Molecular dynamics simulations of sodium silicate glasses: Optimization and limits of the computational procedure

• Published in: Computational materials science Vol. 47; no. 3; pp. 739 - 751
• Main Authors: Pota, M., Pedone, A., Malavasi, G., Durante, C., Cocchi, M., Menziani, M.C.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 2010; Elsevier
• Subjects: Alkali silica glasses; Condensed matter: structure, mechanical and thermal properties; Design of experiments; Disordered solids; Exact sciences and technology; Glasses; Molecular dynamics simulations; Physics; Structure of solids and liquids; crystallography; 81.05.Kf; 02.70.Ns; 07.05.Fb; Alkali silica glasses; Molecular dynamics simulations; 61.43.Bn; Design of experiments; 31.15.xv; 02.50.Sk; Cooling; Bond angle; Molecular dynamics method; Optimization method; Medium range order; Binary systems; Glass structure; Regression analysis; Silicates; Soda-lime glasses; Design; Radial distribution function; Short-range order; Sodium oxide; Chemical composition
• ISSN: 0927-0256; 1879-0801
• DOI: 10.1016/j.commatsci.2009.10.017

The performance of the molecular dynamics (MD) simulations to obtain the structure of silica glasses containing different concentrations of alkali oxides has been tested. An optimal MD simulation procedure (including cooling cycle, MD constants and ensemble used) has been developed by means of experimental design methodologies (DOE), firstly restricting the study to the 30% Na 2O silica glass, for which experimental data are available to allow the comparison of the results. The optimization procedure led to simulations that well predict experimental density and short-range structure of glasses with different sodium content. On the contrary, the medium-range structure has been badly reproduced and it was not possible to determine a reliable correlation with the parameters of the simulation procedures used. Therefore, the correlation of medium-range properties with the structure has been studied by means of the PLS methods. The results showed to be useful to highlight the relationships among structural elements, such as radial distribution functions of specific bonds and angles, and Q n species, suggesting possible directions for force fields improvements.