a twist boundary

A twist boundary is the second clean way to misorient two grains. Take the tilt picture and change one thing: now rotate one crystal about an axis that stands PERPENDICULAR to the boundary plane, like twisting the top half of a stack of papers around a pin driven straight down through the pile. The two crystals meet on the same plane, but one is rotated in that plane relative to the other, as if you gave it a twist.

A twist boundary is defined by a rotation axis normal to (perpendicular to) the boundary plane. For a small twist angle, the misfit is taken up not by edge dislocations but by a crosshatched grid of screw dislocations: two sets of parallel screw dislocations crossing at an angle, forming a mesh in the boundary plane. As with tilt, tighter twist means a finer mesh, and the screw-dislocation spacing shrinks roughly as b/theta. Away from the dislocation lines the atoms match well; along them the twist is concentrated.

Tilt and twist are the two pure building blocks; describing any real low-angle boundary means resolving its misorientation into a tilt part (edge dislocations) and a twist part (screw dislocations). A twist boundary is a lovely direct demonstration that screw dislocations, not just edge ones, live in boundaries. Honest note: because a general grain boundary has a rotation axis pointing in some arbitrary direction, pure twist boundaries are as idealised as pure tilt ones; most real boundaries are mixed.