1. Introduction

Ultraviolet (UV) light-emitting diodes (LEDs) have recently attracted much attention due to their potential applications in sterilization, disinfection, water/air purification and excitation sources for portable optical analysis [1–3]. However, UV LEDs present problems such as low internal quantum efficiency, the high electrical activation energy for Mg-doped p-GaN and the low transmittance of transparent conductive electrodes at short wavelengths. Indium tin oxide (ITO) has been generally used for transparent conductive electrodes in LEDs and solar cells. However, ITO layers exhibit some problems because of the low chemical / mechanical stability, limited resources of indium and low transmittance in the UV region [4]. Therefore, an alternative to ITO-based transparent conductive electrodes in UV LEDs is required.

Graphene, which is a two-dimensional monolayer of sp²-bonded hexagonal carbon atoms, has been considered a replacement to ITO due to its superior properties such as high electrical and thermal conductivity, good chemical and mechanical stability, excellent mobility and high transmittance from the UV to near-infrared spectral regions [5, 6]. Graphene transparent conductive electrodes for GaN-based UV LEDs have been previously investigated [7, 8]. Large-area graphene grown by the chemical vapor deposition (CVD) method demonstrated effective current spreading and very high transparency in the UV region as transparent conductive electrodes. However, the high sheet resistance (>1000 Ω/sq for a single layer, 800~900 Ω/sq for bi-layer, and 450~600 Ω/sq for tri-layer un-doped graphenes) and high forward turn-on voltage of un-doped graphene transparent conductive electrodes still limits the implementation in UV LEDs [9–12]. Therefore, significant effort has been focused on reducing the high sheet resistance by increasing the electrical conductivity and lowering the high forward turn-on voltage by increasing the work-function of graphene [9–11]. The sheet resistance of the graphene layer can be usually reduced by adding various chemical p-type dopants such as nitromethane, HNO₃, AuCl₃, H₂SO₄ and HCl [13–16]. Bae et al. reported the effects of various chemical dopants on graphene electrodes, where they showed an AuCl₃/nitromethane solution was one of the most effective methods to decrease the sheet resistance of the graphene layer [16]. Also, the high turn-on voltage in graphene transparent conductive electrodes can be improved by p-type doping of graphene. Choe et al. investigated the work function of AuCl₃-doped graphene layers [11]. They reported that the work function of p-doped graphene layers exposed to 20 mM AuCl₃ solutions increased from 4.42 eV to 5.12 eV due to the electron transfer from the graphene layer to the Au nanoparticles.

In this work, we report the integration of large-area doped-graphene transparent conductive electrodes with UV LEDs. Optical and electrical characterizations, including transmittance, electroluminescence (EL), current-voltage (I-V) characteristics, sheet resistance and micro-Raman spectroscopy, were performed to study the effects of AuCl₃ solutions on graphene transparent conductive electrode properties and performance in UV LEDs.

2. Experiment details

Metal-organic chemical vapor deposition was used to grow the UV LED structure on sapphire substrates, where the growth was initiated with a 25 nm AlN layer grown at 680 °C and 50 Torr pressure. Then, the temperature and pressure were raised to 1025 °C and 150 Torr, respectively, to deposit 2 μm GaN:Si, a AlGaN/GaN/AlGaN 8 nm/5 nm/8 nm single quantum well (SQW), and a 200 nm GaN:Mg layer. A large-area graphene monolayer was grown on Cu foil (300 µm thick, Alfa Aesar) by a CVD technique utilizing methane and hydrogen gases at 1000°C for 30 minutes. The graphene / Cu foil was spin-coated with Poly(methyl methacrylate) (PMMA), followed by etching the Cu foil in a 1wt.% (NH₄)₂S₂O₈ solution for six hours. After the graphene layer was transferred onto UV LED / sapphire substrate, the PMMA layer was removed by an acetone solution. The transfer process of a monolayer graphene was repeated three times to achieve tri-layer graphene. After all the transfer processes of the graphene layers were completed, the mesa pattern was defined by dry-etching with an inductively coupled plasma etcher using BCl₃ and Cl₂ etching gases. The n-type Ti/Al/Ni/Au (20/40/30/80 nm) Ohmic metallizations were deposited by e-beam evaporation method and then heated by rapid thermal annealing at 750°C for 30 seconds to form an Ohmic contact to the n-GaN layer. Finally, the p- and n-contact pads (Ti / Au (40 / 100 nm)) were deposited by e-beam evaporation method. An AuCl₃ solution was synthesized by dissolving AuCl₃ powder (Aldrich) in a nitromethane solvent (Sigma-Aldrich), followed by filtering AuCl₃ solution with polytetrafluoroethylene (PTFE) filter (Advantec). The fully fabricated UV LED chips were dip-coated in 5, 10 and 20 mM concentrations of AuCl₃ solutions for 30 seconds, followed by dipping in deionized water for 5 seconds and drying on a hot plate at 100°C for 1 minute.

