Appl. Phys. Lett. 104, 123107 (2014); https://doi.org/10.1063/1.4870087 104, 123107
© 2014 AIP Publishing LLC.
Indirect and direct optical transitions in
In0.5Ga0.5As/GaP quantum dots
Cite as: Appl. Phys. Lett. 104, 123107 (2014); https://doi.org/10.1063/1.4870087
Submitted: 15 January 2014 • Accepted: 20 March 2014 • Published Online: 27 March 2014
G. Stracke, E. M. Sala, S. Selve, et al.
ARTICLES YOU MAY BE INTERESTED IN
Growth and structure of In0.5Ga0.5Sb quantum dots on GaP(001)
Applied Physics Letters 109, 102102 (2016); https://doi.org/10.1063/1.4962273
230 s room-temperature storage time and 1.14 eV hole localization energy in In0.5Ga0.5As
quantum dots on a GaAs interlayer in GaP with an AlP barrier
Applied Physics Letters 106, 042102 (2015); https://doi.org/10.1063/1.4906994
Growth of quantum dots on GaP utilizing a GaAs interlayer
Applied Physics Letters 101, 223110 (2012); https://doi.org/10.1063/1.4768294
Indirect and direct optical transitions in In
0.5
Ga
0.5
As/GaP quantum dots
G. Stracke,
1,a)
E. M. Sala,
1
S. Selve,
2
T. Niermann,
3
A. Schliwa,
1
A. Strittmatter,
1
and D. Bimberg
1
1
Institut f€
ur Festk€
orperphysik, Technische Universit€
at Berlin, Hardenbergstrasse 36, 10623 Berlin, Germany
2
Zentraleinrichtung Elektronenmikroskopie, Technische Universit€
at Berlin, Straße des 17. Juni 135,
10623 Berlin, Germany
3
Institut f€
ur Optik und Atomare Physik, Technische Universit€
at Berlin, Straße des 17. Juni 135, 10623 Berlin,
Germany
(Received 15 January 2014; accepted 20 March 2014; published online 27 March 2014)
We present a study of self-assembled In
0.5
Ga
0.5
As quantum dots on GaP(001) surfaces linking
growth parameters with structural, optical, and electronic properties. Quantum dot densities from
5.0 10
7
cm
2
to 1.5 10
11
cm
2
are achieved. A ripening process during a growth interruption
after In
0.5
Ga
0.5
As deposition is used to vary the quantum dot size. The main focus of this work lies
on the nature of optical transitions which can be switched from low-efficient indirect to
high-efficient direct ones through improved strain relief of the quantum dots by different cap
layers. V
C2014 AIP Publishing LLC.[http://dx.doi.org/10.1063/1.4870087]
The direct-bandgap semiconductor InGaAs represents
the basis of the active area of a multitude of commercial pho-
tonic devices, mostly on GaAs and InP substrates. InGaAs
based quantum dots (QDs), confining charge carriers in all
three dimensions of space largely improve laser properties
like material gain, threshold current density, or temperature
stability.
1
Recently, GaP has attracted research interest
for its potential to realize defect-free III/V semiconductor
heterostructures on silicon substrates.
2,3
Being an indirect
semiconductor, GaP cannot serve as active medium for effi-
cient photonic devices. Consequently, several studies of
InGaAs
4–8
and InGaAsN
9
QDs on GaP have been published
during the last years, including an InGaAs/GaP QD light
emitting diode on Si substrate.
10
The extreme lattice mis-
match of 11.2% between InAs and GaP on the one hand repre-
sents a challenge for the growth of dislocation-free QD
structures. On the other hand, together with a strong quantum
confinement, it produces indirect bandgaps in (In, Ga)As/GaP
nanostructures,
7,11–13
yielding poor luminescence unsuited
for photonic devices. Here, we report on direct-bandgap
In
0.5
Ga
0.5
As/GaP(001) QDs with strong optical emission via
strain engineering by different QD cap layers.
Another promising application are QD-based memory
cells, combining fast access times with long retention times.
