Optical properties of InGaN quantum dots grown by Si N x nanomasks
L. L. Huang, H. J. Chang, Y. Y. Chou, C. H. Wang, T. T. Chen, Y. F. Chen, J. Y. Tsai, S. C. Wang, and H. C. Kuo
Citation: Journal of Applied Physics 101, 083501 (2007); doi: 10.1063/1.2717258
View online: http://dx.doi.org/10.1063/1.2717258
View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/101/8?ver=pdfcov Published by the AIP Publishing
Articles you may be interested in
Ge 1 y Sn y Si ( 100 ) composite substrates for growth of In x Ga 1 x As and Ga As 1 x Sb x alloys J. Appl. Phys. 101, 013518 (2007); 10.1063/1.2407274
Self-assembled In 0.22 Ga 0.78 As quantum dots grown on metamorphic Ga As Ge Si x Ge 1 x Si substrate J. Appl. Phys. 100, 064502 (2006); 10.1063/1.2337770
Prospective emission efficiency and in-plane light polarization of nonpolar m -plane In x Ga 1 x N Ga N blue light emitting diodes fabricated on freestanding GaN substrates
Appl. Phys. Lett. 89, 091906 (2006); 10.1063/1.2337085
Characteristics of InGaAs quantum dots grown on tensile-strained GaAs 1 x P x J. Appl. Phys. 97, 093518 (2005); 10.1063/1.1884249
Study of the dominant luminescence mechanism in In Ga N Ga N multiple quantum wells comprised of ultrasmall InGaN quasiquantum dots
Appl. Phys. Lett. 86, 121915 (2005); 10.1063/1.1891291
Optical properties of InGaN quantum dots grown by SiN
xnanomasks
L. L. Huang, H. J. Chang, Y. Y. Chou, C. H. Wang, T. T. Chen, and Y. F. Chena兲
Department of Physics, National Taiwan University, Taipei 106, Taiwan
J. Y. Tsai, S. C. Wang, and H. C. Kuo
Institute of Electro-optical Engineering, National Chiao-Tung University, Hsin-Tsu 300, Taiwan
共Received 15 November 2006; accepted 7 February 2007; published online 16 April 2007兲 InGaN quantum dots共QDs兲 deposited on SiNxnanomasks have been investigated by atomic force
microscopy, photoluminescence 共PL兲, and photoluminescence excitation 共PLE兲 measurements. It was found that the size of QDs can be well controlled by SiNx nanomasks, enabling the
manipulation of quantum confinement effect. The PL spectra of InGaN QDs contain several fine structures, and the main peaks can be attributed to families of QDs with different sizes. The emission arising from InGaN QDs and GaN buffer layer can be clearly distinguished based on PLE measurement, which can be used to improve the interpretation in the previous reports. Our study indicates that the quantum confined Stark effect due to piezoelectric field plays a very important role in the optical properties of InGaN QDs, which is very useful for the application of optoelectronic devices. © 2007 American Institute of Physics.关DOI:10.1063/1.2717258兴
I. INTRODUCTION
The group III-nitride wide-band-gap semiconductors have been recognized as very important materials for many optoelectronic devices, such as blue ultraviolet 共UV兲 light emitting diodes 共LEDs兲, laser diodes 共LDs兲, and high-temperature/high-power electronic devices.1–3 For example, InGaN / GaN quantum well 共QW兲 structures have been suc-cessfully used as the active layers in LEDs and LDs.1–3 How-ever, the threshold current density is still high for LDs with InGaN QW structures. It is expected that using quantum dots 共QDs兲 instead of QWs as the active layer can improve the performance of LDs.4 Consequently, growth and character-ization of InGaN QDs become very important for highly efficient blue and violet semiconductor lasers.4 To ensure suitability for the optoelectronic applications, high density and uniform size of QDs must be achieved. Recently, several growth modes have been employed to grow InGaN QDs, including the Stranski-Krastanow mode and the deposition of antisurfactant such as Si or a SiNxnanomasks facilitating the
