Structural and electrocatalytic properties of copper clusters: A study via deep learning and first principles

Abstract

Determining the atomic structure of clusters has been a long-term challenge in theoretical calculations due to the high computational cost of density-functional theory (DFT). Deep learning potential (DP), as an alternative way, has been demonstrated to be able to conduct cluster simulations with close-to DFT accuracy but at a much lower computational cost. In this work, we update 34 structures of the 41 Cu clusters with atomic numbers ranging from 10 to 50 by combining global optimization and the DP model. The calculations show that the configuration of small Cu n clusters ( n = 10–15) tends to be oblate and it gradually transforms into a cage-like configuration as the size increases ( n > 15). Based on the updated structures, their relative stability and electronic properties are extensively studied. In addition, we select three different clusters (Cu13, Cu38, and Cu49) to study their electrocatalytic ability of CO2 reduction. The simulation indicates that the main product is CO for these three clusters, while the selectivity of hydrocarbons is inhibited. This work is expected to clarify the ground-state structures and fundamental properties of Cu n clusters, and to guide experiments for the design of Cu-based catalysts.