Optimising cell design for in-situ/operando computed tomography of PEM fuel cells and electrolysers

EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476265 B0602 Page 1/9 B0602 Optimising cell design for in-situ/operando computed tomography of PEM fuel cells and electrolysers Olivia Linley, Jennifer Johnstone-Hack* School of Chemical, Materials and Biological Engineering, University of Sheffield, Sheffield/UK; *Contact corresponding authors: www.EFCF.com/ContactRequest Abstract In-situ and operando computed tomography (CT), using both neutrons and X-rays, are increasingly popular methods for characterising transport, degradation and materials in both polymer electrolyte membrane (PEM) fuel cells and electrolysers. The use of neutrons and X-rays as imaging sources are highly complementary, enabling imaging across length and time-scales. Where neutrons are highly suited to studying water distribution in the flow channels and transport layers, X-rays are well suited for studying morphology evolution and architecture in the membrane electrode assembly (MEA) components. To achieve the highest quality data, cell design is of crucial importance. The design must balance the need for good electrochemical performance, with achieving high-resolution imaging datasets. This work will discuss the evolution of cell designs suitable for X-ray (Figure 1a) and neutron (Figure 1b) CT, from “Generation 1” to “Generation 2” cells. As shown in Figure1c and d, small modifications to the cell design can result in marked performance improvements, although materials choice clearly plays a significant role in Ohmic resistance in the cell. EIS was performed across the neutron and X-ray cells, with the resistance across the graphite X-ray cell (229 mΩ) around 50× greater than the aluminium neutron cell (4 mΩ). Finally, this work will discuss recent “Generation 3” cells, that are intended to be reversible for operation in both fuel cell and electrolyser mode. With a strive towards open access research, this work also discusses efforts to make these cell designs available through open data repositories. This ensures that both new and established CT users can work together to build best-practices around imaging of these critical technologies. Figure 1 a) Graphite “X-ray” cell and b) gold-coated aluminium “neutron” cell, showing the imaging and compression regions in both; c) Corresponding Gen 1 and Gen 2 X-ray and d) Gen 1 and Gen 2 neutron cell performance. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476265 B0602 Page 2/9 Introduction Electrolysers for green hydrogen production and fuel cells for zero-carbon energy conversion are expected to be rapidly scaled-up over the next decades [1]. In order to meet these ambitious global scale-up targets, whilst remaining mindful of the need to effectively utilize critical materials resources like platinum, iridium and titanium, significant innovation is required. Polymer electrolyte membrane (PEM)-based technologies are currently the most widely developed and commercialized technologies, thanks to their fast response times and high performance [2]. However, there is still an ongoing need to discover new materials and components to lower their cost, as well as investigate the degradation mechanisms that cause the components to fail. In recent years, imaging techniques have become increasingly popular for studying the three-dimensional morphology inside PEM electrolysers (PEMWEs) and PEM fuel cells (PEMFCs). Because the reactions and transport of species occur at hidden interfaces between the components in these devices, methods like tomography that can effectively ‘see-inside’ PEMWEs and PEMFCs are highly attractive. Furthermore, the ability to conduct in-situ and operando imaging experiments, with the ever-increasing imaging capabilities of both lab-based and national facility sources, has allowed for the advanced study of interfacial evolution during operation [3] and failure [4]. Because electrochemical data can be collected simultaneously, it is also possible to correlate changes in electrochemical performance to the materials morphology within the PEM components. Neutrons and X-rays are widely considered as complementary imaging methods for studying PEMWEs and PEMFCs [5]. Where neutrons are well suited to studying water evolution and gas distribution in the flow field and porous transport layer (PTL)/gas diffusion layer (GDL) components [6,7], the lower flux of neutron sources limits their spatial and temporal resolution. Conversely, X-rays have high spatial and temporal resolution (and high attenuation of high-Z elements) making them well suited for studying changes in the catalyst layer (CL) during operation and degradation [8,9]. Thus, when used together in a complementary way it is possible to build up a full picture of how new materials and component designs influence PEMWE/FC electrochemical performance and degradation. An essential consideration for achieving the highest-quality imaging datasets is the cell design. Whilst a range of designs, with varying orientations, materials and sizes, exist in the literature [3-9], the interplay between cell design and experimental conditions is less often discussed. Because of the often limited space or field-of-view (FOV) of imaging experiments, cell design (and catalyst coated membrane (CCM) active area) is often a balance between achieving good electrochemical performance and high-image quality. Whilst some work has demonstrated the ability to conduct simultaneous X-ray and neutron imaging [10], most cell designs are tailored to suit the source properties, i.e. either neutrons [3, 6, 7] or X-rays [4, 8, 9]. The field of cell design for imaging experiments hence presents a rich opportunity for developing universal open-access designs, with performance similar to lab-based cells, that can move towards multi-modal imaging experiments. 