Experimental design for the optimization of the derivatization reaction in determining chlorophenols and chloroanisoles by headspace-solid-phase microextraction–gas chromatography/mass spectrometry

• Tools to optimize a derivatization reaction and interpret 8 responses in conflict.
• Tools to optimize a derivatization reaction of chlorophenols acetylation.
• The joint use of PARAFAC and Doehlert designs with gas chromatography multiway data.
• A multiresidue procedure based on HS-SPME–GC/MS is optimized by a Doehlert design.
• The study of the Pareto-optimal front gives insight into the derivatization reaction.

The paper shows some tools (its interpretation and usefulness) to optimize a derivatization reaction and to more easily interpret and visualize the effect that some experimental factors exert on several analytical responses of interest when these responses are in conflict. The entire proposed procedure has been applied in the optimization of equilibrium/extraction temperature and extraction time in the acetylation reaction of 2,4,6-trichlorophenol; 2,3,4,6-tetrachlorophenol, pentachlorophenol and 2,4,6-tribromophenol as internal standard (IS) in presence of 2,4,6-trichloroanisole, 2,3,5,6-tetrachloroanisole, pentachloroanisole and 2,4,6-trichloroanisole-d5 as IS. The procedure relies on the second order advantage of PARAFAC (parallel factor analysis) that allows the unequivocal identification and quantification, mandatory according international regulations (in this paper the EU document SANCO/12495/2011), of the acetyl-chlorophenols and chloroanisoles that are determined by means of a HS-SPME–GC/MS automated device. The joint use of a PARAFAC decomposition and a Doehlert design provides the data to fit a response surface for each analyte. With the fitted surfaces, the overall desirability function and the Pareto-optimal front are used to describe the relation between the conditions of the derivatization reaction and the quantity extracted of each analyte. The visualization by using a parallel coordinates plot allows a deeper knowledge about the problem at hand as well as the wise selection of the conditions of the experimental factors for achieving specific goals about the responses. In the optimal experimental conditions (45 °C and 25 min) the determination by means of an automated HS-SPME–GC/MS system is carried out. By using the regression line fitted between calculated and true concentrations, it has been checked that the procedure has neither proportional nor constant bias. The decision limits, CCa, for probability a of false positive set to 0.05, vary between 0.221 and 0.420 µg L-1.