Design of Polymeric Nickel Catalysts for Suzuki–Miyaura Type Cross–Coupling Reaction Using Machine Learning

Suzuki–Miyaura cross-coupling reactions are extensively employed in sectors such as pharmaceuticals. Traditionally, palladium catalysts have been used for this reaction; however, these cause an increased depletion of precious metals and have adverse environmental effects. Consequently, polymeric Ni catalysts─which are heterogeneous catalysts─are attracting attention as an alternative. In this study, we build a yield prediction model for the Suzuki–Miyaura type cross-coupling reaction using polymeric Ni catalysts and developed highly active polymeric Ni catalysts with various structures. This model predicted yield values for both homogeneous and heterogeneous catalysts─which included experimental samples as well as polymeric Ni catalysts developed using this model─with high accuracy. Bayesian optimization was used to determine experimental conditions that achieve high yields, even at low additive loadings. The proposed method facilitates the design of polymeric Ni catalysts with diverse, previously unexplored structures and the prediction of their corresponding yields, which is expected to result in a yield improvement.