1. Scientific Approach With the need to optimize cell design for imaging in mind, the overall aim of this work was to understand how cell design can influence cell performance and imaging quality, using PEMFCs as a case study, then use this experience to design a new, open-access universal cell design that can be used for both X-ray and neutron computed tomography (CT). EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476265 B0602 Page 3/9 First, two generations of PEMFC cells were designed: one suitable for X-ray imaging and one suitable for neutron imaging. Electrochemical data was collected on both Generation 1 and Generation 2 designs to evaluate how changing the cell design influences performance. The two Generation 2 designs were then used in two case-study experiments. The X-ray design was used to study the effect of degradation on the PEMFC CL along the length of the flow channel [9]. Identical location X-ray CT was used on three different locations along the channel and the rate of degradation was compared. The neutron design was used to demonstrate the first example of high-speed operando neutron tomography experiments [7, 11], where the volume of water evolved in the cathode flow channel, anode flow channel and membrane electrode assemble (MEA) were studied. When analyzed together, it is possible to correlate how inhomogeneities in water evolution and distribution can influence and affect end-of-life degradation in the cell. Finally, this work discusses more recent developments in cell design, which target a ‘multimodal’ cell design. The aim of this work is to design a modular cell, that can function as either a PEMFC or PEMWE by interchangeable parts, and that is suitable for both X-ray and neutron CT experiments. This would allow for truly correlative imaging of microscale morphology changes occurring in the CL to the macroscale design and morphology of the PTL/GDL and flow channel design. Lastly, the efforts to make cell designs open access will be discussed, to help increase discussion and collaboration surrounding cell designs for imaging and making designs available for the community to use. 2. Experiments 2.1 Individual cell design for X-ray and neutron imaging separately Two generations of cell designs for both X-ray and neutron imaging of PEMFCs were designed and manufactured. Each imaging source led to different design requirements for the cell, where the X-ray cell was manufactured from graphite to ensure good transmission of X-rays. The neutron cells were manufactured from gold-coated aluminium, since aluminium is a poor attenuator of neutrons. The two generations of cell designs for the Xray and neutron design are shown in Figure 2. Figure 2 a) Generation 1 and b) Generation 2 X-ray designs manufactured from graphite; c) Generation 1 and d) Generation 2 neutron designs manufactured from gold-coated aluminium. It can be seen that both Generation 1 cells have a deep single channel (Figure 2a, c), with an ‘imaging region’ in the centre of the cell, where the amount of cell material has been reduced, and a ‘compression’ region, where clamps that are outside the imaging FOV provide compression to the cell. Flexible tubing was attached to the nozzles on each design for hydrogen and air gas supply on the anode and cathode, respectively. Developments to the Generation 2 cells include the incorporation of serpentine flow channels, to better EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476265 B0602 Page 4/9 represent real-world cell flow channel designs, and holes through the compression region to improve the method for compression and closing the cell and to provide better contact between the plates. All four cells were tested on a custom-built rig in the research lab, with mass flow controllers to control hydrogen and air flowrates (20 mL min-1 and 100 mL min-1, respectively) and a Gamry Interface 5000 potentiostat (Gamry, USA) for electrochemical control. MEAs were prepared with electrode areas of 1.7 cm2 for the Generation 1 cells and 1.6 cm2 (X-ray) and 2 cm2 (neutron) for the Generation 2 cells. HyPlat GDEs (HyPlat, South Africa) with a loading of 0.4 mg cm-2 on both anode and cathode were hot-pressed onto a Gore membrane (Gore, USA) to prepare the MEAs. Polarisation curves and electrochemical impedance spectroscopy (EIS) measurements (at 100 mA cm-2) were collected on all four cells. 