Abstract

The prediction of a specific chemical property across a vast library of derivatives represents a formidable challenge. Conventional computational methodologies typically rely on brute-force calculations involving the computation of the property of interest for the entire library or a significant subset. In this study, we present a novel phenomenological approach to address this challenge, employing a perturbation theory-like framework to describe substituent effects. This proposed methodology has the potential to forecast the molecular properties of millions of compounds based on information derived from just a few hundred. This method is applied to the design of molecular solar thermal (MOST) systems, which are devices permitting harvesting solar energy and storing it in a chemical form. The optimization of MOST performance is a critical issue in practical applications of this technology, so exploration of large libraries of derivatives at low computational cost is an interesting approach to tackle the problem. To accomplish this objective, we explore the functionalization of the norbornadiene/quadricyclane (NBD/QC) system utilizing the proposed perturbational approach predicting the energy of 350 derivatives from small sets of 5 and 50 calculated compounds.