Hydrogen diffusion on metals: Full-Dimensional Ring Polymer Molecular Dynamics Rate Calculations
- 1Lehrstuhl für Theoretische Chemie, Egerlandstr. 3, Erlangen, Germany, 91058
Abstract
The diffusion of hydrogen atoms on nine different metals: Co, Ni, Cu, Rh, Pd, Ag, Ir, Pt, Au, has been simulated by ring polymer molecular dynamics. For Co, the (0001) surface facet has been chosen, while for all other metals, (100) surfaces are investigated. For the nine systems, message-passing atomic cluster expansion (MACE) machine-learned interatomic potentials (MLIPs) were set up by means of self-consistent biased fine-tunings. The H diffusion was sampled first with the MACE MP-0 foundation model by means of umbrella samplings along the diffusion path; then, a subset of sampled structures was recalculated by periodic density-functional theory (DFT) and the MACE was fine-tuned with the obtained reference data. This was repeated once to correct for biases of the foundation model concerning the sampling of configuration space. Thorough benchmarking revealed that the fine-tuned MLIPs were reliable and of high quality, also for situations not covered by the training set. The fine-tuned MACE MLIPs were then used to sample the full-dimensional diffusion rate constants with the Caracal program for temperatures between 20 and 300 K, in 10 K steps, and from them obtain the temperature-dependent diffusion coefficients. Crossover temperatures to the deep tunneling regime were calculated as well.
Related
- Steric Scaling of Hydrogen Transport in FCC Transition Metals for Mechanically Consistent Molecular Dynamics Simulations
- Multimodal Spectroscopy–Driven Reconstruction of Three-Dimensional Molecular Structures via Conditional Diffusion Model
- Interpretable Machine Learning of Hydrogen Production from Reactive Molecular Dynamics Simulations of Polyethylene Oxidative Gasification
- Ammonia-Hydrogen Phase Equilibria in Planetary Envelopes from First-Principles-Based Machine-Learning Molecular Dynamics
- Studying Protein-Ligand Interactions with In-droplet Hydrogen-Deuterium Exchange Mass Spectrometry and Molecular Dynamics Simulations
BibTeX
@misc{steffen2026hydrogen,
title = {{Hydrogen diffusion on metals: Full-Dimensional Ring Polymer Molecular Dynamics Rate Calculations}},
author = {Julien Steffen},
howpublished = {ChemRxiv},
year = {2026},
month = sep,
doi = {10.26434/chemrxiv.15008827/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008827/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Hydrogen diffusion on metals: Full-Dimensional Ring Polymer Molecular Dynamics Rate Calculations
AU - Julien Steffen
PY - 2026
DA - 2026/09/14
PB - ChemRxiv
DO - 10.26434/chemrxiv.15008827/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008827/v1
AB - The diffusion of hydrogen atoms on nine different metals: Co, Ni, Cu, Rh, Pd, Ag, Ir, Pt, Au, has been simulated by ring polymer molecular dynamics. For Co, the (0001) surface facet has been chosen, while for all other metals, (100) surfaces are investigated. For the nine systems, message-passing atomic cluster expansion (MACE) machine-learned interatomic potentials (MLIPs) were set up by means of self-consistent biased fine-tunings. The H diffusion was sampled first with the MACE MP-0 foundation model by means of umbrella samplings along the diffusion path; then, a subset of sampled structures was recalculated by periodic density-functional theory (DFT) and the MACE was fine-tuned with the obtained reference data. This was repeated once to correct for biases of the foundation model concerning the sampling of configuration space. Thorough benchmarking revealed that the fine-tuned MLIPs were reliable and of high quality, also for situations not covered by the training set. The fine-tuned MACE MLIPs were then used to sample the full-dimensional diffusion rate constants with the Caracal program for temperatures between 20 and 300 K, in 10 K steps, and from them obtain the temperature-dependent diffusion coefficients. Crossover temperatures to the deep tunneling regime were calculated as well.
ER -