Melting Transitions in Small Aluminum Clusters Simulated with Energies Approaching DFT Accuracy

Area of Science:

• Computational physics and chemistry
• Materials science
• Statistical mechanics

Background:

• Modeling melting-like transitions in atomic clusters is crucial for understanding material properties.
• Traditional methods often lack the efficiency or accuracy needed for complex systems.

Purpose of the Study:

• To develop and validate a novel computational framework for simulating melting-like transitions in atomic clusters.
• To investigate the melting behavior of various aluminum cluster ions and neutral clusters.

Main Methods:

• Combines global optimization, Density Functional Theory (DFT) energy calculations, and machine-learned interatomic potentials (MLIPs).
• Utilizes an Allegro E(3)-equivariant neural network potential for accurate energy fitting.
• Employs parallel tempering Monte Carlo simulations for efficient modeling.

Main Results:

• The MLIP achieves accuracy of 10 meV/atom or better.
• Simulations of Na20 validate the methodology against prior results.
• Melting points of Al_n^+ (n=9-16) clusters are studied, with some exceeding the bulk melting point.
• Al13^- exhibits an exceptionally high melting point near 2100 K.

Conclusions:

• The developed framework enables efficient and accurate simulations of melting-like transitions.
• Aluminum cluster ions show distinct melting behaviors compared to bulk aluminum.
• Specific cluster structures, like Al13^-, possess remarkably high thermal stability.