Virtual Ligand-Assisted Optimization: A Rational Strategy for Ligand Engineering

Abstract

Ligand engineering is one of the most important, but labor-intensive processes in the development of transition metal catalysis. Historically, this process has been guided by ligand descriptors such as Tolman’s electronic parameter and the cone angle. Analyzing reaction outcomes in terms of these parameters has enabled chemists to identify the most important properties for controlling catalytic pathways and thus designing better ligands. However, typical strategies for these analyses rely on regression approaches, which often require extensive experimental studies to identify trends across chemical space and understand outliers. Here, we introduce the virtual ligand-assisted optimization (VLAO) method, a computational approach for reactivity-directed ligand engineering. In this method, important features of ligands are identified by simple mathematical operations on equilibrium structures and/or transition states of interest, and derivative values of arbitrary objective functions with respect to ligand parameters are obtained. These derivative values are then used as a guiding principle to optimize ligands within the parameter space. The VLAO method was demonstrated in the optimization of monodentate and bidentate phosphine ligands including asymmetric quinoxaline-based ligands. In addition, we successfully found an optimal ligand for the α-selective hydrogermylation of a terminal ynamide, applying the design principle suggested by the VLAO method. These results highlight the practical utility of the VLAO method, with the potential for directed optimization of a wide variety of ligands for transition metal catalysis.