Experimental Background

As well as the experiments described above, there is some work specifically relevant to the (CH)_iO_4i defect.

Recent experimental work by Markevich and Suezawa has identified an EPR active centre that forms in irradiated material containing both H and O [217]. This EPR `D1 centre' is stable in the singly negative charge state, but has an alternative structure which is only marginally stable in the neutral charge state before spontaneously forming a metastable +1 charge state structure. The -1 charge state is a deep level defect, however both the neutral and +1 charge state defects are shallow (identified using DLTS). It is present in concentrations of up to 10¹⁵ cm^-3, and is a negative-U centre with U $\sim$ -0.065 eV [218]. It has a g-tensor remarkably similar to that of NL10(H), and an activation energy for generation of 1.68 eV [219], suggesting it grows by dimer addition (since the dimer migration barrier is approximately 1.7 eV, see Chapter 6). They are produced in hydrogenated Cz-Si crystals after electron irradiation and annealing at 300-400$^\circ$C [220,217]. Above around 400$^\circ$C they transform into further D2 and D3 centres, all of which anneal out at 550-600$^\circ$C. Their formation corresponds with the loss of C_iO_i absorption signal. D1 displays well-resolved hyperfine interactions of the unpaired electron with a ²⁹Si atom which shows isotropic s-type character; this is not consistent with either our calculations or published EPR results on the NL10 centres. The electronic IR signals for the D centres are in some cases very close to those of NL10(H), in other cases identical (for 1s $\rightarrow$2p₀ transitions, the D centres absorb at 249.7, 204.4 and 198.8 cm^-1, compared to 208.7, 204.2 and 198.2 cm^-1 for NL10(H)). In addition the D series donor levels lie at 42.6, 37.0 and 36.3 meV [217], whereas the nearest NL10(H) members lie at 37.8, 37.0 and 36.3 meV [200].

Thus it appears that the D1 centre is a precursor to NL10(H), whereas D2 and D3 correspond to two of the NL10(H) centres. D1 dies away at the same temperature as NL10(H) increases, which adds weight to this suggestion. The isotropic EPR signal of D1 suggests it is quite different from any of the centres modelled here.