Rapid Detection of Aflatoxin B1 in Peanut Oil by Surface-enhanced Raman Spectroscopy
Yue-Xiang Zhang,Yong-Yu Li,Yan-Kun Peng,Shao-Jin Ma,Wei Wang,Ji-Feng Wu
DOI: https://doi.org/10.19756/j.issn.0253-3820.231291
IF: 1.193
2023-01-01
Chinese Journal of Analytical Chemistry
Abstract:The characteristic Raman shifts at 544, 691, 1278 and 1621 cm-1 of aflatoxin B1 (AFB1) in peanut oil were determined by comparative analysis of surface-enhanced Raman spectra of AFB1 standard reserve solution and other surface-enhanced Raman spectra in this study, based on the self-developed surface-enhanced Raman spectral acquisition system for automated mixing of liquid samples in the laboratory. The surface-enhanced Raman characteristic information of AFB1 in peanut oil was obtained directly by using QuEChERS (Quick, easy, cheap, effective, rugged and safe) pretreatment technique with KI solution at a concentration of 1.2 mol/L as a coagulant promoter with a detection limit of 0.02 mg/kg, and the prediction models based on AFB 1 Raman feature shifts of unary linear regression, multiple linear regression (MLR) and partial least squares regression (PLSR) prediction models based on full spectral variables were established, respectively. Because the AFB 1 characteristic peaks shifted along with the changes of solution polarity and substrate, the advantage of the quantitative prediction model using PLSR of full spectrum was more significant. To effectively eliminate irrelevant variables in the full spectrum and ensure the fingerprint information of AFB1 molecular structure and functional groups, the quantitative PLSR prediction model of AFB1 was established by screening the feature variables based on three algorithms, namely, competitive adaptive reweighted sampling (CARS), random frog (RF), and successive projections algorithm (SPA), respectively, among which CARS algorithm was the most effective, with the model determination coefficient of correction set of 0.9961, and the root mean square error of correction set of 0.02213 mg/kg. Finally, the model was externally validated, and the coefficient of determination was 0.9792, and the root mean square error (RMSE) was 0.04426 mg/kg. The results showed that the established method could be used for rapid quantitative detection of AFB 1 in peanut oil.