Abstract:
A thermal procedure and an ab initio molecular-dynamics method based on the Harris functional, applied to originally crystalline, periodically continued 64-atom cubic cells, is used to generate random networks of four different materials of varying degrees of covalency: C, Si, Ge, and a nearly stoichiometric sample of Si-N. We obtain their radial distribution functions (RDF's) for four different time steps, one for each material, using densities dictated by experiment. The simulated RDF's for amorphous C, Si, and Ge show the four characteristic radial peaks observed experimentally. For the nearly stoichiometric SiN1.29 sample two runs were performed and averaged. The agreement between simulated and experimental RDF's is very good.