Type-P defects (mobile phase defects on dimer rows, which had a structure that was similar to that of the type-C defect) were observed on the (100) surface at 80K. However, the observed surface structure was mainly of c(4 x 2)-type; contrary to published data for 6K. The complete p(2 x 2) arrangement was unstable, and type-P defects tended to form pairs with other type-P defects on neighboring dimer rows. This led to a reduction in the area which exhibited a complete p(2 x 2) arrangement. This was thought to be the first observation of interacting phasons. The results were analyzed by using an Ising model, and domain boundaries between c(4 x 2) and p(2 x 2) arrangements were found to play an important role in the dynamics of type-P defects at 80K.

H.Shigekawa, K.Miyake, M.Ishida, K.Hata: Japanese Journal of Applied Physics, 1997, 36[2-3A], L294-7