Environmental challenges impose a rapid energy transition. The renewable power capacity is expected to increase by 50% in the next 5 years, and recent scenarios plan that photovoltaics (PV) will be leading the new energy sources. Combining high efficiencies and simple fabrication process, silicon heterojunction (SHJ) solar cells have the potential to be the main technology for PV world wide deployment. Challenges yet remain, notably due to the use of a-Si:H as contact layers. MoOx appears to be one of the most promising alternative to tackle some of these challenges, owing its transparency and excellent electronic properties. This thesis focuses on the analysis and development of MoOx-based hole contact for SHJ solar cells. Contact passivation, selectivity and transport are assessed by varying the contact parameters, using standard (p)a-Si:H as a benchmark, and using characterization adapted for contact-limited SHJ.
By thinning down theMoOx layer from the standard 10 nm thickness to 4 nm, excellent passivation and selectivity were maintained. A current gain of 1.3mAcm-2 was obtained as compared with the standard a-Si:H contact. A 2x2cm2 screenprinted certified 23.5% world record device was achieved through this analysis, outperforming the reference (p)a-Si:H-based device. Further thinning the MoOx layer provoked several losses. Using many characterization methods combined to fittings using an innovative analytical description of contact-limited solar cells proposed by Roe et al., we could conclude that the losses stem from a passivation decrease combined with a selectivity reduction inducing a drop of the majority hole and an increase of parasitic electron currents in the contact, due to the reduction of the contact's work function.
Doping and deposition power of the ITO top layer was then modified and tested over several MoOx layer thicknesses, to estimate MoOx screening length towards ITO work function and investigate the influence of ITO on transport. 4-nm-thickMoOx showed to efficiently screen the ITO work function on the tested range while the 2-nm-thick-MoOx-based contact was affected. The ITO deposited at low power drastically enhanced the FF (» 1%absolute). In situ work function estimation of the ITOs (4.6 to 4.8 eV) and hole contact (» 5.0 eV) was conducted using fittings according to Roe's formalism. PECVD-deposited-a-SiOx:H, commonly used on top of fully-processed SHJ to boost the JSC, degraded theMoO3-x-based devices FF, ascribed to the presence of H in the plasma reducing MoOx.
MoOx-based devices showed to be remarkably stable over time. An annealing step performed on solar-cell samples prior toMoOx deposition was shown to slightly damaged the passivation but protected MoOx selectivity degradation for processes temperature up to 170 °C. Similarly to standard SHJ devices, current injection in dark in forward bias significantly improved the FF of the devices by 1.3%absolute. The cells were encapsulated and degraded by UV radiations, which showed to strongly affect selectivity.
A review of main transport mechanisms was gathered in the last chapter to parallelize with temperature-dependent J-V measurements (-160 to 70 °C) performed for MoO3x- and (p)a-Si:H-based SHJ. Voc(T) and FF(T) evolution showed similar trends in both cases, confirming the expected transport behavior from comparing band-diagrams. An extension of the method based on Roe et al. model is proposed to rapidly identify selectivity loss from S-shaped J-V curves.