A novel conducting filament (CF)-embedded indium tin oxide (ITO) film is fabricated using an electrical breakdown method. To assess the performance of this layer as an ohmic contact, it is applied to GaN (gallium nitride) light-emitting diodes (LEDs) as a p-type electrode for comparison with typical GaN LEDs using metallic ITO. The operating voltage and output power of the LED with the CF embedded ITO are 3.93 V and 8.49 mW, respectively, at an injection current of 100 mA. This is comparable to the operating voltage and output power of the conventionally fabricated LEDs using metallic ITO (3.93 V and 8.43 mW). Moreover, the CF-ITO LED displays uniform and bright light emission indicating excellent current injection and spreading. These results suggest that the proposed method of forming ohmic contacts is at least as effective as the conventional method.
© 2015 Optical Society of America
Transparent conductive electrodes (TCEs) have been extensively researched owing to their wide-ranging applications in optoelectronic devices, such as flat panel displays, solar cells, sensors, and light-emitting diodes (LEDs) [1–3]. TCE materials having high optical transmittance and low electrical resistance that can serve as ohmic contacts are essential to further improve the electrical and optical properties of conventional optoelectronic devices. In general, to form high-quality ohmic contacts with semiconductor materials, the Schottky barrier height between the metal and the semiconductor should be made as low as possible to minimize the work function difference between the metal and semiconductor (heavy doping of the semiconductor can help with this). Indium tin oxide (ITO) is an attractive TCE due to its superior electrical conductivity and optical transparency in the visible range. For this reason, ITO has been most widely used as a transparent ohmic electrode in optoelectronic devices, particularly on p-type gallium nitride (p-GaN) used in commercial GaN-based LEDs, because of its excellent current spreading and effective photon extraction characteristics . However, conventional ITO is not effective in the ultraviolet (UV) regime because of its large light absorption. With the optical band gap energy of ~3.4 eV, the cut-off wavelength of conventional ITO is known to be 354 nm . In addition, a limited supply of indium and increasing costs of mining and refinement have led to a push in recent years to find alternatives or replacements for ITO (i.e., indium-free TCEs) . Many scientists and engineers have worked with new functional materials such as silver nanowires , carbon nanotubes , thin metal films , graphene layers , and conductive polymers , but this area still remains quite challenging.
Ideally, any alternative TCE material should have good ohmic properties when used with various types of semiconductors as well as high transmittance across a wide optical spectrum – including UV light. However, it is difficult to simultaneously solve a trade-off problem between optical transmittance and electrical conductivity, particularly for p-(Al)GaN. The work function of p-(Al)GaN (~7.5 eV) is much larger than that of metals, and it is difficult to achieve heavy doping in p-(Al)GaN . Therefore, it is highly desired to develop an alternative TCE whose optical and electrical device performance is similar to, or better than, that of ITO even in the UV wavelength region. To provide a fundamental solution to these issues, we have reported a universal method of producing both high electrical conductivity and high optical transmittance using an electrical breakdown (EBD) method , and its successful application to vertical-type GaN LEDs .
In this study, we investigate the feasibility and usefulness of the proposed ohmic method by applying the conducting filament (CF)-embedded insulating ITO (I-ITO) to lateral-type GaN LEDs as a TCE and directly comparing their device performance to GaN LEDs with the conventional metallic ITO (M-ITO).
