Background Correction in Multicomponent Spectroscopic Analysis Using Target Transformation Factor Analysis

Abstract

A principal components regression (PCR) is used to construct a calibration matrix for multicomponent spectroscopic analysis which corrects for background absorption due to variable concentrations of unknown species in a sample matrix. Mixed standards are used in the method, and spectra of the pure background components are not needed. The background correction capability of the method is demonstrated with the use of the UV spectra of aqueous standards prepared from cobalt (II) nitrate and nickel (II) nitrate. Known concentrations of chromium (III) nitrate were added to simulate a background which gave rise to relative concentration errors as high as 70% without the background correction capabilities of the method. With the background correction capabilities of the method, the relative concentration errors were reduced to less than one percent. Preliminary results are presented from an application of PCR to determine the concentration of two active ingredients in a pharmaceutical product in the presence of weakly absorbing tablet excipient materials.

Principal components regression for routine multicomponent UV determinations: A validation protocol

“…is the RMS spectral error for the reproduced data matrix. 8 In Table 1, if a one-factor model is used, the estimated spectral error is about 0-014 absorbance units, which is much greater than the measurement noise expected from the HP8452A diode array spectrometer used in this work. If a two-factor model is used, the estimated spectral error is about 040008 absorbance units, which corresponds nicely with the anticipated spectral error for the HP8452A.…”

Factor analysis and Kalman filter studies of severely overlapped amino acid derivatives in thin-layer chromatography

“…Background Calibration-Factor Analysis-Target Transformation. This method of factor analysis has been described by Gemperline and co-workers in a recent publication (13). These authors reported good results for the determination of component concentrations in the presence of a variable background contribution, provided that the background components were present in a calibration data set.…”

Mixture analysis using factor analysis I: Calibration and quantitation

“…The probability level from the F-test was used to identify unacceptable test vectors. Fredericks el al.," Brown et al 43 and Gemperline et al 44 recognized that the residual sum of squares for response variables could be used to detect invalid samples in principal component regression. Haaland and Thomas also reported an F-test for PCR and PLS that can be used to detect invalid samples.…”