The various surface texturing effects of InGaN light emitting diodes (LEDs) have been investigated by comparison of experimented data and simulated data. The single-layer and double-layer texturing were performed with the help of ITO nanospheres using wet etching, where the ITO ohmic contact layer and the p-GaN layer are textured using ITO nanospheres as an etch mask. In case of single-layer texturing, p-type GaN layer texturing was more effective than ITO ohmic contact layer texturing. The maximum enhancement of wall-plug efficiency of double-layered textured LEDs is 40% more than conventional LEDs, after packaging at an injected current of 20 mA. The increase of light scattering at the textured GaN surfaces is a major reason for increasing the light extraction efficiency of LEDs.
© 2011 OSA
High-efficiency GaN-based LEDs are widely used in many applications such as full color displays, solid-state lighting, and backlighting in liquid crystal displays. The fast-growing LED market has thus captured the attention of leading semiconductor firms. However, the extraction efficiency of GaN-based LEDs continues to remain low because of total reflection that occurs at the LED surface because of the large difference in refractive indices between the semiconductor and air [1,2]. Considering the disparate refractive indices of GaN (n = 2.5) and air (n = 1), the internal light has difficulty escaping into the air from the semiconductor, and the critical angle for the total reflection is as small as 24° . Consequently, several studies have focused on improving the light extraction efficiency by decreasing total reflection. In addition, facilitating light scattering from the device by surface texturing and patterning techniques is also an important parameter to improve the light extraction efficiency at the interface between GaN and air. For example, a GaN and an indium-tin-oxide (ITO) contact layer were textured through photolithography , holographic lithography [5,6], and colloidal lithography . Recently, Kim et al. reported light extraction for nanoscale surface textures using colloidal lithography . It has hole-patterned p-GaN and pillar-patterned ITO contact layers by the polystyrene sphere as an etching mask for the surface texturing, respectively. According to Kim et al., the light output power of the LED with the textured p-GaN layer is higher than that of the LED with the textured ITO contact layer. However, it is very difficult to discern why p-GaN layer texturing exhibits a more effective light extraction effect than ITO contact layer texturing. And also, those researches are used photo-resist (PR) or polymer as an etch mask. In those cases, an etch mask must be removed after texturing processes because PR and polymer would interrupt current spreading between transparent ohmic contact and GaN surface.
In this work, we have fabricated LED devices with single- and double-layered textured LEDs using ITO nanospheres. The various surface texturing effect of InGaN light emitting diodes (LEDs) have been investigated by comparison of experimented data and simulated data. For double-layered texturing, surface texturing has been created on both the ITO contact layer and the p-type GaN layer using ITO nanospheres and dry etching, respectively. Here, we have chosen ITO nanospheres as the surface texturing material, because it can act as a surface texturing layer when dispersed on the uniform surface and can also be used as an etch mask to create surface texturing on the p-GaN layer via dry etching. The significance of this methodology is that the use of same material (ITO) as a transparent ohmic contact layer and mask would not cause any damage to the interface, especially in the electrical and optical properties. In addition, we compare the experimental results with ray tracing simulation data to study the enhancement trend of the light extraction efficiency for each textured layer.
Samples used in this study are grown on sapphire substrates using the process of metal organic chemical vapor deposition (MOCVD). The GaN nucleation layer is deposited on sapphire at a temperature of 560°C, followed by a 2 μm thick layer of undoped-GaN epilayers at 1,100°C, and a 2 μm thick layer of Si-doped n-type GaN at 1,120°C. After completing the growth of the n-type GaN template, seven-period multi-quantum wells (MQWs) of InGaN/GaN pairs, which emit at approximately 460 nm, are subsequently grown at 810°C. Next, a 230 nm thick p-GaN layer is grown at 1,050°C, and finally, a 10 nm thick highly doped p-GaN layer is grown at 1,050°C.
After the growth, mesa etching of the exposed n-type GaN is carried out using inductively coupled plasma, and a 200 nm-thick ITO layer is evaporated onto the p-type GaN surface using an electron-beam evaporator and by the photolithography process. A conventional LED (Sample A), an ITO contact layer textured LED (Sample B), a p-type GaN layer textured LED (Sample C), and a double-layered textured LED (Sample D) were fabricated using the same process steps except for the texturing step. For the texturing process, as-deposited 200-nm-thick ITO films were dipped inside diluted HCl (5%) for about 30 s to induce ITO nanosphere formation with a diameter of 200 nm . The driving forces that facilitate the formation of the ITO nanospheres are a weak binding energy at the grain boundary of the as-deposited ITO films and Ostwald ripening. This etching process is further initiated by the fact that the as-deposited ITO layer is polycrystalline. The schematic fabrication processes of each sample were described in detail, as shown in Fig. 1 .
