Multicomponent determinations by partial least-squares regression analysis of fast-reaction multiwavelength profiles...

Partial least-squares regression (PLS) analysis of multiwavelength reaction profiles, obtained by continuous addition of a reagent to a binary mixture of analytes, giving fast reactions, is presented. The methodology has proved to be useful for the simultaneous determination of the mixture components. Octacyanomolibdate(V) (Mo(V)) has been used as a reagent for mixtures of hydroquinone (HQ) and pyrogallol (PG) in acidic medium. When the reaction profiles are followed spectrophotometrically at several wavelengths, and they are handled with PLS, the mixture can be resolved. The final spectrum of the mixture, after reaction with Mo(V), is enough to predict the HQ concentration, but in the case of PG successive spectra along the reaction are necessary to predict properly the concentration. The concentrations for HQ in the calibration set ranged between 1.8x10(-6) and 1.7x10(-5) M and for PG between 2.5x10(-6) and 2.5x10(-5) M. The concentration ratio, [HQ]:[PG], in the validation set ranged between 0.17 and 4.10. Mean validation errors of 1.6% for HQ and 4.2% for PG were found. Satisfactory results were obtained when independent studies of accuracy and precision were accomplished. The effect of p-phenylenediamine and aminophenols was studied as interferences and, in general, hydroquinone is more affected than pyrogallol. The tolerated concentration of interferences ranged between 10(-6) and 10(-4) M.