Export citation

Export citation

Choose format for download:

Download Citation

    Machine learning enhanced quantum molecular dynamics assessment of the phase diagram of uranium

    P. V. Chirkov1,*, G. S. Eltsov1, R. M. Kichigin1, A. A. Mirzoev2, A. V. Karavaev1, and V. V. Dremov1

    • *Contact author: chirkovpv@vniitf.ru

    Phys. Rev. B 112, 024104 – Published 8 July, 2025

    DOI: https://doi.org/10.1103/8w5w-f3kd

    Abstract

    Two approaches to the calculation of phase diagrams for materials under high pressures and temperatures with the use of ab initio molecular dynamics are discussed. The first is based on calculations of vibrational free energies of crystal phases with the temperature-dependent effective potential, and the second is based on calculations of thermodynamic potentials with the thermodynamic integration technique combined with machine learning accelerated quantum molecular dynamics with moment tensor potential. Detailed consideration is given to methodical issues. The emphasis of the paper is on phase diagram calculation for uranium in the region where its three solid phases (orthorhombic α, body-centered-cubic γ, and body-centered-tetragonal γ′) exist (P<400GPa and T<6000K). It is shown that the position of equilibrium lines on the phase diagram is sensitive to the choice of theoretical approximation [density functional theory (DFT), DFT+U], but phase diagram topology (the phases and their relative positions) on the P−T plane is not if the accuracy of thermodynamic potential calculations is below 3–5meV/at. Comparison with calculations by other authors and with available experimental data is provided. The prediction of a triple point α−γ−γ′ at P∼100GPa and T∼2400K in the calculation may be of interest in view of its possible verification in experiments in laser-heated diamond anvil cells.

    Physics Subject Headings (PhySH)

    Authorization Required

    We need you to provide your credentials before accessing this content.

    References (Subscription Required)

    Outline

    Information

    Sign In to Your Journals Account

    Filter

    Filter

    Article Lookup

    Enter a citation