Understanding Composition-Structure-Property Relationships in Multicomponent Silicate Glasses

Multicomponent silicate glasses underpin both commodity and specialty glass industries, yet their compositions have traditionally been developed through costly and time-intensive trial-and-error approaches. Advances in computational materials science now enable the integration of experimental measurements with molecular dynamics simulations and artificial intelligence to establish predictive quantitative structure–property relationship (QSPR) models. Such models rely on high-quality experimental datasets that directly link molecular-level structural descriptors—such as silicate and borate speciation—to measurable physical properties. In this work, we measure key physical properties, including density, coefficient of thermal expansion, refractive index, and elastic moduli, for more than 50 multicomponent silicate glasses and correlate them with structural information obtained from MAS NMR spectroscopy. These data provide a foundation for integrating experiments with molecular dynamics and machine-learning approaches toward the rational design of glass compositions.