Machine Learning Prediction of BCS Superconductors without BCS Theory

Area of Science:

• Materials Science
• Condensed Matter Physics
• Computational Chemistry

Background:

• Superconducting critical temperature (Tc) is key for technological applications.
• BCS superconductors' pairing mechanism is understood via Eliashberg function and McMillan equation.
• First-principles calculations for predicting BCS superconductors are computationally expensive.

Purpose of the Study:

• To develop a machine learning (ML) framework for rapid screening of potential BCS superconductors.
• To provide a computationally inexpensive alternative to density functional theory (DFT) for BCS superconductor prediction.
• To accelerate the discovery of novel BCS superconductors.

Main Methods:

• Curated a database of experimentally relevant crystal structure data.
• Selected material descriptors suitable for experimental data.
• Trained ML models to classify materials as BCS superconductors.

Main Results:

• Demonstrated that ML approaches can expedite BCS superconductor discovery.
• ML provides a faster, computationally inexpensive alternative to DFT.
• Successfully trained models to predict BCS superconductivity.

Conclusions:

• ML framework accelerates the discovery of novel BCS superconductors.
• Eliminates the need for expensive first-principles calculations.
• Offers a roadmap for superconductor discovery beyond traditional BCS theory.