André Weber

Obtaining a detailed insight into the internal states of a solid oxide cell (SOC) stack via experimental methods is challenging due to the multitude of processes occurring across different scales. Physics-based models offer a complementary alternative, providing access to the desired information, by coupling operating conditions with relevant physical effects on the various spatial and temporal scales. To guarantee the trustworthiness of the model predictions adequate experimental parametrization is mandatory.In this work a spatially-resolved three-dimensional (3D) SOC stack model is developed, accompanied by an extensive experimental characterization of a commercial state-of-the-art cell. Physical consistency and reliability of the model results are ensured through experimental validation on industrial-size cells in a custom-developed short stack.The developed 3D stack model captures macroscopic distributions across the stack as well as microscopic, localized effects. This is achieved by coupling a homogenized full stack with fully resolvable regions. The coupling substantially reduces total computational time 15-fold and increases computable stack size 6-fold, while enabling the detailed analysis of critical regions. For a non-uniform gas supply to the stack only minor differences in global stack performance compared to a uniformly supplied stack are found. However, assessment of local variables (e.g., cell voltages, current density, overpotentials and molar fractions) in the fully resolved regions shows significant deviations between the cases. This investigation demonstrates the importance of evaluating global, but also local operation conditions when defining safe, economically viable and lifetime-compatible stack operating conditions.

A pulse-fitting methodology for time-domain analysis of lithium-ion battery impedance is presented and systematically extended to address automotive-relevant measurement constraints under controlled test conditions, including non-ideal current profiles, limited stabilization times, measurement noise, low sampling frequencies, and SOC-dependent OCV contributions. Since no universally applicable test pulse exists for large-format automotive lithium-ion cells, pulse-based diagnostics require evaluation methods that remain robust under application-specific excitation conditions. In this work, a pulse-fitting methodology for time-domain analysis of lithium-ion battery impedance is systematically extended. The approach models the voltage response to current pulses using a physically motivated RC network, enabling the extraction of diffusion-related time constants without relying on frequency-domain techniques. Key enhancements include numerical convolution for non-ideal current profiles, incorporation of a dynamic open-circuit voltage, Tikhonov regularisation, and weighting and scaling strategies to improve robustness against noise and low sampling frequencies. Furthermore, a quadratic programming solver is introduced to increase stability for ill-conditioned problems, and an extrapolation method is proposed to account for insufficient voltage stabilization times. The methodology is validated using both synthetic and experimental pulse data, demonstrating reliable access to low-frequency impedance characteristics. The results highlight the potential of the approach for advanced state estimation and model development.

For concentrated lithium-ion battery electrolytes, ion association and ion crowding cause the ionic conductivity σ to deviate from the classical Walden rule, and existing empirical equations require dense grid-based measurement campaigns to capture these deviations. We derive a density-corrected Walden (DCW) equation that introduces the solution density ρ as a second transport descriptor alongside the dynamic viscosity η ; the model contains two free parameters and is calibratable by linear regression. A complementary mass-action model, based on the classical ion-association equilibrium, separates ion pairing from ionic mobility and yields a physically interpretable dissociation profile. Both models are validated on LiPF ₆ in EC:DMC:EMC spanning 0.25 mol L ⁻¹ to 3.00 mol L ⁻¹ and −20 to 45 °C. Using only three calibration concentrations, the DCW model reconstructs the full σ ( c , T ) surface with a root-mean-square error of 0.27 mS cm ⁻¹ , approaching the accuracy of established benchmark Eqs. that require 5 to 10 calibration concentrations. Cross-validation on LiFSI in three pure solvents confirms transferability to other salt–solvent systems. The deterministic σ – η – ρ link additionally enables real-time quality control of conductivity measurements.

Concentration-related changes of the Nernstian voltage inevitably contribute to the resistance observed during DC operation of solid oxide cells. In electrochemical impedance spectroscopy (EIS), concentration losses arising from gas diffusion and gas conversion often overlap with electrochemical polarization processes, which is particularly true for high-capacity mixed-ionic–electronic conducting electrodes, complicating the interpretation of impedance spectra and extracted activation energies. This work presents a comprehensive in-operando methodology to deconvolute gas diffusion and gas conversion resistances in electrolyte supported cells with screenprinted electrodes. Gas diffusion losses are quantified using an extended inert-diluent variation approach, enabling accurate determination of an effective microstructure parameter and prediction of diffusion resistances at arbitrary operating conditions. An alternative method based on replacing hydrogen/steam with carbon monoxide/carbon dioxide is introduced to access the same parameter. Furthermore, a novel approach for determining gas conversion resistances is presented, avoiding reliance on idealized analytical models, and enabling accurate prediction of concentration losses over a wide range of gas compositions and temperatures. The combined prediction of diffusion and conversion losses accurately reproduces the total concentration impedance over a broad steam partial pressure range and outperforms existing analytical approaches. The methodology is demonstrated on a commercial full cell, containing a Ni/GDC fuel electrode, allowing reliable extraction of intrinsic activation energies even in the presence of increased, setup-induced concentration losses.

The accumulation of hydrogen and oxygen within the flow channels of proton exchange membrane water electrolysis cells results in increased gas fractions and pronounced two-phase flow effects at higher current densities. Spatially resolved segmentation along the channel permits investigation of such gradients but necessitates complex hardware. In this study, a simplified method is introduced to emulate local gas production in an incremental cell through controlled injection of well-defined amounts of hydrogen and oxygen into the respective inlet streams. The impact of the gas-to-water ratio on ohmic resistance and polarization phenomena is examined using electrochemical impedance spectroscopy combined with subsequent distribution of relaxation times analysis. This reveals membrane dehydration effects as well as an impact on activation-related losses on the cathode side. A detailed comparison between an incremental (zero gradient) cell and an along-the-channel cell demonstrates a good overall performance correlation. In the medium current density region (2-3 A cm-2), the along-the-channel cell shows significantly increased polarization losses, which might be attributed to contamination from ion release from the CCM and/or mass transport issues.