Open Access
Add to BibSonomyAbstract
Three green light emitting InGaN/GaN multiple quantum well (MQW) structures with different In composition grown by metal-organic chemical vapor deposition are investigated by the X-ray diffraction and the temperature-dependent photoluminescence (PL) measurements. It is found that when the In composition increases in the InGaN/GaN MQWs, the PL spectral bandwidth may anomalously decrease with increasing temperature. The reduction of PL spectral bandwidth may be ascribed to the enhanced non-radiative recombination process which may lower the light emission efficiency of the localized luminescent centers with shallow localization energy in the high-In-content InGaN quantum wells and also cause a reduction of integrated PL intensity.
© 2015 Optical Society of America
1. Introduction
InGaN/GaN multiple quantum well (MQW) heterostructures are widely used as active layers of solid-state light emitting devices such as light emitting diode (LEDs) [1] and laser diodes (LDs) [2], where the color and intensity of emitted light can be tuned by changing the InGaN alloy composition and well layer width [3]. It is believed that the optical emission of InGaN quantum well (QW) is originated from the radiative recombination of carriers in the quantum-dot-like localized luminescent centers (LLCs) (known as localization effect) [4], induced by the compositional fluctuations of indium within the InGaN alloy or the thickness fluctuations of InGaN well layers [5]. The lateral size of LLC regions may be larger than conventional quantum dots in a wide range from several tens to hundreds of nanometers [6]. From the view of the energy landscape, such LLCs can be considered as localization states or band-tail states with different depth of localization energy statistically distributed below the nominal band edge of InGaN [7]. The carriers confined in the LLCs can better avoid to be captured by the non-radiative recombination (NRR) centers, and the matrix element for optical transition is enlarged by attracting carriers within the same localized site. Thereby the internal quantum efficiency of InGaN/GaN MQWs is relatively high despite the large density of dislocations in InGaN QWs [8].
However, as the emission wavelength of InGaN/GaN MQWs extends to green range, the indium content in InGaN well layer increases. More indium incorporated in InGaN QWs will give rise to a larger lattice mismatch between InGaN wells and GaN barriers, resulting in the stronger quantum confined Stark effect (QCSE) and the enhanced NRR process induced by more dislocations or other defects, leading to a serious reduction of luminescence efficiency in high-In-content InGaN/GaN MQWs [9, 10 ].
Many experimental methods have been used to study the impact of NRR process caused by defect centers or dislocations on the luminescence efficiency of InGaN QWs, such as X-ray diffraction (XRD) [11] and time-resolved photoluminescence (PL) spectroscopy [12]. Furthermore, the direct observations of threading dislocations and V-pits in the regions of LLCs are realized by the means of transmission electron microscope (TEM) and high-resolution cathodoluminescence (CL) hyperspectral images [6,13 ]. Therefore, the NRR process induced by defect centers or dislocations may compete against the radiative recombination (RR) process in LLCs, and result in a contribution to the reduction of luminescence efficiency in green light emitting InGaN/GaN MQWs. In other words, in the high-In-content InGaN/GaN MQWs structures the NRR process may exist in the LLCs and weaken the light emission form LLCs.
To elucidate the impact of NNR process on the LLCs in InGaN QWs, three green light InGaN/GaN MQWs samples grown with different flux of trimethylindium (TMIn) during epitaxial growth process are studied. The temperature-dependent PL measurements are used to study the emission from LLCs in InGaN QWs. It is found that the variation of PL spectral bandwidth with increasing temperature can be analyzed to investigate the detailed influence of NRR process on the LLCs. Combined with XRD and PL results, it is found that, with increasing In composition in the green light InGaN/GaN MQWs, the increased NRR centers may exist in the regions of those LLCs with shallow localization energy, causing the reduction of integrated PL intensity as well as a reduction of PL spectral bandwidth with increasing temperature.
