Tong Liua,
Shujie Jiao*ab,
Hongwei Liangc,
Tianpeng Yangd,
Dongbo Wanga and
Liancheng Zhaoa
aSchool of Materials Science and Engineering, Harbin Institute of Technology, Harbin 150001, China. E-mail: shujiejiao@gmail.com; Tel: +86 451 86417763
bKey Laboratory for Photonic and Electric Band gap Materials, Ministry of Education, Harbin Normal University, Harbin 150001, China
cSchool of Physics and Optoelectronic Engineering, Dalian University of Technology, Dalian 116024, China
dEpiTop Optoelectronic Co., Ltd., Pingxiang 337000, China
First published on 30th March 2015
The structural and optical properties of near-ultraviolet (UV) multiple quantum well (MQW) structures using quaternary AlInGaN as the well layers have been investigated. The composition of the barrier layers is determined by three In0.08Ga0.92N/AlxInyGa1−x−yN multiple quantum well samples with varying Al content in the barrier layers. The compositions of the well and barrier layers are estimated from the high-resolution X-ray diffraction (HRXRD) results. In spite of the larger lattice mismatch, the remarkable enhancement of the photoluminescence (PL) intensity of the MQWs sample with AlInGaN as the well layers is attributed to the increase in the carrier localized states induced by the increase in the compositional fluctuation in the AlInGaN well layers. The S-shaped temperature-dependence of the PL peak energy indicates the existence of localized states induced by the potential fluctuations. The magnitude of the carrier localization, which is estimated by the fitting results, is significantly increased in the Al0.11In0.13Ga0.76N/Al0.16In0.045Ga0.795N MQWs due to the improvement of the spatial potential fluctuations using quaternary AlInGaN as the well layers.
Recently, it has been suggested that the optical properties and carrier confinement were enhanced when quaternary AlInGaN was used in place of GaN barrier layers in near-UV LEDs,7 as the lattice constant and band gap of the quaternary AlInGaN can be independently controlled by adjusting the composition of AlN, InN and GaN.8–11 We have reported that the carrier localization can be improved by optimizing the Al composition of the barrier layers in blue In0.20Ga0.80N/AlInGaN MQWs with an emission wavelength around 450 nm.9
In recent years, several research groups have reported AlInGaN/AlInGaN near-UV MQWs LEDs with greater carrier localization12,13 due to a uniform growth condition and small lattice mismatch between the well and barrier layers. Mee-Yi Ryu et al. suggested that the PL properties dominated by localized carriers in the AlInGaN MQWs agree well with those of InGaN/GaN MQWs with a high density of localized states.14,15 P. Lefebvre et al. reported that the in-plane localization of carriers induced by local potential fluctuations could be enhanced by the optimum well width.16 However, the mechanism of the great carrier localization in the quaternary well layers is not yet clear. In this work, we study the carrier localization effect of the Alx′Iny′Ga1−x′−y′N/AlxInyGa1−x−yN MQW structures. As the carrier localization can also be influenced by the composition of the barrier layers, three In0.08Ga0.92N/AlxInyGa1−x−yN MQWs samples were first grown with a similar emission energy to optimize the barrier composition.
High-resolution X-ray diffraction was employed to analyze the structural parameters. Raman measurements were carried out to verify the existence of the In-rich clusters. Temperature-dependent photoluminescence spectra were measured to investigate the carrier localization using an iHR320 spectrometer with a He–Cd laser at 325 nm as the exciting source.
