Anisotropic Evolution of Anode Microstructure During Long-Term Operation of Solid Oxide Fuel Cells

Abstract

Long‑term operation of a solid oxide fuel cell (SOFC) can drive significant anisotropic changes in its anode. Using electron nanotomography, we examined the microstructure of a commercial SOFC stack before and after 3 800 h of continuous operation. Analysis of the 3D digital reconstruction revealed that both the nickel (Ni) and pore phases developed pronounced anisotropy, whereas the yttria‑stabilized zirconia (YSZ) phase remained isotropic. These micro‑scale transformations alter electron and gas transport pathways, underscoring the need for anisotropic parameters in accurate SOFC models. Our findings provide essential data for realistic simulation of aged SOFCs.

Background

A solid oxide fuel cell converts the chemical energy of hydrogen directly into electricity. The cell is a flat plate in which a dense, ion‑conducting electrolyte is sandwiched between porous anode and cathode electrodes. Fuel is introduced at the anode, while air is supplied to the cathode. Gases are prevented from mixing; instead, they react at the catalyst surfaces, generating electrons that flow through the external circuit. Because the cathodic reaction is slower, a potential difference develops between the electrodes, and oxygen ions migrate from the cathode to the anode through the electrolyte. The electrode microstructure governs all transport processes, making its design critical for SOFC performance.

Typical anodes consist of three phases: Ni, which conducts electrons; YSZ, which conducts oxygen ions; and a pore network, which allows gas transport. Electrochemical activity occurs only at the triple‑phase boundary (TPB) where all three phases meet. Figure 1 illustrates these transport pathways.

A schematic view of the transport phenomena across a typical solid oxide fuel cell with a highlighted role of the microstructure.

Due to the complex composite nature of the anode, a microstructure‑oriented design is essential. In this study, we track microstructural evolution during extended SOFC operation, focusing on the anisotropic tortuosity factor—a metric that captures the directional complexity of transport pathways. We employ scanning electron microscopy combined with a focused ion beam (FIB‑SEM) to generate high‑resolution 3D reconstructions, enabling direct measurement of microstructural parameters relevant to long‑term degradation.

Our work is the first to demonstrate anisotropic evolution in SOFC anodes during long‑term use, highlighting the role of nickel particle migration, growth, and coarsening.

Experimental Apertures

Modular Stack Testing Bench

The aging test was performed on a Modular Stack Test Bench (MSTB) built by SOLID Power. Figure 2 shows the bench layout, with the stack housed inside an electric furnace.

A schematic view of the Modular Stack Test Bench.

Fuel and air were delivered via mass‑flow controllers and preheaters. Depending on the fuel composition, the gas stream either passed through a catalytic partial oxidation (CPOX) reactor (for methane) or bypassed it (for a hydrogen‑nitrogen mixture). Air was preheated and fed to the cathode channel, where part of it was consumed electrochemically and the rest was used to cool the stack and combust excess fuel in the afterburner. The fuel—hydrogen mixed with nitrogen—was preheated before entering the anode channel. After combustion, the gas was cooled, condensed water was removed, and the dry stream was vented to ambient air.

Seven thermocouples (labelled “T” in Figure 2) monitored temperature distribution. Each bipolar plate was connected to a wire and potentiostat, enabling current‑voltage characterization for every cell. The cell dimensions were 60 × 80 mm, with an active area of 48 cm². The stack operated in a co‑flow configuration, delivering up to 75 % fuel utilization and achieving power densities exceeding 1 W cm⁻². Detailed bench specifications are reported in references 27 and 28.

Focused Ion Beam–Scanning Electron Microscope

A dual‑beam system merges a scanning electron microscope (SEM) with a gallium ion source (FIB) in a single chamber. The SEM provides imaging, while the FIB mills the sample. The system allows sequential sectioning of the specimen, enabling reconstruction of a 3D microstructure. Figure 3 illustrates the setup and workflow.

The configuration of a dual‑beam system.

After depositing a protective carbon layer, a trench is milled to expose the intersection of interest. The cross‑section is polished with a low‑energy Ga⁺ beam, and a secondary‑electron image is captured using an in‑lens detector. This “cut‑and‑see” cycle is repeated to acquire 200–300 images, which are then reconstructed into a 3D volume (~1 µm³) with distinct contrast among Ni, YSZ, and pores.

