Vacancy-cluster-driven cation self-diffusion in <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msub> <mml:mi>UO</mml:mi> <mml:mn>2</mml:mn> </mml:msub> </mml:math> and <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msu

Cation self-diffusion plays a key role in the microstructure evolution of a wide range of materials, including oxide nuclear fuels. Growing evidence indicates that small defect clusters can make a substantial contribution to cation transport in these oxides. Quantifying this contribution is nevertheless challenging because of complex defect chemistry, strong correlations between cation and anion migration, and the existence of multiple competing migration mechanisms. In this work, we compute bulk cation self-diffusion coefficients in stoichiometric ${\mathrm{UO}}_{2}$ and ${\mathrm{PuO}}_{2}$ by explicitly accounting for small vacancy clusters (cation--anion divacancies and Schottky defects) in addition to isolated cation vacancies. Our approach combines a systematic exploration of cluster migration pathways using ART-nouveau with the calculation of cluster transport coefficients through the KineCluE code, which rigorously incorporates the influence of local atomic environments and kinetic correlations on defect migration. These transport coefficients are then coupled with equilibrium cluster concentrations to obtain self-diffusion coefficients directly comparable with experimental data. We find that small vacancy clusters are significantly more mobile than isolated cation vacancies and dominate the overall diffusion process. In ${\mathrm{UO}}_{2}$, the resulting self-diffusion coefficients are consistent with the most reliable experimental datasets, whereas in ${\mathrm{PuO}}_{2}$ the predictions are discussed in light of the limited data currently available. Although vacancy clusters exhibit higher mobility in ${\mathrm{PuO}}_{2}$, the overall self-diffusion coefficient is lower than in ${\mathrm{UO}}_{2}$ because of the larger cation vacancy formation energy predicted by our defect model. Beyond actinide oxides, this work demonstrates how defect clusters can be systematically integrated into diffusion calculations and provides transferable insights and input data for fuel behavior models.