APPLICATION OF CHEMOMETRIC ALGORITHMS FOR SPECTROPHOTOMETRIC DETERMINATION OF SYNTHETIC FOOD COLORS E110 AND E124

A method is proposed for the spectrophotometric determination of synthetic food dyes Yellow "Sunset" and Ponceau 4R in model mixtures using chemometric methods (projection on latent structures, PLS, regression on principal components, PCR). 16 binary mixtures of dyes with different concentrations (in the range from 2 to 20 mg/l) were investigated. The absorption spectra and the first derivative of the absorption spectra (hereinafter, the derivatives of the absorption spectra of the zero and first orders, 0D and 1D, respectively) for 11 mixtures were used as a training dataset, and for 5 mixtures as test ones. The optimal number of principal components (latent variables) was selected for each chemometric method and the order of the derivative of the absorption spectra, which ranged from 2 to 6, depending on the type of model. The smallest root mean square errors of calibration (RMSEC) were: for E110 - 0.0518 (PLC, 1D); for E124 -0.0077 (PLC, 0D). The smallest root mean square errors of prediction (RMSEP) were 0.0332 (RGC, 1D) and 0.0153 (PLC, 0D) for E110 and E124, respectively. It was found that the equations of the "measured-predicted" dependencies have a slope close to 1 and a free term close to zero. The correlation coefficient is close to 1 which speaks of the high quality of the models. Values of recovery of E110 and E124 in test mixtures ranged from 95.6 to 101.0%. The greatest error is observed when determining E110 at a concentration of 2 mg/ml. Thus, it has been shown that the use of chemometric methods provides high accuracy of spectrophotometric determination of components in mixtures with overlapping absorption bands.