近红外光谱结合膜富集技术测定大米中毒死蜱农药残留

Determination of Chlorpyrifos Pesticide Residues in Rice by NIR Spectroscopy Coupled with a Membrane Enrichment Technique

本文以纤维滤膜富集大米中的微量农药残留,提高近红外光谱技术的检测限。向阴性大米样本中喷洒不同浓度毒死蜱标准溶液,制备含农药残留大米样品,以乙腈为溶剂提取大米中的毒死蜱农药,用氮吹仪将提取液浓缩后,使用滤纸富集提取液中的农药,真空冷冻干燥,采集滤纸的近红外漫反射光谱。运用特征波长筛选方法优选特征变量,建立大米中毒死蜱农药残留的近红外光谱分析模型。结果表明,利用联合区间偏最小二乘法方法从全光谱区优选出子区间[3 4 5 10],进一步用遗传算法从子区间中优选80个变量时,所建模型性能最好。在0.46~11.20 mg/kg浓度范围内,模型对预测集样本的相关系数为0.9798,预测均方根误差为0.604 mg/kg,将该模型预测4个未知农药含量的大米样本,其预测值与实际测量值具有较好的一致性。研究表明该方法能较好地快速检测大米中微量农药残留。

Fiber filters were used in this study to accumulate trace pesticide residues in rice in order to improve the detection limit for near-infrared（NIR） spectroscopy. Rice samples containing pesticide residues were prepared by spraying different concentrations of chlorpyrifos standard solutions onto non-contaminated rice. Acetonitrile was used to extract the chlorpyrifos from the rice, the extracted liquid was concentrated using a nitrogen-blowing instrument, and filter papers were used to collect the pesticide from the extracted liquid. After vacuum freeze-drying, the diffuse reflectance NIR spectra of filters were recorded using an NIR spectrometer. The results indicated that a model with a good performance could be established when the subinterval [3 4 5 10] was selected from the full spectral region using synergy interval partial least square（si PLS） algorithm, and 80 optimal variables were selected from the subinterval using a genetic algorithm（GA）. Within the concentration range of 0.46~11.20 mg/kg, the correlation coefficient of the model for the samples in the prediction set（Rp） was 0.9798 with a root mean square error of prediction（RMSEP） of 0.604 mg/kg. The contents of chlorpyrifos pesticide in four rice samples were predicted by this model, and the prediction values were consistent with the measured values. The results show that this method can detect trace pesticide residues in rice effectively and quickly.