IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 48, NO. 6, JUNE 2001 1065 Effects of Current Spreading on the Performance of GaN-Based Light-Emitting Diodes Hyunsoo Kim, Seong-Ju Park, and Hyunsang Hwang Abstract Effects of current spreading on the performance of multiple quantum wells (MQWs) GaN/InGaN light-emitting diodes (LEDs) were investigated. For the theoretical prediction of device performance, we developed a model using two device-design parameters, which consist of the applied current density and the effective length for the lateral current path. A comparison of the theoretical and experimental results clearly showed that the reliability characteristics and the optical efficiency of device are heavily dependent on the applied current density. In addition, the effective length for the lateral current path was found to have a profound effect on the uniform current spreading. Based on these findings, an ideal geometrical design of the highly efficient LED is proposed. Index Terms Current spreading, GaN, geometrical design, light-emitting diode (LED), model, reliability. I. INTRODUCTION GaN-BASED materials have many attractive electronic, optical, and thermal properties, which make them promising materials for optoelectronic, high-power, and high-temperature electronic devices. Recently, significant progress has been made on our understanding of III V nitride light-emitting diodes (LEDs) and laser diodes (LDs) [1], [2]. Typically, it is necessary to employ a lateral carrier injection type with respect to these diodes due to the insulating sapphire substrate. However, an employment of the lateral injection type can exhibit a problem of nonuniform current spreading during device operation. Recently, Eliashevich et al. [3] reported that the conductivity of an n-type GaN layer has a profound effect on uniform current spreading. It was shown that the nonuniform current spreading could significantly degrade the properties of LED performance due to the current crowding in a localized region of the device. However, the role of current spreading in the GaN-based LEDs has not been clearly understood as yet. In this paper, we report on the effects of current spreading on the electrical, optical, and reliability characteristics of the multiple quantum well s (MQW s) LEDs using the current-voltage, light output power-injected current density, and the lifetime measurements. In order to study this in detail, we propose the device operation model using two important parameters such as the applied current density and the effective length for the lateral current path. It is shown that the LED performances are Manuscript received September 8, 2000. This work was supported in part by the Korea Energy Management Cooperation and the Brain Korea 21 Project. The review of this paper was arranged by Editor P. K. Bhattacharya. H. Kim, S.-J. Park, and H. Hwang are with the Department of Materials Science and Engineering, Kwangju Institute of Science and Technology, Kwangju 500-712, Korea (e-mail: hwanghs@kjist.ac.kr). Publisher Item Identifier S 0018-9383(01)03237-3. Fig. 1. Schematic cross-sectional view of the fabricated LED. Current paths A and B represent two extreme paths from p- to n-pad. sensitively dependent upon these parameters. Finally, based on our experimental and theoretical results,the highly efficient and reliable LED design is suggested. II. MODEL Fig. 1 illustrates a theoretical prediction of an ideal current, which spreads during LED operation using two main current paths for the cross-sectional LED structure. (It should be noted that and quoted in this model indicates the extreme current path from p- to n-pad). The total voltage drop across an arbitrary current path between two pads is given by [3], [4] where represents the current density, and and represent the thickness and the resistivity of the respective layer, respectively. The geometrical parameters ( and ) represent the arbitrary length of the lateral current path through the transparent layer and the n-type layer, respectively. In order to create a simpler model, the resistances of the p- and n-ohmic contacts were ignored. In addition, the component of voltage drops which are related to the vertical aspects of the transparent layer and the n-layer can also be neglected. Furthermore, that of the voltage drop related to the lateral current path through the p-layer can be also ignored. Therefore, the relation of the voltage drops for path can be written as where and represent the extreme length of the lateral current paths through the transparent and the n-type layer, respectively. Similarly, the relation of the voltage drops for path can be written as (1) (2) (3) 0018 9383/01$10.00 2001 IEEE
