Content uploaded by Sumi Bhuyan
Author content
All content in this area was uploaded by Sumi Bhuyan on Jul 20, 2014
Content may be subject to copyright.
Light emission despite doubly-forbidden radiative transitions in AlP/GaP quantum
wells: Role of localized states
Sumi Bhuyan, Richarj Mondal, Pradip Khatua, Mykhaylo Semtsiv, W. T. Masselink, Jean Léotin, Bipul Pal, and
Bhavtosh Bansal
Citation: Journal of Applied Physics 114, 163101 (2013); doi: 10.1063/1.4825328
View online: http://dx.doi.org/10.1063/1.4825328
View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/114/16?ver=pdfcov
Published by the AIP Publishing
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Mon, 23 Dec 2013 12:14:09
Light emission despite doubly-forbidden radiative transitions in AlP/GaP
quantum wells: Role of localized states
Sumi Bhuyan,
1
Richarj Mondal,
1
Pradip Khatua,
1
Mykhaylo Semtsiv,
2
W. T. Masselink,
2
Jean L
eotin,
3
Bipul Pal,
1
and Bhavtosh Bansal
1,a)
1
Indian Institute of Science Education and Research Kolkata, Mohanpur, Nadia 741252, West Bengal, India
2
Department of Physics, Humboldt University Berlin, Newtonstrasse 15, D-12489 Berlin, Germany
3
Laboratoire National des Champs Magn
etiques Puls
es, 143 avenue de Rangueil, 31400 Toulouse Cedex 4,
France
(Received 18 August 2013; accepted 1 October 2013; published online 22 October 2013)
The GaP/AlP/GaP heterostructure has an indirect gap both in real as well as momentum space,
making the first order radiative recombination doubly forbidden. Nevertheless, we have observed
relatively efficient emission from these structures. This paper comprehensively studies the origin of
this improved light emission through a detailed analysis of the photoluminescence (PL) spectra. Our
observations suggest that localized excitons within the acceptor states in GaP close to the
heterostructure interface are enough for efficient light emission in these structures, doing away with
the need for more complicated structures (superlattices or neighboring confinement structures). This
real space localization of holes, close to the interface, apart from increasing the wave function
overlap, also relaxes the delta-function momentum selection rule. Independent experimental evidence
for this assertion comes from (i) the PL spectrum at high excitation power where transitions from
both the localized as well as extended states are independently observed, (ii) the observation that
extended states emission has the expected band-bending-induced blue-shift with increase in excitation
power, whereas the localized states do not, (iii) observation of phonon replicas for PL from localized
states, and (iv) observation of persistent photoconductivity at low temperature. Finally, we propose a
simple analytical model that accounts for both the type-II nature as well as the indirect bandgap to
explain the improvement of radiative recombination efficiency with increased localization. The
experimental observations are reproduced within an order of magnitude. The model is very general
and it also provides a framework to study the optical properties of other such (type-II and/or
indirect gap) heterostructures. V
C2013 AIP Publishing LLC.[http://dx.doi.org/10.1063/1.4825328]
I. INTRODUCTION
Despite the near lattice match, AlP/GaP is among the
least studied of the group III–V heterojunction pairs. This is
because both AlP and GaP have an indirect gap. To make
things bleaker for optoelectronics, the band alignment
between AlP and GaP is type-II; the X-valley electrons (C-
valley holes) see a lower potential in AlP (GaP) [Fig. 1]. As
a result, the overlap of electron and hole wave functions is
not expected to be optimal. Non-optimal overlap would lead
to poor radiative efficiency.
Due to the expected poor emission efficiency, AlP-based
materials have been largely ignored. The little research
reported, mostly in 1990’s, has primarily focused on innova-
tive bandgap engineering ideas which may improve the poor
light emitting efficiency expected of these structures. Most
studies have been made on (AlP)
m
/(GaP)
n
superlattices,
where mand nare the number of monolayers.
1–5
Like for the
Si/Ge system, zone folding is predicted to yield a direct gap
also in AlP/GaP for suitable mand n.