2.1 Fabrication of GaN-based lateral LEDs
Lateral-type GaN-based LEDs emitting at ~450 nm (dimensions: ~1150 μm × 700 μm) were fabricated on LED templates. The LED structures were grown by metal organic chemical vapor deposition on a c-plane sapphire substrate, which consists of a 2-μm-thick undoped GaN layer, a 3-μm-thick Si-doped n-GaN layer, a layer of InGaN/GaN multiquantum wells (MQWs) with five periods, a 10-nm-thick p-AlGaN blocking layer, a 0.2-μm-thick Mg-doped p-GaN layer, and a current blocking layer. In this sample, a hole concentration of 1 × 1019 cm−3 was designed to be achieved after thermal annealing of the p-GaN layer. Subsequently we performed a series of LED fabrication processes including photolithography and ICP etching. Following this, 30-nm-thick TCE films with isolated mesa structures were deposited on the p-GaN layers using a radio frequency magnetron sputtering system. M-ITO was deposited in Ar gas environments at a base pressure of ~2 × 10-7 Torr and a working pressure of ~1 × 10-3 Torr whereas I-ITO was deposited in Ar and O2 mixture gas environments at a base pressure of ~2 × 10-7 Torr and a working pressure of ~1 × 10-2 Torr to increase the resistivity of ITO. Subsequently, the EBD process involving the application of electrical voltages was conducted along the current blocking layer patterns on the p-GaN layer to form CFs within the I-ITO films. Finally, Cr/Al/Ni/Au was deposited by electron-beam evaporation on the n-GaN and p-GaN surfaces to form the n- and p-type electrodes, respectively. For the purpose of comparing the performance of different types of device contacts, LEDs without CF-embedded TCEs were also prepared using the same LED epitaxial wafer under the same fabrication conditions.
2.2 Electrical and optical characterizations
First, transmission line method (TLM) was used to evaluate the ohmic contact characteristics of the proposed CF-embedded TCEs deposited on the p-GaN layers. To fabricate the TLM patterns, mesa isolation was performed using an inductively coupled plasma etcher. Then, the TLM patterns (100 × 100 μm2) on the contact pad were formed with varied spacing (ranging from 2 μm to 10 μm) using photolithography and a development process. Finally, 30-nm-thick ITO films with isolated mesa structures were deposited on the p-GaN layers using a radio frequency magnetron sputtering system with ambient Ar–O2 gas at a base pressure of ~2 × 10−7 Torr and a working pressure of ~1 × 10−3 Torr. Afterward, a lift-off process was performed using acetone and the electrical characteristics of the contacts were measured using a Keithley 4200 semiconductor parameter analyzer.
Next, to quantitatively evaluate the transmittance of the proposed CF embedded TCEs, 30-nm-thick I-ITO films were deposited on the quartz substrates using the radio frequency sputtering system. Then, the transmittance of the ITO films on the quartz substrates was measured as a function of wavelength using a Lambda 35 UV/VIS Spectrometer with an operating wavelength range from 240 nm to 700 nm.
Finally, the light output power and the operating voltage were measured for full-structure LED chips using a wafer-level LED measurement system (OPI-150, With Light Co., Ltd.). More specifically, the light output power of each LED was measured from the top-side of the LEDs using a Si photodiode connected to an optical power meter. In addition, light-emission images of the lateral LED chip surface were acquired using a charge-coupled device camera.
3. Results and discussion
Figure 1(a) presents the schematic drawing of a GaN-based LED having a CF-embedded I-ITO electrode after application of the EBD, and the magnified atomic structure detail shows that CFs can be formed by generating oxygen vacancies in the I-ITO to provide CFs between the TCE and the p-GaN layer for carrier injection. To obtain the CFs within the I-ITO, we performed a direct current (DC) bias sweep from 0 to 20 V between two-point probe contacts on the TCE using a Keithley 4200 semiconductor parameter analyzer. As can be seen in Fig. 1(b), the current in the current versus voltage (I–V) curve abruptly increased at about 13 V (blue line), and then in second DC bias sweep, the linear I–V curve (red line) is observed with 5.9 mA at 1 V. In other words, the virgin state of the cell remained in the high resistance state. However, after the EBD process (i.e., sudden increase in the current), the cell was changed to a low resistance state (LRS) . The more detailed fabrication steps are shown in our previous work . In addition, it was found that the current level at 1 V increased from ~850 nA to ~5.9 mA after EBD (Fig. 1(b)). To evaluate the long-term stability in the LRS, we then measured the retention characteristics at 1 V, as shown in Fig. 1(c). The results of this measurement indicate that the LRS could be maintained for >105 s. Using measured retention data, we also extrapolated the results against the read delay time to estimate the long-term retention of the cell. As a result, we find that the LRS of the cell could be stably maintained for ~10 years.
