Garnet-type solid electrolytes such as Li_6.25Al_0.25La₃Zr₂O₁₂ (LLZO) have received considerable attention for use as separators in solid-state batteries, since they combine high ionic conductivities and chemical compatibility with lithium metal. Although their nominal composition is often the same in many studies, interlaboratory reproducibility of experimental results is not necessarily guaranteed. The sintering protocol strongly affects the polycrystalline microstructure such as grain size distribution and porosity. However, the effect of the actual microstructure is usually not properly considered in data interpretation.

For example, the electrical transport properties of electroceramics such as LLZO are typically investigated using impedance spectroscopy (IS). The macroscopic transport parameters are determined using simple 1D equivalent circuit models. The parameter pairs (R_i, C_i) are then interpreted in terms of microscopic parameters (σ_i, ε_i) using a brick layer model (BLM) approach. However, it is usually ignored that the BLM cannot provide error-free microscopic transport quantities: The BLM assumes a dense polycrystalline sample with a highly ordered arrangement of identical grains in a matrix, separated by a thin intergranular phase. The real system, however, typically has a non-uniform grain size distribution and pores. Consequently, the lack of dedicated studies regarding microstructure-related uncertainties in the transport properties obtained from a BLM analysis hinders interlaboratory comparability.

Thus, we performed a comprehensive analysis of experiments and simulations to estimate the uncertainty in the BLM-derived transport quantities.[1] Different sintering protocols are used to deliberately manipulate the microstructure of LLZO pellets. We then statistically analyzed the sintering induced microstructural changes using a machine learning assisted image segmentation approach combined with (FIB)-SEM data. Experimental IS characterization of the individual LLZO samples using the standard BLM approach suggests continuous changes of material-specific transport properties with sintering time. However, by performing 3D transport computations through statistical twin microstructures using an impedance network,[2] we show that the material-specific properties have not necessarily changed during sintering. In the case of a dense microstructure, the grain bulk conductivity evaluation is not much affected by the use of a BLM. In contrast, the uncertainty in the BLM-derived grain boundary conductivity can be up to 150% due to variations in grain size in the real microstructure alone. The effect of porosity on bulk and grain boundary transport parameters is even more severe and strongly depends on the spatial distribution of pores in the ceramic.

The study as a whole emphasizes the importance of microstructural analysis for a proper interpretation of macroscopic (impedance) measurements.[1] It provides a guide to estimate the true value range of transport quantities, helping experimentalists distinguishing between microstructural effects and true changes of the material-specific properties. This will improve the understanding and optimization of the characteristics of solid electrolytes, and it will also increase interlaboratory compatibility of reported results.