Abstract

The need for rapid estimates for the superconducting onset grows exponentially with the increasing capabilities of first-principles and artificial intelligence methods to explore ever-larger numbers of candidate materials. In this context, chemical bonding-oriented approaches have emerged as a promising proxy for predicting the value of the superconducting critical temperature (Tc) at a substantially reduced computational cost. The concept has so far been developed and restricted to hydride superconductors through the heuristic networking value method[1,2,3]. The latter estimates the interactions responsible for conventional superconductivity by quantifying the overlap between localized electron basins and bypassing the elevated computational cost of the phonon and electron-phonon properties necessary for the estimate of Tc. This seminar reviews the foundation of this method and presents a formulation to consolidate its formal basis, demonstrating that a reliable estimate of Tc is achievable from a real-space perspective. Here is also discussed how electron localization is an essential aspect to achieve high Tc in conventional superconductors. An extension of this approach is also proposed to broaden the applicability of the method beyond hydride to a wider range of conventional materials. [1] Strong correlation between electronic bonding network and critical temperature in hydrogen-based superconductors. Nat. Commun. 12, 5381 (2021) [2] TcESTIME: predicting high-temperature hydrogen-based superconductors. Chem. Sci. 2024 Nov 13;16(1):57–68 [3] Refining Tc Prediction in Hydrides via Symbolic‐Regression‐Enhanced Electron‐Localization‐Function‐Based Descriptors. Ann. Phys. (Berl.) 537 (11), e00280