To understand the origin of the CFs in the I-ITO, we have investigated the change of atomic concentration in the TCE film using Auger electron spectroscopy analysis before and after the EBD process, in comparison with the change in M-ITO film (inset), as shown in Fig. 1(d). In general, in the related research area [16–19], it is believed that CFs consist of oxygen vacancies (or oxygen ions) within the insulating films created through the EBD process. Efforts have been made to analyze the origin of CFs in the insulating material and some mechanisms have been revealed. First, Wei et al. reported that the formation of CFs, acting as a conduction path, can be observed after EBD using transmission electron microscopy analysis . They also found that the oxygen concentration is reduced at CFs and around CFs, compared to the virgin state, using Electron Energy-Loss Spectroscopy. Then, Qi et al. reported that oxygen vacancy-based CFs can be generated by redox reaction under the application of a bias . In addition, they measured 100 nm-wide CFs using conductive atomic force microscopy, and investigated their vertical structures using piezoresponse force microscopy to analyze phase images. On the other hand, Kamiya et al.  demonstrated that oxygen structural phase transition with oxygen vacancy cohesion-isolation is the physical origin for resistive switching of binary oxide films using first-principle calculations based on density functional theory. In this study we took a materials chemistry approach to find other evidence for oxygen vacancy-based CFs in the insulator materials. First, we observed a reduced oxygen concentration of 4% in I-ITO films after EBD in the Auger electron spectroscopy analysis, particularly at the interface between the ITO films and the p-GaN layer. These phenomena can be explained by the fact that the generated CFs consisting of oxygen vacancies in I-ITO films might help the formation of ohmic contact between the TCEs and the p-GaN layers. In addition, according to the literatures [20,21], the oxygen vacancies generated at the interfacial layer can effectively reduce the energy barrier height between the TCEs and the p-GaN layer due to doping effects. We also measured x-ray photoelectron spectroscopy results at the surface of the metallic ITO (M-ITO) and I-ITO before and after the EBD, to analyze further the generation of CFs related to the oxygen vacancies. In some previous results [20–23], after the formation of CFs, the peaks of each element in the spectra are shifted in the high-energy direction because of microstructural changes, compared with normal insulating films. In Fig. 1(e), the results indicate that In, Sn, and O elements coexist on the entire ITO films. In the case of the M-ITO films, binding energies of a photoelectron in In 3d5/2, In 3d3/2, Sn 3d5/2, Sn 3d3/2, and O 1s are determined to be 445.0, 452.6, 487.0, 495.5 and 530.7 eV, respectively. The observed values reveal that the valence states of In, Sn, and O in the sample are mainly + 3 and + 4, respectively, which are consistent with previous results . In addition, as can be seen in the enlarged data for each atom at right of Fig. 1(e), the In 3d5/2, In 3d3/2, Sn 3d5/2, Sn 3d3/2, and O 1s signals in the I-ITO films are shifted slightly upwards, compared with the M-ITO films, due to the microstructural change in the films [19,24]. From these results, we can deduce that binding energies of each atom might be dependent on oxygen concentration in the ITO films. In addition, in the CFs embedded in the I-ITO films after the EBD, all signals are also shifted downwards compared to that of the I-ITO films, even that of the M-ITO films. As a result, it was found that the CFs within the I-ITO films were generated by the microstructural change of the ITO structures, and thereby binding energies of In, Sn, and O elements have been reduced by the microstructural changes via oxygen vacancy generation in the ITO films. Consequently, on the basis of electrical and material property analyses, we propose a possible structure of the CFs within the TCEs as schematically illustrated in Fig. 1(a). As shown in this figure, the CFs are composed of the chain of oxygen vacancies generated with a high conductivity by out-diffusion of oxygen ions and then the carrier can be injected from the p-metal electrode to the p-GaN layer across the CFs path in the I-ITO films.
In order to evaluate the contact property of the CF-embedded TCEs on the p-GaN layer, we investigated the specific contact resistances (ρc) of CF-embedded p-GaN/TCEs using a TLM of the I-ITO films before and after EBD, as shown in Fig. 2(a). Before EBD, we observed a very low current level (~850 nA) at 1 V in the I–V curves (inset), probably due to the large Schottky barrier height between the I-ITO and the p-GaN layer as well as the extremely low conductivity of the TCEs. However, after EBD, we observed a perfect ohmic behavior with a ρc of 1.38 × 10−4 Ω·cm2 for I-ITO cells deposited on the p-GaN layer. For comparison, we measured the I–V curves for M-ITO cells deposited on the p-GaN layers, as shown in Fig. 2(b); the ρc was measured to be 5.94 × 10−3 Ω·cm2. The improvement in the ρc could be explained by the reduced Schottky barrier height between the TCEs and p-GaN layer due to the locally generated energy levels in the CF-embedded I-ITO, which allows effective carrier injection from the metal electrode to the p-GaN layer across the TCEs via the CFs.
