ABSTRACT
For grain size estimation, a prototype system was developed by integrating a vision-acquiring hardware and a vision-assistant-processing module based on the platform software package of LabVIEW, to systematically estimate the average grain size of solar-grade multicrystalline (mc)-Si wafers. Three groups of 156 x 156 mm mc-Si wafers were selected to produce the average grain sizes of 3.4 mm (Group 1), 3.8 mm (Group II), and 4.6 mm (Group III), and were used for the fabrication of mc-Si solar cells by employing the standard mc-Si cell fabrication procedure of the 30 MW mass production line. The conversion efficiency including Jsc and Pmax, showed a quasi linear dependence on the mean grain size, with a correlation factor of 0.525%/mm. By combining the EL image and the grain size/position-dependent EQE spectra in a wavelength range of 400-1100 nm, the conversion efficiency of uniformly-surface-texturized mc-Si solar cells with larger grain sizes can be made much higher as a result of the much-reduced spatial density of the nano/microscope grain boundaries acting as recombination centers or traps.
