Hydrogen diffusion in amorphous silicon

Abstract

Hydrogen diffusion in doped and compensated a-Si:H has been measured by secondary-ion mass spectrometry profiling in the temperature range 155-300°C. Doping reduces the activation energy and enhances the diffusion coefficient by up to three orders of magnitude at 200°C, and a correlation between the diffusion coefficient and the dangling-bond density is found. An analysis of three different diffusion models indicates that the breaking of weak Si – Si bonds by hydrogen may be an important process. The relation between the diffusion results and the thermal equilibration of the electronic structure is discussed.

The hydrogen diffusion coefficient in a-Si:H has been measured over the temperature range 155-300°C, with particular emphasis on the effects of doping and compensation. In all cases D _H is thermally activated with an energy 1⋅2-1⋅5 eV. The diffusion coefficient decreases slowly with time which is attributed to the disorder-induced variation in site energies. We find that D _H is greatly enhanced by doping, and its magnitude correlates with the dangling-bond density in the samples. At a doping level of 10⁻² B₂H₆ in SiH₄, D_H is about 10³ times larger than in undoped a-Si:H. The diffusion coefficient is very low in compensated material, again showing the correlation with the dangling-bond density. We have also measured the boron diffusion and found that it does not follow the same trends as the hydrogen.

In order to explore the diffusion mechanisms, we have analysed three models. A model in which hydrogen can only move if there is a neighbouring dangling bond to receive it is able to explain the data but is doubtful because of the large distance over which the Si–H interaction must occur. A second model in which hydrogen is released into an interstitial site where it diffuses until it is captured by a dangling bond seems to be unable to account for the data. However, a third model in which interstitial hydrogen breaks weak Si–Si bonds results in agreement with experiment. This latter model is attractive because it can be related both to the process of thermal equilibration and to the Staebler-Wronski effect. However, a much more detailed understanding of the atomic bonding is needed before the diffusion mechanism can be conclusively determined. Irrespective of the details of the mechanism, we argue that the observed activation energies are consistent with the breaking of Si–H bonds. Furthermore, the energy will be influenced by the presence of band-tail electrons or holes in doped a-Si:H. Thus we anticipate that a full analysis of diffusion will include electronic transitions.