Zakaria Philip, Macka Miroslav, Haddad Paul R
Department of Chemistry, University of Tasmania, GPO Box 252-75, Hobart, Tasmania 7001, Australia.
J Chromatogr A. 2003 May 16;997(1-2):207-18. doi: 10.1016/s0021-9673(03)00064-5.
The electrokinetic chromatographic (EKC) separation of a series of aromatic bases was achieved utilising an electrolyte system comprising an anionic soluble polymer (polyvinylsulfonic acid, PVS) and a neutral beta-cyclodextrin (beta-CD) as pseudo-stationary phases. The separation mechanism was based on a combination of electrophoresis, ion-exchange interactions with PVS, and hydrophobic interactions with beta-CD. The extent of each chromatographic interaction was independently variable, allowing for control of the separation selectivity of the system. The ion-exchange and the hydrophobic interactions could be varied by changing the PVS and the beta-CD concentrations, respectively. Additionally, mobilities of the bases could be controlled by varying pH, due to their large range of pKa values. The separation system was very robust with reproducibility of migration times being <2% RSD. The two-dimensional parameter space defined by the two variables, [beta-CD] and %PVS, was modelled using a physical model derived from first principles. This model gave very good correlation between predicted and observed mobilities (r2=0.999) for the 13 aromatic bases and parameters derived from the model agreed with the expected ion-exchange and hydrophobic character of each analyte. The complexity of the mathematical model was increased to include pH and this three-dimensional system was modelled successfully using an artificial neural network (ANN). Optimisation of both the two-dimensional and three-dimensional systems was achieved using the normalised resolution product and minimum resolution criteria. An example of using the ANN to predict conditions needed to obtain a separation with a desired migration order between two of the analytes is also shown.
