Variants of garnet-type Li7La3Zr2O12 are being intensively studied as separator materials in solid-state battery research. The material-specific transport properties, such as bulk and grain boundary conductivity, are of prime interest and are mostly investigated by impedance spectroscopy. Data evaluation is usually based on the one-dimensional (1D) brick layer model, which assumes a homogeneous microstructure of identical grains. Real samples show microstructural inhomogeneities in grain size and porosity due to the complex behavior of grain growth in garnets that is very sensitive to the sintering protocol. However, the true microstructure is often omitted in impedance data analysis, hindering the interlaboratory reproducibility and comparability of results reported in the literature. Here, we use a combinatorial approach of structural analysis and three-dimensional (3D) transport modeling to explore the effects of microstructure on the derived material-specific properties of garnet-type ceramics. For this purpose, Al-doped Li7La3Zr2O12 pellets with different microstructures are fabricated and electrochemically characterized. A machine learning-assisted image segmentation approach is used for statistical analysis and quantification of the microstructural changes during sintering. A detailed analysis of transport through statistically modeled twin microstructures demonstrates that the transport parameters derived from a 1D brick layer model approach show uncertainties up to 150%, only due to variations in grain size. These uncertainties can be even larger in the presence of porosity. This study helps to better understand the role of the microstructure of polycrystalline electroceramics and its influence on experimental results.
Read full abstract