Generation of library of dyes for photovoltaic applications: A machine learning assisted framework

Abstract

The generation of a comprehensive library of dyes and the prediction of their properties is a critical task in the development of novel materials for various applications. This paper presents a machine learning-assisted framework designed to streamline this process by using the power of machine learning. The ultraviolet–visible (UV–Vis) absorption maxima of dyes are predicted using multiple models, Random Forest model has emerged as the most effective. A library of 10,000 dyes is generated and evaluated based on the predicted absorption maxima values, resulting in the selection of 30 promising dyes. An assessment of synthetic accessibility has shown that most of the selected dyes can be synthesized with relative ease. Our proposed farmwork has potential to reduce the need for extensive experimental synthesis and testing.