Global Optimization of HPLC Separations

Abstract

Abstract A method suitable for automated optimization of HPLC separations is developed. A sequential strategy based on a non-linear interpolation between measured retention data permits a stepwise approximation to a global optimum. This scheme provides a smooth representation of the elution behaviour from an unspecified number of regularly or irregularly spaced data points. The accuracy of the so defined curves can be gradually improved by successive addition of new experimental data as they become available. It is possible, therefore, to employ variables yielding multimodal response hypersurfaces in the optimization work. Further emphasis is given to an individual weighting of all detectable compounds with respect to their analhtical relevance. Examples including an isocratic separation of fat soluble vitamins as well as gradient separations of dansylated amino acids and steroid hormones demonstrate some applications of the proposed procedure.