 108-109 (2005), p. 267
[13] Y.-H. Lee, J. C. Corelli and J.W. Corbett: Phys. Lett. Vol. 60A (1977), p. 2653
[14] C.P. Ewels, R. Jones and S. Oberg: Mater. Sci. Forum. Vol. 196-201 (1995), p. 1297.
[15] M. Pesola, J. von Boehm, T. Mattila and R.M. Nieminen: Phys. Rev. B Vol. 60 (1999), p. 11449
[16] Y.V. Pomozov, L.I. Khirunenko, V.I. Shakhovtsov and V.I. Yashnik: Fiz. Tekh. Poluprovodn. Vol. 24
(1990), p. 993 [Sov. Phys. Semicond. Vol. 24 (1990), p. 624]
[17] A.R. Bean, R.C. Newman, R.C. Smith: J. Phys. Chem. Solids Vol. 31 (1970), p. 739
[18] L. Khirunenko, V. Shakhovtsov, Y. Pomozov, and V. Yashnik, in Early Stages of Oxygen
Precipitation in Silicon, edited by R. Jones (Kluwer Academic Publishers, Dordrecht, 1996), p. 403.
[19] N.V. Sarlis, C.A. Londos and L.G. Fytros: J. Appl. Phys. Vol. 81 (1997), p. 1645
[20] C.A. Londos, N.V. Sarlis and L.G. Fytros: J. Appl. Phys. Vol. 85 (1999), p. 8074
[21] L.J. Cheng., J.C. Corelli, J.W. Corbett and G.D. Watkins: Phys. Rev. Vol. 152 (1966), p. 761
[22] J.H. Svensson, B.G. Svensson and B. Monemar: Phys. Rev. B Vol. 38 (1988), p. 4192
[23] G. Alfieri, E.V. Monakhov, B.S. Avset and B.G. Svensson: Phys. Rev. B Vol. 68 (2003), 233202
[24] J.W. Corbett, J.C. Bourgoin, L.J. Cheng., J.C. Corelli, E.H. Lee, P.M. Mooney and C. Weigel, in
Radiat. Effects in Semic., edited by N.B. Urli and J.W. Corbett (Institute of Phys., Conf. Ser. N31,
1976), p.1.
[25] Y.H. Lee and J.W. Corbett, Phys. Rev. B Vol. 13 (1976), p. 2653
134 Gettering and Defect Engineering in Semiconductor Technology
XIII
A preview of this full-text is provided by Trans Tech Publications Ltd.
Content available from Solid State Phenomena
This content is subject to copyright. Terms and conditions apply.