Accounting for Impurities in Spectroscopic Multicomponent Analysis Using Fourier Vectors

Abstract

Two methods are presented for dealing with impurities in unknown samples subjected to spectroscopic analysis. In one method, a residual is calculated, and its relative length is used to “flag” that particular sample as a suspect. In the second method, a representation is found for the impurity from the analysis of standard samples without knowing its concentration. This representation is used to deal with the impurity in unknowns. All spectra are Fourier transformed, and terms of the transforms are selected as coordinates of vectors. The vectors for standard mixtures are factor analyzed to determine eigenvectors which are used to represent both the analytes and impurities.