Abstract
We successfully demonstrated that the Arplasmatreatment p-GaN surface increased the contact resistance of ITO/P-GaN serving as injection current deflection layer under the electrode pad. It was found that the values of the two LEDs at 20 mA were approximately 3.3 V. Under a 20 mA current injection, it was found that output powers of conventional LED and Ar-plasma-treatment LED on p-GaN surfaces were 9.8 and 11.08 mW, respectively. We can increase the output power of GaN LEDs in 13% due to current blocking on the surface of p-GaN under the electrode pad by inserting the treatment with Ar plasma. It was also found that, after the reliability test for 72 hours the half lifetimes of conventional LEDs and LEDs with Ar-plasma treatment on p-GaN surface were about 49% and 55%, corresponding to the initial intensity, respectively.
1. Introduction
Nitride-based materials have recently emerged as important semiconductor materials, leading to the realization of high performance light emitters from ultraviolet (UV) to blue and green spectral regions [1, 2]. For example, GaN-based blue and green light-emitting diodes (LEDs) have already been extensively used in full-color displays and as highly efficient light sources for traffic-light lamps. Highefficiency nitride-based LEDs are also potentially useful for solid state lighting. However, the external quantum efficiency (EQE) of LEDs, which is defined as the product of internal quantum efficiency and light-extraction efficiency, is still required to be further improved. To improve the EQE of LEDs, the internal quantum efficiency and light-extraction efficiency must be increased. To approach solid state lighting, however, one needs to further improve the output efficiency of these LEDs. There are several parameters that affect the output efficiency of the LEDs, such as light extraction efficiency, internal quantum efficiency, and current distribution of LEDs. It has been demonstrated that several methods can be used to improve output efficiency of GaN-based LEDs by enhancing the light-extraction efficiency, such as textured surfaces [3–5], a highly transparent p-contact layer [6], a proper substrate design [7], and flip-chip packaging [8]. Moreover, the crystal quality, device band engineering structure design, and doping profile would affect the internal quantum efficiency of LEDs. Besides these factors, current distributed throughout the LEDs also plays one of the key roles in the efficiency of the LEDs as well. In other words, current distributed throughout the LEDs determines the real light-emitting area of the devices, and light output power is proportional to the light-emitting area. Recently, there are several approaches for improving the current spreading and the light-extraction efficiency of GaN-based LEDs by current blocking layer (CBL) [9, 10]. Current distribution had been wildly studied in the GaAs-based LEDs [11–13] to increase output power. Current blocking is one of the ways to change current distribution in the LEDs. The current injected by the top contact enters the active region predominantly under the top contact. The extraction of the light generated in the active region is thus strongly hindered by the opaque metal contact. The current block layer would block the entering current from top contact to the active region. The light-extraction efficiency of LEDs is improved because the current is deflected away from the top contact. In this study, we would apply the different methods of current-blocking layer formation on the InGaN/GaN multiple quantum wells (MQWs) LEDs. We would also compare the electrical, optical and reliability characteristics of those different current blocking layers.
2. Experimental
Samples used in this study were all grown on 2-inch (0001) sapphire substrate by vertical metal organic chemical vapor deposition (MOCVD). The detailed layers and growth procedures have been described in a previous publication [13]. The thicknesses of the p-AlGaN and p-GaN layers were 50 nm and 150 nm, respectively. After growing the samples, postannealing was needed to activate the top p-GaN layer of InGaN/GaN MQWs LED at 700°C and 20 min. The partial p-GaN and active layers were removed to expose the n-GaN layer for the n-electrode ohmic contact by using the Inductive Coupled Plasma (ICP) etcher. The current blocking layer was formed right before the deposition of the indiumtinoxide (ITO) transparent contact layer (TCL). An Ar-plasma treatment on surface of p-GaN underneath the p-electrode pad was introduced to form the current-blocking layer in this study with the following conditions: Ar: 20 sccm, working pressure: 3 mtorr, and ICP power: 120 W. Then, the ITO TCL was deposited on the p-GaN surface to cover the current blocking layer. The Cr/Au (50/200 nm) bilayer metal contact was deposited on the ITO TCL and exposed n+-GaN layers to form the p-type electrode (anode) and the n-type electrode (cathode), respectively, at the same time. The chip size of the LEDs in our study was mil2. The conventional LED is LED I and the ICP plasma treatment LED is LED II in the paper. The current-voltage (-) characteristics of the current blocking layer and experimental LEDs were measured by using the HP-4156C semiconductor parameter analyzer, and the output powers of the LEDs were measured by using calibrated integrating sphere.
3. Results and Discussion
Figure 1 indicated the - characteristics of the ITO/p-GaN contact with and without Ar plasma treatment on p-GaN surface. It was found that the resistance of ITO/p-GaN contact increased after the Ar plasma treatment of the p-GaN surface. The ITO/p-GaN contact with Ar plasma treatment on p-GaN surface was shown as a Schottky-like contact and the ITO/p-GaN contact without the plasma treatment was exhibited as an ohmic-like contact. - characteristics of the two curves in Figure 1 do not show Schottky characteristic or perfect ohmic characteristic. The resistances of the contact without and after Ar plasma are 3.1 10−3 Ω cm2 and 5.0 10−1 Ω cm2. It was found that the resistance of the contact after Ar plasma is larger than that of the conventional contact. The increase of the resistance of the ITO/p-GaN contact on Ar plasma treatment p-GaN surface should be attributed to the Ar plasma damage resulting in a decrease the carrier concentration on the p-GaN surface. It should be noted that high contact resistance of the Ar plasma-damaged p-GaN surface could be served as the current deflection resistance right under the electrode pads.
