PHYSICAL REVIEW B 66, 085331 共2002兲

Defect states in red-emitting Inx Al1Àx As quantum dots
R. Leon and J. Ibáñez
Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, California 91109

S. Marcinkevičius and J. Siegert
Department of Microelectronics and Information Technology, Royal Institute of Technology, Electrum 229, 164 40 Kista, Sweden

T. Paskova and B. Monemar
Department of Physics and Measurement Technology, Linköping University, S-581 83, Linköping, Sweden

S. Chaparro, C. Navarro, S. R. Johnson, and Y.-H. Zhang
Center for Solid State Electronics Research and Department of Electrical Engineering, Arizona State University, Tempe, Arizona 85287
共Received 13 May 2002; published 30 August 2002兲
Optical and transport measurements carried out in pn diodes and Schottky barriers containing multilayers of
InAlAs quantum dots embedded in AlGaAs barriers show that while red emission from quantum dot 共QD兲
states is obtained at ⬃1.8 eV, defect states dominate the optical properties and transport in these quantum dots.
These defects provide nonradiative recombination paths, which shortens the carrier lifetimes in QD’s to tens of
picoseconds 共from ⬃1 ns兲 and produce deep level transient spectroscopy 共DLTS兲 peaks in both p and n type
structures. DLTS experiments performed with short filling pulses and bias dependent measurements on InAlAs
QD’s on n-AlGaAs barriers showed that one of the peaks can be attributed to either QD/barrier interfacial
defects or QD electron levels, while other peaks are attributed to defect states in both p and n type structures.
DOI: 10.1103/PhysRevB.66.085331

PACS number共s兲: 78.66.Fd, 73.40.Gk

INTRODUCTION

The role that defects play on quantum dot properties is
interesting from several perspectives. Radiation induced defects appear to have much lesser effects in diminishing radiative emission from quantum dot 共QD兲 luminescent structures and QD laser diodes1,2 than in quantum well 共QW兲 or
bulk structures. Higher intensities have even been observed
after proton and ion induced displacement damage in QD
structures. These can be explained by defect assisted carrier
relaxation, and by defect assisted tunneling in cases with
potential barrier surrounding QD’s.3
The effects of dislocation defects on QD properties are
also of interest due to several promising technological applications. The growth of multistacked planes containing quantum dots is used to increase gain in lasers. If the cumulative
strain from several dot layers exceeds the critical thickness
for plastic relaxation4,5 then a misfit dislocation can be
formed at the buffer layer/quantum dot interface in multistacked dot layers. Such misfit dislocations might form spontaneously in some structures, however, strong radiative emission from QD states is still present in these cases. Another
application where dislocation defects are relevant involves
the attempts to obtain positional order in QD’s. In this type
of growth, QD’s are grown on strain relaxed GeSi or InGaAs
epitaxial films, and they have been shown to have rectangular alignment, presumably from the dislocation shear steps at
the surface.6,7 Another interesting application where dislocations can play an important role is in the growth on InAs
QD’s on GaAs/Si.8,9 QD device applications in such structures have the potential of enabling Si optoelectronic devices, and of integrating logical and optical functions in
system-on-a-chip applications.
0163-1829/2002/66共8兲/085331共7兲/$20.00

Impurities introduced unintentionally during growth can
have an important role in the properties of QD containing
structures, and are believed to be responsible for the short
luminescence lifetimes observed in some InAlAs/AlGaAs
QD’s.10 It is also well known, that interfacial defects as those
formed during ex-situ processing in etched quantum dots
have a strong role on their optical properties as well.11,12
The use of deep level transient spectroscopy 共DLTS兲 in
the characterization of deep levels in semiconductors is well
established. DLTS has also been used to attempt determination of eigen-states from electron and hole levels in InAs
quantum dots.13–15 The literature on this topic has given
some conflicting information, and the interpretation of results is not always straightforward. To complicate matters, it
is quite possible to have defect states in samples that also
contain QD’s; therefore, differentiation of DLTS signals
originating from defects or from electron or hole energy levels can be ambiguous.
Here we use optical and electrical measurements to study
the role of impurities and unintentional defects introduced
during growth in InAlAs QD’s in both p and n AlGaAs barriers. The results shown here demonstrate that unlike radiation induced damage, or dislocation effects from the barrier
layer under the dots, these defects have a prominent role in
the optical and transport properties of these QD structures.
DLTS experiments show different behaviors from peaks in p
and n type samples. Measurements carried out at different
reverse biases and experiments performed varying the filling
pulse times allowed determination of trap activation energies, trap densities, and capture cross sections. In this work
we show that these defects affect the optical properties of
these quantum dots due to their localization either within the
dots or at their interfaces. This is in contrast to what is ob-

