Among a variety of solid surfaces, Si(001) and Ge(001) have been most
extensively studied. Although they seem to be rather simple systems, t
here have been many conflicting arguments about the atomic structure o
n these surfaces. We first present experimental evidence indicating th
at the buckled dimer is the basic building block and that the structur
al phase transition between the low-temperature c(4x2) structure and t
he high-temperature (2x1) structure is of the order-disorder type. We
then review recent theoretical work on this phase transition. The real
system is mapped onto a model Ising-spin system and the interaction p
arameters are derived from total-energy calculations for different arr
angements of buckled dimers. The calculated critical temperature agree
s reasonably well with the experimental one. It is pointed out that th
e nature of the phase transition is crucially affected by a small amou
nt of defects on the real surfaces.