Appl. Phys. Lett. 101, 211119 (2012); https://doi.org/10.1063/1.4767525 101, 211119
© 2012 American Institute of Physics.
Electrically driven single photon source
based on a site-controlled quantum dot with
self-aligned current injection
Cite as: Appl. Phys. Lett. 101, 211119 (2012); https://doi.org/10.1063/1.4767525
Submitted: 22 October 2012 • Accepted: 31 October 2012 • Published Online: 20 November 2012
W. Unrau, D. Quandt, J.-H. Schulze, et al.
ARTICLES YOU MAY BE INTERESTED IN
Electrically driven quantum dot-micropillar single photon source with 34% overall efficiency
Applied Physics Letters 96, 011107 (2010); https://doi.org/10.1063/1.3284514
Invited Review Article: Single-photon sources and detectors
Review of Scientific Instruments 82, 071101 (2011); https://doi.org/10.1063/1.3610677
Electrically driven quantum dot single-photon source at 2 GHz excitation repetition rate with
ultra-low emission time jitter
Applied Physics Letters 102, 011126 (2013); https://doi.org/10.1063/1.4774392
Electrically driven single photon source based on a site-controlled quantum
dot with self-aligned current injection
W. Unrau, D. Quandt, J.-H. Schulze, T. Heindel, T. D. Germann, O. Hitzemann,
A. Strittmatter, S. Reitzenstein, U. W. Pohl, and D. Bimberg
Institut f€
ur Festk€
orperphysik, Technische Universit€
at Berlin, Hardenbergstrasse 36, D-10623 Berlin, Germany
(Received 22 October 2012; accepted 31 October 2012; published online 20 November 2012)
Electrical operation of single photon emitting devices employing site-controlled quantum dot (QD)
growth is demonstrated. An oxide aperture acting as a buried stressor structure is forcing site-
controlled QD growth, leading to both QD self-alignment with respect to the current path in vertical
injection pin-diodes and narrow, jitter-free emission lines. Emissions from a neutral exciton, a neutral
bi-exciton, and a charged exciton are unambiguously identified. Polarization-dependent measurements
yield an exciton fine-structure splitting of (84 62) leV at photon energies of 1.28–1.29 eV. Single-
photon emission is proven by Hanbury Brown and Twiss experiments yielding an anti-bunching value
of g
(2)
(0) ¼0.05 under direct current injection. V
C2012 American Institute of Physics.
[http://dx.doi.org/10.1063/1.4767525]
Future quantum-communication systems depend on the
availability of low-cost sources of single polarized photons or
entangled photon pairs. Low-dimensional semiconductor
materials such as quantum dots are suitable candidates for
electrically driven single-photon sources (SPS) since they
show very narrow emission linewidth and can be easily inte-
grated with mainstream semiconductor diode technology.
1–3
Pure single-photon characteristics as expressed by second
order auto-correlation values g
(2)
(0) <0.02 have been demon-
strated.
4
Electrical operation of QD-based single photon emit-
ters has been previously demonstrated with g
(2)
(0) <0.02 and
repetition rates up to 1 GHz.
5–7
Moreover, the exploitation of
quantum electrodynamics such as the Purcell effect in QD-
microcavities allows one to strongly enhance the extraction
efficiency of SPSs and overall efficiencies of up to 34% have
been reported for electrically pumped devices.
8,9
Quantum
optical systems strongly benefit from QDs with their low
excitonic dephasing rates
10
and narrow emission linewidths.
Excitonic transitions exhibiting linewidths down to few leV
at low temperatures are obtained, if QDs are embedded into
defect-free matrix material such that no jitter or spectral diffu-
sion induced by stochastic fluctuations of electric fields in the
QD environment occurs. Thus, QD-based single photon emit-
ters are key elements of future quantum communication sys-
tems and are already used for quantum key distribution.
3,11,12
One important problem of QD-based SPS prohibiting
broad use is related to the random spatial distribution of QDs
on the growth surface resulting from the probabilistic nature
of the self-organized Stranski-Krastanow growth mode, which
is typically used during epitaxy. The SPS device yield is con-
sequently low, and the performance of QD-SPSs remains arbi-
trary as long as QD alignment is not precisely controlled. In
recent years, much effort has therefore been directed towards
so-called site-controlled growth of QDs.
13–18
The common
task is to locally modify the free energy of the growth surface
in order to increase the probability for QD nucleation at pre-
defined sites. Many techniques for site-controlled growth
have been developed which provide the intended determinis-
tic nucleation characteristics. A few groups also demonstrated
electrical operation of diode structures with site-controlled
QDs embedded into the intrinsic region of pin-diodes.
