Machine-learning interatomic potentials with global charge embeddings and multifidelity training predict the ground-state structures and thermodynamics of charged defects in the solar absorber ${\text{Sb}}_2{\text{Se}}_3$ with hybrid-DFT accuracy.

Synthesizing iron catalysts on an alumina support via simple ball-milling enables effective methane decomposition into hydrogen and graphitic carbon.

By exploring lithium-ion diffusion in a garnet-type oxide across a wide temperature range, this work offers insight into the ion conduction mechanism in these solid-state electrolytes.

This work establishes a practical route to include electron-phonon contributions to the total energy in first-principles simulations, unlocking accurate phase-stability predictions for energy materials.

Machine-learning molecular dynamics, supported by high-temperature X-ray diffraction, enable accurate prediction of thermal expansion and capture structural evolution in complex perovskite oxides for high-temperature electrochemical devices.