This thesis reports on the study and use of low temperature processes for the deposition of indium gallium nitride (InGaN) thin films in order to alleviate some of the present drawbacks of its monolitic deposition on silicon for photovoltaic applications. The first of these processes is plasma-enhanced chemical vapor deposition (PECVD), with which it was proven that lowering the growth rate generally improves the quality metrics of the layer characteristics (Urbach energy EU , crystallite size, resistivity). It is also observed that the InGaN bandgap (BG) follows the quadratic Vegard's law with respect to indium content XIn. The use of a time-of-flight mass spectrometer (TOFMS) allowed for a deeper analysis of the plasma dynamics, and the key importance of different factors leading to a deposition of higher quality layer is presented. In particular, this higher quality layer is achieved by promoting a stronger dissociation, by a more efficient evacuation of the plasma by-products, and by starting the plasma before injecting the precursors containing carbon. Eventually, most of the different layer characteristics and quality metrics are modeled as a function of all but one (the radio-frequency) deposition parameters with more or less accuracy, hinting towards the deposition conditions that should lead to an optimized layer. It is concluded that the optimized InGaN PECVD layer doesn't perform well enough to act as an absorber at that stage of the research. This is due to p-doping that could not be achieved, to optical properties that were not satisfactory enough, as well as crystallinity and resistivity ones. However the performances were good enough to still hold some potential for a contact application.

The second of these studied processes is physical vapor deposition (PVD) sputtering, which highlights the strong role of the pressure on the layer basic characteristics, reaches acceptable electrical performances with germanium doping but with a too high germanium content of the layer, and worse electrical performances with silicon doping but with a lower silicon content than with germanium. Epitaxy of undoped gallium nitride (GaN) on sapphire is demonstrated, but remains unstable under an increasing doping level of silicon. Increasing the indium content of the layer helps decrease the resistivity, but still to values at least two orders of magnitude greater than the standard (n)a-Si:H layer usually used for contacting the absorber. For the exact same reasons as with PECVD, but especially because p-doping could not be achieved as well, it is concluded that the optimized InGaN layer doesn't perform well enough to act as an absorber, but has acceptable characteristics for a contact application. This thesis also reports on the integration of these optimized layers in solar cell architectures as contacts. It is observed that indeed, the "barely sufficient" layer properties (be it optical, electrical or structural) doesn't improve the cell performances compared to state-of-the-art literature values. However, for each technique used and presented (undoped In/GaN by PECVD, doped GaN and undoped InGaN by PVD), an electrical power is produced by the solar cell under standard test conditions, in one case even stronger than the reference cell it is compared to. The main two issues with these contacts are their too high resistivity (decreasing the open-circuit voltage (VOC) and worsening the fill factor (FF)) and their bad selectivity.