19–21
Best reported values show excitonic linewidths of 170 leV
and photon auto-correlation values with g
(2)
(0) ¼0.4.
21
The surface patterning technique used to achieve site-
controlled growth has a significant impact on the optical
quality of the QDs. Depending on the way of patterning and
the subsequent growth procedure, defects may be created in
close vicinity to optically active QDs. Such defects typically
contribute to severe line broadening due to spectral diffu-
sion.
22
As one possible solution to the problem of patterning-
induced defects we recently proposed and demonstrated site-
control of QDs using a buried stressor layer. This technology
employs the strain field of a buried oxide aperture to trigger
the nucleation of QDs within the aperture. By embedding
QDs into this defect-free environment, their optical proper-
ties have become now comparable to the ultimately high op-
tical quality of QDs randomly grown on planar surfaces.
Additionally, this approach is very attractive for the process-
ing of deterministic devices since perfect self-alignment of
the site-controlled QDs to the vertical current path as con-
fined by the oxide aperture is obtained. In this work, we dem-
onstrate electroluminescence (EL) spectra of single QDs
integrated into a pin-diode structure via buried stressor, fea-
turing linewidths of excitons down to 25 leV and electrically
driven single-photon emission showing anti-bunching with
g
2
(0) ¼0.05.
A two-step metal organic vapor phase epitaxy procedure
is applied to embed site-controlled QDs into a vertical pin-
diode structure as shown in Fig. 1. During the first step, the
bottom part comprising a 20-period n-doped AlGaAs/GaAs
distributed Bragg-reflector (DBR) and the stressor structure
are grown. For the stressor, a 120 nm thick AlGaAs/AlAs/
AlGaAs sandwich structure buried by GaAs is deposited.
Afterwards, arrays of circular mesas with diameters ranging
between 22.0 lm and 24.8 lm in 200 nm steps are realized
by standard photolithography and dry etching. An etching
depth of 300 nm is chosen to yield access to the AlAs layer
via the mesa sidewalls. Subsequently oxide apertures are
0003-6951/2012/101(21)/211119/4/$30.00 V
C2012 American Institute of Physics101, 211119-1
APPLIED PHYSICS LETTERS 101, 211119 (2012)
formed using selective oxidation of the AlAs layer in a fur-
nace under water steam/nitrogen atmosphere at T ¼420 C.
In situ control of the oxidation rate is used to stop the process
when sub-lm apertures are obtained. Arrays of mesa struc-
tures are found to exhibit open apertures within the oxide
layer for d
mesa
>23 lm. A detailed description of this
method has recently been reported.
23
The second epitaxial growth sequence is initiated by oxide
desorption at 720 C using arsine stabilization. Prior to the QD
layer, about 50 nm of GaAs is deposited at 685 C. Parameters
for QD growth are independently calibrated on planar and ref-
erence mesa templates. The InGaAs QD layer and a GaAs cap
layer are grown at 485 C. Except for the layer thickness which
is close to the 2D/3D transition of the Stranski-Krastanow
growth regime, all other parameters are equal to those applied
for QD layers optimized for laser devices.
24,25
The planar QD
density is tuned to below 1 10
8
cm
2
to provide good selec-
tivity between QD nucleation on the aperture region and sur-
rounding areas. After GaAs cap layer deposition, the growth
temperature is raised to 600 C. Then, another 50 nm thick
undoped GaAs layer and a 110 nm p-doped GaAs contact layer
are grown.
Important for site-controlled nucleation of QDs is the
tensile surface strain within the aperture region, which
reduces the lattice mismatch between InGaAs/GaAs. Tensile
strain is generated upon volume reduction of the oxidized
layer. This lattice distortion of 1%–2% lowers locally the
free energy for QD growth, thus preferring QD nucleation
above the oxide apertures.
23
By choosing an appropriate
layer thickness and aperture size, precise control of the QD
number within the aperture region is possible.
For device fabrication, each mesa diameter is reduced
by 5 lm via dry etching, which removes any parasitic depos-
its on mesa side-walls and also QDs located at the mesa edge
where an area of parasitic QD nucleation is present. The
structure further comprises a backside n-contact metalliza-
tion. Front-side insulation using SiN
x
is applied prior to top-
side p-contact metallization with circular openings of around
15 lm aligned to the mesa center. Fig. 1(a) illustrates sche-
matically the cross section of the device structure. As dis-
played, the aperture within the oxide layer defines
simultaneously the QD position and the vertical current path
through the device. Thereby, optimum charge carrier injec-
tion into a target QD is achieved. Fig. 1(b) depicts a top view
of a fully processed device taken by a scanning electron
microscope (SEM).
Optical characterization of the device was performed in a
high-resolution micro-electroluminescence (l-EL) setup at low
temperatures (10–50 K). Bonding of devices onto chip carriers
is required prior to mounting into a continuous-flow helium
cryostat. The emission is collected through a 20microscope
objective with a numerical aperture of 0.4 providing a focal
FIG. 1. (a) Schematic cross section of a device indicating QD position, aperture size, and current flow (green lines); (b) SEM image of the top side of the de-
vice area. In view: uncovered semiconductor, p-contact metallization, and several circular topographic profiles from first mesa etch, mesa diameter reduction,
and SiN
x
passivation opening; (c) EL emission as recorded by a digital camera using an IR filter; (d) electroluminescence spectra recorded from the aperture
region as a function of the injection current. Emission from excitons (X), bi-excitons (XX), and charged excitons (X
) are labeled.
211119-2 Unrau et al. Appl. Phys. Lett. 101, 211119 (2012)
spot size of about 3 lm. The l-EL emission is dispersed by a
spectrometer with a focal length of 0.75 m and a spectral reso-
lution of 25 leV. Coarse and fine positioning of devices is
achieved by a combination of stepper motors and piezoelectric
actuators. Photon auto-correlation measurements are per-
formed using a fiber coupled Hanbury-Brown and Twiss
(HBT) configuration based on a 50:50 multimode fiber beam
splitter and two Si-avalanche photo diodes with 0.8 ns timing
resolution.
Fig. 1(c) shows low-temperature EL of a device at a cur-
rent of 1 lA recorded by a digital camera attached to the
l-EL setup. The emission is filtered by a 930 nm edge-type
optical filter to suppress contributions at higher energies.
The total output area of the device is indicated by a dashed
white circle. As can be seen, long-wavelength (>930 nm)
emission is generated from a region above the current confin-
ing aperture (in the following it is referred to as aperture
region) which is also the area of site-controlled QD growth.
Diode-like current-voltage characteristics with an onset volt-
age of V
ON
¼1.4 V and a reverse saturation current <0.1 nA
are found (not shown). From overview spectra taken between
1.270 and 1.320 eV the onset of QD emission at injection
currents of 0.01 lA is determined to be located in a spectral
window between 1.280 eV and 1.295 eV. Fig. 1(d) shows
successive l-EL spectra in the spectral range of 1.275–
1.305 eV taken from the aperture region for increasing cur-
rents from 0.01 lA to 0.60 lA at a temperature of 12.5 K.
For the device under test with a mesa diameter of 24.6 lman
aperture size of 0.2–0.4 lm is estimated from the oxidation
rate and time.
In Fig. 2(a), individual l-EL spectra at different injection
currents are shown in a waterfall presentation for better dis-
play of the power dependence of individual lines. At an injec-
tion current of 0.01 lA, only one excitonic transition is visible
whereas up to nine optical transitions are observed at higher
injection currents (0.60 lA). The low total number of emission
lines is attributed to a very low QD number directly above the
aperture region. All lines exhibit narrow spectral linewidths
below 50 leV, with best values of 25 leV (still resolution lim-
ited), which constitutes a major progress as compared to previ-
ous reports on site-controlled QD growth.
13–17,20,21
The high
spectral purity of the emission lines also facilitated the identifi-
cation of several characteristic electronic transitions associ-
ated with the same QD, namely, the bi-exciton (XX), the
neutral exciton (X), and the negatively charged exciton
(X
).
26
Moreover, fine structure splitting (FSS) of the exciton
level system is well resolved. At 0.01 lA current injection
level, a doublet transition at 1.2886 eV is dominant. A second
doublet at 1.2893 eV appears at around 0.10 lA which shows a
superlinear intensity increase with current, whereas the
1.2886 eV transition saturates for currents above 0.1 lA. Such
behavior is typical for exciton and bi-exciton emission of a sin-
gle QD.
The relationship between these transitions is further ana-
lyzed by polarization-dependent measurements shown in
Fig. 2, which reveals the paired nature of the XX-X cascade.
