Y-lids - Tri-valent oxygen

It seems that for oxygen to move to a trivalent position in Si there are two constraints:

1.

Firstly there has to be some compression from at least one side, increasing the strain on a `BC' O_i atom and bringing it closer to a potential third neighbour.

2.

Secondly there has to be a strong quadrupole attraction that binds it tightly to one of its Si neighbours. In this work that has been induced by the next neighbour presence of another highly electronegative element, in these cases always O. This pulls the O atom close to this shared neighbour, but crucially weakens the second O-Si bond in turn. This allows the O sufficient freedom to move out of its divalent site and form a weak third bond.

Thus these results do not show trivalent oxygen to form three uniform bonds; instead it seems to form one strong bond, one intermediate, and a third much weaker bond. This is contrary to previous CNDO/S calculations which show three equivilent bonds [105]. The exception to this is tri-valent oxygen atoms on the C₂ axis of defects, such as the core of the 5O thermal donor and to a lesser extent the Snyder-Stavola structure. These have bond lengths that are more uniform, particularly the 5O species. This is probably because these atoms have two next neighbour O atoms which roughly equalise the Coulombic attraction along both top bonds.