CO2 Electrocatalyst Design Using Graph Theory

Abstract The electrochemical reduction of CO2 (CO2RR) to higher-order hydrocarbons or oxygenates using low-carbon electricity offers a promising path to generate renewable fuels and chemicals; however, the low selectivity of present-day CO2RR catalysts toward more valuable C3 products limits technoeconomically compelling avenues toward productization. Systematically enumerating possible intermediates and reactions to the set of possible {C1, C2, and C3} CO2RR products entails 3206 intermediates and 4506 reactions. Here we use graph theory to enumerate these possibilities comprehensively, treating intermediates as graph nodes and pathways as graph edges. C3 products fall into two groups, group A (1-propanol, allyl alcohol, and propionaldehyde) and group B (acetone and hydroxyacetone); and we find that an early branch reaction, three steps after C1-C2 coupling, bifurcates these two groups of C3 products. We develop a set of C3 descriptors and screen catalysts for CO2RR to C3 products: specifically, we screen bimetallic (doped Cu) catalysts for their combination of CO binding, C1-C1 coupling and C1-C2 coupling. Cu and also Au- and Ag-doped Cu fulfill the first two requirements; but the set of promising C1-C2 coupling catalysts (Ni-doped Pb and Al-doped Pb) needed to get to C3 is nonoverlapping with that for CO binding and C1-C1 coupling. Our findings agree with the experimental picture that Cu, while among the most productive to propanol, has been limited in F.E. to the 10-20% range; and that, to date, no single catalyst has achieved exceptional C3 productivity. We discuss tandem catalyst designs where a first catalyst promotes CO binding and C1-C1, and where these C2 and C1 intermediates can come together for coupling on a second distinct class of integrated catalysts.