Nature of the Effect of the Type E Process

The AES results provide some insight into how the surface of the diamond may be modified by the application of the type E process. The carbon peak arises from a KVV Auger process whilst the three maxima which arise in the valence band structure of diamond may give rise to a number of Auger processes with slightly differing energies. Peaks arising from from the two strongest, K V iV i and K V2V2 transitions are

Chapter 8: The Origin o f the Improvements in the Performance o f CVD D iam ond Photodetectors

com m only observed along with surface and bulk plasmon losses from the Auger electrons of the primary (K V iV i) transition [8.10, 8.11]. The positions o f Aq, A% and A2, the three highest energy o f these peaks observed by others are marked on figure (8.6). Close agreement is evident between these lines and the N(E) peak positions that can be predicted from the differential spectrum plotted for curves (i) and (ii). The CVD diamond film surfaces represented by these traces thus comprise good quality diamond over the probe depth o f the AES technique (3-4 atomic layers [8.12]) and within the sensitivity o f this technique («1% o f a monolayer [8.12]). Spectrum (iii) in figure (8.6) shows the primary peak minima to be shifted to 271eV. This shift and the change in the nature o f the fine structure on the low energy side o f the peak is consistent with the presence o f a graphitic surface [8.10]. In addition, a new peak with a minimum at an energy o f «290eV is apparent. This feature has not been observed during studies on well defined graphite surfaces although Hoffman [8.13] reported a weak peak at 290eV for a type Ua natural diamond crystal whilst noting that a similar peak was not observed in AES spectra from CVD diamond film s; no assignm ent was offered for this observation, the wide energy scan spectra o f figure (8.5) indicate that the samples that had just been cleaned and those that had undergone both steps in the the type E process support a persistent surface oxide phase. The intensity o f the peak suggests that this involves a sub-monolayer quantity o f oxygen in both cases, but more oxygen is present on the treated surface than the cleaned one.

The oxygen peak which is evident on all 'as inserted' samples arises from simple air contamination and is readily removed by heating in vacuo, however the the type E process leaves the surface more heavily oxidised. The graphitic nature o f the type R surface after methane treatment alone suggests that it is this oxidation that removes the graphitic phase during the second step o f the type E process. This hypothesis is supported by the visual observation that a type R sample looks darker than an untreated sample, but that a fully processed type E sample is restored to a brighter appearance. A photograph o f this effect is reproduced as figure (8.8), however the subtlety o f the change, the subjectivity o f its interpretation and the limitations o f the printing process com bine to prevent this image being offered as formal evidence - it is sim ply a supporting observation.

At the visible wavelengths explored in this study photoluminescence will emerge from the bulk o f the diamond film, it can therefore be concluded that the strongly modified PL spectra recorded following type E treatment indicates that the processing acts upon more than just the outermost few atom layers o f the material. The peak at 1.68eV has been observed in PL recorded for a number o f types o f thin film diamond. Fong and

Chapter 8: The Origin o f the Improvements in the Performance o f CVD D iam ond Photodetectors

Schwartz [8.3] found the peak to be strongly m odified by the methane concentration used to grow film s by MPACVD: increased methane concentrations, which are normally asssociated with decreasing diamond quality, led to a strongly reduced PL intensity at 1.68eV. These authors suggest that residual stress quenches the PL emission intensity but does not effect the peak position or half width. Vibronic bands associated with this peak were primarily due to lattice phonons. It was broadly concluded that this result agreed with earlier work which suggested that the peak was associated with Si impurities.

A study by Kania and Oelhafen [8.8] using M PACVD diamond films showed that the 1.68eV peak was stronger in randomly textured films than in (100) aligned ones, whilst Brown and Rand [8.7] have suggested that the defect responsible for this em ission is (111) orientated. Bergman and co-workers [8.4, 8.6] found that isolated diamond nuclei grown by CVD emitted this PL band which was again attributed to Si defects. The strong 1.68eV peak evident in figure (8.7) is therefore to be expected even in high quality CVD diamond. Although recorded at room temperature, it is possible to identify the vibronic structure on the low energy side o f the peak. The peak intensity is significantly reduced by the type B treatment (plot (ii)) but the peak position is unchanged: this implies that there is an increase in the stress level within the film.

Bergman et al. [8.4, 8.6] also studied broad band PL em ission from CVD diamond film s over a similar energy range to that presented in figure (8.7). The authors concluded that that amorphous material within the film , containing sp^ bonding was responsible for the broad peak; it was proposed that a wide distribution o f continuous in-gap states would result from the random nature o f this form o f defect. PL measurements made on amorphous hydrogenated carbon fim s (DLC) have shown similar broad band characteristics over this energy range [8.14]. The fact that the type E treatment strongly modifies the intensity and position o f this broad band implies that the nature o f the amorphous material within the film is being changed.

In summary it can therefore be identified from the AES experiments that the type E process acts by donating carbon or a carbon containing species to the surface o f the film whilst the subsequent oxidation step removes surface graphitic material. This model is supported by the earlier observation (§6.4.1) that the methane processed surface is highly conductive but that after oxidation the resistance is dramatically increased. The PL analysis suggests that increased stress and m odified non-diamond carbon regions within the film may result from the treatment, and that the effects o f the process must extend beyond the exposed and near surface regions o f the film. As diffusion within

C h apter 8: The Origin o f the Improvements in the Performance o f CVD Diamond Photodetectors

diamond grains is highly improbable at the temperatures involved, it would appear reasonable to anticipate a model involving transport of a carbon containing species along grain boundaries to achieve passivation of electro-optically active defects which are local to the grain boundaries.

Figure 8.8: Photographic 'observation' of the change in darkness exhibited by the CVD diamond surface at each stage of type E processing. Left to right: untreated, methane only, methane and oxidation.