It was recalled that B2O3 was an encapsulant, for the melt growth of monocrystals, which was effective in reducing the density of twins and contributed only a small degree of B contamination (1016 to 6 x 1018/cm3). Thus, the structure of thin twin lamella which was typical of vertical Bridgman samples disappeared, but other structural defects (such as twin domains with low-angle grain boundaries) was observed. It was suggested that a so-called type-I structure (thin twin lamellae) resulted from a periodic slip mechanism which spontaneously nucleated the 3C phase at the crystal surface. A so-called type-III structure was expected to result from non-random nucleation of the 3C phase, followed by polygonization during annealing. A type-II structure was expected to be formed under conditions which were intermediate between those that produced type-I and type-III: that is, fewer nucleation sites on the crystal surface, and moderate supercooling.

H.Okada, T.Kawanaka, S.Ohmoto: Journal of Crystal Growth, 1997, 172[3-4], 361-9