In summary, it appears that the (CH)i-O2i defect forms with the oxygens in a pair on one side of the (CH)i. This defect is not a shallow donor defect but has a partially filled level deep in the gap. It is unclear whether it would be naturally neutral or -1 charged since this depends on the respective locations of the defect level and the fermi level, and our Kohn-Sham eigenvalues are not quantitatively accurate enough for such a prediction.
When this defect attracts further oxygen atoms to form (CH)i-O4i this acts as a single shallow donor. This may happen through the trapping of a fast diffusing oxygen dimer. The donor level is effective mass like, AIMPRO predicts it to sit very close to the conduction band. However, on trapping of another electron, filling the core Si p-type orbital and making the defect -1 charged, it is able to undergo a structural change by rotating two of the oxygen atoms on one side of the defect core so that they bow out of their bond centred sites away from the defect core (in effect, one side of the wonderbra defect is `popping out'). This allows the p- lobe of the Si orbital to redistribute the majority of its charge away from the remaining O atom in the core, and the defect level drops to mid-gap. This is indicative of negative-U behaviour, but within the accuracy of our code we cannot determine whether the defect is negative-U based on our cluster energies.
Photo-luminescence work has observed a number of lines such as the I- and M-lines in Si, which are believed to be C-H-O defects of some kind . It is possible that these are associated with (CH)iOn defects such as these, possibly either (CH)iOi or (CH)iO2i (or (CH)iCOn). Further work would be required to correctly assign these, however. The I-line forms after 30 minutes annealing at 600 C, and at this temperature dimers are unstable. Its symmetry was found with uniaxial stress to be monoclinic-I . This would exclude (CH)iO2i but could still be (CH)iOi since (CH)i can also sit in a bond centre with an sp2 coordinated C atom, in which case it could form a tight pair with the Oi similar to the dimer, and all the defect atoms would lie in the same 110 plane. If these PL-lines could be assigned to a (CH)iOni defect this would be important, as it would show proof of pre-cursors to the (CH)iO4i STD complex.