These findings ultimately identify the low-energy doublet as
exciton (X) and the high-energy doublet as anti-binding bi-
exciton (XX) transitions. Correspondingly, the bi-exciton
binding energy of 0.72 meV and fine structure splitting of
the exciton levels of (84 62) leV are extracted. Both values
are in very good agreement with a statistical analysis of QDs
for which a correlation of QD size and shape to ground-state
optical transition levels was established.
26
The specific val-
ues for the exciton transition energy and bi-exciton binding
energy suggest a 3 ML high QD with 10 nm base length.
Spectrally stable transition energies enable to associate the
third line at 1.2817 eV to the same QD by comparing the spa-
tial dependence of emission intensity of all lines with regard
to objective lens position. Neither FSS nor polarization de-
pendence as seen in Fig. 2(c) is characteristic of this line
indicating recombination of charged excitons. Typically, a
red shift of the emission wavelength relative to correspond-
ing exciton emission is associated with negatively charged
excitons (X
).
26
The binding energy of the charged exciton
state is 6.96 meV which also agrees well with the estimated
size and shape of the QD. A remarkably low resolution lim-
ited linewidth of only 25 leV is measured for the X
line
confirming that the QD is embedded in a defect-free matrix
environment.
Narrow linewidth, spectral separation from other transi-
tions, and high intensity make the X
emission very attrac-
tive for generation of single photons on demand. Analysis of
FIG. 2. (a) Injection current dependent l-EL spectra. Emission lines from
exciton (X), bi-exciton (XX), and charged exciton (X
) transitions are indi-
cated; (b) high resolution spectrum at 0.6 lA current injection level with
annotations of linewidths and binding energies. Insets (c) and (d) show the
polarization dependence.
211119-3 Unrau et al. Appl. Phys. Lett. 101, 211119 (2012)
photon statistics on the X
line is performed with a fiber
coupled HBT setup. A fit of the second order autocorrelation
function g
(2)
(s) to the experimental data is shown in Fig. 3as
red line. Clear photon anti-bunching with g
(2)
(0) ¼0.05 is
obtained at injection current of 0.32 lA. It is noteworthy that
single photon emission with such high purity has not yet
been reported for site-controlled QDs. For electrically oper-
ated QD-based single photon emitters, the result compares
well to best values of g
(2)
(0) ¼0.02 measured on devices
with randomly distributed QDs.
27
Moreover, future combin-
ing our device concept with suitable microcavity structures
will enable the development of deterministic SPS with high
outcoupling efficiency.
In conclusion an electrically driven single photon source
based on site-controlled QDs is presented. Perfect self-
alignment of QDs to current injection path is obtained using
a buried stressor technology. Very narrow, resolution-limited
emission linewidths down to 25 leV and pure single photon
emission yielding g
(2)
(0) ¼0.05 are observed. Several exci-
tonic features originating from the same QD are resolved
by power- and polarization-dependent l-electroluminescence
spectroscopy.
The authors acknowledge E. Schneiderwind and P.
Moser for technical assistance and SFB 787 of DFG for its fi-
nancial support.
1
P. Michler, A. Imamoglu, M. D. Mason, P. J. Carson, G. F. Strouse, and S.
K. Buratto, Nature (London) 406, 968 (2000).
2
V. Zwiller, T. Aichele, and O. Benson, New J. Phys. 6, 96 (2004).
3
D. Bimberg, E. Stock, A. Lochmann, A. Schliwa, J. A. T€
offlinger, W.
Unrau, M. M€
unnix, S. Rodt, V. A. Haisler, A. I. Toropov, A. Bakarov, and
A. K. Kalagin, IEEE Photon. J. 1, 58 (2009).
4
A. J. Bennett, D. C. Unitt, A. J. Shields, P. Atkinson, and D. A. Ritchie,
Opt. Express 13, 7772 (2005).
5
A. Lochmann, E. Stock, J. A. T€
offlinger, W. Unrau, A. Toropov, A.
Bakarov, V. Haisler, and D. Bimberg, Electron. Lett. 45, 566 (2009).
6
A. J. Shields, Nat. Photonics 1, 215 (2007).
7
E. Stock, W. Unrau, A. Lochmann, J. A. T€
offlinger, M. €
Ozturk, A. I. Toro-
pov, A. K. Bakarov, V. A. Haisler, and D. Bimberg, Semicond. Sci. Tech-
nol. 26, 014003 (2011).
8
T. Heindel, C. Schneider, M. Lermer, S. H. Kwon, T. Braun, S. Reitzen-
stein, S. H€
ofling, M. Kamp, and A. Forchel, Appl. Phys. Lett. 96, 011107
(2010).
