There has long been an interest in using molecular simulation methods to examine diffusion behavior of moderately complex organic molecules containing up to a few tens of heavy atoms. Areas in which knowledge of such behavior is invaluable include drug formulation and delivery, dye diffusion in synthetic or natural polymers, food storage and spoilage, pervaporation, and toxin diffusion in protective fabrics or films.

In view of of the size of penetrant molecules in systems of interest, which are often highly viscous liquids or glassy amorphous materials, diffusion coefficients tend to be of order 10^-9 cm²/s or (much) smaller, which is at least two orders of magnitude lower than can be routinely predicted using the classical molecular dynamics approach in which the mean squared particle displacement is followed as a function of time and analyzed using the standard Einstein relationship. Moreover the structural and chemical complexity of the molecules of interest also tends to limit the application of other techniques such as those based on transition state theory, which in principle are capable of handling larger penetrant species in more rigid environments. Consequently a number of workers have adopted an alternative approach in which an artificial external force is applied to a diffusing penetrant molecule while simultaneously performing molecular dynamics simulation. The resulting trajectory is then analyzed with a view to identifying a regime in which the penetrant moves with a steady state velocity, which can then be used to infer a diffusion coefficient.

Potential limitations of the forced diffusion approach are that measurable displacements may be detectable only when an unphysically large force is applied, leading to distortion of the penetrant and adoption of improbable conformations or orientations. Moreover, reproducibility may be impaired owing to the tendency for the 'diffusing' species to become trapped in the heterogeneous atomistic environments associated with materials in the vicinity of the glass temperature. The current situation with regard to this approach is that there have been reports of positive results [1], though it should be remarked that the available evidence is limited and there do not appear to have been attempts to systematically evaluate the approach by application to systems for which extensive experimental data exist.

The present talk will discuss application of the forced diffusion approach to examine translational diffusion in viscous matrices, using systems such as fluorescein in concentrated sucrose solution, for which related simulations have suggested that equilibrium properties can be well predicted using accurate force fields, such as the COMPASS force field used in this work. Using a proven force field and method thereby allows us to concentrate on assessing the merits of the forced diffusion approach to the maximum extent possible.

[1] see, for example, Jacobson, S.H., Pharmaceutical Technology, 23, 120 (1999)