2. Experiments
Three green light emitting InGaN/GaN MQWs samples were grown on c-plane sapphire substrate by AIXTRON close-coupled showerhead 3 × 2 in. vertical reactor metal-organic chemical vapor deposition (MOCVD) system. The epitaxial growth was conducted by using trimethylgallium (TMGa), trimethylindium (TMIn), and ammonia (NH3) as the precursors for Ga, In, and N, respectively, with H2 and N2 as the carrier gas for GaN and InGaN growth, respectively. A low temperature GaN nucleation layer was deposited on a sapphire substrate at first, and then a Si-doped GaN with a thickness of 2 μm was grown on the top of the nucleation layer at 1080 °C as template. Then the InGaN/GaN MQW layers were grown. All samples consist of five periods of MQWs with the same well width (3 nm) and barrier thickness (8 nm). For GaN quantum barrier layers, the growth temperature was increased by 100 K with respect to well layer growth temperature in order to improve the material quality of barrier layers. Finally, a 150 nm-thick p-type GaN contact layer doped with Mg was deposited at 1040 °C. Three samples with different nominal In content in InGaN alloy layers were deposited and the TMIn flux was employed as the modulating parameter for changing the In concentration. All other growth parameters except the flux of TMIn were identical during the epitaxial growth process. These structures were denoted as Samples W24, W33, and W39 for the flux of TMIn were 240, 330, and 390 sccm, respectively.
The X-ray diffraction (XRD) measurement with a Rigaku X-ray diffractometer was used to determine the In content and material quality of InGaN/GaN MQW samples. For temperature-dependent PL measurements, the excitation power was relatively low. A λ = 325 nm continuous wave He–Cd laser was used as excitation source, with an incident optical power of 4 mW and a spot size of 0.5 mm2, i.e. an excitation density of 0.8 W/cm2. A closed-cycle refrigerator of CTI Cryogenics was used for the temperature-dependent measurements.
3. Results and discussion
The (0002) ω–2θ scan XRD curves of three Samples W24, W33, and W39, denoted by black, red, and blue lines, respectively, are depicted in Fig. 1 . Two superlattice (SL) satellite peaks, i.e., 1st and 2nd order SL peak, can be observed for all samples in Fig. 1. Since the 0th order SL satellite peak cannot be well distinguished from the GaN main peak, the angle position of 0th order SL peak is calculated from the angle distance between 1st and 2nd order satellite peaks. According to the angle position of 0th order peak, the nominal In composition in InGaN well layers can be determined, which are 15.4%, 19.5%, and 21.4%, for Samples W24, W33, and W39, respectively.
Fig. 1 (0002) ω–2θ scan curves of Samples W24, W33, and W39, denoted by black, red, and blue lines, respectively. The inset shows the relationship between the satellite peaks’ FWHM and the peaks’ order for all samples.
It is known that for the MQW structures the full width at half maximum (FWHM) of satellite peaks will be broadened with increase of the order by the rough interfaces between wells and barriers [14]. Therefore, the interface roughness can be obtained according to the change of satellite peak’s FWHM with the satellite peak’s order. If the interface roughness is described by a Gaussian distribution function with standard deviation, the FWHM of the n-th satellite peak will be expressed as [15]:
where n is the order of satellite peak, Λ is the period length, ΔθM is the angle distance between adjacent satellite peaks, and σ/Λ is the interface roughness. W0 and Wn are the FWHM values of 0th and n-th order satellite peaks, respectively. The inset shows the relationship between the satellite peaks’ FWHM and the peaks’ order for the Samples W24, W33, and W39, denoted by black squares, red circles, and blue triangles, respectively. According to Eq. (1), the interface roughness can be determined from the slope of linear fitting of the data. It is found that the slope values for Samples W33 and W39 are larger than that for Sample W24. Therefore we can judge that Sample W24 has the flattest interfaces, while the interfaces of MQWs for Samples W33 and W39 are rougher. It is deduced that the rougher interfaces of Samples W33 and W39 due to more In atoms incorporated in InGaN well layers may cause the deterioration of the material quality of InGaN/GaN MQWs structures [11].In Fig. 2 the room temperature (RT) PL spectra of 3 InGaN/GaN MQW samples with different In composition are plotted by black, red, and blue lines for Samples W24, W33, and W39, respectively.
Fig. 2 RT PL spectra of Samples W24 (black), W33 (red), and W39 (blue) measured under the same conditions. The peak energy (spectral bandwidth) of Samples W24, W33 and W39 is 2.44 eV (200 meV), 2.36 eV (256 meV) and 2.32 eV (308 meV), respectively, which are obtained by Gaussian fitting.
