Interest in silicon heterojunction solar cells is growing due to their manufacturing simplicity and record efficiencies. A significant limitation of these devices stems from parasitic absorption in the amorphous silicon layers. This can be mitigated replacing the traditional (p) and (n) doped amorphous silicon selective layers by other materials. While promising results have been achieved using molybdenum oxide (MoOx) as front-side hole-selective layer, charge transport mechanisms in that contact stack have remained elusive and device efficiencies below predictions. By carefully analyzing the influence of the MoOx and intrinsic a-Si:H thicknesses on current-voltage properties, we discuss transport and performance-loss mechanisms. In particular, we find that thinning down the MoOx and (i) a-Si:H layers (down to 4 nm and 6 nm respectively) mitigates parasitic optical sub-bandgap MoOx absorption and drastically enhances charge transport, while still providing excellent passivation and selectivity. High-resolution transmission microscopy reveals that such thin MoOx layer remains continuous and close to a MoO3 stoichiometry in spite of the reactive sputtering and annealing steps involved in the electrode deposition. Based on these insights, a screen-printed device reaching a certified efficiency of 23.5% and a fill factor of 81.8% is demonstrated, bridging the gap with traditional Si-based contacts and demonstrating that dopant-free selective contacts can rival traditional approaches.

Dopant-free passivating contacts for photovoltaics have the potential to be deposited at low costs while providing excellent surface passivation and low contact resistance. However, one pressing issue of dopant-free carrier selective contacts is their lower environmental stability compared to conventional silicon-based contacts. In this contribution, we study the degradation in the ZnO/LiF x /Al electron selective nanocontact with experiments and simulations and suggest design modifications for higher performance and stability. Using a thicker metallization and optimal ZnO deposition temperature (130 °C), we improved open-circuit voltage and fill factor, together with improved stability with retention of over 93 and 88% of the initial open-circuit voltage and fill factor after storage in air for 380 h. The champion device has reached an efficiency of 21.3% with V OC of 727 mV, J SC of 37.6 mA/cm2, and FF of 78.0%. Furthermore, the enhanced stability in vacuum, scanning transmission electron microscopy (STEM) images, and the current-exchange simulation suggests that the degradation of the a-Si:H­(i)/ZnO/LiF x /Al contact is caused by a drop of the LiF x /Al work function, due to interaction with air. This work has developed a deep understanding of the degradation mechanism and the methodology of stability analysis for dopant-free silicon solar cells.

Herein, challenges in the fabrication of full dopant‐free bifacial silicon solar cells are discussed and efficient devices utilizing a MoO3/ indium tin oxide (ITO)/Ag hole‐selective contact and ZnO/LiFx/Al electron‐selective contacts with up to 79% short‐circuit current bifaciality are demonstrated. The ZnO/LiFx/Al rear electron contact features a full‐area ZnO antireflective coating and a LiFx/Al finger contact, allowing sunlight absorption from the back side, thus producing more overall power. The ZnO/LiFx/Al electron contacts with a thinner ZnO layer and a larger contact fraction display a better selectivity and a lower resistance loss. When considering rear‐side irradiance of 0.15 sun, the dopant‐free bifacial solar cell with 60 nm ZnO and 50% LiFx/Al metal contact fraction achieves a 3% estimated output power density improvement compared with its monofacial counterpart (21.0 mW cm−2 compared to 20.3 mW cm−2) using the full‐area back contact. Both the efficiency and bifaciality factor of this dopant‐free device are still significantly lower than those of state‐of‐the‐art devices relying on doped‐silicon‐based layers. The required improvement for this technology to become industry‐relevant is discussed.

Many silicon heterojunction solar cells tend to suffer modest fill factors due to high series resistance compared to homojunction or high-temperature-passivated-contact-based solar cells. Loss analysis indicates that this limitation lies in the high contact resistance between the wafer and the electrode (through the intrinsic (i) amorphous silicon (a-Si:H) passivating layer and amorphous silicon doped layers), mostly originating from the p-type contact. We implement p-type microcrystalline doped layers in heterojunction cells and demonstrate with 2-side contacted devices that a low CO2 concentration silicon oxide (SiO) plasma treatment on the a-Si:H (i) passivating layer allows to reach high crystallinity with thin layers without impeding passivation. The influence of the treatment time and CO2 concentration on lifetime and nucleation of the microcrystallites is discussed. We then show the potential of such a SiO treatment by comparing, in 2side contacted devices, amorphous (without SiO treatment) and microcrystalline player , the latter showing efficiency up to 21.5% prior to optimization.

The window-layer stack limits the efficiency of both-side-contacted silicon heterojunction solar cells. We discuss here the combination of several modifications to this stack to improve its optoelectronic performance. These include the introduction of a nanocrystalline silicon-oxide p-type layer in lieu of the amorphous silicon p-type layer, replacing indium tin oxide with a zirconium-doped indium oxide for the front transparent electrode, capping this layer with a silicon-oxide film, and applying a post-fabrication electrical biasing treatment. The influence of each of these alterations is discussed, as well as their interactions. Combining all of them finally enables the fabrication of a highly transparent and electrically well-performing windowlayer stack, leading to a screen-printed silicon heterojunction solar cell with 24.1% efficiency. Paths towards industrialization and for further improvements are finally discussed.