Content uploaded by S. P. Watkins
Author content
All content in this area was uploaded by S. P. Watkins on Jul 22, 2015
Content may be subject to copyright.
InAs/InAsSb strain balanced superlattices for optical detectors: Material
properties and energy band simulations
D. Lackner, M. Steger, M. L. W. Thewalt, O. J. Pitts, Y. T. Cherng et al.
Citation: J. Appl. Phys. 111, 034507 (2012); doi: 10.1063/1.3681328
View online: http://dx.doi.org/10.1063/1.3681328
View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v111/i3
Published by the American Institute of Physics.
Related Articles
Modeling and analysis of intraband absorption in quantum-dot-in-well
mid-infrared photodetectors
J. Appl. Phys. 111, 033713 (2012)
Waveguide-integrated near-infrared detector with self-assembled metal silicide nanoparticles embedded in a
silicon p-n junction
Appl. Phys. Lett. 100, 061109 (2012)
Impact ionization in quantum well infrared photodetectors with different number of periods
J. Appl. Phys. 111, 034504 (2012)
Effect of overgrowth temperature on the mid-infrared response of Ge/Si(001) quantum dots
Appl. Phys. Lett. 100, 053507 (2012)
Improved photoresponse of InAs/GaAs quantum dot infrared photodetectors by using GaAs1−xSbx strain
reducing layer
Appl. Phys. Lett. 100, 043512 (2012)
Additional information on J. Appl. Phys.
Journal Homepage: http://jap.aip.org/
Journal Information: http://jap.aip.org/about/about_the_journal
Top downloads: http://jap.aip.org/features/most_downloaded
Information for Authors: http://jap.aip.org/authors
Downloaded 14 Feb 2012 to 142.58.122.43. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
InAs/InAsSb strain balanced superlattices for optical detectors:
Material properties and energy band simulations
D. Lackner,
1
M. Steger,
1
M. L. W. Thewalt,
1
O. J. Pitts,
1,a)
Y. T. Cherng,
1,2
S. P. Watkins,
1,b)
E. Plis,
3
and S. Krishna
3
1
Department of Physics, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada
2
4D LABS, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada
3
Center for High Technology Materials, Department of Electrical and Computer Engineering,
University of New Mexico, Albuquerque, New Mexico 87106, USA
(Received 7 June 2011; accepted 7 January 2012; published online 10 February 2012)
InAsSb/InAs type II strain balanced superlattices lattice matched to GaSb have recently been
proposed as an alternative to InAs/(In)GaSb short period superlattices for mid- to long infrared
photodetectors. Photoluminescence data at 4 K of OMVPE grown InAsSb (multi-) quantum wells
in an InAs matrix on InAs and GaSb substrates is presented for Sb compositions between 4% and
27%. The measured transition energies are simulated with a self-consistent Poisson and
Schroedinger equation solver that includes strain and band-offsets. The fitted parameters are then
used to predict the type II transition energies of InAsSb/InAs strain balanced superlattice absorber
stacks at 77 K for different compositions and periods. The optical matrix element was calculated
and compared with InAs/(In)GaSb superlattices. The InAsSb/InAs structures can be designed with
higher or equal matrix elements for longer periods. Finally, the initial optical response data of an
unoptimized strain balanced InAs
0
.
79
Sb
0
.
21
/InAs detector with a 40 nm period are shown. Its cutoff
wavelength is 0.15 eV (8.5 lm), in good agreement with the predicted type II transition energy of
0.17 eV. V
C2012 American Institute of Physics. [doi:10.1063/1.3681328]
I. INTRODUCTION
Recently, strained superlattice (SL) type II infrared
detectors based on InAs/GaSb have been improved
enormously.
1
Detectivities comparable to those of mercury-
cadmium-telluride (MCT), the dominant detector material
for 5–20 lm, have been reported.
2
Operation at room tem-
perature for certain designs and diode arrays has been
developed.
3–5
Part of this progress is due to the extensive
modeling efforts, which have improved over time to include
even Sb segregation on the interfaces.
6–12
Most of this pro-
gress has been based on the use of molecular beam epitaxy
(MBE) growth, as it is very challenging to grow the mono-
layer interfaces needed for short period GaSb/InAs superlat-
tices via organometallic vapor phase epitaxy (OMVPE).
13
OMVPE growth technology is desirable, as it has the
advantage of higher volume production than MBE, and thus
the cost advantage compared to that of MCT technology
would be improved significantly.
Grein et al. proposed InAsSb/InAs superlattices lattice
matched to GaSb substrates as a potential type II alternative
to the InAs/GaSb system for mid-infrared photodetectors.
14
Preliminary modeling of the band structure, detectivity, and
Auger recombination indicated a favorable response at 10 lm;
however, no growth or device fabrication was reported in that
work. This system can be grown strain balanced on GaSb, and
thus it is possible to grow thick absorber stacks without crystal
defects from strain relaxation. There have been a few reports
of the growth of InAsSb/InAs superlattices grown on InAs
substrates for mid-infrared emitter applications; however, this
epilayer–substrate combination is not strain balanced and
therefore is not suitable for thick detector structures. In addi-
tion, various authors have offered several different experimen-
tal and theoretical reports on the band lineups between
InAsSb and InAs. Various conflicting reports of the band line-
ups have been reported, including type I,
15
type IIa (Ref. 16)
(electrons in the InAsSb layers), and type IIb (Ref. 17)(elec-
trons in the InAs layers). Mid-infrared electroluminescence
18
and electrically pumped lasers
19
with InAsSb/InAs active
layers have been demonstrated.
