Machine-Learning-Driven Simulations on Microstructure and Thermophysical Properties of MgCl₂-KCl Eutectic

Area of Science:

• Computational Materials Science
• Physical Chemistry
• Machine Learning in Materials

Background:

• Ab initio calculations (Density Functional Theory - DFT) are crucial for MgCl2-KCl eutectic studies but limited by computational cost for large-scale and long-time simulations.
• Accurate prediction of thermophysical properties and microstructure evolution requires methods that overcome these computational barriers.

Purpose of the Study:

• To develop a more efficient and accurate computational scheme for studying the MgCl2-KCl eutectic.
• To investigate the microstructure and thermophysical properties of the MgCl2-KCl eutectic using the developed method.
• To provide a deeper understanding of the Mg2+ ion coordination, medium-range order, and transport properties.

Main Methods:

• Constructed a deep potential (DP) using ab initio calculations to describe interatomic interactions.
• Employed molecular dynamics (MD) simulations powered by the DP for higher efficiency and comparable accuracy to DFT.
• Analyzed structural evolution using partial radial distribution functions, coordination numbers, angular distribution functions, and structural factors.

Main Results:

• The DP-based MD simulations successfully predicted thermophysical properties (density, thermal expansion, viscosity, diffusion, specific heat) in agreement with experimental data.
• Revealed that Mg2+ ions exhibit a distorted tetrahedral coordination, not octahedral, and possess a strong local structure.
• Observed medium-range order in the microstructure, which is enhanced at higher temperatures.

Conclusions:

• The combined ab initio-DP-MD approach offers a computationally efficient alternative to pure DFT for complex material systems.
• The study provides valuable insights into the structural characteristics and thermophysical behavior of the MgCl2-KCl eutectic.
• Enhanced understanding of ion coordination and diffusion mechanisms, crucial for molten salt applications.