Grain boundary segregation in multicrystalline silicon studied by correlative microscopy

Silicon represents the most economic feedstock for an environmentally friendly energy generation by photovoltaic and dominates the global solar cell market since decades. The vast majority of installed photovoltaic modules is based on multicrystalline Silicon (mc-Si) due to its low productions costs. However, the energy conversion efficiency of mc-Si based solar cells is significantly lower compared to monocrystalline Silicon, since it contains a much higher defect density. Crystallographic defects like dislocations and grain boundaries (GBs) are known to be particularly harmful to the cell efficiency as they can serve as energetically favorable traps for impurity atoms, which in turn can significantly increase the recombination activity of charge carriers. The present study quantifies and localizes impurity species at different types of GBs at highest spatial resolution with highest sensitivity and correlates this information to the crystallographic structure and the electrical performance of the interfaces. To this aim, GBs were characterized by electron backscatter diffraction (EBSD) and electron beam induced current (EBIC) and were subsequently analyzed by high resolution scanning transmission electron microscopy (HR-STEM) and atom probe tomography (APT). The correlation of EBIC and EBSD at a mesoscopic scale showed a trend towards higher recombination activity at higher order coincidence site lattice (CSL) GBs (higher Σ values). The chemical compositions measured by APT at the corresponding GBs represents one of the first direct quantification of GB segregation in mc-Si and show a similar trend towards higher impurity decoration with growing Σ of the CSL GBs. However, a more careful look on the recombination activity revealed that several exceptions from these trends exist and that these trends are no longer valid when the GBs planes are deviating from known low-energy configurations. Complex atomic structures of these interfaces could be resolved by HR-STEM, for example the previously unknown atomic configuration of an asymmetrical Σ9{111|115}c interface, which despite its complexity, showed a clear periodicity and a surprisingly high stability. Furthermore, in case of GBs with ambiguous habit planes at the mesoscale, a faceting of the GB at the nanoscale could be resolved by HR-STEM. The correlative APT analysis revealed that the impurity atoms are no longer distributed homogenously over the GB plane, but showed clear evidence for preferential segregation towards specific facets and specific linear facet junctions. This novel type of segregation, referred to as topological segregation, was analyzed by a unique 1:1 correlation of atomically resolved STEM and APT data obtained from the same specimen. These experimental results suggest a direct influence of the atomic configuration and coordination at the GB on its chemical composition. These observations are in agreement with recent theoretical calculations found in the literature. In turn, the preferential segregation, in particular with transition metal impurities, explains the increased recombination activity at these interfaces.The experimental results of this work show a consistently high impact of the GB orientation on its atomic structure, and thus on its local chemical composition and electrical performance. The observed GB structures and compositions can serve as an ideal starting point for further computational simulations, which have been initiated already to some extent, to better understand the theoretical background of the observed phenomena