Particle.news

Download on the App Store

Machine Learning Cracks Low-Temperature Structure of FAPbI3 Perovskite

Cryogenic experiments verified the modelled phase to guide design of more stable perovskite mixtures.

Overview

  • Researchers at Chalmers University identified and described the previously elusive low-temperature phase of formamidinium lead iodide (FAPbI3).
  • Machine learning–augmented simulations ran thousands of times longer than earlier efforts and scaled models from hundreds to millions of atoms.
  • The University of Birmingham confirmed the predicted structure by cooling samples to about −200°C to match simulation conditions.
  • The team observed formamidinium molecules becoming trapped in a semi-stable state as the material cools.
  • The findings, published in the Journal of the American Chemical Society, provide atomic-scale insight aimed at improving stability in future solar and optoelectronic materials.