Key elements involved in the fabrication of Micromorph thin-film silicon solar cells, a tandem device including an amorphous silicon top cell and a microcrystalline silicon bottom cell, are studied in this manuscript. Due to the very short diffusion lengths of photogenerated carriers in both materials, the photoactive layers of both sub-cells of Micromorph devices must be kept thin enough to ensure carrier collection. Due to this limited thickness, not all valuable light (i.e. light of higher energy than the band-gap) can be fully absorbed after one pass through the absorber layers. Advanced light-harvesting schemes are thus mandatory to achieve high conversion efficiencies. Random rough interfaces are typically used to induce light scattering in the photoactive layers, thus elongating the light path through these layers, enhancing their absorption. A simple and analytical way of modeling light harvesting in thin-film solar cells is developed. Its validity is demonstrated by comparing with experimental measurements involving different types of rough interfaces. It is shown that present light-scattering schemes come close to the best theoretically achievable scattering from random rough interfaces. With the morphology of a state-of-the-art rough ZnO layer, 32mA/cm2 can be obtained for a 1-μm-thick μc-Si:H layer, compared to 33.2 mA/cm2 for the Yablonovitch limit. Most of the gains in terms of light management are therefore to be made by making non-active layers more transparent (since these layers are presently responsible of ∼ 7 mA/cm2 of losses for a 1-μm-thick μc -Si:H layer). Parasitic absorption in non-active layers is also shown to be equally detrimental on both sides of the cell in the infrared part of the spectrum, corresponding to the wavelength range for which light trapping is most important. To improve significantly light trapping, complementary strategies to random rough interfaces must therefore be applied. For such a strategy, part of the light has to be prevented to escape from the cell. This can be obtained for example by using an angular filter, transmitting all light up to a certain incidence angle, and reflecting all light of higher incidence angle. An experimental optical setup based on spatial filtering is presented. It is shown to prevent 80% of light from escaping the cell, additionally to other light trapping strategies. A strong absorption enhancement of the complete device is demonstrated, at the cost of a reduction of the acceptance angle. However, most of the spared light is shown to be absorbed in non-active layers. A drastic reduction of parasitic absorption from these layers is therefore identified as a prerequisite to benefit from a better light trapping. Turning then to complete device analysis, the requirements of the front electrode for high- efficiency Micromorph devices are discussed point-by-point, focusing both on optical and electrical requirements. The need for sharp and relatively small features for an efficient coupling of light in the top cell is notably pointed out, as well as the need for features large enough to scatter light of wavelength up to 1100 nm for a high bottom cell current. The impact of sharp and large features on the electrical performances of the cells is also underlined. An optimal morphology is proposed, exhibiting features that are the smallest enabling scattering of light up to 1100 nm, and the sharpest that do not harm the electrical quality of the bottom cell. A device with a stable efficiency of 11.8% could be obtained on such a substrate. To avoid this trade-off, the combination of features with different sizes in a multi-scale elec- trode architecture is studied. It is shown that even though no gain in terms of light scattering is seen compared to a state-of-the-art single-scale morphology, a better electrical quality can be obtained, making the multi-scale approach of interest. A noteworthy 14.1% initial efficiency is demonstrated with such an architecture, with a short-circuit current density of 14 mA/cm2 for a total thickness of the silicon layers of 3 μm. Another way to add a degree of freedom in the design of a Micromorph device is also presented, employing a smoothening intermediate layer between both sub-cells. A separate tuning of the morphology inducing light scattering in each sub-cell can be made. Several routes are explored, out of which promising results are obtained by using a spin-coated lacquer on the top cell. A slight etching uncaps the tips of the top cell surface and allow for electrical conduction, whereas the smoothening effect is preserved. A large top cell current boost (up to+2.5mA/cm2)andopen-circuitvoltageimprovement(+50mV)areobtained,withmany adjustable processing parameters to obtain various morphologies. The newly developed concepts enable a better understanding of the present limitations of Micromorph devices. We believe that, by implementing these concepts in Micromorph tandem devices employing state-of-the-art amorphous and microcrystalline single-junction solar cells, stable 13.5% to 14% efficiency are within reach. Yet, increasing further the efficiency of Micromorph devices will likely require material improvements.
EPFL2012
