Large Bandgap Photovoltaic Devices and Tandem Integration with Silicon Solar Cells

In this thesis, I present the main results of my PhD project, which focuses on the study of selenium as a photovoltaic material. Selenium, the world’s oldest photovoltaic material, is experiencing a research renaissance due to its irresistible monoatomic simplicity, as well as its wide bandgap, making it a potential candidate for the top cell absorber in tandem devices. However, achieving high-­efficiency selenium thin-­film solar cells remains a challenge, and despite being the key driver of this renewed attention, a tandem device featuring selenium has yet to been demonstrated. Addressing these issues constitutes the core of my work. Here, I list what I consider to be the main highlights.
First, we developed and optimized a process flow to fabricate tunnel oxide passivated contact (TOPCon) silicon solar cells, which serve as the bottom cells in our tandem devices. While silicon technologies are well­-established in industry, our in­house processing allows us to modify the device structure to accommodate the top cell processing. The optimized silicon solar cells demonstrate implied open-­circuit voltages greater than 0.72 V and implied fill-factors exceeding 83%, all while remaining resilient to the back­end processing.
Second, we reproduced the state-­of­-the-­art selenium solar cell device architecture presented in a recent publication by IBM. From this baseline, we replaced the gold electrode with a transparent conductive oxide to fabricate the first bifacial selenium solar cells, and discovered that our selenium solar cells are strongly polarity dependent. To address this polarity dependence, we investigated potential device inversion strategies and the band alignment in our devices. This study lead us to the conclusion that our ZnMgO/poly­-Se heterostructure forms an ideal pn-heterojunction. However, due to the optimal elemental composition of ZnMgO, the FTO/ZnMgO heterointerface is prone to form charge carrier transport barriers. To resolve this issue, we explored the potential of multi­layered buffers and native doping strategies, guided by SCAPS-1D device simulations. Additionally, to improve reproducibility, we developed a closed-­space annealing strategy, which significantly enhances the collection efficiency in our devices.
Third, we demonstrated a record open­-circuit voltage of 0.99 V, providing an ideal starting
point for investigating the origin of the still significant open­circuit voltage deficit. In this study, we found that trigonal selenium possesses a quasi-­indirect bandgap, the thin-­films exhibit no detectable photoluminescence at room temperature, the absorption onset is quite broad, and the photoconductance decay features two characteristic lifetimes. Following the associated publication, I will present new, unpublished results, where the mobility-­lifetime product is interpreted differently. This new interpretation leads to our device simulations quantitatively matching our experimental data.
Fourth, we studied the defect physics in selenium. After concluding an ideal band alignment in our pn-­heterojunction, no detectable photoluminescence at room temperature, and that the effective carrier lifetime is likely to fall within the picosecond range, I have no doubt that the most significant losses in our devices originate from non­-radiative recombination within the absorber. However, as point defects and their properties may be challenging to probe experimentally, we used first­-principles methods based on density functional theory to support and guide our experimental investigation. We examined the effects of both intrinsic and extrinsic defects, and detected halogen impurities in our absorber layer originating from the tellurium source material. We suspect that the halogen species could potentially form killer defects in selenium, but this has yet to be proven.
Fifth, we explored the potential of laser­-annealing as a defect­engineering tool. Unlike thermal annealing, the charge carriers in selenium are not in thermal equilibrium with their surroundings during laser-­annealing. This non-­equilibrium state has implications for the formation energies of both extrinsic and intrinsic defects, and hence their concentrations. To prevent sublimation of selenium from the surface of the film, the laser is guided through the semitransparent substrate, resulting in the formation of a selenium crystallites in the vicinity of the carrier­-separating junction. This seed layer served as a growth template for solid-­phase epitaxy, facilitating the formation of larger, more preferentially oriented crystal grains with negligible surface roughness.
Finally, although the efficiencies of our selenium thin­-film solar cells are still too low for tandem integration to be viable, we have successfully fabricated and demonstrated the first monolithically integrated selenium/silicon tandem solar cell. The tandem device consolidates all the individual conclusions drawn throughout this project: polarity dependence, device inversion, the ideal band alignment of the ZnMgO/poly­-Se pn­heterojunction, the formation of charge transport barriers, the low­-energy photon collection efficiency issues, and the optoelectronic quality of our selenium thin­-films limiting the overall device performance. Additionally, during the crystallization process in this study, we observed the formation of large crater­like holes in the selenium thin-­films, a phenomenon that has been elaborated on in greater detail following the presentation of the tandem paper.