SciGroveBeta
Materials

AI-accelerated metallized -bonding screening for superconductor discovery

Zechen Tang, Wen-Han Dong, Baochun Wu, Jian-Feng Zhang, Yuxiang Wang, Yang Li, Honggeng Tao, Qiyu Zeng, Chong Wang, Chen Si, Zhong-Yi Lu, Wenhui Duan, Tao Xiang, Yong Xu

Featured July 4, 2026

Original Paper

This analysis was generated by SciGrove. Upload your own PDFs or enter a DOI — and get the same AI breakdown on any paper.

Get started

AI-generated analysis — This is SciGrove's AI interpretation of the paper, not peer-reviewed content. Always refer to the original paper.

Simply

A new AI-powered shortcut helps scientists quickly find materials that become superconductors at high temperatures by focusing on a special kind of electron bond instead of slow, traditional calculations.

In depth
The authors introduce a novel, efficient descriptor called σ-bonding density of states (σDOS) to identify high-transition-temperature superconductors from standard electronic structure calculations, circumventing the computationally expensive Density Functional Perturbation Theory (DFPT). This evaluation is further accelerated by integrating a deep-learning DFT Hamiltonian (DeepH) method, enabling rapid, large-scale screening of millions of materials for promising candidates.

Key Takeaways

  • 1
    The study proposes σ-bonding density of states (σDOS) as a computationally inexpensive yet reliable descriptor for identifying high-temperature superconductors, bypassing costly DFPT calculations.
  • 2
    The evaluation of σDOS is significantly accelerated by a deep-learning DFT Hamiltonian (DeepH), enabling high-throughput screening of millions of materials.
  • 3
    Applying this physics-guided, AI-accelerated workflow, the authors successfully identify B$_{13}$Se as a novel ambient-pressure superconductor candidate with a predicted transition temperature () exceeding 40 K.

Conceptual Flow

HIGH LEVEL
1
Methodology: Faster Superconductor Search

Instead of slow, traditional calculations, the authors use a smart shortcut based on electron bonds, sped up even more by AI, to find new superconductors.

Material Structure
AI-Accelerated Bond Analysis
Superconductor Score
2
Results: New Superconductor Family

This new method quickly found a whole family of materials, including a specific one called B13Se, that are predicted to be good superconductors.

Millions of Materials
Find Best Candidates
B13Se Superconductor
B13X Family