Improved electrode performance in gallium nitride LEDs
Light-emitting diodes based on III-nitride semiconductors have made their way into applications such as backlighting for displays, and solid-state white lighting. High-power LEDs with efficacies larger than conventional fluorescent lamps are now commercially available, bringing the world closer toward the realization of energy-efficient solid-state lighting.1, 2 However, despite significant breakthroughs in light extraction and internal quantum efficiencies, the formation of low resistance, transparent ohmic contact to the highly resistive p-GaN (gallium nitride) layer continues to be a challenge.3 Modern commercial LEDs employ either translucent nickel-gold or indium tin oxide (ITO) as a transparent conductive electrode. Nevertheless, the low transparency of a nickel-gold stack—as well as increasing concerns over the soaring prices and future demand for indium—have initiated a search for an alternative material.4 In addition, next-generation LEDs require transparent conductive electrodes to be flexible (unlike ITO, which is brittle), cheap, and compatible with large-scale manufacturing methods.
Graphene, the most recently discovered form of carbon, offers exceptional characteristics such as high transparency, mechanical flexibility, and superior thermal and electrical conductivities.5, 6 Despite its comparatively high sheet resistance (Rs), recent work has shown that graphene can be applied to the p-GaN layer in GaN-based LEDs as a transparent conductive electrode in place of ITO.7 However, because of the large difference in work function (Φ), integrating pristine graphene directly with p-GaN forms a Schottky barrier height (SBH, qϕB) at the interface. A large SBH is undesirable because it can create low contact resistance, which leads to strong current crowding and a high operating voltage.
To improve the interface properties, we sought to enhance the graphene conductivity as well as decrease both qϕB and the specific contact resistance (ρc) of the graphene-p-GaN junction. We therefore investigated the effects of chemical charge transfer doping of multilayer graphene (MLG) and the use of an ultrathin gold interlayer.8
We doped graphene, grown by chemical vapor deposition on nickel, with an aqueous gold (III) chloride (AuCl3) solution of different concentrations (1–30mM). Hall measurements of the transferred MLG films before and after doping indicated a significant reduction in Rs from 1150Ω/K for the pristine MLG down to 476Ω/K for the 20mM AuCl3 concentration. Moreover, the doping caused an increase in Φ from 4.24eV for the pristine MLG to 4.93eV for the one doped at an AuCl3 concentration of 20mM (see Figure 1). Using a circular transfer line model, we evaluated the ρc of the MLG/p-GaN contacts from current-voltage (I-V) measurements. The ρc of the contacts made with pristine and doped MLG were 1.3 and 0.4Ω-cm2, respectively. By introducing a thin gold layer between the MLG and p-GaN followed by rapid thermal annealing, the ρc was further reduced to 0.24Ω-cm2.9
We fabricated InGaN/GaN multi-quantum-well LEDs with both ITO electrodes and MLG (pristine and doped graphene both with and without a gold interlayer). Figure 2 compares the I-V characteristics of the devices. The forward voltage (Vf, defined at an injection current of 20mA) for the ITO and pristine MLG LEDs was 4 and 6.59V, respectively. Moreover, the MLG device exhibited a strong current crowding around the p-metal pad (see the optical image in Figure 2). We attribute the comparatively high Vf value observed in the pristine MLG LED to poor lateral conductivity, i.e., the high Rs of pristine MLG7 and the high ρc value caused by a large difference in Φ between graphene and p-GaN.
When we applied a doped MLG to the LED, the Vf dropped to 5.55V and the current spreading significantly improved. These improvements are ascribed to the decrease of ρc as a result of the lower work function difference as well as the reduced Rs. The forward voltage was further reduced to 3.96V (slightly lower than the value obtained with the ITO electrode) when we inserted a thin gold layer between the p-GaN and doped MLG electrode. This could be the result of inhomogeneous SBHs at the MLG/p-GaN interface caused by the nanomorphology of the gold interlayer.9 We also studied the injection current–dependent electroluminescence (see Figure 3), which reveals a successful implementation of LEDs with a modified graphene electrode.
In conclusion, the doping of graphene and the use of a thin metal interlayer improved the graphene/p-GaN interface characteristics in a GaN-based LED. By combining these two approaches, the current spreading and device forward voltage can be considerably improved and made comparable to ITO. Our future work will explore the graphene/GaN interface properties by varying the number of layers and adopting different doping means to develop high-power LEDs with a graphene electrode.
This project was supported by the Priority Research Centers Program through the National Research Foundation of Korea funded by the Ministry of Education, Science, and Technology (2010-0029706), and by the IT R&D program of MKE/KEIT (KI002163, Development of Core Technology for High Efficiency Light Emitting Diode based on New Concepts).
Chang-Hee Hong obtained his PhD in electrical engineering from the Korea Advanced Institute of Science and Technology. He is presently director of the LED Fusion Technology Center at Chonbuk National University. His research is focused on growth of strain-relaxed GaN, development of high-efficiency LEDs, and LED-based plant technology.
S. Chandramohan joined the Semiconductor Physics Research Center in 2009. His research interests include graphene-based materials and fabrication of plasmonic and hybrid LEDs based on III-nitride semiconductors. He received his PhD in physics from the Bharathiar University, India.