66 085331-1

©2002 The American Physical Society

PHYSICAL REVIEW B 66, 085331 共2002兲

R. LEON et al.

served from radiation induced defects,1–3 which are randomly distributed, and the majority are formed in the barrier
material, and spatially separated from the QD region, where
they do not contribute significantly to nonradiative recombination due to wave function confinement in the QD’s.
EXPERIMENTAL DETAILS

Two QD structures were grown by molecular beam epitaxy. The structures consist of 50-nm layers of n-type 共ptype兲 Al0.33Ga0.67As terminated with In0.45Al0.55As QD’s 共using five monolayers coverage兲, repeated eight times on top of
a 300-nm-thick n ⫹ doped (n ⫹ ⫽1⫻1018 cm⫺3 ) AlGaAs
layer and a 100-nm-thick n ⫹ doped (n ⫹ ⫽1⫻1018 cm⫺3 )
GaAs buffer layer. Si and Be were used for n and p dopants
with nominal doping n⫽p⫽1017 cm⫺3 . A final Al0.3Ga0.7As
layer with the same doping level was deposited and capped
with a thin GaAs layer to prevent oxidation of the AlGaAs. A
top Schottky diode and back Ohmic contacts were formed
for the n-type sample and back and top Ohmic contacts were
formed on the p-type sample. Two additional samples with
doped AlGaAs barriers and no InAlAs quantum dots were
also grown as control structures, and processed in a similar
manner. Analysis of island sizes and densities using atomic
force microscopy in air gives average diameters of 20 nm 共5
nm heights兲 and concentrations of 1⫻109 cm⫺2 for uncapped InAlAs QD’s grown under the same conditions. Capacitance voltage measurements were carried out at variable
temperature 共20 to 300 K兲. Deep level transient spectroscopy
was carried out from 20 to 315 K at delay time ␶ in the
共0.02–100兲 ms range and at a rate window of 4.3⫻ ␶ . Lowtemperature cathodoluminescence 共CL兲 imaging and spectroscopy were carried out using a scanning electron microscope equipped with a cryogenic stage and a monochromator
attachment for CL spectroscopy. CL was performed imaging
the structures in cross section, at temperatures between 4.5
and 10 K using an accelerating voltage of 5 KeV. Carrier
dynamics were studied by time-resolved photoluminescence
共TRPL兲 at 80 K after excitation by frequency-doubled femtosecond laser pulses at 400 nm from a Ti:saphire laser. The
PL was detected using a synchroscan streak camera, combined with a 0.25 m spectrometer 共temporal resolution 3 ps兲.

FIG. 1. Cathodoluminescence spectra from the eight-layer QD
regions in n- and p-InAlAs/AlGaAs structures.

grown InAlAs/AlGaAs QD’s.16,17 The control structures
showed no CL from the active region. Emission from the
GaAs at 831 nm was seen in all structures, with a stronger
intensity from the GaAs buffer layer below the AlGaAs barrier.
Time resolved photoluminescence

Figure 2 shows PL transients from these QD structures.
Some weak PL emission from the control samples 共containing only the p and n doped AlGaAs barrier films兲 can be
detected by TRPL, although these were not observable with
the CL setup that was used at temperatures of 4.5 K. Figure
2 shows results from p- and n-InAlAs quantum dots as well

