─ 7 ─gle crystal and continuous rings of ne Ni polycrystalline lm. Diffraction spots or rings corresponding to intermetallic compounds do not appear on the pattern.By combining the results shown in Figs. 7 and 8, it can be mentioned that the improvement in electric conductivity is achieved without changing the material adjacent to GaN, i. e., without lowering the Schottky barrier height. The improve-ment is brought by thinning of the Schottky barrier. The heat treatment at 673 K increases the apparent carrier concentration in GaN beneath the Ni electrodes and the application of volt-age enhances the increase. The van der Pauw measurement re-vealed that holes are the major carrier after annealing. There-fore, the results shown above lead to an answer: the heat treatment at 673 K applying 30 V between the electrodes acti-vates the acceptor dopants in GaN by evacuating the hydrogen atoms. Although acceptor dopants can be activated without voltage application, it takes a long time to achieve a certain conduc-tivity. This fact indicates that the heat evaporates hydrogen through GaN surface. The voltage application enhances the evacuation by the following steps. First, the voltage accumu-lates hydrogen in a narrow zone beneath an electrode. Then, the diffusion ux of hydrogen through the electrode becomes larger, since the ux is determined by the diffusion coefcient and the gradient of hydrogen composition. Therefore, the hy-drogen permeability of the electrode material is important. Fig. 9 compares three types of the electric conduction pro-les between two Pd electrodes formed on GaN. Comparison of Fig. 9 with Fig. 7 reveals the following two points. One is that the as deposited electrodes of Pd is conductive at high voltage. Another point is that the conductivity is improved by heat treatment and can be enhanced by application of voltage. The last point is that the conductivity of Pd electrode is lower than Ni, except the high voltage zone of as deposited state. Fig. 10 illustrates schematically the difference in the hydro-gen evacuation behavior between Ni and Pd. Ni electrodes does scarcely dissolve hydrogen. The hydrogen accumulated in the vicinity of the cathode by heat treatment applying volt-age must be evacuated to the atmosphere through the GaN/Ni interface, as shown in Fig. 10(a). On the other hand, Pd elec-trodes allow evacuation through both the Pd lm interior and the GaN/Pd interface, as shown in Fig. 10(b1). In addition, Pd electrodes can store a considerable amount of hydrogen. The hydrogen atoms stored in the Pd electrodes can return back into GaN, as shown in Fig. 10(b2). Low conductivity of Pd electrodes can be caused by those back-diffused hydrogen, since the inactivation of acceptor dopants by the mechanism will be more signicant in the volumes closer to the GaN/Pd interface.3.4Short summary The present paper proposed and demonstrated a new meth-od to improve the conductivity of an metallic electrode on p-type GaN. The following points became clear. Fig. 9 Differences in I-V characteristics among three GaN/Pd electrodes with or without heat treatment and voltage application.-2-1012-0.06-0.04-0.0200.020.040.06Current, I [mA]Voltage, V [V] as deposited annealed (673 K, 3.6 ks) annealed & voltage applied (673 K, 3.6 ks, 30 V)Fig. 10 Schematic illustratiion of the difference in hydrogen evacuation behavior between Ni and Pd.
元のページ ../index.html#9