Materials genes of heterogeneous catalysis from clean experiments and artificial intelligence

Heterogeneous catalysis is an example of a complex materials function governed by an intricate interplay of several processes. While modeling the full catalytic progression via first-principles statistical mechanics is impractical, we show how a tailored artificial-intelligence approach can be applied, even to a small number of materials, to determine the key descriptive parameters or materials genes reflecting the processes that trigger, facilitate, or hinder catalyst performance. We start from a consistent experimental set of clean data,[1] containing 12 selective-oxidation catalysts which were synthesized, fully characterized, and tested according to standardized protocols. Then, we apply the symbolic-regression SISSO approach[2] to identify the few most relevant materials properties that correlate, in a possibly nonlinear way, with the reactivity.[3]