Open Access
Add to BibSonomyAbstract
GaN-based light-emitting diodes (LEDs) grown on c-plane vicinal sapphire substrates are fabricated and characterized. Based on the material quality and electrical properties, the LED with a 0.2° tilt sapphire substrate (device A) exhibits the lowest defect density and high performance, while the LED with a 1.0° tilt sapphire (device D) exhibits the highest one. At 2 mA, the extremely enhanced output power of 23.3% indicates of the reduction of defect-related nonradiative recombination centers in active layers for the device A. At 60 mA, the improved value is up to 45.7%. This is primarily caused by the formation of indium quantum dots in MQW which provides an increased quantum efficiency.
©2010 Optical Society of America
1. Introduction
Recently, high-brightness GaN-based light-emitting diodes (LEDs) have attracted great interest for various applications [1,2]. The sapphire substrate is the most commercialized material due to the cheaper cost and mature technique. However, due to the significantly high defect density (108~1010 cm−2) within epilayers [3], the structural and optical properties of GaN LEDs are still limited. It’s revealed that around 16% lattice mismatch and larger difference of thermal expansion coefficients exist between GaN layer and sapphire substrate [4]. Previously, as compared with the exactly c-axis oriented sapphire (i.e., 0°-tilt c-plane sapphire), the tiny c-plane misoriented sapphire substrates were employed by many groups to reduce threading dislocations and native defects in epilayers [5–11]. Both the relieved effect on lattice mismatch at sapphire/GaN interface [5] as well as more constrained crystalline orientation [6] in GaN epilayers, are induced by vicinal sapphires. Previous reports have demonstrated that the crystal quality of GaN epilayer is very sensitive to the tilt angle. Despite of the specific tilt directions (some tilt toward m-plane [5,6,8,10] and the other a-plane [7]), experimentally, the optimized tilt angle is around 0.15~0.3°. In addition, it is known that the GaN growth mode shifts form two-dimentional island/spiral mode toward microstep/macrostep mode with increasing the sapphire tilt angle [5]. Defect formation exists in the region where two-dimentional islands coalesce during GaN growth process [9]. Also, the growth mode is gradually changed by vicinal sapphires with tilt angle larger than 0.15° [5]. Thus the defect density can be reduced. On the other hand, when the microstep shifts to macrostep growth mode with increasing the tilt angle larger than 0.5° [8], the rougher GaN surface morphology is found. This causes the excess degradation of GaN crystal quality. Yet, the previous discussion was focused just on the qualitatively material/optical analyses. Few report touched limited electrical analyses of LEDs grown on misoriented sapphires [10]. Moreover, an opposed hypothesis was reported by Shen et al. [11]. They suggested that the crystal quality was always improved with increasing the sapphire tilt angle, even at 2°. Accordingly, for obtaining a more accurate trend, some quantitatively electrical characteristics must be carried out. In this work, GaN LEDs grown on vicinal sapphire substrates (0.2°-, 0.35°-, 0.5°-, 1.0°-tilt c-plane sapphire) are fabricated and characterized. The complete electrical and optical characterizations are clearly presented in this work. All improved properties including better turn-on voltage (@ 20mA), reverse-biased leakage current, optimized ideality factor (1.41), larger breakdown voltage (−55.5 volt), and reduced junction capacitance are obtained when a 0.2°-tile sapphire substrate is used. This certainly indicates the improved crystal quality. In addition, from the output power performance under a low current level of 3 mA, the excellent improvement (at most 23.3%) can be directly related to the reduction of defects in MQW layers of LEDs. When the driving current is up to 60 mA, the improved value is up to 45.7%. This is primarily caused by the formation of indium quantum dots in MQW layers which provides an increased quantum efficiency.
