Application of statistical design for the optimization of amino acid separation by reverse-phase HPLC

Modified resolution and overall separation factors used to quantify the separation of complex chromatography systems are described. These factors were proven to be applicable to the optimization of amino acid resolution in reverse-phase (RP) HPLC chromatograms. To optimize precolumn derivatization with phenylisothiocyanate, a 2^5-1 fractional factorial design in triplicate was employed. The five independent variables for optimizing the overall separation factor were triethylamine content of the aqueous buffer, pH of the aqueous buffer, separation temperature, methanol/acetonitrile concentration ratio in the organic eluant, and mobile phase flow rate. Of these, triethylamine concentration and methanol/acetonitrile concentration ratio were the most important. The methodology captured the interaction between variables. Temperature appeared in the interaction terms; consequently, it was included in the hierarchic model. The preliminary model based on the factorial experiments was not able to explain the response curvature in the design space; therefore, a central composite design was used to provide a quadratic model. Constrained nonlinear programming was used for optimization purposes. The quadratic model predicted the optimal levels of the variables. In this study, the best levels of the five independent variables that provide the maximum modified resolution for each pair of consecutive amino acids appearing in the chromatograph were determined. These results are of utmost importance for accurate analysis of a subset of amino acids.