Materials Letters 57 (2003) 4218 – 4221
www.elsevier.com/locate/matlet
A novel method to realize InGaN self-assembled quantum dots by
metalorganic chemical vapor deposition
Liang-Wen Ji a,*, Yan-Kuin Su a, Shoou-Jinn Chang a, Liang-Wen Wu a,b, Te-Hua Fang c,
Qi-Kun Xue d, Wei-Chi Lai a, Yu-Zung Chiou a
a
Institute of Microelectronics and Department of Electrical Engineering, National Cheng Kung University, Tainan 701, Taiwan, ROC
b
South Epitaxy Corporation, Hsin-Shi 744, Taiwan, ROC
c
Department of Mechanical Engineering, Southern Taiwan University of Technology, Yong-Kan 710, Taiwan, ROC
d
Institute of Physics, Chinese Academy of Science, Beijing 100080, PR China
Received 19 August 2002; received in revised form 5 March 2003; accepted 16 March 2003
Abstract
We report the use of an interrupted growth method in metalorganic chemical vapor deposition (MOCVD) to control the
growth of InGaN layers and to grow nanoscale InGaN self-assembled quantum dots (QDs). With a 12-s growth interrupt, we
successfully formed InGaN QDs with a typical lateral size of 25 nm and an average height of 4.1 nm. The QDs density is about
2 1010 cm 2. Strong photoluminescence (PL) emission of InGaN nanostructure was observed at a room temperature with a
full-width-half-maximum (FWHM) of about 92 meV. These results suggest that such QDs are potentially useful in nitride-based
optoelectronic devices.
D 2003 Elsevier Science B.V. All rights reserved.
Keywords: InGaN; Quantum dot; Atomic force microscopy; PL; FWHM
III –V nitride semiconductor materials have a wurtzite crystal structure and a direct energy band gap. At
room temperature, the band gap energy of AlInGaN
varies from 1.95 to 6.2 eV depending on its composition. Therefore, III – V nitride semiconductors are
particularly useful for light-emitting devices in the
short wavelength region [1 –3]. Indeed, III – V nitridebased blue and green high-brightness light-emitting
diodes (LEDs) made from InGaN/GaN quantum wells
(QWs) structures are now commercially available
with output power larger than 2.5 mW at 20 mA,
* Corresponding author.
i.e., external quantum efficiency >10%. Such a performance is quite amazing considering the high 108 –
1010 cm 2 dislocation density in these LEDs [4].
There is strong evidence to support that InGaN alloy
inhomogeneities play a key role in the high efficiency
of nitride based LEDs grown on sapphire [5 –7]. It has
been proposed that nanoscale indium composition
fluctuation due to InGaN phase separation results in
the formation of indium-rich clusters, which act as
quantum dots (QDs) [5 –7]. Hence, the InGaN system
acts as an extremely sophisticated quantum capture
system, in In(Ga)N QDs, and the charge carriers are
deeply localized so as to hinder their migration toward
non-radiative defects (dislocations) [7,8]. Therefore,
0167-577X/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0167-577X(03)00293-3
L.-W. Ji et al. / Materials Letters 57 (2003) 4218–4221
high luminescence efficiency could be expected if the
density of QDs is much higher than that of dislocations. QD structures are being intensively investigated
not only for device applications but also for studying
fundamental physics. It has been shown that nitride
QDs can be self-organized using the strain-induced
Stranki– Krastanov (S –K) growth mode [8– 10]. Another way to form nitride QDs is to take advantage of
surfactants or anti-surfactants, which are often used to
change the surface free energy of epilayers [11 – 14]. It
had been reported that nitride-based QDs could be
formed by using Si as the anti-surfactant [11– 13].
