DC Breakdown in gases for complex geometries from high vacuum to atmospheric pressure
Lausanne, EPFL, 2013.
DOI : 10.5075/epfl-thesis-5962.
A Grid Reactor with Low Ion Bombardment Energy for Large Area PECVD of Thin Film Silicon Solar Cells
Lausanne, EPFL, 2013.
DOI : 10.5075/epfl-thesis-5686.
Arcing in Very Large Area Plasma-Enhanced Chemical Vapour Deposition Reactors
Lausanne, EPFL, 2011.
DOI : 10.5075/epfl-thesis-5090.
This thesis presents an investigation into arcing and parasitic plasmas in large area plasma enhanced chemical vapour deposition reactors. Two types were investigated: RF breakdown in millimetric gaps in absence of plasma (e.g. dark space shielding), in a millibar pressure range, and RF hollow cathodes in glow discharges. RF breakdown curves (voltage vs. pressure) for parallel plate electrodes generally show a steep left-hand branch at low pressures and a flatter right hand branch at higher pressures. Introducing protrusions or holes in parallel plate electrodes will lower the breakdown voltage in certain conditions. This is, however, not due to the increased electric field at sharp edges or ridges. Instead, both experiments and simulation show that breakdown at high pressure will occur at the protrusion providing the smallest gap, while breakdown at low pressure will occur in the aperture providing the largest gap. This holds true as long as the feature in question is wide enough: Features that are too narrow will lose too many electrons due to diffusion, either to the walls of the apertures or to the surroundings of the protrusion, which negates the effect on the breakdown voltage. An analytical approximation of breakdown in parallel plates with cylindrical protrusions supports this argument. The simulation developed to study breakdown in structured parallel plate electrodes also presents a tool to aid the design of complex RF parts for dark-space shielding. A method was developed to measure the pressure-limits of ignition for RF hollow cathodes, and it could be shown that these limits not only depend on gas type, diameter and depth of the hollow cathode, but also on the presence and/or absence of other hollow cathodes in the vicinity. It could also be conclusively shown that hollow cathodes damage the electrode by sputtering and/or evaporation.
Study of radio-frequency plasma deposition of amorphous silicon for the improvement of solar cell production
Lausanne, EPFL, 2004.
DOI : 10.5075/epfl-thesis-2900.
Plasma enhanced chemical vapour deposition (PECVD) of thin films such as amorphous silicon has widespread applications especially in the field of photovoltaic solar cells and thin-film transistors for flat screen production. Industrial applications require high depositions rates over large areas with a good uniformity in layer thickness. In this thesis, some aspects of PECVD in large surface, industrial type, capacitive radio frequency reactor are investigated. The aim of this work is to study the plasma process conditions to increase the deposition rate of uniform, good quality, a-Si:H layer for solar cell application in a single chamber reactor. The studies realized during this thesis have necessitated the development and the comprehension of diagnostics such as deposition rate measurement by in-situ interferometry, plasma power measurement and layer density measurement by ellipsometry. During the thesis, we have also elaborated a matching-box circuit for using process frequencies of 27.12 MHz and 40.68 MHz. Deposition of a-Si:H in small electrode gap reactor has been studied. At present industrial reactors have a standard electrode gap of 2.4 cm. We modified a reactor to reach a small gap of 1.7 cm. It appears that we have obtained faster deposition rate in the small gap reactor but non-uniformity problems increase due to edge focusing and powder effects. One solution, based on a teflon jigsaw to keep the plasma away from the edge confined spaces, is proposed to suppress focusing effect but the operation parameter space is still reduced by the powder effect. Systematic measurements of a-Si:H layer density was also done by ellipsometry. It is shown that the layer density decreases when the deposition rate increases, independently of pressure, gas flow and frequency (27.12 MHz/40.68 MHz) of the plasma. At high deposition rate, 6 Å/s, only an increase of the process temperature from 200°C to 230°C can significantly improve the layer density. We have noted also a slight improvement of layer density for layers deposited in the small gap reactor. Nevertheless industrial constraints impose a process temperature of 200°C and a standard gap reactor. By optimising the process parameters, keeping the temperature process at 200°C, good quality, uniform, a-Si:H layer were deposited at 3 Å/s on 37 cm × 47 cm glass substrates at 40.68 MHz. A particular source of non-uniformity in large area reactor has been examined. In large area reactors, a perturbation in RF plasma potential, due to the electrode edge asymmetry, propagates towards the plasma center with a characteristic damping length λ. The variation of RF plasma potential at the edge implies a variation of the deposition rate across the reactor area and then a non-uniformity of the deposited layer. A theoretical study was done to understand of this phenomenon and experimental results confirmed the model. Finally, for solar cell applications, a study of the boron cross-contamination during solar cell deposition in a single chamber process has been done. During the deposition of the intrinsic layer, for a p-i-n cell, i-layer is contaminated by the residual boron radicals present in the reactor after the deposition of the boron-doped layer. This contamination decreases cell performance. Several reactor treatments have been tested to solve this contamination problem. The effectiveness of these treatments was evaluated by secondary ion mass microscopy (SIMS) measurements. It appears that an ammonia flush or a water vapour flush of a few minutes, between the deposition of the p-layer and i-layer, reduces the boron contamination at the p-i interface. The performance of cells made with these treatments, in a single chamber process, are comparable to performance of cells done in a multi-chamber process.