14
1.6 s hole retention time were demonstrated for InAs/GaAs
QDs with additional Al
0.9
Ga
0.1
As barrier. In order to increase
retention times decisively, little explored material combina-
tions like InGaAs QDs in (Ga, Al)P must be considered.
15
We
determined the hole retention time in In
0.25
Ga
0.75
As/GaP QDs
to 3 ls,
4
more than 3 orders of magnitude longer than in
InAs/GaAs QDs,
16
validating the potential of In
x
Ga
1x
As/
(Ga, Al)P QDs as a basis for nanomemories. The retention
time of charge carriers in QDs is proportional to the inverse of
their capture cross-section, which varies over 3 orders of mag-
nitude for different QD materials.
15
Yet, methods modifying
the capture cross-section in favor of retention time for a given
material system are still missing. The switching between
indirect and direct QDs in the same material system reported
in this paper allows for future studies of the impact of the
electronic QD structure on the capture processes of charge
carriers into QDs.
In our previous work,
4,17
we demonstrated the growth of
In
0.25
Ga
0.75
As QDs on GaP(001) by metalorganic vapour
phase epitaxy (MOVPE). To initiate the three-dimensional
growth (Stranski–Krastanow mode) of In
x
Ga
1x
As on GaP,
the GaP surface needs to be covered by 2-3 monolayers
(ML) of GaAs prior to In
x
Ga
1x
As deposition. Here, we
present a systematic study of growth parameters affecting
density and size, as well as, emission wavelength and inten-
sity of the QDs, based on photoluminescence (PL) spectros-
copy, atomic force microscopy (AFM), and transmission
electron microscopy (TEM). Finally, we propose a model of
the electronic structure of In
0.5
Ga
0.5
As/2.2 ML GaAs/GaP
QDs based on experimental results and eight-band kp
calculations.
Samples are grown in a horizontal MOVPE reactor on
GaP(001) substrates using H
2
as carrier gas. First, 500 nm
thick undoped GaP buffer layers are grown at a substrate
temperature of 750 C. The temperature is then lowered to
500 C for the growth of GaAs and In
0.5
Ga
0.5
As layers. If
not mentioned otherwise, after deposition of the In
0.5
Ga
0.5
As
layer, growth interruptions (GRI) ranging from 1 s to 400 s
without any precursor supply are applied, followed by the
deposition of 6 nm GaP. Buried QDs for PL and TEM meas-
urements are additionally overgrown by 50 nm GaP at
600 C. The TEM measurements were performed at the FEI
TITAN 80–300 Berlin Holography Special operated at
300 kV.
Fig. 1shows AFM images of samples with varying
In
0.5
Ga
0.5
As coverage on 2.2 ML GaAs/GaP. No additional
cap layers were grown after In
0.5
Ga
0.5
As deposition. For 0.42
ML In
0.5
Ga
0.5
As, the surface has a smooth two-dimensional
structure (Fig. 1(a)). QD formation sets in for deposition of
0.52 ML In
0.5
Ga
0.5
As. QD densities rapidly increase from 5.0
10
7
cm
2
to 1.5 10
11
cm
2
for increasing deposition
from 0.52 ML to 0.90 ML In
0.5
Ga
0.5
As (Fig. 1(d)). Average
a)
E-mail: gernot.stracke@tu-berlin.de
0003-6951/2014/104(12)/123107/4/$30.00 V
C2014 AIP Publishing LLC104, 123107-1
APPLIED PHYSICS LETTERS 104, 123107 (2014)
QD height and base length are 3.1 60.5 nm and 21.7
62.0 nm, respectively. For >1.0 ML In
0.5
Ga
0.5
As coverage
large defective clusters are observed (Figs. 1(c) and 1(d)). The
defect density is 4.2 10
8
cm
2
for 1.14 ML In
0.5
Ga
0.5
As.
These defects have an average height and diameter of
7.6 61.2 nm and 42.3 62.9 nm, respectively.