1,6
Relatively, efficient
light emission has indeed been observed from these struc-
tures but conclusive evidence for direct transitions has never
been found. It is now generally believed that the
disorder-related relaxation of the momentum selection rules
and carrier localization are responsible for unexpected bright
emission from these samples. Other ideas for improving light
emission include making the AlP quantum well (QW) only a
few atomic layers thin so that there may be large spillover of
the electron wave function outside the AlP QW and the C-X
valley mixing due to the abrupt heterointerface may lead to a
phonon-free pathway for light emission.
7,8
Furthermore, for
the GaP/AlP/GaP type-II QWs, since the holes in GaP would
effectively be free, neighboring confinement systems to
confine the holes have also been tried.
9–11
Finally, there was
also a report of insertion of an ultrathin AlP QW between
the GaAsP/GaP heterostucture to improve the radiative effi-
ciency of GaP.
12
In most of these studies, a relatively effi-
cient photoluminescence (PL) has been observed which has
been tentatively attributed to carrier localization.
4,11
But this
localization and its origin have never been clearly estab-
lished. It is hence surprising that apart from some very pre-
liminary work,
13
there is no reported luminescence study
on more conventional, wider QWs of AlP/GaP. Without
the added complications of the above mentioned engineering
ideas, wider QWs are perhaps better suited to probe the
nature of localization. In this paper, we have studied
GaP/AlP/GaP multiple QW (MQW) samples of 3, 4, and
5 nm well widths. Through a systematic study of the PL in
a)
Electronic mail: bhavtosh@iiserkol.ac.in
0021-8979/2013/114(16)/163101/6/$30.00 V
C2013 AIP Publishing LLC114, 163101-1
JOURNAL OF APPLIED PHYSICS 114, 163101 (2013)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Mon, 23 Dec 2013 12:14:09
these MQWs, we have attempted to elucidate the primary
mechanism responsible for the relatively efficient light
emission.
II. EXPERIMENT AND OBSERVATIONS
Multiple quantum wells composed of Si-doped AlP wells
and GaP barriers were grown using gas-source molecular-
beam epitaxy on unintentionally doped n-type GaP(001) sub-
strates with n81015 cm3at 300 K.
14
Low temperature
(15 K) PL spectra were measured by nonresonantly exciting
the samples with a 405 nm diode laser with the power inci-
dent on the sample varied between 20 lW and 20 mW. The
spectra were dispersed in a 0.5m monochromator and were
recorded using a silicon charge coupled device with a spectral
resolution of 0.5 meV.
A simplified schematic band diagram of the AlP
QW sandwiched between GaP layers is shown in Fig. 1.
The type-II band alignment makes the transition energy
(2.01–2.05 eV) smaller than the energy gaps of both the
parent compounds. For simplicity, the strain-induced split-
ting of the three-fold degenerate X-valley of AlP is not
shown.
15,16
It is important to note that there will be an elec-
trostatically produced band-bending.
17
This will be espe-
cially important for the holes in the GaP layer. The valence
band holes should observe a carrier density-dependent con-
finement potential which may be approximated by a triangu-
lar well.
17
It was perhaps not realized that this potential
makes the extra QWs created in the neighboring confinement
structures
9–11
redundant. There will also be an accompany-
ing band bending in the AlP QW layer, but since this would
only add a small perturbation to the already present strong
QW confinement potential felt by electrons, the band bend-
ing inside the QW is not shown.
Figure 2shows PL spectra from three different AlP/GaP
MQW samples (well widths 3, 4, and 5 nm). All the spectra
have been scaled to the height of the strongest peak and the
abscissa shifted such that the strongest peak is centered at
zero energy. This clearly shows that all the three samples
have a second peak at approximately the same energy
(45 meV) below the first peak. This energy difference is
close to various longitudinal optical (LO) phonon energies in
GaP, 50 meV at the Cpoint, and about 46 meV at L and X
valleys.
18
Hence, the second lower energy peak is from a
phonon-assisted transition and the first peak is identified as
the no-phonon (NP) peak. Note that the wider QWs have rel-
atively weaker NP peaks. The actual transition energies for
NP peaks are shown in Fig. 2(inset). The systematic blue-
shift of the spectra with the decreasing well width confirms
quantum confinement of carriers and hence our first impor-
tant conclusion is that at least the electrons participating in
PL are indeed from the AlP QWs. This rules out the trivial
defect recombination from GaP barrier layers alone.
