Temperature Extensible Deep Potential Model for Molten NaF-BeF₂-ZrF₄: Predicting Transport Properties and Local...

Area of Science:

• Materials Science
• Computational Chemistry
• Nuclear Engineering

Background:

• Molten NaF-BeF2-ZrF4 (FNaBeZr) is a key candidate for nuclear fuel carrier salts in molten salt reactors.
• Accurate simulation of its properties is crucial for reactor design and safety.

Purpose of the Study:

• To systematically investigate the transport properties and local structures of molten FNaBeZr using deep potential molecular dynamics (DPMD).
• To establish correlations between structural features and temperature-dependent transport properties.
• To demonstrate the computational efficiency and scalability of the DPMD approach for molten salt systems.

Main Methods:

• Deep Potential Molecular Dynamics (DPMD) simulations integrating first-principles calculations, machine learning, and molecular dynamics.
• Systematic investigation across a broad temperature range (773–973 K and beyond).
• Analysis of densities, ionic self-diffusion coefficients, pair/cluster structures, ionic conductivity, and shear viscosity.

Main Results:

• DPMD accurately reproduces known properties and successfully predicts others at extended temperatures.
• Quantifiable correlations between structural features (coordination number, peak positions) and temperature were established.
• Demonstrated significant reduction in computational cost and enhanced system scalability compared to traditional methods.

Conclusions:

• The DPMD framework provides a computationally efficient and scalable method for simulating molten fluoride salts.
• Results offer critical technical parameters and engineering data for molten salt reactor design and optimization.
• The deep potential model shows temperature extensibility for molten FNaBeZr, enabling future simulations.