Chiral hierarchical porous ZIF-8/au@ag bipyramids integrated with deep learning for ultrasensitive detection and discrimination of mixed pesticide residues

The widespread use of combined pesticides has led to increasing concerns over mixed pesticide residues in food, posing significant challenges for rapid and reliable detection. Herein, we develop an intelligent surface-enhanced Raman spectroscopy (SERS) platform by integrating a bichiral nanostructured substrate with deep learning for multiplex pesticide analysis. A hierarchical porous chiral ZIF-8 (HPZ) was engineered using L-histidine as a pore-expanding agent and assembled with L-cysteine-directed Au@Ag bipyramids to form the HPZA substrate, enabling synergistic adsorption enrichment and electromagnetic enhancement. The resulting platform exhibits high sensitivity and stability, achieving a detection limit of 9.54 ng/mL for quinalphos in citrus samples. Moreover, when coupled with a fully connected neural network (FCNN), it enables accurate discrimination of multiple pesticides and their mixtures, reaching an identification accuracy of 93.33%. This work presents an integrated material–algorithm strategy for rapid and intelligent detection of pesticide residues, offering a promising approach for food safety monitoring.