Figure 1(a) shows the structure of a conventional LED device (Sample A) with a flat ITO surface, which is used for comparison. Figure 1(b) shows the fabrication process of Sample B. In this case, a 200-nm-thick ITO film is deposited on ITO nanospheres 200 nm in diameter to form a hemispherical textured ITO ohmic contact layer . In order to generate p-type GaN surface texturing, ICP treatment was used for about 15 s as in Sample C, with the help of ITO nanospheres, as an etch mask. Following this, residual ITO nanospheres were removed by HCl for about 5 min, as shown in Fig. 1(c). Then, the 200-nm-thick ITO film was deposited on the textured p-type GaN surface to form the ITO ohmic contact layer. Finally, in Sample D, for double-layered texturing, a combination of the above-prescribed two different techniques was incorporated, as shown in Fig. 1(d). The major significance of this process is the adoption of a simple and realistic process, which does not require any additional process to eliminate the residues of ITO nanospheres after dry etching, because of usage of similar materials as the ohmic contact layer. The hemispherical ITO ohmic contact layer of double-layered texturing was achieved by depositing a 200-nm-thick ITO film on a textured p-type GaN surface and residual ITO nanospheres. The characteristics of ITO nanospheres are similar to the ITO films. When ITO nanospheres were used as an embedded mask for surface texturing, the entire thickness and absorption of ITO layer are increased. But, that is not important issue to enhanced light output because the decrease of internal reflection effect is dominant in textured layer. After the texturing processes, with the aim of completing a typical LED process, the ITO ohmic contact layer was annealed at 600°C for 30 s in air. Then, the Cr/Ni/Au and Ni/Au layers were deposited, as metal contacts, to the n- and p-type layers, respectively. The surface morphologies were examined by scanning electron microscopy (SEM). The current-voltage (I-V) and light output-current (L-I) measurements have been carried out using a probe station system and the beam profile of the LED characterized by an OL 770 multichannel spectroradiometer. Moreover, simulated data were generated using LightToolsTM for comparison with the experimental data.
3. Results and discussions
Figure 2 demonstrates the top surface views of scanning electron microscope (SEM) images of Samples B, C, and D, and the receptive cross sectional views of the SEM images are placed as insets. As shown in Fig. 2(a), a large number of ITO nanospheres are observed on the surface of the p-type GaN layer after wet etching. Figure 2(b) shows the uniformly distributed nano-textured hemispherical ITO ohmic contact layer surface. These hemispherical shapes are the result of deposition of a 200-nm-thick ITO film on the ITO nanospheres. Figure 2(c) shows the nano-textured p-type GaN surface after ICP treatment, in which ITO nanospheres are used as an etch mask. The shape and size of the texturing is almost identical to the ITO nanospheres, and the etch depth is about 80–90 nm. As illustrated in Fig. 2(d), the top surface of the ITO ohmic contact layer is almost flat because the textured p-type GaN surface was covered by the 200-nm-thick ITO films. Here, there will be no light scattering on the ITO ohmic contact layer in Sample C because of flat ITO film formation. However, the residual ITO nanospheres are observed on the textured p-type GaN layer after dry etching, as shown in Fig. 2(e). The residual ITO nanospheres and the depth of textured p-type GaN were totally covered after the evaporation of the 200-nm-thick ITO film, which resulted in a hemispherical, textured ITO ohmic contact layer.
The texturing density of the textured p-type GaN layer and textured ITO ohmic contact layer were measured as 12 × 10−9 cm2 and 8 × 10−9 cm2, respectively. Additionally, Samples B and D have the same aspect ratio of height over radius of approximately 0.55. The enhancement of light output power of Sample B and D in textured transparent ohmic contact layer is same because these samples have a same aspect ratio of transparent ohmic contact texture. Thus, four different samples help us to investigate the effect of light scattering on a planar ITO ohmic contact layer, textured ITO ohmic contact layer, textured p-type GaN layer, and double textured layer, in Samples A, B, C, and D, respectively.