Next, we investigated the optical transmittances as a function of the wavelength to examine the transparency of the M-ITO and I-ITO on quartz substrates, as shown in Fig. 3(a), respectively. In addition, using the results of the optical transmittance, we have calculated the optical band gap energy (Eg) of the M- and I-ITO films, as shown in Fig. 3(b). As a result, the increased Eg is found to undergo a blue-shift with an increase in the oxygen concentration and BEs of the ITO films, as shown in Fig. 1(d) and 1(e). Consequently, comparing with the M-ITO films, the optical transmittances of the I-ITO films is increased from 91.5% to 95.4% at 450 nm. Especially in the I-ITO films, it is found that the higher optical transmittance of 87% at 365 nm and 80% at 300 nm is stably maintained in the UV region as well, which is the additional benefit of CF-embedded I-ITO in terms of transmittance, compared to the M-ITO films. These results indicate that the CF-embedded I-ITO TCEs are suitable for use as transparent ohmic contacts with a p-GaN layer. By way of opto-electrical and material analyses, we investigated the band diagram of lateral LEDs structures having the CF-embedded I-ITO using a SiLENSe simulator, as shown in Fig. 3(c). In this band diagram, we have modified the Eg of CFs embedded I-ITO based on the results of ohmic and Eg characteristics of I-ITO as well as a related result reported in a previous work . As a consequence, employing the CF-embedded I-ITO via EBD method, we could achieve an improvement in both the ohmic contact property between p-GaN and TCEs and the transmittance properties as a result of having a smaller Schottky barrier height and a higher Eg in the CF-embedded I-ITO, compared to M-ITO.
Finally, we fabricated lateral-type GaN LEDs using three different types of TCEs: M-ITO, CF-embedded I-ITO after EBD, I-ITO. The chip size was 1150 μm × 700 μm as shown in Fig. 4(a). To demonstrate the validity of the proposed concept on the basis of device performance, three types of LED device characteristics, including the I–V curve, light output power, and emission profile characteristics, were compared as shown in Fig. 4(b)-4(d). It is seen that the light output power and the operation voltage at an injection current of 100 mA for LEDs having CF-embedded I-ITO were 8.49 mW and 3.93 V, respectively, which was nearly equivalent to the voltage and power of conventional LEDs with M-ITO (8.43 mW and 3.93 V). This similarity in device performance can be explained by the high optical transparency as well as having a sufficiently low contact resistance between CF-embedded I-ITO and p-GaN, which leads to efficient current injection and excellent current spreading from the metal electrodes into the LEDs. These results show the effectiveness of the proposed CF-based ohmic method and also imply that the EBD process used for CF generation in TCEs does not cause any severe degradation in device performance. Moreover, for the same current levels, the light output power and the operating voltage of the LEDs with I-ITO were measured to be 7.42 mW and 4.83 V, respectively. The optical and electrical performance of the I-ITO based devices is not as good as conventional LEDs with M-ITO and LEDs with CF-embedded I-ITO. The I-ITO devices’ decreased light output power and higher operating voltage are attributed to the low conductivity of I-ITO and the high contact resistance between the interface of I-ITO and p-GaN. Although the overall device performance of the blue-emitting LEDs with CF-embedded I-ITO is comparable to that of the blue-emitting LEDs with M-ITO, when we consider the fact that the CF-embedded TCEs have the benefit of good contact ability and also a larger optical Eg than that of the M-ITO in the UV regime, we believe that the proposed approach of CF-embedded TCEs might be effectively applied to UV LEDs as an improvement to conventional UV LEDs using M-ITO. Moreover, to evaluate the influence of CFs on the current spreading and injection behaviors, we investigated the light emission images of the three types of LEDs at injection currents of 5, 10, 15, 20, and 30 mA, respectively. As shown in Fig. 4(d), the LEDs with CF-embedded I-ITO and M-ITO exhibited the brightest and most uniform light emission over the entire top area of the chip surface, whereas the LEDs with I-ITO had a high light emission intensity around the p-contact region due to incomplete current spreading. Such observations agree with the results shown in Fig. 4(b) and 4(c). In this figure, the difference in brightness can be clearly distinguished when the injection current is below 10 mA. This difference might be a result of both the efficient current injection and spreading from the metal to p-GaN through the CFs across the I-ITO as well as the low optical loss of CF-embedded I-ITO.
