Exploring the stable structures of cerium oxide nanoclusters using high-dimensional neural network potential

Area of Science:

• Materials Science
• Computational Chemistry
• Nanotechnology

Background:

• Cerium oxide clusters are vital industrial catalysts due to oxygen storage and redox properties.
• Atomically precise structural determination of these clusters is challenging due to complex structures and lack of efficient sampling methods.

Purpose of the Study:

• To develop a reliable computational method for identifying the global minimum energy structures of cerium oxide clusters.
• To investigate the structural stability and electronic properties of specific cerium oxide cluster compositions (Ce14O28, Ce14O27, Ce14O29).

Main Methods:

• Combined neural network algorithms with density functional theory (DFT) calculations to create a high-dimensional potential.
• Employed neural network potential-enhanced molecular dynamics for structural stability analysis at high temperatures.
• Explored over 100,000 configurations to identify low-energy structures.

Main Results:

• Identified pyramid-like structures as the lowest energy configurations for Ce14O28, Ce14O27, and Ce14O29 clusters (approx. 1.0 nm).
• Demonstrated the structural stability of these pyramid-like clusters at elevated temperatures.
• Electronic structure analysis revealed high catalytic activity and facile oxygen release.

Conclusions:

• Neural network potentials offer an effective approach for exploring stable structures of metal oxide nanoclusters.
• The findings provide insights into the structure-property relationships of cerium oxide clusters for catalytic applications.