axis in graphs like those in Figure 8.3 does not represent the water activity of
the food, at least at lowaw. In an actual food sample of low water content,
one may imagine that at any spot a certainawprevails, but it cannot be
determined and it would probably vary with place and time.
Many foods show even stronger hysteresis than in Figure 8.4. It is
further seen in Figure 8.4c that intermediate sorption curves are obtained
when one starts drying or wetting somewhere in between, especially when
starting at fairly lowaw; other materials show similar, though not precisely
the same, behavior. The explanation of the hysteresis is far from clear. It can
be stated that the drying involved in determining the isotherm alters the
physical state of the food in such a way that it cannot relax again to the
original state after taking up water, or only very sluggishly. Both curves may
be considered to represent a metastable state, in the sense that the values do
not change over the time scale of interest. This is true enough, but gives little
further understanding. In the Note at the end of Section 8.3, a possible
mechanism (closing of pores in the material) is mentioned.
Another factor would be that equilibrium is not reached because the
rate of diffusion of water through the sample is too small. Figure 5.9 gives,
as an example, a diffusion coefficientD& 10 ^15 m^2 ?s^1 at about 2.5%
water, and as water content becomes smaller,Dbecomes ever smaller.
Applying the simple Eq. (5.13) (x^02 ¼D?t 0 : 5 , withx^0 ¼diffusion distance—
FIGURE8.4 Sorption isotherms—w^0 in g per g dry starch versus water activityaw—
of native potato starch, obtained when decreasing water content (desorption) and
when increasing it (absorption). (a) Linear scales. (b) Part of the same data on a log/
log scale. (c) Part of the same data, also showing intermediate curves (broken). (After
results by C. van den Berg, Ph.D. thesis, Wageningen University, 1981.)