Identification and quantification of adulterants in Arnebia euchroma (Royle) Johnst. using portable near infrared spectroscopy

In the present study, a rapid, non-destructive and real-time detection of the adulteration in Arnebia euchroma (Royle) Johnst. ( A . euchroma ) samples were established using a portable near infrared spectrometer in combination with chemometrics. 37 batches of A . euchroma , 3 batches of Onosma paniculatum Bur. et Franch ( O . paniculatum ), 8 batches of Arnebia benthamii (Wall. ex G. Don) Johnston ( A . benthamii ) samples were collected. In addition, the adulterated A . euchroma samples were prepared by mixing A . euchroma with O . paniculatum or A . benthamii in different ratios, respectively. Afterwards, the near infrared spectra of the samples were acquired by a portable near infrared spectrometer. Firstly, a classification model was established by using data driven-soft independent modeling by class analogy (DD-SIMCA) and partial least squares-discriminant analysis (PLS-DA) algorithms, respectively, for differentiating the A . euchroma from its adulterants. Secondly, a quantitative model for detecting the adulteration concentration in adulterated A. euchroma samples was constructed by applying PLS and support vector machine (SVM) algorithms, respectively. The classification models of DD-SIMCA and PLS-DA achieved 100% of accuracy in calibration sets, as well as an accuracy of 95.6% and 100%, respectively, in the prediction sets. Moreover, the PLS regression model registered a ratio of prediction to deviation (RPD) value and root mean square error of prediction (RMSEP) value of 2.2, 1.6 and 8.70%, 12.03%, respectively, for O. paniculatum and A. benthamii adulterants, while the SVM model yielded the RPD value and the RMSEP value of 8.2, 3.4 and 2.29%, 5.55%, respectively. In summary, portable near infrared spectrometer combined with chemometrics was verified to be capable of achieving rapid, real-time detecting of adulteration of A. euchroma .