2.2 Application of the in-situ X-ray CT cell The Generation 2 X-ray cell was used to conduct in-situ X-ray CT on three regions of the serpentine flow channel, namely the cathode gas inlet, a middle region and the cathode gas outlet. The results of this study have been published, and full experimental details can be found in the publication [9] but are briefly discussed here to highlight the capability of in-situ X-ray imaging for studying PEMFCs. A Zeiss Xradia 620 Versa instrument was used to conduct X-ray CT imaging. After assembly, the inlet, middle and outlet regions were initially imaged using 4X optics, 3 s exposure time and 9 W beam power. Following this, a carbon-specific AST was conducted, where the voltage was cycled between 1 and 1.5 V at 500 mV s-1 for 5000 cycles. After 2000 and 5000 cycles, the AST was interrupted, a polarisation curve was collected, and the identical location at inlet, middle and outlet were imaged again. Thus, nine datasets were collected in total at 0, 2000 and 5000 cycles. Post-processing was done using Avizo Fire (Thermo Fisher Scientific, USA) software. Datasets of identical locations were automatically aligned, then segmentation was carried out. Local thickness of both the CL and the cracks in the CL was conducted, so that the CL thinning and crack expansion could be quantified. Various other metrics were quantified, such as the cathode and anode crack ratios (CCR and ACR, respectively) and the crack connectivity. 2.3 Application of the operando neutron cell The Generation 2 neutron cell was used to conduct the first high-speed operando neutron tomography experiments to visualise water evolution in the PEMFC in 4D [7]. As with the Xray CT experiment in Section 2.2, full experimental details can be found in the paper. Two sets of experiments were conducted: the preliminary experiments on the CONRAD beamline at the Helmholtz-Zentrum Berlin, Germany [7], and a set of further experiments investigating different flow field designs using the NeXT instrument at the Institut LaueLangevin, France [11]. In both cases, the custom-built rig described in Section 2.1 was used for experiments. The cell was held at varying current densities between 100 and 700 mA cm-2 for either 10 minutes [7] or 1 hour [11] depending on the experiment. Tomograms were collected every 36 s [7], and the incorporation of slip rings into the experimental setup allowed for a tomogram every 18 s in the second set of experiments [11]. Post-processing of tomography datasets was done with a mixture of Python code (to quantify the volume of water in the cathode flow field, MEA and anode flow field) and using Avizo Fire software (to visualise the distribution and evolution of water in the flow channels). 2.4 Towards a combined cell and open data EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476265 B0602 Page 5/9 Whilst the two Generation 2 cells designs are well suited for imaging using separate X-ray and neutron methods, due to design limitations, it is not possible to use either cell for both methods (i.e. multimodal, correlative imaging using X-rays first and then neutrons). Furthermore, developments at beamline facilities, such as the concurrent X-ray and neutron capabilities of the NeXT instrument at ILL [12], mean that it is now possible to conduct simultaneous X-ray and neutron imaging [10]. This gives the motivation to design one cell that could be used for both experiments. Autodesk Inventor 3D CAD software (Autodesk, USA) was used to design and build the multimodal cell files. The next steps will be to get the cell manufactured and test the performance both electrochemically and from an imaging perspective. Furthermore, as mentioned in the Introduction, there are a wide range of cell designs, geometries and materials used in the literature, but sharing of these designs is often limited. The past and current generations of designs will be made available via an open access repository [13], along with a bill of materials and notes on the design. 3. Results 3.1 Cell design generations and electrochemical performance Results of polarization curve (Figure 3a,c) and EIS (Figure 3b, d) characterisation are shown in Figure 3 for the two Generations of the X-ray and neutron design. It can be seen that in both cases, the performance of the Generation 2 cell is significantly higher than the Generation 1 cell. This is more pronounced for the graphite X-ray design, where the limiting current density at 0.3 V more than doubles from 432 mA cm-2 to 908 mA cm-2. For the aluminium neutron design, the limiting current density increased by nearly a third from 600 mA cm-2 to 782 mA cm-2. The most significant improvement can be seen in the Ohmic region (above 50 mA cm-2), indicating that the improvement in contact due to the serpentine flow field and addition of compression screws has reduced the cell resistance. This was further confirmed by the EIS measurements. The Nyquist plots of EIS measured at 100 mA cm-2 for both the X-ray and neutron designs show a decrease in the high-frequency resistance (HFR) where the curves cross the X-axis. This is accompanied by a decrease in the charge-transfer arc between Generation 1 and Generation 2 designs for both X-ray and neutron cells. The EIS data confirms that the improved clamping mechanism and the change in the flow field design contributed to the improved cell performance between generations. Furthermore, the