The effects of doping the tri-layer graphene were optically characterized by micro-Raman spectroscopy, transmittance measurements, and scanning electron microscopy (SEM, Hitachi, S-4700). The micro-Raman spectra (LabRam HR, Horiba JY) were obtained in a backscattering geometry using a 514 nm Ar-ion laser with approximately 0.5 mW power. The transmittance spectra before/after p-type doping of graphene layers on quartz substrates were measured by a spectrophotometer (Cary 5000, Varian) from 300 nm to 800 nm. Electrical characterizations were performed by four-point probe measurements, I-V characteristics and electroluminescence (EL) measurements. Sheet resistances of graphene layers before/after p-type doping were obtained with four-point probe measurement equipment (CMT-SR1000, Changmin Tech Co.). The changes in the electrical properties of UV LEDs with graphene transparent conductive electrodes before/after various concentrations of p-type dopings were monitored by an Agilent 4155C semiconductor parameter analyzer connected to the probe station.

3. Results and discussion

Figure 1(a) is a schematic of a UV LED device with graphene transparent conductive electrodes p-doped by AuCl₃ solutions. The mesa area of the UV LED was 200 µm × 500 µm (inset of Fig. 1(b)). Figure 1(b) is an EL spectrum from the fabricated UV LED device with AuCl₃-doped graphene transparent conductive electrodes at an injection current of 10 mA, where the concentration of AuCl₃ was 10 mM. UV emission at a wavelength of 363 nm was observed, as shown in Fig. 1(b). Figures 2(a)-2(c) are SEM images of tri-layer graphene after doping with various concentrations of 5, 10 and 20 mM AuCl₃ solutions, respectively. Figures 2(d)-2(f) are high resolution SEM images after p-type AuCl₃ doping using 5, 10 and 20 mM, respectively. In our experiments, the diameter of the Au nanoparticles varied from 10 nm to a few micrometers. Since large Au nanoparticles can greatly reduce the transmittance, we used a 0.2 µm PTFE filter to screen out Au nanoparticles larger than a diameter of 200 nm in the AuCl₃ solutions. Consequently, the diameters of Au nanoparticles ranged from 10 nm to 100 nm, as shown in Figs. 2(a)-2(f). The packing density of Au nanoparticles gradually rose by increasing the concentration of AuCl₃ solutions from 5 mM to 20 mM, as shown in Figs. 2(a)-2(f). After p-type doping using AuCl₃ solutions, Au nanoparticles were well adhered to the graphene layer.

 

Fig. 1 (a) Schematic image of GaN-based UV LED with AuCl₃-doped graphene transparent conductive electrode (b) EL spectrum from UV LED with AuCl₃-doped graphene transparent conductive electrode. Inset is EL image at an injection current of 10 mA.

Download Full Size | PDF

 

Fig. 2 SEM images of tri-layer graphene after p-type doping using (a) 5 mM, (b) 10 mM and (c) 20 mM AuCl₃ solutions. High resolution SEM images after p-type doping using (d) 5 mM, (e) 10 mM and (f) 20 mM AuCl₃ solutions.

Download Full Size | PDF

The doping effects on a graphene layer using an AuCl₃ solution were investigated by various methods, as shown in Figs. 3(a)-3(d). The transmittance and the sheet resistance of a graphene layer were decreased by increasing the concentration of the AuCl₃ solutions, as shown in Figs. 3(a) and 3(b). Figure 3(a) is the transmittance spectra of tri-layer graphene before and after p-type doping using 5, 10 and 20 mM AuCl₃ solutions. At a wavelength of 363 nm, the transmittance of tri-layer graphene before and after p-type doping using 5, 10 and 20 mM AuCl₃ solutions was 87.9, 86.4, 85.1 and 82.9%, respectively. As shown in Fig. 3(a), the transmittance after p-type doping using a 20 mM AuCl₃ solution was significantly decreased by ~5%. However, this is still substantially higher than has been reported for ITO (150 nm thick) layer at a wavelength of 363 nm, approximately 61% [8]. Figure 3(b) is the sheet resistance of a tri-layer graphene before and after p-type doping using 5, 10 and 20 mM AuCl₃ solutions. Y-axis in right (red color) and y-axis in left (black color) represent the sheet resistance and the change of sheet resistance, respectively. The sheet resistance of the tri-layer graphene before and after p-type doping using 5, 10 and 20 mM AuCl₃ solutions was 466.1, 175.5, 158.5 and 112.4 Ω/sq, respectively. As shown in Fig. 3(b), the sheet resistance of the graphene layer after p-type doping using 5, 10 and 20 mM AuCl₃ solutions was decreased by ~62.3, ~66.0, and ~75.9% compared to un-doped pristine tri-layer graphene. The reported sheet resistances of the graphene layers were calculated as the average data of five different spots.