RESULTS
Cathodoluminescence

A typically broad emission in the visible 共red兲 region is
apparent in the CL spectra displayed in Fig. 1. The peaks
found are at 660– 675 nm 共1.843 and 1.857 eV with inhomogeneous broadening of 37 and 26 meV, respectively兲. Monochromatic CL imaging 共not shown here兲 using the peak
wavelengths in both n and p samples 共imaged in cross section兲 did show that the signal originated from the 40 nm
region containing the multilayered InAlAs QD’s, hence we
attribute these CL peaks to radiative recombination from QD
states. The absence of any emission from AlGaAs is apparent
as a dark region from the layer under the QD structures in
CL imaging mode, even though AlGaAs related peaks ⬃2
eV have been observed in previous studies of similarly

FIG. 2. Time resolved photoluminescence from p- and
n-InAlAs/AlGaAs quantum dots. TRPL is also shown from the p
and n control samples.

085331-2

PHYSICAL REVIEW B 66, 085331 共2002兲

DEFECT STATES IN RED-EMITTING Inx Al1⫺x As . . .

as from the control sample barrier emission. These measurements indicate PL decay times of 6 and 29 ps for the n- and
p-type InAlAs/AlGaAs QD’s, respectively. These are significantly shorter than the decay from the weaker AlGaAs emission in the control structures, which were 88 and 85 ps for
the n and p structures, respectively. In InAs and InGaAs QD
structures, where carrier dynamics are not dominated by nonradiative recombination, PL decay times measured at similar
temperatures have been reported in the range 共1–5兲 ns.18,19
Decay times from InAlAs/AlGaAs QD’s grown by MOCVD
and MBE have been reported in the range 共300–500兲 ps.18,20
Electrical measurements „C-V and DLTS…

In order to further characterize the origin of the nonradiative recombination centers responsible for the short QD PL
decay times observed in both n and p InAlAs/AlGaAs quantum dot structures, electrical measurements were performed
using DLTS and C-V analysis. Capacitance voltage was performed at various temperatures in order to determine the
electron and hole concentration 共majority carriers兲 in these
samples. These measurements allowed quantitative determination of electron and hole trap densities and estimations of
the space charge region at the various values of reverse bias
used in DLTS. From C-V analysis it was found that the
carrier concentration drops sharply below 180 K for the n
type samples, and below 150 K for the p type structures
共from an initial shallow acceptor or donor concentration of
2⫻1017 cm⫺3 at room temperature兲. This rapid drop is due
to carrier freeze-out and it accounts for the variation in DLTS
peak intensity found at different time windows in the DLTS
measurements.

FIG. 3. DLTS spectra for from p-InAlAs/AlGaAs QD structures,
for delay time of 0.2 ms 共emission rate of 0.86 ms兲, at ⫺1 V applied
bias, and 1 ␮s pulse duration. The inset shows Arrhenius plot for 12
values of delay times from the main peak.

effects from the main 共more intense兲 peak did not allow determination of activation energy from this peak.
DLTS measurements were also performed on the control
p-AlGaAs samples. No signal was observed even at much
higher gains than used in the measurements on the structures
containing the InAlAs quantum dots.
DLTS in InAlAsÕn-AlGaAs quantum dot structures

DLTS in InAlAsÕp-AlGaAs quantum dot structures

DLTS of the p structures was performed for various values of reverse bias, filling pulse duration and delay time
windows.21 Figure 3 shows a typical spectrum taken at delay
times of 0.2 ms, and reverse bias of ⫺1 V. Spectra were
collected from values of time windows between 0.02 and
100 ms, and at reverse bias voltages from ⫺0.5 to ⫺1.5 V.
Activation energies were evaluated from all spectra, a typical
Arrhenius plot is shown in the inset of Fig. 3. The values for
the activation energy (E a ) did not vary significantly with
reverse bias 共E a ranged from 0.35 to 0.39 eV兲 averaging 0.37
eV. The peak shape did not change with filling pulse duration, but its intensity changed. This was used to measure the
value of the main trap capture cross section directly. As is
expected and common in most DLTS spectra, the peak shape
did not change with applied bias, and the peak intensity increased monotonically with increased bias. Trap concentrations (N t ) were evaluated for this peak from the relation
N t ⫽ 共 2⌬C/C 兲 N d,a .
N t was found to be ⬃6⫻1016 cm⫺3 .
Figure 3 also shows a lower intensity peak at lower temperature 共130 K兲. This peak was only present when DLTS
spectra were acquired using longer delay times, and it did not
change intensity in proportion to applied bias. Convolution

