Noble gas electrochemistry-driven luminescence integrated with machine learning for environmental monitoring.

Noble gases are traditionally known as not being easily involved in chemical reactions owing to their stable electronic configuration. Here, we study the electrochemical redox and electrochemiluminescence (ECL) mechanisms of a series of noble gases (radon, xenon and krypton) for the first time and demonstrate their ability to produce an ECL signal with an emission peak in the range of 600-640 nm. Among them, the key intermediates Rn˙+/Xe˙+ are observed via HPLC-MS to confirm their electrochemical redox and ECL process. Additionally, the anodic/cathodic ECL behavior and the ECL intensity of radon in an annihilation state are confirmed to depend on the interaction between solvent molecules and Rn˙-/Rn˙+, which can be summarized as a free radical-solvent interaction-dependent ECL (FSDE) mechanism. A machine learning framework with SHapley additive exPlanations analysis is also integrated to enable high-fidelity differentiation of Rn/Xe for accurate determination. This work advances our knowledge of the electrochemistry of noble gases and develops a machine-learning-assisted trace radon/xenon monitoring method with high selectivity and sensitivity. The ECL behavior is successfully employed for noble gas determination in various real-world atmospheric samples, indicating its application potential in the fields of environment, geoscience, and public health.