Determination of Dicofol in Tea Using Surface-Enhanced Raman Spectroscopy Coupled Chemometrics

Abstract

Dicofol is a highly toxic residual pesticide in tea, which seriously endangers human health. A method for detecting dicofol in tea by combining stoichiometry with surface-enhanced Raman spectroscopy (SERS) technology was proposed in this study. AuNPs were prepared, and silver shells were grown on the surface of AuNPs to obtain core-shell Au@AgNPs. Then, the core-shell Au@AgNPs were attached to the surface of a PDMS membrane by physical deposition to obtain a Au@AgNPs/PDMS substrate. The limit of detection (LOD) of this substrate for 4-ATP is as low as 0.28 × 10^-11 mol/L, and the LOD of dicofol in tea is 0.32 ng/kg, showing high sensitivity. By comparing the modeling effects of preprocessing and variable selection algorithms, it is concluded that the modeling effect of Savitzky-Golay combined with competitive adaptive reweighted sampling-partial least squares regression is the best (Rp = 0.9964, RPD = 10.6145). SERS technology combined with stoichiometry is expected to rapidly detect dicofol in tea without labels.

Keywords: Au@AgNPs/PDMS; SERS; chemometrics; dicofol; tea.

Figure 1

Schematic diagram of Au@AgNPs/PDMS as a surface-enhanced Raman scattering substrate for the determination of dicofol in tea.

Figure 2

(a) UV spectra and (b) Raman spectra of AuNPs of different sizes; (c) particle size distribution of 34 nm AuNPs; (d) UV spectra and (e) Raman spectra of core–shell Au@AgNPs with Ag shell thicknesses of different sizes; (f) particle size distribution of core–shell Au@AgNPs with Ag shell thickness of 9 nm; (g) TEM results of the Au@AgNPs/PDMS substrate; (h) SERS intensity of 4-ATP at 1078 cm⁻¹ on 3 mm × 2 mm Au@AgNPs/PDMS; (i) SERS intensity of 4-ATP (at 1078 cm⁻¹) on Au@AgNPs/PDMS substrates stored for 0, 10, 20, and 30 days, respectively.

Figure 3

(a) Raman spectra of gradient concentration of 4-ATP; (b) standard curve of Raman intensity and logarithmic concentration of 4-ATP at 1078 cm⁻¹; (c) Raman spectra of eight gradient concentrations of dicofol; (d) standard curve of Raman intensity and concentration of dicofol at 489 cm⁻¹; (e) raw Raman spectra of dicofol under eight concentration gradients; (f) PLS modeling results of the original Raman spectrum.

Figure 4

Four variable-filtering graphs and PLS modeling results of dicofol spectral data. (a,b) Variable-filtering and modeling results of CARS-PLS model; (c,d) variable-filtering and modeling results of Si-PLS model; (e,f) variable-filtering and modeling results of SPA-PLS model; (g,h) variable-filtering and modeling results of UVE-PLS model.