The theory for the asymmetrical flow field-flow fractionation channel was evaluated experimentally. The assumption of a constant crossflow along the channel was satisfactory. The zone-broadening theory was extended to take Into account the longitudinal diffusion and the width of the starting sample zone. In the present apparatus the longitudinal diffusion was negligible but may have to be considered for lower outlet flow rates. The axial flow velocity gradient has the advantage that the contribution by the width of the starting sample zone to the final zone width decreases. For rectangular channels with a crossflow rate to outlet flow rate ratio of 10, a 1.5cm-wide starting zone contributes to the plate height by 0.001 cm only. The effect of channel length was Investigated. Although theory predicts that the channels can be as short as 10 cm, a 0.01-0.04-cm contribution to the plate height not accounted for by theory suggests that It Is better to use longer channels. An upper limit In length If found around 60 cm. The effect of crossflow rate, outlet flow rate, and channel thickness on speed, efficiency, and dilution was demonstrated.

A new trapezoidal geometry of the asymmetrical flow fieldflow fractionation (FFF) channel Introduces an additional means to regulate the longitudinal flow velocity. The trapezoidal geometry, where the breadth decreases on going toward the channel outlet, makes It possible to level out the steep linear velocity gradients that can appear In the rectangular channels. Equations for the channel flow velocity gradient and the void time are derived. In addition, an equation for the nonequilibrium plate height Is derived for asymmetrical channels. The equation suggests that the same efficiencies should be obtained In asymmetrical as In the original symmetrical channels under similar flow conditions. The relative peak dilution occurring In channels of different designs Is evaluated. The results favor the trapezoidal versus the rectangular channel and asymmetrical versus symmetrical channels. The plate numbers calculated from the derived equations were compared with those found experimentally. The latter were found to deviate from the calculated numbers at high retention levels. Approximately 2600 plates have been reached.

No abstract

No abstract

No abstract