Optical properties and modal gain of InGaN quantum dot stacks

Abstract

We present investigations of the optical properties of stacked InGaN quantum dot layers and demonstrate their advantage over single quantum dot layer structures. Measurements were performed on structures containing a single layer with quantum dots or threefold stacked quantum dot layers, respectively. A superlinear increase of the quantum dot related photoluminescence is detected with increasing number of quantum dot layers while other relevant GaN related spectral features are much less intensive when compared to the photoluminescence of a single quantum dot layer. The quantum dot character of the active material is verified by microphotoluminescence experiments at different temperatures. For the possible integration within optical devices in the future the threshold power density was investigated as well as the modal gain by using the variable stripe length method. As the threshold is 670 kW/cm2 at 13 K, the modal gain maximum is at 50 cm–1. In contrast to these limited total values, the modal gain per quantum dot is as high as 10–9cm–1, being comparable to the IIVI and III-As compounds. These results are a promising first step towards bright low threshold InGaN quantum dot based light emitting devices in the near future (© 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)

Full text

Page 1

arXiv:0812.0301v1 [cond-mat.mes-hall] 1 Dec 2008
Optical Properties and Modal Gain of
InGaN Quantum Dot Stacks
Joachim Kalden, Kathrin Sebald, and J¨ urgen Gutowski
Christian Tessarek, Timo Aschenbrenner, Stephan Figge, and Detlef Hommel
Institute of Solid State Physics, University of Bremen
P.O. Box 330 440, 28334 Bremen, Germany
accepted 2008/11/27 for publication in physica status solidi (c)
Abstract
We present investigations of the optical properties of stacked InGaN quantum dot
layers and demonstrate their advantage over single quantum dot layer structures.
Measurements were performed on structures containing a single layer with quantum
dots or threefold stacked quantum dot layers, respectively. A superlinear increase of
the quantum dot related photoluminescence is detected with increasing number of
quantum dot layers while other relevant GaN related spectral features are much less
intensive when compared to the photoluminescence of a single quantum dot layer.
The quantum dot character of the active material is verified by microphotolumi-
nescence experiments at different temperatures. For the possible integration within
optical devices in the future the threshold power density was investigated as well
as the modal gain by using the variable stripe length method. As the threshold is
670kW/cm2at 13K, the modal gain maximum is at 50cm-1. In contrast to these
limited total values, the modal gain per quantum dot is as high as 10-9cm-1, being
comparable to the II-VI and III-As compounds. These results are a promising first
step towards bright low threshold InGaN quantum dot based light emitting devices
in the near future.
1 Introduction
Semiconductor quantum dots (QDs) have attracted much interest in recent years due to
the high stability of their emission properties when changing external conditions [1, 2].
Due to the three-dimensional electron confinement, QDs as active material allow for
the realization of low-threshold light emitting and laser diode structures [3, 4, 5]. Fur-
thermore, it is possible to employ the quantum-confined Stark effect (QCSE) occurring
due to the built-in piezo-electric field [6, 7] to shift the PL to longer wavelengths, since
1

Page 2

the transparency carrier density is reduced compared to quantum well (QW) based de-
vices [8]. However, it is necessary to increase the QD density by using stacked layers to
achieve sufficient active material for appropriate lasing operation. This method has been
very successful for the group-III arsenides [9, 10] and the II-VI wide bandgap semicon-
ductors [11]. In addition, stacking of the QD layers is expected to result in an increasing
uniformity of the QD size throughout the layering process [12], resulting in an increasing
modal gain [13].
2 Sample structure and experimental setup
The investigated samples were grown by metal-organic vapor phase epitaxy. On the
(0001)-sapphire substrate a 2µm thick GaN buffer layer was grown, followed by a 540nm
Al0.12Ga0.88N cladding layer. Subsequently, the lower waveguide layer was realized by
104nm GaN, followed by the threefold stack of InGaN QD layers. Every QD layer
was grown by a two-step method [14], applying a 1.5nm nucleation layer consisting of
In0.18Ga0.82N overgrown by a 5nm formation layer of In0.08Ga0.92N, and finally capped
by 14nm GaN. The GaN capping of the top QD layer is 117nm thick, representing the
upper waveguide layer as well. The upper cladding is omitted since the index contrast
from GaN (n ≈ 2.5) to air (n ≈ 1) is sufficiently large.
Reference samples without cladding layers are investigated via microphotoluminescence
(µPL) experiments with a spatial resolution of about 2µm given by the laser spot di-
ameter. To further reduce the active area, mesa structures are processed by focused ion
beam (FIB) milling, which then can be individually studied via µPL. The excitation
source is a HeCd laser operating in cw mode at 325nm (3.815eV). A reflecting micro-
scope objective with a numerical aperture of 0.5 is used to focus the laser beam on the
sample and to detect its PL signal. The threshold of the waveguide structures is deter-
mined by power density dependent measurements while the modal gain was deduced via
the variable stripe length (VLS) method [15, 16]. For these experiments, the excitation
source is a XeCl excimer laser operating at 308nm (4.025eV) in pulsed mode with a
pulse width of 5ns pumping a dye laser operating at 360nm (3.444eV). To achieve a
well defined stripe-like spot as demanded for VLS experiments, the beam is shaped via
an rectangular pin hole and a cylindrical lens. The variable stripe length is conrolled by
stepper-driven razor blades in the range from 0 – 500µm.
3 Results and discussion
Photoluminescence measurements are presented in Fig. 1 for a single QD layer sample as
well as a threefold QD stack. Both spectra show a modulation of the QD emission band
which is caused partly by Fabry-Perot oscillations between the interfaces air/GaN and
GaN/sapphire. Additionally, the spectral position of each layer in the stacked sample
may slightly differ which could also yield some local maxima in the QD emission band.
For the stacked sample the QD emission intensity dominates the PL spectrum relative
to the donor-bound exciton (D0X) and the donor-acceptor-pair (DAP) recombination
2

Page 3

?
???
?
?
?
?
?
?
Figure 1: Comparison of the PL of a single QD layer (top) and a threefold QD stack
(bottom). The QD PL is significantly increased for the stack, while D0X and
DAP intensities are less intensive than for the single layer (see text).
band intensity. In order to quantitatively describe the relative intensity of the QD PL,
the integrated PL intensities of D0X and QD emission for both samples are compared.
As we find the QD emission to be 3 times more intensive than the D0X for the single
layer, it is determined to be 65 times stronger for the QD stack. This large increase
cannot be explained just by the superposition of three independent QD layers, thus it
may be due to a decreased exciton lifetime reported in vertically coupled QD stacks [17]
leading to a relatively higher PL intensity. For QD stacks, it is further reported that the
QD size slowly increases layer by layer, leading to an overall redshift of the QD PL [18]
as seen in Fig. 1. However, up to now no transmission electron microscopy studies have
been performed to prove this assumption of vertical coupling.
When the active area of the stack sample is reduced to about 1.2µm2via FIB, µPL
reveals that the QD ensemble emission band splits up, and several sharp lines become
visible at 4K, indicating the QD origin of the PL emission. Temperature dependent
3

Page 4

?
?
?
?
???
?
?
?
?
?
???
?
?
?
?
?
???
?
Figure 2: a) PL of an active area of about 1.2µm2for different temperatures. b) Arrhe-
nius plot of QD band maximum [19]. Eais the activation energy.
experiments show that the emission can still be traced up to room temperature for this
small amount of active material (Fig. 2a), demonstrating the high PL stability with
respect to temperature increase of the stacked QD layers.
activation energy of about 24meV deduced by using a thermal activation model [19] as
shown in Fig. 2b.
This is confirmed by an
Raising the excitation power density, the PL intensity of the stack sample increases
decently up to a critical excitation power density usually called threshold density. A
further raise of the excitation power density leads to a steeper increase (Fig. 3). The PL
emission also gets spectrally narrower and stimulated emission sets in (not shown here).
From measurements with different excitation power densities, this threshold density is
deduced to 670kW/cm2at 13K. For higher temperatures, this value is expected to be
increased due to nonradiative recombination processes such as electron-phonon scattering
which raise the losses that have to be overcome before optical gain is achieved. Also the
crystalline quality of the GaN spacer layers grown at low temperatures in order to not
dissolve the QDs could raise the threshold. Although the threshold power density for
QD based devices is predicted [4] and for the group-III arsenides demonstrated [20] to be
much smaller than for QW devices, it is here found to be still higher compared to InGaN
QW structures [21]. This may be due to the fact that the QD layers contain much less
active material than QW structures, since the QD density is limited to 5×109cm-2per
layer. For CdSe QD stacks containing five layers with a QD density of up to 1011cm-2per
layer, stimulated emission was demonstrated for 32 to 60 kW/cm-2[11] while threshold
4

Page 5

??
???
?
?
?
???
Figure 3: Power density dependent PL intensity at 13K. The threshold power density
Pthis found to be 670kW/cm2.
densities of below 1kW/cm-2were recently reported for arsenide based structures at
room temperature [22].
The modal gain of the waveguide structure is investigated via the VLS method. A
rectangular laser spot with a fixed width and a variable length is focussed on the sample.
The PL is detected for several stripe lengths. From these data the modal gain can be
deduced [15, 16]. For guaranteeing success of the VLS method, saturation effects have to
be avoided. One reason for saturation effects can be that the stripe length becomes too
large [23, 24]. In this case the light generated on the stripe end opposite to the emitting
edge will no longer be amplified since all excited states have already been depleted by
photons coming from nearer distances.
Here, the modal gain maximum is gmod= 50cm-1at 12K, and the saturation stripe
length is measured to Ls= 150µm. These values are mainly limited by the background
absorption visible at lower energies where no absorption should occur in the ideal case.
The product of modal gain maximum and saturation length is gmodLs = 0.75, which
5

Page 6

?
?
?
?
?
?
?
Figure 4: Modal gain spectrum at 12K calculated from VLS measurements with stripe
lengths up to 150µm.
is roughly one order of magnitude less compared to QW structures, where the product
has been measured to be 4 – 10 [23, 25]. InGaN QWs reach values of gmod= 180 –
300cm-1and Ls= 250 – 350µm [21]. However, as the QD density is taken into account
by calculating the modal gain per QD, it is found that our results are comparable with
established materials [11], yielding 10-9cm-1per QD. This gives evidence to the fact that
rather the surrounding structure needs further improvement as well as the filling factor
being still very low. For structures with higher QD densities up to 9×1010cm-2per
layer, a significantly lower threshold has recently been presented [26]. Hence it should
be possible to improve our results and achieve an even higher modal gain by increasing
the QD density in our samples.
4 Conclusions
Gallium nitride based waveguide structures containing stacked quantum dot layers as
active region grown by metal-organic vapor-phase epitaxy were presented and analyzed
6

Page 7

in detail. Photoluminescence experiments documented the quantum dot character of the
stacked layers’ spectral features. Temperature dependent measurements revealed a slight
variation of the photoluminescence up to room temperature, which is a crucial fact for
commercial applications in the future. The threshold density of the structure was found
to be 670kW/cm2. Modal gain of up to 50cm-1was detected at low temperatures while
a saturation length of 150µm was identified. Nevertheless, the modal gain per quantum
dot of 10-9cm-1is found to be in the same regime as reported for the well established II-
VI and III-As compounds. The findings presented are promising for future applications
of light emitting devices based on InGaN quantum dot stacks. Next challenges are a
further increase of the filling factor by stacking more layers and increasing the density per
layer as well as further improvement on the structural quality to lower the background
absorption in the structure. Another subsequent step in the future is the embedding
of nitride based QD stacks in vertical emitting microcavity structures to combine the
advantages of quantum dot stacks with features like a symmetric beam profile as well as
single mode operation, the latter leading to a further threshold reduction [27].
Acknowledgements
This work has been supported by the Deutsche Forschungsgemeinschaft in the framework of the
Research Group 506: ”Physics of Nitride-Based Nanostructured Light-Emitting Devices”.
References
[1] M. Senes, K.L. Smith, T.M. Smeeton, S.E. Hooper, and J. Heffernan,
Phys. Rev. B 75, 045314 (2007).
[2] K. Sebald, H. Lohmeyer, S. Herlufsen, J. Kalden, J. Gutowski, C. Tes-
sarek, T. Yamaguchi, and D. Hommel, phys. stat. sol. (c) 5, 1883 (2008).
[3] Y. Arakawa and H. Sakaki, Appl. Phys. Lett. 40, 939 (1982).
[4] M. Asada, Y. Miyamoto, and Y. Suematsu, IEEE J. Quantum Electron. 22,
1915 (1986).
[5] C. Kruse, S. Figge, H. Dartsch, C. Tessarek, D. Hommel, H. Lohmeyer,
J. Kalden, K. Sebald, and J. Gutowski, phys. stat. sol. (c) 5, 2320 (2008).
[6] W.W. Chow, M. Kira, and S.W. Koch, Phys. Rev. B 60, 1947 (1999).
[7] T. Bretagnon, P. Lefebvre, P. Valvin, R. Bardoux, T. Guillet, T. Tal-
iercio, B. Gil, N. Grandjean, F. Semond, B. Damilano, A. Dussaigne,
and J. Massies, Phys. Rev. B 73, 113304 (2006).
[8] W.W. Chow and H.C. Schneider, Appl. Phys. Lett. 81, 2566 (2002).
7

Page 8

[9] J. Bloch, J. Shah, L. Pfeiffer, K. West, and S. Chu, Appl. Phys. Lett. 77,
2545 (2000).
[10] O. Schmidt, N. Kirstaedter, N. Ledentsov, M.H. Mao, D. Bimberg,
V. Ustinov, A. Egorov, A. Zhukov, M. Maximov, P. Kop’ev, and
Z. Alferov, Electron. Lett. 32, 1302 (1996).
[11] K. Sebald, P. Michler, J. Gutowski, R. Kr¨ oger, T. Passow, M. Klude,
and D. Hommel, phys. stat. sol. (a) 190, 593 (2002).
[12] E. Mateeva, P. Sutter, J.C. Bean, and M.G. Lagally, Appl. Phys. Lett.
71, 3233–3235 (1997).
[13] Y. Qiu, P. Gogna, S. Forouhar, A. Stintz, and L.F. Lester, Appl. Phys.
Lett. 79, 3570–3572 (2001).
[14] T. Yamaguchi, K. Sebald, H. Lohmeyer, S. Gangopadhyay, J. Falta,
J. Gutowski, S. Figge, and D. Hommel, phys. stat. sol. (c) 3, 3955 (2006).
[15] K.L. Shaklee, R.E. Nahory, and R.F. Leheny, J. Lumin. 7, 284 (1973).
[16] M. R¨ owe, P. Michler, J. Gutowski, V. K¨ ummler, A. Lell, and V. H¨ arle,
phys. stat. sol. (a) 200, 135 (2003).
[17] N.N. Ledentsov, V.A. Shchukin, M. Grundmann, N. Kirstaedter,
J. Bhrer, O. Schmidt, D. Bimberg, V.M. Ustinov, A.Y. Egorov, A.E.
Zhukov, P.S. Kopev, S.V. Zaitsev, N.Y. Gordeev, Z.I. Alferov, A.I.
Borovkov, A.O. Kosogov, S.S. Ruvimov, P. Werner, U. Gsele, and
J. Heydenreich, Phys. Rev. B 54, 8743 (1996).
[18] A. Hospodkov, V. Kpek, K. Kuldov, J. Humlek, E. Hulicius, J. Oswald,
J. Pangrc, and J. Zeman, Phys. E 36, 106–113 (2007).
[19] G. Bacher, H. Schweizer, J. Kovac, A. Forchel, H. Nickel, W. Schlapp,
and R. L¨ osch, Phys. Rev. B 43, 9312 (1991).
[20] G. Liu, A. Stintz, H. Li, K. Malloy, and L. Lester, Electronics Letters 35,
1163–1165 (1999).
[21] T. Swietlik, G. Franssen, C. Skierbiszewski, R. Czernecki, P. Wis-
niewski, M. Krysko, M. Leszczynski, I. Grzegory, P. Mensz, S. Jursenas,
T. Suski, and P. Perlin, Semicond. Sci. Technol. 22, 736–741 (2007).
[22] T.D. Germann, A. Strittmatter, J. Pohl, U.W. Pohl, D. Bimberg,
J. Rautiainen, M. Guina, and O.G. Okhotnikov, Appl. Phys. Lett. 93,
051104–3 (2008).
[23] M. Vehse, P. Michler, O. Lange, M. R¨ owe, J. Gutowski, S. Bader, H.J.
Lugauer, G. Bruderl, A. Weimar, A. Lell, and V. H¨ arle, Appl. Phys.
Lett. 79, 1763 (2001).
8

Page 9

[24] K. Kyhm, R.A. Taylor, J.F. Ryan, T. Someya, and Y. Arakawa, Appl.
Phys. Lett. 79, 3434–3436 (2001).
[25] J. Mickevicius, G. Tamulaitis, M.S. Shur, Q. Fareed, J.P. Zhang, and
R. Gaska, J. Appl. Phys. 99, 103513 (2006).
[26] Q. Wang, T. Wang, J. Bai, A.G. Cullis, P.J. Parbrook, and F. Ranalli,
J. Appl. Phys. 103, 123522–7 (2008).
[27] K.A. Shore and M. Ogura, Opt. Quantum Electron. 24, S209–S213 (1992).
9