English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Reactive chemistry of molecular hydrogen at surfaces, notably dissociativesticking and hydrogen evolution, plays a crucial role in energy storage andfuel cells. Theoretical studies can help to decipher underlying mechanisms andreaction design, but studying dynamics at surfaces is computationallychallenging due to the complex electronic structure at interfaces and the highsensitivity of dynamics to reaction barriers. In addition, ab initio moleculardynamics, based on density functional theory, is too computationally demandingto accurately predict reactive sticking or desorption probabilities as itrequires averaging over tens of thousands of initial conditions.High-dimensional machine learning-based interatomic potentials are starting tobe more commonly used in gas-surface dynamics, yet, robust approaches togenerate reliable training data and to assess how model uncertainty affects theprediction of dynamic observables are not well established. Here, we employensemble learning to adaptively generate training data while assessing modelperformance with full uncertainty quantification for reaction probabilities ofhydrogen scattering on different copper facets. We use this approach toinvestigate the performance of two message-passing neural networks, SchNet andPaiNN. Ensemble-based uncertainty quantification and iterative refinement allowus to expose the shortcomings of the invariant pairwise-distance-based featurerepresentation in the SchNet model for gas-surface dynamics.

Machine learning interatomic potentials for reactive hydrogen dynamics at metal surfaces based on iterative refinement of reaction probabilities