Effect of liquid phase composition on the experimental determination of binary gas diffusivities in an isothermal Stefan diffusion column

The Stefan diffusion column was designed in the 19th Century to determine binary gas diffusivities, D-AB?s. Typically, pure liquid A is overlaid by gases A and stagnant B, with a steady gas B sweep at the top to remove the evaporated-diffused A. Experimental diffusivities D-AB,D-exp may be obtained from transient interfacial descent data, but column end effects impacting such determination have been ignored or neglected in the literature. This study addresses experimentally and theoretically for the first time the role played by liquid phase composition on D-AB determination using the Stefan column. Specifically, changes in interfacial curvature were sought by adding a nonvolatile liquid I (glycerol) to a volatile liquid A (ethanol). Mixtures of known initial composition x(A0) were tested in isothermal evaporation-diffusion experiments at ?65??C using atmospheric air (B). A one-dimensional transport model for gas A was used to analyze the interfacial descent data. The average D-AB,D-exp errors did not differ significantly between the liquid mixture groups and those for pure ethanol (range for the latter of +9.9 to +17.1%). However, considerable scatter in D-AB,D-exp occurred as the mixtures became more dilute in ethanol, with the coefficient of variation increasing more than ten-fold in the x(A0)?=?0.146 group. This variability cannot be explained by differences in surface tension exclusively, but may also result from changes in liquid physical and transport properties as well as diffusion of liquid A. A more comprehensive modeling of the liquid phase in conjunction with gas phase transport is necessary to obtain accurate D-AB?s.