Network Ripples

An aggregated emulsion in the bulk might appear to be doing very little, but our investigations show that at a microscopic level there is a great deal of activity. In particular, this is true of so-called transient gel states in which “nothing happens” prior to the gel collapsing.

Dynamics exist on a variety of length and time scales. For example, an aggregated emulsion (with oil denser that water) actually sediments very slowly, in our experiments giving a vertical movement between stacks of a few microns every 5 minutes. This corresponds to a bulk sedimentation rate of 10 centimetres in 130 days, which is why the bulk sample appears quiescent.

At the other end of the scale, at a fine level it is possible to witness “rattlers”, single droplets that move quickly back and forth between two locations. The image sequence below shows an example:

We interpret rattlers as droplets oscillating between two potential wells of similar depths.

We have also detected some much more dramatic movements. Some of these take the form of relocations of a few slices relative to those before and after, amounting to a “swirl” within a stack that is otherwise assumed to be a snapshot. A typical swirl is a large scale movement (spanning more than 100 microns square) with a time frame of 20 seconds. We see these in samples comprising 30% oil and 0.9% PEO. Swirls are a challenge to feature location and subsequent tracking since they amount to a “dislocation” in a single image stack, and individual features cannot be identified in the region of a dislocation.

At lower PEO concentrations the system moves more quickly. The sequence below shows three slices from the mid-region of a single confocal stack of an aggregated emulsion comprising 30% oil and 0.1% PEO. This system creams.

In this sequence time is advancing from left to right. Gravity is down. The middle frame shows a region where droplets appear distorted. It is tempting to interpret this as a shear band of some kind, but we do not think this is correct. We believe that the distorted region is due to a large scale “heave” occurring on a time frame considerably shorter than the time taken to scan a single frame. In the sequence above, scanning takes 3.7 seconds, with the scan occurring in each frame from right to left. Recall that in a given frame what happens before the scanning beam reaches a particular region may not be recorded. Likewise what happens within a region after the beam has passed over may not be captured.

These distorted regions are common. One source may be large (> 100 microns) network distortions due to relative motion between significantly larger regions of network elsewhere in the sample, though we have no evidence of this. We believe that some, if not all, of these effects are due to bubbles in the system.

Sedimentation experiments show evidence of foam layers, a sign that the system spontaneously forms gas bubbles that we believe to be due to the volatile nature of the oil. We have been lucky enough to capture bubble movement using the confocal. The series of still images below shows a bubble (showing black) entering the bottom of the frame and travelling to the top, with a second bubble following:

These images are from an emulsion system comprising 30% oil and 0.2% PEO. Time runs from left to right. The physical dimensions are again 107.3 microns square, but this time each image is only 256 by 256 pixels, so the slice acquisition time is approximately 1 second. The larger bubble is about 30 microns in diameter, and moves at roughly 1/10 000 metre per second.

Droplets adhere to the bubbles, though not the “leading edge” of the large bubble. The bubbles distort the droplet network, which partially recovers after the passage of the bubbles and continues sedimenting. There’s a trail of disruption in the bubble’s wake which is easier to see in this movie - Download movie clip [AVI, 300Kb].

The movie is actually a slice by slice journey into a single 32 slice stack, so embraces less than 40 seconds of real time though not at fixed z.

Automated quantitative 3D imaging of such fast dynamics is difficult. It is possible to extract interesting dynamical insight in 2D by careful manual analysis, however.

The image (to the right) shows the deformation in the droplet network due to a bubble rising from beneath.

The image corresponds to a few seconds before the bubble sequence above: the bubble is not yet in the field of view. The arrows indicate individual droplet velocities, with the arrow length proportional to droplet velocity, and show that droplets move in response to the bubble even before the bubble has reached them