近红外光谱快速检测痰咳净散中咖啡因的含量

目的: 研究光纤近红外漫反射光谱快速检测痰咳净散中咖啡因的方法。 方法: 按照处方自制了34批痰咳净散实验室样品,收集了31批痰咳净散工业化生产样品。以高效液相色谱法的分析值作为参考,运用偏最小二乘法建立近红外光谱与咖啡因参考含量的校正模型;进一步用马氏距离法建立定性模型鉴别不同阴性样品和不同厂家的样品。 结果: 定量模型校正集的R²为0.975,RMSECV为0.239%(w/w);预测集样品的R²为0.926,RMSEP为0.128%(w/w),平均(最大)相对误差为1.0% (3.2%);t-检验结果表明,近红外光谱预测的咖啡因含量与参考含量无显著性差异;还能正确鉴别痰咳净散和各种阴性样品,并能提示不同生产工艺的样品。 结论: 近红外光谱方法可用于痰咳净散中咖啡因的快速检测,同时为其生产过程的在线分析打下了坚实的基础。

Objective: Fibre optic near infrared (NIR) diffuse reflectance spectroscopy method was developed to detect rapidly caffeine content in Tankejing powder. Methods: 34 laboratory samples were prepared according to the formulation of Tankejing powder. And 31 batches of industrial samples were collected.The contents of caffeine in Tankejing powder were determined using HPLC.Fibre optic NIR diffuse reflectance spectroscopy method was developed to determine qualitatively and quantitatively Tankejing powder. Results: For quantitative analysis,a PLS algorithm was used to construct the multivariate calibration model.The R²and RMSECV of the calibration set were 0.975 and 0.239%(w/w),respectively.The R² and RMSEP of the validation set were 0.926 and 0.128%(w/w),respectively.And the mean (maximal) relative error was 1.0% (3.2%).Furthermore,a paired t-test was also performed to confirm that the NIR values and the reference values were not significantly different.The results indicated acceptable prediction accuracies for industrial on-line monitoring.For qualitative analysis,Mahalanobis distance algorithms were utilized to construct the qualitative model which could distinguish the samples from the negative samples and those manufactured by another factory. Conclusion: The present NIR spectroscopy is rapid,simple and reliable,which can be applied to rapid detection and on-line monirtoring of Tankejing powder.