2. Experimental
In this work, the epitaxial layers of GaN-based LED samples were grown on sapphire substrates by a Thomas Swan metalorganic chemical vapor deposition (MOCVD) system. The substrates employed were scribed from 2 in. exactly_(0001)-oriented (c-plane) sapphire wafers and misoriented ones with their c-axis offset by a very small angle of 0.2°, 0.35°, 0.5° and 1.0°, respectively, toward the direction (m-plane). Herein, precise X-ray diffraction measurements were used to check accurate sapphire orientations. All samples were grown for the same run and with the same growth process regardless of the various sapphire substrates. Trimethyl gallium (TMGa), and trimethyl indium (TMIn) were used as the group-III sources. Ammonia (NH3) was used as the group-V source. Disilane (Si2H6) and bis (cyclopeantadienyl) magnesium (Cp2Mg) were n- and p-type doping sources, respectively. H2 and N2 were served as carrier gases for GaN and InGaN epilayer, respectively. The LED structure consisted of a 2 μm GaN nucleation layer, a 2 μm Si-doped n-GaN layer (n = 1 × 1018cm−3), 15- period Si-doped InGaN/GaN multiple quantum well (MQW) as active layers, and a 0.5 μm Mg-doped p-GaN layer (p = 4 × 1017cm−3). After the epitaxial growth, the structural properties of as-grown films were characterized by high-resolution (HR) XRD (Rigaku 18 kW Rotating Anode X-ray Generator), micro-Raman (Labram HR, with a light source 532 nm laser, 1800 gr/mm grating and 0.47 cm−1/pixel), and AFM measurements. In this study, the GaN films grown on 0.2°-, 0.35°-, 0.5°-, 1.0°-tilt c-plane sapphires were denoted as film A, B, C, and D, respectively. The wafers were first cleaned by acetone and deionized water sequentially. Then, an inductively coupled plasma (ICP) system was utilized to define mesa regions. A 250 nm-thick indium-tin oxide (ITO) layer was deposited on the p-GaN layer by an electronic beam evaporator. Cr/Pt/Au (50/150/2000 nm) metal was deposited as n- and p- pad Ohmic contacts. The ITO layer and n-p pads were activated for 30 min. in a nitrogen ambient at 470°C and 380°C, respectively. These wafers were diced into individual chips with the dimension of 300 × 300μm2. The fabricated devices A, B, C, and D corresponded to the related films A, B, C, and D, respectively. The chips were attached and boned to TO-3 submounts for electrical and optical tests. The current-voltage (I-V) and capacitance-voltage (C-V) characteristics, breakdown voltages, and junction capacitance of studied devices were measured at room temperature by a semiconductor parameter analyzer (Agilent 4156C). The output powers were measured using an integrating sphere instrument built on a Keithley 2400 power source.
3. Results and discussion
Figure 1(a) shows the HR XRD (2θ-scan) spectra of full-width at half maximum (FWHM) of symmetric (002) and asymmetric (102) diffractions for the n-GaN films grown on various vicinal substrates. It is known that the XRD (002)-reflecting plane is related to screw and mixed dislocations while XRD (102)-reflecting plane is related to edge dislocations in GaN films. The measured (002) ((102)) FWHMs are 0.12° (0.09°), 0.114° (0.087°), 0.111° (0.084°), and 0.09° (0.069°) for GaN films D, C, B, and A, respectively. A gradually decreased FWHM indicates that an improved crystal quality of the as-grown film. The defect density Ddis in GaN film can be approximately evaluated as [11]:
where β is the FWHM of XRD peaks, and b the length of the Burger vector of the corresponding dislocation. The defect density Ddis is proportional to the square values of FWHM. Based on this correlation, the estimated mixed (edge) defect densities of GaN films B, C, and D are 1.52 (1.48), 1.61 (1.59), and 1.78 (1.70) times as compared with that of the film A. Thus, a similar tendency caused by the two types of defects is shown and an improvement of the optimized angle (0.2°) is observed. This result doesn’t agree well with the previous report [11], since once the tilt angle is larger than 0.2°, a degraded crystal quality is presented. In other words, the improvement in defect density is hence limited. Figure 1(b) illustrates Raman spectra of the studied LED films. It is known that the frequency of nonpolar E2 (high) mode is sensitive to the internal strain of GaN films which is dominantly induced by lattice mismatch and thermal coefficients difference between the as-grown film and sapphire substrate [12]. The E2 mode of a strain-free GaN sample has a frequency of 568 cm−1, and the shift of the E2 phonon with biaxial strain is 4.1 cm−1/GPa [7]. In this work, all E2 phonon modes are located at 569.4 cm−1. The same compressive strain of around 0.341GPa is generated in all studied LED films regardless of the sapphire tilt angle. Experimentally, the shift of E2 mode in this work is not clear. This is probably due to the comparably small variation of internal strain (<0.114 GPa) (induced by the minor difference, at most 1.78 times of magnitude of defect density) between studied LED films and the limited pixel resolution of the employed Raman instrument. On the other hand, based on the A1 (LO) modes (located at 734 cm−1) in Fig. 1(b), the Raman peaks become weak and broad with increasing the sapphire tilt angle. This is attributed to the stress effect [12].
