Designing high-TC superconductors with BCS-inspired screening, density functional theory, and deep-learning

Abstract

We develop a multi-step workflow for the discovery of conventional superconductors, starting with a Bardeen–Cooper–Schrieffer inspired pre-screening of 1736 materials with high Debye temperature and electronic density of states. Next, we perform electron-phonon coupling calculations for 1058 of them to establish a large and systematic database of BCS superconducting properties. Using the McMillan-Allen-Dynes formula, we identify 105 dynamically stable materials with transition temperatures, TC ≥ 5 K. Additionally, we analyze trends in our dataset and individual materials including MoN, VC, VTe, KB6, Ru3NbC, V3Pt, ScN, LaN2, RuO2, and TaC. We demonstrate that deep-learning(DL) models can predict superconductor properties faster than direct first-principles computations. Notably, we find that by predicting the Eliashberg function as an intermediate quantity, we can improve model performance versus a direct DL prediction of TC. We apply the trained models on the crystallographic open database and pre-screen candidates for further DFT calculations.

High-Throughput DFT-Based Discovery of Next Generation Two-Dimensional (2D) Superconductors

“…We also observe that a broadening parameter of 0.68 eV (0.05 Ry) is sufficient in converging the T c result within our desired tolerance, similar to our previous work. 78 This behavior is similar to the additional convergence tests we performed on a selected group of monolayers (Mg 2 B 4 N 2 , W 2 N 3 , TiClO, NbO 2 , 2H-NbS 2 , ZrBrO, 2H-NbSe 2 ) shown in Figure S2, which all display a converged T c value with respect to a broadening value of 0.68 eV (0.05 Ry). Additional q-point convergence (using the converged k-grid from JARVIS-DFT) for data is presented in Table S1 for 2D 2H-TaS 2 , 2H-NbSe 2 , W 2 N 3 , Mg 2 B 4 N 2 , and TiClO.…”

“…After rerelaxing the structures with Quantum Espresso (using PBEsol and GBRV potentials), we obtained similar results to the original structure from JARVIS relaxed with VASP using OptB88vdW (i.e., there was about an average ≈1−2% difference in in-plane lattice constant between the two methods). Similar to our previous work, 78 we used a 610 eV (45 Ry) plane-wave cutoff since we find that higher values have minimal effect on the calculated EPC parameters. The EPC parameter is derived from spectral function α 2 F(ω), which is given by…”

“…For this reason, we had to modify the BCS screening workflow that was used in our previous work. 78 Instead of screening the density of states (DOS) at the Fermi level and the Debye temperature (as we did for bulk materials), we screened materials based on the density of states (DOS) at the Fermi level, the total magnetic moment, and the electronic band gap (calculated with the OptB88vdW functional). This screening process is based on the notion that a potential 2D superconductor will have a high density of states at the Fermi level (and therefore be metallic), which is inspired by the BCS equation for T c (see eq 1).…”

“…In our previous work, we extensively investigated the effects of these parameters on the converged results. 78 We adopt a similar, less extensive procedure for selected 2D materials to confirm that our convergence criteria for bulk materials is valid in the 2D case. In Figure S1 we show the convergence of λ, ω log , and T c with respect to different k-point and q-point grids with respect to broadening for the 2H-NbSe 2 compound, which is a well-known 2D superconductor experimentally with a high T c value.…”

Systematic DFT+U and Quantum Monte Carlo Benchmark of Magnetic Two-Dimensional (2D) CrX3 (X = I, Br, Cl, F)

“…One of the main components, JARVIS-DFT, is a comprehensive database of DFT-computed material properties for over 3000 2D and 2D-like materials and over 60 000 bulk materials, with results from multiple functionals such as PBE and vdW-DF-OptB88. JARVIS-DFT contains DFT-calculated structural, energetic, elastic, optoelectronic, thermoelectric, piezoelectric, dielectric, infrared, solar-efficiency, topological, anomalous quantum confinement, and superconducting properties. As a response to the limitations of DFT, a limited number of beyond-DFT data, such as hybrid functionals (HSE06, PBE0) or many-body (GW, DMFT) results have been added to JARVIS, to increase accuracy and reliability of data.…”