Now onwards, we shall mostly be discussing the results
from the 3 nm MQW sample. However, note that the other two
samples (4 and 5 nm MQWs) also showed nearly identical
results. Low temperature PL from the 3 nm MQW sample with
the excitation power varied over three orders of magnitude is
plotted in Fig. 3on a semilogarithmic scale. Phonon replicas of
the no-phonon peak due to longitudinal optical (LO) phonon-
assisted indirect transitions are seen at all excitation powers.
Tentatively, we assign these peaks as A-B(NP) [acceptor-to-
band—NP transition], A-B(LO) [acceptor-to-band—one LO
phonon-assisted transition], and A-B(2LO) [acceptor-to-
band—two LO phonon-assisted transition], respectively. We
shall justify the assignment of acceptor-to-band transitions in
Sec. III. At higher powers, a second peak emerges at slightly
higher energy than the A–B(NP) peak. While the energy
positions of the first three peaks are independent of the exci-
tation power, the new peak, labeled as B–B(NP) [band-to-
band - NP transition] shows a blue-shift with power.
This is further analyzed in Fig. 4, where the power-
dependent PL data are re-plotted in the inset on a linear
scale. Here, the signal is normalized at the A-B(NP) peak for
a clear comparison. It is evident that the energy position of
the A–B peak is virtually constant with excitation power,
whereas the B–B peak shows a clear blue-shift. The energy
FIG. 2. Photoluminescence spectra from three AlP/GaP MWQ samples
measured at 15 K under very low excitation power (20 lW). The spectra
have been scaled such that the prominent no-phonon (NP) peak in different
samples has the same intensity and lies around zero energy. In each of the
three cases, the lower intensity (phonon-assisted) peak is about 45–50 meV
below the NP peak. Note that the intensity of the phonon peak relative to the
NP peak increases with increasing well width. (Inset) The actual energy
position of the NP peak as a function of the well width. The blue-shift in the
NP peak energy with decreasing well width ascertains that the electrons are
indeed those confined in the AlP QWs.
FIG. 1. Schematic band diagram of GaP/AlP/GaP QWs under low optical
excitation. Note that the spatially-indirect optical transition between X-
valley electrons in AlP and holes in GaP occurs at energy smaller than the
bandgaps of both GaP and AlP.
163101-2 Bhuyan et al. J. Appl. Phys. 114, 163101 (2013)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Mon, 23 Dec 2013 12:14:09
of B-B peak plotted in Fig. 4as a function of [power]
1=3
nicely fits a straight line. This is well-known in type-II
systems
19
and arises due to carrier induced band-bending
and resultant nearly triangular potential well (schematically
shown for the valance band in Fig. 1) whose steepness
increases with the excitation power leading to the [power]
1=3
dependence of the blue-shift of PL peak.
17
The temperature dependence of the PL from the same
3 nm MQW sample measured under 5 mW excitation power
is presented on a semi-logarithmic scale in Fig. 5.We
observe a rapid drop in intensity of the three lower energy
A-B peaks with increasing temperature and beyond 80 K,
only the high energy B-B peak can be clearly identified.
We also investigated time-resolved photoconductivity.
The device was made with gold wires bonded on the sample
surface to indium contacts which were annealed to make
direct contact to the underlying QW layer. We measured
in-plane two-probe photoresistance when the sample main-
tained at 15 K was exposed to 532 nm laser light. The time
evolution of photoresistance was recorded as the light source
was switched on and then abruptly switched off after a
steady state in the light-on condition was reached. Time-
resolved photoconductivity was inferred from the photore-
sistance measurement. Figure 6shows the time evolution of
normalized photoconductivity from 3 nm MQW sample. The
decay of the photoconductivity rðtÞafter the abrupt switch-
ing off of light is replotted in Fig. 6(inset). The experimental
data can be fitted
20
to the Kohlrausch stretched exponential
form: rðtÞ¼rð0Þexpðt=sÞbwith b¼0:12 and with a very
large effective decay time s3:5107s at 15 K. Hence,
we have observed persistent photoconductivity.
