Selenization of metallic components with Se vapour

As the Se evaporation and annealing did not result in a uniform depth profile of the Se content of the layers, a selenization in selenium vapour was attempted. For this purpose the 0.5 cm² sized samples were sealed in evacuated glass ampoules with Se pellets and annealed at 500°C for 15 minutes. The number of Se pellets was chosen so that during annealing the Se would be in excess.

Fig. 7.5. shows SEM micrographs of the samples after the selenization.

Fig. 7.5. a: the typical morphology of sample 5 and 6, b: the typical morphology of sample 7 and 8

All the samples became laterally homogeneous in composition. Samples 5 and 6 have a generally similar morphology (see Fig. 7.5.a). The circular („cauliflower-like”) grains usually found in CIGS films cannot be found here, therfore a denser, more uniform surface is resulted which is generally considered to yield more efficient

devices. Samples 7 and 8 have a similar morphology (Fig. 7.5b), except for the hexagonal crystallites scattered all over the surface.

The composition of the layers according to the EDS analysis is shown in Table 7.3.

The composition of the layers corresponds to that of the CIGS material. As mentioned before, CIGS absorbers are not as sensitive to the precise composition, as e.g. the silicon based technologies. A few percents of differences could still result is functioning devices, but certain qualities may differ. The composition of sample 7 is nearest the ideal CIGS composition.

Sample no./component Cu In Ga Se

5 35.7% 15.5% 5.1% 43.7%

6 35.5% 15.7% 2.6% 46.1%

7 26.9% 17.8% 6.6% 48.8%

8 16.5% 29.5% 2.5% 51.5%

Ideal composition 25% 17.5-20% 5-7.5% 50%

Table 7.3. The composition of the sequentially evaporated and post selenized CIGS layers

Fig. 7.6. shows the SNMS results and the cross sectional images of the samples.

All the layers are built up from two sub-layers with different morphologies: a bottom layer with a finer structure (grain size of a few tens of nanometers), and a top layer with a larger grain size in the micron range. This corresponds to a Ga rich sub-layer at the bottom of the film. In the films where the Cu was sputtered first (see sample 7 and 8 in Fig. 7.6. 3^rd and 4^th row) this difference is even more prominent. On the other hand the SNMS depth profiles (see Fig. 7.5 first column) show that the Se diffused into the full depth of all the layers, although their elemental distributions along the depth are not homogeneous. To understand the background of this phenomenon, the XRD results need to be analysed.

Fig. 7.6. The SNMS Depth profiles (1. column) and the cross sections (2. column) of the samples selenized with Se vapour: 1, 2, 3 and 4 in the 1., 2., 3. and 4. rows respectively.

The XRD experiments (see Fig. 7.7.) showed that all the layers contained chalcopyrite phase, some also had one of the metals in excess. A summary of the crystalline quality of the layers can be seen in Table 7.4.

Fig. 7.7. The XRD results of the CIGS layers.

5 Predominantly CuIn0.7Ga0.3Se2 with a little CuInSe2 phase present.

6 Both CuInSe2 and CuIn0.7Ga0.3Se2 phases present, the CuInSe2 phase more dominant than the CIGS.

7 Only one chalcopyrite phase present, that of CuIn0.9Ga0.1Se2 with some hexagonal CuSe.

8 CuInSe2 and CuIn0.7Ga0.3Se2 phases with a little hexagonal CuSe present.

Table 7.4. The XRD results of the layers.

The absence of binary phases and the presence of the CuInGaSe2 phase in the case of samples 5 and 6 indicate that the reactions were completed. The two different layers shown in the cross sectional images may be the CIS and the CIGS phases, therefore a phase separation must have taken place with the CIS at the top and the CIGS dominantly at the bottom of the layers. This was also proven as an elemental analysis along the cross sections of the layers showed higher Ga concentration toward the bottom of the layer.

In samples 7 and 8 the CuSe is still present. Therefore the hexagonal phase CuSe is the crystallites seen in the SEM micrographs 7.5.c and d. This statement was also

shown by EDS analysis. As in the case of these samples the copper was sputtered first, this was the bottom layer in the layer structure, which is probably why some CuSe was present in the samples. This layer structure is either less favourable, or it requires a longer annealing time. Sample 8 also showed a separation of the CIGS and CIS phases. It can be seen from the XRD results that in the cases where there is less Ga in the layer than In the phase separation is more pronounced.

XPS measurements were also performed on sample 8. The binding energies were evaluated using the NIST XPS binding energy database, and are summarized in Table 5.

Se 3d In 4d 5/2 Cu 2p 3/2

Sample 8 54.4 eV 444.5 eV 932.5 eV

Reference CIGS sample (own

measurement) 54.45 eV 444.6 eV -

Reference [7.1.] CuInSe2 54.3-54.5 eV 444.6-444.8 eV 932.1-932.6 eV

Cu metal (own measurements) 932.35 eV

In metal 443.9 eV

Se 55.2 eV

Table 7.5. XPS Binding energies of Se, In and Cu measured on sample 8 and on reference samples

The accuracy of our own measurements is in the range of ± 0.1eV (similar to the accuracy of ref.[7.1]). The XPS binding energies are in good agreement with the XRD data and show the presence of CuInSe2 and CuIn0,7Ga0.3Se2 phases in sample 8.

As a conclusion this selenization method resulted in more homogeneous depth profiles of the layers, and CIGS layers were formed. On the other hand, even the best layers produced by this method contain additional binary or ternary phases.