1066 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 48, NO. 6, JUNE 2001 Considering the relationships between (2) and (3), it is evident that a perfectly uniform current spreading across the active region of the LED device can be achieved when the total voltage drop of path is equal to that of path. That is, the following condition must be satisfied, and its relation can be expressed as From (4), it can be known that uniform current spreading is a function of and. Based on this condition, the uniform current spreading can be achieved when the values of and are identical. Basically, this is a function of a material quality and can be attained via the growth of a highquality n-type layer. In this paper, using the obtained parameter (the applied current density) and (the effective length for the lateral current path) in this model, we will mainly discuss the effects of current spreading on the LED performances. (4) III. EXPERIMENT Metalorganic chemical vapor deposition was used to grow 1.5- m thick n-gan:si layer on a (0001) sapphire substrate. This was followed by the growth of 0.05- m thick In Ga N/GaN MQW layers with five periods and then a 0.25- m thick p-gan:mg layer. In terms of fabrication of the LED device, the p-gan layer was selectively etched to expose the n-type layer using an inductively-coupled plasma system. An Ni/Au (2 nm/6 nm) transparent layer was then deposited onto the surface of p-gan layer. This was followed by the deposition of an Ni/Au (30 nm/80 nm) layer, in order to achieve a p-ohmic pad. For an n-ohmic pad, a Ti/Al (30 nm/80 nm) layer was deposited on n-gan, and the metal-deposited samples were then annealed at 450 C for 40 s in a rapid thermal annealing system. All electrical and optical properties of the LED performance were evaluated via on-wafer probing of the devices. The current-voltage ( - ) characteristics were measured using a parameter analyzer (HP 4155A). The light output power of the LED was measured using a UV/VIS 818 PD. Under a constant stress current, the lifetime of the LED was determined from a complete turn-off of the light output power. IV. RESULTS AND DISCUSSIONS A. The Applied Current Density The dependence of the applied current density on the LED characteristics was investigated. The output power of light and the device lifetime as a function of the current density is shown in Fig. 2(a) and (b), respectively. These measurements were performed using an LED chip with an overall device size of m m. The - characteristic of the LED [Fig. 2(a)] shows that the output power increases with increasing current density, reaching a maximum value at critical current density of 270 A/cm, and then decreases. This indicates that the output power is considerably dependent on the current density. Fig. 2(b) shows the relationship between the device lifetime and the stress current density. It can be seen that LED lifetime significantly decreases with increasing current density. According to the theory for a contact electromigration (EM) failure Fig. 2. (a) Light output power (L) versus current density (J ) at room temperature continuous-wave operation. (b) The lifetime (t ) as a function of forward current density. The dotted line represents the extrapolation of lifetime, based on the current range with n value of 2.06. model, the relationship between the lifetime and the current density can be explained by the empirical expression [5] [7] where is a constant and the exponent. From this theory, it is well known that is a function of Joule heating and approaches 2 for the case of negligible Joule heating at a low stress current [5], [6]. As shown in Fig. 2(b), it can be seen that the value of approaches 2 at approximately the critical current density of 270 A/cm, indicating that Joule heating is negligible. However, the value of significantly increases with increasing current density. This shows that a high current density could lead to current crowding in a localized area of the device, resulting in the degradation of the device lifetime as the result of significant Joule heating. Using the value of 2.06 in the negligible Joule heating current range, the LED lifetime is predictable. The calculation shows that the lifetime of the device under normal operating condition of 20 ma at room temperature is s. B. The Effective Length for the Lateral Current Path Effects of the effective length for the lateral current path on the LED performance were investigated. For this work, the electrical, optical, and reliability characteristics of the LEDs were evaluated as a function of device size, as shown in Table I. The results of the threshold voltage and the leakage current density obtained from a variety of devices are listed (5)
