Rapid Screening of Amorphous Fluorinated Polymer Electret Based on Machine Learning

In the present study, a rapid screening method of amorphous fluorinated polymer based on graph network is investigated for development of a novel electret material with high charging performance. Solid-state ionization potential (IP) of amorphous polymer is chosen as the index of charge stability, and determined by the density functional theory (DFT) in conjunction with polarizable continuum model (PCM). Surface charge density of CYTOP (Cyclic Transparent Optical Polymer) polymer with different end groups are measured both for positive and negative charging. The order of solid-state IP estimated with the present PCM/DFT is in quantitative accordance with the order of experimental data, indicating the present simulation gives appropriate quantitative estimation on the charge trapping performance of CYTOP electrets. In order to significantly reduce the heavy computational cost required by PCM/DFT and to develop a rapid screening method toward optimization of the end group for better charging performance, MatErial Graph Network (MEGNet) model is also employed for predicting the solid-state IP based on molecular structures.