Machine learning potential-based study of irradiation defect evolution in U–Mo–Nb ternary alloys

As candidate materials for nuclear reactors, uranium alloys are prone to irradiation-induced defects under neutron exposure. Molybdenum and niobium serve as effective stabilizers because of their favorable solubility and compatibility with U. This study investigates the evolution of irradiation defects in the U–Mo–Nb (uranium–molybdenum–niobium) ternary alloy fuel using molecular dynamics simulations based on a newly developed machine learning potential. We systematically simulated cascade damage under varying temperatures, primary knock-on atom (PKA) energies, and incident angles to analyze their impact on defect production. Our results show that the peak defect count varied by 7.6% with the incident angle. Higher temperatures increase the initial peak defect count but enhance the subsequent annealing, reducing the final number of defects. Higher PKA energies lead to more defects and longer time-to-peak. Crucially, the U–Mo–Nb alloy demonstrated superior irradiation resistance compared with the U–Mo and U–Nb alloys. It produces only ∼75% of the defects in the U–Mo binary alloy at similar temperatures and 50% at the same PKA energy, while also forming smaller vacancies and interstitial clusters. These findings highlight the exceptional potential of the U–Mo–Nb ternary alloy for advanced nuclear fuel applications.