This question comes from observation that there are no known half-lifes in range from 10−10 seconds to 10−21 seconds, (Except 8Be, which has a half-life of 7×10−18 seconds.)
Excuse me, but a short look at the chart of the nuclides will show you a some more cases with half lifes in the cited region. Examples are Ne-15, Ne-16.
But you are right, the examples are rare. I think there is a problem to distiguish nuclear reaction times with "real" half lifes of the observed process. Compound reactions need times in the range of 10-16 s, direct reactions correlate to the particle transfer times through the nucleus (about 10-22 s).
In principle, there is no sense in talking about a "nucleus" as such, if the system has not enough time to equilibrate its internal temperature.
The very short "half lives" in the order of 10e-21 in the chart of nuclides do not correspond to measured lifetimes of certain nucleon ensembles but are calculated from line widths of resonances (lifetime/linewidth correlation as a result of the uncertainty principle).