Optimizing telescoped heterogeneous catalysis with noise-resilient multi-objective Bayesian optimization

Abstract

This study evaluates the noise resilience of multi-objective Bayesian optimization (MOBO) algorithms in chemical synthesis, an aspect critical for processes like telescoped reactions and heterogeneous catalysis but seldom systematically assessed. Through simulation experiments on amidation, acylation, and S_NAr reactions under varying noise levels, we identify the qNEHVI acquisition function as notably proficient in handling noise. Subsequently, qNEHVI is employed to optimize a two-step heterogeneous catalysis for the continuous-flow synthesis of hexafluoroisopropanol. Remarkable optimization is achieved within just 29 experimental runs, resulting in an E-factor of 0.125 and a yield of 93.1%. The optimal conditions are established at 5.0 sccm and 120 °C for the first step, and 94.0 sccm and 170 °C for the second step. This research highlights qNEHVI’s potential in noisy multi-objective optimization and its practical utility in refining complex synthesis processes.