An Adaptation of Kubista's Method for Spectral Curve Deconvolution

A chemometric approach to spectral curve deconvolution is described, evaluated, and applied to micellar systems. The technique is based on the method of principal component analysis of a spectral matrix followed by transformation of the abstract vectors into real spectra and concentrations. The approach reported here is similar to that of Kubista et al. (Anal. Chem. 1993, 65, 994-998). In the present study, however, more spectral information is known about the system of interest This information is included in the deconvolution, which should, in general, increase the reliability of the method. From this method we obtain very reliable (noise-insensitive) λ_max values of indicator molecules in the micellar pseudophase free from contributions of the indicator in the aqueous phase. The water-to-micelle partition coefficients are also determined. The effects of noise and the extent of indicator partitioning on the reliability of the method are evaluated using model data. The application of the method to the study of eight indicators in a prototypical micellar system (sodium dodecyl sulfate) is presented. Extension of the method to other types of chemical studies such as the determination of kinetic rate constants and product spectra is briefly discussed.