Molecular Dynamics Study of Hydrogen Release from NaH Using Machine Learning Potential

Area of Science:

• Materials Science
• Computational Chemistry
• Nuclear Engineering

Background:

• Sodium hydride (NaH) is a hazardous byproduct in sodium-cooled fast reactors.
• Understanding NaH thermal decomposition is vital for reactor safety.
• Conventional simulations struggle with the atomic-scale mechanisms due to scale limitations.

Purpose of the Study:

• To develop an accurate deep neural network potential (DP) for simulating NaH.
• To investigate the atomic-scale thermal decomposition mechanism of NaH.
• To provide insights into hydrogen gas release relevant to nuclear reactor safety.

Main Methods:

• Developed a deep neural network potential (DP) for NaH using the DP-GEN framework.
• Validated the DP model against density functional theory (DFT) calculations and experimental data.
• Performed large-scale Deep Potential Molecular Dynamics (DPMD) simulations on NaH clusters and a slab model.

Main Results:

• The DP model accurately reproduced DFT results for NaH properties, outperforming ReaxFF.
• Simulations showed model-dependent thermal decomposition behavior.
• The slab model exhibited H2 formation and release at elevated temperatures (300-1500 K), while clusters showed transitions and dissociation at higher temperatures (1000-1500 K).

Conclusions:

• The developed DP model is a reliable tool for studying NaH decomposition.
• Atomistic insights into NaH thermal decomposition and hydrogen release were obtained.
• Findings are crucial for assessing safety hazards in sodium-cooled fast reactors.