formation of auto-organized QDs.4–6 However, a clear pic-ture to describe luminescence from InGaN QDs is still not yet achieved. The detailed luminescence mechanism is quite complicated, which strongly depends on several different factors, including material composition, internal strain, QD size, excitation condition, as well as the method of QD for-mation. For example, the true indium content of small clus-ters is difficult to determine due to the shape and the small size of clusters. The large mismatch between InN and GaN lattices causes difficulties in their growth while film instabil-ity against spinodal decomposition often causes nonuniform distribution of indium aggregation.7 Besides, due to large piezoelectric effect in this material, the quantum-confined Stark effect could be very important.8It has been controver-sial whether the dominant luminescence mechanism in
InGaN / GaN originates from the presence of the In-rich clus-ter or from the inclus-ternal electric field. In this paper, we inves-tigate optical properties of InGaN QDs grown by selective growth on SiNx-deposited GaN epilayer. By changing the
duration of the SiNxtreatment, the wavelength covering
vis-ible region can be easily controlled. We demonstrate the fine structures in the photoluminescence 共PL兲 spectra arising from families of QDs with different sizes. The emissions from InGaN QDs and from GaN buffer layer can be unam-biguously distinguished based on the photoluminescence ex-citation 共PLE兲 measurement. In addition, our measurements clearly indicate that the properties of InGaN QDs can be significantly influenced by the piezoelectric field. In view of the strong piezoelectric共PZ兲 field in InGaN QDs, the results shown here should be very important for their practical ap-plication.
II. EXPERIMENT
The studied ultra-high-density InGaN QDs were grown by metal-organic chemical vapor deposition 共MOCVD兲. A 30- nm-thick undoped GaN layer was first deposited on
c-face sapphire substrates at 550 ° C, followed by a
2-m-thick Si-doped GaN underlying layer at 1000 ° C. Then, NH3 and the diluted Si2H6 were flowed through the sample to grow a rough SiNx layer on the n-type GaN
un-derlying layer with increasing the SiNxtreatment time from
120 to 195 s. The temperature was then reduced to 800 ° C to grow the InGaN layers, and a 10 nm undoped GaN cap layer was grown on the InGaN layer to complete the sample structure. Atomic force microscopy 共AFM兲 images of the uncapped samples are shown in Fig. 1, which indicate that the morphology of the surface changes from nanoislands to sharp QDs as the duration of the SiNxtreatment increases.
The average widths, heights, and densities of the QDs in the samples are estimated to change from 33 to 35 nm, from 0.64 to 2.46 nm, and from 2.1⫻1011 to 2.9⫻1011cm−2with
increasing thickness of the SiNxlayer. Therefore, the
forma-a兲Author to whom correspondence should be addressed; electronic mail:
JOURNAL OF APPLIED PHYSICS 101, 083501共2007兲
0021-8979/2007/101共8兲/083501/5/$23.00 101, 083501-1 © 2007 American Institute of Physics
tion of InGaN QDs could be controlled by just changing the duration of the SiNxtreatment. A detailed description of the
growth process has been published elsewhere.6 For the PL measurement, the sample was excited by a He–Cd laser working at 325 nm, dispersed by a Spex 0.85 m double-grating spectrometer, and the spectra were detected by a pho-tomultiplier tube. Similar arrangements were used for the PLE measurement except the dispersed light from a Xe lamp was used as the continuous radiation.
