Igor Abrikosov

B. O. Mukhamedov, A. V. Ponomareva, M.P. Belov, A. V. Shapeev, F. Tasnádi

National University of Science and Technology "MISIS", Russia; Linköping University, Sweden

Temperature‐dependent lattice dynamics in ab initio simulations of precipitation thermodynamics

Ab initio calculations have become an established tool for simulations of thermodynamic and mechanical properties of alloys. We illustrate their usefulness in studies of multicomponent Fe-Cr based alloys [1-3], including the tendency towards the spinodal decomposition of the solid solutions. However, in high-throughput simulations effects of lattice vibrations are most often neglected. Considering Zr₃Fe and C15-ZrFe₂ compounds, we show that the quasi-harmonic approximation (QHA) can be employed to calculate the free energy of the studied systems as a function of volume and temperature, and a good agreement can be observed between theoretical and experimental heat capacities within validity range of the QHA [4]. However, the latter is not directly suitable for systems with the so-called dynamical instabilities, the imaginary frequencies of the lattice vibrations at temperature T=0K. Here the T-dependence of the lattice dynamics must be treated explicitly. We investigate it for antiferromagnetic and ferromagnetic phases of cubic (B2) FeRh using the temperature-dependent effective potential (TDEP) method. We show that already at low temperature the cubic structure of the antiferromagnetic phase becomes dynamically stable, which eliminates the contradiction between experimental observations and previous theoretical results showing its dynamical instability at temperature T = 0K. The calculated difference between the vibrational entropies of the FM and AFM phases at a metamagnetic transition temperature turns out to be comparable with experimental value of the total entropy change [5]. Importance of the explicit treatment of the temperature dependent lattice vibrations is further illustrated considering simulations of the mixing thermodynamics [6] and diffusion [7-8] in dynamically unstable systems. Finally, we argue that employing on-the-fly trained moment tensor potential, an impressive efficiency of simulations at elevated temperatures can be combined with very high accuracy [9], promising a transformation of the temperature‐dependent treatment of the alloy thermodynamics into a state-of-the-art tool of materials modeling from first principles.