Direct determination of the silicon donor ionization energy in homoepitaxial
AlN from photoluminescence two-electron transitions
B. Neuschl, K. Thonke, M. Feneberg, R. Goldhahn, T. Wunderer et al.
Citation: Appl. Phys. Lett. 103, 122105 (2013); doi: 10.1063/1.4821183
View online: http://dx.doi.org/10.1063/1.4821183
View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v103/i12
Published by the AIP Publishing LLC.
Additional information on Appl. Phys. Lett.
Journal Homepage: http://apl.aip.org/
Journal Information: http://apl.aip.org/about/about_the_journal
Top downloads: http://apl.aip.org/features/most_downloaded
Information for Authors: http://apl.aip.org/authors
Downloaded 23 Sep 2013 to 13.1.101.37. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
APPLIED PHYSICS LETTERS 103, 122105 (2013)
Direct determination of the silicon donor ionization energy in homoepitaxial
AlN from photoluminescence two-electron transitions
B. Neuschl,1,a) K. Thonke,1 M. Feneberg,2 R. Goldhahn,2 T. Wunderer,3 Z. Yang,3
N. M. Johnson,3 J. Xie,4 S. Mita,4 A. Rice,5 R. Collazo,5 and Z. Sitar5
1
Institute of Quantum Matter/Semiconductor Physics Group, University of Ulm, Albert-Einstein-Allee 45,
89069 Ulm, Germany
2
Institut f€
ur Experimentelle Physik, Otto-von-Guericke-Universit€
at Magdeburg, Universit€
atsplatz 2, 39106
Magdeburg, Germany
3
Palo Alto Research Center, Inc., 3333 Coyote Hill Road, Palo Alto, California 94304, USA
4
Hexatech, Inc., 991 Aviation Pkwy., Suite 800, Morrisville, North Carolina 27560, USA
5
Department of Materials Science and Engineering, North Carolina State University, Raleigh, North Carolina
27695-7919, USA
(Received 15 July 2013; accepted 27 August 2013; published online 18 September 2013)
We report on the identification of a two-electron transition for the shallow donor silicon in
homoepitaxial aluminum nitride (AlN). One c-oriented sample was analyzed by low temperature
photoluminescence spectroscopy on multiple excitation spots. We find a unique correlation of one
single emission band, 76.6 meV below the free excitonic emission, with the luminescence
of excitons bound to neutral silicon proving the identity as a two-electron transition. The
assignment is confirmed by temperature dependent photoluminescence investigations. We find a
C 2013 AIP Publishing LLC.
donor ionization energy of ð63:5 6 1:5Þ meV for silicon in AlN. V
[http://dx.doi.org/10.1063/1.4821183]
Recent interest in light emitting devices operating in the
deep UV spectral region1 is due to promising applications
such as the deactivation of microorganisms by exposure to
light with short wavelengths. Several research groups are
working on light emitting diodes (LEDs) based on aluminum
gallium nitride (AlGaN) fabricated on foreign substrates such
as sapphire2–4 or silicon carbide.5 As a consequence, these
structures suffer from high defect densities due to the lattice
and thermal expansion coefficient mismatch between the substrate material and the epitaxial film, directly degrading the efficiency of the device.6 One approach is to deposit the diode
layer, with high Al content, directly on an AlN single crystal.
This kind of substrate is promising for an improvement of the
crystal quality,7,8 and recently, well operating LEDs9 and
lasers10,11 have been presented. Nevertheless, many fundamental properties such as the exact electronic structure for
typical dopants are not known. Only recently, homoepitaxial
AlN layers of good crystal quality became available allowing
the proper identification of single bound excitonic emission
bands.12,13 However, for the mostly present donor silicon (Si),
there is no agreement on the ionization energy. There is an
ongoing discussion, whether SiAl undergoes a strong lattice
relaxation and forms a deep DX center, where the dopant
atom captures a second electron. Several theoretical studies
predict DX center formation for SiAl,14,15 whereas others find
silicon to be a shallow donor in wurtzite AlN.16,17
Experimentally, the majority of studies shows n-type conductivity after silicon doping,18–24 although they find very different values for the ionization energy ranging from 86 to
570 meV. On the other hand, there are two detailed reports
about DX center formation found by spectrally resolved photoconductivity and electron paramagnetic resonance.25,26
a)
Author to whom correspondence should be addressed. Electronic mail:
[email protected]
0003-6951/2013/103(12)/122105/5/$30.00
In this letter, we present a detailed investigation of a
homoepitaxially grown wurtzite AlN layer by means of temperature dependent photoluminescence (PL) spectroscopy.
