Open Access Version

Abstract:
In order to transfer the very good laboratory results for Cu(In,Ga)(S,Se)2 (CIGSe) solar cells with efficiencies above 20% to a mass producing industry a cost effective and efficient technology for a large area deposition of the CIGSe absorber film is needed. In this thesis a reactive magnetron sputtering (RMS) process is investigated for the deposition of these absorber films. By simultaneous sputtering from CuGa and In targets in an Ar:H2S or Ar:H2Se atmosphere onto heated substrates a direct growth of the Cu(In,Ga)S2 repectively Cu(In,Ga)Se2 thin films is achieved. This approach combines the advantages of the two conventional preparation technologies for CIGSe films. The well-established large area deposition method of sputtering is used and at the same time a direct and well controllable growth of the chalcopyrite films can be realized. Although the principle applicability of the RMS technology for the deposition of chalcopyrite films has already been shown, a more detailed process understanding is necessary to enable reproducible depositions. A further aim of this work was, to evaluate, if the relatively unique approach of sputtering from metallic targets in an Ar:H2Se atmosphere is suitable for the deposition of Cu(In,Ga)Se2 thin films, which can achieve a sufficient material quality for high solar cell efficiencies. The behavior of the magnetron sputtering process in dependence of the reactive gas content has been investigated and mainly analyzed by the discharge voltages of the two targets, while sputtering at constant target power. The deposited films show a pure chalcopyrite phase if the reactive gas content during the process was sufficiently high. For higher substrate temperatures and Cu richer growth conditions the expected trends of increased crystallization and better grain growth have been found. However, for Cu-poor deposition conditions the film composition is significantly influenced by the re-evaporation of indium if the substrate temperature Tsub exceeds a critical value Tcrit in the range of 300…450°C. For Tsub = Tcrit + 100°C nearly all excess In is re-evaporating and the film composition is self-adjusted to a [Cu]/([In]+[Ga]) ratio close to 1 (stoichiometric composition). For Cu(In,Ga)S2 absorber layers an optimization of the deposition sequence was very difficult due to reproducibility problems. The best solar cell achieved an efficiency of 10.6% with an open circuit voltage of 876 mV for a bandgap of 1.72 eV. Cu(In,Ga)Se2 absorber layers for solar cells could be reproducibly deposited by using a 1 stage deposition sequence with Cu poor supply conditions during the whole process. Due to significant re evaporation of indium for the used substrate temperatures of about 500°C the composition is self-adjusted to a [Cu]/([In]+[Ga]) ratio close to 1 and good solar cells can be prepared with a relatively wide range of supply ratios from the CuGa and In target. The best efficiencies are 9.3% on normal Mo back contacts and 11.7% on sodium doped Mo back contacts (Mo:Na). The higher efficiency on Mo:Na substrates is not only caused by the additional Na doping but also by a less pronounced re evaporation of In, which leads to a Cu poorer final composition. For the 1 stage process a significant In-excess is necessary to achieve a final composition which is Cu-poor enough to lead to cells with high open circuit voltage Voc and low interface recombination. However, at the same time Cu richer deposition conditions are advantageous for a better material quality (larger grains and larger diffusion length) and higher short circuit current densities (Jsc). Therefore, it was tried in first experiments to reduce the [Cu]/([In]+[Ga]) ratio of an intermediate (nearly) Cu-rich film with a final Cu poor stage. This led to cells, which achieved high Jsc and high Voc at the same time with a maximum efficiency of 12.2% (Voc = 558 mV, Jsc = 31.1 mA/cm², FF = 70.6%).