Partial least squares resolution of multianalyte flow injection data

Abstract

The development of a multivariate calibration model to resolve the complex spectra produced by a mixed chemistry flow-injection manifold is presented. The combination of established spectrophotometric reactions for the simultaneous determination of diverse analytes is discussed and the value of applying partial least squares regression, particularly in the interpretation of reaction combination effects, is considered. The approach is demonstrated with reference to the simultaneous determination of phosphate and chlorine using a combined reaction system that incorporates a mixed molybdate–o-tolidine reagent. This has been achieved using a physically simple flow-injection manifold with photodiode array detection. Calibration is discussed in terms of the experimental design, pre-processing and the number of wavelengths used in modelling. Finally, results of the prediction of a new and independent test set are presented with relative root mean square values of 4.0% for phosphate and 2.4% for chlorine.