Kinetic and Mechanistic Discrepancies of Single/Dual-Atom Nanozymes Drive a Triple-Channel Sensing Array for Machine Learning-Assisted Antioxidant Discrimination

Current colorimetric sensing arrays for antioxidant detection often struggle with discrimination due to cross-reactive signals from individual nanozymes. These signals are typically modulated by external factors such as pH or chromogenic substrates, offering limited kinetic and mechanistic diversity. To overcome this, we present a novel triple-channel colorimetric sensing array utilizing two distinct single-atom nanozymes (Cu SA and Fe SA) and one dual-atom nanozyme (CuFe DA). Our approach leverages the inherent kinetic and mechanistic differences between these atomically dispersed catalysts. We observed measurably distinct oxidase-like activities through variations in their Michaelis–Menten constants (KmKm) and specific activities. Furthermore, detailed mechanistic studies revealed differences in their active metal sites, leading to varied reactive oxygen species (ROS) production. This intrinsic functional divergence creates unique “fingerprint” responses for improved antioxidant differentiation. When integrated with machine learning algorithms (Principal Component Analysis and Hierarchical Cluster Analysis), our array successfully identified and quantified six common antioxidants: ascorbic acid, glutathione, cysteine, tea polyphenol, gallic acid, and tannin. The array exhibited excellent sensitivity, with a low detection limit of 2.01 μM for cysteine. This research offers a robust strategy for developing high-performance sensing arrays by exploiting fundamental atomic-scale kinetic and mechanistic variations, holding significant promise for food safety and health monitoring applications.