III. RESULTS AND DISCUSSION A. Dependence on SiNx treatment time
Figure 2 shows the normalized PL spectra of InGaN QDs with the SiNx treatment time of 120, 160, 178, and
195 s. The InGaN QDs exhibit a very strong visible PL sig-nal at room temperature, which even exceeds that of device quality of InGaN quantum wells. The strong PL signal re-sults from the fact that the confinement of electron-hole pairs is very beneficial for enhancing radiative recombination, and the confined carriers have less probability to diffuse to non-radiative recombination centers. The PL spectra also clearly show that the signal arising from InGaN QDs is much stron-ger than that from GaN band gap emission. As shown in the inset of Fig. 2, the main peak has a redshift of about 232 meV when SiNx treatment time increases from
120 to 195 s, which covers the entire visible region. Accord-ing to the above AFM images, the redshift of the spectra could be simply attributed to the small quantum confinement effect with increasing the average height of InGaN QDs. It is worth noting that the PL spectrum of each InGaN QD sample contains two main peaks, which may arise from the transi-tions involving ground state and excited states of QDs, or families of QDs with different sizes.9,10As shown in Fig.2, these two peaks have a comparable intensity and their energy separation is much larger than the thermal energy at room temperature; the possibility of involving ground and excited transitions can be ruled out because if the Boltzman distri-bution is taken into account, the carrier population in the excited state and the corresponding PL intensity should be very small. Apparently, this result clearly indicates that the double main peaks are due to different dot families.
B. Dependence on excitation intensity
To further investigate the optical properties of InGaN QDs, we have performed the PL measurements under differ-ent excitation densities, as shown in Fig. 3. It is found that the PL intensity is linearly proportional to excitation density, as shown in the inset of Fig.3, indicating that the transition involves an excitonic recombination. It is also found that the main peak position of the PL spectra shows a blueshift as the FIG. 1. 共Color online兲 Atomic force microscopy images of InGaN layers
with共a兲 120 s, 共b兲 160 s, 共c兲 178 s, and 共d兲 195 s of SiNxtreatment on the
underlying GaN layers.
FIG. 2. Normalized photoluminescence spectra at room temperature of InGaN QDs with the SiNxtreatment time at 120 s, 160 s, 178 s, and
195 s. The inset shows the main peak energy of the PL spectra as a function of SiNxtreatment duration.
083501-2 Huang et al. J. Appl. Phys. 101, 083501共2007兲
optical excitation density increases. In the typical case of the quantum confined Stark effect, such as in InGaN / GaN mul-tiple quantum wells,11,12 the blueshift was frequently attrib-uted to the screening of built-in PZ field. The underlying physics is based on the spatial separation of photoexcited electrons and holes due to the built-in PZ field. Because the induced field is opposite to the internal PZ field, the resultant field is therefore reduced. This result demonstrates that the PZ field is important in determining the optical property of InGaN QDs.
C. Temperature dependence
Figure4共a兲shows temperature dependent PL spectra of InGaN QDs with the SiNx treatment time of 195 s under
high excitation density. When the temperature increases from 20 to 30 K, the peak shift of InGaN QDs is about 30 meV, while that of the InGaN bulk is about 90 meV. It is well known that the redshift of the PL peak position of a semi-conductor with increasing temperature originates from electron-phonon interaction. Thus, the smaller redshift ob-served in InGaN QDs than that in bulk material can be ex-plained by partial decoupling of the phonon-exciton interac-tion in QDs.13 The structure covering from 3.0 to 3.3 eV at low temperature can be attributed to donor-acceptor pair 共DAP兲 bands of GaN buffer layer. A clear evidence to sup-port our assignment is that the phonon replica is in exact agreement with the LO phonon energy of GaN. Another fea-ture is the yellow band at about 2.2 eV. The band is associ-ated with deep levels due to point defects such as Ga or N vacancies.14It is worth noting that the integrated PL intensity from InGaN QDs is about 19 times higher than that from GaN at low temperature, and the difference increases with temperature. The improvement of optical recombination in QDs at high temperature can be related to the quantum con-finement effect, in which the strong wave function overlap between electron and hole leads to an enhanced exciton bind-ing energy and transition probability.
Figure4共b兲 shows the PL peak position of InGaN QDs with the SiNxtreatment time of 195 s as a function of
tem-perature under high and low excitation intensities. Under high excitation intensity, the piezoelectric field is severely screened and the PL temperature dependence is similar to that of bulk material, i.e., the band gap decreases with in-creasing temperature. On the other hand, under low excita-tion intensity, a blueshift in the PL peak is observed when temperature increases from 20 to 90 K. We can also under-stand this abnormal PL blueshift based on the screening of the piezoelectric field due to thermal excitation. At low tem-perature, carriers are frozen in nonradiative centers, and they will be thermally activated as temperature increases. This process will induce partial screening of the piezoelectric field and lead to the observed blueshift in the PL spectra. The temperature dependent PL spectra thus reflect a competition between the blueshift induced by the thermal screening of the piezoelectric field and the redshift induced by electron-phonon interaction.
