Deposition of microcrystalline silicon: Direct evidence for hydrogen-induced surface mobility of Si adspecies

Hydrogenated microcrystalline silicon thin films can be deposited at low substrate temperatures using plasma enhanced-or hot wire-chemical vapor deposition using silane, reactive magnetron sputtering of silicon, or related techniques. Microcrystalline silicon is deposited when a large quantity of molecular hydrogen is added to the process gas such that a large flux of atomic hydrogen impinges on the film growth surface; otherwise, the films are amorphous. Three different microscopic mechanisms have been hypothesized to explain the formation of the microcrystalline phase. In essence, the hypotheses are that atomic hydrogen: (i) enhances the surface diffusion of Si adspecies, which in turn raises the probability of crystalline phase formation, (ii) promotes a subsurface transformation of amorphous into microcrystalline Si, or (iii) preferentially etches amorphous regions such that only microcrystalline Si survives to produce film growth. In this work, we critically test mechanism (i) as follows. We deposit films using dc reactive magnetron sputtering of a Si target in an argon-hydrogen plasma, which yields very poor adspecies mobility at low rates of hydrogen injection. We then increase the hydrogen injection and measure the increase in adspecies motion via the enhanced rate at which the surface smoothens for film growth on substrates with a calibrated roughness of ∼80 Å. The dynamic surface roughness and the structural phase are determined by real-time spectroscopic ellipsometry. The combination of high atomic hydrogen flux and high surface hydrogen coverage uniquely correlates with microcrystalline phase formation. Higher substrate temperatures do not increase adspecies mobility, and actually decrease it when the rate of thermal desorption becomes sufficient to decrease the surface hydrogen coverage. These results also suggest that the original identity of the Si-bearing growth species is relatively unimportant, because the atomic hydrogen flux appears to produce mobile adspecies via surface reactions. We have previously shown that subsurface transformations, mechanism (ii), can also occur. However, we find no evidence for competitive etching, mechanism (iii), under our experimental conditions.