Classification and quantification of multiple adulterants simultaneously in black tea using spectral data coupled with chemometric analysis

Black tea is a popularly consumed beverage across the world. However, the adulteration of tea powder with extraneous colors is causing serious health threats. In the present study four such color adulterants viz., tartrazine, sunset yellow, carmoisine and ponceau 4 R were simultaneously detected using FT-IR spectral data coupled with chemometric tools. PLS-DA (partial least squares discriminant analysis) was used for the classification of adulterants, whereas, PLS2 (multi response PLS) and LS-SVM (least-squares support vector machines) were used for quantification purpose. RCGA (real coded genetic algorithm) was used as a feature selection algorithm to obtain fewer fingerprints from the FT-IR spectrum of the adulterated tea powders. PLS-DA was able to predict with 100% accuracy for the samples spiked with all four adulterants when only 30 fingerprints for each adulterant were used. Amongst the regression algorithms, LS-SVM was observed to be superior over PLS2 having lower root-mean-square error of prediction (RMSEP) for carmoisine and ponceau 4 R when randomized statistical test was conducted. The prediction results had high regression coefficient (R2p >0.89) with REP values ranging between 10.90–18.40%, and RPD values in the range of 2.82–5.81, for simultaneous quantification four adulterants using 30 variables LS-SVM model. To study matrix effect, an additional experiment was carried to observe the effectiveness of above established methodologies. Towards this, six different indigenous commercial tea brands were selected and the robustness of these chemometric methods were demonstrated. This study provides a robust tool for infrared based simultaneous detection and quantification of four different color adulterants in black tea powders.