📝 SummarXiv

Abstract

Radiation damage in structural materials is a major challenge for advanced nuclear energy systems, and niobium is of particular interest due to its high melting point, mechanical strength, and corrosion resistance. To better understand its radiation response, we carried out large-scale molecular dynamics simulations of collision cascades in pure niobium at 300 K over a primary knock-on atom (PKA) energy range of 1-75 keV, employing four interatomic potentials: an embedded atom method (EAM), two Finnis-Sinclair models (FS-1 and FS-2), and a machine learning-based spectral neighbor analysis potential (SNAP) we developed. All reproduce the general features of cascade formation but differ significantly in defect production, clustering, and morphology. At low energies, defect generation follows trends governed by threshold displacement energy (TDE) and the stiffness-to-range ratio (|S/R|). At higher energies, subcascade formation makes defect evolution dependent on the combined effects of |S/R|, average TDE, and other material-specific factors. Vacancy clustering dominates over interstitial clustering across all cases: EAM produces the largest vacancy clusters and the highest clustering fraction, while SNAP shows the strongest interstitial clustering. Morphological analysis indicates that EAM forms a balanced mix of 1/2<111>, 1/2<110> loops, C15 rings, and hybrid structures; FS-2 favors extended 1/2<111> dumbbells, crowdions, and dislocation loops; whereas FS-1 and SNAP generate more compact or disordered clusters, with SNAP produces a high fraction of C15-like rings (maximum size up to nine atoms) that may evolve into dislocation loops of 1/2<111> and <100>. These findings give clear insights into how niobium reacts when exposed to irradiation, especially at high energies.