Quantitative analyses of multicomponent fluorescence data by the methods of least squares and non-negative least sum of errors

Abstract

Quantitative information can be obtained from fluorescent mixtures using the methods of least squares or linear programming, based on previously determined calibrated excitation-emission matrices (EEM) of known components, even in cases of severe overlap and poor slgnal/noise ratio. The least squares fitting technique was found satisfactory for analysis of a three-component mixture if all components were known. Linear programming, used to minimize the total error with Individual errors constrained to be positive, is preferred when not all components are known. Instrument stability Is a major limitation of accuracy if slow scanning instruments are used. © 1977, American Chemical Society. All rights reserved.