Application of linear solvation energy relationships and principal component analysis methods for the prediction of the retention behaviour of E-resveratrol analogues with substituted silica hydride stationary phases

Highlights: • Retention behaviour of 22 compounds related to E-resveratrol determined. • Selectivity profiles established for different substituted silica hydride sorbents. • Unique retention observed for different stationary and mobile phase conditions. • Molecular descriptors derived from linear solvation energy relationship concepts. • Structure-retention dependencies classified by principal component analysis methods. -- Abstract: In this investigation, application of linear solvent energy relationships (LSERs) and principal component analysis (PCA) methods have been employed to investigate the structure-retention dependencies of E-resveratrol analogues separated under different stationary and mobile phase conditions. To this end, the retention of 22 analogues have been determined with phenyl, diol, bidentate anchored C18 (BDC18) and Diamond Hydride™ C18 (DHC18) substituted silica hydride stationary phases under isocratic chromatographic conditions using mobile phases containing 0.1 (% v/v) formic acid and different acetonitrile or methanol contents from 10 to 90% (v/v) in 10% increments. In general, these compounds showed decreasing retention with increased acetonitrile or methanol content in the mobile phase with all the stationary phases. The retention order generally followed their log P values, although some unique selectivity variations were apparent depending on the nature of the selected stationary and mobile phases. These 22 compounds contained different backbone functionalities linking the phenyl ring A to phenyl ring B and different numbers of hydroxyl groups in the phenyl ring A/phenyl ring B. Structure-retention descriptors, derived according to LSER concepts, were analysed by PCA methods to provide group classification of these resveratrol analogues from the associated PC1 versus PC2 score plots. These results revealed that the selectivity of these compounds was dominated by hydrophobic and steric interactions. Based on the number and position of hydroxyl groups in a specific resveratrol analogue, a reliable curve fitting approach (indicated by R² > 0.99 for the correlation between experimental and predicted log k values) was derived for prediction of the retention of these analytes under different mobile phase isocratic separation conditions. The application of similar methods are anticipated to find general utility for the analysis of diverse classes of other low molecular mass compounds in the different modes of liquid chromatography, permitting enhanced levels of prediction and evaluation of the retention attributes of polar and non-polar compounds.