Critical comparison of retention models for use in computer aided interpretive optimisation of separation in anion chromatography

This work deals with the optimisation of the eluent composition for the separation of anions using Ion Chromatography, specifically using an Iterative Interpretive Optimisation Strategy. Firstly a series of retention models is introduced that are suitable for use in an Interpretive Optimisation strategy. These models include the Linear Solvent Strength model (Dominant Equilibrium and Competing Ion Effective Charge Approaches), Hoover model, Dual Eluent Species model, Kuwamoto model, Extended Dual Eluent Species model, Multiple Species Eluent/Analyte model and the newly developed Empirical End Points model. Also included is the commercially available 3-point model of DryLab. The theory and mathematics behind these models are also discussed. These models have been critically compared to determine their suitability for use in an Iterative Interpretive Optimisation strategy. This comparison is broken into three parts, non-suppressed ion chromatography, suppressed ion chromatography, and single eluent species ion chromatography. For single eluent species ion chromatography the list of models was reduced to the Linear Solvent Strength model, the Empirical End Points model and the 3-point model of DryLab, which were found to be the only models suitable for this type of chromatography. The 3-point model of DryLab was found to be unsuitable for multiple species eluent ion chromatography. A number of trends were noted across the comparison, the most significant of which were that the Empirical End Points and Linear Solvent Strength models were the only two models that were applicable to the entire data set. The Empirical End Points model was consistently the best performed model while the Linear Solvent Strength model was consistently the worst, and the solution requirements for the Empirical End Points model were found to be the most suitable for an optimisation process. Thus the Empirical End Points model was chosen as the retention model for the optimisation process. Next, a procedure for an Iterative Interpretive Optimisation process was developed and a computer program written to handle the large number of calculations required for the process. The computer program was for the Iterative Interpretive Optimisation process of the eluent concentration only. A case study using a carbonate eluent was conducted to test the program and procedure, resulting in the separation of 20 anions within 12 minutes. Finally, the use of an Artificial Neural Network as a retention model was studied. Artificial Neural Networks were found to require a large amount of training compared with theoretical retention models, but their high level of accuracy and low system knowledge requirements make them a prospect worthy of further evaluation.