Enhancement of light trapping in thin film silicon solar cells

Abstract

Recently photovoltaic techniques have drawn extensive attention both in research and industry. Thin film silicon solar cells are becoming more promising due to their low material and energy consumption, when compared to wafer-based crystalline silicon solar cells. However, the efficiency of thin film solar cells is still relative low. Light trapping is one of the approaches through which the efficiency can be considerably improved. To achieve light trapping in the thin film silicon solar cells, rough surfaces presented on transparent conductive oxide (TCO) are included, either on top of or at the bottom of solar cells. The rough surface enables strong light scattering which enhances the absorption in the thin absorptive layers. In this way the short circuit current can be improved as well as the efficiency. In this thesis, we investigated three types of TCOs, including fluorine doped tin oxide (FTO), aluminum doped zinc oxide (AZO) and boron doped zinc oxide (BZO). Light scattering properties of these TCOs and their surface morphologies were studied. Morphological and optical measurements were carried out to determine surface root-mean-square (rms) roughness, haze and angular distribution functions (ADFs). The results showed that with an increased surface rms roughness, an enhancement in light scattering was observed. This effect becomes more pronounced especially when the rms roughness is comparable to the shorter wavelength, i.e. in ultraviolet/visible region. However, it is found that rms roughness could not completely describe the scattering behaviors, the lateral features are also required. In particular, pyramidal structures are more likely to scatter light into angles close to the surface normal, while crater-like surface structures are more favorable into angles out of the incident light direction. With a better theoretical understanding of the scattering behaviors, one can propose optimized surface structures for future manufacturing process.