Recently, we explored the growth of InAsSb/InAs super-
lattices strain balanced on GaSb substrates.
20
By means of
photoluminescence measurements and energy band modeling
studies, we were able to confirm that the band alignment for
this material is type IIb, in agreement with theoretical
predictions.
14,17
In order to explore this concept further, in this work we
have grown a series of InAsSb single quantum wells with
InAs barriers, as well as InAsSb/InAs SL structures. These
were measured via photoluminescence (PL) in order to deter-
mine their transition energies, which were also modeled. The
fitted energy band parameters were then used to simulate
possible SL absorber structures for detector applications in
order to guide the design of such devices. The simulation
results are compared with the InAs/GaSb approach. The opti-
cal response of such a strain balanced SL device with a cut-
off below 7 lm at 77 K is presented.
a)
Currently at Canadian Photonics Fabrication Centre, NRC Institute for
Microstructural Sciences, 1200 Montreal Road, Ottawa, Ontario K1A 0R6,
Canada.
b)
Author to whom correspondence should be addressed. Electronic mail:
simonw@sfu.ca.
0021-8979/2012/111(3)/034507/9/$30.00 V
C2012 American Institute of Physics111, 034507-1
JOURNAL OF APPLIED PHYSICS 111, 034507 (2012)
Downloaded 14 Feb 2012 to 142.58.122.43. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
II. EXPERIMENTAL METHODS
The PL samples were grown via OMVPE with an opti-
cal showerhead design on InAs:S (001) and GaSb:Te (001)
wafers with a 2miscut toward (111)B at 500 C, with all
liquid precursors (trimethylindium, trimethylantimony, and
tertiarybutylarsine). Hydrogen was used as a carrier gas with
a flow rate of 3 standard liters per minute (SLM). For the
growth on InAs substrates, the V/III ratio was kept constant
for all samples at 6. The distribution coefficient k¼([Sb]/
[As]
solid
)/([Sb]/[As]
vapor
) was 0.4. This resulted in a growth
rate of 1.1 lm/h. The growth conditions for the samples
grown on GaSb are described elsewhere.
20
On InAs substrates, three series of samples were grown
on an undoped InAs buffer of 20 nm. The single quantum
well (SQW) samples consist of a 20 nm InAsSb layer sand-
wiched between two 100 nm layers of InAs, and the multi-
quantum well (MQW) samples consist of 6 periods of a
20 nm InAsSb and a 20 nm InAs layer grown on an InAs
buffer. The double quantum well (DQW) samples consist of
a buffer, the first quantum well (QW), a 50 nm barrier, the
second QW, and a 100 nm cap. High resolution x-ray diffrac-
tion (HR-XRD) measurements were used to determine the
Sb content and the layer thicknesses, and it was determined
that none of the layers showed strain relaxation. Table I
shows a summary of the samples, including their thicknesses
and the Sb content in the InAsSb layers.
A set of strain balanced InAsSb/InAs SLs was grown on
lattice matched InAsSb buffer layers grown on GaSb sub-
strates. The SL consists of 6 periods of InAs/InAsSb/InAs
layers with a period of 40 nm. The Sb composition of the
InAsSb was varied from 0.14 to 0.27. Again, no relaxation
of the strained layers was detected in x-ray measurements.
A strain-balanced SL device sample was grown on a
GaSb:Te (100) substrate miscut 2toward (111)B, with a
V-III ratio of 6 and a growth rate of 0.3 nm/s at 500 C. A
50 nm InAs
0
.
91
Sb
0
.
09
lattice matched buffer was grown on
the GaSb wafer, followed by a 1 lm thick InAsSb/InAs strain
balanced SL absorber with a Sb composition of 0.21 and a
period of 40 nm. A 100 nm p-doped (Zn: 2 10
18
cm
3
)
InAs
0.91
Sb
0.09
top layer was then grown to form the p
þ
n
junction, as the non-intentionally doped layers are n-type
(typically 2 10
15
cm
3
). The doping concentration of the
substrate was 1 10
18
cm
3
, which forms a tunnel junction
with low n-doped InAsSb, as shown in an earlier work.
21
HR-XRD was employed to verify the layer composition,
thickness, and strain state.
Device fabrication was achieved by using optical photo-
lithography to define 410 lm410 lm square mesa devices
with apertures ranging from 20 to 300 lm. Etching was per-
formed using an inductively coupled plasma reactor with
BCl
3
gas. The sample was physically clamped to the sub-
strate electrode in order to reduce lateral etching. The flow
of He to the backside of the wafer maintained its temperature
at 25 C. Next, Ohmic contacts were evaporated on the
bottom and top contact layers using Ti (50 nm)/Pt (50 nm)/
Au (300 nm) in both cases. The device was fabricated at the
University of New Mexico.
The lattice matched homojunction detector consisted of
1lm of non-intentionally doped InAs
0.91
Sb
0.09
grown on
a GaSb:Te miscut wafer with a 0.1 lm p-doped (Zn:
210
18
cm
3
) top layer. This device was fabricated with a
wet chemical process in 4D LABS at Simon Fraser Univer-
sity. The details of the growth and fabrication can be found
elsewhere.
