It has been shown that domain boundaries may be formed in the monoclinic I-
w phase of Y-Si-Al-O-N or Er-Si-Al-O-N as a consequence of its ordered stru
cture. Three classes of orientational domain boundaries have been identifie
d: the (001) twin, a 60 degrees rotation about [010], and a 60 degrees rota
tion about [010] followed by (001) twinning. High-resolution transmission e
lectron microscopy images are shown of all these classes of orientational d
omain boundaries. It is shown that they arise from two different sources. F
irstly, most twin boundaries seem to have arisen from growth accidents duri
ng crystallization. In contrast with this, the rotational domain boundaries
and disordered boundaries between twin-related domains seem to have arisen
from crystal coalescence during the glass crystallization heat treatment.
Evidence is presented that suggests that adjacent crystals rotate during th
e heat treatment to a similar crystallographic orientation of the hexagonal
Y or Er cation sublattice, followed by coalescence. This could then lead t
o the formation of these domain boundaries if the orientations of the monoc
linic unit cell with respect to the hexagonal sublattice are different in t
he two crystals. Finally, it is suggested that van der Waals forces acting
across the viscous glass may be responsible for the alignment of adjacent c
rystals during heat treatment.