Understanding Performance Decay in Solid Oxide Electrolysis Cells: Benchmarking and Analyzing a Model for Experiment-Driven Modeling

ECS Meeting Abstracts, vol. MA2025-03, pp. 404

Abstract

With the global push toward a sustainable hydrogen economy, Solid Oxide Electrolysis Cells (SOECs) have emerged as a promising technology for efficient hydrogen production. Especially their ability to utilize both electrical and thermal energy and the absence of scarce and expensive catalyst materials is a striking advantage for their integration into renewable energy systems or industrial processes. However, their long-term degradation significantly limits their operational reliability and scalability. Building on previous fundamental studies, this work presents advancements in establishing a simulation-based framework to understand degradation mechanisms in commercial, fuel-electrode-supported Ni-YSZ SOECs. Using the commercial COMSOL Multiphysics ® software, an integrated approach is developed that combines a finite element method (FEM)-based multiphysics model with experimental data. This study paves the way for a framework bridging critical gaps in linking microstructure evolution to performance degradation. Experimental performance data derived from i-V characterization and FIB-SEM tomography serve as the basis for model validation and refinement. The developed FEM model simulates coupled physicochemical processes within the SOEC, allowing for deeper insights into the interplay between microstructural parameters and cell performance metrics. Special attention is paid to microstructural changes during degradation, with key parameters such as porosity, tortuosity, and active surface area being explicitly resolved. The model attempts to elucidate critical aspects such as diffusion limitations and the pivotal role of Triple Phase Boundaries (TPBs) in electrochemical processes. Building on prior studies that examined the model’s general ability to describe degradation processes, [1] this study focuses on refining and optimizing the simulation framework for SOEC button cells. Therefore, the aim is to develop a robust framework for the simulation of SOEC button cells, based on which targeted investigations of microstructural influences will be possible. As a first step, the model was therefore subjected to thorough benchmarking, including rigorous analysis of mesh size and granularity, as well as the fraction of the cell domain considered. Essentially, the benchmarking of the model aimed at a balance between computational efficiency and accuracy. During the benchmarking procedure curves for various predefined and custom mesh settings were compared to identify the optimal mesh resolution beyond which further mesh refinement no longer detectably influenced the results. This analysis identified a mesh configuration with element sizes ranging from 4 µm to several hundred µm, yielding average element sizes that are slightly larger but still comparable to the Ni-YSZ cermet particle sizes reported in established literature. [2-6] Further mesh refinement only led to marginally improved accuracy, and thus demonstrated particle-level resolution to be unnecessary for accurate process depiction. The developed mesh configuration allows the resolution of degradation phenomena like Ni depletion, which according to Hauch et al. [4] as well as Sciazko et al. [5] occur in the electrolyte-adjacent 5 µm region, while avoiding the exponentially increasing computational cost of ultra-fine meshes. The identified configuration thus strikes an optimal balance between computational cost and accuracy, as coarser meshes, employing maximum element sizes larger than 200 µm resulted in significant accuracy losses. On the other hand, persistent convergence challenges occur under lean feed-gas conditions when reducing the maximum element size below 190 µm. Combined with particle-level considerations being demonstrated unnecessary for accurate modeling, further mesh refinement thus offers no significant benefit. Benchmarking across varying mesh settings underscores the importance of capturing diffusion and transport processes. While high-current regimes exhibit a strong dependency on the mesh resolution, at lower current densities mesh refinement had negligible effects on simulation outcomes. Further analysis explored the impact of reducing the cell domain considered on simulation accuracy and computational time. Simulation of half-cell domains showed no differences compared to full-cell considerations within experimental accuracy, while quarter-cell domains introduced deviations that rendered such approaches impractical. This highlights the feasibility of extrapolating to a full cell using smaller domains for computational efficiency without compromising accuracy, provided that the reductions remain within optimal bounds. Studies of different gas compositions and temperature ranges during the benchmarking ensured the coverage of a wide parameter space, extending the model’s applicability beyond static operating points. This allows the exploration of different conditions from experimental procedures - demonstrating the robustness of the model to explore and decipher experimentally hardly accessible phenomena. One such phenomenon is microstructure evolution during SOEC operation. Parameter variations revealed the profound impact of microstructural features, particularly TPBs, on cell performance. These results confirm the central role of TPBs in electrochemical reactions and are consistent with established literature findings. [7] Other key parameters, such as exchange current density and active surface area, were also found to be influential in determining cell performance and degradation. The applied parameter variations demonstrated that the model is able to account for any type of loss mechanism that may occur within experiments. These results position the FEM model as a robust tool not only for understanding degradation mechanisms but also for exploring strategies to mitigate them. Despite its strengths, the model highlights areas for improvement. In particular, comparative studies between simulation and experimental data reveal an incomplete consideration of diffusion phenomena, emphasizing the need for refinements in this area. However, preliminary results suggest that the parameterization of surface areas can partially replicate diffusion limitations, providing a pathway to refined diffusion modeling. Further progress will require more detailed experimental data on real-cell microstructures to better calibrate the model. In addition, comparison of experimental and simulation data will allow identification of bottleneck reactions where further refinement of the given framework is required. References Eyckeler, M. Nohl, J. Uecker, Multiphysics modeling of degradational behavior in high-temperature electrolysis cells based on experimental studies , Master Thesis, RWTH Aachen University, Aachen, 2024 Kawashima, S. Miyoshi, Y. Shibuta, S. Yamaguchi, J. Power Sources 234 (2013), 147–153 Keane, H. Fan, M. Han, P. Singh, Int. J. Hydrogen Energy 39 (2014), 33, 18718–18726 Hauch, K. Brodersen, M. Chen, M. B. Mogensen, Solid State Ion. 293 (2016), 27–36 Sciazko, T. Shimura, Y. Komatsu, N. Shikazono, JTST 16 (2021), 1, JTST0013-JTST0013 Wang, C. Wu, B. Zu, M. Han, et al., J. Power Sources 516 (2021), 230660

Authors 3

1. Niklas Eyckeler Aachen Institute of Physical Chemistry

   RWTH Aachen University · Forschungszentrum Jülich

   Affiliation as printed

   Institute of Energy Technologies (IET-1), Forschungszentrum Jülich

   Institute of Physical Chemistry, RWTH Aachen University

2. Till Frömling

   Forschungszentrum Jülich

   Affiliation as printed

   Inst. of Energy Technologies (IET-1), Forschungszentrum Jülich GmbH

3. Rüdiger‐A. Eichel Aachen Institute of Physical Chemistry

   RWTH Aachen University · Forschungszentrum Jülich

   Affiliation as printed

   Institute of Energy Technologies (IET-1), Forschungszentrum Jülich

   Institute of Physical Chemistry, RWTH Aachen University