Machine learning modeling of the atomic structure and physical properties of alkali and alkaline-earth aluminosilicate glasses and melts

Abstract

The first version of the machine learning greybox model i-Melt was trained to predict latent and observed properties of K2O-Na2O-Al2O3-SiO2 melts and glasses. Here, we extend the model compositional range, which now allows accurate predictions of properties for glass-forming melts in the CaO-MgO-K2O-Na2O-Al2O3-SiO2 system, including melt viscosity (accuracy equal or better than 0.4 log10 Pa⋅s in the 10−1-1015 log10 Pa⋅s range), configurational entropy at glass transition (≤ 1 J mol−1 K−1), liquidus (≤ 60 K) and glass transition (≤ 12 K) temperatures, heat capacity (≤ 3 %) as well as glass density (≤ 0.02 g cm−3), optical refractive index (≤ 0.006), Abbe number (≤ 4), elastic modulus (≤ 6 GPa), coefficient of thermal expansion (≤ 1.1 10−6 K−1) and Raman spectra (≤ 25 %). Uncertainties on predictions also are now provided. The model offers new possibilities to explore how melt/glass properties change with composition and atomic structure.