Machine Learning-Driven Biomimetic MXene-Based Sensor Array for Ultraefficient Multiplexed Oxyanions Detection

The pervasive contamination of groundwater by highly toxic oxyanions necessitates advanced sensing technologies capable of high-throughput screening. To overcome the intrinsic single-analyte limitation of conventional ion-selective electrodes (ISEs), we developed a biomimetic sensor array inspired by the cross-reactive principles of biological taste systems, which was constructed by self-assembling quaternary ammonium salts (QACs) with tailored alkyl chain lengths (Cn) within MXene interlayers (MXene/QAC-Cn). When integrated with a machine learning model, the array successfully enabled the simultaneous identification of four kinds of oxyanions (CrO42-, SeO42-, BrO3-, and ClO4-) with a discrimination accuracy of 97.1%. The sensor demonstrated exceptional sensitivity, achieving a detection limit of 1.2 × 10-9 M, and a 32-fold enhancement in analytical throughput compared to single-analyte approaches. Theoretical calculations revealed that the confined microenvironment of the MXene/QAC-Cn composite promotes partial dehydration of the oxyanions, reducing the adsorption energy by 1.09 eV. This mechanism is pivotal for generating high-quality and differentiable response fingerprints, which in turn ensures the high accuracy of the machine learning model. The programmable design of this platform underscores its potential for extension to a broader spectrum of oxyanions, offering a versatile and scalable solution for comprehensive water quality monitoring.