Use of multiple emission lines and principal component regression for quantitative analysis in inductively coupled plasma atomic emission spectrometry with charge coupled device detection

Abstract

A procedure for the automatic selection of spectral lines, based upon principal component regression (PCR), is described. The procedure analyses the consistency of the pattern of emission signals from a number of lines in both a series of standard solutions and the sample solution. From the calibration experiment the principal component associated with the largest eigenvalue is shown to be linearly dependent upon concentration and hence can be used to determine the unknown analyte concentration in the sample solution. The detection limit for this principal component is shown to be lower than the detection limit of any of the individual spectral lines. For example, a set of five Fe lines has a detection limit of 5.4 ng ml^–1 for the first principal component, compared to the lowest detection limit of 11 ng ml^–1 for Fe II 260.709 nm. In addition, the remaining principal components are shown to be independent of analyte concentration, but dependent upon the presence of unknown, additive spectral interferences. The procedure was used to determine 0.5 µg ml^–1 of Cr, Mn and V in a matrix containing Ce, La, Dy, Fe, Mo and Co. Six Cr lines, three Mn lines and seven V lines were used for quantitative analysis. Spectral interferences were correctly predicted for 2 Cr lines and 3 V lines, whilst the predicted concentrations for the three analytes were 0.51 ± 0.01, 0.49 ± 0.01 and 0.51 ± 0.01 µg ml^–1, respectively.