Cracks generated due to desiccation
of wet colloidal systems are
ubiquitous, examples being nanomaterial films, painted walls, cemented
floors, mud fields, river beds, and even giant rocks. In all such
cases, crack patterns are often appreciably similar but for the length
and time scales, which can be widely differing. In this work, we have
examined the crack formation more closely to see if there exists some
generality with regard to the length scale of parameters and the formation
time. Specifically, using a commonly used colloidal dispersion and
optimized conditions to form polygonal network patterns rather than
isolated cracks (films of subcritical thickness), we have studied
the time evolution of the pattern parameters, the area occupied by
the cracks, their lengths, and the widths. As is well known, initially,
a network of cracks forms, which we term as the primary generation,
followed by interconnecting cracks inside the polygonal regions (secondary)
and, later, cracks spreading in local regions (tertiary). We find
that the area and the width increase nearly linearly with time with
the change in the slope corresponding to the change in the generation.
When normalized with respect to the final values, the trends obtained
for different film thicknesses overlap, the only exception being the
pattern containing unconnected cracks. Thus, the time evolution of
cracks is shown to be predictable based on width filtering. Including
the angle between cracks as further input into the recursive model,
the possibility of identifying the hierarchy of crack segments is
also shown. The approach may be useful in determining the age, authenticity,
and details of old paintings, understanding the stress profile of
geological rocks, and analyzing various natural and manmade hierarchical
structures.