A Method for Quantitative 3D Mesoscale Analysis of Solid Oxide Fuel Cell Microstructures Using Xe-plasma Focused Ion Beam (PFIB) Coupled with SEM

Abstract

Three-dimensional (3D) distributions of microstructural features in electrodes of solid oxide fuel cells (SOFCs) directly influence their electrochemical performance. Commercial grade SOFCs produced in cost-constrained environments can be expected to exhibit variations on the mesoscale (≈ hundreds of µm) of the microscale features (≈ tens of µm) that determine the local electrochemical activity. We are interested in quantitatively determining the mesoscale distributions of microscale properties in SOFC electrodes, especially quantifying the tails of distributions to establish correlations between such outliers and long-term performance. In this talk, we describe a new method that allows for the fast reconstruction of incredibly large volumes while maintaining a resolution on the order of tens of nm, allowing for fine-scale features to be accurately quantified. We use a Xe plasma focused ion beam (FIB) for slicing and a scanning electron microscope (SEM) for imaging. The pFIB allows for an increase in material removal rates by a factor of ≈ 100 compared to Ga ion FIBs, while the SEM resolution remains very high (herein ≈ 50nm), comparable to nano-CTs. In the same time as required for collecting Ga-FIB or nano-CT data sets, we collected a pFIB volume that is ≈ 70-80 times larger than attainable with Ga-FIB-SEM and ≈15-20 times larger than with nano-CT. From this data, we quantified the mesoscale distributions, containing both the mean and the variance, of key microscale features including volume fraction, average particle size, tortuosity, and triple phase boundary density. The commercial SOFC electrodes investigated were found to exhibit significant variations in the mesoscale distributions of their local microstructural features, which will be discussed. Importantly, these mesoscsale variations in microstructural features are expected to result in significant variations in electrochemical activity throughout the electrode. Using an effective medium theory model of polarization resistance, we quantified the distribution of microscale polarization resistance and mapped it spatially. Though the average value of polarization resistance is within range of expectation, the measoscale distribution of the microscale/local value of polarization resistance around the average indicates that several local areas are well outside the average, and potentially operating in regions expected to impact long-term cell performance.