Classical molecular dynamics simulations are widely used to facilitate the modeling and design of silicate glasses. To this end, several empirical have been developed—each of them focusing on distinct features of silicate glasses (e.g., structure, mechanical properties, etc.). Available forcefields relying on fixed partial charges can be divided into “soft” and “hard” depending on the partial charge attributed to Si atoms (around 2.0 and 2.4, respectively). Here, we use machine learning to explore the “landscape of silica forcefields,” that is, the evolution of the overall forcefield accuracy as a function of the forcefield parameters. Interestingly, we show that soft and hard forcefields correspond to two distinct, competitive local minima in this landscape. By analyzing the structure of the silica glasses predicted by soft and strong forcefields, we show that these two families of potential yield a distinct medium-range order structure. This work provides a theoretical foundation for the rational modeling of ionocovalent glasses.
Han Liu*, Zipeng Fu, and Mathieu Bauchy, University of California, Los Angeles