Crispness, Mechanical and Acoustic Properties of Flakes
As shown by Fig. 9, the mechanical properties measured by this test
cannot fully cope with the mapping issued from the sensory analysis, at least
its first component which is “crispness” (Fig. 5A). This is also reflected by
the low correlation obtained between crispness, evaluated by the first sensory
dimension CVA1, with maximum forces reached during complete compressions
or with slopes of first compression step curves (r2 = 0.4 and 0.6, respectively).
Conversely, all samples belonging to the group of most crispy products
exhibited lower Fmax and Slmin values, which support the relevance of the
mechanical test. The main discrepancy arises from samples G and L, having
very different Fmax and Slmin values, although they belong to the same group
of crispness. They are separated from the other samples by the second sensory dimension, i.e., sensory attributes “sound duration” and “loudness”
(Fig. 5A,B). The average amplitude of acoustic signal, AAS, and the fraction
of peaks of higher amplitude, Rp+, are fairly correlated to the first component
of the sensory analysis which is “crispness” (r2 = 0.63, P = 0.0012 and
r2 = 0.65, P = 0.001, respectively), and suggest that acoustic measurement
may be used for assessment of crispness (Fig. 13). Moreover, the comparison
between Figs. 5A and 10 is encouraging for crispness prediction by an acoutic measurement, even if the analysis in acoustic temporal domain fails into
separating samples G and L from the others. They have AAS and Rp+ values
that are very close to each other, but the sensory analysis clearly separates
these samples from the others in terms of loudness. This limitation may be
due to the speed of biting, which is different from mechanical compression.
Due to the difficulty of simulating human mastication, a more profound
analysis of the possible relation between this sensory dimension and emitted
acoustic signal would require the use of an electronic sensing system that is
directly located on the cheeks and jaws of the panelists, as done by Peng et al.
(2002) for assessing the fracturability of various solid foods. The way fractures
propagate during compression depends on the material structure
(Nicholls et al. 1995), and has an important role in the transmission of vibrations
in the air by acoustic waves. Less crispy products crushing leads to low
frequencies of acoustic signals, like for samples J and E. Moreover, these
samples emitted high-amplitude acoustic signals, while for samples H, K, M
and D, the acoustic emissions are balanced between low and high frequencies,
well split in the studied domain from 1 to 10 kHz.
As previously mentioned, the jaggedness of the signal is often used to
assess “crispness.” This variable was measured for the mechanical signal by
1 - r2), but no significant correlation was found here with CVA1. This apparent
contradiction can first be due to the use of the Kramer cell, which has a
large volume, and that fluctuations are limited by averaging and cushioning
effects. However, the testing volume (about 100 cm3) is close to the one
leading to the largest jaggedness obtained in the study by Nixon and Peleg
(1995). The explanation may rather rely on the fact that, in our study, the
samples only exhibit tiny differences in moisture content (Table 1). In most
studies on crispness of cereal products, the samples are equilibrated at different
moisture contents in order to emphasize the changes of this sensory
property. Thereby, crispness variations and the associated variations of jaggedness
are interpreted by modifications of the temperature of glass transition,
Tg. Differences between mechanical behaviors are mostly due to material
matrix and its components as the amount of glassy starch, or the one of water,
having influenced by lowering the Tg (Attenburrow et al. 1992; Le Meste
et al. 1996). In our case, the samples are all from commercial brands that were
expected to have satisfactory sensory properties, so changes in crispness are
not so significant, although sufficient to be recognized by the sensory panel.
These tinier variations of crispness may not rely on changes of Tg. Texture
behavior cannot be predicted only from the knowledge of the glass transition
temperature alone (Nicholls et al. 1995). Differences on mechanical properties
measured for flakes beds may also be due to the intrinsic properties of
the constitutive material of the flakes. The brittle behavior of starch–zein
systems has been shown not to depend on the Tg of components, but rather on the microstructural morphology of the matrix/particles biopolymer blends
(Chanvrier et al. 2005).