Machine Learning-Accelerated Molecular Dynamics Simulations of PuF3 in LiF-NaF-KF: Unraveling Coordination Chemistry and Ionic Transport Mechanisms.

A thorough understanding of the microstructural evolution and thermophysical properties of Pu-containing molten salts is crucial for optimizing the performance of molten-salt reactor (MSR) fuel salts and spent-fuel reprocessing. This study combines active learning strategies with deep potential molecular dynamics to investigate the relationship between microstructure and thermophysical properties of the FLiNaK-PuF3 molten salt system. First, a high-precision deep potential model was constructed, with density predictions in excellent agreement with experimental data, validating the model's reliability. Microstructural analysis revealed that the relative interaction strength of ion pairs follows the order Pu-F > Li-F > Na-F > K-F, weakening with increasing PuF3 concentration or temperature. Thermodynamic studies showed that the system's density decreases linearly with increasing temperature, whereas heat capacity (Cp) remains constant over the 773-1173 K range. With increasing PuF3 concentration, the density of the molten salt system rises significantly, while Cp decreases notably. Kinetic analysis found that the self-diffusion coefficients of the five ions follow the order Li+ > Na+ > F- > K+ > Pu3+, all monotonically decreasing with increasing PuF3 concentration. This work systematically elucidates the correlation between microstructural features and macroscopic properties of the FLiNaK-PuF3 molten salt system, particularly elucidating the coordination chemistry of Pu3+ ions in fluoride molten salts and their influence on mass transport mechanisms. The results provide important theoretical insights into the composition-property relationships of MSR fuel salts, offering guidance for performance optimization.