Similar DLTS experiments as described in the previous
section were performed on n type InAlAs quantum dot structures. In these experiments, several ‘‘anomalies’’ were observed. At long values of filling pulse time 共from 1 to 100
ms兲 the DLTS spectra shows a complex structures, with at
least three convoluted peaks. Variation of time windows
showed a change in peak shape, indicating different values
for activation energies from the different peaks. The complex
nature of the spectra did not allow determination of single
activation energies for spectra acquired using these long filling pulse times.
DLTS measurements performed on the control n-AlGaAs
structures did not show any of the peaks found in the
InAlAs/AlGaAs structures. A different peak was found at
higher temperatures, however, its intensity was at least an
order of magnitude lower than the ones reported for the
InAlAs/AlGaAs structures.
Further experiments performed using shorter ‘‘filling’’
pulses were carried out in these structures. The aim was to
eliminate the signal from traps with small capture cross sections by using short filling pulses, and possibly analyze any
remaining peaks that might have larger capture cross sections. Surprisingly, when the filling pulses were reduced 共1 to
10 ␮s兲, a dramatic change in the spectra was observed. The
complex peak structure disappeared, and instead, a simple
peak 共but broader than most defect related peak兲 remained.

085331-3

PHYSICAL REVIEW B 66, 085331 共2002兲

R. LEON et al.

FIG. 4. DLTS spectra from n-InAlAs/AlGaAs QD structures
obtained for very short filling pulses 共1 ␮s兲 at various applied reverse bias voltages for a delay time of 0.5 ms.

This peak appeared at lower temperature, and it was much
more intense than what was observed as a very weak shoulder in the experiments performed with long filling pulses.
DLTS spectra were then taken using different delay times
and different applied reverse biases. The experiments at different reverse biases showed several unusual features. The
peak shape changed with reverse bias, becoming narrower at
higher values of applied bias. These changes are shown for
several bias values in Fig. 4. The intensity of the signal was
compared for the various biases by integrating the peaks in
order to account for the changes in the peak shapes. The
signal intensities do increase with applied bias, but they
show at least three well-defined plateaus. Activation energies
obtained from the different Arrhenius plots were evaluated. A
general increase in activation energy can be seen with increased applied biases ranging from 0.5 to 3.25 V. The increase is not monotonic but the general trend is towards
higher activation energies with larger applied biases. Measurement of trap concentrations from this peak was also
evaluated from the signal intensity and measured carrier concentration and found to be ⬃3⫻1014 cm⫺3 at ⫺1 V applied
reverse bias.
Experimental measurements of trap carrier cross sections

Capture cross sections can be measured for traps by varying the ‘‘fill’’ pulse duration and recording the change in
signal intensity for a given peak. As reported in earlier
studies,22,23 capture cross sections can be measured directly
in DLTS using the relation
⌬C 共 t p 兲 /⌬C 共 t p →⬁ 兲 ⫽1⫺e 共 ⫺ ␴ e 具 v th典 t p 兲 ,
where ␴ is the capture cross section, e the carrier concentration, 具 v th典 is the thermal velocity, and t p is the filling pulse
duration.

FIG. 5. DLTS signal intensity as a function of filling pulse for
the main peak in the p-type InAlAs/AlGaAs quantum dots and for
the peak obtained at short filling pulse times in the n-InAlAs/
AlGaAs QD structures. Measurements were done at the temperatures for maximum peak intensity 共120 K in n-InAlAs/AlGaAs and
186 K for p-InAlAs/AlGaAs兲 and 0.5 ms delay times in both cases.

