Molecular dynamics simulation of metallic Al-Ce liquids using a neural network machine learning interatomic...

Area of Science:

• Materials Science
• Computational Materials Science
• Metallurgy

Background:

• Aluminum-cerium (Al-Ce) alloys show promise for high-temperature applications, potentially replacing heavier materials like steel.
• Understanding the liquid state structures and properties of Al-Ce alloys is essential for optimizing their manufacturing processes.
• Accurate interatomic potentials are currently lacking for reliable molecular dynamics simulations of Al-Ce alloy systems.

Purpose of the Study:

• To develop a reliable interatomic potential for Al-Ce alloys using artificial neural network (ANN) deep machine learning (ML).
• To validate the developed potential against ab initio calculations for both liquid and crystalline Al-Ce phases.
• To utilize the validated potential for molecular dynamics simulations of Al90Ce10 liquid to gain insights into its structure and properties.

Main Methods:

• Utilized ab initio molecular dynamics simulation data for Al-Ce liquid and crystalline compounds to train an ANN-ML interatomic potential.
• Employed a small unit cell size (approximately 200 atoms) for ab initio simulations during the training phase.
• Applied the developed ANN-ML potential in molecular dynamics simulations to investigate the Al90Ce10 liquid.

Main Results:

• The ANN-ML model accurately reproduced energies, forces, and atomic structures of Al90Ce10 liquid and crystalline Al-Ce compounds compared to ab initio results.
• The developed interatomic potential demonstrated reliability for simulating Al-Ce alloy systems.
• Molecular dynamics simulations using the new potential provided insights into the structures and properties of liquid Al90Ce10.

Conclusions:

• An accurate Al-Ce interatomic potential was successfully developed using ANN-ML, overcoming a significant simulation challenge.
• The validated potential enables reliable molecular dynamics simulations of Al-Ce alloys.
• This work provides valuable insights to guide experimental processes for producing desired Al-Ce alloys for high-temperature applications.