Statistical design of experiments for efficient performance characterization of protonic-ceramic electrolysis cells

Abstract

Obtaining a cohesive understanding of performance in protonic-ceramic electrolysis cells is difficult due to the wide operating space coupled with low-throughput diagnostic techniques, sluggish system dynamics, and cell degradation. In this work, design of experiments (DOEs) methods are implemented to provide an efficient framework for understanding the phenomena that most strongly dictate cell performance. In addition to a more robust description of cell-level phenomena, mathematical equations are generated that accurately describe the complex relationship between the cell operating variables and cell performance metrics such as faradaic efficiency, cell potential, resistances, and energy conversion efficiency. Here, DOE is realized without the need to pre-select the most important operating variables based on a priori rationalizations. This is particularly valuable for system-level and technoeconomic analyses where the accurate prediction of cell/stack response over many operating conditions is required. The demonstrated experimental framework consists of a screening design and subsequent optimization design. The Plackett–Burman factor-screening design identifies temperature, current density, and steam content as having the largest impacts on cell performance, particularly faradaic efficiency. Increasing the electrolytic current density from 0.2 to 0.5 A cm ^−2 decreases polarization resistances by 74% due in large part to a negative-capacitance element that dominates at low frequency and high electrolysis bias. Impedance data highlights the connection of this negative feature to electronic leakage through the electrolyte and gas diffusion limitations. Additionally, increasing cell temperature from 500 to 600 ^∘ C is shown to decrease faradaic efficiency by 9% due to electrolyte dehydration and oxygen incorporation at high temperatures. The Box–Behnken optimization design then enables generation of regression equations to be used in response surfaces for data visualization and cohesive, multivariate analysis of cell operation.

Keywords

• proton-conducting ceramic
• electrolysis
• design of experiments
• faradaic efficiency