At times, grain growth proceeds in a "normal" manner—characterized by a parabolic increase in the average grain size and self-similar evolution of the size distribution—but, in other instances, the microstructure of a polycrystalline material changes "abnormally," with a few crystallites growing much faster than the rest, resulting in rapid emergence of a bimodal size distribution. In many cases, it is still a mystery why a particular sample or set of annealing conditions leads to abnormal grain growth, but even the seemingly well-behaved normal growth mode harbors secrets that have yet to succumb to powerful experimental probes and computational algorithms. For example, the grain size distributions found in real materials tend to be more asymmetric than those predicted by analytic models under the assumption of self-similar scaling, and computer simulations of normal grain growth are able to reproduce measured changes in 3D microstructure only over very short annealing times.
In light of these shortcomings, which appear to occur whenever grain growth is modeled in the "forward direction" (i.e., starting from boundary mobilities and driving forces derived from physical principles), we decided to try out the opposite approach: that is, to "reverse engineer" the phenomenon, much like an industrial spy might dissect the interplay between working parts in a competitor's machine or microchip! Using 3D x-ray diffraction (3DXRD) microscopy, we have investigated thermally induced grain growth in two different Al alloys. Through a sequence of microstructural snapshots taken between isothermal annealing steps, we followed the morphology, misorientation and migration of thousands of GBs in a single sample, forming the basis for a robust statistical analysis of local growth kinetics. The results allow us to extract dependencies of reduced mobility (the product of GB mobility and energy) on GB misorientation and inclination. In one particular specimen, this dependency is consistent with expectations for normal grain growth, but, in another case, we find evidence for abnormal kinetics that lie beyond the scope of standard models.