Integrated Computational and Experimental Approach to the Generation of Ketyl Radicals from Alkyl Ketones

The catalytic generation of alkyl ketyl radicals remains challenging due to the very high reduction potentials of alkyl ketones, which impedes their reduction under catalytic conditions. We first employed a photoexcited palladium species with a sufficiently high reduction potential to enable the single-electron reduction of alkyl ketones. Mechanistic studies revealed that back-electron transfer (BET) competes with the productive radical-coupling pathway, providing a key design principle for ligand optimization. This insight motivated us to use a virtual ligand-assisted screening (VLAS) to computationally map the electronic and steric ligand space required to suppress BET. Among the 38 computationally evaluated candidates, VLAS predicted a narrow region capable of minimizing BET and promoting a productive ketyl radical coupling. Experimental evaluation of only four ligands validated these predictions and identified tri(pmethoxyphenyl)phosphine as the optimal ligand. This theory-first strategy enables the general catalytic generation of alkyl ketyl radicals and establishes a predictive framework for computational ligand design using VLAS.