In order to determine the activation energy of the elec-trons confined in InGaN QDs, we have performed integrated PL intensity as a function of the reciprocal temperature in InGaN QDs under different SiNxduration times, as shown in
Fig. 5. As the temperature increases, the variation of inte-grated PL intensity can be expressed as15
I共T兲 = I共0兲/1 + C exp共− EA/kBT兲, 共1兲
where I共0兲 is the integrated PL intensity at 0 K, EA is the
activation energy, C is a fitting parameter, and KB is the
FIG. 3. Photoluminescence spectra of InGaN QDs with different excitation densities at room temperature. The inset shows the PL intensity of main peak as a function of excitation power density.
FIG. 4. 共a兲 Temperature dependent photoluminescence spectra of InGaN QDs with the SiNxtreatment time at 195 s.共b兲 Temperature dependence of
the photoluminescence peak position under high and low excitation intensities.
083501-3 Huang et al. J. Appl. Phys. 101, 083501共2007兲
Boltzmann constant. Using Eq.共1兲, the activation energy EA
is 48, 35, 28, and 17 meV in InGaN QDs with SiNxduration
time of 120, 160, 178, and 195 s, respectively. Higher acti-vation energy in InGaN QDs originates from a reduction in the dimensions of QDs, which is a consequence of the in-creased Columb interaction due to carrier confinement.
D. Photoluminescence excitation measurement
Figure6shows the PL and PLE measurements in InGaN QDs with the SiNxtreatment time of 195 s at room
tempera-ture. The PL and PLE spectra were measured by using the dispersed light from a xenon lamp as the excitation source. The PL spectrum is excited by a 3.35 eV radiation and the PLE spectra is monitored at different PL peaks, as shown in Fig. 6. Quite interestingly, the optical properties of PLE spectra with different monitoring wavelengths show very in-teresting behaviors. Especially, the emissions arising from InGaN QDs and GaN defects are quite different. The peak
energies at 2.19 and 2.32 eV can be attributed to the transi-tions in the GaN layer, because the PLE intensity can be greatly enhanced when the excitation energy is in resonance with the band gap transition of the GaN layer. However, PLE spectra monitored at 2.62 and 2.73 eV are quite different from those monitored at 2.19 and 2.32 eV. The PLE intensi-ties monitored at 2.62 and 2.73 eV become larger even when the excitation energy does not belong to the transitions in the GaN layer. Therefore, they can be attributed to the excitonic transition in InGaN QDs. The spectral modulation below 3.36 eV in the PLE spectra is very interesting. Previously, its absorption has been identified as excitonic resonance in InGaN QDs, and therefore a giant Stokes shift was inferred.16 As shown in Fig. 7, we summarize the PL and PLE spectra of InGaN QDs with the SiNx treatment time
from 120 to195 s at room temperature. All the PL spectra were excited by 3.36 eV using a xenon lamp as the excita-tion source, and the PLE spectra were monitored at the main peak positions, as shown in Fig.7. Here, we only show the PLE spectra taken at the main PL emission peak in InGaN QDs with the SiNxtreatment time from 120 to 195 s. Similar
behavior in PLE spectra was observed when the detection energies were varied in the region of QD radiation. All the PLE spectra in the region below 3.36 eV clearly show an oscillatory behavior, and the oscillatory structure does not shift with the change of QD size. Therefore, our result clearly indicates that the oscillatory behavior does not arise from the excitonic resonance in QDs, as reported previously.16Instead, it may originate from the absorption of the SiNxor the wetting layer. We thus conclude that the PLE
spectra monitored at different wavelengths are very useful to identify the origin of optical properties in multilayer struc-tures.
