(649d) Multiscale Water Diffusion Measurements in Poly(ethylene glycol) and Glycerol Solutions

A comprehensive understanding of water and solute transport and separation in membranes requires detailed knowledge of the diffusion behavior of water in polymers. Many studies that have probed water and solute diffusion in polymer membranes have employed gradient-driven flux measurements, which can only capture macroscopic transport behavior. Opportunity exists to combine both macro-scale and micro-scale diffusion measurements to investigate the molecular underpinnings of water and solute transport.

This work employs three complementary techniques that probe water self-diffusion at significantly different length and timescales to elucidate diffusion phenomena at both the macroscopic and microscopic level. Pulsed-field gradient (PFG) NMR measures solvent self-diffusion in the bulk at time scales on the order of milliseconds to seconds and length scales on the order of micrometers. Overhauser Dynamic Nuclear Polarization (ODNP), an electron paramagnetic resonance technique, probes microscopic water dynamics within ~1 nm of a spin probe on the order of picoseconds to nanoseconds. ODNP is thus more sensitive to the local chemical environment around a spin probe, which can be tethered to a polymer chain, while PFG-NMR captures diffusion in the bulk. Molecular dynamics (MD) simulations probe the structure and dynamics on the order of 1-100s of picoseconds for waters in hydration layers to further contextualize ODNP results.

Water self-diffusion coefficients in both PEG and glycerol solutions were measured across a broad range of PEG or glycerol concentrations (0% to nearly 100%). PFG-NMR and ODNP water diffusivities exhibited distinct behavior in dilute solutions whereas their behavior was more similar as solutions became concentrated. Diffusivities of hydration shell water molecules probed by MD simulations exhibited similar behavior to diffusion in the bulk. In both solutions, PFG-NMR diffusivities exceeded Stokes-Einstein predictions based on solution viscosities, with greater deviations observed in the PEG solutions. Modeling the solutions using a mixture free volume model was shown to provide close agreement with experimental self-diffusivities.