The aim of this work was to study the physicochemical interactions between chiral stationary phases and chiral molecules and to elucidate which of the specific interactions are more or less important. The HPLC separation of 58 aryloxyaminopropanols was performed on two chiral stationary phases containing the macrocyclic antibiotics teicoplanin or vancomycin and using a methanol/acetonitrile/acetic acid/triethylamine mobile phase (volume ratios 45/55/0.3/0.2). The resolution of enantiomers (Rij) as the target variable was predicted for the mentioned kind of compounds by means of thoroughly selected descriptors provided by the applied Dragon software. The created QSPR models can be considered as a way to explore and discover new relationships or interactions between the quantitative structure and resolution of enantiomers. For calculation and validation of the QSPR models, different modelling methodologies were applied based on MLR (multiple linear regression) and ANN (artificial neural network) techniques. Both methods exhibit an ability for successful prediction of the enantioresolution characteristics of the studied molecules. The results seem to demonstrate that it is possible to predict resolution values of enantiomeric separations of related compounds on given chromatographic systems.
Keywords: Artificial neural network; Aryloxyaminopropanol derivatives; Quantitative structure–property relationship; Resolution of enantiomers; Teicoplanin; Vancomycin.
Copyright © 2013 Elsevier B.V. All rights reserved.