The 889 cm^{-1} defect also anneals out around 450C,
whereupon LVMs around 910, 976 and 1005 cm^{-1} (10K) are found.
The intensities of these LVMs appear to be correlated. Another band at
986 cm^{-1} also arises but its intensity is not correlated with the
other three. Corbett et al [141] assign the three LVMs to
VO_{3} where O_{i} is trapped near the VO_{2} centre, since at this
temperature O_{i} is now mobile. Thus VO_{2} does not dissociate but
traps O_{i}. Support for this assignment is that the rate of growth
of 976 cm^{-1} line is proportional to the product of the
intensities of the 889 cm^{-1} line and that due to O_{i}
[90]. However, the interpretation here is dependent on
the assignment of the 889 cm^{-1} defect.

Further lines have also been seen and tentatively assigned to VO_{4}
complexes, and a study of higher order VO_{n} defects is currently
underway by Lindström and Hallberg.