STATISTICS OF BRANCHED FRACTURE SURFACES

A statistical analysis of fracture surfaces of the polycrystalline int ermetallic compound Ni3Al is reported. Although these surfaces contain secondary branches, a roughness exponent xi can be defined, and is fo und close to 0.8. The number of branches is shown to have nontrivial f luctuations, which exhibit a power-law increase with an exponent stron gly dependent upon the dynamics of crack branching during crack propag ation. Moreover, the probability distributions of both heights and ave raged heights are shown to slowly decrease, i.e., like power laws, for high enough altitudes. Dynamical effects could be responsible for the se ''anomalous'' statistics.