Room temperature PL spectra of buried In
0.5
Ga
0.5
As
layers with varying thickness grown on 2.2 ML GaAs/GaP
are shown in Fig. 2. For 0.41 ML In
0.5
Ga
0.5
As, the emis-
sion around 685 nm is attributed to a wetting layer (WL)
formed by the GaAs and In
0.5
Ga
0.5
As layers. The PL peak
shifts abruptly to 739 nm for 0.46 ML In
0.5
Ga
0.5
As (Fig.
2(b)), in good agreement with the onset of QD formation
observed in AFM (Fig. 1). Thus, this second peak is attrib-
uted to QD luminescence and the critical In
0.5
Ga
0.5
As cover-
age for QD formation is determined to 0.46 ML. Upon
increasing the In
0.5
Ga
0.5
As coverage to 1.03 ML, the QD lu-
minescence shifts to 800 nm, and the intensity rises, reflect-
ing increasing QD size and density also observed in AFM
(Fig. 1). The integrated PL intensity plotted in the inset of
Fig. 2(a) drops for >1.03 ML In
0.5
Ga
0.5
As indicating
non-radiative recombination caused by the onset of cluster
formation as seen in the AFM images. Fig. 2(c) depicts the
peak wavelength versus In
0.5
Ga
0.5
As coverage.
Fig. 3shows AFM images of two QD samples with
1.27 ML In
0.5
Ga
0.5
As/2.2 ML GaAs/GaP. The structures are
capped by 6 nm GaP after a GRI without any precursor sup-
ply of 10 s and 200 s, respectively. For a GRI of 10 s, only
small modulations <1 nm of the surface are visible, indicat-
ing that the underlying QDs are very small. For a GRI of
200 s, the surface exhibits distinct hills with a height of
1.3 60.4 nm. This implies a ripening process during the
GRI resulting in larger QDs for longer GRIs.
18
Fig. 4
shows a TEM micrograph of a sample containing 0.85 ML
In
0.5
Ga
0.5
As on 2.2 ML GaAs in GaP. The GRI was set to
200 s. The micrograph was obtained under strong-beam dark
field conditions using the {200} reflection in growth direc-
tion. The image intensities under these conditions are inter-
pretable as specimen composition projected along the beam
direction (InGaAs dark; GaP bright). The specimen was
rotated by 13.25with respect to the beam in order to visual-
ize the QD in-plane distribution and avoid overlapping of
several QDs in the micrograph. The QDs exhibit the typical
shape
19
of a truncated pyramid and comparable size as the
not-overgrown QDs.
The impact of the GRI on the QD luminescence is shown
in Fig. 5for QDs without (a) and with (b) an additional 1.6 ML
FIG. 1. AFM images of (a) 0.42 ML, (b) 0.57 ML, and (c) 1.03 ML
In
0.5
Ga
0.5
As deposited on 2.2 ML GaAs on GaP. (d) QD and defect density
versus In
0.5
Ga
0.5
As coverage.
FIG. 2. (a) Room temperature PL spectra of buried In
0.5
Ga
0.5
As layers of
varying thickness on 2.2 ML GaAs/GaP. The In
0.5
Ga
0.5
As is capped by
1.6 ML GaAs and 56 nm of GaP to improve the PL intensity (see Fig. 6).
The curves are offset vertically for better visualization. The inset shows the
integrated PL intensity versus In
0.5
Ga
0.5
As coverage. (b) Spectra for low
In
0.5
Ga
0.5
As coverage of 0 to 0.52 ML reflecting the 2D-3D transition. (c)
Spectral position of WL and QD PL peaks.
FIG. 3. AFM images of 1.27 ML In
0.5
Ga
0.5
As/2.2 ML GaAs/GaP, capped
with 6 nm GaP after a GRI of (a) 10 s, (b) 200 s.
123107-2 Stracke et al. Appl. Phys. Lett. 104, 123107 (2014)
GaAs layer above the QDs. The additional GaAs layer is
grown after the GRI and before the low temperature 6 nm
GaP cap layer and will be motivated below. The insets show
the integrated luminescence versus GRI. The luminescence
of QDs without additional GaAs layer exhibits a rapid evolu-
tion of the QD luminescence intensity with the GRI duration
up to 30 s and remains nearly constant up to GRIs of 300 s.
