Consequences of zero-point motion to the radial distribution function of amorphoussilicon

Abstract

While there have been many studies based on models of amorphous silicon, there have beensurprisingly few (perhaps only one) that have seriously addressed the radial distributionfunction at low temperature. Our work is based in part on the so-called NRLtight binding method using parameters for silicon determined by Bernstein et al. As we have recently shown in the case of 216-atom models, upon includingzero-point motion good agreement is obtained with very accurate low temperaturex-ray diffraction measurements by Laaziri et al of the radial distribution function,although, as also found by Herrero who used the Stillinger–Weber potential, a slightasymmetry of the first peak in the RDF is predicted and this asymmetry has not beenobserved experimentally. Upon use of an estimate of zero-point broadening from ourprevious work we show here that 1000-atom models lead to good agreement withexperiment for the RDF. Perhaps fortuitously, we obtain models that agree withthe experimentally determined second peak in the RDF for both annealed andunannealed samples: our tight binding relaxed models based on topologies derived fromthe Wooten–Winer–Weaire method and the Barkema–Mousseau method yieldunannealed-sample results, whereas our tight binding relaxed model based on an MDquench of the liquid using the semi-empirical interatomic potential, EDIP, of Kaxiras andcoworkers yield the annealed-sample results. Finally, the significant effect of zero-pointmotion on the first peak in the radial distribution that we obtain in the case ofamorphous silicon could also have implications for other amorphous materials, e.g.SiO2.