Study On Pesticide Residues Detection Using Terahertz Time-domain Spectroscopy

In order to control pests and increase the production of agricultural products, a large number of pesticides gain widespread applications recently, which result in a lot of residues left on crops. The quality of farm commodities is so bleak that it is a matter of great urgent to find a reliable way to detect residues of pesticide and thus control the phenomenon of abuse of them. Given the shortcomings of traditional means, it is necessary to study a simple, fast and efficient method, such as spectroscopy detection technology. THz spectroscopy, because of its low energy, strong penetrability through the non-polar material and the ability to reflect the nuances of the molecule, turns into a hot spot in security detection, communications, medical imaging, and nondestructive testing. In THz band, a large number of molecules, especially the organic molecules, have rotational and vibrational transition, showing optical specificity in frequency domain, and a typical terahertz time-domain spectroscopy system only need 1~2 minutes to scan a sample. So THz spectroscopy can be applied to qualitatively and quantitatively analyze the substance.Combing THz-TDS technology, this thesis selected Cyfluthrin as the representation of commonly-use pesticides to explore residues detection method. A specific experimental design was proposed and some key issues were addressed, then this paper analyzed the processing method of THz spectroscopy data and quantitative detection approach. Traditional spectroscopic technology and chemometric method were applied to establish the quantitative model of the spectrum, and the characteristics of each model were summed up. In the end, the difficulties of pesticide residue detection using THz spectroscopy technology and the next research direction were given.The results showed that the spectral absorption coefficient of liquid pesticides in THz band was in a linear relationship with its concentration, which can be used as a theoretical foundation of quantitative detection. For a single component model, step-wise Multi-Linear Regression(SMLR) is slightly better than the Partial Least Squares(PLS) in quantitative test. The innovation of this paper includes: building a system for THz pesticide detection, supplying the extraction method of optical information and modeling algorithm of quantitative and qualitative analysis, providing theoretical and experimental basis for the initial application of THz spectroscopy techniques on agricultural product quality inspection.