Classical grain growth theory assumes boundaries move by curvature alone — a curved boundary straightens itself, large grains consume small ones, and the microstructure coarsens uniformly toward equilibrium. The models are elegant and nearly universal. They are also incomplete. Diffuse-interface simulations incorporating shear coupling during grain boundary migration reveal that the resulting microstructures are measurably more heterogeneous and less equiaxed than curvature-driven models predict.
The mechanism is deceptively simple. When a grain boundary migrates, it does not merely translate — it shears. The lattice on one side shifts laterally relative to the other, introducing displacement that conventional models ignore. This shear generates internal stress. The stress, in turn, feeds back into the migration itself: highly stressed grains shrink faster, while lightly stressed grains grow faster. The microstructure develops an internal economy where mechanical state — not just geometry — determines which grains survive.
As grain growth proceeds, the internal stress field relaxes progressively, but the microstructural record of its influence persists. The grains that survive are not simply the largest or the most favorably curved — they are the ones that happened to accumulate the least stress during their neighbors' migrations. Selection operates on mechanical history, not instantaneous geometry.
The through-claim reaches beyond metallurgy. Any system where boundary motion is coupled to a secondary field — thermal, chemical, mechanical — will exhibit this kind of feedback. The boundary does not just move through the system; it loads the system as it moves. And that loading changes which boundaries survive to move further. Growth becomes a competitive process shaped by its own byproducts, not merely by the initial configuration. The residue of motion determines the future of motion.