Speaker
Description
The search for novel conventional superconductors is increasingly turning towards ternary and quaternary systems with large unit cells. However, because the number of local minima scales exponentially with the number of atoms, purely DFT-driven crystal structure prediction struggles to explore these complex energy landscapes and reliably determine thermodynamic stability. To address this challenge, we employ the recently developed hot-AIRSS methodology [1]. By leveraging machine-learned interatomic potentials to perform short molecular dynamics runs between structural relaxations, this approach greatly accelerates investigations of these larger cells. We apply this workflow to search for promising ternary hydrides at 100 GPa, coupling the hot-AIRSS search with high-throughput electron-phonon coupling calculations - accelerated by up to two orders of magnitude via density of states rescaling [2] - to solve the Eliashberg equations for the superconducting transition temperature. We will present the results of this search, discussing the structural diversity, thermodynamic stability, and predicted $T_{\mathrm{c}}$ values of the most promising candidate phases.
[1] CJ Pickard. “Beyond theory-driven discovery: introducing hot random search and datum-derived structures”. In: Faraday Discussions 256.0(2025), pp. 61–84.
[2] K Bozier et al. “High-throughput superconducting Tc predictions through density of states rescaling”. In: Physical Review B 113.6 (2026), p. 064507.