Laser-Induced Fluorescence Multicomponent Analysis in Full Spectra and Their Fourier Transforms by Principal Component Regression

Abstract

Laser-induced fluorescence spectroscopy (LIFS) has been used to analyze two- and three-component systems. Equations describing the fluorescent emission process have been incorporated into the Beer's law absorption analog to make multicomponent analysis applicable to LIFS. Principal component regression (PCR) has been employed to predict concentrations of components having severe peak overlap. Results obtained from using a pulsed nitrogen laser and a cw argon-ion laser as excitation sources are compared in order to elucidate some of the important features in applying fluorescence for multicomponent analysis. Furthermore, PCR analysis was also applied to Fourier transformed fluorescence spectra with the use of only 8 and 16 Fourier terms for prediction. The improvements from the Fourier processing are 50.1% and 19.8% in a two-component system, and 22.4% and 8.3% in a three-component system, for cw and pulsed laser excitation, respectively. The time required for calibrating with the Fourier transformed fluorescence spectral data is dramatically reduced to 67 s compared to 551 s for wavelength domain data. Multicomponent analysis of the four dyes (FITC, NBD, TMRITC, T.RED) and dye-amine conjugates used in fluorescence-based DNA sequencing, as well as the analysis of a highly overlapped polyaromatic hydrocarbon (pyrene, anthracene, fluoranthene) system is demonstrated. Standard error of prediction (SEP) values for cross-validation calculation of all training sets were <5%. When applied to time-resolved fluorescence measurements, multicomponent analysis can provide an alternative means of resolving fluorescence peak overlap output as numerical representations, especially in mixtures where components have similar fluorescence lifetimes and overlapped spectra.