Capture cross section measurements were performed for
the main peak seen in the p-QD samples and for the main
peak studied in the n-QD samples 共the remaining peak after
1 ␮s fill pulse兲. These are plotted for both peaks in Fig. 5.
For the n-type structures, the peak intensity reaches a maximum and then decreases with long filling pulses, which is
consistent with the unusual low temperature peak increase
shown in Fig. 6. The maximum peak intensity in this case
occurs for filling pulse time of 30 ␮s.
Better fits were obtained for the shorter times. In the
n-type samples, this can be explained with the unusual behavior of the lower temperature peak for longer fill times,
and also by peak distortion due to the appearance of the
multipeak structure due to the increased importance of defects with smaller capture cross sections at longer fill times.
Fits to the experimental data shown in Fig. 5 show larger
capture cross sections from the traps found in the n-type
samples with small filling pulses, as compared to the traps
found in the p-type samples. Values for the capture cross
sections were ⬃1⫻10⫺17 cm2 in the n-type samples vs 1
⫻10⫺19 cm2 in the p-type samples.
DISCUSSION

These results, which combine optical and transport measurements, clearly show that a strong interaction exists between defect centers, and quantum dot states in both types of
structures 共with n- and p-doped barriers兲. The very short PL
decay times originating from the QD emission was the first
indication of the strong role of defect levels on the properties
of these QD’s, since PL decay times in the QD’s are much
shorter than values obtained in earlier measurements of PL
lifetimes in InAs, InGaAs, and InAlAs quantum dots. Since

085331-4

PHYSICAL REVIEW B 66, 085331 共2002兲

DEFECT STATES IN RED-EMITTING Inx Al1⫺x As . . .

FIG. 6. DLTS spectra 共at ⫺4 V reverse bias and 100 ms delay
time兲 from n-InAlAs/AlGaAs QD structures. The peaks are shown
at their corresponding relative intensities, for very short filling pulse
times and long filling pulse times.

the PL decay times in the QD’s are also much shorter than in
the control samples 共without the quantum dots兲, it is probable that defect levels responsible for the short carrier lifetimes in the dots are either within the dots or at the dot/
barrier interface. PL decay times in doped samples gives
minority carrier lifetimes.24 Estimation of trap densities in pand n-type structures give 6⫻1016/cm⫺3 for the InAlAs/pAlGaAs samples vs 3⫻1014/cm⫺3 for the InAlAs/n-AlGaAs
samples. However, the trapping rates 共1/␶兲 are proportional
to trap concentrations multiplied by capture cross sections
共and the thermal velocity兲. Here we have more traps in the p
samples, but also a much smaller cross section. The product
of the concentration and cross section is only twice as large
in the p structure than in the n structures. If we multiply
further by the thermal velocity, the rate in the n sample is
larger than in the p sample, which can explain the shorter
lifetime in the n than in the p structures.
One of the difficulties using DLTS to characterize quantum dots in an environment that also contains traps due to
defect related levels, is the ambiguity in differentiating
which signal originates from defects and what is due to the
QD electron or hole levels, since both of them can produce
DLTS signals. We believe the main peak found in the DLTS
spectra from the p-type structure, with E a ⬃0.37 eV, originates from defect levels, rather than QD levels. The two
main arguments for attribution of this hole trap to a defect
related trap, rather than from hole levels in quantum dots, are
the trap activation energy (E a ) and the trap level concentration (N t ). If these originated from QD hole levels, 0.37 eV
would be the energy spacing from the barrier valence band to
the hole level; and this is much too large to be explained by
a hole level. Also, the hole and electron level concentrations
would have to scale with the known quantum dot concentration in the small volume probed by our measurements

FIG. 7. Integrated DLTS signal 共hollow symbols兲 as a function
of applied reverse bias from n-InAlAs/AlGaAs QD structures. Also
plotted are activation energies extracted from Arrhenius plots of
DLTS spectra taken at various values of applied reverse bias.

