Fig 4.

(A) To explore the role of fluid densities and channel lengths the operation of the density sorter chip, the bead-sorting experiment in Fig 3 was repeated using a single fluid density in all three inlets (1.035 g/mL) and three different channel lengths (22, 118, and 207 mm). Without layers of different fluid densities as in Fig 3, the purity of the beads at the outputs of the straight channel chip (22 mm) dropped from an average of 99.8% to 83.5% for the top outlet and from an average of 87.5% to 79.0% for the bottom outlet. Increasing the channel length increases the purity of the beads at each output, with an average of 95.5% polyethylene beads at the top outlet and an average of 91.0% polystyrene beads at the bottom outlet of the 8-turn (207 mm) density sorter chip. The uncertainties in percentage points represent the standard deviation of three replicates of each experiment. (B) For insights into these results, we simulated beads in the 207 mm chip using our density_sorter.py software (S1 File) at 30 second intervals. The larger but less-dense polyethylene bead (green circle) floats to the channel ceiling in about 30 seconds, and the smaller but more-dense polystyrene bead (red circle) sinks to the channel floor in about 3000 seconds. These results suggest that without multiple density layers, particles are free to sink or float to the channel walls where they may stick irretrievably (thus reducing sorting efficiencies compared to the results in Fig 3), though lengthening the channel can improve sorting efficiencies somewhat.