 

Fig. 3 (a) Transmittance spectra of tri-layer graphene on quartz substrate before and after p-type doping (b) Sheet resistance and the change in sheet resistance before and after p-type doping using 5, 10 and 20 mM AuCl₃ solutions. (c, d) Raman spectra before and after p-type doping using various concentrations of AuCl₃ solutions.

Download Full Size | PDF

The doping effects induced by charge transfer between Au nanoparticles and graphene layers were also characterized by micro-Raman spectroscopy, as shown in Figs. 3(c) and 3(d), which show the Raman spectra of the graphene layers before and after p-type doping using AuCl₃ solutions. Figure 3(d) focuses on the G peak in the Raman spectra to investigate the doping effects on the graphene layers. The two prominent peaks in the Raman spectra of high quality graphene layers were the G peak (~1585 cm⁻¹) and the 2D peak (~2700 cm⁻¹), as shown in Fig. 3(c). The charge transfer on graphene layers can be confirmed by the G-peak shift in the Raman spectra [17]. As shown in Figs. 3(c) and 3(d), the G peaks were gradually upshifted by increasing the concentration of the AuCl₃ solutions, and finally upshifted by ~11 cm⁻¹ after p-type doping using a 20 mM AuCl₃ solution due to an increase in the hole charge carrier density, which could be attributed to the reduction potential difference between the graphene and ionic AuCl₄⁻ [14].

Figures 4(a)-(d) are the EL images of UV LED chips with graphene transparent conductive electrodes before and after p-type doping using 5, 10 and 20 mM AuCl₃ solutions, respectively, at a forward voltage of 6.5V. Figures 4(e)-(g) are I-V characteristics before and after p-type doping using 5, 10 and 20 mM AuCl₃ solutions. As can be seen by comparison of Figs. 4(a) through 4(d), the brightness of UV emission, which was emitted over the whole p-GaN region, was dramatically increased by p-type doping of the graphene layers. The currents at an injection voltage of 8 V were gradually increased by increasing the concentration of the AuCl₃ solution, as shown in Figs. 4(e) through 4(g). The current level at an injection voltage of 8 V after p-type doping using 5, 10 and 20 mM AuCl₃ solutions was increased by ~48, ~63 and ~73%, respectively. The large difference of work function between p-GaN (~7.5 eV) and graphene (4.2~4.4 eV) causes the low electrical contact properties and high turn-on voltage [11, 18]. However, the work function of graphene layer after AuCl₃ doping can be increased up to 5.1 eV by the electron transfer from the graphene layer to Au nanoparticles [9, 11]. Therefore, as shown in Figs. 4(e)-4(g), the I-V characteristics of UV LEDs were significantly improved by both the decreased sheet resistance and increased work function of Au-doped graphene layer. However, the UV emission in Fig. 4(c) was slightly brighter than that in Fig. 4(d) because there was a trade-off between the transmittance and sheet resistance in the case of p-type doping using AuCl₃ solution, as shown in Figs. 3(a) and 3(b).

 

Fig. 4 EL images of GaN-based UV LEDs with graphene transparent conductive electrodes (a) before and after p-type doping using (b) 5 mM, (c) 10 mM and (d) 20 mM AuCl₃ solutions at a forward voltage of 6.5 V. I-V characteristics of UV LEDs with graphene transparent conductive electrodes before and after p-type doping using (e) 5 mM, (f) 10 mM and (g) 20 mM AuCl₃ solutions.

Download Full Size | PDF

4. Conclusion

Large-area p-doped graphene transparent conductive electrodes were successfully integrated in GaN-based UV LEDs with an emission peak of 363 nm. A graphene layer was p-doped by 5, 10 and 20 mM AuCl₃ solutions, then characterized by SEM, transmittance measurements, micro-Raman spectroscopies, I-Vs, ELs and sheet resistance measurements. Sheet resistance was significantly decreased by AuCl₃ doping, and transmittance was slightly decreased by AuCl₃ doping. I-V characteristics and EL intensities were dramatically improved by AuCl₃ doping. In our experiments, p-type doping of the graphene by a 10 mM AuCl₃ solution was optimal due to the trade-off between the transmittance and the sheet resistance of the graphene layers. Doped graphene transparent conductive electrodes can be a good alternative to ITO-based transparent conductive electrodes in UV LEDs because of their high transmittance in the UV region and lowered sheet resistance via p-type dopings.