AQCat Adsorption Spin

AQCat is a Large Quantitative Model for catalyst discovery, built as a machine learning interatomic potential that predicts the key properties determining catalyst performance. It was trained on 13.5 million high-fidelity quantum chemistry calculations spanning 47,000 catalyst systems, which lets it approach the accuracy of physics-based methods such as density functional theory while running up to 20,000 times faster. This speed changes the economics of screening: instead of the fewer than 100 candidates per week that conventional laboratory methods typically evaluate, research teams can rank vast candidate libraries computationally and reserve costly laboratory and supercomputing work for only the strongest leads.

Two design choices set AQCat apart in fidelity and reach. First, it explicitly accounts for spin polarization, a physical effect that governs the electronic structure of widely used industrial metals such as iron, nickel, and cobalt and that many comparable models omit. Second, it supports six additional industrially relevant elements (barium, cerium, fluorine, lithium, lanthanum, and magnesium) that are absent from other datasets, extending the model to catalysts and critical-material pathways that older methods could not reach.

Together these capabilities compress R&D timelines and de-risk investment across the discovery pipeline, with direct applications in green hydrogen, sustainable aviation fuel, fertilizer production, and plastics recycling. Teams can run AQCat on demand within infrastructure they already operate, keeping every calculation inside their own environment for security and governance.