Determining the uncertainty in microstructural parameters extracted from tomographic data

In the past few years, tomography has emerged as an indispensable tool to characterize and quantify the microstructural parameters of porous media, including lithium ion battery electrodes and separators, fuel cell parts, and supercapacitor electrodes, as well as electrolysis diaphragms. However, tomographic data often suffer from low image contrast between the pore space and the solid phase. As a result, data binarization is error-prone, which translates into uncertainties in the calculated parameters, such as porosity, tortuosity, or specific surface area. We systematically investigate this error propagation on a set of seven commercial negative lithium ion battery electrodes imaged with X-ray tomographic microscopy, and find that, for low porosity electrodes, the selection of binarization criteria can lead to uncertainties in the calculated tortuosity exceeding a factor of two. Our analysis shows that Otsu's interclass variance, which can be determined from a simple and computationally cheap analysis of the histogram of gray values, is a good predictive indicator of the expected uncertainty in the microstructural characteristics.