Fig. 1 (a) XRD FWHM measurements about (002) and (102)-reflecting planes of studied films, and (b) Raman spectra of the studied films.
Figure 2 illustrates the turn-on voltages of the studied LED devices under the operation current of 20 mA at room temperature (RT). Around 5~6 chips are tested for the same type of LEDs. It’s inferred that a small but uniform variation is obtained between different types of samples (i.e., different sapphire tilt angles). It is believed that the defect density in LED films is not related to the difference of high-operation (>20mA) turn-on voltages [13], especially for the nearly comparable crystal qualities in this work. On the other hand, these values are suggested to be dominantly determined by the surface contact properties. Thus, the measurement of root-mean-square (RMS) roughness of as-grown films are developed by an AFM instrument. About 5 random regions (with area of 20 × 20 μm2) on the same type of as-grown film are scanned (with the same scanning frequency). This result is shown in the inset of Fig. 2. Obviously, the result is highly correlated with Fig. 2, which confirms the prediction as mentioned above. The gradually formed macrostep growth mode will degrade the surface property of as-grown films [5,6,8] with increasing the sapphire tilt angle. Thus a higher turn-on voltage of a LED device is observed.
Fig. 2 Turn-on voltages of studied devices as a function of sapphire tilt angle. The surface root-mean-aquare roughness of the studied films as a function of sapphire tilt angle is illustrated in inset.
Figure 3(a) demonstrates current-voltage (I-V) characteristics of studied LED devices under the reverse bias condition at RT. The reverse-biased leakage current prominently corresponds to crystal and electrical properties of a LE D device. It can be seen that under lower operation voltage (|V| <20 volt), an irregular variation of leakage currents is found. From the previous results, due to the small change in defect densities (no larger than 1.78 times of magnitude), it’s possible that the variation of leakage currents between studied LEDs is affected mainly by individual samples over the minor quality change induced by sapphire substrates with different tilt angles. On the other hand, under higher operation voltage (|V| >20 volt), the linear-like dependences of the logI-V relations are clearly shown due to the saturation of leakage paths induced by defects within LEDs. A reasonably and significantly lower leakage current of the device A (0.2°-tilt) is obtained. The leakage current becomes higher with increasing the sapphire tilt angle, which demonstrates the increased defect density in LEDs. Interestingly, a comparably lower leakage current of the device A is measured, as compared with those of other devices. For instance, at −40 volt, the leakage current of device A is around 10−6 A, while those of other LEDs are around 10−4 A. This is caused by the improved crystal quality (at most 50% reduction in defect density) for the optimized device A, as revealed in the XRD results in Fig. 1(a). In addition, under a very high voltage level (|V| >50 volt), the breakdown regions of LEDs are presented. For a nitride-based diode grown on a sapphire substrate, it’s believed that under a high reverse-biased voltage level, the defect-related tunneling current becomes prominent and finally dominates the Zener breakdown over the avalanche breakdown. This is caused by the remarkably increased defect density generated by a high electrical field in active layers. The phenomenon is dominantly observed especially in a MQW structure with abrupt barriers/wells, and high carrier concentration InGaN well layers. About 5 chips for the same type of LEDs are selected randomly to measure the breakdown voltage operated at the temperature of 300 K, as shown in Fig. 3(b). The average breakdown voltages are around 55, 53.5, 52, and 51 volts, for devices A, B, C, and D, respectively. It can be concluded that, the higher breakdown voltage can be observed with decreasing the sapphire tilt angle (until the optimized 0.2°) due to the improved crystal quality. Furthermore, in the Fig. 3(b), additional 2~3 random chips for the same type of LEDs are selected to measure the breakdown voltage operated at 400 K. Experimentally, the breakdown voltage is lower due to the higher probability of defect-assisted tunneling current with increasing the temperature. The Zener breakdown mechanism is dominant as it is expected.
