On the insensitivity of liquid diffusivity measurements to deviations from 1D transport

Abstract

In liquid diffusion experiments, the diffusion path can be partially obstructed by bubbles/voids, or oxide layers between the segments of a diffusion couple. In addition, there have been claims of macroscopic manifestations of `wall effects', i.e. a perceived dependence of the diffusivity D on distance from the container wall over macroscopic dimensions. We have numerically simulated the evolution of the concentration field in a 2D diffusion sample in response to such deviations from 1D transport. We found that, owing to the smoothing of the concentration distribution by radial diffusion, significant deviations from the ideal 1D concentration field occur only in the axial direction. On evaluation of the concentration field for an apparent diffusivity , either by the semi-infinite capillary methodology or by the method of Codastefano et al (1977 Rev. Sci. Instrum. 48 1650), we found that local obstructions of the transport path have to be in excess of half the transport path's cross-section to result in . Furthermore, we conclude that the radial concentration gradients found in solidified diffusion samples are no evidence for a `wall effect' but are likely to indicate convective transport contamination.