Defect–Interface Engineering of Machine Learning-Enhanced Au–VSe2–MXene Nanocomposites with Tuned Electronic States for Trace Antibiotic Sensing

Ensuring food safety requires engineered hybrid composites with hierarchical interfacial architectures capable of detecting trace antibiotics in complex food and environmental matrices. Here, we report an electrochemical sensing platform based on gold nanoparticle (AuNP)-decorated VSe ₂ -MXene hybrid nanocomposite for on-site detection of furazolidone (FZD). The VSe ₂ -MXene scaffold, synthesized via a one-step hydrothermal method, enables in situ anchoring and near-atomic-scale AuNP deposition through chemical reduction. This structural integration induces lattice compression, generating abundant active sites and enhanced electron mobility that facilitate efficient interfacial charge transfer. Comprehensive physicochemical analyzes show that hydrothermal synthesis followed by chemical reduction enables uniform AuNP anchoring on the VSe ₂ -MXene framework while converting surface M–OH groups into M–O bonds. This interfacial electronic modulation enhances electron-transfer kinetics and selective FZD adsorption. The resulting Au-VSe ₂ -MXene sensor exhibits significantly reduced charge-transfer resistance (~ 45 Ω) and an enhanced heterogeneous electron-transfer rate (1.68 × 10 ^− 2 cm s ^− 1 ), enabling a low cathodic peak potential of -0.40 V (vs. Ag/AgCl). The sensor demonstrates a wide linear detection range (6-255 nM, R ²  = 0.9383), an ultralow detection limit (0.21 nM), high sensitivity (0.597 µA nM ^− 1 cm ^− 2 ), and excellent long-term stability (> 30 days), enabling reliable trace-level FZD detection in real samples even in the presence of common interferents. Machine learning-driven analysis of amperometric i-t currents enabled precise FZD quantification, reducing calibration errors; the Random Forest model (R ²  = 0.997) integrated with IoT supports real-time monitoring for diagnostic and environmental applications.