Coalescence of Expanding Bubbles: Effects of Protein Type and Included Oil Droplets - Food Colloids: Self-Assembly and Material Science - page 375

probability of coalescence is, on average, higher for the larger bubbles. This is
probably because of the greater buoyant force acting on larger bubbles, which
increases as the cube of the bubble diameter. Nonetheless, this factor does not
significantly alter the main trends or the differences between the proteins. For
example, for C b ¼ 1-4 wt%, the mean bubble diameter with WPI was 17 mm,
while for OA it was 9 mm, even though the value of F c for OA-stabilized
bubbles was higher overall than for WPI-stabilized bubbles, i.e., contrary to the
expected trend if stability were to be solely determined by size.
Similarly, the variable tendency for the injected bubbles to cluster together
on injection, either in the centre of the barrier, or at its walls, did not seem to
have a significant effect on the measured values of F c . Using a high-speed
camera we have confirmed the findings of Liger-Belair 8 that each bubble
coalescence event at the planar interface, even at these high protein concen-
trations, is extremely rapid (complete within o 0.5 ms) and locally violent,
causing plumes of aqueous phase to be ejected into the air and inducing
pronounced (short-term) distortion of the neighbouring bubbles.
By analysing the sequence of individual coalescence events in some bubble
clusters in detail, we have shown 1 that coalescence of one bubble (small or
large) does not significantly increase the probability of a near neighbouring
bubble also coalescing. In fact, in the short term, it actually appears to reduce
slightly the probability of the neighbour subsequently coalescing. Possibly this
is due to a local decrease in the stress in the interface on expansion that
triggered the coalescence event in the first place. Alternatively, coalescence of
one bubble may raise the local bulk concentration of protein, thereby reducing
the local interfacial tension gradients caused by the original expansion. While
these effects cannot be easily proved, it seems reasonable to consider them as
likely to originate from the generation or existence of local variations in surface
composition as the system expands.
Even down to C b ¼ 1 wt%, based on various measurements made else-
where, 9,10 one expects diffusive adsorption to the newly created interface on
expansion to be easily fast enough to replenish regions depleted in adsorbed
protein at the expansion rates employed, assuming the decrease in the surface
protein load on expansion is evenly distributed throughout the adsorbed layer.
Another key feature of the results in Figure 3 is that, even at the highest
protein concentrations studied, the measured value of F c levels off, but it does
not reach zero. Thus, significant instability remains (e.g., F c 4 20%). This
would represent, e.g., a significant change in the bubble-size distribution and/or
loss of gas from a product. It should be emphasized here that, under all the
experimental conditions relevant to Figure 3, the bubbles were completely
stable to coalescence for several hours in the absence of expansion. Only on
expansion does instability arise, as most of the bubbles that do coalesce do so
during the expansion; the remainder that coalesce typically do so within 1 min
of the expansion ceasing. 1 In other measurements we have shown 1 also that
when the expansion rate is reduced by a factor of 10, to 0.007 s 1 , the measured
values of F c are similar, although as expected F c does increase roughly in
proportion to the final extent of expansion (A/A 0 ).