Presentation Information
[9a-N304-9]Deep Potential molecular dynamics analysis of surface enrichment of boroxol rings in molten B2O3
〇Jun Otsuka1 (1.Sumitomo Electric Industries, Ltd.)
Keywords:
B2O3 glass surface,machine-learning potential,first-principles calculation
We investigated the surface structure of B2O3 glass and melt by molecular dynamics simulations using a machine-learning potential trained on first-principles DFT data. The potential was developed through active learning for both bulk and slab configurations, enabling accurate large-scale simulations of free surfaces. Analysis of 240-atom slabs over 300–1300 K showed that the boroxol-ring fraction in the bulk decreases monotonically with increasing temperature, consistent with previous diffraction studies. In contrast, boroxol rings remain enriched at the surface even at high temperatures. In particular, a boroxol-rich surface termination exhibits a ring density higher than that in the bulk at all temperatures, reaching about 2.5 times the bulk value at 300–800 K. This surface enrichment is associated with BO2 dangling bonds and coplanar arrangements of large 8–10-membered rings. Dynamical analysis further revealed that boroxol rings, which are quasi-static at low temperature, become highly dynamic above 1100 K and repeatedly form and dissociate on the picosecond timescale.