Recently, (1 0 0) highly textured diamond thin films were successfully synthesized on Si(1 0 0) [1], but complete heteroepitaxial diamond films have not yet been synthesized. This is because the mechanisms of nucleation and growth of diamond by chemical vapour deposition (CVD) are not fully understood. To clarify the mechanism of the nucleation process at the various nucleation sites, continuous observation of the nucleation and growth process in the same area, ideally in situ observation, is required. Scanning tunnelling microscopy (STM) is one of the candidates for this purpose, since STM uses a microscope with atomic resolution which can be operated in various environments, even in CVD conditions. Moreover, STM can also be used for atomic scale spectroscopy (scanning tunnelling spectroscopy) [2] and for the fabrication of nanometre scale structures [3]. For these reasons, STM is expected to be applied not only to the observation of the nucleation process but also to the fabrication of artificial nucleation sites. STM observations of nucleation in diamond CVD have been reported [4, 5]; however, roughened Si wafers scratched by diamond powder have been used exclusively as the substrates. In addition, different areas were observed before and after the depositions. Therefore, it has been impossible to correlate the nucleation process and nucleation sites so far, although information about growth on the nanometre scale have been reported. In this letter, we report the observation of diamond nucleation in a hot-filament assisted CVD process at the nanometre scale holes fabricated artificially on highly orientated pyrolytic graphite (HOPG) by STM. Fig. 1 shows a schematic diagram of the apparatus used in this study. A commercial STM (NanoScope II, Digital Instruments Inc.) was in the STM part, and the deposition was performed by a hot-filament assisted method in the CVD part. To transfer a substrate from the STM part to the CVD part, and vice versa, an Invar rod fixed on the end of a micrometre head was used with a guiding rail and a stopper. The deviation of the substrate position before and after transferring the substrate was less than 3 im, which was enough to access the same area of the substrate surface by the STM. To minimize the deviation of the position, the distance between the STM part and the CVD part should be as short as possible. However, the piezoelectric elements of STM do not function well in a high temperature environment, so they were separated by 10 mm with a thermal shield mounted between them. Deposition time was controlled by a shutter inserted between the filament and the substrate. All of the STM studies were carried out in the constant-current mode at room temperature after exposure to air with etched Pt–Ir tips (Materials Analytical Services). The tunnelling current (It) and the bias voltage (Vb) were 0.5 nA and 100 mV, respectively. HOPG was used as a substrate, since large atomically flat and inert surfaces can be obtained easily by cleaving. However, nucleation of diamond hardly occurs on the as-cleaved (0 0 0 1) graphite surface [6]. Prior to deposition in this study, therefore, 3 3 3 holes at intervals of 200 nm were fabricated on an HOPG substrate by STM in air by raising Vb to 4 V with It 0.5 nA for 1 s, with the aim of forming preferential nucleation sites [7]. Fig. 2a shows an STM image of the nine holes, of about 40 nm diameter. After transferring the substrate to the CVD part, the chamber was evacuated by a rotary pump, and then CVD was performed on the substrate without auxiliary heating. The experimental conditions were as follows: gases: 2% CH4 in H2 (CH4 1 sccm and H2 49 sccm); total pressure: 6.7 3 103 Pa (50 torr); Ta filament temperature: 2100 8C (measured by an optical pyrometer); fila-
Read full abstract