It can be seen that the spectral profile of each sample is not a regular Gaussian or Lorentz profile, but an asymmetric and irregular one, often with many dips modulated by Fabry-Perot interference. It is known that the detected PL spectral lines from GaN-based MQW film structures grown on sapphire are often strongly disturbed by the Fabry–Perot interference effect. Furthermore, the spectra may be broadened inhomogeneously due to the non-uniform distribution of In composition and the fluctuation of well thickness in the InGaN QWs. Therefore, we should find some way to better describe the spectral physical properties based on the measured PL spectra.
For simplification and a semi-quantitative analysis, the PL spectral lines are fitted by Gaussian functions to obtain the spectral parameters, e.g., the peak energy and spectral bandwidth. Actually, the spectral lines are strongly disturbed by the Fabry-Perot interference and the inhomogeneous broadening effect. Thus the fitted parameters only can be semi-quantitatively used to represent the characteristics of the PL spectra, and in certain extent to reflect the peak position and spectral bandwidth of the obtained PL spectra. Since the same fitting method is employed to the PL analysis of all studied samples, it is reasonable to compare the variation trend of PL spectral parameters among these different samples. Therefore, in this paper, the peak position and FWHM of fitted Gaussian functions are used to describe and study the measured PL spectra.
In Fig. 2 the fitted PL peak energy (or wavelength) of Samples W24, W33, and W39 is 2.44 eV (508 nm), 2.36 eV (525 nm), and 2.32 eV (534 nm), respectively. The smaller peak energy of W39 than that of W24 can be attributed to both smaller effective band gap of InGaN alloy and larger QCSE in QWs due to an increase of In composition in InGaN QW layers [10,14 ]. Compared with Sample W24 in Fig. 2, Sample W39 exhibits an approximate five-fold decrease in PL intensity. Such decrease of PL intensity can be attributed to both the stronger QCSE and the more NRR centers induced by the increased defect centers or dislocations [16].
It is also noticed that the PL spectral bandwidth of Samples W24, W33, and W39, obtained through Gaussian fitting, is 200 meV, 256 meV, and 308 meV, respectively. It is well known that the broadening of PL spectra, when the In content increases in the InGaN alloy layers, can be attributed to the increased inhomogeneity of In composition [4], including random alloy composition fluctuations, and phase separation during the QW growth (due to InGaN alloy instability) [17]. Additionally, it is suggested that higher density of dislocations and increased interface roughness of InGaN/GaN MQWs may also broaden the PL spectra in InGaN/GaN MQW structures [11]. Furthermore, this may also cause a reduction of PL spectral bandwidth with increasing temperature, which will be analyzed later.
The PL peak energy as a function of temperature is plotted in Fig. 3 for all samples. Since a relatively low excitation power of PL measurement is used in our experiments, an anomalous S-shape shift of PL peak energy can be observed, which can be seen as a fingerprint of the existence of localization states or LLCs [7]. Generally, the S-shape temperature-dependent curves can be explained as follows [18]: at very low temperature, e.g. 10 K, the injected carriers are randomly distributed in all LLCs in QWs. Since in this case for most of the photo-generated carriers the mobility of carriers is too small to reach the LLCs with deepest localization energy, i.e. the deepest LLCs, they recombine and emit photons in all LLCs. As temperature rises, more carriers are able to migrate into the deepest LLCs and recombine from the lowest energy levels, leading to a red shift of the PL peak. On further increasing temperature, more and more carriers are thermally excited out from the deepest LLCs, and may hop into the LLCs with shallower localization energy, i.e. the shallower LLCs, where carriers can populate the higher energy states. This process leads to a temperature-induced blue shift of the emission peak when the rate of the shift overcomes the rate of temperature-induced band-gap shrinkage. When temperature continuously increases to approach to RT, most of the carriers cannot be trapped in the deepest LLCs and become randomly redistributed in all LLCs again, due to the increased thermal energy. Thus, the blue-shift behavior becomes less significant. And then the temperature-induced band-gap shrinkage starts to dominate in the high temperature range, giving rise to the red-shift process as predicted by the well known Varshni equation [19].
Fig. 3 Temperature-dependent PL peak energy for three Samples W24, W33, and W39, denoted by black, red, and blue squares, respectively. The dot lines are guide for eye.
