Global optimization and structural evolution of platinum clusters (Pt_N, N = 6-50) via deep potential and hybrid...

Area of Science:

• Computational materials science
• Nanotechnology
• Catalysis

Background:

• Platinum (Pt) clusters exhibit unique properties crucial for catalysis and nanoelectronics.
• Determining the lowest energy structures of Pt clusters computationally is challenging, especially for larger sizes.
• Existing methods struggle with accuracy and scalability for predicting global minimum structures.

Purpose of the Study:

• To develop and implement an efficient computational strategy for finding ground-state structures of Pt clusters.
• To extend the size range of accurately predicted Pt cluster structures beyond previous limitations.
• To analyze the structural evolution and identify stable configurations of Pt clusters.

Main Methods:

• Constructed a high-quality dataset using ab initio molecular dynamics (AIMD) simulations (300-1500 K).
• Trained a deep potential (DP) model with high energy prediction accuracy (MAE < 0.011 eV/atom) and cross-size generalization.
• Integrated the DP model with an improved hybrid differential evolution (HDE) algorithm for structure searching.

Main Results:

• Successfully extended ground-state structure searches for Pt clusters up to 50 atoms with near first-principles accuracy.
• Identified more stable structures compared to those predicted by empirical potentials (S-C potential).
• Discovered magic number sizes (10, 17, 30, 36, 42, 43) and elucidated structural transitions (cage-like to layer-like to multi-shell).

Conclusions:

• The DP-HDE framework provides an efficient and reliable tool for studying medium-to-large Pt clusters.
• This method overcomes the computational bottleneck of traditional high-precision calculations for cluster structure prediction.
• The findings offer insights into the structure-property relationships and evolutionary patterns of Pt clusters.