[21p-B103-1~8] State of the art of glass materials for solid electrolytes ~2022 International Glass Year IYOG Memorial Symposium~

[21p-B103-6] Atomistic simulations for modeling lithium borosilicate glasses

〇Shingo Urata¹ (1.AGC Inc.)

Keywords:molecular simulation, glass, machine learning

Lithium-containing glasses are expected to decrease interfacial resistance between electrolyte and electrode of all-solid-state batteries. However, flexible structural change of boron from threefold to fourfold-coordinated complicates computational simulations on the structural and dynamical properties of lithium borosilicate (LBS) glasses. Specifically, three-membered rings, namely boroxol ring, is difficult to be modeled using molecular dynamics (MD) simulations with an empirical force-field. Therefore, we tried to model LBS glasses using (1) ab initio MD simulations, (2) classical MD simulations with a composition-dependent force-field, and (3) MD simulations enriched by machine-learning potential. In this talk, comparisons among the three modeling methods will be discussed.