High-Throughput Optimization of a High-Pressure Catalytic Reaction

摘要

High-throughput optimization of a hydroformylation reaction using CO2 instead of CO was performed through Bayesian optimization in combination with a high-throughput screening system. CO2 and H2 pressure as well as catalyst composition were efficiently optimized by transferring a surrogate model, constructed through catalyst composition optimization, for the comprehensive optimization of the entire search space. This method successfully increased the aldehyde yield by 1.5 times compared to that reported in the literature with a combination of small amounts of Rh and Ru catalysts combined with ionic liquid with chloride ions. The optimization was completed within 1–2 months through the combination of AI, robotics, and human expertise, demonstrating the feasibility of rapid catalyst development, even for high-pressure reactions.