DESIGN OF EXPERIMENTS AND RESPONSE SURFACE METHODOLOGY AS POWERFUL TOOLS FOR THE OPTIMIZATION OF SAMPLE PREPARATION AND INSTRUMENTAL ANALYSIS

The Design of Experiments (DoE) is a multivariate approach aimed at a rational plan of the experiments, which allows to compute mathematical models able to describe our system. It permits to understand how the variables influence a response and interact among each other, avoiding the costly and long “one variable at a time” strategy [1]. Depending on the purpose, it is possible to choose among a selection of experimental designs, such as Plackett-Burman, Box-Behnken and central composite designs. In our lab, DoE has been successfully implemented to both sample treatment and instrumental analysis, in different applications, herein presented. If several variables are involved in complex problems, a first screening followed by a response surface design may be necessary. This was the case of the optimization of the analysis of five phytoestrogens in soy-food. A Plackett-Burman design allowed to select the significant variables affecting the accuracy; then, Box-Behnken design was employed to build the response surfaces and find the optimal experimental conditions. In the analysis of 21 bioactive compounds by non-aqueous reverse phase chromatography, a simple central composite design allowed to see how flow, temperature and gradient velocity influenced chromatographic separation. Just seventeen runs (less than one day of analyses) were necessary to gather the data and then define the best compromise to separate coeluted peaks. The optimization of a dispersive solid phase extraction for polycyclic aromatic hydrocarbons (PAHs) in seawater was rapidly achieved thanks to the Plackett-Burman design. This screening design (just 12 experiments) allowed to reveal that only one variable out of 9 significantly affected the recovery, thus avoiding the performance of a large number of experiments. In a similar application to botanical supplements, the D-optimal design was employed for optimization. The knowledge acquired in the previous study permitted to select only 4 variables: their effect on the PAHs recovery was studied by performing 18 experiments. The response surfaces allowed to rapidly identify the significant variables and optimal conditions. In all these case studies DoE demonstrated to be a powerful tool to obtain the maximum information with the minimum experimental effort.