Magnetic Resonance Imaging of Aggregated Emulsions

Using a high resolution microscope is useful since it reveals individual droplets and aggregates, but the price is a very restricted field of view. Increasing that field of view by using low resolution microscopy introduces a new problem, which is that even for a near-index matched system there is significant light scattering and therefore limited optical penetration. It is not possible to explore the whole bulk sample in this way. To look at larger length scales and to be able to “see” right into the sample, including in the bulk, we use MRI ( see section on NMR technique).

For successful application of MRI to emulsions it is to necessary to enhance the contrast between the dispersed oil phase and the continuous (“water”) phase. To achieve this we dope the continuous phase with copper sulphate. This is a standard trick that works by shortening the T1 relaxation time in the aqueous phase, so that when we run fast repetitive sequences the oil signal saturates and only the water signal remains. In our experiments, we use 3D Fourier spin echo sequencing rather than the much faster “slice” techniques typical of medical imaging. This means that every one of our 70x70x256 image elements is averaged over the entire acquisition time, which is typically 90 minutes for a complete 3D image. The long acquisition time is a consequence of imaging at high resolution using image voxels of side length ~60 microns. The image quality could be further improved at the expense of an even longer acquisition time, but this may in turn reduce the likelihood of spotting interesting dynamics. Our experiments are performed at 23 and 300 MHz, using a sample housed in a glass cylinder of 4.2 millimetre internal diameter.

To get an appreciation of the likely image quality and resolution, we imaged a test system of crushed glass microscope cover slips.

crushed cover slips	corresponding MRI picture

The left-hand image shows a sample tube inside the spectrometer loaded up with a pile of crushed 130 micron-thick cover slips immersed in the same doped continuous phase that we use in our emulsions. The right-hand image shows the corresponding MRI picture. Since glass cover slips are invisible to MRI they mimic the role of the oil in an emulsion. The image on the right is actually of the continuous phase, but has been rendered to show the regions free of continuous phase, in other words the cover slips. The jagged edges of the cover slips are clearly visible, giving a qualitative demonstration of the available imaging resolution. This level of resolution should be adequate to capture significantly sized long-lived void regions in aggregated emulsions.

Based on sedimentation studies we have investigated emulsion compositions spanning oil volume fractions from 30% to 40% and PEO concentrations from 0.6% to 0.9% w/w of continuous phase.

We do not see any evidence for the spontaneous formation and growth of cracks or voids. There are two important provisos, however. One is our limited spatial resolution, which means that small cracks might exist but we cannot see them. The other is temporal – the imaging technique amounts to a system-wide averaging over the entire acquisition time, so that for a crack to be visible it must occupy the same pixel for at least 90 minutes. Any cracks which were to open up and then close within 90 minutes, or wholly relocate in that time, would also be invisible to us. So a more precise statement is that we have not seen any spontaneously-formed large and persistent cracks. The original hypothesis envisaged cracks that formed spontaneously and gradually opened up and grew in time. Out MRI work therefore finds against this hypothesis.

The emulsion sedimentation observed using MRI is consistent with what we see visually. We note in passing that MRI does not average across the sample, unlike optical and ultrasonic methods, and could be helpful in looking for non-uniformities in flow across the sample diameter. In addition, we also observe foaming at the sample surface. We believe that the foam is due to the spontaneous formation of gas bubbles in the bulk of the emulsion.

MRI still showing bubble, taken from video

The following image shows the upper reaches of a single vertical slice through a sedimenting emulsion, extracted from the full 3D image.

The emulsion is 30% oil aggregated with 0.9% PEO. Continuous phase (“water”) shows as white. The darkest region on either side of the image is the sample container and beyond. The speckled grey is the bulk emulsion itself. The dark spherical features at the top of the sample are gas bubbles. Bright regions occur wherever continuous phase is predominant. This emulsion sample evolved in time, as the following time lapse movie shows.

Download MRI sedimentation video showing bubble [AVI, 500Kb]

The movie follows the time evolution over a total of 76 hours for the single vertical slice displayed above. At the top of the sample is a region dominated by foam, evidenced by the large and numerous black bubbles. This matches visual observation of bubbles forming at the top of emulsion samples, and also the passage of bubbles as captured in our confocal microscopy images (see section on Network ripples). The foam is very active, with bubbles that grow, collapse and otherwise change throughout the movie. As the emulsion sediments, a bright (water-rich) region forms both around the bubbles in the foam and above the remaining emulsion. This region expands as the emulsion sediments and the boundary between continuous phase and the remaining emulsion moves downwards.

One point of particular interest captured in this sequence is the bright intrusion extending from the boundary into the emulsion. It moves downwards with the boundary layer, its shape slowly changing, and it becomes smaller over time. Since it is bright this intrusion is a region of continuous phase.

MRI bubble schematic

We believe that this feature arose from a gas bubble present even as the droplet network first formed, persisting as the network became rigid. This bubble shows as the black circular region immediately above the intrusion. As the boundary fell, slowly at first, the space in the droplet network once occupied by the bubble filled with water, eventually forming a closed water-filled region, as depicted in the following schematic.

The fact that such a region could be sustained at all indicates the presence of a structure having some degree of rigidity. Crucially, this void region shrank in time as the network slowly “healed” and filled it. For fractures to exist in the mode hypothesised this filling presumably would not happen, since it suggests that any spontaneous voids that might be formed within the system would close up, consistent with the network recovery we observe following the passage of bubbles in confocal sequences (see section on Network Ripples). The MRI bubble sequence is therefore further evidence that the collapse of a network is not a consequence of some increasing population of propagating fractures. However since we have studied only part of the emulsion life cycle we cannot exclude the possibility that large and persistent voids might be relevant at some other stage in the life of the emulsion system, though we have no evidence to support this proposition.