Exposure at multiple excitation spots and the temperature dependent behavior of the emission bands observed allow for
an identification of a two-electron transition with shallow silicon donors involved. We further discuss the presented
results in the framework of possible DX center formation.
The sample under investigation was grown by metal organic vapor phase epitaxy on a physical vapor transport
grown AlN single crystal.7,8 The c-oriented, nominally
undoped layer of wurtzite AlN is 500 nm thick and reveals a
very low defect density <104 cm2. For PL investigations,
the sample was cooled down in a helium flow cryostat,
allowing for sample temperatures between 8 K and room
temperature, and excited by an ArF excimer laser (193 nm).
The luminescence light was spectrally resolved in a grating
monochromator SPEX1704 having 1 m focal length, and the
signal was detected by a liquid nitrogen cooled, backilluminated charged coupled device camera. UV optimized
optics allow for a spectral resolution better than 100 leV
at 6 eV.
Figure 1 shows a representative PL spectrum around the
excitonic bandgap. By analyzing multiple PL spectra taken
at different spots on the sample, a statistical approach was
used to determine the transition energies of the observed
emission bands. Differences for the emission energies were
found, probably due to local strain variations. For this reason, all absolute emission energies given in the following are
averaged over all the energy values measured, weighted by
the intensity of the free exciton emission in the individual
spectrum. A standard deviation of 3 meV for absolute energies and below 1 meV for energy differences was deterat 6:0428 eV is
mined. The free exciton emission Xn¼1
A
expected to have C5 symmetry, i.e., it should correspond to a
103, 122105-1
C 2013 AIP Publishing LLC
V
Downloaded 23 Sep 2013 to 13.1.101.37. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
122105-2
Neuschl et al.
FIG. 1. Photoluminescence spectrum of the excitonic bandgap at low temand first
peratures. Emission from the free excitonic ground state Xn¼1
A
is visible besides two donor bound exciton bands Do X
excited state Xn¼2
A
and Sio X. At lower energies, longitudinal optical phonon replica can be
seen, as well as a contribution which we assign to the two-electron satellite
emission Sio X : 2s; p.
c-oriented dipole and thus be observable in detection perpendicular to the c-oriented sample surface.27 From the first
n¼2
, the free exciton binding
excited free exciton state XA
energy of 51.7 meV can be derived in agreement with earlier
results.12 Two donor bound exciton emission bands show up
n¼1
emission, where the
13.4 meV and 28.6 meV below the XA
o
deeper one is labeled Si X and arises from excitons bound to
neutral silicon donors.12 They dominate the emission spectrum, although the sample is not intentionally doped. Si is a
common trace impurity for AlN growth.8 The full width at
half maximum of the Sio X emission band is 500 leV demonstrating the excellent crystal quality of the sample. The
chemical nature of the donor responsible for the Do X emission with 13.4 meV exciton localization energy is not known.
