Gradient elution in aqueous normal-phase liquid chromatography on hydrosilated silica-based stationary phases

The possibility of applying a theoretical model in the prediction of the retention of phenolic acids on hydrosilated silica, in aqueous normal phase mode was studied. The actual gradient of the aqueous component in acetonitrile may fluctuate from the pre-set program, as even the gradient-grade acetonitrile contains some water. Hence, the actual concentration of water during the gradient run is higher than pre-set by the gradient program, which leads to lower than expected sample retention. Furthermore, the actual gradient profile may be affected by an increase in water uptake on a polar column during the gradient run. These effects were investigated using the using frontal analysis method and Karl-Fischer titration, for the determination of water in the initial mobile phase, and in the column effluent. Preferential adsorption of water on the Silica hydride, Diamond hydride, UDC Cholesterol, Bidentate C18, and Phenyl hydride columns can be described by Langmuir isotherms. At the column saturation capacity, less than one monomolecular water layer is adsorbed, with a further decrease in coverage density for modified materials. Parameters of semi-logarithmic and logarithmic model equations, describing the dependence of retention factor on the concentration of water, were determined under isocratic conditions. These parameters and linear gradient profiles corrected for the actual water concentrations were used in calculation of gradient retention data. The corrections for the actual water concentration greatly improved the agreement between the experiment and the predicted gradient elution volumes. Generally, the semi-logarithmic model provides slightly better prediction of the gradient data, with respect to the logarithmic retention model.