Optimization and Scale-up of Solid Oxide Cell Manufacturing for Hydrogen Generation

Introduction The imminent energy crisis caused by the heavy dependence on fossil fuels has led to the search for a net zero-emission scenario that requires the development of reliable and efficient solutions for energy conversion and storage for the intermittent renewable energies. In this context, hydrogen has been identified as a potential energy carrier not only for energy storage, but also for the decarbonization of key industries and transport [1], [2]. There are various hydrogen production methods in different stages of development, being the water splitting process one of the most environmentally friendly, as it does not require fossil fuels as starting material. Thermochemical water splitting, photoelectrochemical methods and water electrolysis are among these techniques [3]. Water electrolysis is considered the cleanest technology when combined with renewable energies, producing the so-called green hydrogen. There are several electrolysis technologies with varying degrees of development, being the most common alkaline electrolysis (AEL), followed by proton exchange membrane electrolysis (PEMEL) and solid oxide electrolysis (SOEC) [3]. In the last decade, solid oxide cells (SOC) have generated great interest due to its high conversion efficiency, the possibility of working in fuel mode or electrolysis mode and the fuel flexibility (H 2 O+CO 2 co-electrolysis mode) [1], [2]. The operating temperature in SOC technology rises >650 ºC, favouring thermodynamics and kinetics of reactions, but demanding the use and development of cell and system materials, as well as designs, that stand the elevated temperatures and optimize the heat management. This is a point of improvement together with the scaling up of cell and stack manufacturing processes, to obtain stacks of higher power output that allow to obtain powerful electrolysis plants [4]. A dense electrolyte is required for a proper anion conduction in SOC, being yttria-doped zirconia (YSZ) the most used compound. The widely employed material for the hydrogen electrode is Ni used together with YSZ, getting the Ni-YSZ cermet [1]. For the air electrode, perovskite-type oxides are often used in the literature, such us lanthanum strontium manganite (LSM), combined with YSZ. Different fabrication methodologies have been used in SOC manufacturing, being the most standardized techniques tape casting and screen printing. Tape casting is used for thicker layers (20-250 µm), while screen printing is applied for thinner ceramic layers (2-50 µm). Both are processing methods that can be scaled up by adapting the production requirements [5]. During this work, planar anode supported cells were fabricated, employing tape casting for the hydrogen electrode and electrolyte and screen printing for the air electrode. The process for scaling-up each step is described, both in terms of volume produced as well as deposition method, making use of pre-industrial scale equipment. Methodology Cell fabrication In this work, a hydrogen electrode supported cell was prepared, being the thickest layer and acting as mechanical support of the cell. The hydrogen electrode is divided into a dense functional layer, which acts as the active part of the electrode, and a thick and a porous support layer, which acts as the mechanical support layer and facilitate the flow of gases. First, for the preparation of the cell components, the slurry for the different layers was prepared by ball milling a water solvent, Duramax D3005 (Dow Chemicals) dispersant and a 5M KOH basifying agent with powders: YSZ (Tosoh) for the electrolyte and YSZ and NiO (Vogler) for the hydrogen electrode. Subsequently, the obtained electrolyte and functional hydrogen slurries were mixed with the Duramax B1000 binder (Dow Chemicals) in a planetary mixer. For the porous support layer, the binder and cornstarch poreformer were added to the slurry to generate porosity, and this combination was firstly mixed in a helix stirrer prior planetary mixer. Subsequently, the defoaming program was optimized to allow a correct deairing. The electrolyte and hydrogen electrode layers were casted with a tape caster adjusting the gap to achieve the target thicknesses, being 20-30 µm for the electrolyte, 70-80 µm for the dense hydrogen electrode, and 175-185 µm for the porous hydrogen electrode. The electrolyte was firstly casted on a susbtrate film and dried. Then, the functional layer slurry was casted on top of it and left drying, obtaining a bilayer. Separately, the porous layer slurry was casted on a substrate film and dried. The casting speed was 1,8 m/min in electrolyte and 4,8 m/min in the functional and support layers. The drying temperature was 40 ºC in all cases. Subsequently, the green layers were detached and cut at the desired size. A bilayer and three support layers were calendered at 40 ºC to join them together to obtain the green half-cell. Finally, the green half-cell was heated up to 1500ºC necessary for sintering and densification of the ceramic layers. The air electrode layer was also divided into a functional layer and a support contact layer. The ink was prepared initially by mixing an ink vehicle with the powders LSM (Fuell cell materials) and YSZ in the ball milling in different ratios. Afterwards, these inks were further processed with a three-roll-mill. These inks were printed by screen printing with a target thickness of 20 µm in each layer (functional and support layer respectively) on the electrolyte of the sintered half-cell. The final cell was finally sintered up to 1100ºC. Up-scaling process The upscaling of cell manufacturing is of great interest in terms of simplicity, repeatability, versatility and especially cost-effectiveness. However, the transition to larger manufacturing scales requires optimization of both the process with respect to operating conditions and working dynamics. In this work, cell fabrication follows the standard protocol previously explained with some changes in the process, which can be divided into volume scale-up and process scale-up. Volume scaling refers to the increase in slurry volume production. When manufacturing at low volume scale, a ball milling machine that can hold one milling jar was used, whereas a high volume pre-industrial ball mill admitted up to 4 milling jars simultaneously. The filling proportions in each jug were maintained but, in this case, double quantity can be produced. Initially, the slurry was casted with a lab scale tape caster which was the same used for low volume production. The working parameters used for casting were the same as for low volume production. Once the produced volume had been increased, the up-scaling process was performed using a pre-industrial roll-to-roll tape caster (HaikuTech, Lab-Scale Tape caster, Model CAM-L252). This equipment continuously cast the slurry on top of the substrate, having a 3,5 m long casting capacity and 2 m long drying tunnel. The slurry produced within the previous step was used to cast in this equipment. For this purpose, the working conditions to obtain the desired thickness were optimized. Subsequently, the tapes obtained followed the same calendering and sintering protocol as in lower scale production but optimizing the process to reduce time and harness the tape dimensions. Characterization and test Volume scaling In a first step, viscosity measurements of the slurries were carried out to study their stability over time and as a tool for quality control of the process. These measurements were useful not only to characterize the stability over time but also to identify the different viscosity values between low and high-volume production. Viscosity values were taken immediately after the preparation of the slurry and after 1 and 2 hours at rest. For low and high production volume, the viscosity values were 33.0 ± 0.7 Pa.s and 5.7 ± 0.1 Pa. s for the electrolyte, 148.0 ± 14,1 mPa.s and 94.2 ± 3.5 mPa.s for the functional hydrogen layer and 188.0 ± 4.5 Pa.s and 89.8 ± 4.0 Pa.s, respectively. These results showed that, regardless of the production volume, the slurry viscosities remained stable over time (up to 2 hours). This conclusion was interesting because it indicates that the slurries can be used without requiring additional mixing steps over time and without affecting the properties of the tapes. This finding could be appreciated on tapes produced after keeping at rest the slurry during these period of times (0, 1 and 2 h), and no differences in appearance or thickness were observed. The stability of the slurries will allow large volumes to be casted over longer periods of time during process upscaling. Moreover, the results showed a decrease in viscosity values when produced at higher volume. Scaling in process