Generation 1 designs had a deep (3.6 mm) channel, meaning that a large amount of reactant gas would flow through the cell without reaction. Thus, the Generation 2 designs also constitute an improved utilization of the reactant gases. EIS measurements were collected on the ‘empty’ Generation 2 cells, i.e. cells without MEA and gasket where the two end plates were in direct contact (Figure 3e). This was to investigate the extent to which the cell itself (discounting any variation in MEA performance) contributed to the overall cell performance. It was found that the aluminium neutron cell has a significantly lower resistance than the graphite design, with calculated values of 4 ± 1 mΩ and 229 ± 1 mΩ for the aluminium and graphite cells, respectively. Clearly, the poorer conductivity of graphite compared with the gold-coated aluminium metal results in larger resistances in the graphite cell. This indicates that future cell designs should aim to minimize the amount of graphite in the cell, and instead aim to manufacture the cell from a different X-ray and neutron-transparent material (such as PTFE) where possible, with only a thin layer of graphite for conductivity. This will be discussed further in Section 3.4. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476265 B0602 Page 6/9 Figure 3 a) polarisation curves and b) EIS spectra of the Generation 1 and 2 graphite X-ray cells; c) polarisation curves and d) EIS spectra of the Generation 1 and 2 aluminium neutron cells; e) EIS of the empty cells. 3.2 In-situ X-ray CT along the flow channel during degradation The graphite X-ray cell was used to study the differences in degradation rates along the serpentine flow channel [9]. Whereas electrochemical data can give us global information about the overall cell performance, in-situ X-ray CT allows us to spatially resolve the effect of degradation on different regions of the cell. Figure 4 shows the fully segmented tomograms of the ‘cracks’ (Figure 4a-i), which represent the empty space between regions of the CL where material has been degraded and lost as a result of the AST. Figure 4 a-i) Volume renderings of the cathode crack evolution from 0-5000 cycles from inlet, middle and outlet region, j) cathode and k) anode crack ratios showing the extend of cracking is greater at the outlet than inlet. Figure adapted from [9] It can be seen that for all three regions of the serpentine, the extent of cracking increases with increasing cycle number. This is a result of the carbon-specific AST that leads to EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476265 B0602 Page 7/9 collapse of the pore structure in the CL, crack growth and widening, and crack network evolution. What was most notable was that the rate of degradation was significantly greater at the cathode gas outlet than inlet. This was quantified by calculating the ratio of crack volume fraction to crack+CL volume fraction (Figure 4j). It can be seen that in the starting material, the volume of cracks in all three regions is similar, but after 2000 and 5000 cycles, the ratio of cracks at the outlet is greatly increased. It is thought that this is due in part to starvation of reactant gas at the outlet leading to larger polarization and an accelerated rate of degradation. The ratio of cracks on the anode side (Figure 4k) remained constant throughout cycling, indicating that the AST was only effecting the cathode (as intended). Overall, these results highlight the importance of conducting multi-location tomography in both PEMFCs and PEMWEs, since effects such as the inhomogeneity of degradation could be missed by only investigating one region. 3.3 Operando neutron CT of water evolution and distribution The gold-coated aluminium neutron cell was used to conduct high-speed 4D tomography of water evolution in the flow channels of PEMFCs during galvanostatic holds [7]. Each tomogram represented 36 s of operation, and a set of galvanostatic holds were conducted for 10 minutes each. The results of visualization of the water evolved in the anode (red) and cathode (blue) flow channels, as well as in the MEA (green) during the 400 mA cm-2 hold are shown in Figure 5a. It can be seen that there is significantly more water evolved in the cathode, which is expected, but there has been some back diffusion due to flooding into the anode flow channel. As has been reported before [14], the water preferentially pools at the bends of the serpentine before filling the straights. Furthermore, on closer inspection of the water droplets themselves, it was found that in some cases a ‘micro-channel’ was formed when the serpentine became fully flooded [11], meaning that gas could still flow along the length of the channel, and water wetted the outer surface of the channel. Figure 5 a) Volume rendering of water in the anode, MEA and cathode, b) volume of water measured in the cathode and c) volume of water measured in the anode between 100-700 mA cm-2. Figure adapted from [7]. It was also possible to quantify the volume of water residing in each of the flow channels at 36 s intervals for the 10 minute hold (Figure 5b, c for cathode and anode respectively). As expected, with increasing current density the volume of water filling the channel increases. In the case of the anode, it was found that at the higher current densities, the water volume began to drop after ~400 s, indicating that after the initial start-up of the cell after applying the current, an equilibrium is being reached where water can be effectively removed from the cell. In summary, these results highlight that advances in cell design, along with improvements in instrument capabilities, have enabled operando tomography to study water EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476265 B0602 Page 8/9 evolution in PEMFCs. There is still scope for further cell optimization, both to enable a larger active area to be studied, as well as attempting to study 4D water/gas dynamics in PEMWEs. 