Machine learning-assisted chemical space generation of small molecule organic semiconductors for efficient photodetectors

Abstract

Organic photodetectors (OPDs), the emerging next-generation candidates for light sensing applications, could compensate deficiencies of the traditional inorganic photodetectors. To get OPDs of highly improved performance, and broader applications, machine learning (ML) technique is employed. A chemical space of 20,000 organic semiconductors (OSCs) by RDKit via BRICKS (Breaking Retrosynthetically Interesting Chemical Substructures) method was achieved, from the selected building blocks of the OSC data obtained from the literature, contributing to big data. Out of these, top twenty OSCs were screened based on synthetic accessibility score, chemical similarity analysis, and structural behavior understanding.