The light emission from LLCs cannot only induce the S-shape shift of PL peak energy, but also influence the variation of PL spectral bandwidth with increasing temperature. The temperature-dependent PL spectral bandwidth, obtained by fitting the PL spectra with Gaussian functions, are normalized individually and exhibited for all samples in Fig. 4 . For comparison, the normalized temperature-dependent PL spectral bandwidth of bulk GaN obtained from the GaN template are also shown in Fig. 4 and denoted by green diamonds.
Fig. 4 Normalized PL spectral bandwidth as a function of temperature for Samples W24, W33, W39, and bulk GaN, denoted by black squares, red circles, blue triangles, and green diamonds, respectively. The dot lines are guide for eye.
It is widely known that the increase of temperature can provide more energy for carriers to populate higher energy states, leading to a monotonic increase of the PL spectral bandwidth toward RT, as shown in Fig. 4 for bulk GaN. However, the temperature dependence of PL spectral bandwidth of our InGaN/GaN MQW samples is totally different from that of bulk GaN. It can be seen in Fig. 4, as temperature increases, the spectral bandwidth of W24 decreases at first, and then increases a little at high temperature. This phenomenon can be understood by taking into account the existence of a large amount of LLCs in high-In-composition InGaN QWs [20,21 ]. At very low temperature, photo-generated carriers do not have enough thermal energy to migrate into the deepest LLCs. They randomly distribute and recombine radiatively in all LLCs with different depths of localization energy in QWs, and correspondingly emit photons with a broad range of energy. As temperature rise, the mobility of carrier increases, the thermally activated carriers prefer to reach to the deepest LLCs. More carriers can mainly concentrate and recombine radiatively in the deepest LLCs, causing a reduction of PL spectral line bandwidth. Further increase of temperature provides enough energy for carriers to escape out from the deepest LLCs into the shallower LLCs with higher energy states and redistribute among all LLCs in the whole QW, leading to an increase of PL spectral bandwidth, e.g., in the case of Sample W24 in Fig. 4. However, for Samples W33 and W39, the increase of PL spectral bandwidth is not observed at high temperature. Instead, the spectral bandwidth of both Samples W33 and W39 decrease monotonically toward RT.
To analyze the variation of spectral bandwidth with increasing temperature in detail, the PL spectra of all samples at different temperatures are exhibited in Fig. 5 , where the PL intensity is normalized individually at each temperature in order to clearly observe the variation of the PL spectral shape with increasing temperature. As shown in Fig. 5, the PL spectra of all 3 samples in general narrow with increasing temperature, i.e., the spectral bandwidth at low temperature is larger than that at high temperature, which is consistent with the results in Fig. 4. However, for Sample W24 at high temperature, since the increase of spectral bandwidth is slight (shown in Fig. 4), the spectral broadening of W24 cannot be distinctly observed in Fig. 5.
Fig. 5 Narrowing of the PL spectra of Samples W24, W33 and W39 with temperature rising. The PL peak intensity is normalized individually in order to clearly observe that the narrowing mainly occurs at the higher energy side with temperature increasing. The arrows indicate the direction of temperature rising.
On the other hand, it is also noticed in Fig. 5 for Samples W33 and W39 that the lower energy side of PL spectra are almost unchanged with varying temperature, while the narrowing of the PL spectra with increasing temperature occurs mainly on the higher energy side, which contradicts the commonly expectation that the thermally activated carriers prefer to populate higher energy states and the higher energy side of PL spectra should broaden when temperature rises. We attempt to explain such an unusual result by assuming the existence of NRR centers in the regions of individual LLCs in InGaN QWs with high In content. Indeed, this assumption is reasonable since the individual LLC regions may span few tens to hundreds of nanometers in lateral dimension [22] and can accommodate NRR centers induced by, e.g. V-pits [23]. It is known that in the LLC regions electron-hole pairs can better avoid to be captured by the NRR centers outside the LLC regions and recombine radiatively over there. However, the existence of V-pits or dislocations in the LLC regions has been observed directly [6,13 ]. The defects which act as NRR centers in the LLC regions will weaken the light emission of individual LLCs [6]. With increasing In content in InGaN QWs, the enhanced NRR process in the LLC regions may even makes the individual bright LLCs turn to be dark at high temperature [24]. On the other hand, it is suggested that in the regions of shallow LLCs which emit high-energy photons, few NRR centers may exist and lower the energy barrier for carriers to recombine non-radiatively, while in the regions of deep LLCs the NRR process can be ignored [22,23 ]. The difference of activation barrier to NRR process between deep and shallow LLCs may depend on the local strain energy of individual localization centers. The local strain is extremely sensitive to In content in the vicinity of NRR centers, resulting in higher barriers for In-rich localization centers which act as deep LLCs [23].