Fig. 3 (a) The I-V characteristics of LED devices at the reverse-biased voltage region, and (b) The breakdown voltage, operated at 300 and 400 K respectively, as a function of sapphire tilt angle.
It’s known that the current transports in active layers are dominated by diffusion and recombination components when the ideality factors approach to 1 and 2, respectively [14]. Moreover, ideality factor >>2 indicates a deep-level-assisted tunneling component [14]. Ideality factors extracted from lower current levels (<10−6 A in this work) are mainly influenced by shunt resistances, while those extracted from higher current levels (>10−4 A) are influenced by series resistances of a diode. Ideality factors are valid and give insights into the I-V characterization of a single LED junction (i.e., the active layer itself) only in the linear region of the logI-V relation, which corresponds to the minimum ones at the intermediate current region [15]. In order to further study the effect of slight sapphire misorientations on active layers of LEDs, ideality factors at RT for studied LEDs as a function of forward currents (10−10~10−2 A) are extracted and shown in Fig. 4 . The minimum ideality factors are 1.41, 1.44, 1.46, and 1.62 for devices A, B, C, and D, respectively. Clearly, with decreasing the tilt angle, the reduced ideality factors show the improved crystal quality of the active layer. The current transport in active layers belongs to a drift/diffusion emission over defect-assisted recombination. The optimized ideality factor of 1.41 is dramatically lower than previously reported InGaN/InGaN and InGaN/GaN MQW systems [15-16].
In order to further study the effect of sapphire misorientations on MQW of the studied devices, the capacitance-voltage (C-V) measurement is also carried out. Previously, the information about junction defects of a diode comes from the dependence of capacitance on applied voltage (C-V diagram), where the junction capacitance and apparent charge distribution (ACD) profiles were extracted [17]. Junction capacitance and ACD profiles reflect the generation/propagation of defects, especially in the interface of heterojunctions or the semiconductor-based border with high carrier concentration [17]. Generally, the total junction capacitance in a diode includes two components, i.e., space-charge and diffusion capacitances [18]. The space-charge capacitance is built by the depletion region in MQW, while the diffusion capacitance by the storage of minority carriers in MQW. In this work, the relation of junction capacitance as a function of applied voltage (−25~5 volt) is plotted in Fig. 5(a) . The measuring frequency of 1 M Hz is chosen for obtaining a good signal-to-noise ratio as well as limiting the contribution caused by the diffusion capacitance (hence providing a reasonable space-charge capacitance) [17,18]. A gradually decreased junction capacitance is obtained at an elevated reverse-biased voltage, suggesting that the depletion region swept through MQW extends. Oppositely, when an increased forward voltage is applied, due to the suppression of depletion layer, an increased junction capacitance is observed. Prior to reaching the threshold voltage [19] (around 2.5 volt in this work), a maximum junction capacitance is shown. Obviously, these maximum capacitances of studied LEDs, i.e., 305, 268, 241, and 99.2 pF for device B, C, D, and A, respectively, are characteristic values and worthy to be further studied. As the applied voltage is larger than the threshold voltage, the followed drastic precipitation of the junction capacitance is inferred by the strong recombination (rapid decrease in the amount of charge carriers) in MQW [20]. The junction capacitance, extracted at −25 volt operation voltage as a function of vicinal sapphire, is shown in Fig. 5(b). A larger reverse-biased voltage is chosen to limit the diffusion capacitance. About 3~5 random chips are selected for individual type of the studied LEDs. As can be noticed, the junction capacitance become larger with increasing the sapphire tilt angle, indicating that the charged defects/deep-level states in MQW are increased [17,21]. This result is highly correlated with the previous statements in Figs. 1(a), 3 and 4. It is worth noting that the variation of junction capacitance is regular but relatively small (clearly shown in Fig. 5(a)), due to the small change in defect density of MQW as mentioned above. However, the comparably larger change occurs in device A, agreeing well again that the more improvement in crystal quality is found in device A.
Fig. 5 (a) C-V characteristics of LED devices. The measuring frequency is kept at 1 MHz, and (b) Junction capacitance operated at −25 volt, as a function of sapphire tilt angle. The measuring frequency is kept at 1 MHz.
