Harnessing Electro-Descriptors for Mechanistic and Machine Learning Analysis of Photocatalytic Organic Reactions

Abstract

Photocatalysis has emerged as an effective tool for addressing the contemporary challenges in organic synthesis. However, the trial-and-error-based screening of feasible substrates and optimal reaction conditions remains time-consuming and potentially expensive in industrial practice. Here, we demonstrate an electrochemical-based data-acquisition approach that derives a simple set of redox-relevant electro-descriptors for effective mechanistic analysis and performance evaluation through machine learning (ML) in photocatalytic synthesis. These electro-descriptors correlate to the quantification of shifted charge transfer processes in response to the photoirradiation and enabled construction of reactivity diagram where high-yield reactive "hot zones" can reflect subtle changes of the reaction system. For the model reaction of photocatalytic deoxygenation reaction, the influence of varying carboxylic acids (substrate A, oxidation-intended) and alkenes (substrate B, reduction-intended) and varying reaction conditions on the reaction yield can be visualized, while mathematical analysis of the electro-descriptor patterns further revealed distinct mechanistic/kinetic impacts from different substrates and conditions. Additionally, in the application of ML algorithms, the experimentally derived electro-descriptors reflect an overall redox kinetic outcome contributed from vast reaction parameters, serving as a capable means to reduce the dimensionality in the case of complex multiparameter chemical space. As a result, utilization of electro-descriptors enabled efficient and robust quantitative evaluation of chemical reactivity, demonstrating promising potential of introducing operando-relevant experimental insights in the data-driven chemistry.