In the following, we discuss one specific emission band
located 76.6 meV below the free exciton showing up in the
region of the longitudinal optical (LO) phonon replica
Appl. Phys. Lett. 103, 122105 (2013)
sideband. For the normal recombination of donor bound
excitons ðDo XÞ, a competing process is expected, where the
donor is left in an excited final state, e.g. the 2s or 2p state of
the donor bound electron. In this process, a photon is emitted
with an energy difference (E2s;p E1s ) below the normal
(Do X) recombination. In a simple hydrogenic model for the
donor binding states, the photon energy is lowered by three
quarters of the donor ionization energy, giving rise to the
two-electron satellite (TES) lines tentatively labelled
Sio X : 2s; p. The relative intensities of the TES and normal
Sio X emission are expected to be independent of the sample
spot. To investigate this point, we excited the sample on
more than 50 different spots. Local fluctuations of the dopant
atoms are expected but cannot be directly detected by secondary ion mass spectroscopy due to its spatial resolution
limit. Figure 2 shows on a series of such spectra that a more
pronounced Sio X emission leads to a stronger Sio X : 2s; p luminescence band. We quantitatively analyzed this behavior
by fitting multiple Gaussian curves to each spectrum shown
in Figure 2. The result is shown in Figure 3, where the linear
correlation of both emission bands under investigation
proves their linear interdependence in intensity. From the
crossing point with the intensity axis, the relative transition
probability of the TES and the Sio X can be estimated to
1:230, which is in line with observations for GaN and many
other semiconductor materials. Except for this specific emission, no further emission band was found to correlate linearly
with the Sio X luminescence. We consider this as the first
direct evidence for two-electron transitions, besides some
earlier speculations based on rather noisy temperature dependent cathodoluminescence spectra recorded on AlN.28 To
verify our identification, we carried out temperature dependent PL investigations. Figure 4 summarizes spectra recorded
at temperatures between 8 and 60 K. Apparently, both the
Sio X and the corresponding TES decrease simultaneously.
Further emission bands found in the low energy range in the
left graph of Figure 4 decrease as well with increasing
temperature. Again, we quantified the measurements by
FIG. 2. PL investigation for multiple
excitation spots. The spectra within
both graphs are ordered the same such
that the Sio X and the Sio X : 2s; p intensity increase from bottom to top simultaneously. For clarity, the spectra were
shifted vertically as well as horizontally such that the position of the free
exciton band matches 6:0428 eV.
Downloaded 23 Sep 2013 to 13.1.101.37. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
122105-3
Neuschl et al.
FIG. 3. Comparison of emission intensities of the Si bound exciton with the
according TES emission. A linear correlation is obvious due to the slope of
1 in the double logarithmic plot.
Gaussian fitting and compared the emission intensities. It
turned out that only the emission labeled Sio X : 2s; p shows
a linear correlation with the Si bound exciton, confirming
our identification. In analogy to the TES emission in GaN,
the donor electron of Si either remains in the excited 2s
state29 or in the 2p state.30 Following the discussion in a
comment given by Freitas et al.,31 we probably cannot sufficiently resolve both spectral contributions, but for the sample
temperature of 8 K the transition 1s-2s is believed to dominate over the 1s-2p process. From the 48 meV energy difference between the Sio X : 2s; p and the Sio X luminescence,
the donor ionization energy for silicon can be derived assuming a simple hydrogen-like model. This yields
Ed ¼ ð64 6 1Þ meV, whereas effective mass theory (EMT)
predicts 62 meV using an effective electron mass of
me ¼ 0:32m0 ,32 and the static dielectric constant s ¼ 8:39.33
The small discrepancy is due to the chemical shift, which
lowers the ground state of the donor electron. For this reason,
we probably slightly overestimate, and EMT underestimates
the total donor ionization energy. Alternatively, the sum of
48 meV energy difference between the ground and the first
excited state of the donor electron and one quarter of
62 meV, which is the calculated energy distance between the
Appl. Phys. Lett. 103, 122105 (2013)
first excited state and the conduction band according to
EMT, yields ð63:5 6 1:5Þ meV for the donor ionization
energy. Hence, the chemical shift amounts to 1.5 meV. To
calculate the coefficients for Haynes rule is still not possible,
because pairs of both the ionization energy and the exciton
localization energy are needed for at least two different types
of donors. However, we find that the linear coefficient
between both energies of 0.45 exceeds 0.214 as given from
Freitas et al.29 for gallium nitride.