In this study, lateral-type GaN-based blue-emitting LEDs using CF-embedded TCEs with high conductivity and high optical transmittance were successfully fabricated by a novel EBD process. The CF-embedded I-ITO enabled direct-ohmic contacts to p-GaN layers with a specific contact resistance of 1.38 × 10−4 Ω·cm2 and exhibited high optical transmittance of 95.4, 87.5, and 80% at 450, 365, and 300 nm, respectively. Consequently, compared to LEDs with I-ITO, the light output power and the operating voltage were significantly improved for the LEDs with CF-embedded I-ITO, which were comparable to those of the conventional LEDs with M-ITO. Based on these device results, we demonstrated that the proposed CF-based ohmic method was effective in spreading and injection characteristics of LED devices and indicates its potential for use as a TCE in LED device applications, as well as its possible use as a replacement for conventional M-ITO contacts used in the visible regime. In particular, significant utility for this method may be found with UV LEDs because of the good ohmic contact properties and higher than usual optical transmittance in the UV regime.
This work was supported by a Korean government National Research Foundation of Korea (NRF) grant (No. 2011-0028769).
References and links
1. D. S. Hecht, L. Hu, and G. Irvin, “Emerging transparent electrodes based on thin films of carbon nanotubes, graphene, and metallic nanostructures,” Adv. Mater. 23(13), 1482–1513 (2011). [CrossRef] [PubMed]
3. E. Fortunato, D. Ginley, H. Hosono, and D. C. Paine, “Transparent conducting oxides for photovoltaics,” Mater. Res. Soc. Bull. 32(03), 242–247 (2007). [CrossRef]
4. D. J. Chae, D. Y. Kim, T. G. Kim, Y. M. Sung, and M. D. Kim, “AlGaN-based ultraviolet light-emitting diodes using fluorine-doped indium tin oxide electrodes,” Appl. Phys. Lett. 100(8), 081110 (2012). [CrossRef]
5. S. Ghosh, H. S. Kim, K. P. Hong, and C. M. Lee, “Microstructure of indium tin oxide films deposited on porous silicon by rf-sputtering,” Mater. Sci. Eng. B 95(2), 171–179 (2002). [CrossRef]
9. D. S. Ghosh, T. L. Chen, and V. Pruneri, “High figure-of-merit ultrathin metal transparent electrodes incorporating a conductive grid,” Appl. Phys. Lett. 96(4), 041109 (2010). [CrossRef]
10. S. Bae, H. Kim, Y. Lee, X. Xu, J.-S. Park, Y. Zheng, J. Balakrishnan, T. Lei, H. R. Kim, Y. I. Song, Y.-J. Kim, K. S. Kim, B. Özyilmaz, J.-H. Ahn, B. H. Hong, and S. Iijima, “Roll-to-roll production of 30-inch graphene films for transparent electrodes,” Nat. Nanotechnol. 5(8), 574–578 (2010). [CrossRef] [PubMed]
11. Y. H. Kim, C. Sachse, M. L. Machala, C. May, L. Müller-Meskamp, and K. Leo, “Highly conductive PEDOT:PSS electrode with optimized solvent and thermal post-treatment for ITO-free organic solar cells,” Adv. Funct. Mater. 21(6), 1076–1081 (2011). [CrossRef]
12. H. Morkoç, “Materials properties, physics and growth,” in Handbook of Nitride Semiconductors and Devices (Willey-VCH, 2008).
