arXiv Science⌕ Search

arXiv · 2611.15627

Machine-accelerated extraction of experimental superconducting parameters for physical insight and prediction across thousands of materials

Abstract

New superconductors could unlock next-generation technologies from nuclear fusion to lossless electronics. Machine learning can predict candidate superconductors but lacks validated experimental training data beyond the critical temperature. Here we present AutoSuperCon, a database of eight superconducting parameters, containing 20354 entries for 6793 superconductors. The data were extracted from over 47000 arXiv preprints using a large language model and validated against reference datasets ($R^2 = 0.96$ with NIMS SuperCon). Using AutoSuperCon, we show that median gap ratios rise from 3.6 in conventional superconductors to 6.4 in cuprates, 28% of upper critical fields exceed the Pauli limit, and critical temperatures lie below a ceiling proportional to the superfluid density. Further, we train composition-based predictors for the critical temperature ($R^2_{\log}=0.83$) and the upper critical field ($R^2_{\log}=0.69$), boosted by transfer learning. Within known superconducting families, composition alone is sufficient to screen candidates for both parameters. AutoSuperCon thus enables physical insight and data-driven discovery across thousands of superconductors.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Mihir Manium, Benjamin Geisler, Peter J. Hirschfeld, Amalia I. Coldea. 2026-11-11. Machine-accelerated extraction of experimental superconducting parameters for physical insight and prediction across thousands of materials. https://arxiv.science/abs/2611.15627

Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite arXiv Science by Indiciae: BibTeX · RIS.

KEEP EXPLORING

Related papers

From the Full Fermionic Hilbert Space to the BCS Paired Subspace

Conventional Bardeen--Cooper--Schrieffer (BCS) and number-projected BCS variational treatments are formulated within a paired Hilbert subspace, where modes with opposite momenta and spins are either both empty or both occupied. Here we ask whether this paired subspace can itself be selected energetically from the full fermionic Hilbert space of the attractive reduced BCS model. Using conserved projectors that distinguish singly occupied states from empty and doubly occupied ones, we decompose the full Hilbert space into invariant sectors. A direct energy comparison between sectors shows that, at zero temperature, every ground state at fixed average particle number lies in the paired subspace. The same holds at fixed even particle number, with the comparison preserving the total particle number, while at fixed odd particle number one fermion remains unpaired, as required by parity. Within the resulting paired subspace, minimization over positive generalized-$P$ representations of pseudospin coherent states selects a common phase for the nonzero pair amplitudes and determines their magnitudes. BCS and number-projected BCS states arise as particular minimizers under the respective particle-number constraints.

cond-mat.supr-con↗

Ab initio Engineering of Topological Superconductivity in Substrate-Supported Magnetic Chains

In this paper, we employ a combination of several first-principles-based approaches to systematically scan chains of several types of magnetic atoms in different chain geometries on top of various overlayers on the Nb(110) surface, to develop a pathway to engineer topological Majorana edge states. Using atomistic spin-model calculations, we first identify a rich variety of magnetic configurations, including ferromagnetic, antiferromagnetic, spin-spiral, and exotic up-up-down-down states stabilized solely within a second-order Heisenberg model. In the next step, we solve the Bogoliubov-de Gennes equations using a band-theoretical method. Depending on the magnetic structure and atomic configuration, the superconducting spectra may exhibit topologically trivial minigaps, broad overlapping Shiba bands, or localized zero-energy edge states accompanied by sizable topological minigaps. In particular, Fe chains on Ag/Nb(110) are found to display pronounced signatures associated with topological superconductivity, including strongly localized zero-energy edge states and indications of Shiba band inversion. We further demonstrate that changes in the spin structure or a small amount of spatial relaxation can substantially modify the superconducting band structure and the localization properties of the edge states.

cond-mat.supr-con↗

Atomic-scale Imaging of Time Reversal Breaking Supercurrents in $\mathrm{Bi_2Sr_2CaCu_2O_8}$

A persistent mystery of CuO_2-based high temperature superconductivity has been the mechanism and consequences of time reversal ($\mathcal{T}$)-symmetry breaking. Although theoretical hypotheses focus upon persistent electronic currents as a potential cause, these have been invisible to direct experimental probes. That impasse has recently been removed by development of techniques for atomic-scale supercurrent visualization. Here, by imaging the energy-resolved density of electronic states $N(r,E)$ in $\mathrm{Bi_2Sr_2CaCu_2O_8}$, we search for the specific signatures of flowing supercurrents in a disordered $d$-wave superconductor. Employing the strategy of artificial neural networks (ANN) for complex quantum matter visualization, we designed a dual-ANN (VMAP) to efficiently implement $d$-wave supercurrent visualization, achieving imaging of supercurrent speeds $V(r)$ at magnetic fields $|B| \leq μT$ in $\mathrm{Bi_2Sr_2CaCu_2O_8}$. Correspondence between specific characteristics of $N(r,E)$ around each supercurrent loop and a quantum "Rosetta Stone" which translates $N(r,E)$ to $V(r)$, emphasizes the validity of $V(r)$ detection and visualization. Then, for hole-densities 0.17 $\leq p \leq$ 0.22 we observe statistically repeatable networks of supercurrent flows. Crucially, these supercurrents satisfy Kirchhoff's law at junctions within each network. Further, theoretical predictions for supercurrent speeds $V$ in disordered $d$-wave superconductors are consistent with measured 1250 m/s $< V <$ 4000 m/s, and with reported magnetic field amplitudes $|B| \approx$ 0.1 mT Thus, detection and visualization of these supercurrent loop networks provides unique perspectives on $\mathcal{T}$-symmetry breaking in cuprates. It also emphasizes the accelerating utility of ANN techniques for physics discovery from high data volume quantum matter research.

cond-mat.supr-con↗