Grégory Bugnon
This thesis investigates the link between the plasma deposition conditions and microcrystalline silicon (μc-Si:H) material quality for thin-film silicon photovoltaic applications. The role of interfaces and the μc-Si:H material quality on the device performance are analyzed in detail. The low absorption of μc-Si:H at long wavelengths requires the deposition of absorber layers with thicknesses of typically a few micrometers for use in multi-junction TF Si solar cells. The growth typically takes place on highly textured surfaces, which provide increased light absorption—often called light trapping—but which potentially induce structural defects in the film during its growth. Therefore, to further improve the TF Si technology, one of the main challenges is the identification of the determinant plasma deposition parameters that result in the growth of very high-quality μc-Si:H at an increased deposition rate on textured substrates that guarantee efficient light trapping. As a first approach to better understand the plasma conditions necessary for the growth of high-quality μc-Si:H, the roles of both the silane depletion fraction and the deposition pressure are studied in an industrial-type large-area KAI reactor. With increasing pressure and silane depletion, the μc-Si:H defect density is significantly lowered leading to improved solar cell performance. An estimation of the average energy with which ions impinge on the substrate supports the hypothesis that ion bombardment is mainly responsible for the observed differences. Then, a fundamental aspect of μc-Si:H deposition on highly textured substrates is highlighted: two different phases of μc-Si:H material contribute to the overall solar cell efficiency, both of which can drive cell performance. The first phase relates to the bulk material and dominates the performance of cells on flat substrates. However, on rough morphologies, substrate-induced defective localized nanoporous regions—the second phase—develop and are found to be significantly more sensitive to the plasma process conditions and substrate morphology than the bulk phase. The relative importance of this secondary defective phase is shown through the use of new damp-heat experiments. Silicon oxide doped layers are demonstrated to mitigate the influence of these nanoporous regions on the solar cell performance. Next, a comparative study of the plasma excitation frequencies of 13.56 MHz (RF) and 40.68 MHz (VHF) shows that, while both allow for the growth of very good-quality bulk material, the efficiency of VHF-prepared cells is always poorer compared to that of RF-prepared cells within the range of our study for growth rates below 5 Å s−1. This decrease in solar cell performance is related to a higher density of nanoporous regions in the VHF-prepared cells as evidenced by damp-heat experiments, leading to strong open-circuit voltage instabilities. Still, the use of VHF is shown to be beneficial at increased deposition rates, thanks to reduced ion bombardment and improved bulk material quality. The crucial interplay between μc-Si:H growth rate and substrate morphology with regard to the formation of nanoporous regions is further discussed for regimes with high deposition rates of around 10 Ås−1. It is shown that high-silane-depletion regimes with significantly reduced H2 flow rate or increased pressure lead to a denser μc-Si:H material but are associated with increased secondary gas-phase reactions and powder formation. The use of a reduced interelectrode distance is demonstrated to allow for the growth of μc-Si:H with significantly improved bulk material quality at higher growth rates. Plasma simulations performed in collaboration with the University of Patras are presented and suggest that improvements observed in the μc-Si:H material quality are related mainly to an increased contribution of less reactive silane monoradicals, such as SiH3 and Si2H5, to the growing film, as compared to highly reactive ones such as SiH2 and SiH. Then, in an effort to better understand the formation of these two distinct μc-Si:H phases, the intrinsic stress within μc-Si:H i-layers is studied and correlated with the bulk defect density. Further improvements to both μc-Si:H and a-Si:H solar cells are obtained by introducing a novel intrinsic silicon oxide buffer layer at the p -i interface. For μc -Si:H solar cells, all electrical parameters are improved unless the i -layer is significantly more amorphous-rich and high quality, in which case an improvement only in carrier collection in the blue region is observed. In a-Si:H solar cells the silicon oxide buffer is shown to lower light-induced degradation, which is one of the weak points of TF Si technology. Furthermore, for both a-Si:H and μc-Si:H solar cells, the buffer can also act as an efficient barrier to boron cross-contamination, eliminating the need for additional time-consuming processing steps such as a water flush for single-chamber processes. Overall, this work contributes to a better understanding of the μc-Si:H material requirements for PV applications and how they relate to the plasma deposition conditions. Based on all the aforementioned developments, significant progress has been made in the understanding and the fabrication of thin-film silicon solar cells based on μc-Si:H. An outstanding single-junction μc-Si:H solar cell of 10.9% was attained; to our knowledge this is the highest reported in the literature. This work also contributed to the development of very high-efficiency tandem and triple-junction thin-film silicon solar cells at PV-Lab.
EPFL2013