Nevertheless, the resultant energy is relatively small as
compared to the findings in various Hall investigations.18–24
To bridge this gap, large lattice relaxation has to be considered, as it was found earlier for arsenide material systems.34,35
By capturing a second electron, the silicon donor atom is displaced away from one of the surrounding nitrogen atoms by
approximately one third of the binding length. The ground
state for this configuration, called DX center, is lower in
energy than the neutral shallow donor ground state d , lowered
by the so-called formation energy. Following the configuration
scheme given by Zeisel et al.25 and the findings of Son
et al.,26 we believe that silicon is in the DX– ground state for
low temperatures in darkness. The higher d8 state is populated
under laser excitation with a photon energy above 1.5 eV
(Ref. 25) via an excited neutral DX level in the conduction
band. The formation energy was estimated to be 78 meV,26
but in fact the difference between the d8 level and the Fermi
energy EF was measured. The DX– level of another residual
donor oxygen (O) is expected to be below DX
Si , and EF
should be in between those two levels, depending on the donor
concentrations. In Hall experiments, the electrons are excited
from EF into the conduction band. The more O is incorporated
in the sample, the lower the Fermi level and the larger the
derived activation energy Ea should be. This assumption is
confirmed by Hall investigations, where Ea increases from
180 to 570 meV for increased O concentrations.19,22,23
A thermal barrier of around 5 meV was found,25,26
where Si in the d8 state undergoes a large lattice relaxation
and populates the deep DX– state. Furthermore, their electron paramagnetic resonance (EPR) signal, where the d8 concentration is measured, was again increased even in darkness
FIG. 4. PL investigation for various
temperatures, increasing from top to
bottom. For clarity, the spectra were
shifted vertically. We find a linear correlation of the Sio X : 2s; p and the
Sio X emission intensity.
Downloaded 23 Sep 2013 to 13.1.101.37. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
122105-4
Neuschl et al.
Appl. Phys. Lett. 103, 122105 (2013)
FIG. 5. PL investigation at various temperatures, increasing from top to bottom. The Sio X luminescence quenches quickly above 60 K whereas the Do X
emission appears to be more stable.
above T 95 K. This could be explained by thermal activation from Si in the DX– state, which would therefore be
located 10 meV below the shallow donor level.
A temperature dependent PL spectra series is shown in
Figure 5. The emission intensity of both bound exciton luminescence bands (Sio X; Do X) decreases relative to the free
exciton, when the sample temperature is increased. This is
due to the delocalization of bound excitons following a
Boltzmann population law. The deeper bound, the more stable an exciton is expected to be. Apparently, this fact seems
to be violated in the present case, where the Sio X emission
quenches much faster with increasing temperature as compared to the other Do X emission, which has a shallower exciton localization energy. The integrated emission intensity of
the different luminescence bands shown in Figure 5 was
determined and is shown as a function of temperature in
Figure 6. Additionally, the temperature dependency was fit
using Eq. (1) taking into account two activation energies E1
and E2 as well as the parameters I0 for the intensity at low
temperature, and different weighting factors A1 and A2 for
the competing processes
I0
IðTÞ ¼
1 þ A1 e
E1
BT
k
E2
þ A2 ekB T
:
(1)
For Sio X, the deactivation energies are E1 ¼ ð29 6 2Þ meV
and E2 ¼ ð5 6 2Þ meV, whereas we obtain E1 ¼
ð12 6 2Þ meV and E2 ¼ ð69 6 2Þ meV for the other Do X. E1
is almost identical to the localization energy of each donor
bound exciton, which is 28.6 meV and 13.4 meV, respectively. E2 is much larger for Do X compared to the Si bound
exciton, explaining the different thermal delocalization
behavior. The fact that the activation energies E1 match the
localization energies quite well hints to a minor effect of
nonradiative recombination processes even for high temperatures. The thermal quenching at around 60 K for Sio X is
described well by the thermal barrier of E2 ¼ 5 meV in accordance to the findings of Zeisel et al.25 and Son et al.26
Analogously, we assume that silicon in the shallow donor
state overcomes this thermal barrier via a metastable donor
state and falls into the deeper DX– level. An alternative explanation for this observation would be a lower amount of
silicon atoms incorporated in the crystal as compared to the
respective impurity or defect causing the Do X emission. The
efficiency of thermal dissociation of the luminescentE1Sio X
complex not only depends on the Boltzmann factor ekB T but
also on the prefactor A1 in Eq. (1). This factor is large, if the
total number of neutral Si donors is much lower than the
number of available states in the free exciton band. Another
aspect of Figure 6 is that the free exciton emission increases,
when the excitons bound to Si quench. In the initial state of a
donor bound exciton, the donor electron attracts the electron
of the former free exciton quasiparticle. The according
exchange interaction leads to a binding or anti-binding state
with total electron spin 0 or 1. In the binding state, the
energy of the bound exciton is lowered by the so-called exciton localization energy. Bound exciton emission arises from
annihilation of excitons in the binding state. We observe an
increase of the free exciton emission intensity, when the
localization energy of the bound exciton state is overcome
thermally. Therefore, we suggest that no excitons are localized to silicon in the DX– state, and, indeed, the deep DX–
center probably does not contribute to the emission spectra.