20
III. DISCUSSION
A. Evidences for hole localization
We shall now try to infer the nature of the light emitting
states from these observations. The first conclusion which
had already been made earlier on the basis of the shift in the
transition energy with well width [Fig. 2(inset)] is that the
electron states are quantized within the QW.
Let us now pay attention to the hole states. Based on ex-
perimental observations stated above, we shall argue in the
following that at low excitation densities, the PL emission
FIG. 4. Energy of the B-B(NP) peak in the 15 K PL spectra from 3 nm
MQW sample is plotted as a function of [power]
1=3
. The solid line is a linear
fit showing the expected [power]
1=3
dependence of blue-shift. Inset: Power
dependent PL spectra normalized at the A-B(NP) peak at 2.05 eV. Note that
there is no energy shift in the A-B(NP) peak with power.
FIG. 5. Temperature dependence of the PL spectra for 3 nm MQW sample
measured under 5 mW excitation power. Note that at higher temperature,
band-to-band recombination dominates over the acceptor-to-band transitions.
FIG. 6. Time evolution of normalized photoconductivity under abrupt switch-
ing on and off of a green (532 nm) laser light exciting the MQW sample at
15 K. Inset: Decay of photoconductivity is fitted to Kohlrausch stretched expo-
nential: rðtÞ¼rð0Þexpðt=sÞbwith b¼0:12 and s3:5107s.
FIG. 3. Photoluminescence spectra from 3 nm MQW sample measured at
15 K as a function of excitation power (20 lW20 mW). The peaks due to
NP acceptor-to-band [A-B(NP)], one LO phonon-assisted acceptor-to-band
[A-B(LO)], two LO phonon-assisted acceptor-to-band [A-B(2LO)], and NP
band-band [B-B(NP)] transitions are marked. The A-B transitions do not
show blue-shift, whereas the B-B(NP) transition is strongly blue-shifted
with increasing excitation power. The peak at 2.2 eV is not identified.
163101-3 Bhuyan et al. J. Appl. Phys. 114, 163101 (2013)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Mon, 23 Dec 2013 12:14:09
involves the localized hole states in GaP leading to the local-
ized hole-to-conduction band transition [A-B(NP)] in Fig. 3.
Strong localization in real space leads to fuzziness in mo-
mentum owing to the position-momentum uncertainty princi-
ple. This facilitates electron-hole wave-function overlap in
momentum space and enhances optical transition probability
in indirect gap systems. This is schematically shown in
Fig. 7. To ensure nonzero wave-function overlap in real
space in the type-II systems, only the hole states localized
very close to the QW interfaces can participate in PL. Such
localization centers are much smaller in number than the
band states. At high excitation powers, these get saturated
and the band-to-band (B-B) emission also becomes visible.
One may obtain a rough estimate of binding energy of these
localized states from the difference in the observed PL
energy of the B-B(NP) and A-B(NP) peaks. This is about
35 meV
21
and is most likely related to a shallow acceptor
state. It is emphasized that unlike the type-I structures, here
the holes cannot be localized by the interface fluctuations
(there is no confinement potential for holes). The extent of
the bound hole wave function, d, can be estimated from the
uncertainty argument, dh=ffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2mhEB
p.
22
With the heavy
hole mass in GaP mh¼0:67m0, where m
0
is the free electron
mass, and localization energy EB35 meV, we get
d1:2 nm, that is about two lattice constant. Such strongly
localized hole states would have a considerable spread in
their momentum states which would support phonon-free
recombination in this indirect gap system. Even the electron
states in these narrow QW samples are localized within the
thin QW region. Stronger localization in narrower QWs leads
to enhanced no-phonon transition relative to the
phonon-assisted transitions compared with wider QWs as is
seen in Fig. 2.
We have observed a large oscillator strength for the pho-
non satellites connected with the A-B transitions, whereas
there are no phonon satellites at all for the B-B peaks. While
in an indirect gap material, phonon-assisted transitions are
expected, these are usually extremely weak because the long
radiative recombination time associated with the second
order pathway has to compete with nonradiative processes.