Self-assembled nitride QDs had already been fabricated by molecular-beam epitaxy (MBE) in InGaN/
GaN system [8,10] using S –K growth mode. Recently, Tachibana et al. [9] reported the growth of InGaN/
GaN QDs by metalorganic chemical vapor deposition
(MOCVD) without using any surfactant. However, it
remains uncertain if the formation of QDs was due to
strain-induced mode or phase separation since small
three-dimensional (3D) InGaN islands were found
after the deposition of up to 20 monolayers (MLs)
of InGaN [10]. In this letter, we report an approach to
grow InGaN QDs using a novel method in a commercial MOCVD reactor (EMCORE D180). An interrupted growth method was proposed to grow selfassembled InGaN QDs without the use of surfactants
or anti-surfactants. Atomic force microscopic (AFM,
Shimmadzu SPM-9500JZ) images reveal that InGaN
self-assembled QDs were successfully obtained by
using interrupted growth method, and their dimensions were small enough to expect zero-dimensional,
quantum effects. The optical properties of such InGaN
QDs are studied with photoluminescence (PL) results
measured at room temperature.
All InGaN QD samples used in this study were
grown on (0001)-oriented sapphire (Al2O3) substrates
in a vertical low-pressure (MOCVD) reactor with a
high-speed rotation disk [15 –26]. The gallium, indium and nitrogen sources were trimethylgallium
(TMGa), trimethylindium (TMIn) and ammonia
(NH3), respectively. After a 30-nm-thick low-temperature GaN nucleation layer was deposited onto the
sapphire substrate at 500 jC, the temperature was
raised to 1000 jC to grow a 2-Am-thick undoped GaN
buffer layer with a growth rate of 2 Am/h. During the
growth of GaN buffer layer, the flow rate of TMGa
was kept at 88 Amol/min while the flow rate of H2
4219
carrier gas was kept at 10 l/min. The growth temperature was then reduced to 730 jC to grow InGaN
QDs. During the growth of InGaN QDs, the flow rates
of TMGa and TMIn were kept at 10 and 35 Amol/min,
respectively. The carrier gas was switched to N2 and
we fixed the N2 carrier gas flow rate at 20 l/min.
According to our calibration, the average indium
composition in the InxGa1 xN layer should be around
x = 0.3. Knowing the lattice mismatch between GaN
and InN equals 11%, the mismatch between In0.3
Ga0.7N and the underneath GaN buffer should be
around 3.3% (Da/a = 11% 0.3 = 3.3%). Previously,
Grandjean and Massies [27] reported that only 3 MLs
could be grown layer by layer before islanding when
InxGa1 xN was deposited on top of the GaN buffer.
In our experiment, the growth rate of the InGaN layer
is estimated to be 0.04 nm/s. Here, an interrupted
growth method was employed during the deposition
of InGaN. We first deposited a thin 4.5 MLs of InGaN
on top of the GaN buffer. We then stopped the growth
from 12 s. After growth stop, another 4.5 MLs of
InGaN was again deposited so as to achieve an InGaN
layer with a total thickness of about 9.0 MLs. For
comparison, samples without the InGaN layer were
also prepared. The surface morphologies of these
samples were then characterized as ex situ by an
AFM system with a sharpened Si3N4 tip at room
temperature. Photoluminescence (PL) was also used
to study the optical properties of these samples.
During PL measurement, a He –Cd laser was used
as the excitation source. The collected luminescence
signal was dispersed by a monochromator, and
detected by a cooled charge-coupled device (CCD)
camera also at room temperature.
Fig. 1a and b shows AFM images of the undoped
GaN buffer layer and the InGaN QDs, respectively.
As shown in Fig. 1a, it can be seen that the surface
morphology of the GaN buffer layer (i.e., without
InGaN layer) is very smooth. In contrast, a rough
surface with the formation of QDs was clearly observed when InGaN was deposited on top of the GaN
with growth stop, as shown in Fig. 1b. From the AFM
picture, the diameter of QDs is estimated to be in the
range of 20– 38 nm, with an average height of 4.1 nm.
In addition, the density of QDs is estimated to be
2 1010 cm 2. It is known that clusters grow primarily in two ways: ripening and coalescence. An
ensemble of clusters on a surface may undergo a
4220
L.-W. Ji et al. / Materials Letters 57 (2003) 4218–4221
Fig. 1. AFM images of (a) undoped GaN buffer layer, and selfassembled QDs formed by the deposition (b) at 9 MLs.
ripening process with mass transport from smaller
cluster to larger cluster by atomic surface diffusion.