In conclusion, we revealed the contribution of a twoelectron transition caused by silicon impurities to the emission spectra of wurtzite AlN. Applying PL spectroscopy, the
linear correlation of the TES luminescence intensity with the
Sio X for either various excitation spots or sample temperatures proved this interpretation. An ionization energy of
ð63:5 6 1:5Þ meV for Si in AlN was derived. Beyond, we
argued that SiAl behaves as a shallow donor at low temperatures under exposure to laser radiation and probably undergoes a large lattice relaxation for temperatures above 60 K.
Therefore, we emphasize previous studies25,26 that SiAl in
wurtzite AlN forms a DX center with its DX– ground state
located closely below the shallow donor state.
Part of the work was supported by the Defense
Advanced Research Projects Agency CMUVT Program
(PM: Dr. John Albrecht) under U.S. Army Cooperative
Agreement No. W911NF-10-02-0102.
FIG. 6. Arrhenius plot of the PL intensity for the luminescence shown in Figure
5. The symbols refer to the measured data points, whose temperature dependency is well described by the solid lines generated according to Eq. (1).
1
M. Kneissl, T. Kolbe, C. Chua, V. Kueller, N. Lobo, J. Stellmach, A.
Knauer, H. Rodriguez, S. Einfeldt, Z. Yang, N. M. Johnson, and M.
Weyers, Semicond. Sci. Technol. 26, 014036 (2011).
Downloaded 23 Sep 2013 to 13.1.101.37. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
122105-5
2
Neuschl et al.
J. P. Zhang, A. Chitnis, V. Adivarahan, S. Wu, V. Mandavilli, R.
Pachipulusu, M. Shatalov, G. Simin, J. W. Yang, and M. A. Khan, Appl.
Phys. Lett. 81, 4910 (2002).
3
H. Hirayama, N. Noguchi, and N. Kamata, Appl. Phys. Express 3, 032102
(2010).
4
C. Pernot, M. Kim, S. Fukahori, T. Inazu, T. Fujita, Y. Nagasawa, A.
Hirano, M. Ippommatsu, M. Iwaya, S. Kamiyama, I. Akasaki, and H.
Amano, Appl. Phys. Express 3, 061004 (2010).
5
Y. Taniyasu, M. Kasu, and T. Makimoto, Nature 441, 325 (2006).
6
K. Ban, J. Yamamoto, K. Takeda, K. Ide, M. Iwaya, T. Takeuchi, S.
Kamiyama, I. Akasaki, and H. Amano, Appl. Phys. Express 4, 052101
(2011).
7
A. Rice, R. Collazo, J. Tweedie, R. Dalmau, S. Mita, J. Xie, and Z. Sitar,
J. Appl. Phys. 108, 043510 (2010).
8
R. Dalmau, B. Moody, R. Schlesser, S. Mita, J. Xie, M. Feneberg, B.
Neuschl, K. Thonke, R. Collazo, A. Rice, J. Tweedie, and Z. Sitar,
J. Electrochem. Soc. 158(5), H530 (2011).
9
J. R. Grandusky, S. R. Gibb, M. Mendrick, and L. J. Schowalter, Phys.
Status Solidi C 8, 1528 (2011).
10
T. Wunderer, C. L. Chua, J. E. Northrup, Z. Yang, N. M. Johnson, M.
Kneissl, G. A. Garrett, H. Shen, M. Wraback, B. Moody, H. S. Craft, R.
Schlesser, R. F. Dalmau, and Z. Sitar, Phys. Status Solidi C 9, 822 (2012).
11
J. Xie, S. Mita, Z. Bryan, W. Guo, L. Hussey, B. Moody, R. Schlesser, R.
Kirste, M. Gerhold, R. Collazo, and Z. Sitar, Appl. Phys. Lett. 102,
171102 (2013).
