Modeling of chromatographic lipophilicity of food synthetic dyes estimated on different columns

• Published in: Journal of separation science 2010-08, Vol.33 (15), p.2219-2229
• Main Authors: Sârbu, Costel, Casoni, Dorina, Kot-Wasik, Agata, Wasik, Andrzej, Namieśnik, Jacek
• Format: Article
• Language: English
• Subjects: Algorithms; Analytical chemistry; Biological and medical sciences; Chemistry; Chromatographic methods and physical methods associated with chromatography; Chromatography, High Pressure Liquid; Chromatography, Reverse-Phase; Computed log P; Exact sciences and technology; Food additives; Food Analysis - instrumentation; Food Analysis - methods; Food Coloring Agents - chemical synthesis; Food Coloring Agents - chemistry; Food Coloring Agents - isolation & purification; Food industries; Food synthetic dyes; Fundamental and applied biological sciences. Psychology; General aspects; Hydrophobic and Hydrophilic Interactions; Linear Models; Lipophilicity indices; Models, Chemical; Other chromatographic methods; QSPR; RP-HPLC; Software
• ISSN: 1615-9306; 1615-9314
• DOI: 10.1002/jssc.201000099

The retention behavior of some food synthetic dyes has been studied by RP-HPLC on chemically bonded C18, C8, C16 and CN stationary phases. Using methanol-ammonium acetate (0.08 mol/L, pH=6.76) as mobile phase, a linear behavior of retention parameters throughout the methanol fraction variance was obtained in all cases (r>0.99). The patterns of chromatographic behavior of the compounds illustrate high similarities between the C18, C8 and C16 columns, respectively. Highly significant correlations were obtained between experimental lipophilicity indices log kw and φ₀ estimated on C18 and C8 stationary phases and some computed log P-values. An extensive investigation made for quantitative structure-property (lipophilicity) relationships of studied dyes, using descriptors from Dragon software, multiple linear regression and genetic algorithm, revealed that the molecular descriptors appearing in the best models combine 2-D and 3-D aspects of the molecular structure. The most significant descriptors can be classified as radial distribution function, GETAWAY (autocorrelation), 3D-MoRSE signal, Burden eigenvalues and edge adjacency descriptors.