22
A Bomem DA8 Fourier transform spectrometer with a
liquid nitrogen cooled HgCdTe photodetector (12 lmcutoff
wavelength) and KBr beam-splitter was used to obtain the PL
spectra. The samples were mounted strain-free in liquid He at
4.2 K. PL spectra were collected using frequency doubled
Nd:YVO
4
laser excitation at 532 nm with 200 mW of cw
power in a 3 mm diameter spot size. Blackbody radiation was
rejected by chopping the laser at 3 kHz using a phase sensitive
detector to provide the signal input to the interferometer.
For the optical response measurements of the detector
samples, the HgCdTe detector was replaced with the InAsSb
based detector sample mounted on a leadless ceramic chip
carrier attached to a coldfinger in a dewar with a ZnSe win-
dow, which was cooled to liquid nitrogen temperature. A
globar was used as the IR source. A KBr beam splitter was
used in the interferometer for the SL sample, and a CaF
2
beam splitter was employed for the measurement of the lat-
tice matched device. The diodes were operated in photovol-
taic mode under zero bias conditions.
III. RESULTS AND DISCUSSION
A. Photoluminescence results
Previously, we explored the optical properties of approxi-
mately strain balanced superlattices of InAsSb/InAs on GaSb
substrates.
20
In the present work we expand upon these pre-
liminary results by measuring the optical properties of InAsSb
QWs grown on InAs substrates in order to expand the range
of Sb composition toward the lower side, to provide a consis-
tency check for our detailed energy band simulations.
InAsSb can easily be grown on InAs substrates, and some
challenges, such as the GaSb-InAs(Sb) interface,
23
are
avoided. Due to the lattice mismatch, only thin films or layers
with very little Sb content can be grown before relaxation
occurs. Figure 1(a) shows the 004 x-ray rocking curves of
single quantum well samples (1-5) with increasing Sb content
in the well. The strong pendellosung fringes are a sign of
good structural layer quality and sharp growth interfaces.
TABLE I. Summary of the PL samples grown on InAs substrates. Composi-
tion and thickness values were measured via HR-XRD.
Sample x
Sb
(%) Type t
InAs
Sb (nm) t
InAs
(nm)
1 4 SQW 20 104
2 6 SQW 21 103
3 8 SQW 21 105
4 10 SQW 21 108
5 12 SQW 21 107
6 3 MQW 21 21
7 6 MQW 21 21
8 7.5 SQW 61 199
9 7.5 (5.5) DQW 16, 5 52, 102
10 7.5 DQW 31, 10 52, 102
034507-2 Lackner et al. J. Appl. Phys. 111, 034507 (2012)
Downloaded 14 Feb 2012 to 142.58.122.43. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
Figure 1(a) shows the 004x-ray rocking curve of MQW sam-
ple 6, which consists of 3 periods of 20 nm InAsSb/20 nm
InAs with a Sb composition of 3%. This is compared to a
simulated XRD curve, which was offset for clarity. The meas-
ured fringes are hardly broadened compared to the simulation,
and all pendellosung fringes are clearly visible over a wide
diffraction angle range, which indicates good layer quality
and very low Sb segregation into the InAs wells.
Figure 2(a) shows the measured low temperature PL
spectra for the series of SQW samples. A strong dependence
of the transition energy and linewidth on the Sb content in the
InAsSb QWs is evident, which is typical for spatially indirect
type II transitions. Figure 2(b) shows the PL spectra of the
two MQW structures. The peak positions of the PL lines are
plotted as a function of Sb composition in the InAsSb layers
in Fig. 3. Here the transition energy depends linearly on the
Sb content in the InAsSb layers, as was found by Liu et al.
24
for MBE grown InAsSb (7 nm)/InAs (50nm) MQWs on InAs
substrates. The measured transitions are compared to the cal-
culated bandgap of strained (unstrained) InAsSb, which is
plotted as the dashed-dotted (solid) line. The discrepancy
between the calculated bandgap and the measured PL transi-
tion energy is due to the type II band alignment, as predicted
by theory
17
and experimentally confirmed.
16,20,24
For the simulations, the software nextnano
3
(Ref. 25)was
used, and rectangular quantum wells with periodic boundary
conditions were assumed for the MQW samples. The program
provides a self-consistent solution of the Schrodinger, Pois-
son, and current equations. In order to find the quantization
energies, the carriers were treated within the effective mass
approximation, and the dependence of band offsets with strain
follows the van de Walle treatment,
26
which includes the
effect of strain and temperature on the bandgap. The exciton
binding energy was neglected because it is such a small cor-
rection for InAs (1.4 meV) due to the light electron mass. The
III-V material parameters used to compute the simulations are
the standard parameters found in the work of Vurgaftman,
27
with the exception of the InAsSb split-off band bowing
FIG. 1. (Color online) HR-XRD spectra of the 004 reflection. (a) Measured
data of SQW samples 1-5. The scans are offset for clarity. (b) Comparison
between measured data of MQW sample 6 and a simulation.
FIG. 2. (Color online) The 4 K PL spectra of InAsSb layers grown on InAs
and labeled according to the Sb content in the QWs. (a) A series of samples
consisting of one SQW with a width of 20 nm embedded in InAs. (b) Six pe-
riod SL structures with InAs and InAsSb layer thicknesses of 20 nm per
40 nm period.