Experimental Methodology

The study comprised two parts: a long‑term power generation test and a microstructural analysis. To accelerate aging, the stack operated at 800 °C with an imposed current of 19.4 A, delivering 90 W at the outset. Fuel utilization was maintained at 75 %. Full test conditions are summarized in Table 1. After 3 800 h, the stack was disassembled, and nine samples were extracted—three from cells 1, 3, and 5 (upstream, center, downstream) and one fresh reference cell that had not been cycled.

The locations of the selected samples in a cell and in the stack.

Each 5 mm × 5 mm sample was impregnated with epoxy resin and polished to expose the pore network. Using the FIB‑SEM, we acquired 3D reconstructions for all nine specimens. The raw SEM images underwent an intensive segmentation process—labeling each voxel as Ni, YSZ, or pore—which required up to one month per sample. Post‑processing involved resampling to cubic voxels and generating triangular surface meshes with AVIZO (ThermoFisher Scientific). The resulting digital models are shown in Figure 6.

Digital material representation of anode microstructure before and after the aging test. Reference sample, Cell 5 upstream, Cell 5 center, Cell 5 downstream, Cell 3 upstream, Cell 3 center, Cell 3 downstream, Cell 1 upstream, Cell 1 center, Cell 1 downstream.

To quantify microstructural complexity, we calculated the anisotropic tortuosity factor using a random‑walk algorithm. In this approach, numerous virtual walkers are randomly displaced within the pore network; the mean‑square displacement over time yields the effective diffusion coefficient, from which the tortuosity factor τ = D₀ / D(t) is derived. Anisotropic tortuosity components (τₓ, τᵧ, τ𝓏) are obtained by constraining walkers to single directions. The anisotropy factor ξ = √[(τₓ−τᵣ)²+(τᵧ−τᵣ)²+(τ𝓏−τᵣ)²] captures the deviation from isotropy. Phase mirroring ensures walkers remain within the finite reconstruction volume.

Results and Discussion

Figure 7 shows the stack’s terminal voltage over the 3 800 h aging test. The voltage remained stable, and polarization dropped during the first thousand hours. Despite this, our prior work revealed a significant decline in TPB area, which was spatially non‑uniform. Here, we demonstrate that microstructural evolution is not only heterogeneous but also anisotropic.

Terminal voltage as a function of operational time during long‑term operation.

Due to experimental artifacts such as curtaining and redeposition, the volume of interest varied among samples (typically ~1 µm³). For visualization, all reconstructions were cropped to a common size of 10 µm × 8 µm × 5 µm to enable side‑by‑side comparison (Figure 6).

Figure 8 presents the anisotropy factors for each location, compared with the reference. Key observations are:

• Reference anodes were isotropic; anisotropy grew markedly after aging.
• Anisotropy increased progressively downstream within each cell.
• Only Ni and pore phases exhibited strong anisotropy; YSZ remained isotropic.

The anisotropy factor at different locations in the stack and in a cell (UP, CE, DW, REF).

Nickel particle coarsening and directional migration—likely driven by vaporization–deposition of volatile species such as nickel hydroxide—are the most plausible drivers of anisotropy. Because migration is predominantly along the cell flow direction, it creates elongated Ni clusters and constricts pore connectivity, thereby increasing directional tortuosity. These changes explain why only Ni and pores become anisotropic while YSZ, a stable ionic conductor, does not.

Most diffusion models for SOFCs assume homogeneous porous electrodes. Our results suggest that this assumption fails after long‑term operation. When incorporating aged microstructural data into simulations, it is essential to account for direction‑dependent tortuosity, especially for gas diffusion calculations. Ignoring anisotropy may lead to inaccurate predictions of cell performance.

General Conclusions

We have demonstrated, for the first time, that prolonged SOFC operation induces pronounced anisotropy in the anode microstructure. 3D FIB‑SEM reconstructions, coupled with random‑walk tortuosity analysis, reveal that nickel and pore phases evolve anisotropically, while YSZ remains isotropic. Nickel migration, growth, and coarsening are the dominant mechanisms. These findings highlight the importance of incorporating anisotropic transport parameters into realistic SOFC models, particularly when simulating aged cells.

Availability of Data and Materials

The raw and processed data required to reproduce these results are not available at this time, as they are part of an ongoing study.

Abbreviations

CPOX

Catalytic partial oxidation

FIB

Focused ion beam

MSTB

Modular Stack Test Bench

SEM

Scanning electron microscope

SOFC

Solid oxide fuel cell

TPB

Triple phase boundary

YSZ

Yttria stabilized zirconia