Multidimensional Engineering of Single-Atom Iron Nanozymes for Machine Learning-Assisted Multiplexed Sensing

Single-atom nanozymes have been recognized as promising alternatives to natural metalloenzymes, but challenges remain in precise structure engineering for efficient biochemical reactions. Herein, we demonstrate a facile strategy for the efficient production of iron single-atom nanozymes with multidimensional coordination engineering, where atomically dispersed iron sites are modulated by axial F and first-shell B (Fe-N₃B₁/F). Combined theoretical and experimental studies reveal that the introduction of F and B could effectively break the stringent linear scaling relations of conventional Fe-N₄ sites, converting a prohibitive *OH desorption penalty into a spontaneous exergonic process, thereby accelerating H₂O₂ activation and reactive oxygen species generation. Benefiting from the excellent peroxidase-like activity of Fe-N₃B₁/F nanozymes, we developed a cross-reactive sensor array. Coupled with machine learning algorithms, the sensor platform could achieve the intelligent pattern recognition of four homologous tetracycline (TC) antibiotics with about 80% classification accuracy. Moreover, integrating this sensing array with a smartphone-based paper analytical device could enable the rapid and instrument-free quantification of TCs in real environmental water samples. The present study highlights the significance of dimensionality engineering and establishes a multifunctional paradigm for environmental monitoring.