034507-3 Lackner et al. J. Appl. Phys. 111, 034507 (2012)
Downloaded 14 Feb 2012 to 142.58.122.43. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
parameter and the valence band bowing. The InAsSb split-off
band bowing parameter was set to zero according to the work
of Cripps et al.
28
The vacuum energy levels of the valence
bands of the bulk materials, which are used to calculate the
band offsets, were taken from the work of Wei and Zunger.
29
All relevant material parameters are tabulated in Tables II.
The second parameter that is not standard is the distribution
of the bowing between the conduction and valence bands.
Assuming that all the bandgap bowing occurs in the conduc-
tion band, as suggested by Wei and Zunger,
17
leads to too
shallow a slope of the energy dependence on Sb content com-
pared with our results. This was also found by Liu et al.,
24
who assumed 40% bowing in the conduction band and 60%
bowing in the valence band. In our work, 30% bowing in the
conduction band and 70% in the valence band was needed in
order to get satisfactory agreement between measurement and
simulation of the data for SQW and MQW samples on InAs
substrates. The simulated energy and the measured PL peaks
are plotted in Fig. 3.
The same model and parameters were then applied to a
series of 6 period superlattice InAsSb/InAs samples with a
40 nm period grown on GaSb substrates (published earlier
20
).
The Sb composition in the InAsSb layers ranged from 0.14
to 0.27. Again, the simulated and measured transition ener-
gies are in good agreement, as shown in Fig. 4. Thus, with
one set of parameters, the type II transition energy at 4 K can
be simulated for Sb compositions in the InAsSb layers
between 4% and 27% on InAs and GaSb substrates.
The type-IIb alignment is in agreement with theory
14,17,26
but in disagreement with certain experimental results.
15,16
FIG. 3. (Color online) 4 K PL peak energies of InAsSb layers grown on
InAs are plotted as a function of Sb content in the QWs. The symbol ($)
denotes the SQW (MQW) samples. The InAsSb unstrained bandgap is dis-
played as a solid line. The dashed-dotted line shows the InAsSb bandgap
strained on the InAs substrate. The dotted (dashed) line is the result of the
type IIa transition energy simulation for the case of the SQW (MQW)
samples.
TABLE II. Parameters used for simulations. Parameters are taken from the
work of Vurgaftman (Ref. 27) unless otherwise stated. The bandgap (E
g
)as
a function of temperature (T) is given in the Varshni form: E
g
(T)¼E
g
aT
2
/(Tþb). cdescribes the linear expansion coefficient of the lattice param-
eter. Note that parameters such as band gap and carrier masses are regarding
the C-point.
InAs InSb GaSb
a(A
˚) 6.0583 6.479 6.0959
c/(10
5
A
˚/K (T-300 K)) 2.75 3.48 4.72
E
g
(eV) 0.417 0.235 0.812
a(meV/K) 0.276 0.32 0.417
b(K) 93 170 140
E
v,vac
(eV) 1.39
a
1.75
a
1.77
a
D
so
(eV) 0.39 0.81 0.76
m
e0.026 0.0135 0.039
m
hh 0.41
b
0.405
c
0.34
c
m
lh 0.026
b
0.016
b
0.045
b
a
c
(eV) 6.66
a
6.4
a
9.33
a
a
v
(eV) 1.0 0.31
d
1.32
a
b (eV) 1.8 2.0 2.0
d (eV) 3.6 4.7 4.7
c
11
(GPa) 832.9 684.7 884.2
c
12
(GPa) 452.9 373.5 402.6
E
p
(eV) 21.5 23.3 27
a
Wei et al. (Ref. 29).
b
Yu and Cardona (Ref. 31).
c
Landolt-Bo¨rnstein (Ref. 32).
d
Qteish et al. (Ref. 33).
TABLE III. Bowing parameters of ternary alloys used for simulations. Only
the non-zero bowing coefficients (c
bow
)definedasY(ABC)¼xY
AB
þ
(1 x)Y
AC
x(1 x)c
bow
are displayed. Parameters are taken from the work
of Vurgaftman (Ref. 27) unless otherwise stated.
InAsSb InGaSb
E
g
(eV) 0.67 0.415
E
v,vac
(eV) 0.47
a
D
so
(eV) 0.0
b
0.1
m
e0.035 0.009
m
lh 0.011
a
Present work.
b
Cripps et al. (Ref. 28).
FIG. 4. The 4 K PL peak energies are plotted as a function of Sb content for
the InAsSb QWs (). Also included are the calculated bandgaps of
unstrained InAsSb (solid line) and InAsSb strained on GaSb (dashed-dotted
line), as well as the predicted type IIb transition energies (dashed line).
Good agreement with the measured data was achieved by assuming that
30% of the InAsSb bandgap bowing occurs in the conduction band, and
70% in the valence band.
034507-4 Lackner et al. J. Appl. Phys. 111, 034507 (2012)
Downloaded 14 Feb 2012 to 142.58.122.43. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
Type-IIb alignment is also consistent with the fact that the
electron affinity of InSb is accepted to be 300 meV less than
that of InAs.
30
In order to test our model assumptions, we have grown a
series of InAs
0.925
Sb
0.075
QWs within an InAs matrix. The
quantum well widths are between 5 nm and 60 nm. In the
assumed type IIb case, the confinement of the heavy holes is
modified, and thus very little dependence of the transition
energy QW widths above 10 nm is expected. The measured
PL energies are plotted in comparison with the model predic-
tions (solid line) in Fig. 5, and the predicted behavior is seen.
