Polyfluoroalkyl Substance-Induced Nanoclusters Immobilized MOF-on-MOF Architecture Dissociation-Driven Machine Learning-Assisted Ratio Fluorescence Sensor Array

The widespread use of polyfluoroalkyl substances (PFASs) threatens the ecosystem and human health, while their rapid multicomponent identification remains challenging. Here, for the first time, a metal nanocluster (MNC)-immobilized metal-organic framework (MOF) anchored on MOF (ZIF-on-MIL) architectures (M@ZIF-on-MIL) assembled machine learning-assisted ratio fluorescence sensor array was developed for the rapid, specific screening and discrimination of multiple PFASs. Wherein, AuNCs and AgAuNCs were confined on the surface of ZIF-on-MIL, which exhibited significantly enhanced fluorescent features. Interestingly, due to the interactions of PFASs and ZIF-8 in the M@ZIF-on-MIL, the PFASs can induce M@ZIF-on-MIL dissociation to produce diverse ratio fluorescent "fingerprints", which were further identified by pattern recognition methods. The array, integrated with machine learning algorithms, successfully identified and distinguished all eight PFAS species with 100% accuracy. Additionally, it utilizes statistical analysis methods to achieve the highly accurate detection of individual PFASs across various concentrations and mixtures of PFASs at different ratios. More importantly, this method proves to be highly effective in detecting PFASs in natural waters from various sources, highlighting its potential for practical applications. This study provides a pioneering ratio fluorescence sensor array for the rapid PFAS screening and lays the groundwork for the application of MNC-immobilized MOF-on-MOF architectures in array technology fields.