A NEW APPROACH TO THE RESOLUTION OF THE EXCITATION-EMISSION SPECTRA OF MULTICOMPONENT SYSTEMS

A method has been developed for the calculation of the emission, absor ption, and concentration characteristics of chemical species from the fluorescence spectra recorded for a series of solutions of varying com position at different excitation wavelengths. The three-dimensional fl uorescence-intensity array (whose ijkth element is the fluorescence in tensity of the ith solution at the jth excitation and the kth emission wavelength) is assumed to be a trilinear product of the excitation (X ), emission (Y), and concentration (C) characteristics of the emitting species. The decomposition of experimental spectra is carried out in three steps. In the first step, an eigenvector analysis of the Gauss t ransform of the fluorescence array is performed, whose rank is the low er bound to the rank of the matrices X, Y, and C, and thereby to the n umber of independently emitting chemical species present in the system . In the second step, factorization of the eigenvectors of the Gauss t ransform of the fluorescence array is performed, which gives approxima te factor matrices X, Y, and C. The obtained factors correspond to exa ct decomposition, if there is no error in the data. In the third step, the factors are refined by means of a nonlinear least-squares method, which is directed at minimizing the sum of the squares of the differe nces between the elements of the observed and calculated fluorescence array. Application of the method to the fluorescence spectrum of mixtu res of anthracene and 9-methylanthracene resulted in perfect resolutio n of the fluorescence spectra into the component excitation and emissi on spectra and correct profiles of component concentrations. It is als o shown that if the numbers of absorbing and emitting species are not equal, as occurs for the beta-naphtol-beta-naphtolate system in acid a nd neutral aqueous solution, unambiguous decomposition of the spectrum is impossible to carry out. Copyright (C) 1996 Elsevier Science Ltd