Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
The authors constructed SoLiD26, an open dataset containing 15.4 million DFT-computed atomic configurations across 15 chemical elements and up to 576 atoms per cell to resolve the severe data scarcity at solid-liquid interfaces. Simulating these boundaries is critical for electrochemistry, catalysis, and corrosion, where machine-learned interatomic potentials (MLIPs) must simultaneously capture ordered solids, disordered liquids, and their dynamic interactions. Baseline evaluations using equivariant MACE architectures validate the dataset's efficacy for training and benchmarking MLIPs across chemically and structurally heterogeneous interfaces.
Electrochemical and catalytic interfaces have long been an unaddressed blind spot for atomistic ML; 15.4 million ab initio configurations now supply the training scale required to model them accurately.
Machine-learned interatomic potentials (MLIPs) for solid-liquid interfaces in advanced materials applications, e.g., electrochemistry, catalysis and corrosion, require training data that samples both liquid environments, the solid and the interface itself. We present SoLiD26, a curated solid-liquid interface dataset, containing 15.4 million first-principles atomic structures with up to 576 atoms and 15 chemical elements for training and evaluating MLIPs. The structures were compiled from density functional theory (DFT) calculations performed in studies of solid-liquid interfaces, with most configurations originating from ab initio molecular dynamics (AIMD) simulations. Each record contains atomic species, positions, simulation cell, periodic boundary conditions, potential energy and atomic forces. SoLiD26 includes aqueous coinage metal interfaces, electrode-electrolyte systems, and selected bulk reference structures, calculated with VASP using the PBE functional and D3 dispersion corrections. We describe the data ingestion and preparation pipeline used to construct the dataset. The application of SoLiD26 for training and evaluating MLIPs is demonstrated with a suite of MACE models on a simple training, validation and test split. The dataset enables development and benchmarking of MLIPs for structurally and chemically heterogeneous solid-liquid interfaces.