Aggregation processes in CVD diamond

The concentration of substitutional nitrogen in sample GC2 does not significantly change during the annealing study until it is annealed at 2200 ◦C. This nitro- gen concentration has an average value of 3600 (400) ppb. After annealing at 2200 ◦C the concentration of Ns is reduced to 2200 (250) ppb, where the rest of

the substitutional nitrogen has presumably gone on to form aggregates. It is pos- sible to compare this rate of aggregation to that in natural and HPHT synthetic samples by considering the aggregation rates presented in Figure 6.1. After an- nealing at temperatures up to and including 1800◦C, for a sample with a starting Ns concentration of 3600 ppb, the model predicts very little aggregation. Anneal-

ing at 2000 ◦C would result in ∼ 10% aggregation, a value which is still within the error bounds of the values calculated by EPR and IR absorption. Annealing at 2200 ◦C, however, results in an aggregation of ∼ 39%, leaving 61% of the sub- stitutional nitrogen in an unaggregated form. The final nitrogen concentration of this sample after an anneal at 2200 ◦C is 2200 (220) ppb while that predicted by 2nd order kinetics is also 2,200 ppb.

The substitutional nitrogen concentration as measured by EPR and IR spec- troscopy remaining after treatment for three more samples is detailed in Table6.5

along with details of the maximum annealing treatments these samples received. It can be seen from this table that the experimentally calculated remaining ni- trogen concentrations, [Ns], matches very closely to those predicted by the rate

constants of Evans and Qi. The measured Ns concentration after treatment is

systematically high for samples JB1 and JE1, however if the starting concentra- tion of these samples was actually 13,000 ppb (i.e. FTIR measurements in the as-grown state have over estimated the Ns concentrations) then the predicted con-

centrations come in line with the measured values for both samples. Assuming that the as-grown concentration of samples JB1 and JC1 should have been 13,000 ppb does not invalidate the results presented in Table 6.4, on the contrary, it would bring the total nitrogen assay concentrations post treatment closer to the as-grown concentrations. This suggests that the rate of nitrogen aggregation in

CVD diamond is similar to that in natural and HPHT synthetic diamond but that the products formed by aggregation are different (i.e. A-centres are not formed). Sample HE1 has not been included in this analysis because the additional irradi- ation and annealing which this sample has received may have significantly altered its aggregation.

It is clear from the results of Figure6.5 and from the details of samples presented in§6.3.2that N2VH0 is regularly produced in nitrogen doped CVD diamond after

annealing at 1800 ◦C. The lowest temperature treatment from which N2VH0 was

still observed was that of sample GG1 after 120 hours of annealing at 1500 ◦C. From these results it seems that N2VH0 is produced at temperatures during which

the NVH centre is annealing out.

Figure 6.5, however, shows that there is a reduction in NVH total concentration after annealing at 1600 ◦C but no corresponding production of N2VH0. Further,

all detectable NVH is removed upon annealing at 1800 ◦C however an increase in the N2VH0 concentration is still observed after annealing at 2000 ◦C. These

details suggest that there is an intermediate step between the loss of NVH and the production of N2VH0. Perhaps this is the energy at which the hydrogen is

able to escape the vacancy (at which point the NV unit would then continue to aggregate as in natural or HPHT synthetic diamond) or the temperature at which the vacancy dissociates from the nitrogen with the hydrogen still trapped inside. Under this process the vacancy hydrogen centre may go on to be captured by A-centres which are produced via the normal vacancy assisted process where vacancies are released from extended defects as discussed earlier.

Figure6.5shows that the N3VH0defect continues to increase in concentration after

annealing at 2200 ◦C. As discussed earlier, there has been no evidence in CVD,

natural or HPHT diamond of the removal of 3107 cm-1 _{upon annealing. This}

suggests that where large hydrogen concentrations are involved in the aggregation process, the N3VH0 defect is the final product of the aggregation chain.

The 1295 cm-1 _{feature is also present in all samples presented in Figure 6.12 but}

Figure6.15 shows that since the exact shape of the absorption feature centred at 1295 cm-1 is unknown it is not possible to deconvolve even 5 ppm of A-centres from the experimental spectrum. If a further study of annealed CVD diamond could deconvolve the overall shape of the broad feature centred at 1295 cm-1 then more information about the formation of A-centres in diamond could be discerned from experimental spectra.

By comparing the untreated single substitutional nitrogen concentration provided in Table 6.2 and the total nitrogen which can be accounted for by quantitative techniques provided in Table 6.4 it is possible to calculate how much nitrogen is unaccounted for after treatment. These results show that, to within error, all substitutional nitrogen can be accounted for by the defects which can already be quantified by EPR, IR and UV-Vis absorption spectroscopy. In natural and HPHT diamonds most of the single substitutional nitrogen goes on to form A- centres, as previously discussed, however in these samples the results show that

1000 1100 1200 1300 1400 1500 0 1 2 3 4 5 Residual Absorption Co efficien t (cm -1 ) Wavenumber (cm-1₎ Experiment A-centre reference

Figure 6.15: IR spectra of sample JB1 after UV -illumination. The original

experimental spectrum is shown in black, a reference spectrum scaled to show an absorption resultant from 5 ppm of A-centres is shown in red and the residual of the experimental spectrum minus the A-centre reference is shown in blue.

this is not the case. Instead the nitrogen goes on to form N2VH0 and then N3VH0

centres. The accuracy with which these results calculate the untreated Ns con-

centrations also suggests that the calibration coefficient proposed for N3VH0 is a

good approximation.