Distributed-activation kinetics of

Martensitic nucleation under normal circumstances is a heterogeneous process, and the heterogeneity can be explained by the potency distribution of defects that provide the nucleation sites, including both preexisting and autocatalytically generated defects. The potency distribu- tion of the preexisting defects has been shown earlier by small-particle studies to follow an exponential function, whereas that of the autocatalytic defects is found in the present work to obey a Gaussian function. A major distinction between these two distributions is that the number density of the preexisting defects increases monotonically with decreasing defect potency, while the number density of the autocatalytic defects is distributed about a mode, giving a saturation level of its cumulative distribution. As a result of these potency distributions, the nucleating sites of both origins exhibit distributions in their activation energies for nucleation, and a distributed- activation kinetic model is now proposed to take these variations into account for martensitic transformations. The features of this model are tested with considerable success in experiments on an Fe-32.3Ni alloy, leading to calculations of the entire course of martensitic transformation curves (including the athermal, anisothermal, and isothermal contributions), the changing shapes of time-temperature-transformation (TTT) diagrams for a range of Fe-Ni compositions, and the grain-size dependence of "bursting" behavior.