Machine learning glass transition temperature of polyacrylamides using quantum chemical descriptors

Abstract

Glass transition temperature, T_g, is an important thermophysical property of polyacrylamides, which can be difficult to determine experimentally and resource-intensive to calculate. Data-driven modeling approaches provide alternative methods to predict T_g in a rapid and robust way. We develop the Gaussian process regression model to predict the glass transition temperature of polyacrylamides based on quantum chemical descriptors. The modeling approach shows a high degree of stability and accuracy, which contributes to fast and low-cost glass transition temperature estimations.