Modelling of recombination velocity and doping influence in epitaxial silicon solar cells

Abstract

This work presents a study of n +pp + type epitaxial solar cells developed on different wafers derived from an ingot UMG-Si recrystallized by HEM method with efficiencies of about 10%. More precisely it focuses on the development of software adapted to this type of cells, which expresses the variations of photovoltaic parameters as a function of epilayer thickness calculated for different values of recombination velocity and doping concentration, in order to optimize the epilayer thickness and efficiency of the cells. For this purpose, the one-dimensional (1D) model is evaluated through a simulation program which takes into account the interaction between several parameters as well as the restrictions between them. By modelling short circuit current density, open circuit voltage and efficiency, cells of different grain sizes have been studied. In cases of low recombination velocity calculated results have shown that the photocurrent density and conversion efficiency ( J sc ∼30 mA/cm 2, η∼13.8%) saturate when epilayer thickness values are higher than ∼65 μm and the gain is minimal, while they are heavily affected from epilayer thickness in cases of high recombination velocity. Calculated results also show that in cases of low doping concentration values photocurrent density and efficiency saturate for epilayer thickness values higher than 65 μm. However, for higher values of doping concentration higher photocurrent density and efficiency can be achieved by thinner epilayers. This study also suggests a second best value of epilayer thickness (≤50 μm), which is significantly lower and induces minor reductions in photocurrent density and efficiency values. A further comparison between simulated and experimental curves of quantum efficiency under 1000 W/m 2 illumination shows good agreement for wavelengths longer than 0.8 μm. However, near the blue part of the solar spectrum the measured quantum efficiency is significantly lower than the simulated one, due to the absence of surface passivation and the differences in the reflection coefficient between experimental and simulated devices.