Figure 3 shows the I-V characteristics of Samples A-D. Following the dry etching process for 15 s, the forward voltages measured at 20 mA are 3.45, 3.45, 3.55, and 3.55 V for the Samples A, B, C and D, respectively. Based on I-V characteristics of each sample, series resistance was calculated using diode equation and MATLAB. For forward bias, the following equation describes the I-V characteristic of the Schottky diode:11]. After the dry etching processes, the ohmic contact area between ITO and heavily-doped p-GaN is decreased . But, it does not adversely affect the LED performance. The insets in Fig. 3 are the images of respective samples captured during light emission at 0.1 mA. Comparison of all the images show that all samples have same behavior of current spreading without any current crowding.
In general, the photons generated at the multi-quantum well (MQW) region can be emitted in all possible directions. For measuring the luminance intensity of an unpackaged chip, it is necessary to consider the front top side (top side through the ITO ohmic contact layer), the backside (bottom side through the sapphire substrate/transparent glass chuck), and the side walls of the device. However, in practice, the available measurement techniques for the unpackaged chip are only from the front and back sides of the LED device. In addition, the luminance intensity measurement at the side wall region is meaningless without the separation of the chip. However, this can be examined only following the packaging step, which will generate a considerable impact of light emission at the side wall using an integrating sphere. Here, the light extraction trend was studied with respect to different types of surface texturing in LED chips.
The respective L-I characteristics of four different LEDs are shown in Fig. 4 . The light output powers of Samples B, C, and D are found to be 23.8%, 45.6%, and 70% higher than Sample A at 20 mA, respectively, at the front side of the LED chip, as shown in Fig. 4(a). Similarly, in Fig. 4(b), the improvement of light output powers of Samples B, C, and D are also found to be 17.9%, 73.5%, and 71.5% higher than Sample A at 20 mA, respectively, from the backside of the LED chip. Therefore, based on these results, the light output powers of the textured samples were increased significantly not only at the front side but also at the backside because of the light scattering tendency at the textured surface. As a result, the light output power was increased for all samples with respect to surface texturing. The major reasons for the improvement at the front side were as follows. At the front side, the textured surface induced the increase of light extraction because of the increased light scattering at the textured surface. Further, at the backside, the light reflection at the bottom side was increased because of the increased light scattering at the textured top surface. This measurement is important because it has the possibility of light extraction toward the front side because of the light reflections at the GaN/sapphire interface and sapphire/metal reflector interface after packaging. However, among Samples B, C, and D, the difference in light output power improvements because of the different types of texturing surfaces at different regions is evident. For example, in comparison with Samples B and C, we can assume that p-type GaN layer texturing is more effective than ITO ohmic contact layer texturing . However, the effect of light extraction of each sample was not only induced by the textured layers but also includes the tendency of light scattering and reflection at the interfaces of GaN/ITO, ITO/air, and GaN/sapphire. This information was further validated with theoretical simulated data; it offers a clear picture on light output power improvement, as shown in Fig. 5 . An optical simulation was carried out for each sample with a different surface textured layer, in order to trace the light path within the device structure, using LightToolsTM. Over 100,000 rays are assumed to originate from the MQW region, and the size of the sample used in the simulation is 315 × 315 μm2. Figures 5(a) and 5(b) illustrate the percentage of improvement at the front side and backside, respectively. There is considerable difference between the simulated and the experimental data, which was expected because of the differences in the structural design and detailed components used in the simulation and the experimental test. The simulated structures are formed based on real structure components such as density of texturing, depth of texture, and thickness of each layers. However, the simulated structure cannot follow the random texture of real structure and detailed surface roughness. Moreover, the simulation considers only the number of incident rays approaching the receiver during ray extraction from inside of LED structure and the rays which are weak in power due to scattering losses are not considered. As a result, we observe a significant difference between simulation and experimental results, as shown in Fig. 5. However, the increasing trend of light output intensity was similar to the experimental results. The rate of enhanced light output power of Sample C is larger than that of Sample B.
In addition, Sample D has a result similar to the sum of Samples B and C, as shown in Fig. 5(a). Contrary to Fig. 5(a), the rate of enhanced light output power of Sample B is larger than that of Sample D at the backside, as shown in Fig. 5(b). It was expected that the amount of light incident to the backside detector was relatively less in Sample D, because the amount of incident light to the front side detector was enhanced by secondary light scattering at the textured ITO ohmic contact layer. Consequently, the amount of reflected light at the ITO/air interface and incident light toward the backside were decreased considerably.
