Mark RadarMARK RADAR
About
EN
Sign in
Event File AI

XtalPaint Restores Missing Hydrogen in Crystal Structures

1 reports · First detected 2026-08-12 · Last active 2026-08-12

Hydrogen positions can shape a crystalline material’s electrical, thermal and energy-storage properties, yet the element scatters X-rays weakly and is difficult to locate with conventional diffraction. Neutron measurements can provide better data but require specialized large-scale facilities and are more costly. Crystallographic databases therefore often omit hydrogen atoms or place them using heuristics, creating gaps that can undermine computer simulations and the screening of materials for batteries, hydrogen storage and other applications.

Switzerland’s Paul Scherrer Institute said on June 11, 2026, that its researchers developed XtalPaint by adapting Microsoft’s MatterGen model with TD-Paint, an inpainting technique from computer vision. In tests that removed hydrogen positions from known crystal structures, the open-source system reconstructed the reference arrangement in 87% of cases and found a more energetically stable configuration in another 10%, producing a 97% overall success rate. The researchers said the approach could also be applied to lithium and sodium, two elements central to battery development.

All Coverage

1 original reports

The Backstory

The history behind this event

No historical echoes for this signal

Mark Radar|MARK RADAR
All times are in Taipei time (GMT+8)