It was recalled that studies of grain boundary structure had long assumed that hypothetical boundaries of infinite extent were adequate representations of real interfacial structures. It was pointed out here that this was clearly sufficient for many interfaces, but that the assumption should still be questioned; especially with regard to nano-crystalline materials. Circumstances were identified and investigated under which so-called infinite-boundary models were likely to lead to misleading predictions. Such cases included those of very short boundaries that were terminated by triple lines, and short boundaries that were terminated by free surfaces. In both cases, the end effects were potentially significant and, in extreme situations, might even be predominant features of the boundary structure.

A.H.King, L.Balasubramanian: Materials Science Forum, 1995, 189-190, 143-8