Segment Descriptor Enabling Prediction of Electronic Properties and Photocatalytic Hydrogen Evolution Rate of Alternating Conjugated Copolymers Based on Machine Learning

Alternating conjugated copolymers have been regarded as promising candidates for photocatalytic hydrogen evolution due to the adjustability of their molecular structures and electronic properties. In this work, machine learning (ML) models with molecular fingerprint of segment descriptors (SD) have been successfully constructed to promote the accurate and universal prediction of electronic properties such as electron affinity, ionization potential and optical bandgap. Moreover, without any experimental values, a high-performance prediction classifier model toward photocatalytic hydrogen production of alternating copolymers has been developed with high accuracy (real-test accuracy = 0.91). Consequently, our results demonstrate accurate regression and classification models to disclose valuable influencing factors concerning hydrogen evolution rate (HER) of alternating copolymers.