KIM et al.: EFFECTS OF CURRENT SPREADING ON PERFORMANCE OF GaN-BASED LEDs 1067 TABLE I ALIST OF LEDS WITH THE VARIOUS DEVICES SIZES. THE ACTIVE AREA IS BASED ON THE LIGHT-EMITTING AREA INCLUDING p-type LAYER. THE ACTIVE AREA RATIO IS THE RELATIVE ACTIVE AREA COMPARED TO THAT OF SYMBOL F. THE THRESHOLD VOLTAGE (V ) WAS DETERMINED FROM THE BIAS UNDER CURRENT LEVEL OF 20 ma. J INDICATES THE LEAKAGE CURRENT DENSITY EVALUATED UNDER A FORWARD BIAS OF 0.5 V Fig. 4. The LED lifetime (t ) as a function of the active area ratio of device. To evaluate the LED lifetime, a constant stress current density was applied. The applied current density was 160 A/cm and 200 A/cm for symbol A and B F, respectively. Considering reliability characteristics, native defects should be also taken into consideration. As given in Table I, of the small-size device is much lower than that of the large-size device, which is indicative of the reduced nonradiative recombination centers [8]. It is evident that, whether the origin of the leakage current is due to native bulk defects or etch damage-induced surface defects, the nonradiative recombination centers are also related to the LED lifetime. Therefore, it is concluded that the superior reliability characteristics for the small-size devices are due to combined effects of the uniform current spreading and the reduced nonradiative recombination centers. In Fig. 3(b), the presence of the nonradiative recombination centers could be confirmed from observation of a lower efficiency at low current density [8]. The efficiency begins to increase with increasing current density by means of saturating nonradiative recombination centers. In addition, the efficiency of the device is reduced above the critical current density due to Joule heating of the device. Fig. 3. (a) The current density (J )-voltage (V ) and (b) the differential quantum efficiency ()-current density (J ) characteristics for LEDs with various device sizes. in Table I. It is well known that, as the device sizes (or active area) are scaled down (from the symbol A to F ), the threshold voltages are increased. This is due to the reduced cross-sectional area for current flow with a scaling down of device size. In fact, however, the current density of a small-size device is even higher than those of larger-size devices, as shown in Fig. 3(a). In addition, it can be seen that the differential quantum efficiency can be greatly improved with a scaling down of device size, as shown in Fig. 3(b). Such considerable improvements of the electrical and optical characteristics with a scaling down of device size can be attributed to the reduced effective length for the lateral current path, which leads to the uniform current spreading. These results are consistent with the improved reliability characteristics as the device sizes are scaled down (Fig. 4). C. Geometrical Device Design Based on the condition (4), an ideal geometrical design of LED is possible for the fabrication of the highly efficient device. Basically, this can be realized by minimizing the effective length in terms of reducing the device size. However, it should be noted that the threshold voltage increases with a scaling down of device size as discussed in Table I. This factor prevents the practical application of the sufficiently small-size device. In this regard, it is concluded that device dimensions of m mor m m are appropriate for both aspects of LED characteristics and practical applications. In addition, the LED performance could be considerably improved by modifying the p-type pad structures as shown in Fig. 5. Fig. 5 shows the cumulative failure plot for m m LEDs with different p-type pad structures. The inset (a) is the standard LED design for an evaluation of the scaling effect of LED size. Compared to this design, it is clearly shown that the design (b) has the improved reliability characteristics. This can be attributed to the release of the current crowding problem in a localized region of the device for the case of LED design (b). Considering the LED design
1068 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 48, NO. 6, JUNE 2001 output characteristics, as shown in Fig. 6. It should be noted that the output power characteristic was improved, although the additional p-type pad can reduce the light emitting area. These results show the importance of uniform current spreading by means of the geometrical device design for the ideal device performance. V. SUMMARY Fig. 5. Cumulative failure plots for 300 m 2 300 m LEDs with a different shape of p-type pad structures. The inset (a) represents the plane-view of the standard LED design with a rectangular p-type pad structure. The insets of (b) and (c) show the plane-view of the modified LED design. The dotted arrows in the LED schematic indicate a possible current path for uniform current spreading during LED performance. The LED lifetimes were measured under stress current density of 460 A/cm at 20 C. Two important parameters, both current density and the effective length, were proposed to understand the effects of current spreading on the performance of GaN/InGaN MQW s LED from the presented simple model. It was clearly shown that two parameters are sensitively dependent upon the electrical, optical and reliability characteristics of LEDs, indicating that the proposed model and parameters are valid. Based on these results, the geometrical design of the highly efficient and reliable LED was successful. ACKNOWLEDGMENT The authors would like to thank N.-M. Park and J.