For comparison, the dashed line shows the expected depend-
ence of the transition energies in the case of the opposite band
lineup. These data strongly suggest that the electrons are
within the InAs layers, consistent with a type IIb band lineup.
The measured PL energy of the 5 nm QW cannot be simulated
by any parameter set that is in good agreement with the data
points between 10 nm and 60 nm QW widths when assuming
the same composition. Thus we conclude that the composition
must have changed to x
Sb
¼0.055 for the 5 nm sample. In that
case, it is in agreement with the model. Unfortunately, HR-
XRD is not sensitive to a small composition change in such a
thin QW. The HR-XRD measurement of DQW sample 9
(5 nm and 16 nm) is shown in Fig. 6in comparison with two
simulation assumptions: constant composition for both QWs
(top) and different compositions (bottom). The composition
shift has very little effect on the spectra; however, the simula-
tion on the bottom does give a slightly better fit, which is in
agreement with the above assumption.
B. Superlattice modeling
As seen from the PL results (Figs. 3and 4), the lowest
optical transition energy is strongly dependent on the Sb
composition in the InAsSb layers, which is understood by
assuming a spatially indirect type II transition. Because the
electrons in the conduction band are confined in the InAs
layer, whereas the (heavy) holes are confined to the InAsSb
layer, the transition occurs only where the probability distri-
bution functions overlap (see Fig. 7). This overlap can be
increased by shortening the period of the superlattice, as
shown in Fig. 8, which also results in an increase of the tran-
sition energy.
In order to predict the transition energy and, as a figure
of merit, the optical matrix element of a strain balanced
superlattice of a specific Sb composition and period, a num-
ber of simulations were calculated with nextnano
3
at 77 K.
FIG. 5. (Color online) Measured PL energy () as a function of InAs
0.925
Sb
0.075
QW width on InAs substrates compared with simulation. The solid
(dashed) line is computed with a type IIb (IIa) alignment in which the heavy
holes (electrons) are confined in the well. The fact that there is very little
change in confinement for a QW width <10nm strongly suggests the type IIb
alignment. The inset shows a sketch of the two different band alignment
cases.
FIG. 6. (Color online) HR-XRD (004) measurement of sample 9 compared
to two simulations: (top) assuming both InAsSb QWs (16 nm, 5 nm) have a
Sb content of 0.075, and (bottom) different compositions with the Sb content
of the 5 nm QW changed to 0.055. Unfortunately, XRD measurement is not
very sensitive to a small composition change in a single quantum well of
this width; however, the bottom simulation does appear to be closer to the
measured data. The spectra are offset for clarity.
FIG. 7. (Color online) Band structure simulation of an InAs QW in InAsSb
strained on GaSb with periodic boundary conditions at 77 K. The Sb compo-
sition in the InAsSb layers is 0.21. This structure is strain balanced on a
GaSb substrate. The dotted lines are the 1st quantized energy levels of the
electron confined in the InAs layer and the heavy hole in InAsSb. W
2
is plot-
ted so that the zero of probability coincides with the corresponding quan-
tized energy level.
034507-5 Lackner et al. J. Appl. Phys. 111, 034507 (2012)
Downloaded 14 Feb 2012 to 142.58.122.43. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
The same parameters that successfully predicted the PL tran-
sitions at 4 K were used, and the carriers were again treated
within the effective mass framework. No temperature de-
pendence was assumed for the bandgap bowing parameter.
Figure 9(a) shows the lowest transition energy (e-hh) as
a function of Sb composition in the InAsSb layers for SL
periods between 40 nm and 5 nm. Note that all calculated SL
structures are strain balanced on GaSb. Strain balancing is
achieved by setting the average lattice parameter of one pe-
riod weighted with the layer thickness equal to the lattice pa-
rameter of GaSb. Thus the InAsSb layer thickness (t
InAsSb
)
as a function of the Sb composition (x
Sb
) and SL period (P)
can be calculated by
tInAsSb ¼aGaSb aInAs
aInSb aInAs
P
x
¼0:090 P
x
:(1)
With increasing Sb composition or longer periods, the car-
rier overlap decreases, and thus the transition probability is
reduced accordingly. The transition probability is propor-
tional to the square of the optical matrix element per period
(M
p
). This is used as a figure of merit as a design guide and
for comparison to the InAs/(In)GaSb short period SL sys-
tem. The optical matrix element (M) can be expressed as
M¼ðW
eðrÞð^
pÞWhhðrÞd3r;(2)
where is the unit vector of polarization and ^
pis the momen-
tum operator.
The electron (and hole) wave function can be expressed
as Bloch states with the standard Bloch function (U
cke
) mul-
tiplied by an envelope function F
e
(z) for the particle in the
well:
34
We¼FeðzÞUcke:(3)
Within the so-called envelope-function approximation, the
matrix element is then approximated by
8
Mphcbjpjhhibulk ðP=2
P=2
F
hhðzÞFeðzÞdz;(4)
where hcbjpjhhi
bulk
is the bulk optical matrix element.
