New paradigms for advances in efficiency and stability of CdTe solar cells

Polycrystalline thin film CdTe continues to be a leading material for the development of cost effective and reliable photovoltaics. Although research in CdTe dates back to the 1950s, it is a very innovative and promising technology. Thanks to the continuous development, in 2016 an efficiency of 22.1 % has been achieved, a gain of about 5 percentage point compared to the efficiency of 2012. In these last years many innovations have been made, but yet the theoretical efficiency has not been obtained and there is still a large room for improvement. Moreover many physical and electronic mechanisms in the devices are still not well understood, the understanding of which would lead to further improvement. Thus the research work on CdTe is far from being finished. In our laboratory, superstrate configuration CdTe solar cells, fabricated by low temperature process, have achieved efficiencies exceeding 16 %. However, the continuation of our studies and the optimization of our processes will lead to even better efficiencies. In this thesis I have presented the latest innovations we have introduced in the standard CdTe device: in particular our work and our studies concerning the window layer, the activation treatment, and the back contact. We did not just reproduce the changes introduced in the cells by the world leader First Solar; we looked for and followed different directions. Since it is very difficult to predict what direction will lead to the best results (as achieving the maximum theoretical efficiency and the improvement of the cells stability) it is very important to follow also alternative ways. Study of alternative window/buffer layer The treatment in difluorochloromethane gas of a thin CdS layer, in the order of 30-80 nm, allows to improve the short circuit current of the devices, reducing the open circuit voltage loss. While, generally, cells with thin CdS layers are affected by micro-pinholes resulting in lower open circuit voltage. The insertion of a high resistance transparent HRT layer of Mg-doped ZnO between ZnO or ITO and CdS leads to an improved band alignment with the CdS buffer layer, compared to ITO or ITO/ZnO stacks. In fact, as the band gap of MZO is tunable by changing the amount of Mg in the compound, it is possible to optimize the band alignment. This is probably the cause of the FF improvement up to 74 %, which allows to reach efficiencies up to 16.2%. Because of the tunable band gap of MZO, it is even theoretically possible to tune the MZO band gap and electron affinity in a way to have an optimum match with CdTe, forming MZO/CdTe heterojunction. Thus CdS layer has been totally removed, dramatically increasing the short circuit current. This has been confirmed by the EQE response, with a significant absorption improvement in short-wavelength region. Good efficiencies close to 13 % have been obtained, with Voc above 885 mV and short-circuit current of 27.0 mA/cm2, but low fill factor. This is probably the symptom of the low quality of the MZO/CdTe junction, which could be caused by the presence of a large amount of defects due to the reticular mismatch and possibly by an insufficient doping of MZO. Despite the results obtained so far, regarding the window/buffer layer innovations, a further optimization of the MZO layer and engineering of the MZO/CdTe heterojunction is the simplest way to increase the samples’ efficiency. Study of the alternative MgCl2 activation treatment Solar cells with efficiencies exceeding 14 % have been obtained by replacing the traditional treatment agent CdCl2, with MgCl2. MgCl2 has demonstrated to be a good alternative to the conventional CdCl2, with the advantage to be non-toxic and less expensive. However devices produced with CdCl2 still exhibit a higher efficiency and this could be related to the lower charge density and to a different structure of defects for the MgCl2 treated samples. For this reason, currently CdCl2 activation treatment remains the most effective. Study of alternative back contact The insertion of MoOx at the back contact has led to the fabrication of samples with efficiencies exceeding 13%. However, in order to achieve these efficiencies, the presence of copper at the back contact is still necessary. MoOx deposited by reactive sputtering does not make a good ohmic contact with low substrate temperature CdTe. The insertion of MoOx improves neither the efficiency of our superstrate CdTe solar cells nor their stability. Reducing the thickness of the copper layer in the cells from 2.0 nm (our standard) to 0.1 nm, deposited by vacuum evaporation, the samples average efficiency decreases from 15.6 % to 13.3 %. However the analysis suggest that a 0.1 nm thick Cu layer is enough to dope a 7 µm thick CdTe layer, while the additional copper amount leads to the formation of compensating defects. The fill factor is the main parameter which negatively affects the J-V characteristics of the low Cu samples. It is clearly caused by a marked roll over, symptom of a high back contact barrier. Thus, despite a very thin copper layer succeeds in CdTe doping, this thickness is insufficient to favor the perfect ohmicity of the contact between CdTe and gold. This study leads to interesting conclusions about cells degradation related to copper content. In devices with a large amount of copper (a 1-2 nm thick layer), a higher amount of Cu diffusion from the back contact leads to the formation of a larger amount of compensating defects, as an excessive Cu diffusion can compensate the CuCd- acceptors with Cui donors. While in cells with a 0.1-0.5 nm thick Cu layer, the degradation is mainly due to the decay of the CuCd acceptor into the donor acceptor pair Cui VCd. A CuCl2 wet deposition method for inserting in the CdTe structure a quantity of copper equivalent to a 0.1 nm thick Cu layer has been developed. This process allows reduction of the incorporated copper quantity without any loss in performance compared to a standard Cu contacting route and simple and rapid depositing evenly over the entire area of the sample. It allows the formation of Cu-Cd acceptor defects, reducing the formation of the compensating donors Cui. Moreover, a dramatic improvement in performance stability has been achieved. The introduction of this method in the industrial production could really improve the lifetime of CdTe panels, without having too much efficiency losses.