-S. Jang at Kwangju Institute of Science and Technology for many useful discussions. REFERENCES Fig. 6. I V characteristics for LEDs with modified geometrical designs of 1 and 2. (a), the extremely short effective length can be found between the corners of n- and p-type pad (marked in dotted circle). Thus, it is predictable that the high current density tends to flow heavily in this localized region. Furthermore, this was confirmed by the observation of an optical microscope during the lifetime test under high stress current injection. That is, a strikingly bright luminescence was observed in this localized region during operation of LED. Therefore, it is concluded that the extremely short effect length in a localized region should be avoided for an ideal geometrical design of the LED. Basically, for the uniform current spreading, the effective length should be minimized. As shown in the inset (b), schematically, the parameter can be expressed as the sum of and. It is noteworthy that the parameter can be considerably reduced by adding an extra p-type pad as shown in the inset (c). In this case, it was clearly shown that the lifetimes of devices could be improved by an order of magnitude compared to those of design (b). Furthermore, we were able to observe the improved light [1] S. Nakamura, T. Mukai, and M. Senoh, Candela-class high-brightness InGaN/AlGaN double-heterostructure blue-light-emitting diodes, Appl. Phys. Lett., vol. 64, pp. 1687 1689, 1994. [2] S. Nakamura and G. Fasol, The Blue Laser Diode, Berlin, Germany: Springer-Verlag, 1997. [3] I. Eliashevich, Y. Li, A. Osinsky, C. A. Tran, M. G. Brown, and R. F. Karlicek Jr., InGaN blue light-emitting diodes with optimized n-gan layer, in SPIE Conf. Light-Emitting Diodes: Research, Manufacturing, and Applications-Part III, vol. 3621, 1999, pp. 28 36. [4] H. Kim, J.-M. Lee, C. H, S.-W. Kim, D.-J. Kim, S.-J. Park, and H. Hwang, Modeling of a GaN-based light-emitting diode for uniform current spreading, Appl. Phys. Lett., vol. 77, pp. 1903 1904, 2000. [5] H. Kim, H. Yang, C. H, S.-W. Kim, S.-J. Park, and H. Hwang, Electromigration-induced failure of GaN multi-quantum well light emitting diode, IEE Electron. Lett., vol. 36, pp. 908 910, 2000. [6] C. Y. Chang and S. M. Sze, ULSI Technology. New York: McGraw- Hill, 1996. [7] A. S. Oates, Electromigration failure of contacts and vias in sub-micron integrated circuit metallizations, Microelectron. Reliab., vol. 36, pp. 925 953, 1996. [8] M. Hansen, P. Kozodoy, S. Keller, U. Mishra, J. Speck, and S. Denbaars, The effect of diode area on the luminescence of InGaN quantum well light emitting diode grown by MOCVD, in Proc. 2nd Int. Symp. Blue Laser and Light Emitting Diodes, Chiba, Japan, Sept. 29 Oct. 2 1998, pp. 540 543. Hyunsoo Kim was born in Ulsan, Korea, on October 6, 1974. He received the B.S. degree in metallurgical engineering in 1997 from Pusan National University, Pusan,Korea and the M.S. degree in materials science in 1999 from Kwangju Institute of Science and Technology, Kwangju, Korea, where he is currently pursuing the Ph.D. degree in materials science and engineering. His current research interests are device design, modeling, and reliability of GaN-based LEDs.
KIM et al.: EFFECTS OF CURRENT SPREADING ON PERFORMANCE OF GaN-BASED LEDs 1069 Seong-Ju Park was born in Korea in 1952. He received the B.S. degree in chemistry and the M.S. degree in physical chemistry from Seoul National University, Seoul, Korea, in 1976 and 1979, respectively, and the Ph.D. degree in physical chemistry from Cornell University, Ithaca, NY, in 1985. From 1985 to 1987, he was a Post-doctoral fellow at the IBM T. J. Watson Research Center, Yorktown Heights, NY. From 1987 to 1995, he was with the Electronics Telecommunications Research Institute,Taejeon, Korea, as a Principal Researcher. In 1995, he joined the faculty at Kwangju Institute of Science and Technology, Kwangju, Korea, as a Professor in the Department of Materials Science and Engineering. Presently, he is a Professor of the department and the Director of the Center for Photonic Materials Research. He has engaged in research on growth and characterization of semiconductor epitaxial structures, characterization of nanoelectronic and photonic materials, atomic and electronic structures of semiconductor surfaces, and plasma etching and reaction mechanism. Hyunsang Hwang was born in Korea in 1966. He received the B.S. degree in metallurgical engineering from Seoul National University, Seoul, Korea, in 1988, and the Ph.D. degree in materials science from University of Texas at Austin in 1992. From 1992 to 1997, he was with the LG Semicon Corporation, Cheongju, Korea, as a Principal Researcher. In 1997, he joined the faculty at Kwangju Institute of Science and Technology, Kwangju, Korea, as a Professor in the Department of Materials Science and Engineering. His research interests are process and device design of deep submicron MOSFET, MOSFET device reliability, ultrathin dielectric, and optoelectronic devices.