Because the heavy holes are highly localized in the InAsSb
wells, as opposed to the electron wave functions, which have
considerable leakage into the barriers (see Fig. 8), only the
bulk optical matrix element of InAsSb was chosen for the
FIG. 8. (Color online) Probability distribution for electrons and heavy holes
in InAsSb/InAs SL structures on GaSb substrates at 77 K for a Sb content in
InAsSb of 0.21 and periods of 40 nm (a), 20 nm (b), and 10 nm (c). Periodic
boundary conditions were chosen so as to simulate the SL. The probability
distribution functions are normalized to one over one period. With decreas-
ing period, the wavefunction overlap strongly increases. While electrons
(holes) are confined in the InAs (InAsSb), only the electron wavefunction
penetrates significantly into the barrier.
FIG. 9. Simulation results for InAsSb/InAs strain balanced SL structures on
GaSb at 77 K. (a) Expected transition energy as a function of Sb composition
in InAsSb for different SL periods. (b) Calculated squared optical matrix
elements for one period as a function of Sb composition in InAsSb for dif-
ferent SL periods. m
e
denotes the mass of an electron.
034507-6 Lackner et al. J. Appl. Phys. 111, 034507 (2012)
Downloaded 14 Feb 2012 to 142.58.122.43. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
approximation in Eq. (4). It was linearly interpolated from
the values of InAs and InSb reported by Vurgaftman et al.,
27
which are displayed in Tables II and III.
The results of the simulations are presented in Fig. 9(a),
which shows the predicted transition energy at 77 K as a
function of Sb composition for different periods. With
decreasing period, more Sb is needed in the InAsSb layers in
order to achieve a certain cutoff wavelength. For a period of
5 nm, the lowest transition energy that can be reached for a
Sb composition <50% is 0.21 eV (6 lm); for a 7 nm period
it is 0.12 eV (10 lm), and for a 10 nm period it is <0.06 eV
(>20 lm).
The transition probability, however, is proportional to
the optical matrix element squared, which is plotted in
Fig. 9(b). It clearly shows that by decreasing the period from
40 nm (which is the period of the detector presented below)
to 20 nm, M2
Pcan be increased by a factor of 10. A further
decrease of the period to 10 nm would lead to another
increase by a factor of 5 of the value of M2
P.
The calculation of M2
Palso makes it possible to compare
the expected performance of InAsSb/InAs type II strain bal-
anced superlattice detectors to that of the more common
InAs/(In)GaSb superlattice detectors grown on GaSb.
The growth and control of InAs/GaSb interfaces and
thin layers via OMVPE (Ref. 13) is challenging, and thus
these are usually achieved via MBE. The simulations in
Fig. 10 do not include interface layers and are only for the
case of equal layer thicknesses of InAs and (In)GaSb at 77 K
grown on GaSb (001). In Fig. 10(a), the transition energy
is plotted as a function of period length for InAs/GaSb,
InAs/In
0.2
Ga
0.8
Sb, and InAs/In
0.4
Ga
0.6
Sb superlattices.
Figure 10(b) shows the squared optical matrix element of
one period as a function of the SL period. Again the optical
matrix element was approximated by the product of the over-
lap integral and the optical matrix element of bulk (In)GaSb.
As examples, two cutoff wavelengths (5 lm and 10 lm)
were selected for comparison regarding their squared optical
matrix elements in Table IV. For the InAsSb system, the Sb
content required in the InAsSb layers for different SL peri-
ods is also displayed; for the InAs/GaSb system, the SL pe-
riod needed, assuming equal thicknesses of layers of InAs
and (In)GaSb, is tabulated. These two systems can now be
compared with regard to the optical matrix element. InAsSb/
InAs can be designed with a larger optical matrix element
and a longer period for both cutoff wavelengths relative to
the more common InAs/GaSb approach. Also, the freedom
to adjust the composition and period in order to target a cer-
tain wave-length provides further design flexibility to tune
the energy positions of the bands and minimize Auger
recombination rates. In addition, InAsSb/InAs can easily be
grown strain balanced, and thus it is easier to realize with the
OMVPE growth technique. This makes InAsSb/InAs strain
balanced SL structures a promising approach for infrared
detection based on III-V technology in the wavelength range
between 5 lm and 20 lm.
FIG. 10. Calculated transition energy (a) and optical matrix element for one
period (b) of InAs/(In)GaSb SL structures at 77 K as a function of period.
InAs and (In)GaSb layers are of the same thickness.
TABLE IV. Comparison of InAsSb/InAs with InAs/(In)GaSb SL structures for the case of a detector cutoff at 10 lm and 5 lm at an operating temperature of
77 K.
InAsSb/InAs InAs/(In)GaSb
Period (nm) x
In
in InGaSb
Target k5 7 10 20 0 0.2 0.4
10 lmx
Sb
- 0.49 0.35 0.28 P (nm) 10.5 9 7.5
(2/m
e
)M2
P(eV) - 7.5 5.9 2.3 (2/m
e
)M2
P(eV) 1.7 3.3 6.1
5lmx
Sb
0.28 0.21 0.18 0.16 P (nm) 6.7 5.4 4.2
(2/m
e
)M2
P(eV) 14 12.5 10.8 6.8 (2/m
e
)M2
P(eV) 6.6 10.5 14
034507-7 Lackner et al. J. Appl. Phys. 111, 034507 (2012)
Downloaded 14 Feb 2012 to 142.58.122.43. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
Figure 11 shows the uncalibrated optical response of the
strain balanced SL detector (black) and a comparable lattice
matched structure (gray). The spectra were normalized by
the system response as measured with a HgCdTe photo-
conductive detector. The SL detector shows, as predicted, a
significant sub-bandgap signal. Its cutoff lies just below the
predicted type II transition at 170 meV (7 lm). This cut-off
energy is much lower than what can be reached by the lattice
matched InAsSb composition (280 meV or 3.9 lm).
