Zirconium alloys, which are widely used as cladding materials in nuclear reactors, are prone to react with oxygen (O). Furthermore, the ω-Zr in zirconium alloys can significantly increase the strength and hardness of these alloys, but there is a lack of reports on the behavior of oxygen in ω-Zr in the current literature. To investigate their interactions, we have studied the behavior of O in ω-Zr using the first-principles approach. In this work, we examined the effects of vacancy and alloying elements (Nb, Sn) on the behavior of O in ω-Zr. The results show that O with a formation energy of −5.96 eV preferentially occupies an octahedral interstitial position in ω-Zr. A vacancy reduces the formation energy of O in a tetrahedral interstitial position in ω-Zr. Nb and Sn decrease the formation energy of O in the octahedral interstitial position by 6.16 eV and 5.08 eV. Vacancy effectively reduces the diffusion barrier of O around it, which facilitates the diffusion of O in ω-Zr. Nb and Sn preferentially occupy the 1b and 2d substitution sites in ω-Zr, respectively. Nb makes the diffusion barrier of O in ω-Zr lower and promotes the diffusion of O in ω-Zr. Moreover, Sn makes the diffusion of O around Sn difficult. It was further found that O is less prone to form clusters in ω-Zr and tends to independently occupy interstitial positions in ω-Zr. In particular, a single vacancy would make the binding energy between O atoms to be further reduced.

Zirconium monoxide (ZrO) plays a key role in the water-side corrosion resistance of Zr alloys as cladding materials in nuclear reactors. This study investigates the behavior of intrinsic defects in ZrO through first-principles calculations, and the influence of main alloying elements (Cr, Fe, Nb and Sn) is also evaluated. We focus on the formation and migration properties of vacancies and interstitials. The results show that the formation energy of oxygen vacancy is 5.31 eV. The formation energy of interstitial O_i-tet in ZrO is −4.04 eV, indicating that O_i-tet can be formed spontaneously. Another interstitial oxygen O_i-mid with a formation energy of 0.03 eV can also be found in large quantities in ZrO. As for the migration properties, oxygen vacancy in ZrO without doping tends to diffuse along Path 2, and the diffusion barrier is 2.96 eV. Cr and Fe reduce the migration barriers of oxygen vacancies, while Nb and Sn increase them. In contrast, alloying elements generally hinder the formation of oxygen interstitials and increase their migration barriers, particularly in the case of Cr and Fe. The migration barrier of interstitial oxygen diffusion along Path a in pure ZrO is 2.91 eV. However, the migration barriers of interstitial oxygen in ZrO with Cr or Fe doping could increase to more than 4 eV. These findings provide critical insights into the role of alloying elements in modifying defect dynamics, offering a theoretical basis for improving the corrosion resistance and performance of zirconium alloys in practical applications.

Molybdenum (Mo) is a key structural material for advanced nuclear systems, yet under irradiation, the formation of helium (He) bubbles and dislocation loops leads to severe degradation. Understanding He–dislocation loop interactions at the atomic scale is essential for predicting long-term performance. In this work, we develop a machine-learning Moment Tensor Potential and perform molecular dynamics simulations to systematically study the interactions between He atoms and the primary interstitial-type dislocation loops (1/2 ⟨111⟩ and ⟨100⟩) in BCC Mo. Our results show that dislocation loops act as strong trapping sites for He, modifying its spatial distribution and promoting heterogeneous nucleation of He clusters, with the ⟨100⟩ loop having a stronger effect. Conversely, He enhances loop growth by absorbing self-interstitial atoms, a process that intensifies with He concentration. A pronounced pinning effect of He on loop glide is observed, effectively immobilizing loops at 600 and 1200 K, though thermal activation at 1800 K partially releases this pinning. This study elucidates the mutual enhancement mechanism between He and dislocation loops, offering fundamental insights for designing radiation-resistant Mo alloys.

Machine learning (ML) has emerged as a powerful tool for accelerating the design of refractory high-entropy alloys (RHEAs), yet the black-box nature of conventional models often limits physical understanding. Herein, an interpretable phase-prediction framework based on Kolmogorov-Arnold Networks (KAN) is developed to reveal the governing principles of phase formation in RHEAs. KAN achieves strong predictive performance, with an F1 score of 0.95 for solid-solution classification. Analysis of the learned activation functions shows that complex phase-selection behavior can be approximated by simple low-order mathematical functions. Feature-neuron connections identify atomic size difference, mixing enthalpy, and electronegativity difference as the dominant descriptors. The extracted nonlinear relationships also improve the predictive performance of the logistic regression models. This work highlights KAN as an interpretable framework for alloy design and provides a route for gaining data-driven physicochemical insights from complex materials datasets.