To make a clear illustration, the possible explanation of the temperature-dependent monotonic variation of PL spectral bandwidth for Samples W33 and W39 is depicted in Fig. 6 .
Fig. 6 Schematic diagrams indicating the possible mechanism of the reduction of PL spectral bandwidth with increasing temperature for Samples W33 and W39. (a) At low temperature, carriers exist in all LLCs where the RR process dominate, (b) at middle temperature, more carriers move into the deepest LLCs and the RR process still dominate in all LLCs, (c) at high temperature, carriers redistribute in all LLCs while the NRR process dominate the recombination in shallow LLCs.
Based aforementioned discussions and the schematic diagrams in Fig. 6, for the Samples W33 and W39, at the beginning, the decrease of PL bandwidth from very low temperature to middle temperature can be ascribed to the thermal migration of carriers into deep LLCs, which also cause a red shift of PL peak energy (e.g., from 10 K to 110 K for Sample W33 in Fig. 3 and Fig. 4). As temperature increases further, for example, from middle temperature to high temperature in Fig. 6, more carriers can be thermally activated and immigrate into shallow LLCs. Although the carriers can redistribute in all LLCs with different depths of localization energy, the NRR process starts to dominate the recombination in shallow LLCs at high temperature, while in deep LLCs the RR process is still very efficient. This may lead to the narrowing of PL spectral bandwidth when temperature rises toward RT (e.g., above 110 K for Sample W33 in Fig. 4).
In this paper, Sample W24 with the smallest In composition exhibits the best material quality, where the NRR process can be ignored in all LLC regions and the photo-generated carriers can only be captured by the NRR centers outside the individual LLC regions even at high temperature. Thus, the increase of PL spectral bandwidth, induced by the radiative recombination of thermally activated carriers in shallower LLCs with higher energy stats, is observed at the high temperature range. While for Samples W33 and W39 with higher In content, the NRR centers, induced by more defects, may exist and dominate the recombination process in the shallow LLCs at high temperature, causing the narrowing of the PL spectra on the higher energy side at high temperature in Fig. 5.
In fact, it is understandable that the density of defects is strongly increased with increasing TMIn flux during the epitaxial growth of InGaN well layers. With increasing TMIn flux, since more In atoms will be incorporated in QW layers during MOCVD growth, the lattice mismatch between QWs and barriers becomes larger and the strain relaxation in InGaN well layer may occur [9], generating more dislocations or other defects, e.g., more V-pits generated from the stacking mismatch boundaries induced by stacking faults within the high-In-content InGaN/GaN MQWs [16]. In addition, for the case of high TMIn flow, because of the presence of an excess indium coverage layer at the InGaN surface during the epitaxial growth [25], more indium atom may accumulate and turn to form the randomly-distributed indium-segregated regions near the upper interface of the InGaN well layers. Such instable In-rich InGaN alloy regions are inclined to be thermally decomposed during the subsequent high-temperature growth process of GaN barriers and p-GaN layers, generating more defects and deteriorating the optical properties of green light emitting InGaN/GaN MQWs [26].
4. Conclusion
In summary, the green light emission from 3 InGaN/GaN MQWs samples with different In composition is investigated by the temperature-dependent PL measurements. As the In content increases in InGaN QWs, the luminescence efficiency of MQW samples declines, which can be attributed not only to the stronger QCSE, but also to the increased number of NRR centers, e.g. dislocations or other defects, induced by the strain relaxation of InGaN well layers. It is found that in the InGaN QWs with high In content, the non-radiative centers may exist and dominate the recombination process in the shallower localized luminescence centers (LLCs) at high temperature, while in the deeper LLCs the radiative process is still dominated. As a result, for the green light emission InGaN/GaN MQW structures, the PL spectral bandwidth is decreased in the high temperature region, and at the same time the luminescence efficiency is also reduced.
