In earlier work, WPI-ICReDD scientists showed that a palladium catalyst paired with phosphine ligands could drive light-activated (reaction activated by shining light) transformations of aryl ketones, but the same system did not work for alkyl ketones. Their data indicated that alkyl ketyl radicals did form briefly. However, these radicals immediately returned an electron to the palladium center, a phenomenon known as back electron transfer (BET), before any useful reaction could proceed. As a result, the starting material remained unchanged.

Similar to traditional palladium-based catalysis, the behavior of photoexcited palladium catalysts is highly dependent on the phosphine ligand attached to the metal. The team suspected that choosing the correct ligand might unlock reactivity with alkyl ketones. The difficulty was scale: thousands of phosphine ligands exist, and experimentally screening them for an unfamiliar reaction would be slow, labor-intensive, and generate unnecessary chemical waste.

To overcome these limitations, the researchers turned to computational chemistry to narrow down the field of candidate ligands. They used the Virtual Ligand-Assisted Screening (VLAS) approach developed by Associate Professor Wataru Matsuoka and Professor Satoshi Maeda at WPI-ICReDD. Applying VLAS to 38 phosphine ligands, the method produced a heat map that predicted how well each ligand might promote the desired reactivity by analyzing electronic and steric properties.

Guided by these predictions, the team selected three ligands for laboratory testing and ultimately identified L4 as the most effective option -- tris(4-methoxyphenyl)phosphine (P(p-OMe-C₆H₄)₃). This ligand successfully suppressed BET, allowing alkyl ketones to generate ketyl radicals and participate in high-yield transformations.

The resulting method provides chemists with an accessible way to work with alkyl ketyl radicals and demonstrates how VLAS can rapidly guide the development and optimization of new chemical reactions.