A new self-consistent empirical potential model for multicomponent borate and borosilicate glasses

A new self-consistent empirical potential model based on the BMP potential [Bertani et al., Phys. Rev. Mat. 5 (2021) 045602] has been developed for the simulation of multicomponent borate and borosilicate glasses. We exploited the Bayesian optimization approach to create a set of parameters for the B-O interaction, which depends on the glass composition, and in particular on the R = ([A(2)O]+[AEO])/[B2O3] (A = alkaline and AE alkaline-earth ions) and K = [SiO2]/[B2O3] ratios. The obtained force field (FF) has been applied to several borate and borosilicate glass series containing, as modifier oxide, Na2O, Li2O, CaO, and MgO and tested on experimental data, such as the fraction of BO4 (N-4), density, non-bridging oxygen speciation, neutron diffraction spectra, B-11, Si-29, and O-17 magic angle spinning nuclear magnetic resonance. A comparison with other interatomic potentials available in literature has also been performed. The results show that the FF reproduces well almost all the abovementioned properties, showing excellent agreement with experimental data in a wide range of compositions.