| Abstract Scope |
Understanding how molten salt systems change under extreme conditions is crucial for advanced nuclear reactors and pyrochemical reprocessing. Traditional methods can’t track these fast, high-temperature chemical changes in real time. Here, we introduce a new approach that combines laser-induced breakdown spectroscopy (LIBS) with electrochemical measurements, allowing in situ, real-time analysis of molten salts. By using a probabilistic neural network to merge spectroscopic and electrochemical data, the system separates true signals from noise and improves the accuracy of identifying and measuring chemical components. This integrated method addresses challenges like overlapping signals and localized sampling, delivering reliable, uncertainty-aware information about molten salt chemistry. The framework shows how pairing advanced spectroscopy with machine learning can reveal new details about chemical changes, impurities, and material transport in nuclear environments, and points toward more advanced diagnostic tools for next-generation nuclear systems. |