ENVIRONMENT AND ENERGY

Report

Many materials that could improve solar cells, LEDs or batteries look orderly in textbook drawings. At room temperature, however, their atoms are never still. They vibrate and form short-lived local structures that differ from the average, static crystal. This dynamic disorder is strong in some energy materials such as halide perovskites. It can change how a material absorbs light, how charges move, and how ions migrate through a battery electrolyte.

The project “Dynamic disorder in strongly anharmonic semiconductors investigated by machinelearning accelerated dynamics” described here was set out to capture this moving atomic world realistically. It was led by Prof. Dr. David A. Egger from the Department of Physics, TUM School of Natural Sciences, Technical University of Munich. The work was carried out by the Theory of Functional Energy Materials group with collaborators.

The challenge addressed here was that common calculations often treat atomic movements as small, regular vibrations around fixed positions. This is useful for simple crystals, but misses the large, irregular motions found in highly anharmonic materials. To overcome this, the team used molecular dynamics simulations, comparable to making an atom-by-atom movie. Each frame requires quantum-mechanical calculations of the forces acting on many atoms, and useful movies must be long and large enough to avoid misleading size effects.

This is why supercomputing was essential. The simulations were run on the JUWELS highperformance computing (HPC) center at Forschungszentrum Jülich. First-principles molecular dynamics and density-functional theory calculations were performed with a combination of publicly available code packages, and home-written code, typically using multiple computer nodes and cores. This made it possible to compare different material sizes, temperatures and settings, and to connect simulated motions to measurable spectra and electronic properties.

One outcome was a clearer picture of how finite-temperature motion controls material behavior. In the nitride semiconductor CuTaN₂, simulations showed that tantalum atoms move away from their high-symmetry positions at room temperature. Including these movements changed the predicted band gap, one of the important properties for semiconducting materials applications, from about 0.4 to 0.8 eV, close to the range relevant for solar-energy harvesting. The simulations also helped explain Raman measurements, which detect atomic vibrations.

The team also studied halide perovskites, known for strong light-matter interactions. For CsPbBr₃, regular harmonic vibrations (which resemble a pendulum moving around its equilibrium position) could not explain the observed temperature behavior of the band gap. The electronic structure is instead linked to anharmonic fluctuations: larger and less regular atomic motions. In simulations of the two-dimensional perovskite BA₂PbI₄, the size of the simulated crystal cell affected how phase changes appeared, showing that carefully tuning simulation parameters is crucial to capture the correct physics.