Deconvolution and quantitation of unresolved mixtures in liquid chromatographic-diode array detection using evolving factor analysis

Summary A recently developed self-modeling curve resolution method based in different factor analysis techniques has been applied for the first time to the study of liquid-chromatography-diode array data under situation where the separation of two components is not achieved. Two applications are reported: the resolution and quantitation of a coeluted mixture of carbamate pesticides pirimicarb and 1-naphthol, and the estimation of the concentration profiles of the double peak obtained in the elution of the triazine metabolite chlorodiamino-s-triazine. Different methods of quantitation are compared, including Evolving Factor Analysis and Rank annihilation. Quantitation from the area of the elution profiles once the component spectra have been transformed for their area contribution to the signal, gives a relative composition for pirimicarb and naphthol pesticides which agrees with the known sample composition. In the case of the unknown triazine mixture, an approximate quantitation of the two peaks obtained for this metabolite is obtained by assuming equal signal contribution or equal maximum absorbance of the individual spectra of the two detected components.