Data-Driven Prediction of Flory–Huggins Parameter for Quantifying Polymer–Solvent Interaction

Abstract

The Flory–Huggins interaction parameter is an influential thermodynamic parameter for quantifying polymer–solvent interaction. The data-driven quantitative structure–property relationship (QSPR) model provides a rapid and accurate route to obtain the Flory–Huggins parameter of polymer–solvent mixtures. To evaluate the correlation between the polymer–solvent structure and the Flory–Huggins parameter, we developed a QSPR model based on 29 norm descriptors. To overcome the diversity of structural representations of a polymer, a ring repeating unit is adopted for unique structural representation of the polymers. Considering the effects of polymers, solvents, and polymer–solvent interactions on the Flory–Huggins parameters, the descriptors are divided into three parts. The results of statistical parameter analysis indicate that the QSPR model exhibits good prediction performance (R²_test = 0.9348, AAE_test = 0.1943) and robustness (Q²_LOO-CV = 0.9177). This work offers a quantitative reference and a tool to understand polymer–solvent interactions.