In fact, ripening is a process where larger clusters
grow gradually at the expense of smaller clusters. The
total surface energy of a system is decreased when
small particles combine to form larger clusters with no
loss of mass [28]. In our experiment, we first deposited 4.5 MLs of InGaN on top of GaN buffer and then
stopped the growth of InGaN for 12 s at 730 jC. Since
the critical thickness is only 3 MLs for In0.3Ga0.7N,
the initial growth of 4.5 MLs of InGaN should already
relax to form small clusters according to strain-induced S– K growth mode. Generally speaking, the
strained energy of 3D islands is lower than that of the
wetting layer. Thus, if we employ interrupted growth
method at a reasonable temperature, small clusters
should have sufficient time to combine with each
other so as to from large cluster (i.e. 3D islands) by
atomic surface diffusion. After growth interrupt, the
total surface energy will decrease. We can thus form
InGaN QDs by depositing another 4.5 MLs of InGaN.
In other words, such a growth interrupt method could
be used to prepare InGaN QDs without the use of
surfactants or anti-surfactants, and the size of QDs
formed by such a method should be small enough to
show zero-dimensional quantum effects. One other
advantage of such method is that we should be able to
control the physical properties of InGaN QDs by
adjusting interruption time and/or interruption temperatures easily. Note that this result has been reproduced
in our MOCVD system.
Fig. 2 shows the room temperature PL spectrum
observed from InGaN QDs formed by growth interrupt method. It can be seen that PL peak position of
the growth interrupt sample is located at about 2.68
eV (462 nm) with a narrow full width at half maximum (FWHM) of about 92 meV. According to our
previous experience, typical FWHM of room temperature PL spectrum is around 100 meV for InGaN QW
samples grown in our MOCVD system. Although we
expect that fluctuations in QD size and indium composition could result in a broad PL spectrum, the 92
meV PK FWHM observed from the QF sample is
even smaller than those observed from typical quantum well samples. Such a result could probably be
attributed to the large effective mass in this material
system [9]. The reasonably sharp 92 meV PL FWHM
also suggests that InGaN QDs formed by MOCVD
Fig. 2. Photoluminescence from InGaN QDs at room temperature.
L.-W. Ji et al. / Materials Letters 57 (2003) 4218–4221
growth interrupt are potentially useful in nitride-based
optoelectronic devices. With the optimization of QD
formation, we should be able to further improve the
optical properties of these nitride-based QDs.
In conclusion, it has been demonstrated that we can
use interrupted growth method in MOCVD to control
the growth of InGaN layers and to grow nanoscale
InGaN self-assembled QDs. With a 12-s growth interrupt, we successfully formed InGaN QDs with a typical
lateral size of 25 nm and an average height of 4.1 nm.
The QD density is about 2 1010 cm 2. Strong PL
emission of InGaN nanostructure was observed at
room temperature with a FWHM of about 92 meV.
These results suggest that such QDs are potentially
useful in nitride-based optoelectronic devices.
Acknowledgements
This work was financially supported by the
National Science Council of Taiwan (Project No.
NSC 90-2215-E-006-024).
References
[1] S. Nakamura, M. Senoh, N. Iwasa, S. Nagahama, Jpn. J. Appl.
Phys. 34 (1995) L797.
[2] I. Akasaki, H. Amano, Jpn. J. Appl. Phys. 36 (1997) 5393.
[3] W.C. Lai, S.J. Chang, M. Yokoyama, J.K. Sheu, J.F. Chen,
IEEE Photonics Technol. Lett. 13 (2001) 559.
[4] S. Nakamura, S. Pearton, G. Fasol, The Blue Laser Diode,
Springer, Berlin, 2000.
[5] Y. Narukawa, Y. Kawakami, M. Funato, S. Fujita, S. Fujita,
S. Nakamura, Appl. Phys. Lett. 70 (1997) 981.
[6] S.F. Chichibu, A.C. Abare, M.S. Minsky, S. Keller, S.B.
Fleischer, J.E. Bowers, E. Hu, U.K. Mishra, L.A. Coldren,
S.P. DenBaars, S. Sota, Appl. Phys. Lett. 73 (1998) 2006.
[7] K.P. O’Donnell, R.W. Martin, P.G. Middleton, Phys. Rev.
Lett. 82 (1999) 237.