13. H.-D. Kim, H.-M. An, K. H. Kim, S. J. Kim, C. S. Kim, J. Cho, E. F. Schubert, and T. G. Kim, “A Universal method of producing transparent electrodes using wide-bandgap materials,” Adv. Funct. Mater. 24(11), 1575–1581 (2014). [CrossRef]
14. S. J. Kim, H.-D. Kim, K. H. Kim, H. W. Shin, I. K. Han, and T. G. Kim, “Fabrication of wide-bandgap transparent electrodes by using conductive filaments: performance breakthrough in vertical-type GaN LED,” Sci. Rep. 4, 5827 (2014). [PubMed]
15. H.-D. Kim, H.-M. An, K. C. Kim, Y. Seo, K.-H. Nam, H.-B. Chung, E. B. Lee, and T. G. Kim, “Large resistive-switching phenomena observed in Ag/Si3N4/Al memory cells,” Semicond. Sci. Technol. 25(6), 065002 (2010). [CrossRef]
16. T. Ninomiya, T. Takagi, Z. Wei, S. Muraoka, R. Yasuhara, K. Katayama, Y. Ikeda, K. Kawai, Y. Kato, Y. Kawashima, S. Ito, T. Mikawa, K. Shimakawa, and K. Aono, “Conductive filament scaling of TaOx bipolar ReRAM for long retention with low current operation,” in Symposium onVLSI Tech. Dig. (IEEE, 2012), pp. 73–74.
17. Z. Wei, T. Takagi, Y. Kanzawa, Y. Katoh, T. Ninomiya, K. Kawai, S. Muraoka, S. Mitani, K. Katayama, S. Fujii, R. Miyanaga, Y. Kawashima, T. Mikawa, K. Shimakawa, and K. Aono, “Demonstration of high-density ReRAM ensuring 10-year retention at 85°C based on a newly developed reliability model,” in Conference on IEDM Tech. Dig. (IEEE, 2011), pp. 721–724.
19. K. Kamiya, M. Y. Yang, T. Nagata, S.-G. Park, B. Magyari-Köpe, T. Chikyow, K. Yamada, M. Niwa, Y. Nishi, and K. Shiraishi, “Generalized mechanism of the resistance switching in binary-oxide-based resistive random-access memories,” Phys. Rev. B 87(15), 155201 (2013). [CrossRef]
20. T. Bertaud, D. Walczyk, M. Sowinska, D. Wolansky, B. Tillack, G. Schoof, V. Stikanov, Ch. Wenger, S. Thiess, T. Schroeder, and Ch. Walczyk, “HfO2-based RRAM for embedded nonvolatile memory: from materials science to integrated 1T1R RRAM arrays,” ECS Trans. 50(4), 21–26 (2013). [CrossRef]
21. M. T. Greiner, L. Chai, M. G. Helander, W.-M. Tang, and Z.-H. Lu, “Transition metal oxide work functions: the influence of cation oxidation state and oxygen vacancies,” Adv. Funct. Mater. 22(21), 4557–4568 (2012). [CrossRef]
22. V. Y.-Q. Zhuo, Y. Jiang, M. H. Li, E. K. Chua, Z. Zhang, J. S. Pan, R. Zhao, L. P. Shi, T. C. Chong, and J. Robertson, “Band alignment between Ta2O5 and metals for resistive random access memory electrodes engineering,” Appl. Phys. Lett. 102(6), 062106 (2013). [CrossRef]
23. Y.-T. Chen, T.-C. Chang, J.-J. Huang, H.-C. Tseng, P.-C. Yang, A.-K. Chu, J.-B. Yang, H.-C. Huang, D.-S. Gan, M.-J. Tsai, and S. M. Sze, “Influence of molybdenum doping on the switching characteristic in silicon oxide-based resistive switching memory,” Appl. Phys. Lett. 102(4), 043508 (2013). [CrossRef]
24. S. Kim, S. B. Kim, and H. C. Choi, “Influence of thermal annealing on the microstructural properties of indium tin oxide nanoparticles,” Bull. Korean Chem. Soc. 33(1), 194–198 (2012). [CrossRef]
25. E. Lee, M. Gwon, D.-W. Kim, and H. Kim, “Resistance state-dependent barrier inhomogeneity and transport mechanisms in resistive-switching Pt/SrTiO3 junctions,” Appl. Phys. Lett. 98(13), 132905 (2011). [CrossRef]