Machine-Learned Potentials for Radiation Damage in Superconductors
Research Paper#Materials Science, Superconductivity, Radiation Damage, Machine Learning🔬 Research|Analyzed: Jan 3, 2026 09:29•
Published: Dec 30, 2025 19:21
•1 min read
•ArXivAnalysis
This paper addresses the critical need for accurate modeling of radiation damage in high-temperature superconductors (HTS), particularly YBa2Cu3O7-δ (YBCO), which is crucial for applications in fusion reactors. The authors leverage machine-learned interatomic potentials (ACE and tabGAP) to overcome limitations of existing empirical models, especially in describing oxygen-deficient YBCO compositions. The study's significance lies in its ability to predict radiation damage with higher fidelity, providing insights into defect production, cascade evolution, and the formation of amorphous regions. This is important for understanding the performance and durability of HTS tapes in harsh radiation environments.
Key Takeaways
- •Machine-learned interatomic potentials (ACE and tabGAP) provide more accurate modeling of radiation damage in YBCO compared to empirical models.
- •The models accurately reproduce DFT energies and predict cascade evolution, including defect production and recombination.
- •Total defect production is weakly dependent on oxygen stoichiometry, offering insights into the robustness of radiation damage processes.
- •Simulations reveal amorphous regions comparable to the superconducting coherence length, consistent with experimental observations.
Reference / Citation
View Original"Molecular dynamics simulations of 5 keV cascades predict enhanced peak defect production and recombination relative to a widely used empirical potential, indicating different cascade evolution."