Hydrogen diffusion on Ni(100): A combined machine-learning, ring polymer molecular dynamics, and kinetic Monte Carlo...

Area of Science:

• Materials Science
• Computational Chemistry
• Surface Science

Background:

• Understanding hydrogen diffusion on metal surfaces is crucial for catalysis and materials design.
• Accurate modeling of hydrogen diffusion requires considering nuclear quantum effects and surface coverage.
• Existing methods often struggle to capture the complexity of diffusion processes across various scales.

Purpose of the Study:

• To develop and validate a comprehensive computational framework for studying collective hydrogen diffusion on metal surfaces.
• To accurately model hydrogen adsorption, diffusion energetics, and their dependence on local coverage.
• To investigate the role of nuclear quantum effects in hydrogen diffusion at low temperatures.

Main Methods:

• Coupling machine-learning potentials (neural networks) with ring polymer molecular dynamics (RPMD) and kinetic Monte Carlo (kMC).
• Utilizing density functional theory (DFT) to generate training data for neural network potentials.
• Employing umbrella sampling and trajectory recrossing simulations to calculate diffusion rate constants.

Main Results:

• The neural network potential accurately reproduces DFT energies and forces.
• Effective diffusion rates were calculated for various temperatures and hydrogen coverages.
• The computational results show good agreement with experimental data for hydrogen diffusion on Ni(100).

Conclusions:

• The combined computational approach provides a powerful tool for understanding hydrogen diffusion on metal surfaces.
• This methodology enables accurate predictions of diffusion rates across different conditions.
• The framework has broad applicability in materials science for designing new materials and processes.