The sentence before the one you mention explains it; the only problem with the top diagram in the header image is that it doesn't make it clear that opposite the CO₂ side of the stream, there is an air side. Seems like the diagram represents their experimental model (two exit streams), not the practical application (three exit streams).

As the team hoped, this arrangement caused suspended particles with positive surface charges to concentrate towards the CO₂ side of the water stream, and those with negative surface charges to concentrate towards the air side, leaving the centre of the stream more or less particle-free.