Defect–interface engineering of machine learning-enhanced Au–VSe2–MXene nanoarchitectures with modulated electronic structure for trace antibiotic sensing

Le résumé fourni par la source

Ensuring food safety requires engineered hybrid composites with hierarchical interfacial architectures for trace antibiotic detection in complex food and environmental matrices. Herein, we report an electrochemical sensing platform based on a gold nanoparticle (AuNP)-decorated VSe 2 -MXene hybrid nanocomposite for on-site furazolidone (FZD) detection. The VSe 2 -MXene scaffold was synthesized via a one-step hydrothermal route, followed by in situ anchoring of near-atomic-scale AuNPs through chemical reduction. This structural integration induces lattice compression, generating abundant electroactive sites and enhanced electron mobility for efficient charge transfer. Physicochemical characterization confirmed uniform AuNP anchoring on the VSe 2 -MXene framework and the conversion of surface M–OH groups into M–O bonds, resulting in enhanced electron-transfer kinetics and selective FZD adsorption. Consequently, the Au-VSe 2 -MXene sensor exhibited low charge-transfer resistance (~ 45 Ω) and an enhanced heterogeneous electron-transfer rate constant (1.68 × 10 − 2 cm s − 1 ), enabling sensitive FZD detection at a low cathodic potential of -0.4 V (vs. Ag|AgCl). The fabricated sensor demonstrated a wide linear 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. Furthermore, machine learning-assisted analysis of amperometric i-t responses enabled accurate FZD quantification with reduced calibration error. Integration of the Random Forest model (R 2 = 0.9997) with IoT-based monitoring further highlights the potential of this platform for real-time diagnostic and environmental sensing applications.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.