IV. CONCLUSION
In conclusion, we have reported a highly efficient emis-sion of ultra-high-density InGaN QDs with controllable op-tical properties by SiNxnanomasks. Based on the
tempera-ture dependent PL spectra under high and low optical FIG. 5. Integrated photoluminescence intensity as a function of the
recipro-cal temperature of InGaN QDs with the SiNxtreatment time at共a兲 120 s, 共b兲
160 s,共c兲 178 s, and 共d兲 195 s. The continuous lines are the best fit for the samples according to the Arrhenius plot.
FIG. 6. Photoluminescence共PL兲 and photoluminescence excitation 共PLE兲 spectra of InGaN QDs with the SiNx treatment time at 195 s at room
temperature.
FIG. 7. Photoluminescence共PL兲 and photoluminescence excitation 共PLE兲 spectra of InGaN QDs with the SiNxtreatment time from 120 to 195 s at
room temperature.
083501-4 Huang et al. J. Appl. Phys. 101, 083501共2007兲
excitations, it is shown that the piezoelectric field plays a very important role in determining the emission from InGaN QDs. In addition, with the measurement of PLE spectra monitored at different photon energies, one can clearly dis-tinguish the emission arising from QDs and buffer layer, which enables the clarification of the misunderstanding in previous reports. Our results also suggest that it is possible to fabricate full-color and high-brightness InGaN QDs LED based on the assistance of SiNxnanomasks.
This work was supported by the National Science Coun-cil and Ministry of Education of the Republic of China.
1S. Nakamura and G. Fasol, The Blue Laser Diode共Springer, Berlin, 1997兲. 2H. Morkoc, S. Strite, G. B. Gao, M. E. Lin, B. Sverdlov, and M. Burns, J.
Appl. Phys. 76, 1363共1994兲.
3S. Nakamura, M. Senoh, S. Nagahama, N. Iwasa, T. Yamada, T.
Matsus-shita, H. Kiyoko, and Y. Sugimoto, Jpn. J. Appl. Phys., Part 2 35, L74 共1996兲.
4J. Wang, M. Nozaki, M. Lachab, Y. Ishikawa, R. S. Qhalid Fareed, T.
Wang, M. Hao, and S. Sakai, Appl. Phys. Lett. 75, 950共1999兲.
5K. Tachibana, T. Someya, and Y. Arakawa, Appl. Phys. Lett. 74, 383
共1999兲.
6R.-C. Tu et al., Jpn. J. Appl. Phys., Part 2 43, L264共2004兲.
7Y. L. Lai, C. P. Liu, and Z. Q. Chen, Appl. Phys. Lett. 86, 121915共2005兲. 8T. Saito and Y. Arakawa, Physica E共Amsterdam兲 15, 169 共2002兲. 9H. M. Ng, T. D. Moustakas, and K. F. Ludwig, Jr., J. Vac. Sci. Technol. B
18, 1457共2000兲.
10L. Brusaferri et al., Appl. Phys. Lett. 69, 3354共1996兲. 11T. Y. Lin, Appl. Phys. Lett. 82, 880共2003兲.
12F. Della Sala, A. Di Carlo, P. Lugli, F. Bernardini, V. Fiorentini, R. Scholz,
and J.-M. Jancu, Appl. Phys. Lett. 74, 2002共1999兲.
13P. Ramvall, S. Tanaka, S. Nomura, P. Riblet, and Y. Aoyagi, Appl. Phys.
Lett. 75, 1935共1999兲.
14B. Gil, Group III Nitride Semiconductor Compounds共Clarendon, Oxford,
1998兲.
15H. P. D. Schenk, M. Leroux, and P. de Mierry, J. Appl. Phys. 88, 1525
共2000兲.
16S. Chichibu, T. Sota, K. Wada, and S. Nakamura, J. Vac. Sci. Technol. B
16, 2204共1998兲.
083501-5 Huang et al. J. Appl. Phys. 101, 083501共2007兲
