(Invited) Interface Treatments for High-Efficiency MoOx Based Silicon Heterojunction Solar Cells

Abstract

Silicon based solar cells dominate the market of photovoltaics, which hold the highest potential for green electricity production. A front/back-contacted architecture combined with the silicon-based heterojunction (SHJ) concept realized a world record efficiency of 26.3%1. However, conventional silicon-based doped layers, which work as carrier selective transport layers, are not optically transparent.Transition metal oxides are promising candidates for SHJ solar cells due to their advantageous opto-electronic properties. Molybdenum oxide (MoOx) with high work function (WF) has achieved highly efficient solar cells with 23.5% efficiency2. However, depositing high quality MoOx thin film on top of a-Si:H is challenging due to highly reactive surfaces3. This interaction may result in interfacial oxidation, lower Mo oxidation states and reduced WF4,5. In turn, electrical performances of solar cells featuring MoOx are poorer3 than standard SHJ devices.Several attempts have been made to prevent the reaction between MoOx and (i)a-Si:H. Pre-annealing4 of a-Si:Hand pre-growth of SiOx layer6 have been proposed as effective ways to prevent the degradation of MoOx. Nevertheless, those approaches require extra process steps that make MoOx integration not immediately viable in industrial environment. In our previous work, we proposed a plasma treatment (PT) at MoOx/(i)a-Si:H interface that demonstrated to mitigate dipole effect and resulted in improved cell performance7. After optimizing the interface treatment, even thinner MoOx layer could be applied.The solar cells were fabricated using 260±20 μm 4-inch double-side textured n-type <100> FZ wafers. (i)a-Si:H was deposited by plasma-enhanced chemical vapor deposition on both sides, while (n)a-Si:H only at the rear side. We carried out two different plasma treatments on the (i)a-Si:H interface prior MoOx deposition: one called PT from a mixture of SiH4, H2 and CO2 and another called PTB from a mixture of SiH4, H2, CO2 and B2H6. One sample without plasma treatment (noPT) was used as reference to evaluate the efficacy of the proposed method. A thickness series from 1 to 4 nm of MoOx was thermally evaporated from MoO3 powder at the pressure of 5×10-6 mbar. After that, optimized 50-nm and 150-nm thick tungsten-doped indium oxide (IWO) layers were sputtered through a hard mask at front and rear side, respectively, defining six 2×2 cm2 solar cells per wafer. As metal contact at the front side, we used room temperature Cu plating8 with a metal coverage fraction of 1.575%; at the rear side, 500-nm thick Ag layer was sputtered on the full device area.We studied the impact of different plasma treatment conditions on MoOx film quality and cells’ performance. MoOx film quality was assessed based on O vacancies inside the film4. The samples with interface treatments contain less O vacancies than the sample without treatment. It is noticeable that the cell precursors endowed with PTB yield higher fill factor (FF) than other types of precursors. Instead, PT and noPT samples benefitted from high short-circuit current density (J SC). Ultimately, PTB and PT samples achieved similar conversion efficiency. For the variation of MoOx thickness, we observed that even a slightly change in its thickness influences the solar cells’ electrical performance drastically. An ultra-thin, ~1.7-nm thick MoOx layer could provide sufficient field passivation (open circuit voltage, V OC = 721.4 mV) and good opto-electrical properties (FF = 82.18% and J SC-illuminated-area = 40.20 mA/cm2). With our approach we demonstrate word record cell with ISFH-certified conversion efficiency of 23.83% 9. 1 LONGi Solar 2021. 2 J. Dréon, et al., Nano Energy, 2020. 3 J. Geissbühler et al., Appl. Phys. Lett. 2015. 4 M.T. Greiner, et al., Adv. Funct. Mater. 2013. 5 M.T. Greiner, et al., Adv. Funct. Mater. 2012. 6 J. Tong, et al. ACS Appl. Mater. Interfaces. 2021. 7 L. Mazzarella, et al., Prog Photovolt Res Appl. 2020. 8 C. Han, et al., Solar RRL 2022. 9 L. Cao, et al., Prog. Photovolt. 2022.