Chromatographic separation of free lutein and fatty acids in de-esterified marigold oleoresin and its mathematical modelling

Abstract

The petal of marigold flowers has been reported to be the richest source of a xanthophyll, namely lutein which exhibits strong antioxidant and anticancer properties. Although large amount of lutein can be obtained, the compound in the flowers exists as esterified lutein, which is not readily bio-available. As a result, a number of separation and purification steps are required after solvent extraction of marigold flowers. One of the key steps is de-esterification of lutein fatty acid esters in the extract, or namely marigold oleoresin, by the reaction of the oleoresin with alkali solution, e.g., KOH, to obtain free lutein. The impurities remained in the reaction product, particularly fatty acids, were separated in the further purification process, e.g., chromatography. In this study, we aimed to study chromatographic separation of free lutein and fatty acids in a normal-phase semi-preparative scale with silica gel as a stationary phase and hexane and ethyl acetate mixture as a mobile phase. Based on HPLC, MS, NMR and FT-IR analyses of organic phase of the de-esterification reaction product, fatty acids were found to be the main impurity. The amount of fatty acids was determined to be 49.12 g/g of oleoresin and it was mostly palmitic acid. At the most suitable mobile phase velocity (0.16 cm/s), determined based on HETP evaluation, free lutein of 93.3% purity could be achieved with an isocratic mode of operation, and as high as 99.2% purified lutein was resulted when a single step gradient mode was employed. Further investigation then involved the development of a suitable mathematical model to describe the mass transfer of the two compounds to be separated: free lutein and fatty acids. Required model parameters: adsorption isotherms, overall mass transfer and axial dispersion coefficients, were determined. The adsorption isotherms were determined from a batch isotherm experiment and were to be linear for the lower range of equilibrium concentrations (<100 µg/ml), while other model parameters were determined from empirical correlation from literature or from correlations developed experimentally in this study. The isotherms and model parameters were then applied to the mass transfer model. The transport model was found to best predict the experimental data of both compounds in both semi-preparative and preparative columns. The correlation factor, higher than 0.70, indicated that the model prediction and experimental results were highly consistent.