A reliable quantification of structural defects is crucial to improve the quality of multicrystalline Silicon, in particular to reduce dislocations and to maximize the monocrystalline part of the ingot for cast-mono Silicon. This work contributes tools to this end: Methods were developed to achieve a statistically valid defect analysis via 2D segmentation and 3D reconstruction, applied on more than 6000 wafers from various materials. Image processing techniques were developed for the extraction of grain structure and dislocation clusters in optical and photoluminescence images. A meaningful quantification of these structures via 2D defect characteristics shows that dislocation reduction is correlated with a finer grain structure in the bottom region. Further, 3D defect development is characterized by a vector field which also serves to reconstruct and quantify 3D defect objects. Thereby, it is analyzed how functional grain boundaries reduce dislocations and stop parasitic grains. Use-cases on varied crystal growth settings show how this work could contribute to improve wafer quality.weiterlesen