Structure-retention time modeling of some anthraquinone-based dyes by considering chain lengths of organic modifiers in the eluent of micellar HPLC

Micellar liquid chromatography (MLC) is a green liquid chromatographic technique which is dramatically growing in last decade [1]. In this technique, an aqueous solution of a surfactant along with an alcoholic modifier is usually employed as mobile phase composition. Quantitative predictive/descriptive models are usually useful tools to predict and elucidate chromatographic behavior in MLC systems. In this work, a quantitative structure–activity relationship (QSPR) model was proposed for retention time of anthraquinones in reverse phase MLC system. Retention time of 96 chromatographic samples (16 antraquinones evaluated by using 6 different organic modifiers) were experimentally determined and used as the independent variables of the QSPR model. Five small-chain alcohols (methanol, ethanol, propanol, butanol and pentanol) as well as acetonitrile were used as the eluent modifiers. The matrix of the dependent variables of the model was build using structural descriptors of antraquinones and empirical parameters of organic modifiers in the applied MLC [2]. It should be noted that not-retained chromatographic mixtures were excluded from QSPR modeling. After deleting the not-retained samples (tR 2.5 min), the matrix containing both the structural properties of antraquinones and the empirical scales of the modifies were entered to the variable selection and multiple linear regression model construction steps. A five-parameter model was proposed for the logarithm of the retention time values which covered about 96% and 95% variance of data in training and cross validation, respectively. The correlation coefficient of the external test set was 0.97 which showed the prediction ability of the proposed model as well as its good applicability domain that was checked using standardized residual-leverage plot. The stability and significance of the proposed model were tested by applying model on different random-selected training and test sets. It was concluded that both structural properties of analytes (antraquinones) and properties of organic modifiers are contributed in the proposed quantitative structure-retention time relationship.