Search

We have made direct observations of the Brownian motion of individual nanosized liquid lead inclusions in solid aluminum by in-situ transmission electron microscopy. The process was found to depend strongly on the size of the inclusion and the anisotropy of the interfacial energy. The rate controlling mechanism was the nucleation of steps on facets of the equilibrium shape and the diffusion of Al along the liquid-solid interface. Because the Al-Pb interface undergoes a roughening transition at T-r approximate to 820 K, the step nucleation barrier decreases with increasing temperature and vanishes completely at Tr. By including the temperature dependence explicitly, we demonstrate that the contribution of the step energy to an Arrhenius plot of the data has a slope greater than the actual activation energy at temperature T by a factor T-r/(T-r-T). In addition, we show that the diffusion barrier for interfacial transport makes a substantial contribution. Our analysis reconciles a large discrepancy between activation energies obtained from the temperature and size dependences of the process and solves the apparent paradox posed by the observation that some particles are frozen in non-equilibrium shapes while nearby smaller particles are sufficiently mobile to undergo rapid Brownian motion.

This work investigates the mechanism and dynamics of grain boundary migration driven by capillary forces via in situ electron microscopy, complemented by molecular-dynamics simulations. Using thin films of Au with the mazed bicrystal geometry, the shrinkage of island grains with 90 degrees lt 110 gt tilt grain boundaries was observed by diffraction contrast and high-resolution imaging. The grains remained cylindrical throughout the shrinkage, and there was no measurable grain rotation even at very small sizes. The rate of shrinkage was found to be erratic and inconsistent with parabolic kinetics, accelerating before complete disappearance. Residual defects were found immediately after complete shrinkage, although the type and magnitude of the defects varied from grain to grain. Measurement of the grain boundary shape anisotropy showed a preference for facets on low-index planes of the crystals, including the mirror-symmetry planes of the bicrystal. These facets were also found directly on individual images extracted from high-resolution video recordings of shrinking grains at similar to 300 degrees C. The dynamics of boundary motion were found to be limited by nucleation and propagation of steps on these facets. The cylindrical geometry and size of the experimentally observed island grains allow direct comparison with molecular-dynamics simulations on the same length scale, which reproduced many of the experimentally observed features, including non-parabolic shrinkage, absence of systematic grain rotation, step-controlled migration and dislocation debris after complete grain shrinkage. Differences between model and experiment are discussed in terms of the possible role of impurities, surfaces and interfacial steps.