An efficient computational procedure to obtain a more stable glass structure

A huge gap in time between the experiment and the atomistic simulation restricts us from accessing a realistic glass structure, because the glass state is highly dependent on the cooling rate. In this study, to improve computational efficiency, we propose a simple but effective procedure, which enables us to explore a deeper basin in the energy landscape of glassy materials without a substantial increase in the computational cost. This method combines canonical ensemble molecular dynamics (MD) and energy minimization while controlling the stress of the MD system, and it is called the quasi-slow-quenching (QSQ) method. Herein, we measured the performance of the QSQ method using a binary silicate, (SiO₂)₈₀(Na₂O)₂₀, and we observed that a more stable configuration can be obtained in comparison with the conventional isobaric-isothermal MD method. The stable glass model appears to possess a lower glass transition temperature (T_g), confirming that the QSQ method finds a deeper local minimum closer to the super-cooled glass state. We also conducted further validation tests for various oxide glasses, including silicate, borate, phosphate, and their mixtures, and we verified that the QSQ method consistently enables the glassy materials to attain energetically more stable configurations and denser structures.