9
S. Reitzenstein and A. Forchel, J. Phys. D 43, 033001 (2010).
10
P. Borri, S. Schneider, W. Langbein, and D. Bimberg, J. Opt. A 8, S33
(2006).
11
E. Waks, K. Inoue, C. Santori, D. Fattal, J. Vuckovic, G. S. Solomon, and
Y. Yamamoto, Nature (London) 420, 762 (2002).
12
T. Heindel, C. A. Kessler, M. Rau, C. Schneider, M. F€
urst, F. Hargart, W.
M. Schulz, M. Eichfelder, R. Rossbach, S. Nauerth, M. Lermer, H. Weier,
M. Jetter, M. Kamp, S. Reitzenstein, S. H€
ofling, P. Michler, H. Weinfurter,
and A. Forchel, New J. Phys. 14, 083001 (2012).
13
C. Taylor, E. Marega, E. A. Stach, G. Salamo, L. Hussey, M. Munoz, and
A. Malshe, Nanotechnology 19, 015301 (2008).
14
P. Atkinson, O. G. Schmidt, S. P. Bremner, and D. A. Ritchie, C. R. Phys.
9, 788 (2008).
15
J. Skiba-Szymanska, A. Jamil, I. Farrer, M. B. Ward, C. A. Nicoll, D. J. P.
Ellis, J. P. Griffiths, D. Anderson, G. A. C. Jones, D. A. Ritchie, and A. J.
Shields, Nanotechnology 22, 065302 (2011).
16
M. Felici, P. Gallo, A. Mohan, B. Dwir, A. Rudra, and E. Kapon, Small 5,
938 (2009).
17
A. Huggenberger, S. Heckelmann, C. Schneider, S. Hofling, S. Reitzen-
stein, L. Worschech, M. Kamp, and A. Forchel, Appl. Phys. Lett. 98,
131104 (2011).
18
M. H. Baier, S. Watanabe, E. Pelucchi, and E. Kapon, Appl. Phys. Lett.
84, 1943 (2004).
19
M. H. Baier, C. Constantin, E. Pelucchi, and E. Kapon, Appl. Phys. Lett.
84, 1967 (2004).
20
M. Mehta, D. Reuter, A. Melnikov, A. D. Wieck, S. M. de Vasconcel-
los, T. Baumgarten, A. Zrenner, and C. Meier, Physica E 42, 2749
(2010).
21
C. Schneider, T. Heindel, A. Huggenberger, T. A. Niederstrasser, S. Reit-
zenstein, A. Forchel, S. H€
ofling, and M. Kamp, Appl. Phys. Lett. 100,
091108 (2012).
22
V. T€
urck, S. Rodt, O. Stier, R. Heitz, R. Engelhardt, U. W. Pohl, D. Bim-
berg, and R. Steingr€
uber, Phys. Rev. B 61, 9944 (2000).
23
A. Strittmatter, A. Holzbecher, A. Schliwa, J. H. Schulze, D. Quandt, T.
D. Germann, A. Dreismann, O. Hitzemann, E. Stock, I. A. Ostapenko, S.
Rodt, W. Unrau, U. W. Pohl, A. Hoffmann, D. Bimberg, and V. Haisler,
“Site-controlled quantum dot growth on buried oxide stressor layers,”
Phys. Status Solidi A.
24
R. L. Sellin, I. Kaiander, D. Ouyang, T. Kettler, U. W. Pohl, D. Bimberg,
N. D. Zakharov, and P. Werner, Appl. Phys. Lett. 82, 841 (2003).
25
A. Strittmatter, T. D. Germann, T. Kettler, K. Posilovic, J. Pohl, U. W.
Pohl, and D. Bimberg, J. Cryst. Growth 310, 5066 (2008).
26
S. Rodt, A. Schliwa, K. P€
otschke, F. Guffarth, and D. Bimberg, Phys. Rev.
B71, 155325 (2005).
27
R. B. Patel, A. J. Bennett, K. Cooper, P. Atkinson, C. A. Nicoll, D. A.
Ritchie, and A. J. Shields, Nanotechnology 21, 274011 (2010).
FIG. 3. Trace of the photon auto-correlation measurement signal obtained
from the X
emission. The red line is a fit of the second-order auto-correla-
tion function g
(2)
(s) to raw data.
211119-4 Unrau et al. Appl. Phys. Lett. 101, 211119 (2012)