The drop of the PL intensity for GRIs >300 s indicates the
formation of defects. The luminescence of QDs with addi-
tional GaAs layer increases continuously with the GRI dura-
tion and reaches a maximum for a GRI of 150 s. For longer
GRIs the intensity drops quickly. The luminescence of QDs
with additional GaAs layer shifts from 754 nm for a GRI of
1 s to 828 nm for a GRI of 180 s. Such a red-shift for longer
GRIs is also observed for InAs/GaAs QDs,
18
caused by a
size increase of the QDs. For QDs without additional GaAs
layer, no such wavelength shift is observed. It is noted that
the additional GaAs layer is grown after the GRI and, thus,
cannot influence the QD ripening during the GRI.
Fig. 6shows PL spectra of samples containing buried
0.83 ML In
0.5
Ga
0.5
As/2.2 ML GaAs/GaP QDs capped by 1
ML to 5 ML GaAs after a GRI of 150 s, followed by 56 nm
of GaP. The additional GaAs layer has strong impact on PL
wavelength and intensity. The PL peak shifts from 722 nm
for QDs without additional GaAs layer to 843 nm for QDs
capped with additional 3 ML of GaAs. Such a red-shift can
be attributed to a redistribution of the strain inside the QDs
and was previously observed for InAs/GaAs QDs when over-
grown with In
x
Ga
1x
As
20
(Dot-in-a-Well structures) or
GaAs
x
Sb
1x
layers.
21
The integrated PL intensity shown in the inset of Fig. 6
is 13 times higher for QDs capped with 2 ML of GaAs as
compared to QDs without GaAs capping. The intensity drop
for >2 ML GaAs indicates the formation of defects which
might form due to the additional strain introduced to the
structure by the GaAs layer.
Quantum confinement and strain effects in a QD can
cause a crossing of the electronic level derived from the C
state of the bulk material above those derived from the X
and L states. Tight-binding calculations
7
of InGaAs/GaP
QDs predict that for the QD size found in our work and
an indium amount of 50% the C-derived level lies above the
X- and L-derived levels, resulting in a weak overlap of electron
and hole wave functions. In highly compressively strained
QDs, the lowest electron level can even be pushed above
that of the surrounding matrix, resulting in a type II band
alignment with electrons residing in the matrix and holes in
the QD.
12
If additionally the matrix has an indirect bandgap
like GaP, with the lowest electron states at the X point of the
Brillouin zone, the X level splits at the QD interface due to
strain effects, forming a localized well for electrons at the
interface of QD and matrix (Fig. 7). The resulting bandgap
of the system is indirect in both real and reciprocal space.
Such a switching from a direct to an indirect bandgap along
FIG. 4. Cross-sectional TEM image of 0.85 ML In0.5GaAs/2.2 ML
GaAs/GaP QDs, (strong-beam {200} darkfield). Darker contrasts correspond
to InGaAs rich areas. The specimen was tilted by 13.25with respect to the
beam as indicated by the sketch on the right hand side to avoid imaging
overlapping QDs.
FIG. 5. Room temperature PL spectra
of buried QDs with varying GRI after
In
0.5
Ga
0.5
As deposition. (a) without
and (b) with additional 1.6 ML GaAs
above the QDs. The spectra in (b) are
vertically offset for better visualiza-
tion. The insets show the integrated PL
intensity versus GRI.
FIG. 6. Room temperature PL spectra of buried 0.83 ML In
0.5
Ga
0.5
As/2.2
ML GaAs/GaP QDs with an additional GaAs layer of varying thickness
directly above the QDs. GRI ¼150 s. Dotted lines represent Gaussians fitted
to the data. The spectra are vertically offset for better visualization. The inset
shows the integrated PL intensity versus the additional GaAs layer
thickness.
