Spectrophotometric determination of food dyes in soft drinks by second order multivariate calibration of the absorbance spectra-pH data matrices

Three common food colorants (tartrazine TT, allura red AR, and sunset yellow SY) were simultaneously quantified in powdered soft drinks by employing a combination of absorbance spectra-pH data matrices and multivariate processing of the generated second-order data. Parallel factor analysis PARAFAC and bilinear least squares/residual bilinearization BLLS/RBL were applied for deconvolution of trilinear data to get spectral and concentration profiles of the dyes as a function of pH. BLLS/RBL outperformed PARAFAC for handling rank deficiency and linear dependency problems that often involved in pH-equilibrating systems. BLLS/RBL was able to mange the earlier problems that came from the high spectral overlap between acid and basic forms of dyes and linear dependency in pH dimension. Using BLLS/RBL, the second order advantage was fully accomplished in which TT, SY, and AR were quantified in four soft drinks even in the presence of many uncalibrated interferents. The concentration levels of dyes in commercial samples were within the safe recommended ones, IT 323-480, SY 84-1016 mg kg(-1), and AR was not detected in any product. The comparison of chemometric results with those obtained by standard chromatographic technique proved that the former protocol has a reasonable accuracy with average recoveries of 104% for TT and 104.9% for SY. (C) 2013 Elsevier Ltd. All rights reserved.