mat-qha-thermal-expansion

To calculate thermal expansion using machine learning interatomic potentials, you can use a quasi-harmonic approximation workflow that fits an equation of state and evaluates MLIP-derived vibrational contributions across a temperature grid. This script requires a relaxed structure file, a selected MLIP wrapper, and an EOS model to generate temperature-dependent thermal expansion data.

The quasi-harmonic approximation predicts temperature-dependent Gibbs energy by fitting an equation of state, such as Vinet, to MLIP-derived vibrational calculations. This method evaluates phonon contributions across a user-defined temperature grid to accurately compute Gibbs free energy for solid materials.

Yes, you can compute QHA thermal properties using MACE or MatGL models. The script supports selecting QHA-stable MLIP wrappers including MACEWrapper, MatGLWrapper, and FAIRCHEMWrapper to access the forces and vibrational stability required for the calculations.

To run the quasi-harmonic thermal expansion calculation, you need a relaxed structure file, an MLIP model type, an EOS fitting method, and an output directory. The script processes these inputs to generate gibbs_temperature.dat, thermal_expansion.dat, and a qha_results.json summary.

The best way to get volumetric thermal expansion data over a specific temperature range is to run a QHA script that applies an equation of state and MLIP-derived phonon calculations across a user-defined temperature grid. This approach outputs thermal_expansion.dat files containing the volumetric expansion coefficients.

A key limitation for QHA thermodynamics is that the chosen MLIP wrapper must be QHA-stable to ensure accurate vibrational calculations. If the potential lacks proper force evaluation or vibrational stability, the fitted equation of state and resulting Gibbs energy predictions may be invalid.