The output power and external quantum efficiency (EQE) as a function of operation current (0~100 mA) are depicted in Fig. 6 . Obvious enhancements are found with decreasing the sapphire tilt angle. Figures 7(a) and 7(b) show high-resolution TEM images near InGaN/GaN MQW layers of devices A and D. The corresponding enlarged views of MQW layers of devices A and D are shown in Figs. 7(c) and 7(d), respectively. Clearly, for device A, there are many discontinuous dark dots in InGaN wells, which indicates the serious indium nonuniformity phenomenon. The dark dots, are known as the indium-rich regions, which behave as quantum dots to provide radiative recombination centers [22]. These 0-D structures are believed to decrease the radiative lifetime of free carriers (e.g., increase radiative recombination rate) [22]. Thus the quantum efficiency could be increased. On the other hand, this phenomenon is unclear for device D, which results in a degraded quantum efficiency. In other words, the indium nonuniformity effect become relieved with increasing the sapphire tilt angle. Hence both the output power as well as EQE are reduced. It is interpreted that, as compared with the island formation on a 0.2°-tilt sapphire substrate, the micro-/macro-step formation during the MQW growth process provides more disordered surface states when a 0°-tilt sapphire substrate is used. Based on the step growth mode, the number of dangling bonds for terraces and steps are reasonably different. In contrast, the island/spiral growth mode exhibits more ordered surface states since all islands are grown in the same terrace. As a result, the indium fluctuation induced by bonding preference could be disarranged by the step growth mode easily because the growth process is influenced by different surface states. In other words, the step formation helps to suppress aggregation of indium atoms in a localized region. Hence, the uniformity of indium incorporation in MQW is greatly improved. Moreover, the light emission wavelengths are 457, 453, 451, and 446 nm for devices A, B, C, and D, respectively. Obvious wavelength blue-shift also indicates the suppression of indium-rich regions with increasing the sapphire tilt angle. A similar inference was reported previously by Nakamura et al. [10]. Yet, they did not provide further reasonable explanations about this phenomenon.
Fig. 7 TEM images near the InGaN/GaN MQW regions of devices A (a) and D (b). The enlarged views of MQW layers of devices A and D are also shown in (c) and (d), respectively.
Generally, the injection current Iinj of a LED can be expressed as [23]:
where q is the electronic charge and V the active region volume. A is the monomolecular recombination coefficient, which describes nonradiative recombination through traps and defects. B is the radiative or bimolecular recombination coefficient associated with radiative emission in LEDs. C is the Auger recombination coefficient. I leak is denoted as the leakage current diffused into p- or n-cladding layers under high injection. Based on this equation, at a higher current level (>20 mA), the optical characteristics influenced by saturated defect-related nonradiative recombination centers in MQW are unclear [21,23], while the current spreading [24], quantum-confined Stark effect (QCSE) [25], and indium-rich localized states are dominant factors. On the other hand, the decrease of optical power caused by defects becomes more prominent and convinced when the measurement is operated under a lower current level [17,21,23]. The optical measurement results under a very-low current level (1~3 mA, 0.05 mA/step), are shown in Fig. 8 . At 2 mA, the output power improvements are 23.3, 7.44, and 2.11% for devices A, B, and C, respectively, as compared with the device D. These improvements are due to the enhancement in crystal quality, highly correlating with the statements as mentioned above. Based on this experimental result, the optical property of LEDs is highly sensitive to the defect density. In addition, these improved values at 2 mA are comparably small with those of 38.6 and 45.7% at 20 and 60 mA, respectively.4. Conclusion
In summary, GaN-based LEDs grown on c-plane vicinal sapphire substrates (0.2°-(A), 0.35°-(B), 0.5°-(C), and 1.0°-(D) tilt) are fabricated and characterized. The lower turn-on voltage of around 3.22 volt (@ 20 mA), lower leakage current of 10−6 A (@-40 volt), optimized ideality factor of 1.41, higher breakdown voltage of −55.5 volt, and reduced junction capacitance of 41 pF (@-25 volt, 1 MHz) demonstrate that the device A exhibits the best crystal quality. In addition, for the device A, the excellent optical improvement of 23.3% at 2 mA is related to its superior crystal quality. Based on this experimental result, it is known that the optical property of LEDs is highly sensitive to the defect density. Also, at 60 mA, the improved value is up to 45.7%. This is primarily caused by the formation of indium-rich regions in MQW which provides an increased quantum efficiency.
