Cost-Aware\nBayesian Optimization for Self-Driven Condition\nScreening of Flow Electrosynthesis

Flow electrosynthesis has attracted increasing attention\nas a green\nand sustainable manufacturing method. However, it is still a challenging\nundertaking to determine the appropriate experimental conditions for\ndifferent electrosynthesis systems. As an AI-guided decision-making\nmethod, Bayesian optimization (BO) suffers a lot from neglecting the\ncharacteristics of chemical reactions and one-by-one low-speed screening\nprocesses when it is used to identify the optimal conditions. Herein,\nwe design a cost-aware multi-objective Bayesian optimization framework\nto seek the highest current density for the electrosynthesis of tetrabenzylthiuram\ndisulfide (TBzTD), while keeping the yield and Faraday efficiency\nat high levels. The optimization was finished within only 4 h including\n11 experiments, and the optimized current density became 200% as high\nas the maximum of trial-and-error experiments, which efficiently reduced\nthe experimental cost compared to the standard BO that requires a\nfew days. The approach shows the possibility of utilizing the least\nexperimental budgets for fully optimal condition screening, significantly\nreducing the burden on trial-and-error experimental expenses.