123107-3 Stracke et al. Appl. Phys. Lett. 104, 123107 (2014)
with a decrease in PL intensity by more than two orders of
magnitude was observed for InP/GaP QDs upon increasing
the pressure on the QDs.
22
In
0.5
Ga
0.5
As/2.2 ML GaAs/GaP QDs without GaAs cov-
erage exhibit luminescence between 1.66 eV and 1.72 eV, in
good agreement with the transition energy of 1.67 eV between
the GaP X valley and the QD hole level determined from
eight-band kpcalculations. For the calculation, a truncated
pyramid shaped In
0.5
Ga
0.5
As/2 ML GaAs/GaP QD with
12 nm base length and 10 ML height was assumed, according
to the size and shape of In
0.25
Ga
0.75
As/3 ML GaAs/GaP QDs
measured by cross-sectional scanning tunneling microscopy.
17
The absence of a size-dependent wavelength shift (Fig. 5)for
this type of QDs supports the interpretation of an indirect type
II band alignment, as the QD size has negligible impact on the
GaP-X-valley to QD-hole-level transition energy.
The overgrowth of the QDs by GaAs allows for strain
relief of the QDs, eventually lowering the C-derived electron
levels in the QD below the X- and L-derived levels and
below the X states in the GaP matrix. Eight-band kpcalcu-
lations yield a direct bandgap type I transition energy of
1.51 eV for In
0.5
Ga
0.5
As QDs encapsulated in GaAs on GaP
substrate. This result agrees very well with the emission at
1.48 eV from In
0.5
Ga
0.5
As/2.2 ML GaAs/GaP QDs with
additional 2 ML GaAs on top. The size-dependent wave-
length shift and the increased PL intensity observed for these
QDs are further indications for a direct type I band align-
ment. Upon capping the QDs with GaAs, the PL peak splits
into a doublet which can be fitted by two Gaussians, as repre-
sented by the dotted lines in Fig. 6for 2 ML GaAs capping,
indicating that additional recombination channels are being
activated or a bimodal QD distribution exists.
In conclusion, we have studied in detail In
0.5
Ga
0.5
As/2.2
ML GaAs/GaP QDs grown by MOVPE. The critical cover-
age for QD formation is determined to 0.46 ML In
0.5
Ga
0.5
As
on 2.2 ML GaAs on GaP. The QD density lies between 5.0
10
7
cm
2
and 1.5 10
11
cm
2
depending on the depos-
ited In
0.5
Ga
0.5
As amount. During a GRI after In
0.5
Ga
0.5
As
deposition, a ripening of the QDs leading to increasing QD
size is observed. Improved relief of the high strain in the
InGaAs/GaP QD system is achieved by introduction of a thin
GaAs layer directly above the QDs. The strain relief results
in a strong red-shift of the QD luminescence and an increase
of the luminescence intensity by more than one order of
magnitude. These effects can be explained by a strain
induced switching between indirect and direct band align-
ment both in real and reciprocal space, thus proving the im-
portance of strain engineering for device fabrication using
InGaAs/GaP QDs.
The authors thank the DFG (Contract No. BI284/29-1),
the Federal Ministry of Economics and Technology (BMWi)
(Grant No. 03VWP0059v), and the Federal Ministry of
Education and Research (Grant No. 16V0196 (HOFUS)).
1
D. Bimberg, Electron. Lett. 44, 168 (2008).
2
A. Beyer, J. Ohlmann, S. Liebich, H. Heim, G. Witte, W. Stolz, and K.
Volz, J. Appl. Phys. 111, 083534 (2012).
3
T. J. Grassman, J. A. Carlin, B. Galiana, L.-M. Yang, F. Yang, M. J. Mills,
and S. A. Ringel, Appl. Phys. Lett. 102, 142102 (2013).
4
G. Stracke, A. Glacki, T. Nowozin, L. Bonato, S. Rodt, C. Prohl, A. Lenz,
H. Eisele, A. Schliwa, A. Strittmatter, U. W. Pohl, and D. Bimberg, Appl.
Phys. Lett. 101, 223110 (2012).
