Search papers, labs, and topics across Lattice.
This paper introduces a data-driven workflow for constructing machine-learning interatomic potentials (MLIPs) specifically tailored for gas-surface scattering dynamics, using NO scattering from graphite as a benchmark. They employ SOAP descriptors, PCA for feature reduction, and farthest point sampling to create a compact training set for a Deep Potential model. Refinement is achieved through active learning, enabling large-scale molecular dynamics simulations that reproduce experimental trends and provide insights into scattering dynamics at a fraction of the computational cost of AIMD.
Descriptor-guided sampling and active learning slashes the cost of simulating gas-surface interactions, enabling accurate molecular dynamics at scale.
Accurate atomistic simulations of gas-surface scattering require potential energy surfaces that remain reliable over broad configurational and energetic ranges while retaining the efficiency needed for extensive trajectory sampling. Here, we develop a data-driven workflow for constructing a machine-learning interatomic potential (MLIP) tailored to gas-surface scattering dynamics, using nitric oxide (NO) scattering from highly oriented pyrolytic graphite (HOPG) as a benchmark system. Starting from an initial ab initio molecular dynamics (AIMD) dataset, local atomic environments are described by SOAP descriptors and analyzed in a reduced feature space obtained through principal component analysis. Farthest point sampling is then used to build a compact training set, and the resulting Deep Potential model is refined through a query-by-committee active-learning strategy using additional configurations extracted from molecular dynamics simulations over extended ranges of incident energies and surface temperatures. The final MLIP reproduces reference energies and forces with high fidelity and enables large-scale molecular dynamics simulations of NO scattering from graphite at a computational cost far below that of AIMD. The simulations provide detailed insight into adsorption energetics, trapping versus direct scattering probabilities, translational energy loss, angular distributions, and rotational excitation. Overall, the results reproduce the main experimental trends and demonstrate that descriptor-guided sampling combined with active learning offers an efficient and transferable strategy for constructing MLIPs for gas-surface interactions.
