Asynchronous Transition Across the Crystal-Melt Interface Revealed by Machine Learning Potentials.

Area of Science:

• Materials Science
• Physical Chemistry
• Computational Materials Science

Background:

• Understanding crystal-melt interfaces is key for materials science.
• Atomic-scale structure and thermodynamics are crucial for predicting material behavior.

Purpose of the Study:

• Investigate the atomic-scale structure and thermodynamics of crystal-melt interfaces.
• Utilize a DFT-based machine learning potential for silicon.
• Reveal the nature of the transition from crystal to melt.

Main Methods:

• Employed a DFT-based machine learning potential.
• Analyzed local order parameter ${\overline{q}}_6$
• Computed interfacial free energy via mold integration.
• Applied Wulff construction.

Main Results:

• The crystal-melt transition spans a broad region (> 12 Å) with asynchronous property evolution.
• Identified staged melting with distinct transition areas for structural, dynamic, and thermodynamic properties.
• Calculated anisotropic interfacial free energies (γ(111) < γ(110) < γ(100)).
• Wulff construction predicted equilibrium crystal shapes matching experiments.

Conclusions:

• The crystal-melt interface is a multi-dimensional region, not an abrupt change.
• Staged melting mechanism and interfacial anisotropy govern crystal growth and morphology.
• Findings link atomic-scale dynamics to macroscopic properties, crucial for materials design.