Splitter imperfections in annular split-flow thin separation channels: experimental study of nonspecific crossover.

The separation performance of a split-flow thin (SPLITT) separation device depends on uniformity of channel thickness and the precise placement of the flow splitters at fixed distances between the channel walls. The observation of nonspecific crossover, that is, the transport of sample materials across the channel thickness without the influence of an applied field, has routinely been taken to indicate the presence of irregularities in splitter shape or placement. Computational fluid dynamics software may be used to predict the influence of splitter imperfections on nonspecific crossover, where it is assumed that sample transport is by convection alone. A previous study has shown how small inlet splitter imperfections can account for the relatively low levels of nonspecific crossover observed with typical annular SPLITT devices. This study, however, could not distinguish between the possible sources of nonspecific crossover; hydrodynamic lift or shear-induced diffusion could have contributed. To confirm the validity of the computational approach, a series of experiments has been carried out on a channel having a deliberately and severely bent splitter. Nonspecific crossover was measured for a range of inlet and outlet flow rate ratios, with the bent splitter placed at both the channel inlet and outlet. The severity of the splitter distortion was sufficient to produce significant nonspecific crossover over a wide range of flow conditions. Good agreement was found between experiment and prediction based on computational fluid dynamics, with experiment generally showing only slightly higher crossover than prediction. The quantitative agreement for this extreme case suggests that the contribution to nonspecific crossover due to geometrical imperfections can be well described using computational fluid dynamics.