Acknowledgement
Part of this work was supported by the National Science Council of the Republic of China under Contract No. NSC-97-2221-E-006-238-MY3.
References and links
1. N. C. Chen, C. M. Lin, Y. K. Yang, C. Shen, T. W. Wang, and M. C. Wu, “Measurement of junction temperature in a nitride light-emitting diode,” Jpn. J. Appl. Phys. 47(12), 8779–8782 (2008). [CrossRef]
2. M. A. Mastro, J. D. Caldwell, R. T. Holm, R. L. Henry, and C. R. Eddy Jr., “Design of Gallium Nitride Resonant Cavity Light-Emitting Diodes on Si Substrates,” Adv. Mater. 20(1), 115–118 (2008). [CrossRef]
3. X. A. Cao, J. A. Teetsov, F. Shahedipour-Sandvik, and S. D. Arthur, “Microstructural origin of leakage current in GaN/InGaN light-emitting diodes,” J. Cryst. Growth 264(1-3), 172–177 (2004). [CrossRef]
4. T. Lei, K. F. Ludwig, and T. D. Moustakas, “Heteroepitaxy, polymorphism, and faulting in gain thin-films on silicon and sapphire substrates,” J. Appl. Phys. 74(7), 4430–4437 (1993). [CrossRef]
5. S. W. Kim, H. Aida, and T. Suzuki, “The effect of a slight mis-orientation angle of c-plane sapphire substrate on surface and crystal quality of MOCVD grown GaN thin films,” Phys. Status Solidi 1(10), 2483–2486 (2004).
6. D. Lu, D. I. Florescu, D. S. Lee, V. Merai, J. C. Ramer, A. Parekh, and E. A. Armour, “Sapphire substrate misorientation effects on GaN nucleation layer properties,” J. Cryst. Growth 272(1-4), 353–359 (2004). [CrossRef]
7. Z. Y. Li, M. H. Lo, C. H. Chiu, P. C. Lin, T. C. Lu, H. C. Kuo, and S. C. Wang, “Carrier localization degree of In0.2Ga0.8N/GaN multiple quantum wells grown on vicinal sapphire substrates,” J. Appl. Phys. 105(1), 013103 (2009). [CrossRef]
8. X. Q. Shen, M. Shimizu, T. Yamamoto, Y. Honda, and H. Okumura, “Characterizations of GaN films and GaN/AlN super-lattice structures grown on vicinal sapphire (0 0 0 1) substrates by RF-MBE,” J. Cryst. Growth 278(1-4), 378–382 (2005). [CrossRef]
9. R. W. Chuang, C. L. Yu, S. J. Chang, P. C. Chang, J. C. Lin, and T. M. Kuan, “Crystal growth and characterization of AlGaN/GaN heterostructures prepared on vicinal-cut sapphire (0 0 0 1) substrates,” J. Cryst. Growth 308(2), 252–257 (2007). [CrossRef]
10. A. Nakamura, N. Yanagita, T. Murata, K. Hoshino, and K. Tadatomo, “Effects of sapphire substrate misorientation on the GaN-based light emitting diode grown by metalorganic vapour phase epitaxy,” Phys. Status Solidi 5(6c), 2007–2009 (2008). [CrossRef]
11. X. Q. Shen, H. Matsuhata, and H. Okumura, “Reduction of the threading dislocation density in GaN films grown on vicinal sapphire (0001) substrates,” Appl. Phys. Lett. 86(2), 021912 (2005). [CrossRef]
12. X. L. Tong, L. Li, D. S. Zhang, Y. T. Dai, D. J. Lv, K. Ling, Z. X. Liu, P. X. Lu, G. Yang, Z. Y. Yang, and H. Long, “The influences of laser scanning speed on the structural and optical properties of thin GaN films separated from sapphire substrates by excimer laser lift-off,” J. Phys. D Appl. Phys. 42(4), 045414 (2009). [CrossRef]
13. O. Pursiainen, N. Linder, A. Jaeger, R. Oberschmid, and K. Streubel, “Identification of aging mechanisms in the optical and electrical characteristics of light-emitting diodes,” Appl. Phys. Lett. 79(18), 2895–2897 (2001). [CrossRef]
