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

Area of Science:

• Computational Chemistry
• Materials Science
• Catalysis

Background:

• Determining atomic structures of clusters is computationally expensive using traditional methods like density-functional theory (DFT).
• Deep learning potentials (DP) offer a computationally efficient alternative with near-DFT accuracy for cluster simulations.

Purpose of the Study:

• To update atomic structures for 41 copper (Cu) clusters (n=10-50) using a combination of global optimization and DP.
• To investigate the relative stability, electronic properties, and CO2 electrocatalytic reduction of these Cu clusters.

Main Methods:

• Employed a deep learning potential model integrated with global optimization techniques.
• Calculated and analyzed the atomic structures, stability, and electronic properties of Cu clusters.
• Simulated the electrocatalytic CO2 reduction for selected Cu clusters (Cu13, Cu38, Cu49).

Main Results:

• Updated 34 ground-state structures for Cu clusters (n=10-50).
• Observed a transition from oblate to cage-like configurations as cluster size increases (n=10-15 to n>15).
• Simulations of CO2 reduction showed CO as the primary product, with inhibited hydrocarbon selectivity.

Conclusions:

• The study provides accurate ground-state structures and fundamental properties for Cu clusters.
• Findings are expected to guide experimental design for developing efficient Cu-based catalysts for CO2 reduction.