Since most experiments in the literature were performed on highly doped oxide surfaces, we implanted nitrogen into these TiC>2 films. The resulting surface is almost metallized, similarly to TiN. AFM pictures (Fig. 3(a)) show the same structure as unimplanted surfaces (the z range is higher due to a longer electropolishing time; some grains (depending on their crystal orientation and electronic properties) will be electropolished more strongly than others. STM investigations on these surfaces were possible without difficut. Figure 3(b) shows a picture taken with -3 V and -1 nA setpoint. The bias could be varied up to -1 V (1 V respectively) and a stable tunneling condition was still achieved. The small substructures as seen in Fig. 1(d) were detected on these surfaces as well. In [17] we will demonstrate AFM and STM pictures of the same position on the TiN, which is important for an interpretation of STM effects on differently conducting grains.

Fig. 3. (a) AFM picture of nitrogen implanted Ti02, xy range 75 |im, z^ = 558 mn; (b) STM picture of the same surface, xy range 75 pm, zmax = 393 mn, -3V bias, -1 nA setpoint, W tip.

3.4 Tunneling spectroscopy of Ti02/TiN

An investigation on tunneling spectroscopy on titanium has already been presented in [20]. In addition, a systematic investigation was performed [17], which included the comparison between TiC>2 and TiN films on titanium. Figure 4 shows the voltage tunneling spectra of the two surfaces (with the same substrate as in Fig. 3 ). It is obvious that TiN has a higher conductivity at all voltages than T1O2 meanwhile the band gap is only apparent and cannot be correlated with the surface band gap. The apparent resistances (in branches) are several hundreds of megaohm, increasing to several gigaohms in the „band gap". The higher conductivity is the reason why STM pictures on implanted surfaces are possible in contrast to normal, anodically grown TiC>2, recorded in a dry nitrogen atmosphere. In contrast to the expected behavior we have found a dependency on the oxide thickness, the initial parameters, prepolarization, scan direction, scan rate, scan form, tip material and environmental conditions. Since the apparent band gap depends significantly on the oxide thickness, we have concluded (a) that it cannot be related to the surface band gap and (b) that the tip penetrates into the oxide and the mechanism is not a normal vacuum-type tunnel mechanism. Figure 5 shows some of these dependencies. Figure 5(a) is a comparison of voltage tunneling spectra on 5,15, or 40 nm oxide-covered titanium. The thicker the oxide, the higher the bias voltage needed to establish InA. If the initial bias was set to -IV, it was possible to obtain a stable tunneling current of -InA only on the 5 nm oxide, but not on thicker oxides.

0 0

Post a comment