Predicted Structures of the Active Sites Responsible for the Improved Reduction of Carbon Dioxide by Gold Nanoparticles

Abstract

Gold (Au) nanoparticles (NPs) are known experimentally to reduce carbon dioxide (CO_2) to carbon monoxide (CO), with far superior performance to Au foils. To obtain guidance in designing improved CO_2 catalysts, we want to understand the nature of the active sites on Au NPs. Here, we employed multiscale atomistic simulations to computationally synthesize and characterize a 10 nm thick Au NP on a carbon nanotube (CNT) support, and then we located active sites from quantum mechanics (QM) calculations on 269 randomly selected sites. The standard scaling relation is that the formation energy of *COOH (ΔE_(*COOH)) is proportional to the binding energy of *CO (E^(binding)_(*CO)); therefore, decreasing ΔE_(*COOH) to boost the CO_2 reduction reaction (CO_2RR) causes an increase of E^(binding)_(*CO) that retards CO_2RR. We show that the NPs have superior CO_2RR because there are many sites at the twin boundaries that significantly break this scaling relation.