12
B. Neuschl, K. Thonke, M. Feneberg, S. Mita, J. Xie, R. Dalmau, R.
Collazo, and Z Sitar, Phys. Status Solidi B 249, 511 (2012).
13
M. Funato, K. Matsuda, R. G. Banal, R. Ishii, and Y. Kawakami, Appl.
Phys. Express 5, 082001 (2012).
14
C. H. Park and D. J. Chadi, Phys. Rev. B 55, 12995 (1997).
15
L. Silvestri, K. Dunn, S. Prawer, and F. Ladouceur, Appl. Phys. Lett. 99,
122109 (2011).
16
C. G. Van de Walle, Phys. Rev. B 57, R2033 (1998).
17
S. B. Zhang, S. H. Wei, and A. Zunger, Phys. Rev. Lett. 84, 1232 (2000).
18
Y. Taniyasu, M. Kasu, and N. Kobayashi, Appl. Phys. Lett. 81, 1255
(2002).
Appl. Phys. Lett. 103, 122105 (2013)
19
M. L. Nakarmi, K. H. Kim, K. Zhu, J. Y. Lin, and H. X. Jiang, Appl. Phys.
Lett. 85, 3769 (2004).
Y. Taniyasu, M. Kasu, and T. Makimoto, Appl. Phys. Lett. 85, 4672
(2004).
21
T. Ive, O. Brandt, H. Kostial, K. J. Friedland, L. D€aweritz, and K. H.
Ploog, Appl. Phys. Lett. 86, 024106 (2005).
22
M. Hermann, F. Furtmayr, A. Bergmaier, G. Dollinger, M. Stutzmann, and
M. Eickhoff, Appl. Phys. Lett. 86, 192108 (2005).
23
B. Borisov, V. Kuryatkov, Yu. Kudryavtsev, R. Asomoza, S. Nikishin, D.
Y. Song, M. Holtz, and H. Temkin, Appl. Phys. Lett. 87, 132106 (2005).
24
R. Collazo, S. Mita, J. Xie, A. Rice, J. Tweedie, R. Dalmau, and Z. Sitar,
Phys. Status Solidi C 8, 2031 (2011).
25
R. Zeisel, M. W. Bayerl, S. T. B. Goennenwein, R. Dimitrov, O.
Ambacher, M. S. Brandt, and M. Stutzmann, Phys. Rev. B 61, R16283
(2000).
26
N. T. Son, M. Bickermann, and E. Janzen, Appl. Phys. Lett. 98, 092104 (2011).
27
M. Feneberg, M. F. Romero, B. Neuschl, K. Thonke, M. R€
oppischer, C.
Cobet, N. Esser, M. Bickermann, and R. Goldhahn, Appl. Phys. Lett. 102,
052112 (2013).
28
J. A. Freitas, G. C. B. Braga, E. Silveira, J. G. Tischler, and M. Fatemi,
Appl. Phys. Lett. 83, 2584 (2003).
29
J. A. Freitas, Jr., W. J. Moore, B. V. Shanabrook, G. C. B. Braga, S. K.
Lee, S. S. Park, and J. Y. Han, Phys. Rev. B 66, 233311 (2002).
30
A. Wysmolek, K. P. Korona, R. Stȩpniewski, J. M. Baranowski, J.
Bloniarz, M. Potemski, R. L. Jones, D. C. Look, J. Kuhl, S. S. Park, and S.
K. Lee, Phys. Rev. B 66, 245317 (2002).
31
J. A. Freitas, Jr., W. J. Moore, and B. V. Shanabrook, Phys. Rev. B 69,
157301 (2004).
32
L. C. de Carvalho, A. Schleife, and F. Bechstedt, Phys. Rev. B 84, 195105
(2011).
33
M. Feneberg, M. F. Romero, M. R€
oppischer, C. Cobet, N. Esser, B.
Neuschl, K. Thonke, M. Bickermann, and R. Goldhahn, Phys. Rev. B 87,
235209 (2013).
34
P. M. Mooney, G. A. Northrop, T. N. Morgan, and H. G. Grimmeiss,
Phys. Rev. B 37, 8298 (1988).
35
D. J. Chadi and K. J. Chang, Phys. Rev. B 39, 10063 (1989).
20
Downloaded 23 Sep 2013 to 13.1.101.37. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions