Content uploaded by Jun Shao
Author content
All content in this area was uploaded by Jun Shao on Jun 18, 2014
Content may be subject to copyright.
Competition of compressive strain with substrate misorientation
in CuPt-type ordered GaInP/AlGaInP quantum wells
Liangqing Zhu, Jun Shao,
a兲
Xiang Lü, Shaoling Guo, and Junhao Chu
National Laboratory for Infrared Physics, Shanghai Institute of Technical Physics, Chinese Academy
of Sciences, 200083 Shanghai, China
共Received 2 June 2010; accepted 9 November 2010; published online 6 January 2011兲
Temperature-dependent photoluminescence 共PL兲 measurements are carried out on lattice-matched
and compressively strained Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P quantum wells 共QWs兲 with CuPt-type
long-range 共LR兲 ordering. Experimental data show that compressive strain and substrate
misorientation of the QWs affect the degree of LR ordering. The compressive strain competes with
the misorientation of 6° off toward 关111兴
A
共denoted as 6 °A兲 significantly. It not only affects the
distribution of domains with different degree of LR ordering in the x−y plane but also introduces
fluctuation of the degree of LR ordering along the z direction of the QWs, which in turn causes the
splitting of PL peaks. A phenomenological model is proposed to account for the experimental
phenomena based on the principle of minimum total free energy. The results suggest that 6 °A
misorientation should not be preferable for compressively strained Ga
x
In
1−x
P QWs with LR
ordering. © 2011 American Institute of Physics. 关doi:10.1063/1.3525586兴
I. INTRODUCTION
Ga
x
In
1−x
P is a good semiconductor material for laser
diodes
1
and solar cells,
2,3
in which spontaneous CuPt-type
long-range 共LR兲 ordering
4,5
is an interesting and important
phenomenon. The LR ordering manifests distinct effects of
band-gap reduction and valence-band splitting.
6–9
For bulk
GaInP, the degree of CuPt-type LR ordering,
, depends on
the growth parameters of, e.g., temperature, growth rate, and
substrate orientation.
10
For low dimensional device applica-
tions, on the other hand, adjusting the Ga:In ratio of
Ga
x
In
1−x
P is a common vehicle. Different Ga:In ratio may
result in different degree of strain as well as strain-induced
lateral ordering in the Ga
x
In
1−x
P epitaxial layer,
11,12
which in
turn modifies the electronic band structure of Ga
x
In
1−x
P ac-
cording to the variation in model-solid theory taking LR or-
dering into account.
8,13
It is textbook knowledge that for a laser diode, both
compressive and tensile strain can reduce the threshold cur-
rent density,
14
and the misorientation of substrate affects the
degree of the LR ordering and can hence change the band
alignment. While numerous efforts have been focused on the
formation mechanism of LR ordering and the misorientation
effect on the degree of LR ordering in lattice-matched GaInP
alloy in the past several years,
15,16
less attention has been
paid to possible interaction of strain with LR ordering in
strained Ga
x
In
1−x
P quantum well 共QW兲 systems.
17–19
It is
therefore curious if the interaction is a key factor with re-
spect to the optical properties of the material systems, and if
a particular misorientation of substrate is optimal for com-
pressively strained QWs with possibly low LR ordering?
In this work, temperature-共11–300 K兲 and in x−y plane
position-dependent photoluminescence 共PL兲 measurements
are carried out on lattice-matched and compressively strained
Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs samples with CuPt-
type LR ordering. The results show that in the compressively
strained QWs samples different degrees of PL-peak splitting
occur for different substrate misorientations, which indicates
an obvious competition effect of the compressive strain and
substrate misorientation on the LR ordering. A phenomeno-
logical model is proposed based on the principle of minimum
total free energy, which can account for the mechanism of
the competition reasonably.
II. EXPERIMENT DETAILS
Seven Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs samples
were used in this study, which were prepared by metal-
organic vapor-phase epitaxy 共MOVPE兲 at a temperature of
700 °C.
16
They were sorted into three series, of which series
I 共lattice matched, x = 0.52 and y =0.52兲 and series II 共com-
pressively strained, x=0.40 and y = 0.76兲 each contain three
samples and were grown on Si-doped 共001兲 GaAs substrate
with the misorientations of 0°, 6° off toward 关111兴
A
共denoted
as 6 °A hereafter兲, 6° off toward 关111兴
B
共denoted as 6 ° B
hereafter兲, respectively, with a nine-period of 10-nm-thick
Ga
x
In
1−x
P/ 4 nm-thick 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs structure
and a 2-nm-thick GaInP capping layer, and series III contains
only one sample with 6 ° A misorientations and is similar to
series II samples in the QWs structure but with a 410-nm-
thick zinc-doped GaP capping layer.
For temperature-dependent PL experiments, a Fourier
transform infrared 共FTIR兲 spectrometer-based PL system was
set up.
20
The spectrometer 共Bruker IFS66v/S兲 was equipped
with a Si photodiode detector and worked in so-called
continuous-scan mode
18
rather than step-scan mode.
21,22
An
Ar
+
-ion laser configured at an output of 514.5 nm laser line
was employed. The exciting power density was about
10 W / cm
2
. A spectral resolution of about 1.5 meV 共or
equivalently 12 cm
−1
兲 was warranted in the PL measure-
a兲
Electronic mail: jshao@mail.sitp.ac.cn.
JOURNAL OF APPLIED PHYSICS 109, 013509 共2011兲
0021-8979/2011/109共1兲/013509/6/$30.00 © 2011 American Institute of Physics109, 013509-1
ments. A close-cycle cryostat 共Oxford 1104V兲 was used to
warrant the sample’s temperature in a range of about 11–300
K.
III. RESULTS AND DISCUSSION
Figure 1共a兲 illustrates PL spectra of the series I lattice-
matched QWs samples with different misorientations at 11
K. All the PL spectra are of single-peak structure with rela-
tively narrow peak width. The peak energy is different from
each other, indicating a considerable difference in the degree
of LR ordering.
8,23,24
Figure 1共b兲 shows the PL spectra of the series II com-
pressively strained QWs samples with different misorienta-
tions at 11 K. Similar to the series I samples, the highest
energy of each compressively strained sample reduces mono-
tonically from 6 °A through 0° to 6 °B. Obvious difference
can be identified between the series I and series II samples
that the PL line shape of the series II sample is different from
and the PL peak width of each series II sample is broader
than that of the series I sample with an identical substrate
misorientation. The series II samples manifest different de-
gree of PL peak splitting, especially for the sample with
6 °A misorientation, of which the PL peak splits into three
well-separated features, as denoted by A, B and C in Fig.
1共b兲. This may hint for the effect of compressive strain on
the degree of LR ordering and electronic band structure in
the QWs.
To check if the PL-peak splitting is just an exceptional
case, PL spectrum from the series III sample is also checked
at 11 K. The spectrum is depicted in Fig. 2, together with that
from the series II sample with 6 °A misorientation. While
both the samples manifest PL-peak splitting, obvious differ-
ences are seen that 共i兲 B and C peaks’ energies of the series
II sample are higher than those of the series III sample and
共ii兲 peak C’s intensity of the series II sample is stronger than
that of the series III sample. As a possible reason, thicker
GaP capping layer may produce additional compressive
strain to the GaInP QW layer, and hence reduce the PL tran-
sition energy. In contrast, thin capping layer may be not
enough to keep the GaInP QW layer in completely compres-
sive strain and hence a strain gradient may occur along the
sample’s growth direction, leading to more lattice defects
and dislocations. The capping layer related difference may be
also a hint for the strain effect on electronic band structure
and LR ordering, especially for the sample with 6 ° A sub-
strate misorientation.
According to the established knowledge about Ga
x
In
1−x
P
alloys
13,24,25
and GaP/InP short-period superlattices
共SPSLs兲,
26,27
the splitting of the PL peak may be due to the
following reasons: 共i兲 domains with different degree of LR
ordering distribute along QW’s growth direction and lateral
direction;
28,29
共ii兲 domains with different Ga:In ratio but
identical LR ordering and/or compositional modulations
26,27
exist in compressively strained GaInP QWs; and 共iii兲 the
valence band splitting occurs due to the joint effect of LR
ordering and compressive strain. For a reliable judgment
about the real reason of the PL-peak splitting, temperature-
dependent PL measurements were carried out on the series II
sample with 6 °A misorientation.
Figure 3 depicts the PL spectra taken at different tem-
peratures in a range of 11–250 K for the series II sample with
1.85 1.9 1.95 2 1.75 1.8 1.85 1.9
11K PL Intensity (arb. units)
Energy (eV)
Ener
gy
(eV)
A
B
C
6
0
A
0
0
6
0
B
Compressive
Matched
6
0
A
0
0
6
0
B
(a)
(b)
11K
11K
FIG. 1. 共Color online兲 PL spectra of 共a兲 series I lattice matched 共x= y
=0.52兲 and 共b兲 series II compressively strained 共x =0.40 and y = 0.76兲
Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs with different substrate misorienta-
tions at 11 K. Solid lines are experimental data and dashes are curve fittings.
1.75 1.8 1.85 1.9
A
B
C
A
B
C
Energy (eV)
PL Intens
i
ty
(
arb. un
i
ts
)
Cap:2nm GaInP
Cap:410nm GaP(Zn)
11K
11K
6
0
A
6
0
A
(a)
(b)
FIG. 2. 共Color online兲 PL spectra of the compressively strained
Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs 共x = 0.40 and y =0.76兲 with 6 °A mis-
orientation and different capping layer at 11 K, with A, B, and C fitting
peaks plotted in dashes.
013509-2 Zhu et al. J. Appl. Phys. 109, 013509 共2011兲
6 °A substrate misorientation. Three separate PL peaks can
be identified at low temperatures by curve-fitting treatment.
Below 55 K, the low-energy peak 共peak C兲 is the strongest
one with the broadest full-width at half maximum 共FWHM兲,
and the energy separation between the high-energy peak
共peak A兲 and mediate-energy peak 共peak B兲 is about 25 meV.
As temperature rises, peak A fades out gradually with red-
shift in its energy, while peak B quickly fades out and with
its energy manifesting obvious S-shape shift, and peak C dies
away rapidly and disappears totally at about 110 K with red-
shift in its energy.
The energies are plotted against temperature for the se-
ries II sample with 6 °A misorientation in Fig. 4 for the
three PL peaks. While peak-B energy shows S-shape tem-
perature dependence, which is similar to that evidenced in
the PL spectra of ordered GaInP,
30,31
peak-A energy mani-
fests monotonic redshift, and the energy difference between
peaks A and B also changes with temperature.
These different temperature-dependent characteristics
hint different mechanisms for the three PL peaks. It is clear
that misorientation and compressive strain have a joint effect
on LR ordering in GaInP QWs, especially for the compres-
sively strained sample with 6 ° A substrate misorientation.
And the multi-peak PL may be due to distributions of do-
mains with different LR ordering along the growth direction
and/or lateral direction in the compressively strained GaInP
QWs samples: 共i兲 For the series II sample with 6 °A misori-
entation, the temperature dependence of the PL peaks’ ener-
gies is different. Hence, it should not be due simply to dif-
ferent Ga:In ratio or compositional modulation. 共ii兲 The
energy difference between peaks A and B changes with tem-
perature. If it was due to the valence band splitting, the
change with temperature should be monotonically, which is
in fact not the real case. In addition, model-solid theory sug-
gested that the minimal splitting energy between heavy-hole
and light-hole is about 40–50 meV for series II samples with
different
,
8
which is obviously larger than the maximal
splitting energy between peaks A and B of only about 25
meV.
As to physical mechanisms of three PL peaks, prelimi-
nary assignments are made that peak A comes from excitonic
transitions in the low ordering domains that mostly locate in
the center part of the QW layer with a narrower
distribu-
tion, while peak B originates likely from the high ordering
domains that distribute mostly in the near-interface part of
the QW layer with a broader
distribution,
28
and peak C
may be introduced by defect and/or bandtail states that
mainly exist near the strained interface of the QW layer. It is
worthy to emphasize, however, that it is rather difficult to
identify if the PL peak splitting is resulted mainly from the
fluctuation of LR ordering along sample’s growth direction
or along the lateral direction, as lateral-position dependent
PL spectra also show obvious difference from spot to spot
for the compressively strained sample with 6 ° A substrate
misorientation. Details will be seen shortly afterwards in
Fig. 6.
To make a theoretical justification, a phenomenological
model may be beneficial for simultaneously accounting for
the influence of LR ordering and compressive strain. In the
MOVPE process, the system of GaInP can be taken as a
canonical ensemble when the number of atoms, the volume
and temperature are all kept at constant. The free energy of
the system,
F = U − TS, 共1兲
should be minimal when the system is in equilibrium, where
S is the entropy of the system, U is the intrinsic energy, and
T is growth temperature. In a strained system, S depends on
the completeness of lattice structure, and U directly relates to
the density of deformation energy, which is caused by strain.
It is obvious in Eq. 共1兲 that there are two possibilities for
reducing the free energy of system at a constant T. One is to
increase S and the other is to reduce U of the system. In a
1.7 1.75 1.8 1.85 1.9
11K
16K
25K
40K
55K
77K
90K
110K
150K
200K
250K
Ener
gy
(eV)
PL Intensity (arb. units)
×1
×1
×1.05
×1.3
×1.95
×2.8
×3.25
×4.2
×7.35
×18.3
×55
A
B
C
FIG. 3. 共Color online兲 PL spectra at different temperatures for the series II
compressively strained Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs 共x = 0.40 and
y = 0.76兲 sample with 6 °A misorientation. A, B, and C denote three sepa-
rated PL peaks. Solid lines represent original PL spectra, dashes indicate
sum of curve fittings, and curve fittings for A, B, and C, respectively.
Series I
6
°
111
A
Peak A
Peak B
Peak C
50 100 150 200 250
1.82
1.84
1.86
1.88
1
.
90
T K
E
nergy
e
V
FIG. 4. 共Color online兲 Energy vs temperature for A, B, and C PL peaks of
series II compressively strained Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs 共x
=0.40 and y = 0.76兲 sample with 6 ° A misorientation.
013509-3 Zhu et al. J. Appl. Phys. 109, 013509 共2011兲
strained system, the stress between the barrier and well layer
of the QWs sample may result in lattice defects and disloca-
tions near the interface, which as a consequence increase S
and reduce F of the system. Meanwhile, they usually form
localized and/or bandtail states. At a low temperature, the PL
from the localized and/or bandtail states is usually strong
and the FWHM is broad. As temperature rises, these PL
processes are weakened due to phonon scattering and
state padding effects. In the compressively strained
GaInP/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs with the 6 ° A misorienta-
tion, the characteristics of peak C is very similar to these PL
properties of the localized and/or bandtail states.
The other possibility of reducing the free energy is to
decrease the density of deformation energy w in a strained
system. For the series I samples, the lattice constants of the
QW, barrier, and GaAs substrate are nearly the same, and the
density of deformation energy is hence close to zero. As a
consequence, the degree of LR ordering is determined solely
by the substrate misorientation. For the series II samples, on
the other hand, the lattice constant of the QW layer is larger
than those of the barrier layer and substrate. As a result, the
density of deformation energy is not zero. The local lattice
constant of Ga
x
In
1−x
P may be adjusted by insertion of addi-
tional GaP/InP ordering structure near the interface, i.e., by
regulating the degree of LR ordering as schematically illus-
trated in Fig. 5. Therefore, the formation of in-plane lattice-
matched Ga
x
In
1−x
P structure will be beneficial for reducing
the density of deformation energy in the QW layer, espe-
cially near the interfaces. In addition, CuPt
B
ordering struc-
ture is as indicated by first-principles calculation
32
more
stable in metal-organic chemical vapor deposition samples
共which is similar to the MOVPE technique relevant to this
study兲 than in hydrogen-free molecular beam epitaxy
samples because the energy of forming phosphide 共001兲-共2
⫻2兲 surfaces is lower than that of a rebonded one. Conse-
quently, the formation of Ga
0.4⫾
/2
In
0.6⫿
/2
P LR ordered
structure will further reduce the free energy in the ordered
region.
At this point, it may be curious how the LR ordering will
change if both the compressive strain and misorientation co-
exist in GaInP QWs? The theory of LR ordering has in fact
indicated that the substrate misorientation can significantly
influence the degree of LR ordering in compressively
strained QWs. As a result, the misorientation and compres-
sive strain may introduce a joint effect on the LR ordering in
the QWs. The misorientation of 6 ° B is preferred for LR
ordering as the degree of LR ordering is strong. For the
misorientation of 0°, a weak competition exists between mis-
orientation and compressive strain, and the formation of LR
ordering is depressed. For the misorientation of 6 ° A, mis-
orientation competes strongly with compressive strain, lead-
ing the formation of LR ordering to be strongly hindered.
The competition also results in significant fluctuation to
the degree of LR ordering along QW’s growth direction and
lateral direction, and a possible strain gradient in QW will be
decisive for either compressive strain or substrate misorien-
tation to play a dominant role. In compressively strained thin
GaInP QWs, the z-direction distribution of LR ordering
should be a major effect due to a large strain gradient in this
direction, which is responsible for an obvious PL-peak split-
ting in the series II samples with 6°A substrate misorienta-
tion. It is noteworthy that this framework of hypothesis ap-
proves with the aforementioned preliminary assignments,
i.e., peak A comes from excitonic transitions in low ordering
domains mostly locate in the center part of QW layer with a
narrower
distribution, while peak B originates likely from
high ordering domains distribute mostly in the near-interface
part of the QW layer with a broader
distribution.
28
As a consequence of the competition between misorien-
tation and compressive strain in GaInP QWs, domains with
different degree of LR ordering should exist not only in the z
direction,
29
but also in the x− y plane of the QW layer. Ex-
perimental evidence is established by in x − y plane position-
dependent PL measurements. The representative PL spectra
are illustrated in Fig. 6 for the series I and II samples, which
are taken at 77 K from three or four different spots of the
sample’s surface. In the PL measurements, the laser beam
diameter is about 50
m, which is obviously larger than the
typical domain size of about ⱕ2
m.
33
As indicated in Fig. 6共a兲, the peak energy gap among
different spots, ⌬E, is about 11.2 meV, 3.8 meV, and 1.8
meV, respectively, for the series II samples with 6 ° A, 0°,
and 6 ° B substrate misorientations, and the PL intensity dif-
fers from one to another obviously. From Fig. 6共b兲, it is clear,
however, that the peak energy gap ⌬E is only about 1.0 meV,
0.7 meV, and 0.6 meV, respectively, for the series I samples
with 6 ° A, 0°, and 6 °B substrate misorientations, and the
PL intensity does not manifest obvious difference, which are
in sharp contrast with those of the series II samples. This
result suggests that while the in x − y plane LR ordering is
quite uniform for each of the series I samples, considerable
x− y plane nonuniformity of the LR ordering occurs in the
series II samples, in which rather strong compressive strain
exists. One possible explanation might be the compositional
...
Barrier
Well
Barrier
GaP
InP
GaP
Ga
1x
In
x
P
InP
Ga
1x
In
x
P
GaP
InP
......
