LAUSR

Abstract Terahertz time-domain spectroscopy (THz-TDS) has been used to analyze mixtures of imidacloprid, carbendazim, and flour substrate. A multiple spectra baseline correction (MSBC) method was adopted to eliminate the oblique baseline and improve the signal-to-noise ratio of the THz spectra. The corrected spectra were fitted with a Voigt function which does not require the spectra of all pure materials of mixtures to be measured. A partial least squares (PLS) method was applied to predict the concentration of the pesticides. The predictive results were compared to those using the PLS method alone for different calibration sets. The results show that the developed method provides smaller root mean square error (RMSE) and larger correlation coefficient (R) for small calibration sets. When the number of calibration sets increases gradually, the obtained accuracy became identical. It is demonstrated that THz-TDS combined with Voigt function fitting and PLS can be used for quantitative analysis of various pesticides in agricultural products in case of small calibration sets.