─ 3 ─from the reaction interface, suppressing the diffusion ux of Ti through the phase adjacent to SiC. The Ni layer should be thick for this purpose, since the thickness of the Ni layer de-termines the chemical potential gradient and diffusion ux of Ti. Consequently, the Ti layer is formed on the Ni layer in the present study in order to make Ti form Ni-Ti intermetallic compounds and to supply Ti to the interface by diffusion of Ti through the intermetallic compound. The thickness was set to 100 nm. The phase diagram shown in Fig. 1 presents also another concern. Gülpen et al. reported that the interfacial phase se-quence of a SiC/ Ni diffusion pair becomes SiC/ (Ni2Si + C)/ (Ni31Si12 + C)/ Ni3Si/ Ni17). The sequence corresponds to the diffusion path indicated on Fig. 1 with a broken line. The dif-fusion path implies that NiSi cannot be formed while Ni exists as the end-member of the diffusion pair. Two methods to over-come this problem are considered. One is to change the an-nealing temperature at which SiC, NiSi and C coexist in equi-librium. However, the three-phase equilibrium cannot be achieved at temperatures below the eutectic of Ni and Ni3Si at 1416 K. The other method to form NiSi adjacent to SiC is to make Ni2Si react with SiC by reducing the chemical potential of carbon. Addition of Ti in the lm reduces the chemical po-tential by forming TiC. Fig. 2 shows the original and modied diffusion path on the Ni-Si-C ternary chemical potential dia-gram at 1173 K16). It is clearly understood that the diffusion path reported in the literature goes through a high chemical potential of carbon. Due to this, NiSi is kept away from for-mation. By preventing the formation of the free carbon, the diffusion pair is allowed to take a direct path with monotonous chemical potential gradients of all constituent elements. In ad-dition, also the formation of Ni-Ti intermetallic compound re-duces the chemical potential of Ni at one of the end-member, which magnies the driving force to establish a diffusion path through a low chemical potential of C. Therefore, the key point of the electrode formation process proposed in the present study is to control the interfacial reac-tion behavior of Ti, i. e., to facilitate the TiC formation pre-venting the formation of other Ti-Si-C byproducts. 2.2Experimental procedure SiC substrates used in the present study were nitro-gen-doped n-type 4H-SiC substrates cut to a size of 5.0 mm square from a 360-µm-thick wafer 50.8 mm in diameter of which surface correspond to (0001) crystallographic plane. Only the (0001) Si-face of the SiC substrates was used. Ni and Ti were deposited consecutively on the substrates by radio-fre-quency magnetron sputtering. The thickness of the Ni layer was kept constant at 100 nm, whereas that of the Ti layer was varied from 0 to 640 nm. The samples were then heated up to 1273 K in vacuum of 3×10–3 Pa and immediately cooled down after reaching the temperature (hereinafter described as the annealing time of 0 s). For comparison, TiC electrode was formed on some SiC substrates by deposition of C/Ti bilayer and subsequent annealing at 1273 K for 0 s in vacuum. Since this process can form TiC on the substrate without an interfa-cial reaction between the deposited lm and the substrate, Ti-Si-C byproducts will not be formed at the interface. The interfacial structures were analyzed by X-ray diffrac-tion (XRD). The electrical properties were measured by di-rect-current conduction test. The mechanical properties were evaluated by constant-load scratch test. A Rockwell-C indent-er was plunged into the electrode with a force of 5.0 N and swept along the surface of the electrode. The depth of the trench formed by the scratch was measured by atomic force microscopy at a position 1.0 mm distant from the start point of the scratch test. The reciprocal value of the trench depth was used as the strength index of the contact layer.2.3Results and discussion Fig. 3 shows the XRD patterns of the samples with various thickness of Ti layer after annealing. The initial thicknesses of the Ti layer for the patterns (a), (b), (c) and (d) are 0, 16, 80 and 640 nm, respectively. The pattern (a) consists of the peaks of SiC and Ni2Si. The peaks of NiSi and the free carbon do Fig. 2 Ni-Si-C ternary chemical potential diagram at 1173 K. The broken line indicates the actual phase sequence reported by Gülpen et al.7). The solid line indicates the modied phase sequence by lowering the chemical potential of carbon by insertion of a Ti layer. aC, aNi, and aSi represent the activity of C, Ni, and Si, respectively.
元のページ ../index.html#5