In
Ga
P
z
Η
a
0
FIG. 5. 共Color online兲 Lattice deformation status and favorite Ga
x
In
1−x
P
lattice structure near the interface in strained Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P
QWs.
013509-4 Zhu et al. J. Appl. Phys. 109, 013509 共2011兲
modulation effect in compressively strained samples, which
was previously observed in GaP/InP SPSLs systems.
11,26,27
When the facts are taken into account that 共i兲 the thickness of
QW layer is 10 nm while that of SPSLs is just about 1 nm,
and 共ii兲 the value of peak energy gap ⌬E and its substrate
misorientation dependence, however, the explanation seems
to be unreasonable in this case, and in contrast, the in x − y
plane fluctuation of LR ordering should be the right reason.
Such intensified in x − y plane fluctuation of LR ordering is a
direct consequence of the competition between the misorien-
tation and compressive strain in the 6 °A sample.
It is worthy to mention that it would in principle be
beneficial to use near-field and/or polarized PL measure-
ments for clarification of minor fluctuation of LR
ordering.
4,28,34
For the competition effect of strain and sub-
strate misorientation in macroscopic scale in the x− y plane
of QW layer, however, the results from conventional far-field
PL measurements are reliable in statistics when limited num-
ber of ordered domains is illuminated by the laser beam as
the case in our experiments. The arguments are as follows:
共i兲 When the LR ordering fluctuates drastically and its distri-
bution is no longer uniform, the PL spectra from different
spots may change significantly with the position of laser
spot, which is in fact the experimental observation as de-
picted in Fig. 6 for the PL spectra from different positions in
the x− y plane of the samples. 共ii兲 Strain can generate a pi-
ezoelectric field and produce polarization PL in
semiconductors,
35,36
it may be hard to distinguish either or-
dering or strain/substrate misorientation as the origin of pos-
sible polarization PL signal in our series II samples.
A schematic diagram is finally derived and is depicted in
Fig. 7 for the band-edge structure and distribution of LR
ordering in two dimensions of the compressively strained
Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs with 6 ° A substrate
misorientation. The 6 °A misorientation is helpful in hinder-
ing the formation of CuPt-type LR ordering,
8
while near the
interface, compressive strain is likely to intensify the forma-
tion of CuPt-type LR ordering and may result in obvious
fluctuation to the degree of ordering. The joint effect is the
coexistence of the domains with different degree of LR or-
dering in the z direction and x− y plane of GaInP QW, which
is responsible for the degradation of optical properties and
the PL-peak splitting from one single peak to A, B, and C
three well-separated PL features. It is therefore straightfor-
ward that 6 ° A misorientation should not be preferable for
compressively strained Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs
with ordering so long as the performance of material and
device is favorite.
IV. SUMMARY
To summarize, temperature-dependent and in x− y plane
position-dependent PL measurements were carried out on
three series Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs samples
with different substrate misorientations and degree of CuPt-
type LR ordering. Experimental results indicate that CuPt-
type LR ordering is affected by both substrate misorientation
and compressive strain in QWs. The competition of the mis-
orientation and compressive strain not only introduces
changes to the degree of LR ordering along the z direction,
but also yields domains with different degree of LR ordering
in the x − y plane of the QW layer. A phenomenological
model is proposed based on the principle of minimum free
energy so as to help the understanding of the experimental
observations. The results suggest that the 6 ° A substrate
misorientation should not be considered for compressively
strained GaInP/AlGaInP QWs so long as the CuPt-type LR
ordering may exist.
1.75 1.8 1.85 1.9 1.9 1.95 2
77K PL Intensity (arb. units)
Energy (eV)
Energy (eV)
Compressed
Matched
514.5nm
laser line
6
o
A
6
o
A
6
o
B
6
o
B
0
o
0
o
ΔE
ΔE
ΔE
ΔE
ΔE
ΔE
ΔE≈1.0meV
ΔE≈11.2meV!
