Extraction of lifetime distributions from fluorescence decays with application to DNA-base analogues.

Several important aspects of fluorescence decay analysis are addressed and tested against new experimental measurements. A simulated-annealing method is described for deconvoluting the instrument response function from a measured fluorescence decay to yield the true decay, which is more convenient for subsequent fitting. The method is shown to perform well against the conventional approach, which is to fit a convoluted fitting function to the experimentally measured decay. The simulated annealing approach is also successfully applied to the determination of an instrument response function using a known true fluorescence decay (for rhodamine 6G). The analysis of true fluorescence decays is considered critically, focusing specifically on how a distribution of decay constants can be incorporated in to a fit. Various fitting functions are applied to the true fluorescence decays of 2-aminopurine in water-dioxane mixtures, in a dinucleotide, and in DNA duplexes. It is shown how a suitable combination of exponential decays and non-exponential decays (based on a Γ distribution of decay constants) can provide fits of equal quality to the conventional multi-exponential fits used in the majority of previous studies, but with fewer fitting parameters. Crucially, the new approach yields decay-constant distributions that are physically more meaningful than those corresponding to the conventional multi-exponential fit. The methods presented here should find wider application, for example to the analysis of transient-current or optical decays and in Förster resonance energy transfer (FRET).