There’s been a lot of interest recently in what goes on inside droplets made up of more than one fluid as they evaporate. This can be entertaining with liquids like whiskey or ouzo, but it has practical applications in ink-jet printing and manufacturing as well. And a new experiment suggests that we’ve been fundamentally wrong about what drives the flow inside these drops.

As these drops evaporate, a donut-shaped recirculating vortex forms inside them, as seem in the cutaway views above. Conventional wisdom says that vortex is driven by surface tension. Evaporation of components like alcohol is more efficient at the edges of the drop, and as the alcohol evaporates, it creates a higher surface tension at the drop’s edge than at its peak. Marangoni forces then pull fluid down toward the edges, creating the vortex. That explanation is  consistent with observations of a sessile drop sitting on top of a surface (left side of images).

But those observations are also consistent with another explanation: evaporating ethanol makes the local density higher, so alcohol-rich parts of the drop rise toward the peak while alcohol-poor regions sink. This difference in density would also create a flow pattern consistent with observations. So which is the real driver, surface tension or gravity?

To find out, researchers flipped the drop upside-down (right side of images). When hanging, the preferred flow direction due to surface tension doesn’t change; flow should still go from the deepest point on the drop toward the edge. But gravity is swapped; alcohol-rich areas should be found near the edge and attachment points of the drop because buoyancy drives them there. And that is exactly what’s observed. The flow direction inside the hanging droplet is consistent with the direction prescribed by buoyancy-driven flow, thereby upending conventional wisdom. It turns out that gravity, not surface tension, is the major driver of internal flow in these multi-component droplets! (Image and research credit: A. Edwards et al.; via APS Physics; submitted by Kam-Yung Soh)