Machine learning-assisted triple-emission Ln-MOFs sensor array for detection of multiple PFCs in aqueous environments.

Perfluorinated compounds (PFCs) are persistent environmental pollutants with potential carcinogenicity, posing a major threat to ecosystems and human health. Rapid identification of PFCs in complex environmental matrices remains challenging due to the limitations of conventional single-emission probes in their sensitivity to environmental disturbances. This study developed a three-channel fluorescence sensing platform based on a europium/terbium (Eu/Tb) bimetallic organic framework. By using 2,3,5,6-tetrafluoroterephthalic acid and 1,10-phenanthroline as dual ligands to coordinate europium/terbium ions to construct a new fluorescent probe, the triple characteristic emission of Eu, Tb and ligands was achieved under a single-wavelength excitation, effectively overcoming the drawback that traditional single-emission probes are easily disturbed by environmental variables. The array exhibits a wide linear detection range of 0.1-100 μM for six PFCs, with a detection limit of 42 nM. Hierarchical clustering analysis and linear discriminant analysis accurately identified the six PFCs and their mixtures. Furthermore, the stepwise prediction model developed by integrating machine learning algorithms enables concentration-independent qualitative identification of six PFCs, while also achieving precise quantitative determination. The high accuracy achieved in blind sample validation further confirms the feasibility of the model for practical applications. This technology achieves rapid and high-precision monitoring of trace persistent pollutants through multi-signal collaboration and intelligent analysis, which is highly significant for environmental risk management and pollution reduction.