Content area
Full Text
PERSPECTIVES: POLYCRYSTALLINE MATERIALS
In coarse-grained metals, plastic deformation is mainly carried by dislocations-line defects of the regular crystal lattice--within the individual grains. Dislocations can move through the crystal grains and can interact with each other (1). Grain boundaries often hinder their transmission, creating a dislocation pile-up at the boundary and thereby making the material harder to deform.
One of the best known theories based on dislocation pile-up, described by the Hall-Petch equation (2, 3), predicts the hardness of the material to be inversely proportional to the square root of the grain size. However, as grain sizes are reduced to the nanometer scale and the percentage of grain boundary atoms increases correspondingly, this traditional view of dislocation-driven plasticity in polycrystalline metals needs to be reconsidered. In a sample with grain diameters of 20 nm, 10% of atoms are located at grain boundaries. Dislocation sources and pile-- up are hardly expected to exist in such a material and deformation is believed to be carried mostly by the grain boundaries via a particular accommodation mechanism.
Experimental measurements have shown various deviations from the Hall-Petch equation as grain sizes reach the nanometer scale. Some may be attributed to synthesis or measurement artifacts; others indicate intrinsic properties of the nanostructured materials (4, 5). However, atomic level understanding of the grain boundary accommodation mechanism is limited. No direct experimental visualization technique is available that allows nonintrusive investigation of grain boundary structures during deformarion. Transmission electron microscopy requires samples to be thinned to a thickness comparable to the grain size, inducing structural relaxations and thus changing the grain boundary structure (6).
Massively parallel supercomputers offer the possibility to shed light on the deformation mechanism in atomistic simulations involving millions of atoms, equivalent to computer samples of three-dimensional networks of up to 15 grains with a 20 nm diameter or 100 grains with a 10 run diameter (see the...