On the importance of including intermolecular correlations in the measurement of polymer interdiffusion by fluorescence nonradiative energy transfer

There has been a great deal of recent interest in monitoring polymer interdiffusion with fluorescence nonradiative energy transfer (NRET), a typical experiment consisting of polymer chains labeled with a donor chromophore interdiffusing with chains labeled with an acceptor chromophore, causing an increase in energy transfer efficiency, E. A factor nearly always ignored in models of this process is the fact that chromophore interactions are screened out because of intermolecular correlation effects, leading to a lower E than would otherwise be expected. The model of Dhinojwala and Torkelson (Macromolecules 27, 4817 (1994)) has been modified to include correlations in order to explore the impact that ignoring correlations has on the diffusion coefficients calculated in such experiments. When correlations are included in the model, E is proportional to (D(p)t)(1/2)/x for t < x(2)/(16D(p)) (where x is the thickness of the donor-labeled film and D-p is the polymer diffusion coefficient), similar to when correlations are ignored. Within this linear regime errors in D-p associated with ignoring correlations are typically less than 20% for logically defined systems. which may be insignificant given the uncertainty with which diffusion coefficients are normally reported. However, this study clearly demonstrates that precautions should be used in all experiments of this type to stay in this linear regime, as improperly accounting for or ignoring correlation effects at longer diffusion times can lead to the spurious result that apparent diffusion coefficients decrease with annealing time, by factors of 2 to 5 to as much as an order of magnitude or mon at sufficiently long time. Systems designed to minimize correlation effects should have a large Forster radius and a sufficiently high ratio of accepters to donors; with such a situation and the use of NRET data from the linear regime of E vs. t(1/2), it is possible under many experimental circumstances to determine diffusion coefficients within small error even when the effects of correlations are ignored.