5
R. Leon, C. Lobo, T. P. Chin, J. M. Woodall, S. Fafard, S. Ruvimov, Z.
Liliental-Weber, and M. A. S. Kalceff, Appl. Phys. Lett. 72, 1356–1358
(1998).
6
K. Rivoire, S. Buckley, Y. Song, M. L. Lee, and J. Vucˇkovic´, Phys. Rev.
B85, 045319 (2012).
7
C. Robert, C. Cornet, P. Turban, T. N. Thanh, M. O. Nestoklon, J. Even, J.
M. Jancu, M. Perrin, H. Folliot, T. Rohel, S. Tricot, A. Balocchi, D.
Lagarde, X. Marie, N. Bertru, O. Durand, and A. Le Corre, Phys. Rev. B
86, 205316 (2012).
8
T. Miyamoto, S. Tanabe, R. Nishio, Y. Kobayashi, and R. Suzuki, Jpn. J.
Appl. Phys., Part 1 51, 080201 (2012).
9
K. Umeno, Y. Furukawa, N. Urakami, R. Noma, S. Mitsuyoshi, A.
Wakahara, and H. Yonezu, Phys. E 42, 2772–2776 (2010).
10
Y. Song and M. L. Lee, Appl. Phys. Lett. 103, 141906 (2013).
11
S. Fuchi, Y. Nonogaki, H. Moriya, A. Koizumi, Y. Fujiwara, and Y.
Takeda, Phys. E 21, 36–44 (2004).
12
A. J. Williamson, A. Franceschetti, H. Fu, L. W. Wang, and A. Zunger,
J. Electron. Mater. 28, 414–425 (1999).
13
C. Robert, M. O. Nestoklon, K. P. da Silva, L. Pedesseau, C. Cornet, M. I.
Alonso, A. R. Go~
ni, P. Turban, J.-M. Jancu, J. Even, and O. Durand, Appl.
Phys. Lett. 104, 011908 (2014).
14
A. Marent, T. Nowozin, M. Geller, and D. Bimberg, Semicond. Sci.
Technol. 26, 14026 (2011).
15
T. Nowozin, D. Bimberg, K. Daqrouq, M. N. Ajour, and M. Awedh,
J. Nanomater. 2013, Article ID 215613 (2013).
16
M. Geller, E. Stock, C. Kapteyn, R. Sellin, and D. Bimberg, Phys. Rev. B
73, 205331 (2006).
17
C. Prohl, A. Lenz, D. Roy, J. Schuppang, G. Stracke, A. Strittmatter, U.
W. Pohl, D. Bimberg, H. Eisele, and M. D€
ahne, Appl. Phys. Lett. 102,
123102 (2013).
18
K. P€
otschke, L. M€
uller-Kirsch, R. Heitz, R. L. Sellin, U. W. Pohl, D.
Bimberg, N. Zakharov, and P. Werner, Phys. E 21, 606–610 (2004).
19
A. Lenz, R. Timm, H. Eisele, C. Hennig, S. K. Becker, R. L. Sellin, U. W.
Pohl, D. Bimberg, and M. D€
ahne, Appl. Phys. Lett. 81, 5150 (2002).
20
F. Guffarth, R. Heitz, A. Schliwa, O. Stier, N. Ledentsov, A. Kovsh, V.
Ustinov, and D. Bimberg, Phys. Rev. B 64, 085305 (2001).
21
A. Hospodkov
a, M. Z
ıkov
a, J. Pangr
ac, J. Oswald, K. Kuldov
a, J.
Vyskocˇil, and E. Hulicius, J. Cryst. Growth 370, 303–306 (2013).
22
A. Go~
ni, C. Kristukat, F. Hatami, S. Dreßler, W. Masselink, and C.
Thomsen, Phys. Rev. B 67, 075306 (2003).
FIG. 7. Schematic of the proposed bandstructure for highly strained and
strain relieved InGaAs/GaP QDs.
123107-4 Stracke et al. Appl. Phys. Lett. 104, 123107 (2014)
