On the boroxol ring fraction in melt-quenched B2O3 glass: Insights from machine learning potentials.

An atomistic structural model for melt-quenched B2O3 glass has eluded the simulation community so far. The difficulty lies in the abundance of six-membered boroxol rings-an intermediate-range order motif suggested by Raman and NMR spectroscopy-which is challenging to capture in atomistic molecular dynamics simulations. Here, we report the development of a density functional theory-accurate machine-learned potential and employ quench rates as low as 109 K/s to obtain B2O3 glasses with more than 30% of boron atoms in boroxol rings. Additionally, we show that the pressure, and consequently the boroxol fraction, in the deep potential molecular dynamics simulations critically depends on the range of the geometry descriptor used in the embedding neural network, and it converges beyond a value of 7 Å. The boroxol ring fraction increases with decreasing quench rate. Finally, amorphous B2O3 configurations display a minimum in energy at a boroxol fraction of 75%, remarkably close to the experimental estimate in B2O3 glass.