A mechanism for the origin of screw dislocation sequences, giant screw dislocations, and polytypism in platelet crystals

Abstract

Several workers have observed growth spirals of very large step heights on platelet crystals, sometimes associated with polytypism. These are attributed to spiral growth round a sequence of similar screw dislocations in a close planar array or a single giant screw dislocation. Polytypism can be (but is not necessarily) caused by such giant growth steps. The origin of the dislocations giving rise to them has remained obscure. It is proposed that they are due to radial impurity gradients which cause the lattice parameter to change gradually away from the centre. The hoop stresses which thus arise may be relieved by elastic buckling, twinning, glide or brittle fracture. Since the hoop stresses produce no shear on either the basal plane or the prismatic planes of a crystal growing parallel to the basal plane, as long as the crystal remains planar, the hoop stresses may build up until the platelet buckles, cracks or twins, if those are the only active slip planes in the material. In all cases the dislocations which are formed in glide will be near the screw orientation and slip will be concentrated on or close to those planes which first began to glide. Therefore, if the slip has a component along the relevant axis that is comparable to or larger than the platelet thickness, it will lead to fracture. If the glide has relieved the hoop stresses before that point, it will cease. In the first case, the two edges of the crack will continue to move forward by independent crystal growth and are liable to give rise to growth steps of height equal to the platelet thickness at the point of fracture. Otherwise, the dislocations will crowd together near the head of the slip line which is forming and will also give rise to a giant, growth step centred, however, not on a single giant dislocation but on a planar array of screw dislocations. A simple theory shows that the conditions for the proposed mechanisms must be prevalent. Experimental observations are gcnerally in support of the theory.