针对水果生产中的农药残留问题, 利用表面增强拉曼光谱技术(SERS), 把害虫防治使用较多的有机磷农药亚胺硫磷与毒死蜱作为研究对象, 探索性研究了将金胶用作增强基底检测以脐橙为载体的混合农药残留快速检测。 采集混合农药样品的SERS光谱, 通过对比农药的特征峰可以对混合农药进行定性分析。 同时利用化学计量学方法, 建立混合农药的定量数学模型, 并通过对比不同的预处理方法和建模波段对混合农药样品拉曼光谱的处理结果, 选择出最优预处理方法与算法的组合。 在拉曼光谱范围200～2 300 cm-1内, 利用PLS算法处理经一阶微分预处理后的光谱数据, 建立的脐橙表皮混合农药残留回归模型效果较好, 预测相关系数(Rp)为0.912, 预测均方根误差(RMSEP)为3.601 mg·L-1。 经过波段筛选后并对光谱处理结果对比, 发现光谱在200～620, 830～1 040及1 250～2 300 cm-1范围内, 利用PLS算法处理经一阶微分预处理后的光谱数据, 建立的回归模型效果较好, Rp为0.909, RMSEP为3.338 mg·L-1。 研究表明使用SERS技术, 可以对脐橙表皮上残留的混合农药进行定性与定量的分析。

Pesticide residue in agricultural and sideline products has become a social problem. In order to solve the problem of pesticide residue in fruit production, the rapid detection of mixing pesticide residues with navel orange as carrier was explored by surface enhanced Raman spectroscopy (SERS). The organophosphorus pesticides phosmet and chlorpyrifos were the objects of study because they were used more in the field of pest control, and gold colloid was prepared to get the SERS of the organophosphorus pesticides. The SERS spectra of the mixing pesticide samples were collected, and the characteristic peaks of the pesticides could be used for qualitative analysis of the mixing pesticide. At the same time, the stoichiometric method could be used to establish the quantitative mathematical model of the mixing pesticide, and the combination of the optimal pretreatment method and the wavenumber could be selected by comparing the modeling results of the Raman spectra. In the Raman spectrum range of 200～2 300 cm-1, the partial least square (PLS) was used to build the model of the spectral data after the first derivatives data preprocessing. The results of the combined regression model were better, and it showed that the correlation coefficient (Rp) was 0.912, Root mean square error (RMSEP) was 3.601 mg·L-1. After spectral screening and comparison of the spectral results, it was found that the best model of the spectra was established by the PLS in the range of 200～620, 830～1 040 and 1 250～2 300 cm-1. The regression model had better effect, in which Rp was 0.909 and RMSEP was 3.338 mg·L-1. It had shown that the SERS technology could be used to qualitatively and quantitatively analyze the mixing pesticide residues on navel orange.