Overview
- The study, published in Nature Communications in October 2026, expanded a handpicked set of 152 quantum calculations into a network of 9,389 elementary reactions for CO2 hydrogenation on copper.
- Researchers combined quantum‑mechanical simulations, machine learning to predict activation barriers, automated reaction enumeration and kinetic modelling to scale exploration of 105 surface species affordably.
- In kinetic simulations the larger network raised predicted CO2 conversion by about 40-fold and changed the major predicted products from formic acid to methanol and carbon monoxide, matching experimental trends.
- The expanded map revealed a previously overlooked pathway in which molecular H2 transfers intact to oxygen‑containing intermediates, and targeted quantum calculations confirmed that this route can be favorable.
- Authors present the framework as adaptable to other catalytic problems and a tool to guide catalyst development, while noting that broad experimental validation and engineering translation remain necessary next steps.