Today more than ever the world needs clean energy sources and thus a fast deployment and scaling up of the photovoltaic industry. In this context improving solar cell efficiency plays a major role. In order to achieve the maximum single junction efficiency for this material, technologies featuring carrier selective passivating contacts are foreseen to be the next successors of actual industrial cells. Among them, the silicon heterojunction (SHJ) solar cell is a promising technology which demonstrated efficiencies up to 26.7%. Its main specificity consists in the use of intrinsic and doped amorphous silicon (a-Si:H) to provide surface passivation and carrier selectivity to the c-Si, which allows to reach very high operating voltages in comparison to other c-Si based technology. The major drawbacks of SHJ solar cell is the large parasitic absorption occurring in those layers. To address this challenge, a strong focus on alternative non-silicon based thin films with wider band gap has built up in the literature. A first part of this thesis focuses on reviewing the fundamental process behind carrier selectivity and discuss the ability of different models from the literature to draw informations directly from the peculiar shapes of IV curves commonly observed in the solar cells featuring newly developed materials. We first discuss the simple four ideal diodes model of Roe et al to explain the occurrence of S-shapes depending on the contact quality. We discuss from a theoretical point of view how to improve the model by modifying the ideal diodes by other circuit elements as well as adding bulk recombinations. We test the theory experimentally and identified modifications that should be incorporated in the model. In a second part of the thesis, we explore the development of doped nano-crystalline silicon (nc-Si) layers to replace classical amorphous silicon layers. We investigate the influence of TMB and BF3 as dopant sources on the transparency and contact properties of nc-Si:H(p) layers as well as their integration in solar cells. We expose the roles of both gas to modify the crystallinity of the layer and find different optimum for both of them to lead high efficiencies, illustrated by a certified 23.9%-efficient solar cell. We report also on further optimization steps of this front junction architecture resulting in certified efficiency of 24.44%. We explore next the n-type nc-Si layers for application as window layer. In particular, we perform thickness and doping series to unravel the layer properties along its growth direction. In order to improve the low doping of the nucleation zone of nc-Si:H(n) at the i/n interface, a thin a-Si:H(n) buffer layer is introduced and shown to improve the passivation, selectivity and contact resistivity. Finally, we also report on the beneficial effect of an additional SiOx capping layer which improves both the reflection properties as well as all contact properties, broadening the optimal parameter window. The absorption loss in the front silicon layers remains however the highest share of losses, and therefore we explore theoretically the opportunity of using thinner front side layers or using thin TCO / silicon nitride bilayers to reduce the absorption losses. We finish the discussion by presenting an optimization roadmap, showing a possibility for double-side contacted SHJ cells to reach efficiencies above 26% without requiring any patterning or localization step.