ADEM Publications, Presentations and other output



Hydrogenated amorphous silicon: nanostructure and defects


Melskens, Jimmy;





Publication date


Number of pages


Full text

not available


Since the first report on the fabrication of hydrogenated amorphous silicon (a-Si:H) in 1965, this material has found many applications, for instance in the fabrication of solar cells, sensors and transistors. An especially notable example in the context of this thesis is thin-film silicon solar cells, which have attracted a lot of attention. This is largely due to the fact that this type of photovoltaic technology can be implemented on flexible substrates and is a potentially cheap and lightweight product. However, a-Si:H-based solar cells suffer from light-induced degradation (LID) which can only partially be recovered by annealing. This issue, which has become known as the Staebler-Wronski effect (SWE), is obviously an undesirable material property in the context of photovoltaic applications. Although the solar cell conversion efficiency of thin-film silicon solar cells has seen significant gains over the years, research efforts are still ongoing to achieve further improvements. A significant portion of these improvements has come from the development of new materials or improvements in existing materials. This type of fundamental research is inherently linked to the understanding of the nanostructure and the defects therein to enable further material quality improvements. In this thesis it is aimed to fundamentally improve the understanding the a-Si:H nanostructure and the defects in this material to finally enable a reduction or even elimination of the SWE. Unfortunately, the SWE has proven to be a notoriously difficult problem due to the complexity of the a-Si:H nanostructure, which is the cause of the lacking consensus on the nature of the defects in this material. Because of this complication, there is a two-stage research approach in this thesis and it is not directly aimed to fabricate more stable a-Si:H. The first objective is to improve the fundamental understanding of the nanostructure and the defects in a-Si:H. Only secondly and using this newly gained knowledge, the SWE and the nature of metastable defects are studied to pave the way towards a reduction of the SWE. A wide range of material characterization techniques is used throughout this thesis to conduct extensive, systematic defect studies on both a-Si:H films and solar cells. From the studies conducted on a-Si:H films with widely varying nanostructures it has become clear that Doppler broadening positron annihilation spectroscopy (DB-PAS) and Fourier transform infrared (FTIR) spectroscopy are powerful complementary tools when trying to determine to dominant type of open volume deficiency in the material. For high quality a-Si:H it is concluded that divacancies are the dominant open volume deficiencies, while the dominant type of open volume deficiency can get as large as nanosized voids when the deposition rate is increased sufficiently. When investigating the electrical properties of defects in a-Si:H, Fourier transform photocurrent spectroscopy (FTPS) is used to quantify the sub gap absorption, which contains information about defects in the a-Si:H bandgap. This method can be applied both to a-Si:H films and solar cells. By using voltage biasing, in situ light soaking, or in situ annealing on an a-Si:H solar cell it is possible to manipulate the occupation and density of states in the bandgap of the absorber layer and monitor any induced changes. It is this accuracy that has enabled the first ever reported observation of four sub gap contributions in the absorber layer bandgap of an a-Si:H solar cell. Considering the apparently important role of open volume deficiencies in the a-Si:H nanostructure and the fact that there are more than two distributions in the a-Si:H bandgap, it is unlikely that the commonly assumed continuous random network (CRN), in which isolated dangling bonds (dbs) are the dominant defects, is an accurate description of the nanostructure. Instead, the disordered network with hydrogenated vacancies (DNHV) is proposed as a nanostructural description which is more likely to be correct. When using annealing as a probing tool to study the nanostructure of the as-deposited state, it is shown that a-Si:H films that are deposited from hydrogen-diluted silane gas, which are known to have an enhanced light soaking stability, have a smaller type of dominant open volume deficiency than a-Si:H deposited from pure silane gas, which typically exhibits a substantial LID. Additionally, the hydrogen passivation degree of small open volume deficiencies increases with increasing hydrogen dilution. Irrespective of the used hydrogen dilution, three different processes are observed by means of DB-PAS during annealing: vacancy agglomeration, hydrogen effusion, and crystallization. Since these processes take place at higher temperatures when the a-Si:H film is deposited at increasing hydrogen dilution, a reduced mobility of open volume deficiencies is linked to an enhanced light soaking stability. Finally, FTIR spectroscopy results indicate that the hydrogen effusion is a two-stage process, in which hydrogen initially mostly effuses from the small open volume deficiencies and only secondly predominantly from larger open volume deficiencies. These findings further underline the importance of including open volume deficiencies in an accurate description of the a-Si:H nanostructure and serve as further evidence against the CRN and in favor of the DNHV. Using all these new fundamental insights into the a-Si:H nanostructure, the nature and kinetics of LID in a-Si:H films and solar cells are investigated. Through a combination of DB-PAS, FTIR spectroscopy, and current density-voltage (JV) characterization, it is concluded that the most stable type of a-Si:H is deposited at a low deposition rate, which is associated with the smallest type of dominant open volume deficiency, i.e. a well hydrogen-passivated divacancy. By means of FTPS it is concluded that the four observed sub gap contributions have different light-induced defect creation rates during light soaking and do not follow a single time dependence ~tb. More specifically, it does not generally hold that b = 1/3 or b = 1/2, as has been repeatedly reported in literature. Furthermore, the generation and recombination profiles are spatially correlated, causing the recombination in the top and bulk parts of the absorber layer to be different. Since these findings call for a more advanced SWE model, which involves defects related to open volume deficiencies in addition to the dbs present in the CRN, a new nanoscopic description of the LID is given based on a combined characterization approach involving DB-PAS, continuous wave electron paramagnetic resonance (cw-EPR) spectroscopy and pulsed electron paramagnetic resonance (p-EPR) spectroscopy. It appears that there is a long term process in the LID which does not depend on the a-Si:H nanostructure. More importantly, it is found that there are fast and slowly relaxing defect types, which are predominantly linked to defects created in small and large open volume deficiencies. This newly found link between LID and the a-Si:H nanostructure is an important step on the road towards fully understanding and further reducing the SWE and it once again shows that the DNHV is a more appropriate description of the nanostructure than the CRN.