ΔE≈3.8meV
ΔE≈0.7meV
ΔE≈0.6
meV
ΔE≈1.8meV
(
a
)
(b)
FIG. 6. 共Color online兲 In x − y plane position-dependent PL spectra of 共a兲
series II compressively strained 共x = 0.40 and y = 0.76兲 and 共b兲 series I lattice
matched 共x = y = 0.52兲 Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
P QWs samples with
6 ° A, 0°, and 6 ° B misorientations at 77 K. z denotes QW growth direction.
⌬E is the maximum offset of peak energy among all testing positions.
Band Align and
Η
Distribution
AlGaInP barrier
Lorder
Horder
GaInP
Interface
AlGaInP
AlGaInP
AB BC C
z
z
x
y
Strain Strain
Η
FΗ
A
B
x,y
Η
FIG. 7. 共Color online兲 Schematic for band-edge structure and LR ordering
in compressively strained Ga
x
In
1−x
P/ 共Al
0.66
Ga
0.34
兲
y
In
1−y
PQW共x = 0.40 and
y = 0.76兲 with 6 ° A substrate misorientation. Domains with different degree
of ordering and ordering distribution coexist in x− y plane and z direction
simultaneously.
013509-5 Zhu et al. J. Appl. Phys. 109, 013509 共2011兲
ACKNOWLEDGMENTS
One of the authors 共J.S.兲 thanks Rolf Winterhoff for the
QWs samples, Guoliang Shi for technical support, Tienan
Zhao for help around the Ar
+
laser 共Spectra-Physics Beam-
Lok 2065兲, and Xiaohui Mou and Jihua Li for the mainte-
nance of the FTIR spectrometer. The work was partly spon-
sored by the STCSM 共Grant Nos. 08XD14047,
09JC1415600, and 08ZR1421800兲 and the NSFC 共Grant
Nos. 10927404 and 60723001兲 of China.
1
C. P. Kuo, R. M. Fletcher, T. D. Osentowski, M. C. Lardizabal, M. G.
Crawford, and V. M. Robbins, Appl. Phys. Lett. 57, 2937 共1990兲.
2
J. Olson, S. Kurtz, A. Kibbler, and P. Faine, Appl. Phys. Lett. 56, 623
共1990兲.
3
K. A. Bertness, S. R. Kurtz, D. J. Friedman, A. E. Kibbler, C. Kramer, and
J. M. Olson, Appl. Phys. Lett. 65,989共1994兲.
4
A. Mascarenhas, Spontaneous Ordering in Semiconductor Alloys 共Kluwer
Academic, New York, 2002兲.
5
A. Gomyo, T. Suzuki, and S. Iijima, Phys. Rev. Lett. 60, 2645 共1988兲.
6
D. B. Laks, S.-H. Wei, and A. Zunger, Phys. Rev. Lett. 69, 3766 共1992兲.
7
G. S. Horner, A. Mascarenhas, S. Froyen, R. G. Alonso, K. Bertness, and
J. M. Olson, Phys. Rev. B 47, 4041 共1993兲.
8
J. Shao, A. Dörnen, R. Winterhoff, and F. Scholz, Phys. Rev. B 66, 035109
共2002兲.
9
C. B. Nelson, P. C. Taylor, and W. A. Harrison, Phys. Rev. B 70, 224207
共2004兲.
10
A. Zunger and S. Mahajan, Handbook of Semiconductors, 2nd ed.
共Elsevier, Amsterdam, 1994兲, Vol. 3, p. 1399.
11
K. Y. Cheng, K. C. Hsieh, and J. N. Baillargeon, Appl. Phys. Lett. 60,
2892 共1992兲.
12
A. C. Chen, A. M. Moy, P. J. Pearah, K. C. Hsieh, and K. Y. Cheng, Appl.
Phys. Lett. 62, 1359 共1993兲.
