Speaker
Description
The $\alpha$–$\beta$ phase transition in tin represents a challenging problem for atomistic simulations due to the small free-energy difference between the two phases and the importance of finite-temperature effects. Accurate evaluation of these effects requires simulations beyond the time and length scales typically accessible by first-principles methods. Machine-learned interatomic potentials (MLIPs) provide an efficient alternative, enabling large-scale molecular dynamics simulations with near-DFT accuracy at a fraction of the computational cost of DFT.
In this work, an Atomic Cluster Expansion (ACE) potential trained on density functional theory data is employed to investigate the thermodynamic properties of $\alpha$- and $\beta$-Sn. The temperature dependence of their free energies is studied using both the quasi-harmonic approximation and thermodynamic integration. Comparison of these approaches allows us to evaluate the role of anharmonic contributions and their influence on the predicted phase stability.
Anharmonic free energies are calculated using the Calphy workflow based on molecular dynamics and thermodynamic integration. The free energies of both phases are obtained by reversible switching between the ACE description and suitable reference systems, followed by temperature integration over the investigated temperature range. This approach provides access to the full anharmonic free energy beyond the quasi-harmonic approximation and enables a direct comparison of the thermodynamic stability of $\alpha$- and $\beta$-Sn.