FULLY-AUTOMATIC HIGH-PERFORMANCE LIQUID-CHROMATOGRAPHIC OPTIMIZATION

A good optimization routine should correctly find the best chromatographic separation conditions for mixtures of known or unknown constituents. The chromatographer must define the number of the parameters to be optimized and their ranges. However, the more parameters to be optimized and the more they interact, the more difficult and time-consuming the optimization procedure will be. A system capable of performing fully automated optimization of mobile phase selectivity in reversed-phase liquid chromatography was built. The optimization routine searches for the best conditions (trying to maximize a chromatographic response function) and for the points, inside defined experimental borders, where the least available experimental information is available. By conducting the experiment under the predicted optimum conditions and an additional experiment under conditions corresponding to the least density of information, the system was forced not to search for a local maximum, but to approach the global optimum. Peak tracking, an important part of any optimization process in high-performance liquid chromatography, was an integral part of the optimization software and was based on fuzzy theory. This implementation of an on-line identification of the sample components made a fully automated optimization of the mobile phase composition possible. Once a suitable separation had been achieved, it was necessary to validate the procedure, special attention being focused on robustness. The robustness test appraises the outcome of small variations in method conditions on the analytical performance. An important feature of this robustness analysis was the three-dimensional representation of the data as the hypersurface which helps to relate robustness to elution characteristics.