On-the-Fly Machine Learning Force Field Study of Liquid-Al/Solid-TiB₂ Interfaces

Area of Science:

• Materials Science
• Computational Materials Science
• Chemical Engineering

Background:

• Titanium diboride (TiB2) is a crucial nucleating agent in aluminum alloys.
• Understanding the atomic-level interactions at the liquid-aluminum/solid-TiB2 interface is key to optimizing alloy properties.
• The surface termination of TiB2 significantly influences its nucleation behavior.

Purpose of the Study:

• To investigate the atomic structure evolution at the liquid-Al/solid-TiB2 interface using advanced simulation techniques.
• To elucidate the mechanism by which TiB2 acts as a nucleating particle in aluminum alloys.
• To compare the interfacial properties and nucleation potential of Ti-terminated versus B-terminated TiB2 (0001) surfaces.

Main Methods:

• On-the-fly machine learning force field (MLFF) simulations were employed.
• Simulations were conducted for up to 100 picoseconds (ps) to capture interfacial dynamics.
• Analysis included structural information, dynamic characteristics, mean square displacement, and vibrational density of states.

Main Results:

• The termination of the TiB2 (0001) surface critically influences the formation of ordered structures at the interface.
• The Ti-terminated surface exhibited a greater propensity for forming quasi-solid regions compared to the B-terminated surface.
• Dynamic analyses indicated the liquid layer at the Ti-terminated interface possessed more solid-like characteristics.

Conclusions:

• The nucleation potential of TiB2 particles is directly linked to the interfacial ordered structures, which are termination-dependent.
• Distinct ordered structures arise from different terminations due to variations in diffusion abilities.
• Ti-terminated TiB2 surfaces are more effective nucleating sites for aluminum alloys.