As mentioned above, we need to examine the extent of light scattering and reflection at the interfaces of each layer with respect to different refractive indexes and texturing using simulation. The simulation conditions are the same as mentioned above. As shown in Fig. 6(a) , four different receivers were introduced to detect a number of rays over the entire region (Receiver 1), the front side (Receiver 2), the ITO ohmic contact layer between air and GaN (Receiver 3), and the backside (Receiver 4). The number of detected rays at each receiver is summarized in Table 1 .From the data given in Table 1, the light output of Samples A, B, C, and D were 28.4%, 45.5%, 62.5%, and 76.3%, respectively, when 100,000 rays originated from the MQW according to Receiver 1. As shown in Sample A with flat surfaces, the detected rays at Receivers 2 and 3 were 38% and 31%, respectively. In case of Sample A, as the number of emitted rays at the MQW was assumed to be 100,000, the number of escaped rays from the p-type GaN layer to the ITO ohmic contact layer was 31,050 (31%) and the number of escaped rays from the ITO ohmic contact layer to the air was 11,736 (38%), as shown in Fig. 6(b). This result indicated that the rays that escaped from the p-type GaN layer to the ITO ohmic contact layer dominate over the rays that escaped from the device to air. Figure 6(c) demonstrates the enhancement of emitted rays in Sample B; 65% of rays were extracted from the textured ITO ohmic contact layer to the air. In contrast with Sample B, only 35% of rays were extracted from the flat ITO layer to air in Sample C, as shown in Fig. 6(d). Nevertheless, the number of rays detected at Receiver 2 of Sample C was larger than that of Sample B by 1.4 times. This result was calculated from the difference in the number of rays that escaped from the flat (31%) and textured p-type GaN surface (73%) to the ITO ohmic contact layer. In other words, the texturing effect of Sample B is limited by the number of rays that escaped from the flat p-type GaN layer to the ITO ohmic contact layer. Compared to Sample B, Fig. 6(e) shows that 46% rays were extracted from the textured ITO ohmic contact layer to air in Sample D. A decrease in light extraction efficiency is expected because of the changing direction of rays passing through the textured p-type GaN layer. In any event, the number of detected rays at Receiver 2 of Sample D was increased by 2.9 times, when compared to Sample A. Namely, the light was extracted from MQW to p-GaN layer through textured p-GaN more than that of flat p-GaN, and this light was extracted from p-GaN to air through textured ITO more than that of flat ITO layer. These double-layered texturing effects were occurred at the same time, as a result, the light was extracted more than one-layered texturing effect only.
After packaging the LED chips, the light intensities were measured, as shown in Fig. 7 . According to Fig. 7(a), with an injection current of 20 mA, the wall-plug efficiency of Sample B, C, and D are increased by 15%, 26%, and 40% compared with Sample A. The wall-plug efficiency is calculated by the simple calculation of output power over input power. It is noteworthy to state that the measurement of light output power after packaging is more reliable and accurate than the unpackaged LED chip, because of the considerable impact of sidewall effects of four facets after the separation of the individual chip.
In addition, after packaging, the overall light output power will reduce slightly because of the relatively smaller top surface texturing effect than the light extraction effect at the sidewalls. Furthermore, the beam profiles of the LEDs were measured, and the resulting diagrams are shown in Fig. 7(b). The far-field angles for Samples A, B, C, and D are 160°, 157.5°, 153°, and 153.6°, respectively. The smaller far-field angles of Samples B, C, and D indicate that more rays were extracted at the front side than Sample A because of surface texturing. It was found that the emission intensities of each device with different textured layers were enhanced over all far-field angles than the conventional LED with flat surfaces.
In this study, single-layer textured and double-layer textured LEDs were designed and successfully fabricated. The surface texturing of ITO layer, p-GaN layer and both layer was accomplished with the help of ITO nanospheres using wet etching. We confirmed the extent of light scattering at each textured layer with the experimental results and it was validated by simulation data. In the case of ITO ohmic contact layer texturing, the light extraction efficiency was limited by the number of rays that escaped from the flat p-type GaN layer to the ITO ohmic contact layer. However, p-type GaN layer texturing was more effective than ITO ohmic contact layer texturing in the prescribed structure. As a result, the enhancement of the wall-plug efficiency of Sample B, C, and D are increased by 15%, 26%, and 40% compared with Sample A with the packaged LEDs at an injection current of 20 mA, respectively. The enhancement of light output power can be attributed to the scattering effect at the textured surface. The sample with double-layer texturing at the p-type GaN layer and ITO ohmic contact layer had the opportunity to light scatter twice at each textured layer, which resulted in an increase in light output power.
