On the deduction of silicon-carbide polytypes from screw dislocations

Abstract

Mitchell has deduced the polytypic structures that would arise from theoretical screw dislocations in the 6H, 4H and 15R phases of silicon carbide, by the spiral-growth mechanism of Frank, and compared these with the structures actually observed. In doing so, he assumed that at the instant of the creation of a screw dislocation, the initial platelet consists of a completed number of unit cells and that the exposed ledge, irrespective of its own structure, can wind over itself in a close-packed manner, thereby repeating its own zig-zag sequence indefinitely during the subsequent spiral growth. It is shown that both these assumptions are not justified and that the use of Zhdanov symbols alone leads to incorrect results. The possible polytypic structures that can be derived from theoretical screw dislocations of different Burgers vectors are rededuced with corrected assumptions, by considering the different possible structures of the exposed ledge in terms of ABC layers and investigating the stacking of these layers during the subsequent spiral growth of the ledge. The results so obtained differ markedly from those of Mitchell. In particular a much larger number of structure series are shown to be actually possible. The results are compared with the polytypic structures reported so far.