Electrochemical performance and advanced 3D reconstruction of LSCF-GDC air electrodes for solid oxide electrolysis cells

The performance of solid oxide electrolyzer cell (SOEC) air electrodes depends in a complex way on various compositional and morphological features. This study provides a proof of concept for a knowledge-based AI-assisted design of SOEC air electrodes, which could circumvent the conventional time-consuming and cost-intensive workflow. Symmetrical cells with La _0.6Sr _0.4Co _0.2Fe _0.8O _3-δ (LSCF) − Ce _0.9Gd _0.1O _1.95 (GDC) air electrodes with different phase ratios showed a pronounced minimum in the polarization resistance at 50:50 wt%. Using AI-assisted segmentation of SEM images, the morphological features of the electrodes were extracted with an accuracy >96 %, allowing the peak performance of the 50:50 wt% electrode to be associated with several key morphological features. To validate the derived relationships, the two best electrode designs were transferred to full cells. As predicted, the 50:50 wt% LSCF-GDC electrode delivered an exceptionally high current density of −2.37 A cm ⁻² at 1.2 V and 800 °C, while the cell with the 70:30 wt% electrode exhibited a significantly lower current density. Finally, a 5 × 5 cm ² cell with 50:50 wt% LSCF-GDC electrode was tested for 220 h. The excellent performance demonstrates that AI-assisted image analysis is a powerful tool to accelerate and improve the development of SOEC electrodes and cells in the future.