3.4 Preliminary results of a multimodal cell and open data The first design prototype for a multimodal cell capable of use in both X-ray and neutron imaging experiments is shown in Figure 6. It comprises of two layers that will include the incorporation of O-rings to help with sealing. The conductive ‘flow-field plate’ layer can be made of either graphite or titanium, depending whether the investigation being conducted is on a PEMFC or PEMWE. The goal of the design was to keep this layer as thin as possible, to help optimize transmission of neutrons and X-rays. The outer ‘casing’ layer would be manufactured from a non-hydrogen-containing polymer material, such as PTFE. This would ensure that the casing did not attenuate neutrons, which would be the case were a hydrogen-containing polymer material to be used. Dimensions have been kept similar to the Generation 2 graphite and aluminium designs discussed in previous sections, since they represent a good balance between performance and imaging properties. The imaging region has been made cylindrical to help achieve a near-continuous path length through the casing material, which is particularly important for lab-based X-ray experiments where the beam flux is much lower. The next steps are to get the design manufactured and tested, to compare the performance of the cell to a 5 cm2 lab-based cell, and to test the imaging properties of both neutron and X-ray imaging. Finally, the designs have been made available via and open access data repository [13], which will be continually updated as the project progresses. The goal of this is to stimulate sharing and discussion in the PEMFC and PEMWE imaging community as to how to optimize and improve cell design, and ultimately aim to generate a ‘universal’ design. This would ensure that future studies on new materials or processes would have a common baseline, thus improving repeatability and representivity of results. Figure 6 Concept design for multimodal cell, including two components: a graphite or titanium flow field plate and an X-ray and neutron transparent PFTE casing. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476265 B0602 Page 9/9 References [1] IRENA, IRENA World Energy Transitions Outlook 2023: 1.5 C Pathway, Abu Dhabi, 2023. [2] H Becker, J Murawski, DV Shinde, IEL Stephens, G Hinds, G Smith, Sustainable Energy Fuels, 7, 2023, 1565-1603. [3] M Maier, J Dodwell, R Ziesche, C Tan, T Heenan, J Majasan, N Kardjilov, H Markötter, I Manke, L Castanheira, G Hinds, PR Shearing, DJL Brett, Journal of Power Sources, 455, 2020, 227968. [4] RT White, A Wu, M Najm, FP Orfino, M Dutta, E Kjeang, Journal of Power Sources, 350, 2017, 94-102. [5] SJ Altus, BJ Inkson, J Hack, Journal of Materials Chemistry A, 12, 2024, 23364-23391. [6] B Zhao, C Lee, JK Lee, KF Fahy, JM LaManna, E Baltic, DL Jacobson, DS Hussey, A Bazylak, Cell Reports Physical Science, 2, 2021, 100580. [7] RF Ziesche, J Hack, L Rasha, M Maier, C Tan, TMM Heenan, H Markötter, N Kardjilov, I Manke, W Kockelmann, DJL Brett, PR Shearing, Nature Communications, 13, 2022, 1-11. [8] E Leonard, AD Shum, N Danilovic, C Capuano, KE Ayers, LM Pant, AZ Weber, X Xiao, DY Parkinson, IV Zenyuk, Sustainable Energy and Fuels, 4, 2020, 921-931. [9] J Hack, L Rasha, PL Cullen, JJ Bailey, TP Neville, PR Shearing, NP Brandon, DJL Brett, Electrochimica Acta, 352, 2020, 136464. [10] P Shrestha, JM Lamanna, KF Fahy, P Kim, C Lee, JK Lee, E Baltic, DL Jacobson, DS Hussey, A Bazylak, Science Advances, 9, 2023, eadg8634. [11] J Hack, RF Ziesche, M Fransson, T Suter, L Helfen, C Couture, N Kardjilov, A Tengattini, P Shearing, D Brett, Journal of Physics: Energy, 6, 2024, 025021. [12] A Tengattini, N Lenoir, E Andò, B Giroud, D Atkins, J Beaucour, G Viggiani, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 968, 2020, 163939. [13] Johnstone-Hack, Jennifer (2025). STEP CAD files for the two components. The graphite (or titanium) inner flow field plate and the PFA (or PTFE) outer compression plate. The University of Sheffield. Dataset. https://orda.shef.ac.uk/articles/dataset/_/29087768. [14] Y Wu, JIS Cho, TP Neville, Q Meyer, R Ziesche, P Boillat, M Cochet, PR Shearing, DJL Brett, Journal of Power Sources, 399, 2018, 254-263. Keywords: EFCF2025, H2, Low-Temp. Fuel Cells & Electrolysers, Correlative Imaging, Xray Computed Tomography, Neutron Computed Tomography, Operando Cell Design, PEM Fuel Cells & Electrolysers. Remark: This work is licensed under Creative Commons Attribution 4.0 International