⬃2 – 3⫻1014/cm3 , which is too low to account for N t ⬃6
⫻1016/cm3 .
Unlike the well studied and controversial D-X centers in
n-type AlGaAs, there have been few reports of any important
defects levels in p-AlGaAs. The absence of any DLTS peaks
from the control p-AlGaAs structures used here agrees with
this observation. We believe that this 0.37 eV hole trap is
related to the presence of InAlAs, however, it does not behave similar to other known interface defects, in the sense
that the signal strength increases monotonically with reverse
bias. This is in contrast to the sharply increasing DLTS signal
shown in Fig. 7 for the peak found at short filling pulses in
the n-InAlAs samples, and to the sharply increasing signal
attributed to interfacial defects in other reports that include
AlGaAs based superlattices.25 Hole traps in AlGaAs devices
have been reported in n-type AlGaAs/InGaAs PHEMT’s
which were attributed to surface states at ungated AlGaAs
regions, but their reported activation energies were much
higher than the E a ⫽0.37 eV measured in this work.26
DLTS analysis in the InAlAs/n-AlGaAs structures presents an even more complex picture. The multiple peak
structure found at longer filling pulse times is similar to what
other studies have found for DX-like centers in GaAs/
AlGaAs superlattices.27,28
The increase in intensity with diminishing filling pulse
times observed from the low-temperature peak is very unusual, and to our knowledge, has never been reported. The
observation of plateaus in the integrated DLTS signal intensity rule out a uniform defect distribution, and are consistent
with the signal originating from the different QD layers.
From the measured free electron concentrations, calculations
of depletion widths for Schottky barriers and pn junction
geometries were made at different reverse bias levels. These
give depletion widths of 420, 340, 265, and 152 nm at biases
of ⫺3, ⫺2, ⫺1, and 0 V. These agree quite well with the
interpretation of the signals originating from 2, 3, and 5 lay-

085331-5

PHYSICAL REVIEW B 66, 085331 共2002兲

R. LEON et al.

ers of QD’s at biases of ⫺1, ⫺2, and ⫺3 V 共there are three
quantum dot layers within the space charge region at zero
bias兲. Here, unlike in the case shown earlier for the InAlAs/
p-AlGaAs samples, the trap density is very close to the volumetric estimates of QD in the region probed by the electrical
measurements 共3⫻1014 cm⫺3 vs ⬃2 – 3⫻1010 cm⫺3 兲,
therefore, we cannot rule out the possibility that this trap
level originates from electron levels in the InAlAs quantum
dots. These observations, the peak behavior with applied bias
and the defect trap concentration, lead us to conclude that
this peak originates either from an interfacial defect, most
likely from the InAlAs/AlGaAs interface; or from electron
levels in the InAlAs QD’s. As was reported in earlier
studies,11,12,29 interfacial defects have been known to give
strong DLTS signals and are also known to hamper radiative
recombination from QD states. Interfacial defects have been
observed25,26 to have a rapid increase in signal intensity with
applied bias as the bias sweeps over the interfacial region.
An unambiguous identification of the origin of this broad
low-temperature peak that is seen mainly for short filling
pulses in InAlAs/n-AlGaAs QD’s can only be made if the
defect concentration can be reduced to a much lower density
that the QD volumetric density, which will require additional
growth optimization in future work.
The most unusual features seen in Figs. 6 and 7 are the
increased intensity of the low-energy peak with shorter filling pulses, and the shift in activation energies with increased
reverse bias. This field dependence is different from what
would be expected from the Poole Frenkel effect,30 since the
thermal-emission rates decrease with increasing reverse bias.
Changes in peak temperature 共and correspondingly in DLTS
activation energies or emission rates兲 have been observed in
detailed DLTS studies involving DX centers in AlGaAs of
varying ternary composition.31 This activation energy dependence on electric field also had composition dependences.
The tentative explanation given in that report was that multiple closely related peaks were observed, from slightly different defect configurations. If the concentration of a particular defect configuration is somehow field dependent, then
different activation energies can be obtained from different
bias conditions. Our observation of a broader DLTS peak
from this defect共s兲 and its unusual change in peak shape with
applied bias is consistent with this interpretation, and the
peaks shown in Fig. 4 could very well originate from various
slightly different interfacial defect configurations.
The increased intensity of the low temperature peak with
shorter filling pulses in the n-InAlAs/AlGaAs structures can
perhaps be understood from the larger cross sections measured for this peak. If we assume, that as we proposed, the
low-temperature peak shown in Fig. 6 originates from interfacial defect levels in the InAlAs and AlGaAs barrier, DLTS
experiments show that when the filling pulse is short, the
multiple defect centers, or electron traps shown in Fig. 6,
cannot be filled because they have a smaller capture cross
section. The signal seen at lower temperatures, which we
attribute to interfacial defects, is the only one seen for short
filling pulses because carriers are not trapped by the center
with the smallest capture cross section. The fact that this
simple peak at low temperature can only be observed for

