Modeling of continuous random networks: a case study for vitreous GeO2. II. Topological analysis and refinement

In this paper we present a new method to compare the topological structure of two continuous random network (CRN) models in a quantitative manner. The comparison serves to assess the sampling properties of the molecular model ing program TUMBLEWEED. This program generates spherical CRN clusters and h as been introduced in paper I of this series. The structural diversity of t he models constructed with this program is found to be satisfactory. Furthe r, the models are very different from the alpha-cristobalite and alpha-quar tz structures of GeO2. In addition, we investigate three different techniqu es for the refinement of the clusters presented in paper I: the conjugate g radient (CG) method, the Monte Carlo (MC) method, and the Reverse Monte Car lo (RMC) method. The former two procedures minimize the potential energy, V , whereas the latter minimizes the deviation, R-x, of the model neutron tot al correlation function, T(r), from corresponding experimental data. It is found that neither technique gives satisfactory results by itself, since th e minimization of R-x leads to an increase of V (including the violation of geometric constraints) and vice versa. By combining the MC and RMC methods , i.e. by simultaneous minimization of V and R-x, we obtain models with rea listic geometric properties and an improved fit to the experimental data (R -x = 0.018 +/- 0.003 in the range 0-1 nn). This further supports the validi ty of the CRN approach to the structure of glassy GeO2. New models with a n on-uniform torsion angle distribution have been constructed. It turns out t hat such models cannot be excluded as possible structures (final R-x = 0.02 0 +/- 0.006). (C) 1999 Elsevier Science B.V. All rights reserved.