The response due to the lowest type-II transition is sig-
nificantly weaker than that for the InAsSb direct transition
due to the small matrix element of such a relatively large pe-
riod SL [see Fig. 9(b)]. Also below its bandgap is the type II
transition between the electrons in the InAs and light holes
in the InAsSb, which exhibits an overlap roughly 4 times
larger than the lowest type II (e-hh) transition. In order to
increase the low energy signal, the period of the SL needs to
be decreased; i.e., a reduction from 40 nm to 20 nm should
result in an intensity increase of an order of magnitude [see
Fig. 9(b)]. The devices were measured under zero bias, as
they suffer from a surface accumulation layer in the InAsSb
that shunts the pn-junction and leads to large leakage cur-
rents under bias conditions. This can be avoided by means of
either passivation of the surface or changing to an InPSb het-
erojunction design.
22,35
This improves the diode characteris-
tics, and the higher bandgap layer behaves as barrier, which
should reduce the noise in addition to improving the carrier
collection. Overall, this unoptimized SL diode serves as
proof of principle, and the measured data are in good agree-
ment with the predictions.
IV. CONCLUSIONS
In this work we have presented InAsSb QWs and InAs/
InAsSb strain-balanced SL structures up to 1 lm in thickness
grown via OMVPE without strain relaxation on InAs and
GaSb wafers with InAsSb compositions in the range of
0.04 x
Sb
0.27. The 4 K PL from such structures was
measured, and the transition energies found were compared
to those from simulations made with the nextnano
3
code
using a simple effective mass approach for the carriers.
Good agreement between theory and experiment is found by
assuming that 30% of the band-gap bowing parameter occurs
in the conduction band, and 70% in the valence band. Further
experimental evidence is gathered in order to support the the-
oretical prediction of the exact nature of the band lineup.
The dependence of the PL energy on the InAsSb QW width
supports a type IIb lineup, in which the InAsSb conduction
band is above the InAs conduction band. The same set of pa-
rameters was then used to simulate the type II transition
energy and transition matrix elements for a variety of InAsSb
compositions and periods of InAsSb/InAs strain balanced SL
absorber stacks on GaSb at 77 K. Due to the lower band dis-
continuities between InAsSb and InAs compared to InAs/
GaSb, higher matrix elements can be achieved for the same
transition energies at longer period lengths. As proof of con-
cept, uncalibrated optical response data of an unoptimized
InAs
0.79
Sb
0.21
/InAs strain balanced SL detector are presented
with a cutoff 20 meV below the predicted transition energy
of 170 meV (7 lm). We consider the InAsSb/InAs strain bal-
anced SLs to be a promising alternative material system for
infrared detection in the range of 2 lmto20lm by III-V
materials, which is much more suitable for high volume pro-
duction via OMVPE than the InAs/(In)GaSb system.
ACKNOWLEDGMENTS
The authors acknowledge the support of the Natural
Sciences and Engineering Research Council of Canada and
Perkin Elmer Optoelectronics, Vaudreuil, Quebec. The
authors would like to thank S. Birner for making the simula-
tion tool nextnano
3
available and Dr. G. Kirczenow for help-
ful discussions.
1
A. Rogalski, J. Antoszewski, and L. Faraone, J. Appl. Phys. 105, 091101
(2009).
2
E. H. Aifer, J. G. Tischler, J. H. Warner, I. Vurgaftman, W. W. Bewley,
J. R. Meyer, J. C. Kim, L. J. Whitman, C. L. Canedy, and E. M. Jackson,
Appl. Phys. Lett. 89, 053519 (2006).
3
H. Mohseni, J. Wojkowski, M. Razeghi, G. Brown, and W. Mitchel, IEEE
J. Quantum Electron. 35, 1041 (1999).
4
E. Plis, S. J. Lee, Z. Zhu, A. Amtout, and S. Krishna, IEEE J. Sel. Topics
Quantum Electron. 12, 1269 (2006).
5
R. Rehm, J. Schmitz, J. Fleißner, M. Walther, J. Ziegler, W. Cabanski, and
R. Breiter, Phys. Status Solidi C 3, 435 (2006).
6
L. L. Chang, G. A. Sai-Halasz, L. Esaki, and R. L. Aggarwal, J. Vac. Sci.
Technol. 19, 589 (1981).
7
C. Mailhiot and D. L. Smith, J. Vac. Sci. Technol. A 7, 445 (1989).
8
J. Roslund and T. Andersson, Superlattices Microstruct. 16, 77 (1994).
9
C. H. Grein, P. M. Young, M. E. Flatte, and H. Ehrenreich, J. Appl. Phys.
78, 7143 (1995).