Acknowledgments
The authors acknowledge the support from the National Natural Science Foundation of China (Grant Nos. 61474110, 61377020, 61376089, 61223005 and 61176126), the National Science Fund for Distinguished Young Scholars (Grant No. 60925017), Scientific Research Fund of Chongqing Municipal Education Commission (Grant No. KJ131206), and Natural Science Foundation Project of CQ CSTC (Grant No. cstc2012jjA50036).
References and links
1. S. Nakamura, N. Senoh, N. Iwasa, and S. I. Nagahama, “High-brightness InGaN blue, green and yellow light-emitting-diodes with quantum-well structures,” Jpn. J. Appl. Phys. 34(7), L797–L799 (1995).
2. T. Miyoshi, S. Masui, T. Okada, T. Yanamoto, T. Kozaki, S.-i. Nagahama, and T. Mukai, “InGaN-based 518 and 488 nm laser diodes on c-plane GaN substrate,” Phys. Status Solidi A 207(6), 1389–1392 (2010). [CrossRef]
3. J. Wu, “When group-III nitrides go infrared: new properties and perspectives,” J. Appl. Phys. 106(1), 011101 (2009). [CrossRef]
4. Y. J. Wang, S. J. Xu, D. G. Zhao, J. J. Zhu, H. Yang, X. D. Shan, and D. P. Yu, “Non-exponential photoluminescence decay dynamics of localized carriers in disordered InGaN/GaN quantum wells: the role of localization length,” Opt. Express 14(26), 13151–13157 (2006). [CrossRef] [PubMed]
5. S. De, A. Layek, A. Raja, A. Kadir, M. R. Gokhale, A. Bhattacharya, S. Dhar, and A. Chowdhury, “Two distinct origins of highly localized luminescent centers within InGaN/GaN quantum-well light-emitting diodes,” Adv. Funct. Mater. 21(20), 3828–3835 (2011). [CrossRef]
6. J. Bruckbauer, P. R. Edwards, T. Wang, and R. W. Martin, “High resolution cathodoluminescence hyperspectral imaging of surface features in InGaN/GaN multiple quantum well structures,” Appl. Phys. Lett. 98(14), 141908 (2011). [CrossRef]
7. P. G. Eliseev, M. Osinski, J. Lee, T. Sugahara, and S. Sakai, “Band-tail model and temperature-induced blue-shift in photoluminescence spectra of InxGa1-xN grown on sapphire,” J. Electron. Mater. 29(3), 332–341 (2000). [CrossRef]
8. S. Nakamura, “The roles of structural imperfections in InGaN-based blue light-emitting diodes and laser diodes,” Science 281(5379), 956–961 (1998). [CrossRef] [PubMed]
9. D. Holec, Y. Zhang, D. V. S. Rao, M. J. Kappers, C. McAleese, and C. J. Humphreys, “Equilibrium critical thickness for misfit dislocations in III-nitrides,” J. Appl. Phys. 104(12), 123514 (2008). [CrossRef]
10. V. Fiorentini, F. Bernardini, F. Della Sala, A. Di Carlo, and P. Lugli, “Effects of macroscopic polarization in III-V nitride multiple quantum wells,” Phys. Rev. B 60(12), 8849–8858 (1999). [CrossRef]
11. J. C. Zhang, D. S. Jiang, Q. Sun, J. F. Wang, Y. T. Wang, J. P. Liu, J. Chen, R. Q. Jin, J. J. Zhu, H. Yang, T. Dai, and Q. J. Jia, “Influence of dislocations on photoluminescence of InGaN/GaN multiple quantum wells,” Appl. Phys. Lett. 87(7), 071908 (2005). [CrossRef]
12. T. Langer, A. Kruse, F. A. Ketzer, A. Schwiegel, L. Hoffmann, H. Jonen, H. Bremers, U. Rossow, and A. Hangleiter, “Origin of the “green gap”: Increasing nonradiative recombination in indium-rich GaInN/GaN quantum well structures,” Phys. Status Solidi C 8(7–8), 2170–2172 (2008).