On the other hand, localization of excitons has an effect of
considerably enhancing oscillator strength of these satellites
due to the increase in the Huang-Rhys factor.
23
Notably, the
absence of phonon satellites in the B-B(NP) peak leads us to
independently conclude that the A-B(NP) peak at 2.045 eV
is due to localized excitons.
No blue-shift was observed for the A-B peaks in Fig. 4,
while B-B peak suffered a characteristic [power]
1=3
blue-shift of type-II transition with increasing excitation
power. Absence of blue-shift of the A-B peak with excitation
power implies that corresponding states are too strongly
localized to be affected by the carrier induced band bending
and resultant triangular potential which quantizes the states
participating in B-B transitions. The temperature dependence
of PL (Fig. 5) where the strength of the B-B(NP) peak is
enhanced with respect to the bound exciton (A–B) peaks at
higher temperature also indicates the “ionization” of the
acceptor-bound localized hole states.
Finally, the observation of persistent photoconductivity
(in the QW plane) is another independent indication of hole
localization. Optical excitation with 532 nm laser nonreso-
nantly creates electron-hole pairs mainly in the GaP layers.
Though some of the photogenerated electrons can be depleted
due radiative and nonradiative recombination with holes in
GaP layers, a fraction of them would diffuse to the AlP QW
layer where the electrons see a lower potential. These QW
electrons will control the in-plane photoconductivity of the
sample. Depletion of these electrons due to recombination
with holes will lead to the decay of photoconductivity. Being
a type-II QW system, electron-hole wave function overlap is
not optimal for AlP/GaP QWs. Only the holes available near
the QW interface can recombine with QW electrons and this
cause the initial rapid decay of the photoconductivity in Fig. 6.
With the electrons trapped in the QW, a biased diffusion of
holes from bulk GaP to the regions close to the QW interface
is expected due to the electrostatic attraction between electrons
and holes. If the holes are localized, this transport of holes
from the bulk GaP to the heterointerface will be an activated
process that is exponentially suppressed at low temperature.
This will make the QW electrons available during a very long
period of time after their photocreation. Thus, localized holes
may lead to persistent photoconductivity. Further study of the
dependence of persistent photoconductivity on the sample
temperature and excitation laser wavelength may be useful for
a more conclusive evidence.
B. Relaxation of momentum selection rule
Let us now attempt a back-of-the-envelope estimate of
the extent of the relaxation of the momentum selection rule
due to the localization of holes. The electrons in AlP are at the
three equivalent (ignoring the lifted degeneracy) X-valleys,
where the wave vectors ðkx;ky;kzÞare the three permutations
of (001). Furthermore, since the electron confinement in the
QW is along the zdirection, if we consider only the (001)
X-valley electrons of AlP then we have direct transitions for
k
x
and k
y
, and we only need to worry about a mechanism for
the relaxation of the momentum conservation rule along the z
direction. The problem may thus be effectively treated in one
dimension, with only the wave functions along the growth
direction zbeing the important consideration.
The X-valley electrons are QW bound states with a finite
momentum hk0along one of the ~
kdirections depending on
FIG. 7. Schematic representation of confined electron states in AlP QW and
localized hole states in GaP in (A) real and (B) momentum space. Spreads in
momentum for electron and hole states are related to the inverse of the cor-
responding localization lengths. Extended hole states would correspond to a
momentum d-function.