![]() | (1) |
Egu(AluIn1−uN) = uEg(AlN) + (1 − u)Eg(InN) − u(1 − u)b(AlInN) | (2) |
Egv(InvGa1−vN) = vEg(InN) + (1 − v)Eg(GaN) − v(1 − v)b(InGaN) | (3) |
Egw(AlwGa1−wN) = wEg(AlN) + (1 − w)Eg(GaN) − w(1 − w)b(AlGaN) | (4) |
u = (1 − x + y)/2, v = (1 − y + z)/2, w = (1 − x + z)/2 | (5) |
![]() | ||
Fig. 1 HRXRD (0002) ω-2θ scans of the MQW samples. Samples 1–3 are the MQWs with In0.08Ga0.92N well layers and sample 4 is the MQWs with Al0.11In0.13Ga0.76N well layers. |
Sample | XRD | PL | ||||||
---|---|---|---|---|---|---|---|---|
Well | Barrier | 10 K | 300 K | |||||
Al (%) | In (%) | Al (%) | In (%) | Energy (eV) | FWHM (meV) | Energy (eV) | FWHM (meV) | |
1 | 0 | 8 | 15 | 5.5 | 3.194 | 52 | 3.173 | 123 |
2 | 0 | 8 | 16 | 4.5 | 3.232 | 52 | 3.198 | 123 |
3 | 0 | 8 | 17 | 4 | 3.199 | 54 | 3.169 | 126 |
4 | 11 | 13 | 16 | 4.5 | 3.171 | 61 | 3.150 | 134 |
The lattice mismatch f between the well and barrier layers were numerically evaluated to be 0.65%, 0.78%, 0.84% and 1% for the MQW samples 1–4, respectively, using the following expression:18
f = (aW − aB)/aB | (6) |
It was confirmed that the In-segregation effect taking place in the well layers of InGaN-based blue LEDs is the source of the high radiative recombination efficiency.19,20 Raman scattering measurements were carried out at room temperature to study the In-rich clusters in these near-UV MQW samples. The inset of Fig. 2 shows the Raman scattering of sample 4 over the measurement range of 100–1000 cm−1. The scale was enlarged to get a clear picture of the Raman vibration of the In-rich clusters, as shown in Fig. 2. Two vibration modes, which can be clearly seen in these four samples at around 432 cm−1 and 453 cm−1, were considered to be from the vibration of InN9 and indicate the existence of nearly pure InN regions in the quantum wells. Therefore, local potential fluctuations could be formed around the In-rich regions. Evidently, the intensities of the two vibration modes from sample 4 become stronger as induced by the higher In content in the well layers. Moreover, the shift to lower frequency of the vibration peak positions in sample 4 is considered to be also affected by the increase in the In content.21
The PL spectra measurements of the MQW samples were performed for 10 K (Fig. 3(a)) and 300 K (Fig. 3(b)). At 10 K and 300 K, the PL spectra of the MQW samples are dominated by a sharp emission peak. In the blue In0.20Ga0.80N/AlxInyGa1−x−yN MQWs, the PL emission peak energy exhibits a clear redshift with increasing Al content in the barrier layers due to the increase in the quantum-confined Stark effect (QCSE).9 In this work, however, the PL emission peak energies of samples 1–3 with increasing Al content in barrier layers do not present this redshift. We speculate that the reduced band offsets in the QWs may be responsible for this phenomenon, due to the larger band gap of the In0.08Ga0.92N well layers compared to that of the blue MQWs. It is worth noting that the PL intensity of sample 4 is nearly twice as strong as that of the samples with In0.08Ga0.92N well layers, in spite of the larger lattice mismatch of up to 1% in the QWs. Moreover, the FWHM of the PL curve from sample 4 is larger than that of the other samples at either 10 or 300 K, as shown in Table 1. Therefore, we speculate that the intense emission of sample 4 is attributed to the increase in the local compositional variation by introducing Al atoms into the InGaN well layers, leading to the enhancement of the spatial potential fluctuation and carrier localization effect, and thus, the radiative recombination efficiency is significantly improved.
The temperature-dependent PL spectra were measured in the temperature range from 10 to 300 K to support our speculation that the carrier localization effect is enhanced by increasing the compositional fluctuation. The S-shaped temperature-dependence of the PL peak energy in these MQW samples, that is, the redshift–blueshift–redshift, can be clearly observed in Fig. 4. This behavior is a well-known manifestation of the existence of localized carriers in the QWs.22,23 To estimate the magnitude of the carrier localization effect in these MQW samples, the measurement results of the emission energy were fitted by the modified Varshni empirical equation:24
E(T) = E(0) − αT2/(T + β) − σ2/KBT | (7) |
This equation is only valid for temperatures above 70 K. E(0) is the band gap energy at 0 K; α and β are the Varshni thermal coefficients; KB is the Boltzmann constant; σ indicates the magnitude of the localization effect, namely, a larger value of σ means a stronger carrier localization effect. The fitting results are plotted by the solid lines and the fitting parameters are also given in Fig. 4(a)–(d). The magnitude of the carrier localization increases significantly in sample 4 compared with the In0.08Ga0.92N/AlxInyGa1−x−yN MQW samples, indicating that more carriers are confined in the localized states in the QWs. Therefore, the radiative recombination efficiency will be improved significantly by introducing Al atoms into the InGaN well layers, due to the increase in the local potential fluctuation.
Alternatively, the transition temperature from the redshift to blueshift of the MQW samples 1–4 is 40, 80, 80 and 100 K, and that from blueshift to redshift is 110, 140, 150 and 170 K, respectively. The blueshift energy between the two transition temperatures is about 4, 3, 1 and 6 meV for these four MQW samples. The expected temperature-reduced band gap shrinkage (redshift) is about 5, 9, 11 and 13 meV for the corresponding temperature regions, which were estimated from GaN.25 Therefore, the actual blueshift energy of the emission peaks with respect to the band edge is about 9, 12, 12 and 19 meV. The larger amount of blueshift energy means a stronger carrier localization, which accords well with the variation tendency of the magnitude of the localization effect (σ).
This journal is © The Royal Society of Chemistry 2015 |