Robust estimates of solute diffusivity in polymers for predicting patient exposure to medical device leachables

Medical devices often include polymeric components, which contain additives or contaminants that may leach into patients and pose a health risk. Previously, we proposed a mass transport model that conservatively estimates the leaching kinetics and only requires the solute's diffusion coefficient in the polymer, $D$, to be specified. Because determining $D$ experimentally is time-consuming, we also parameterized empirical models to estimate worst-case $D$ values using only the solute molecular weight, $M_w$. These models were based on a modest database and were limited to 19 polymers and larger solutes ($M_w>100$ g/mol). Here, we assemble a much larger database, which enables us to construct more accurate models using a robust statistical approach, expanding the coverage to 50 device-relevant polymers and smaller solutes ($M_w<100$ g/mol). Then, we demonstrate several applications of these bounds, including modeling the release kinetics. Finally, we observe an interesting phenomenon, a discontinuous drop in $D$ of up to 25,000× for solutes with $M_w>50$ g/mol in glassy polymers. Using molecular simulations and cheminformatics tools, we propose a novel definition of the effective diameter of free volume channels in polymers, and we show that solutes larger than this channel size diffuse much more slowly.