163101-4 Bhuyan et al. J. Appl. Phys. 114, 163101 (2013)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Mon, 23 Dec 2013 12:14:09
which of the three X-valleys they belong. For analytic sim-
plicity, let us assume harmonic confinement potentials for
electrons and holes resulting in Gaussian wave functions for
the ground state electrons and holes [Fig. 8(inset)]. The
ground state wave function for X-valley electrons confined
with a confinement length Lwithin the AlP QW thus has the
form
wX
AlPðzÞ¼ðLpffiffiffiffiffiffi
2p
pÞ1
2exp½ðzz0Þ2=L2ik0z:(1)
The expðik0zÞfactor, with k0¼p=0:545 nm1, takes care
of the fact that the electrons in AlP are not at the Brillouin
zone center and z0ð6¼ 0Þaccounts for the type-II nature of
the heterostructure since the center of the electron wave
function is displaced with respect to the center of the hole
wave function. The latter is assumed to be at the origin. The
envelope wave function of the acceptor-bound holes with a
localization length din the zdirection is taken as
wC
GaPðzÞ¼ðdpffiffiffiffiffiffi
2p
pÞ1
2exp½z2=d2:(2)
The transition probability (P) is proportional to the square of
the overlap integral and can be written as
Pð1
1
wX
AlPðzÞwC
GaPðzÞdz
2
(3)
or
P¼A
2pL
dþd
L
0
B
@1
C
Aexp 2z2
0
L2þd2
exp k2
0
21
L2þ1
d2
2
6
43
7
5
0
B
@1
C
A
:(4)
Here, Ais the constant of proportionality. The transition
probability neatly factorizes into three terms and may thus
be expressed as
P¼P1D P2D P1I:(5)
Here, P1D;P2D , and P1I correspond to the terms in the three
parenthesis, respectively, in Eq. (4). We will only have the
first term P1D for a type-I direct gap QW. The additional
terms P2D and P1I signify the reduction of the overlap on
account of the type-II nature and the off-zone center position
of the electrons in the indirect gap AlP.
To make numerical estimate of these effects, we have
taken z0¼Land Lis estimated at 1.2 nm from the confine-
ment energy (33 meV) of the ground state electron in AlP
QW. Numerical evaluation of various terms in Eq. (5) is
plotted in Fig. 8as a function of hole localization length.
Dramatic enhancement in transition probabilities for indi-
rect gap (both type-II and type-I) transitions relative to that
of type-I direct gap transition is seen with decreasing hole
localization length. As expected, the relative transition
probability for type-II direct gap QW increases with
increasing hole localization length. The localization length
for the acceptor-bound hole is estimated to be one-two lat-
tice constant (0.55–1.1 nm) in GaP. These positions are
marked with thick vertical lines in Fig. 8. In this case, an
enhancement of about 10
5
–10
8
is achieved as compared
with the case for unbound holes. Compared with the direct
gap type-I QW, the transition probability for a type-II indi-
rect gap transition is still smaller by a factor of 10
2
–10
5
.
Note that this calculation is rather approximate. The
assumption of Gaussian wave functions leads to an overlap
integral having exponential dependence on various parame-
ters, making the results in Fig. 8extremely sensitive to the
parameter values. While having exponential wave functions
may be more realistic, the corresponding confinement
potentials are no longer simple to model. Harmonic confine-
ment was chosen for analytical simplicity. Experimentally,
the intensity of the PL signal from AlP/GaP was about two
orders of magnitude smaller than that measured from high
quality GaAs/AlGaAs QW.
IV. SUMMARY
We have shown that there can be relatively efficient
light emission from type-II indirect gap GaP/AlP/GaP
MQW structures. The observed efficiency of luminescence
has been attributed to the breakdown of the momentum
selection rule due to the localization of holes by acceptor
states close to the GaP/AlP interface. The hole localization
energy is estimated at 35 meV. Localization of holes is inde-
pendently established through (i) absence of shift of the low
energy no-phonon peak with excitation power, (ii) large in-
tensity of the phonon replica, (iii) temperature dependence
of the PL spectrum, and (iv) observation of persistent photo-
conductivity at low temperature. Furthermore, a true type-II
emission, as evidenced by the characteristic [power]
1/3
dependence of the PL peak energy was also established at
higher excitation power.
FIG. 8. Numerically calculated transition probabilities for type-II direct
(P2D, dashed line), type-I indirect (P1I , dotted line), and type-II indirect
(P2D P1I, solid line) transitions relative to that of type-I direct transition as
a function of hole localization length. Vertical solid lines mark the localiza-
tion length of one and two lattice constants. Inset: Schematic of electron and
hole confinement potentials (assumed to be harmonic potentials) and the
ground state wave functions (Gaussians).
163101-5 Bhuyan et al. J. Appl. Phys. 114, 163101 (2013)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Mon, 23 Dec 2013 12:14:09
ACKNOWLEDGMENTS
B.B. thanks Gottfried D€
ohler for a useful discussion
regarding the selection rules in indirect gap semiconductors.