This work is financially supported by the Ministry of Knowledge Economy (MKE) and Korea Institute for Advancement in Technology (KIAT) through the Workforce Development Program in Strategic Technology and the Priority Research Centers Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2009-0094032).
References and links
1. I. Schnitzer, E. Yablonovitch, C. Caneau, T. J. Gmitter, and A. Scherer, “30% external quantum efficiency from surface textured, thin-film light-emitting diodes,” Appl. Phys. Lett. 63(16), 2174–2176 (1993). [CrossRef]
2. T. Fujii, Y. Gao, R. Sharma, E. L. Hu, S. P. DenBaars, and S. Nakamura, “Increase in the extraction efficiency of GaN-based light-emitting diodes via surface roughening,” Appl. Phys. Lett. 84(6), 855–857 (2004). [CrossRef]
3. T. H. Hsueh, J. K. Sheu, H. W. Huang, J. Y. Chu, C. C. Kao, H. C. Kuo, and S. C. Wang, “Enhancement in light output of InGaN-based microhole array light-emitting diodes,” IEEE Photon. Technol. Lett. 17(6), 1163–1165 (2005). [CrossRef]
4. C. H. Kuo, H. C. Feng, C. W. Kuo, C. M. Chen, L. W. Wu, and G. C. Chi, “Nitride-based near-ultraviolet light emitting diodes with meshed p-GaN,” Appl. Phys. Lett. 90(14), 142115 (2007). [CrossRef]
5. R. H. Horng, Y. L. Tsai, T. M. Wu, D. S. Wuu, and C. H. Chao, “Investigation of Light Extraction of InGaN LEDs With Surface-Textured Indium Tin Oxide by Holographic and Natural Lithography,” IEEE J. Sel. Top. Quantum Electron. 15(5), 1327–1331 (2009). [CrossRef]
6. H. G. Hong, S. S. Kim, D. Y. Kim, T. H. Lee, J. O. Song, J. H. Cho, C. Sone, Y. Park, and T. Y. Seong, “Enhancement of the light output of GaN-based ultraviolet light-emitting diodes by a one-dimensional nanopatterning process,” Appl. Phys. Lett. 88(10), 103505 (2006). [CrossRef]
7. Y. C. Lee, H. C. Kuo, B. S. Cheng, C. E. Lee, C. H. Chiu, T. C. Lu, S. C. Wang, T. F. Liao, and C. S. Chang, “Enhanced Light Extraction in Wafer-Bonded AlGaInP-Based Light-Emitting Diodes via Micro- and Nanoscale Surface Textured,” IEEE Electron Device Lett. 30(10), 1054–1056 (2009). [CrossRef]
8. T. S. Kim, S. M. Kim, Y. H. Jang, and G. Y. Jung, “Increase of light extraction from GaN based light emitting diodes incorporating patterned structure by colloidal lithography,” Appl. Phys. Lett. 91(17), 171114 (2007). [CrossRef]
9. J. H. Kang, H. G. Kim, H. K. Kim, H. Y. Kim, J. H. Ryu, P. Uthirakumar, N. Han, and C. H. Hong, “Improvement of Light Output Power in InGaN/GaN Light-Emitting Diodes with a Nanotextured GaN Surface Using Indium Tin Oxide Nanospheres,” Jpn. J. Appl. Phys. 48(10), 102104 (2009). [CrossRef]
10. J. H. Kang, H. G. Kim, J. H. Ryu, H. K. Kim, H. Y. Kim, J. Joo, M. S. Lee, Y. J. Park, P. Uthirakumar, and C. H. Hong, “Enhancement of Light Output Power in InGaN/GaN LEDs with Nanoroughed Hemispherical Indium Tin Oxide Transparent Ohmic Contacts,” Electrochem. Solid-State Lett. 13(2), D1–D3 (2010). [CrossRef]
11. X. A. Cao, S. J. Pearton, A. P. Zhang, G. T. Dang, F. Ren, R. J. Shul, L. Zhang, R. Hickman, and J. M. Van Hove, “Electrical effects of plasma damage in p-GaN,” Appl. Phys. Lett. 75(17), 2569 (1999). [CrossRef]
12. T. Mori, T. Kozawa, T. Ohwaki, Y. Taga, S. Nagai, S. Yamasaki, S. Asami, N. Shibata, and M. Koike, “Schottky barriers and contact resistances on p-type GaN,” Appl. Phys. Lett. 69(23), 3537 (1996). [CrossRef]