short filling pulses, and almost disappears at long filling
pulses is unusual, and seems to indicate population of a defect center at the expense of the other center.
A couple of basic, general results can be concluded from
these experiments. One of them is that defect centers in
quantum dots have much different effects on QD optical
properties depending on their atomic configuration, method
of introduction, and most importantly, the spatial positioning
of these defect centers with respect to QD localization. The
defects caused by irradiation, as proton or electron induced
displacement damage, would be expected to have different
impact on optical properties, since the defects do form randomly and are distributed across the entire semiconductor
chip, including the buffer layer and substrates. Most of these
defects are therefore spatially far apart from the region of
QD wave function confinement, and cannot serve as a recombination center for electrons and holes in the QD’s. On
the other hand, when defects are in close proximity to the
dots, within the dots, or at the dot/barrier interface, their
effects on optical emission are very significant, as is shown
in this work. Another clear finding from these experiments is
that even though DLTS can be sufficiently sensitive to detect
electron and hole levels from QD states, it is in all practicality very difficult to detect them when defect concentrations
near the dots 共either at the dots or barriers兲 exceed the volumetric QD concentration, which as seen here can be easily
achieved.

CONCLUSIONS

Red emitting multilayered InAlAs/AlGaAs quantum dots
have been grown using MBE and inserted in pn diodes and
Schottky barriers. We have shown that defects can have an
important role in the optical and transport properties of multilayered QD structures in the InAlAs/AlGaAs material system, and can dominate the optical properties in these InAlAs/
AlGaAs quantum dots; showing that further refinements in
the growth process or in the passivation of interfacial defects
should be performed before InAlAs/AlGaAs QD’s could be
used for efficient red light emitters. Photoluminescence lifetimes in these dots have been measured to be 6 and 29 ps in
InAlAs QD’s embedded in n and p AlGaAs diodes, respectively. Strong DLTS signals are found in both p- and n-type
structures. In the p-type structure, a defect with an activation
energy E a ⫽0.37 eV in concentrations of 6⫻1016/cm3 was
found. In the n-type samples, a complex peak structure was
found for long filling pulses. Such peak structure is similar to
what has been reported for DX-like centers in AlGaAs and
InAlAs. A broader peak in the n-type samples was identified
and isolated after reducing the filling pulse time. Such peak
exhibits plateaus in the signal intensity when the applied bias
is varied, and shows shifts in activation energy with increased applied bias. This peak has been identified as either a
QD/barrier interfacial defect, or as originating from electron
levels in InAlAs/n-AlGaAs quantum dots. Further work reducing the defect density in these n-type samples is needed
for an unambiguous identification of this peak.

085331-6

PHYSICAL REVIEW B 66, 085331 共2002兲

DEFECT STATES IN RED-EMITTING Inx Al1⫺x As . . .
ACKNOWLEDGMENTS

Part of this research was carried out at the Jet Propulsion
Laboratory, California Institute of Technology, under a con-