B.P. thanks DST and INSA, the Government of India for par-
tial financial support.
1
M. Kumagai, T. Takagahara, and E. Hanamura, Phys. Rev. B 37, 898
(1988).
2
H. Asahi, K. Asami, T. Watanabe, S. J. Yu, T. Kaneko, S. Emura, and S.
Gonda, Appl. Phys. Lett. 58, 1407 (1991).
3
A. Sasaki, X.-L. Wang, and A. Wakahara, Appl. Phys. Lett. 64, 2016
(1994).
4
K. Uchida, N. Miura, J. Kitamura, and H. Kukimoto, Phys. Rev. B 53,
4809 (1996).
5
R. K. Soni, S. Tripathy, and H. Asahi, Physica E 21, 131 (2004).
6
C. H. Park and K. J. Chang, Phys. Rev. B 47, 12709 (1993).
7
S. Nagao, T. Fujimori, H. Gotoh, H. Fukushima, T. Takano, H. Ito, S.
Koshihara, and F. Minami, J. Appl. Phys. 81, 1417 (1997).
8
S. Nagao, K. Fujii, T. Fujimori, H. Gotoh, H. Ito, and F. Minami, J. Cryst.
Growth 175/176, 1157 (1997).
9
F. Issiki, S. Fukatsu, and Y. Shiraki, Appl. Phys. Lett. 67, 1048
(1995).
10
N. Usami, F. Issiki, D. K. Nayak, Y. Shiraki, and S. Fukatsu, Appl. Phys.
Lett. 67, 524 (1995).
11
N. Usami, T. Sugita, T. Ohta, F. Issiki, Y. Shiraki, K. Uchida, and N.
Miura, Phys Rev B 60, 1879 (1999).
12
K. Arimoto, T. Sugita, N. Usami, and Y. Shiraki, Phys. Rev. B 60, 13735
(1999).
13
M. Gerhold, K. Kamath, and P. Bhattacharya, Appl. Phys. Lett. 71, 3260
(1997).
14
M. P. Semtsiv, U. M€
uller, W. T. Masselink, N. Georgiev, T. Dekorsy, and
M. Helm, Appl. Phys. Lett 89, 184102 (2006).
15
M. P. Semtsiv, O. Bierwagen, W. T. Masselink, M. Goiran, J. Galibert,
and J. L
eotin, Phys. Rev. B 77, 165327 (2008).
16
M. P. Semtsiv, S. Dressler, W. T. Masselink, V. V. Rylkov, J. Galibert, M.
Goiran, and J. L
eotin, Phys. Rev. B 74, 041303(R) (2006).
17
N. N. Ledentsov, J. B€
ohrer, M. Beer, F. Heinrichsdorff, M. Grundmann,
D. Bimberg, S. V. Ivanov, B. Ya. Meltser, S. V. Shaposhnikov, I. N.
Yassievich, N. N. Faleev, P. S. Kop’ev, and Zh. I. Alferov, Phys. Rev. B
52, 14058 (1995).
18
B. P€
od€
or, Phys. Status Solidi B 120, 207 (1983).
19
B. Pal, K. Goto, M. Ikezawa, Y. Masumoto, P. Mohan, J. Motohisa, and T.
Fukui, Appl. Phys. Lett. 93, 073105 (2008).
20
T. T. Chen, W. S. Su, Y. F. Chen, P. W. Liu, and H. H. Lin, Appl. Phys.
Lett. 85, 1526 (2004).
21
Energy position of the B-B(NP) energy is estimated from the spectra at the
lowest excitation power where it becomes decipherable.
22
B. Bansal, A. Kadir, A. Bhattacharya, and V. V. Moshchalkov, Appl.
Phys. Lett. 93, 021113 (2008).
23
I. Brener, M. Olszakier, E. Cohen, E. Ehrenfreund, A. Ron, and L.
Pfeiffer, Phys. Rev. B 46, 7927 (1992).
163101-6 Bhuyan et al. J. Appl. Phys. 114, 163101 (2013)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
14.139.220.33 On: Mon, 23 Dec 2013 12:14:09