14. J. M. Shah, Y. L. Li, Th. Gessmann, and E. F. Schubert, “Experimental analysis and theoretical model for anomalously high ideality factors (n>>2) in AlGaN/GaN p-n junction diodes,” J. Appl. Phys. 94(4), 2627–2630 (2003). [CrossRef]
15. J. Xu, M. F. Schubert, A. N. Noemaun, D. Zhu, J. K. Kim, E. F. Schubert, M. H. Kim, H. J. Chung, S. Yoon, C. Sone, and Y. Park, “Reduction in efficiency droop, forward voltage, ideality factor, and wavelength shift in polarization-matched GaInN/GaInN multi-quantum-well light-emitting diodes,” Appl. Phys. Lett. 94(1), 011113 (2009). [CrossRef]
16. D. Zhu, J. Xu, A. N. Noemaun, J. K. Kim, E. F. Schubert, M. H. Crawford, and D. D. Koleske, “The origin of the high diode-ideality factors in GaInN/GaN multiple quantum well light-emitting diodes,” Appl. Phys. Lett. 94(8), 081113 (2009). [CrossRef]
17. F. Rossi, M. Pavesi, M. Meneghini, G. Salviati, M. Manfredi, G. Meneghesso, A. Castaldini, A. Cavallini, L. Rigutti, U. Strass, U. Zehnder, and E. Zanoni, “Influence of short-term low current dc aging on the electrical and optical properties of InGaN blue light-emitting diodes,” J. Appl. Phys. 99(5), 053104 (2006). [CrossRef]
18. M. L. Lucia, J. L. Hernandez-Rojas, C. Leon, and I. Martil, “Capacitance measurements of p-n junctions: depletion layer and diffusion capacitance contributions,” Eur. J. Phys. 14, 86–89 (1993). [CrossRef]
19. X. M. Zhang, M. Y. Lu, Y. Zhang, L. J. Chen, and Z. L. Wang, “Fabrication of a high-brightness blue-light-emitting diode using a ZnO-nanowire array grown on p-GaN thin film,” Adv. Mater. 21(27), 2767–2770 (2009). [CrossRef]
20. C. Y. Zhu, L. F. Feng, C. D. Wang, H. X. Cong, G. Y. Zhang, Z. J. Yang, and Z. Z. Chen, “Negative capacitance in light-emitting devices,” Solid-State Electron. 53(3), 324–328 (2009). [CrossRef]
21. M. Meneghini, L. R. Trevisanello, G. Meneghesso, and E. Zanoni, “A review on the reliability of GaN-based LEDs,” IEEE Trans. Device Mater. Reliab. 8(2), 323–331 (2008). [CrossRef]
22. K. S. Ramaiah, Y. K. Su, S. J. Chang, and C. H. Chen, “A comparative study of blue, green and yellow light emitting diode structures grown by metal organic chemical vapor deposition,” Solid-State Electron. 50(2), 119–124 (2006). [CrossRef]
23. A. F. Phillips, S. J. Sweeney, A. R. Adams, and P. J. A. Thijs, “The temperature dependence of 1.3- and 1.5- m compressively strained InGaAs(P) MQW semiconductor lasers,” IEEE J. Sel. Top. Quantum Electron. 5(3), 401–412 (1999). [CrossRef]
24. C. H. Yen, Y. J. Liu, K. K. Yu, P. L. Lin, T. P. Chen, L. Y. Chen, T. H. Tsai, and W. C. Liu, “On an AlGaInP-based light-emitting diode with an ITO direct Ohmic contact structure,” IEEE Electron Device Lett. 30(4), 359–361 (2009). [CrossRef]
25. A. F. Jarjour, R. A. Oliver, A. Tahraoui, M. J. Kappers, R. A. Taylor, and C. J. Humphreys, “Experimental and theoretical study of the quantum-confined Stark effect in a single InGaN/GaN quantum dot under applied vertical electric field,” Superlattices Microstruct. 43(5-6), 431–435 (2008). [CrossRef]
