engleski

QSRR Modeling in Ion Chromatography

Finding a financially acceptable analytical method is an important goal of most modern analytical laboratories. Therefore, method optimization is not only necessary for successful analysis but also unavoidable from financial aspect. Analytical method development and optimization is usually carried out using "trial and error" approach which can be time and resource consuming. Using any other approach is, accordingly, a better solution. In this work, Quantitative Structure-Retention Relationships (QSRR) methodology was applied in order to find a new, cheap and fast strategy for method development and optimization. The QSRR represents methodology which predicts retention time in chromatography on the basis of analyte's molecular structure. QSRR models were developed for prediction of retention for a set of carbohydrates in ion chromatography. The models were developed for several isocratic elution conditions, using three different regression techniques: stepwise multiple linear regression (S-MLR), partial least square (PLS) and uninformative variable elimination - partial least squares regression (UVPLS). The best prediction of retention parameter log k was obtained by using PLS. The obtained predicted values were used for development of general isocratic retention model, i.e. model that is not limited to a specific isocratic run, but can predict retention for any isocratic elution. In order to enable retention prediction for gradient elutions, a gradient model based on isocratic data was applied. By obtaining this model, it was possible to expand the prediction to practically any isocratic or gradient elution condition. Moreover, the use of QSRR methodology allowed application of the model for components which have not been used for modeling or, evermore, have never been analysed before. The predicted retentions at gradient conditions showed good agreement with experimental values (RMSEP=18.34 %) proving great potential of the applied methodology in ion chromatographic method development and optimization.

QSRR; modeling; ion chromatography

nije evidentirano