Time-resolved fluorimetry of two-fluorophore organic systems using artificial neural networks

In this paper, we study the ability of determining the lifetimes tau(1.2) of fluorophores excited states and the ratio of their fluorescent contributions in a two-fluorophore system with the help of time-resolved fluorimetry in its modification when the lifetimes tau(1.2) may be smaller than the exciting pulse duration tau(p) and the receiver gate duration tau(g). The investigation has been performed under the assumption that there are no intermolecular interactions that could influence the times of fluorescence decay. The described three-parameter inverse problem was solved with the help of artificial neural networks (ANN). Numerical modeling and physical experiment with binary dyes solution have been performed. Both have demonstrated that the ANN algorithm can determine with acceptable precision the lifetimes tau(1.2) down to 1 ns at tau(p) and it values equal to 10 ns (the gate delay being changed in 2 ns steps). Practical stability of the ANN algorithms to noise in the input data and to non-controlled variations of shape and duration of the exciting radiation pulse has been investigated. It is shown that for actual level of noise in kinetic curves, the ANN algorithms give significantly better results in solving the studied three-parameter inverse problem than the variational algorithms. It is intended that the considered modification of time-resolved fluorimetry will be used to build the future complex method of fluorimetry of composite multi-fluorophore compounds. (C) 2002 Elsevier Science B.V. All rights reserved.