First, the mechanism of NW growth on polycrystalline CdTe films is discussed in order to explain the differences relative to growth directly on Mo. Second, the non-vertical and apparently random orientation of NWs with respect to the substrate is discussed with reference to the XRD data. Finally, in light of the results, the advantages of metamorphic growth, with reference to the development of NW solar cells, are evaluated.
5.4.4.1: Growth mechanism: NW growth on CdTe/Mo, compared to directly on Mo
99 growth proceeded without delay (< 5s compared to > 5 mins). This behaviour may be
attributed to the more ready formation of liquid droplets. Regardless of whether Au-Cd or Au- Te is responsible for NW growth, the liquid phase is more readily achieved for Au droplets that lie on CdTe films than for Au droplets on Mo, since the former can absorb semiconductor material from both the vapour and the underlying film, whereas the latter may only absorb material from the vapour (in both cases, ‘material from the vapour’ includes atoms arriving directly at the droplet and adatoms that have been deposited onto the substrate and
subsequently diffused to the droplet).’
This also accounts for the increased NW density: Solid droplets may be buried by CdTe film growth and do not nucleate NWs, therefore the more ready acquisition of liquid droplets should result in a lower proportion of them being buried. However, Fig 5.18b shows that even for metamorphic growth, some droplet burial occurs, as there is still some planar CdTe
growth. The NW density achieved (107 – 108 cm-2) therefore falls short of the density of catalyst droplets (108 – 109cm-2) achieved by annealing 5 nm Au films at 360°C – due to droplet burial and possibly due to continued droplet coalescence.
5.4.4.2 NW orientation and lattice parameter: The results presented in Subsection 5.4.3
are now discussed in terms of explaining the apparent random orientation of NWs, which was observed for all cases of NW growth presented in this Chapter.
The reduced degree of preferred orientation - determined from XRD spectra (Fig 5.19 and Table 5.1) - observed for NW arrays, relative to the substrates upon which they were grown, implies either that: a) the NW growth on the substrate is not epitaxial; or b) the underlying substrate of the NW array sample has undergone recrystallisation prior to NW growth, and has templated epitaxial growth of NWs in a variety of crystallographic directions. Firstly, the postulation that growth is non-epitaxial is discussed. If the CdTe substrate remained (111) oriented, and NWs grew epitaxially, then NWs could legitimately still have grown in non <111> directions; a [112] wire would be inclined to the substrate normal by 33°66’ for example. However, as the θ-2θ arrangement only detects reflections from planes that are parallel to the surface, then the reflection detected from a [112] NW would still be from its (111) planes that remain parallel to the substrate surface. The resultant XRD spectra in this case would match that of the substrate, i.e. be highly oriented, which it does not, hence the implication that growth is non-epitaxial.
Secondly, the suggestion that the sputtered CdTe film undergoes recrystallisation prior to NW growth is discussed. As the texture coefficient of the (111) plane, C111, and the degree of
100 minute) were comparable to the as-grown film (grown at 200°C), it can be assumed that recrystallisation had not taken place at the point immediately prior to CdTe sublimation. However, it may have occurred during CSS deposition, with the substrate held at 520°C for 20 minutes; although this would only affect any NWs that nucleate later within the deposition period. It is important to note that X-rays could in principle sample both the NWs and the underlying film however. In contrast to these results, Moutinho et al. report recrystallisation of RF sputtered CdTe upon annealing at 400°C for 30 minutes – inferred from AFM and XRD data49. However, Moutinho’s films were grown at a lower temperature than these, and
annealing was carried out in air rather than nitrogen. Moreover, Moutinho’s annealing included CdCl2; Cl being known to accelerate microstructural changes in CdTe. Indeed,
without the use of CdCl2-activation, higher annealing temperatures (500°C) are typically
required to induce recrystallisation, as shown by Paudel et al.50.
Evidently, these XRD results cannot unambiguously distinguish between the two postulates. Nevertheless, the presence of voids in the sputtered layer of an equivalent NW sample, shown in Fig 5.18a, could be indicative of recrystallisation. Moreover, the lattice parameter, calculated from (111) peak positions, is smaller for the NW sample, and more comparable to the bulk value, than for the two film samples. Depending on whether the main contribution to the XRD data was from the NWs or the film, this may imply either: a)
recrystallisation of the underlying film (only after high temperature sublimation) or merely that the film undergoes strain relaxation; or b) the NWs have a smaller lattice parameter compared to the films. Indeed the lattice parameter of the films is higher than the accepted bulk value, which indicates they are under compressive stress, whereas NWs possess additional degrees of freedom compared to thin films, due to a high surface to volume ratio, which enables them to relieve stress more effectively, as described in Section 3.3.3.
Kinking of the NWs, as shown in Fig 5.20, also makes a minor contribution to the
distribution of observed orientations. The example shown has an external angle of 111°, which is comparable to the angle of 109° expected for a Ʃ = 3 twin involving the [111] and [11¯ 1] directions (where Ʃ is the ‘Friedel index’, defined as the ratio of the number of lattice sites in either of two adjacent grains to the number of coincidence sites for the two interpenetrating lattices). In some cases, NWs kinked so that they would propagate back down towards the substrate, which could be explained by a change from a [111] to a [ 1¯ 1¯ 1] growth direction.
5.4.4.3 Advantages of metamorphic growth for the design of NW solar cells. There are
numerous advantages of using the metamorphic approach of growing CdTe NWs for PV devices. Firstly, it results in the formation of a continuous absorber layer consisting of a thin
101 film and NWs, which provides electrical continuity and prevents short circuiting in the device. Whilst depositing directly on Mo also forms a film, it is discontinuous and there is poor thickness control. Secondly, the metamorphic approach yields an immediate nucleation of NWs (i.e. more reliable control over NW length) and an enhancement in the NW density.
This approach can be replicated on other substrates other than Mo, as demonstrated on CdTe/glass and CdTe/ITO/glass (Fig 5.18a). As the NWs nucleate from the CdTe film, the only requirement for the substrate is that CdTe films can be grown uniformly on them and that they adhere to them. This is extremely significant with regards device fabrication as it enables the selection of a variety of conductive substrates, and subsequent optimisation of the back contact of a device, without the need to re-optimise NW growth conditions. Even highly complex thin film stacks, designed to create an Ohmic back contact, or electron back reflectors, may be integrated between the substrate and the CdTe film, provided they do not subsequently degrade at the NW growth temperatures (500 – 550°C).
However, NW density is still not fully optimised with this approach (i.e. not all Au nanodots catalyse NW growth), and additional VS planar growth still occurs. Nevertheless, the metamorphic approach was used for NW generation in the remainder of this work: a) growth of core-shell heterostructures (Section 5.5); b) NW and core-shell characterisation (Chapter 6); and c) fabrication of NW solar cells (Section 7.5).