The Effect of Fuel Electrode Roughness On the Properties of Plasma Sprayed Solid Oxide Cells

Solid oxide cells (SOCs) fabricated on metal supports offer a number of advantages over electrode and electrolyte supported architectures. Plasma spraying, a type of thermal spray process, can be used to deposit SOC electrode and electrolyte components on porous metal supports [1]. A particularly challenging aspect of manufacturing SOCs with the plasma spraying technique is generating leak tight electrolytes. Leaky electrolytes lead to gas crossover and combustion, low open circuit voltages (OCVs) and electrochemical performance, and rapid metal support degradation. Previous work has shown that the electrode surface roughness prior to deposition of the electrolyte has a strong influence on electrolyte leak rates and OCVs [2]. Electrode surface asperities are known to be created by solidification of the feedstock particles too small to travel through the escaping plasma gases that are moving in-plane with respect to the metal support [3]. Electrode surfaces are often prepared for electrolyte deposition by manual sanding, a technique that inherently produces surfaces of variable quality and that can remove significant fractions of the deposited electrode. Here, we examine the use of a planar jet of compressed air aimed at the plasma plume to remove small feedstock particles prior to contact with the substrate. Ni-YSZ fuel electrodes were deposited on porous metal supports by suspension plasma spraying using a custom made air knife with varying compressed air supply pressures to remove small feedstock particles from the plasma plume. Resulting fuel electrode masses were measured, and the surfaces were characterized with digital stripe projection technique using a 3D surface scanner (LMI MikroCAD premium). Fuel electrode deposition rates decreased with increasing air knife supply pressures, as shown in Figure 1a. From the high fraction of material deposited with the air knife compared to the case where no air knife was used, it can be inferred that feedstock material with large particle sizes at the core of the plasma plume is not prevented from depositing, while some of the finer particles are indeed prevented from being incorporated into the coating. Thus, the resulting fuel electrode masses are comparable to those of fuel electrodes prepared by manual sanding. Here, improving the surface roughness is defined as reducing areal density, height, and depth of surface asperities and removing regions with steep gradients. The average surface roughness of fuel electrodes begin to improve with air knife supply pressures greater than approximately 2 bar. This result is evident in the arithmetic mean average and root mean square of the gradient ISO 25178 roughness parameters Sa, and Sdq, respectively, shown in Figure 1b. Surface roughness reduction is particularly evident when observing motif density shown in Figure 1c, confirming observations made in an optical microscope. Asperity peak heights were also reduced from approximately 30 µm to less than 20 µm, as depicted in Figure 1c. The effect of anode surface roughness on electrolyte microstructure and electrochemical performance of the corresponding SOCs was investigated and correlated to the mass and surface topography measurements. ACKNOWLEDGEMENTS The authors gratefully acknowledge the financial support of the Natural Science and Engineering Research Council of Canada (NSERC). REFERENCES Kesler, O., Cuglietta, M., Harris, J., Kuhn, J., Marr, M., Metcalfe, C. (2013) ECS Transactions, 57 (1), pp. 491-501. Marr, M., Kesler, O. (2012) Journal of Thermal Spray Technology, 21 (6), pp. 1334-1346. VanEvery, K., Krane, M.J.M., Trice, R.W., Wang, H., Porter, W., Besser, M., Sordelet, D., Ilavsky, J., Almer, J. (2011) Journal of Thermal Spray Technology, 20 (4), pp. 817-828. Figure 1: a) Mean anode mass relative to anodes sprayed without air knife. b) Mean areal surface roughness parameters for varying air knife pressures. c) Mean peak heights and motif densities for varying air knife pressures. Figure 1

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.