Automated Microdroplet Reaction Platform for Condition Optimization and Kinetic Investigation of Photocatalytic Benzylic Oxidations

Abstract

Photocatalysis offers a unique redox platform for designing selective transformations under mild and sustainable conditions, yet optimizing reaction conditions and identifying efficient photocatalysts require extensive screening and mechanistic studies. Here, we report an automated multi-channel photochemical platform that integrates oscillatory droplet flow, dynamic task scheduling, and multi-objective Bayesian optimization to accelerate the exploration and kinetic investigation of photocatalytic benzylic oxidation. This platform operates with 30 μL reaction droplets with four-channel parallelization, enabling material- and time-efficient reaction screening. Complete evaluation of 24 aromatic ketone photocatalysts could be finished within one day. In addition, the flexible gas pressure control (0–40 psig) facilitated kinetic studies of the gas–liquid anthraquinone-catalyzed photocatalytic benzylic oxidation system, revealing a dual mechanistic pathway involving hydrogen atom transfer and singlet oxygen-mediated oxidation. The multi-objective Bayesian optimization with Pareto-front mapping identified 2-tert-butylanthraquinone as the most potent photocatalyst to balance reaction yields and productivity simultaneously. The screening outcome was transferred to photocatalytic upcycling of polystyrene to synthesize benzoic acid by activating the benzylic C–H bonds in the polymer backbone.