Robustness testing for a capillary electrophoresis method using the "short-end injection" technique.

A multivariate approach for testing the robustness of a capillary electrophoresis method using the "short-end injection" technique is presented. Firstly, a Plackett and Burman (PB) design with 11 factors (eight real factors and three dummies) was used to identify the critical factors on resolution, plate number, plate count, asymmetry and assay. Then, the factors which were found to be significant were studied in a central composite design to predict the variation of resolution inside the area investigated in the PB design. PB and central composite designs yielded conclusions that were in good agreement with one another. They showed that the separation could be considered as robust, notwithstanding the fact that some factors where found to be statistically significant and should be controlled (injected volume and electrolyte concentration). Using the factor values which gave the worst-case situation for Rs still led to acceptable values for this parameter.