Figure 2 indicated the current-voltage (-) curve of LEDs with and without (i.e., ICP plasma treatment LED: LED II and conventional LED: LED I, resp.) Ar plasma treatment under the electrode pad. As shown in Figure 2, it was found that the forward voltage, () of conventional LED I and Ar-plasma-treatment LED II at 20 mA were approximately 3.31 and 3.39 V, responsively. The lightly increase in 20 mA- results from the increase in current density for Ar-plasma-treatment LED-II. The 20 mA- results indicated that the electrical characteristics of LEDs were not obviously degraded by the Ar plasma treatment under the electrode pad processes.
Figure 3 indicated the intensity-voltage (-) curves of LEDs with and without Ar plasma treatment under the electrode pad. Furthermore, it was found that output powers of these two LEDs were increased as we increased the injection current. Under the same injection current, it was found that the output power observed from ICP-plasma treatment LED II was always larger than the conventional LED I. Figure 3 showed that the output powers were 9.8 and 11.08 mW for conventional LED I and ICP-plasma treatment LED II under 20 mA current injections, respectively. In other words, we can increase the LED output power by 13% by inserting the Ar plasma-damaged p-GaN surface under the electrode pad. The enhancement of output power observed from ICP-plasma treatment LED II should be attributed to the improved current spreading of LEDs by high contact resistance of the ITO/p-GaN under electrode pad. To further investigate the effects of current distribution, we measured near field optical images of the LEDs.
Figure 4 shows the EL spectra of a conventional LED I and ICP-plasma-treatment LED II at an injection current of 20 mA. No significant differences in the EL peak position at 446 nm were detected throughout the three types of LEDs. However, the EL intensity of LED II was higher than that of LED I.
Figure 5 shows the 3D images measured from the LED I and LED II at 20 mA DC injection current into the devices. It can be seen clearly that the output light was emitted uniformly across the chips since current spread uniformly in LED II. In contrast, it was found that light output distribution is nonuniform for the conventional LED I. The better current spreading of LED II should be attributed to the high ITO/p-GaN contact resistance under the electrode pad deflecting the injection current away from pad and then increasing the emission area of devices. In other words, the LED I without high resistive Ar plasma-damaged p-GaN surface under electrode pad would lead the most of injection current to go into devices under the electrode pad. However, the light emitted under the pad would be absorbed by the opaque metal pad and then reduce the output power of LED. At an injection current of 20 mA, the light output power of LEDs with Ar plasma treatment was 13% larger than that of conventional LEDs. At an injection current of 100 mA, the temperature of the p-pad metal on LEDs with Ar plasma treatment is 13°C lower than that of the LEDs with a Ar-plasma CBL. However, the output power of LEDs with Ar plasma treatment is the lower due to the surface damage of p-GaN under the p-pad electrode. Besides, the non-uniform current distribution near the electrode pad might enhance the current crowding while driving at high current. Therefore, it was found that the injection current to reache the highest output power of conventional LED I is less than that of ICP-plasma treatment LED II in Figure 2.
(a) With Ar plasma treatment
(b) Without Ar plasma treatment
Figure 6 shows the reliability test of the LED I and II injection with 50 mA at 80°C. It was found that the decayed intensities of LED I and LED II were about 49% and 55% of the initial intensity, respectively, after aging 72 hours. It should also be attributed to the better current spreading of the LED II with high ITO/p-GaN contact resistance under the electrode pad. The better current spreading could reduce the current crowding under the pad at high current injection. As a result, the intensity degradation of the LED II during the high temperature and high injection current aging is less than that of LED I. The decrease in optical power was closely correlated to an increase in operating voltage due to an increase in the parasitic series resistance during the accelerated current aging test.
4. Conclusion
In summary, we successfully demonstrated that the Ar plasma-damaged p-GaN surface increased the resistance of ITO/P-GaN contact serving as injection current deflection layer under the electrode pad. It was found that the of both LEDs at 20 mA were approximately 3.3 V. It was found that output powers were 9.8 and 11.08 mW for LED I and LED II at 20 mA, respectively. In other words, we can increase the LED output power by 13% by inserting the Ar plasma-damaged p-GaN surface under the electrode pad. It was also found that the decayed intensities of LED I and LED II were about 49% and 55% of the initial intensity, respectively, after aging 72 hours. It was believed that high resistance ITO/P-GaN contact with Ar plasma-damaged p-GaN surface under pad would improve the injection current spreading to the whole device. And better current spreading could lead to improving the photoelectrical characteristics of LEDs.
Acknowledgments
This work was granted in part by the Center for Frontier Materials and Micro/Nano Science and Technology and in part by the Advanced Optoelectronic Technology Center, National Cheng Kung University, under projects from the Ministry of Education, Taiwan. This work was also supported in part by the Ministry of Economic Affairs (MOEA) NSC 98-2622-E-024-001-CC3 and NSC 99-2221-E-024-009. The authors also would like to thank the Bureau of Energy, Ministry of Economic Affairs of Taiwan, for financially supporting this research under Contract no. 98-D0204-6, and the LED Lighting and Research Center, NCKU, for the assistance in related measurements.