Developing a Complete AI-Accelerated Workflow for Superconductor Discovery

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Gibson, Jason B., Hire, Ajinkya C., Prakash, Pawan, Dee, Philip M., Geisler, Benjamin, Kim, Jung Soo, Li, Zhongwei, Hamlin, James J., Stewart, Gregory R., Hirschfeld, P. J., Hennig, Richard G.
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912968944386048
author Gibson, Jason B.
Hire, Ajinkya C.
Prakash, Pawan
Dee, Philip M.
Geisler, Benjamin
Kim, Jung Soo
Li, Zhongwei
Hamlin, James J.
Stewart, Gregory R.
Hirschfeld, P. J.
Hennig, Richard G.
author_facet Gibson, Jason B.
Hire, Ajinkya C.
Prakash, Pawan
Dee, Philip M.
Geisler, Benjamin
Kim, Jung Soo
Li, Zhongwei
Hamlin, James J.
Stewart, Gregory R.
Hirschfeld, P. J.
Hennig, Richard G.
contents The quest to identify new superconducting materials with enhanced properties is hindered by the prohibitive cost of computing electron-phonon spectral functions, severely limiting the materials space that can be explored. Here, we introduce a Bootstrapped Ensemble of Equivariant Graph Neural Networks (BEE-NET), a machine-learning model trained to predict the Eliashberg spectral function and superconducting critical temperature with a mean-absolute-error of 0.87 K relative to DFT-based Allen-Dynes calculations. Intriguingly, BEE-NET achieves a true-negative-rate of 99.4\%, enabling highly efficient screening for the rare property of superconductivity. Integrated into a multi-stage, AI-accelerated discovery pipeline that incorporates elemental-substitution strategies and machine-learned interatomic potentials, our workflow reduced over 1.3 million candidate structures to 741 dynamically and thermodynamically stable compounds with DFT-confirmed $T_{\mathrm{c}} > 5$ K. We report the successful synthesis and experimental confirmation of superconductivity in two of these previously unreported compounds. This study establishes a data-driven framework that integrates machine learning, quantum calculations, and experiments to systematically accelerate superconductor discovery.
format Preprint
id arxiv_https___arxiv_org_abs_2503_20005
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Developing a Complete AI-Accelerated Workflow for Superconductor Discovery
Gibson, Jason B.
Hire, Ajinkya C.
Prakash, Pawan
Dee, Philip M.
Geisler, Benjamin
Kim, Jung Soo
Li, Zhongwei
Hamlin, James J.
Stewart, Gregory R.
Hirschfeld, P. J.
Hennig, Richard G.
Superconductivity
Materials Science
The quest to identify new superconducting materials with enhanced properties is hindered by the prohibitive cost of computing electron-phonon spectral functions, severely limiting the materials space that can be explored. Here, we introduce a Bootstrapped Ensemble of Equivariant Graph Neural Networks (BEE-NET), a machine-learning model trained to predict the Eliashberg spectral function and superconducting critical temperature with a mean-absolute-error of 0.87 K relative to DFT-based Allen-Dynes calculations. Intriguingly, BEE-NET achieves a true-negative-rate of 99.4\%, enabling highly efficient screening for the rare property of superconductivity. Integrated into a multi-stage, AI-accelerated discovery pipeline that incorporates elemental-substitution strategies and machine-learned interatomic potentials, our workflow reduced over 1.3 million candidate structures to 741 dynamically and thermodynamically stable compounds with DFT-confirmed $T_{\mathrm{c}} > 5$ K. We report the successful synthesis and experimental confirmation of superconductivity in two of these previously unreported compounds. This study establishes a data-driven framework that integrates machine learning, quantum calculations, and experiments to systematically accelerate superconductor discovery.
title Developing a Complete AI-Accelerated Workflow for Superconductor Discovery
topic Superconductivity
Materials Science
url https://arxiv.org/abs/2503.20005