3.4. Air bubbles size distribution
The air bubble size was measured by assuming that
the particles are spherical. At least 200 air bubbles for
each sample were analysed to ensure an accurate
representation ofthe whole size distribution. Nevertheless,
the three-dimensional structure was not take
into account: air bubbles were assumed to be fractured
at the equatorial plan, but in reality, the spherical
particles could be fractured anywhere from the top to
the bottom. Contrary to the destructive methods that
are based on sample dilution, the direct method enables
to observe in situ the air bubbles shape and size inside
the frozen ice cream structure which represents a great
advantage ofthis method. So, the qualitative information
given by these images is interesting because it was
possible to observe the relative position ofeach phase.
Furthermore, we observed, for example that air bubbles
are not spread out very uniformly through the other
phases.
Fig. 8 compares the cumulative distribution ofthe airporosity e ¼ 0:29 and the same hardening temperature
Th ¼ 35C, but with different ice cream mix temperatures
at the SSHE exit, namely 0C, 3C and 5C.
So we observed that lowering the SSHE exit temperature
involved an important decrease ofair bubble mean
diameter: from 112 mm for samples only whipped in the
SSHE (Tf ¼ 0C) to 47 mm for the samples partially
frozen in the SSHE (Tf ¼ 5C). As mentioned by
Chang and Hartel (2002c) who observed the same
behaviour with ice cream samples directly removed from
a batch freezer, the freezing of the mix is necessary for
the stabilization ofthe small air cells. Indeed, by
decreasing the SSHE exit temperature, the ice cream
viscosity increases what limits the air bubbles coalescence
and modifies also the flow pattern ofthe frozen
mix.
4. Conclusions
bubbles size for three ice cream samples with the same