13. N. K. van der Laak, R. A. Oliver, M. J. Kappers, and C. J. Humphreys, “Role of gross well-width fluctuations in bright, green-emitting single InGaN/GaN quantum well structures,” Appl. Phys. Lett. 90(12), 121911 (2007). [CrossRef]
14. D. G. Zhao, D. S. Jiang, L. C. Le, L. L. Wu, L. Li, J. J. Zhu, H. Wang, Z. S. Liu, S. M. Zhang, Q. J. Jia, and H. Yang, “Effect of In incorporation parameters on the electroluminescence of blue-violet InGaN/GaN multiple quantum wells grown by metalorganic chemical vapor deposition,” J. Alloys Compd. 540, 46–48 (2012). [CrossRef]
15. D. G. Zhao, D. S. Jiang, J. J. Zhu, H. Wang, Z. S. Liu, S. M. Zhang, Y. T. Wang, Q. J. Jia, and H. Yang, “An experimental study about the influence of well thickness on the electroluminescence of InGaN/GaN multiple quantum wells,” J. Alloys Compd. 489(2), 461–464 (2010). [CrossRef]
16. H. K. Cho, J. Y. Lee, G. M. Yang, and C. S. Kim, “Formation mechanism of V defects in the InGaN/GaN multiple quantum wells grown on GaN layers with low threading dislocation density,” Appl. Phys. Lett. 79(2), 215–217 (2001). [CrossRef]
17. M. Rao, D. Kim, and S. Mahajan, “Compositional dependence of phase separation in InGaN layers,” Appl. Phys. Lett. 85(11), 1961–1963 (2004). [CrossRef]
18. J. Danhof, C. Vierheilig, U. T. Schwarz, T. Meyer, M. Peter, and B. Hahn, “Temperature-dependent photoluminescence measurements on a sub-micrometer length scale on green light emitting InGaN/GaN quantum wells,” Phys. Status Solidi B 248(5), 1270–1274 (2011). [CrossRef]
19. Y. Varshni, “Temperature dependence of the energy gap in semiconductors,” Physica 34(1), 149–154 (1967). [CrossRef]
20. F. B. Naranjo, M. A. Sanchez-Garcia, F. Calle, E. Calleja, B. Jenichen, and K. H. Ploog, “Strong localization in InGaN layers with high In content grown by molecular-beam epitaxy,” Appl. Phys. Lett. 80(2), 231–233 (2002). [CrossRef]
21. Y. J. Yang, P. Ma, X. C. Wei, D. Yan, Y. F. Wang, and Y. P. Zeng, “Design strategies for enhancing carrier localization in InGaN-based light-emitting diodes,” J. Lumin. 155, 238–243 (2014). [CrossRef]
22. J. Bruckbauer, P. R. Edwards, S. L. Sahonta, F. C. P. Massabuau, M. J. Kappers, C. J. Humphreys, R. A. Oliver, and R. W. Martin, “Cathodoluminescence hyperspectral imaging of trench-like defects in InGaN/GaN quantum well structures,” J. Phys. D Appl. Phys. 47(13), 135107 (2014). [CrossRef]
23. S. De, D. Kumar Das, A. Layek, A. Raja, M. Kumar Singh, A. Bhattacharya, S. Dhar, and A. Chowdhury, “Optoelectronic behaviors and carrier dynamics of individual localized luminescent centers in InGaN quantum-well light emitting diodes,” Appl. Phys. Lett. 99(25), 251911 (2011). [CrossRef]
24. A. Vaitkevičius, J. Mickevičius, D. Dobrovolskas, Ö. Tuna, C. Giesen, M. Heuken, and G. Tamulaitis, “Influence of quantum-confined Stark effect on optical properties within trench defects in InGaN quantum wells with different indium content,” J. Appl. Phys. 115(21), 213512 (2014). [CrossRef]
25. J. Northrup, “Impact of hydrogen on indium incorporation at m-plane and c-plane In0.25Ga0.75N surfaces: First-principles calculations,” Phys. Rev. B 79(4), 041306 (2009). [CrossRef]
26. Y.-T. Moon, D.-J. Kim, K.-M. Song, C.-J. Choi, S.-H. Han, T.-Y. Seong, and S.-J. Park, “Effects of thermal and hydrogen treatment on indium segregation in InGaN/GaN multiple quantum wells,” J. Appl. Phys. 89(11), 6514–6518 (2001). [CrossRef]
