Molecular dynamics computation of phase diagrams with superionic phases: the case of Fe and He⁴.

Recent experiments at the EuXFEL (Konopkova et al., Goldschmidt 2024, Chicago, IL, USA) confirmed our prediction (Belonoshko et al., Nature 2003; Nature Geoscience 2017) of the body-centered cubic (bcc) phase of Fe. These experiments close the debate about the stable phase of Fe in the Earth Inner Core. They also offer an opportunity to understand why a number of theoretical and experimental studies did not see the bcc phase of Fe, that becomes superionic at extreme pressures and temperatures. We will report our recent simulations of iron phase diagram using DFT based molecular dynamics with machine learning potentials with and without inclusion of magnetism. The computed phase diagram of Fe when magnetism is considered exhibits remarkable quantitative agreement with the recent experiments. The methods for computing phase diagrams where superionic phases are involved have to be revised. We demonstrate that the competition between bcc and fcc phases of He⁴ modeled with the pair Aziz potential was incorrectly described by D. Frenkel, likely because of the superionic nature of the bcc phase. In contrast, when we simulate the bcc+fcc+liquid 3-phase system we obtain a broad temperature field of the bcc phase stability at high pressure. This suggests that direct simulation of phase equlibria in the cases when Einstein crystal is not a good approximation of a solid is a more reliable method for computing phase diagrams. We shall also report properties of the bcc Fe in the Earth Core.