ZIF-derived N,P,S tri-doped hollow carbon nanofibers for electrochemical determination of tert-butylhydroquinone with machine learning-assisted calibration

ZIF-derived N,P,S-HCNFs provide a porous electroactive interface for TBHQ sensing. • Single- and dual-doped controls with DFT support cooperative heteroatom regulation. • Two linear ranges of 0.9–12 and 12–127 μM were achieved with a 0.075 μM LOD. • Reliable TBHQ quantification was demonstrated in spiked edible-oil samples. • Extra Trees enables validated auxiliary calibration in complex oil matrices. Tert-butylhydroquinone (TBHQ) is a widely used synthetic phenolic antioxidant, and its excessive use in edible oils raises food-safety concerns. Herein, N, P, and S tri-doped hollow carbon nanofibers (N,P,S-HCNF) with interconnected porous channels were fabricated via an electrospinning–pyrolysis strategy. The hollow architecture and heteroatom-induced defects provide abundant electroactive sites and facilitate electron/mass transfer, enabling sensitive electrochemical oxidation of TBHQ. The N,P,S-HCNF/CP sensor exhibited two linear response ranges of 0.9–12 μM and 12–127 μM, with a detection limit of 0.075 μM. Acceptable selectivity, reproducibility, and storage stability were obtained. Single- and dual-doped controls combined with DFT calculations provided mechanistic insight into the enhanced TBHQ adsorption and interfacial electron transfer of N,P,S-HCNF, including preferential interactions involving the phenolic hydroxyl groups. An externally validated Extra Trees calibration model was further employed as an auxiliary strategy to reduce matrix-related prediction bias, integrating hollow carbon nanofiber engineering, multielement heteroatom regulation, mechanistic analysis, and data-driven quantification for TBHQ electrochemical determination in edible oils. A defect-rich N,P,S-doped hollow carbon nanofiber interface enabled sensitive electrochemical oxidation of TBHQ, while DFT supported favorable adsorption and interfacial charge redistribution and machine learning provided auxiliary calibration for improved quantification in edible-oil matrices. Download: Download high-res image (128KB) Download: Download full-size image