13
C. G. Van de Walle, Phys. Rev. B 39, 1871 共1989兲.
14
T. Katsuyama, I. Yoshida, J. Shinkai, J. Hashimoto, and H. Hayashi, Elec-
tron. Lett. 26,1375共1990兲.
15
A. Gomyo, K. Makita, I. Hino, and T. Suzuki, Phys. Rev. Lett. 72, 673
共1994兲.
16
J. Shao, R. Winterhoff, A. Dörnen, E. Baars, and J. Chu, Phys. Rev. B 68,
165327 共2003兲.
17
S.-H. Wei and A. Zunger, Appl. Phys. Lett. 64, 757 共1994兲.
18
J. Shao, X. Lü, F. Yue, W. Huang, S. Guo, and J. Chu, J. Appl. Phys. 100,
053522 共2006兲.
19
J. Shao, A. Dörnen, R. Winterhoff, and F. Scholz, J. Appl. Phys. 91, 2553
共2002兲.
20
J. Shao, W. Lu, X. Lü, F. Yue, Z. Li, S. Guo, and J. Chu, Rev. Sci. Instrum.
77, 063104 共2006兲.
21
J. Shao, F. Yue, X. Lü, W. Lu, W. Huang, Z. Li, S. Guo, and J. Chu, Appl.
Phys. Lett. 89, 182121 共2006兲.
22
J. Shao, W. Lu, F. Yue, X. Lü, W. Huang, Z. Li, S. Guo, and J. Chu, Rev.
Sci. Instrum. 78, 013111 共2007兲.
23
S.-H. Wei and A. Zunger, Appl. Phys. Lett. 56, 662 共1990兲.
24
S.-H. Wei and A. Zunger, Phys. Rev. B 57, 8983 共1998兲.
25
S.-H. Wei and A. Zunger, Phys. Rev. B 49, 14337 共1994兲.
26
K. C. Hsieh, J. N. Baillargeon, and K. Y. Cheng, Appl. Phys. Lett. 57,
2244 共1990兲.
27
H. M. Cheong, Y. Zhang, A. G. Norman, J. D. Perkins, A. Mascarenhas,
K. Y. Cheng, and K. C. Hsieh, Phys. Rev. B 60,4883共1999兲.
28
H. M. Cheong, A. Mascarenhas, J. F. Geisz, J. M. Olson, M. W. Keller,
and J. R. Wendt, Phys. Rev. B 57, R9400 共1998兲.
29
T. Mattila, L.-W. Wang, and A. Zunger, Phys. Rev. B 59, 15270 共1999兲.
30
M. Kondow, S. Minagawa, Y. Inoue, T. Nishino, and Y. Hamakawa, Appl.
Phys. Lett. 54,1760共1989兲.
31
P. G. Eliseev, J. Appl. Phys. 93, 5404 共2003兲.
32
I. G. Batyrev, W. E. McMahon, S. B. Zhang, J. M. Olson, and S.-H. Wei,
Phys. Rev. Lett. 94, 096101 共2005兲.
33
H. M. Cheong, A. Mascarenhas, S. P. Ahrenkiel, K. M. Jones, J. F. Geisz,
and J. M. Olson, J. Appl. Phys. 83,5418共1998兲.
34
H. M. Cheong, Y. Zhang, A. Mascarenhas, J. F. Geisz, and J. M. Olson, J.
Appl. Phys. 83, 1773 共1998兲.
35
T. Maruyama, E. L. Garwin, R. Prepost, G. H. Zapalac, J. S. Smith, and J.
D. Walker, Phys. Rev. Lett. 66,2376共1991兲.
36
T. Maruyama, D.-A. Luh, A. Brachmann, J. E. Clendenin, E. L. Garwin, S.
Harvey, J. Jiang, R. E. Kirby, C. Y. Prescott, R. Prepost, and A. M. Moy,
Appl. Phys. Lett. 85, 2640 共2004兲.
013509-6 Zhu et al. J. Appl. Phys. 109, 013509 共2011兲