[8] B. Damilano, N. Grandjean, S. Dalmasso, J. Massies, Appl.
Phys. Lett. 75 (1999) 3751.
4221
[9] K. Tachibana, T. Someya, Y. Arakawa, Appl. Phys. Lett. 74
(1999) 383.
[10] C. Adelmann, J. Simon, G. Feuillet, N.T. Pelekanos, B. Daudin, Appl. Phys. Lett. 76 (2000) 1570.
[11] S. Tanaka, S. Iwai, Y. Aoyagi, Appl. Phys. Lett. 69 (1996)
4096.
[12] X.Q. Shen, S. Tanaka, S. Iwai, Y. Aoyagi, Appl. Phys. Lett. 72
(1998) 344.
[13] H. Hirayama, S. Tanaka, P. Ramvall, S. Keller, Y. Aoyagi,
Appl. Phys. Lett. 72 (1998) 1736.
[14] J. Zhang, M. Hao, P. Li, J. Chua, Appl. Phys. Lett. 80 (2002)
485.
[15] J.K. Sheu, S.J. Chang, C.H. Kuo, Y.K. Su, L.W. Wu, Y.C. Lin,
J.M. Tsai, R.K. Wu, G.C. Chi, IEEE Photonics Technol. Lett.
15 (2003) 18.
[16] Y.C. Lin, S.J. Chang, Y.K. Su, T.Y. Tsai, C.S. Chang, S.C.
Shei, S.J. Hsu, C.H. Liu, U.H. Liaw, S.C. Chen, B.R. Huang,
IEEE Photonics Technol. Lett. 14 (2002) 1668.
[17] Y.K. Su, S.J. Chang, C.H. Ko, J.F. Chen, W.H. Lan, W.J. Lin,
Y.T. Cherng, J. Webb, IEEE Trans. Electron Devices 49
(2002) 1361.
[18] Y.K. Su, S.J. Chang, C.H. Chen, J.F. Chen, G.C. Chi, J.K.
Sheu, W.C. Lai, J.M. Tsai, IEEE Sens. J. 2 (2002) 366.
[19] S.J. Chang, C.H. Kuo, Y.K. Su, L.W. Wu, J.K. Sheu, T.C.
Wen, W.C. Lai, J.F. Chen, J.M. Tsai, IEEE J. Sel. Top. Quantum Electron. 8 (2002) 744.
[20] C.H. Chen, S.J. Chang, Y.K. Su, J.K. Sheu, J.F. Chen, C.H.
Kuo, Y.C. Lin, IEEE Photonics Technol. Lett. 14 (2002) 908.
[21] T.C. Wen, S.J. Chang, L.W. Wu, Y.K. Su, W.C. Lai, C.H. Kuo,
C.H. Chen, J.K. Sheu, J.F. Chen, IEEE Trans. Electron Devices 49 (2002) 1093.
[22] L.W. Wu, S.J. Chang, T.C. Wen, Y.K. Su, W.C. Lai, C.H. Kuo,
C.H. Chen, J.K. Sheu, IEEE J. Quantum Electron. 38 (2002)
446.
[23] C.H. Kuo, S.J. Chang, Y.K. Su, J.F. Chen, L.W. Wu, J.K.
Sheu, C.H. Chen, G.C. Chi, IEEE Electron Device Lett. 23
(2002) 240.
[24] S.J. Chang, W.C. Lai, Y.K. Su, J.F. Chen, C.H. Liu, U.H.
Liaw, IEEE J. Sel. Top. Quantum Electron. 8 (2002) 278.
[25] C.H. Chen, S.J. Chang, Y.K. Su, G.C. Chi, J.K. Sheu, J.F.
Chen, IEEE J. Sel. Top. Quantum Electron. 8 (2002) 284.
[26] C.H. Chen, Y.K. Su, S.J. Chang, G.C. Chi, J.K. Sheu, J.F.
Chen, C.H. Liu, U.H. Liaw, IEEE Electron Device Lett. 23
(2002) 130.
[27] N. Grandjean, J. Massies, Appl. Phys. Lett. 72 (1998) 1078.
[28] K.N. Tu, J.W. Mayer, L.C. Feldman, Electronic Thin Film
Science, Macmillan, New York, 1992.