Identification of NL10

  The recent work of Newman and Ammerlaan [200,192] suggests that there are actually at least three different types of STD defect. The first, not yet been identified by them, is the N-based defect of Hara, Suezawa, etc. discussed above. This defect is stable to $\sim$900$^\circ$C [165]. The second contains hydrogen and corresponds to at least one of the NL10 EPR signals. This defect is stable to 520$^\circ$C, and has an electronic IR signal distinct from that of STDs found in N-rich material [201]. Finally there is a defect that contains Al on the core C₂ axis but no hydrogen [192], which again has a distinct electronic IR signal. None of these can be attributed to thermal donors, since TDs break down in anneals at 500$^\circ$C, and can only be H-passivated below 200$^\circ$C[203].

We suggest that the first two STD families correspond to those with an N_iO_2i core, and those with a (C-H)_iO_4i core respectively. This would be consistent with the experimental evidence, as well as the thermal stabilities. N_i-O_2i is a tightly bound defect core and should be highly thermally stable (NNO is known to be stable to $\sim$700$^\circ$C). However (C-H)_iO_4i does not have such tightly bound outer O_i atoms, and once these have diffused away to leave (C-H)_iO_2i, this is no longer a shallow donor. Therefore we would expect this to be stable to lower temperatures. In addition, other (C-H) based defects have similar thermal stabilities; the defect responsible for the `T-line' observed in PL experiments, (C-H)_iC_s, is stable to 600 $^\circ$C [196], consistent with the idea that the C-H bond is a strong one.

Recent work by Newman has shown the H-based and N-based STDs are indeed two different types of defect[201], and after our suggestions they are now re-examining their ENDOR results for carbon related signals. In their work correlating NL10 to STDs, they assigned the H-containing NL10 signals to H-passivated TDs [200]. However in light of the work of Weber et al (suggesting H-passivated TDs are only stable at low temperatures) [203], and the discussion above, we believe this to be an incorrect assignment. They have also shown that high carbon concentrations suppress STD formation [201]. This is similar to the result for nitrogen, and is probably due to several mechanisms: C_s forming complexes with dimers, C_i and C_iH.

Strong confirmation of the assignment of (CH)_iO_4i to NL10(H) has come from recent work by Markevich on the D1, D2 and D3 EPR centres, where D2 and D3 correspond to NL10(H) centres that form in hydrogenated irradiated material at the same time that the absorption signal from C_iO_i disappears. This is discussed below in Section 8.8.

Spaeth examined P-doped samples that were subjected to gamma irradiation [216]. This created OV^- centres, which lowered the fermi level and switched the P from a neutral to a P⁺ state. However the concentration of neutral STD centres remained unaffected (and actually appeared to increase a little). They explained this by suggesting the STD levels must lie relatively deep in the gap, over 65meV below E_c. However an alternative possibility is that the gamma irradiation was creating Si_i atoms, which in turn liberated more C_i (through $\rm C_s + Si_i \rightarrow C_s$) and thus created further STDs, similar to the irradiation production method of Markevich discussed above. Their assignment was complicated since the STD absorption only appears as a shoulder on the larger OV^- absorption.

A previous study by Jones et al [111] proposed an Al-based STD consisting of substitutional Al surrounded by two over-coordinated oxygen atoms, which remains a possible candidate for the third type of NL10 defect. However an alternative can be based around thermal donor models with Al substituting for Si in the core. This is examined further in Section 9.5.