采用表面增强拉曼光谱技术结合化学计量方法快速分析脐橙果皮中的三唑磷农药残留。利用乙腈快速提取脐橙果皮的提取液，采用石墨化碳、C18和PSA去除果皮提取液中色素等物质的干扰。以脐橙果皮提取液为基质，配制不同浓度的三唑磷农药溶液，进行拉曼信号采集，得到的拉曼特征峰与三唑磷标准溶液的特征峰基本一致。建立脐橙果皮中三唑磷农药残留的偏最小二乘法预测模型，结果表明，以脐橙果皮提取液为基质的三唑磷溶液最低检测浓度为0.8 mg/L；模型对预测集样本的预测均方根误差为1.60 mg/L，相关系数为0.9669。再利用该方法检测脐橙中实际所含三唑磷农药残留，其模型对样本的预测值与真实值的相对误差绝对值在0.08%~5.98%之间，预测回收率为94.36%~104.36%，模型预测能力和重现性良好。

Surface-enhanced Raman spectroscopy (SERS) combined with chemometrics method was used to rapidly detect triazophos residues in navel orange peel. Acetonitrile was used to rapidly produce navel orange peel extract. Graphite carbon, C18 and Periodic acid-Schiff (PSA) were used to remove interferences, e.g., pigments from the navel orange peel extracts. The obtained extracts were used as the base to prepare triazophos solutions of different concentrations and their signals were collected by laser Raman spectrometer. The resultant Raman characteristic peaks were largely consistent with those of triazophos standard solution. A partial least-squares regression model was developed to predict residual triazophos content in navel orange peel. The results showed that the minimum detection concentration for triazophos in solutions of navel orange peel extract was 0.8 mg/L, the root mean square error of the model’s prediction was 1.60 mg/L, and the correlation coefficient was 0.9669. When this method was used to detect the actual triazophos residue in navel orange peel, the absolute values of the relative error were between 0.08% and 5.98%, while the prediction recoveries were 94.36% to 104.36%. The results of this study demonstrated good predicting ability and reproducibility of the model.

赣鄱英才555工程人选计划，国家自然科学基金项目（31271612）；江西省学术带头人培养计划（20113BCB22001）