Insights into CO dimerization at electrified Cu interfaces from large-scale machine learning simulations

2025/09/22 by Sushree Jagriti Sahoo, Mikael Maraschin, Sahoo, Sushree Jagriti +19 · 1 voice · 4 citations
Materials Science · Physics and Astronomy · #Catalytic Processes in Materials Science #Chemical Physics (physics.chem-ph) #FOS: Physical sciences #Materials Science (cond-mat.mtrl-sci) #cond-mat.mtrl-sci #physics.chem-ph

paper · pdf · doi:10.48550/arxiv.2509.17862

openalex publication_date 2025/09/22 · arxiv published 2025/09/22 · openalex created_date 2025/10/17 · arxiv updated 2026/06/08 · openalex updated_date 2026/08/01

Abstract

Catalysis at solid-liquid interfaces underpins many energy technologies, yet ab initio simulations that capture interfacial dynamics remain prohibitively expensive. Here we introduce Open Catalyst 2025 (OC25), the largest dataset for solid-liquid interfaces. To demonstrate OC25-trained models as practical tools for electrocatalysis, we investigate CO dimerization on Cu surfaces, a key step in CO₂ electroreduction. Using large cells (>800 atoms) and enhanced sampling up to 7 ns - the largest explicit-solvent CO dimerization study to date - we compute free-energy profiles under varied surface charge, cation identity, and surface facet. We find that dimerization is weakly sensitive to charge and cation identity, with appreciable stabilization only at the most negative charge densities, while extension to stepped Cu(310) reveals a more favorable pathway at modest reducing potentials. Our results demonstrate that OC25-trained models provide a scalable tool for investigating electrocatalytic transformations at solid-liquid interfaces, enabling simulations orders of magnitude beyond ab initio methods.

Citations

• UMA: A Family of Universal Models for Atoms
• The Open Molecules 2025 (OMol25) Dataset, Evaluations, and Models
• Machine Learning and Data-Driven Methods in Computational Surface and Interface Science
• Learning Smooth and Expressive Interatomic Potentials for Physical Property Prediction
• Open Catalyst Experiments 2024 (OCx24): Bridging Experiments and Computational Models
• CatTSunami: Accelerating Transition State Energy Calculations with Pre-trained Graph Neural Networks
• EquiformerV2: Improved Equivariant Transformer for Scaling to Higher-Degree Representations
• SPICE, A Dataset of Drug-like Molecules and Peptides for Training Machine Learning Potentials
• MACE: Higher Order Equivariant Message Passing Neural Networks for Fast and Accurate Force Fields
• GemNet-OC: Developing Graph Neural Networks for Large and Diverse Molecular Simulation Datasets
• Towards an accelerated decarbonization of chemical industry by electrolysis
• Methods for nitrogen activation by reduction and oxidation
• Implicit Solvation Methods for Catalysis at Electrified Interfaces
• Absence of CO2 electroreduction on copper, gold and silver electrodes without metal cations in solution
• E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentials
• The atomic simulation environment—a Python library for working with atoms
• A consistent and accurateab initioparametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu
• PACKMOL: A package for building initial configurations for molecular dynamics simulations
• A dimer method for finding saddle points on high dimensional potential surfaces using only first derivatives
• Improved adsorption energetics within density-functional theory using revised Perdew-Burke-Ernzerhof functionals
• From ultrasoft pseudopotentials to the projector augmented-wave method
• Efficient iterative schemes forab initiototal-energy calculations using a plane-wave basis set
• Ab initiomolecular-dynamics simulation of the liquid-metal–amorphous-semiconductor transition in germanium
• Frequency factors and isotope effects in solid state rate processes

Cited by

• Breaking the Training Barrier of Billion-Parameter Universal Machine Learning Interatomic Potentials
• Benchmarking Simulacra AI's Quantum Accurate Synthetic Data Generation for Chemical Sciences
• Cross Learning between Electronic Structure Theories for Unifying Molecular, Surface, and Inorganic Crystal Foundation Force Fields
• Fast and Accurate Foundation Models for Equivariant Machine-Learned Interatomic Potentials
• UMA: A Family of Universal Models for Atoms

Discussions

• Excited to share the OC25 dataset and models for solid-liquid interfaces! Great fun working on this project! We hope OC25 will accelerate the modeling of interfaces and are looking forward to feedback [bsky, 0 points, 0 comments]

Related

• The Open Catalyst 2022 (OC22) Dataset and Challenges for Oxide Electrocatalysts
• Open Challenges in Developing Generalizable Large Scale Machine Learning Models for Catalyst Discovery
• Machine Learning and Data-Driven Methods in Computational Surface and Interface Science
• Catlas: an automated framework for catalyst discovery demonstrated for direct syngas conversion
• A constant potential reactor framework for electrochemical reaction simulations