1

P. G. Piva, R. D. Goldberg, I. V. Mitchell, D. Labrie, R. Leon, S.
Charbonneau, Z. R. Wasilewski, and S. Fafard, Appl. Phys. Lett.
77, 624 共2000兲.
2
R. Leon, G. M. Swift, B. Magness, W. A. Taylor, Y. S. Tang, K. L.
Wang, P. Dowd, and Y. H. Zhang, Appl. Phys. Lett. 76, 2074
共2000兲.
3
S. Marcinkevicius S, R. Leon, B. Cechavicius, J. Siegert, C.
Lobo, B. Magness, and W. Taylor, Physica B 314, 203 共2002兲.
4
J. Zou, D. J. H. Cockayne, and B. F. Usher, J. Appl. Phys. 73, 619
共1993兲.
5
J. W. Matthews and A. E. Blakeslee, J. Cryst. Growth 27, 118
共1974兲.
6
S. Yu. Shiryaev, F. Jensen, J. L. Hansen, J. W. Petersen, and A. N.
Larsen, Phys. Rev. Lett. 78, 503 共1997兲.
7
R. Leon, S. Chaparro, S. R. Johnson, C. Navarro, X. Jin, Y. H.
Zhang, J. Siegert, S. Marcinkevičius, X. Z. Liao, and J. Zou,
J. Appl. Phys. 共to be published兲.
8
J. M. Gerard, O. Cabrol, and B. Sermage, Appl. Phys. Lett. 68,
3123 共1996兲.
9
K. K. Linder, J. Phillips, O. Qasaimeh, X. F. Liu, S. Krishna, P.
Bhattacharya, and J. C. Jiang, Appl. Phys. Lett. 74, 1355 共1999兲.
10
C. Lobo, Ph.D. dissertation, Australian National University, 2000.
11
H. Benisty, C. M. Sotomayor-Torres, and C. Weisbush, Phys. Rev.
B 44, 10 945 共1991兲.
12
U. Bockelmann, Phys. Rev. B 48, 17 637 共1993兲.
13
S. Anand, N. Carlsson, M-E Pistol, L. Samuelson, and W. Seifert,
Appl. Phys. Lett. 67, 3016 共1995兲.
14
H. L. Wang, F. H. Yang, S. L. Feng, H. J. Zhu, D. Ning, H. Wang,
and X. D. Wang, Phys. Rev. B 61, 5530 共2000兲.

tract with the National Aeronautics and Space Administration. Financial support from the Swedish Foundation for International Cooperation in Research and Higher Education
共STINT兲 is gratefully acknowledged.

15

J. Ibañez, R. Leon, D. T. Vu, S. Chaparro, S. R. Johnson, C.
Navarro, and Y. H. Zhang, Appl. Phys. Lett. 79, 2013 共2001兲.
16
R. Leon, P. M. Petroff, D. Leonard, and S. Fafard, Science 267,
1966 共1995兲.
17
R. Leon, S. Fafard, D. Leonard, J. L. Merz, and P. M. Petroff,
Appl. Phys. Lett. 67, 521 共1995兲.
18
S. Marcinkevičius, J. Siegert, R. Leon, B. Čechavičius, B. Magness, W. Taylor, and C. Lobo 共unpublished兲.
19
S. Marcinkevicius, A. Gaarder, and R. Leon, Phys. Rev. B 64,
115307 共2001兲.
20
S. Raymond, S. Fafard, S. Charbonneau, R. Leon, D. Leonard, P.
M. Petroff, and J. L. Merz, Phys. Rev. B 52, 17 238 共1995兲.
21
D. V. Lang, J. Appl. Phys. 45, 3023 共1974兲.
22
C. Hemmingsson, N. T. Son, O. Kordina, E. Janzen, and J. L.
Lindström, J. Appl. Phys. 84, 704 共1998兲.
23
J. Criado, A. Gomez, E. Calleja, and E. Munoz, Appl. Phys. Lett.
52, 660 共1988兲.
24
M. M. Sobolev, I. V. Kochnev, V. M. Lantratov, N. A. Cherkashin,
and V. V. Emtsev, Physica B 273-274, 959 共1999兲.
25
Y. B. Jia, Z. Y. Han, H. G. Grimmeiss, and L. Dobaczewski, J.
Appl. Phys. 80, 2860 共1996兲.
26
K. J. Choi and J. L. Lee, J. Vac. Sci. Technol. B 19, 615 共2001兲.
27
A. Malinin, H. Tomozawa, T. Hashizume, and H. Hasegawa, Jpn.
J. Appl. Phys. 34, 1138 共1995兲.
28
S. Dueñas, R. Pinacho, E. Castan, L. Quintanilla, R. Pelaez, and J.
Barbolla, J. Appl. Phys. 82, 4338 共1997兲.
29
P. Krispin, R. Hey, H. Kostial, and K. H. Ploog, J. Appl. Phys. 83,
1496 共1998兲.
30
J. Frenkel, Phys. Rev. 54, 647 共1938兲.
31
Y. B. Jia and H. G. Grimmeiss, Phys. Rev. B 47, 1858 共1993兲.

085331-7