Combinatorial PED

The unintentional Ga grading is still under investigation. At the same time, intentionally graded CIGS sample are on the way with a double meaning: sharpening the Raman analysis to adapt the model to LTPED-CIGS and understand the feasibility of an intentional double grading of CIGS absorber by means of the LTPED process.

and CuGaSe2. Both the PED sources worked at 16 kV and the substrate temperature was 270 °C. The set-up of this experiment is schematized in Figure 5.24.

Figure 5.24 – Scheme of set-up for CIGS deposition with two PED sources. Here, one CIS and one CGS targets are used.

With this combinatorial deposition, Cu(In1-xGax)Se2 was deposited in 1.6μm-thick film, with x varying between <0.1 on one side of the sample and >0.9 on the other side. The system geometry allowed thickness uniformity distribution of CIGS as shown in Figure 5.25: variation along the sample length was about 2% in the centre and up to 7% at the substrate edges. Thus, the edges were not considered in the solar cells study. Devices with comparable characteristics were investigated from a ~5cm-long central part of the sample.

Figure 5.25 – Simulated thickness distribution by Comsa-David model for samples realized with the combinatorial PED set-up. The film thickness was 1.6 μm over the two targets, had a small decrease (2%) between them and dropped at the edges

(7%).

First of all, compositional and structural features of the whole length-graded CIGS sample were studied in 35 points and are reported in Figure 5.26. The structural properties of the samples were characterized by X-ray diffraction (XRD), performed with a Siemens D500 (Siemens, Berlin, Germany) system in Bragg–Brentano geometry. Morphological and compositional analyses of the samples were conducted by scanning electron microscope (SEM, model Philips 515, Eindhoven, The Netherlands) operating at 25 kV and equipped with an Energy Dispersive X-Ray Spectrometer (EDS) “EDAX” detector.

Compositional analyses by SEM-EDS, made along the whole sample length, indicated the measured data (blue diamonds) had exactly the same trend as the values calculated via the bi-cosine formula of Comsa-David model (green triangles), for the ablation plume distribution applied to a PED system based on 2 guns and 2 different compound as targets. GGI was even calculated from lattice parameters derived by XRD analyses: the c and a axes (red squares and orange crosses, respectively) and the lattice volume (blue crosses). GGI from structural investigation exhibited a very similar trend as SEM-EDS and simulated data, but a little bit shifted towards higher values. Moreover, XRD-GGI were more scattered especially around GGI of 0.6. Some structural modifications, maybe due to strain effects, might be occurred.

Figure 5.26 – GGI data as measured by SEM-EDS (blue diamonds), simulated by Comsa-David model for two PED sources in the experiment geometry (green triangles) and derived by XRD measurements (orange crosses for a axis, red squares for c

axis and blue crosses for lattice volume).

Later, solar cells were realized in the classic solar cell architecture of Figure 5.14 depositing the other layers. CdS was applied in one CBD run on the whole sample, with a thickness of 120 nm. Later, the

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0 10 20 30 40 50 60 70

GGI

Position (mm)

EDS -measured XRD - c axis 2-Plume Model XRD - a axis XRD - Latt volume

sample was cut in 3 pieces and the solar cells were completed in three different runs depositing 120 nm of i-ZnO and 800 nm of AZO by RF-Magnetron sputtering, in off-axis configuration. The samples were then prepared for electrical characterizations completing the structure with 2m-thick aluminium contacts by thermal evaporation and scribing them into 20 devices, with an area of 0.21 cm² each. These 20 solar cells represented CIGS with the GGI from 0.18 to 0.85, as determined by SEM-EDS and Comsa-David simulation. Then, the effect of Ga content on the solar cells behaviour was investigated, studying the parameters of devices with different GGI, all grown in one single deposition run.

Solar cells performance was tested by current–density vs. voltage with a Keithley 2635 system under 1 sun illumination, supplied by a ABET SUN 2000 solar simulator at standard test conditions (AM1.5g at 25 °C).

JV curves showed low values of Jsc, Voc and especially FF. The structure was not optimized since thicker CdS and i-ZnO layers were deposited, to avoid any shunts paths and study the sample in its full length.

Moreover, before CdS growth, the sample underwent compositional and structural analyses during which CIGS surface may be deteriorated. In fact, the aim of this study was not to fabricate high-efficiency solar cells, but to obtain comparable devices grown in the same CIGS run and to find the GGI at which solar cells parameters, like current density and voltage, are maximized. Jsc and Voc trends as a function of GGI are shown in Figure 5.27: Jsc (blue squares) has maximum at GGI≈0.3, while Voc (red dots) is maximized at GGI≈0.57.

Figure 5.27 - Experimental values of Jsc (blue squares) and Voc (red dots) as a function of the GGI, the dotted lines represent their trends.

Since the Ga content determines the band-gap width of CIGS, even if not linearly, the Voc would be maximized for CGS material, i.e. at GGI=1. However, the high amount of Ga brings to stoichiometry

0 100 200 300 400 500 600 700

0 5 10 15 20 25 30 35

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Voc (mV)

Jsc (mA/cm2)

GGI

Jsc Voc

defect density. Interface recombination becomes the dominant recombination mechanism in CGS devices due to increased defects density at the interface, increased tunnelling and negative CBO in the CdS/CGS pn-junction (74). A further Voc drop for large In content (i.e. towards CIS) is indeed caused by the enhanced SRH recombination due to the lower EG.

As said, GGI affects both Voc and Jsc, therefore the most important parameter to be taken in account is the product between Jsc and Voc, called power density. Power density as a function of GGI was calculated and its normalized values are plotted in Figure 5.28. From the trend, power-density behaviour could be foreseen and optimal Ga concentration determined. The maximum value of power density was expected for a Ga concentration of 37.5% in the CIGS target, corresponding to Cu(In62.5Ga37.5)Se2 absorber.

Figure 5.28 - Normalized power density as a function of GGI on the sample. The dots are experimental points while the dotted curve corresponds to their fitting.

This result was an important step for the optimization of solar cells based on CIGS fabricated by means of the LTPED process developed at IMEM-CNR of Parma. Target with composition of Cu(In62.5Ga37.5)Se2 were synthetized and CIGS with GGI=37.5% is currently studied to boost the solar cells efficiency over the present record.

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Normalized power density (Voc x Jsc)

GGI