[20a-P9-1~12] 16.3 Bulk, thin-film and other silicon-based solar cells

[20a-P9-6] Derivation of average minority carrier lifetime of multicrystalline silicon wafers for prediction of photovoltaic conversion efficiency

〇Koji Amazutsumi¹, Shota Toyokura¹, Johannes Giesecke² (1.Kyocera Corporation, 2.Faunhofer ISE)

Keywords:semiconductor, crystalline silicon solar cell, photoluminescence

Precise prediction of potential photovoltaic conversion efficiency of multicrystalline Si wafers at an early stage of production line allows us to accelerate wafer development and to cut expense losses by picking away poor quality wafers in line. We report our recent progress about how to derive single representative excess minority career lifetime of multicrystalline Si wafers for prediction of the potential photovoltaic efficiency by using photoluminescence (PL) imaging. We measured excess minority career lifetime maps of 14 multicrystalline Si wafers with their surface passivated by Al₂O₃, by PL imaging and harmonically-modulated PL under incident photon flux of around 1 sun. 5 sister wafers of each passivated wafer were fabricated into solar cells and thier conversion efficiency was measured by solar simulator. We also measured external quantum efficiency, surface reflectivity and dark current-voltage characteristics of those solar cells to correct the effects of fabrication variation to solar cells such as surface reflectivity and series resistance by finger electrodes, and thus to quantify the immediate impact of wafer electronic quality on solar cell efficiency. The correlation between raw efficiency of solar cells and average lifetime of passivated wafers calculated applying the weighing function presented in literature showed the root mean square deviation (RMSD) of 0.014 in the unit of normalized conversion efficiency. In order to search a better averaging method of lifetime maps in terms of RMSD, dependency of efficiency on bulk lifetime and doping concentration in base Si material was simulated by PC1D device simulation on our base line model. Using the weighing function that best fit to the simulated curve as that of lifetime map and applying the modification on efficiency data stated above, we obtained improved RMSD of 0.0042.