10
G. Dente and M. Tilton, Phys. Rev. B 66, 165307 (2002).
11
Y. Wei and M. Razeghi, Phys. Rev. B 69, 085316 (2004).
12
S. B. Rejeb, M. Debbichi, M. Said, A. Gassenq, E. Tournie, and P. Christol,
J. Phys. D: Appl. Phys. 43, 325102 (2010).
FIG. 11. Optical response of a strain balanced InAs
0.79
Sb
0.21
/InAs SL detec-
tor with a period of 40 nm. The circular detector aperture has a diameter of
200 lm. The vertical dashed lines are the positions of the calculated lowest
four energy transitions. The significant sub-bandgap response is due to the
type II transitions. Due to the low matrix element (see Fig. 9), resulting from
the long period, the type II response is very weak, as predicted. For compari-
son, in gray, we present the response of an InAs
0.91
Sb
0.09
n-p
þ
homojunction
detector lattice matched to GaSb (LM) with a 1 lm thick absorber layer. The
diameter of the circular aperture of this detector is 300lm. Both measure-
ments were done under zero bias condition at 77K. The simulated transitions,
shown as dashed vertical lines, include strain and quantization energies in all
cases.
034507-8 Lackner et al. J. Appl. Phys. 111, 034507 (2012)
Downloaded 14 Feb 2012 to 142.58.122.43. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
13
Y. Huang, J.-H. Ryou, R. Dupuis, V. D’Costa, E. Steenbergen, J. Fan,
Y.-H. Zhang, A. Petschke, M. Mandl, and S.-L. Chuang, J. Cryst. Growth
314, 92 (2011).
14
C. H. Grein, M. E. Flatte, and H. Ehrenreich, Electrochem. Soc. Proc.
95–28, 211 (1995).
15
S. R. Kurtz and R. M. Biefeld, Appl. Phys. Lett. 66, 364 (1995).
16
P. J. P. Tang, M. J. Pullin, Y. B. Li, C. C. Phillips, R. A. Stradling, S. J.
Chung, W. T. Yuen, L. Hart, D. J. Bain, and I. Galbraith, Appl. Phys. Lett.
69, 2501 (1996).
17
S. H. Wei and A. Zunger, Phys. Rev. B 52, 12039 (1995).
18
P. J. P. Tang, M. J. Pullin, S. J. Chung, C. C. Phillips, R. A. Stradling,
A. G. Norman, Y. B. Li, and L. Hart, Semicond. Sci. Technol. 10,1177
(1995).
19
S. R. Kurtz, R. M. Biefeld, A. A. Allerman, A. J. Howard, M. H.
Crawford, and M. W. Pelczynski, Appl. Phys. Lett. 68, 1332 (1996).
20
D. Lackner, O. J. Pitts, M. Steger, A. Yang, M. L. W. Thewalt, and S. P.
Watkins, Appl. Phys. Lett. 95, 081906 (2009).
21
D. Lackner, M. Martine, Y. T. Cherng, M. Steger, W. Walukiewicz, M. L.
W. Thewalt, P. M. Mooney, and S. P. Watkins, J. Appl. Phys. 107, 014512
(2010).
22
D. Lackner, O. J. Pitts, M. Martine, Y. T. Cherng, P. M. Mooney, M. L.
W. Thewalt, E. Plis, and S. P. Watkins, in 22nd International Conference
on Indium Phosphide and Related Materials (IPRM) (IEEE, New York,
2010).
23
D. Lackner, O. J. Pitts, S. Najmi, P. Sandhu, K. L. Kavanagh, A. Yang, M.
Steger, M. L. W. Thewalt, Y. Wang, D. McComb, C. R. Bolognesi, S. P.
Watkins, J. Cryst. Growth 311, 3563 (2009).
24
P.-W. Liu, G. Tsai, H. H. Lin, A. Krier, Q. D. Zhuang, and M. Stone,
Appl. Phys. Lett. 89, 201115 (2006).
25
See http://www.nextnano.de/nextnano3/ for more information about the
nextnano
3
software.
26
C. G. Van de Walle, Phys. Rev. B 39, 1871 (1989).
27
I. Vurgaftman, J. R. Meyer, and L. R. Ram-Mohan, J. Appl. Phys. 89,
5815 (2001).
28
S.A.Cripps,T.J.C.Hosea,A.Krier,V.Smirnov,P.J.Batty,Q.D.Zhuang,
H. H. Lin, P. W. Liu, and G. Tsai, Appl. Phys. Lett. 90, 172106 (2007).
29
S. Wei and A. Zunger, Appl. Phys. Lett. 72, 2011 (1998).
30
S. Adachi, Properties of Group-IV, III-V, II-VI Semiconductors (John
Wiley & Sons Ltd., West Sussex, England, 2005).
31
P. Yu and M. Cardona, Fundamentals of Semiconductors (Springer, New
York, 2001), 3rd ed.
32
Landolt-Bo¨rnstein—Physics of Group IV Elements and III-V Compounds,
edited by O. Madelung (Springer, New York, 1982), Vol. 17.
33
A. Qteish and R. J. Needs, Phys. Rev. B 45, 1317 (1992).
34
C. Weisbuch and B. Vinter, Quantum Semiconductor Structures (Aca-
demic, San Diego, CA, 1991).
35
O. J. Pitts, D. Lackner, Y. T. Cherng, and S. P. Watkins, J. Cryst. Growth
310, 4858 (2008).
034507-9 Lackner et al. J. Appl. Phys. 111, 034507 (2012)
Downloaded 14 Feb 2012 to 142.58.122.43. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions