The use of wavelength dispersive x-ray fluorescence (WDXRF) spectrometry and multivariate techniques for assessment of illegal dyes in spices

Abstract

Sudan dyes are carcinogenic azo and diazo compounds that are banned for use in food worldwide including the European Community. The ability of WDXRF spectroscopic technique to predict the types and levels of adulteration of spices suspected to be adulterated with Sudan I-IV, Para Red and Sunset Yellow dyes was investigated. Logistic regression and discriminant analysis classification models were developed to predict the type of adulteration using WDXRF spectral features. Prediction of the levels of adulteration was assessed by using multiple regression analysis. 83% of the 210 adulterated samples were correctly classified by the logistic regression with 90% sensitivity, 75% specificity with a prediction power of 92% into respective adulteration groups. 86% and 90% correct prediction were obtained for discriminant analysis models with 94% sensitivity and 74% specificity. Three multiple regression models were performed for each data set. The full model based on both the Compton and Rayleigh ratios as well as the Compton and Rayleigh scatter intensities revealed to be the best model to predict the levels of adulteration with an adjusted R² between 95.1 to 99.0% with errors of estimate between 1.6 to 3.7%. WDXRF technique may represent a promising tool for the screening of Sudan dyes-adulterated spices and can be used as an alternative to classical methods for the determination of Sudan dyes present in food prior to confirmation with more sensitive chromatographic-based methods.