Modeling the Flow Behavior of Catalyst Inks for PEM Fuel Cells by an Evolutionary Algorithm
The catalyst coated membrane represents the heart of a polymer electrolyte fuel cell (PEMFC) and is produced by directly applying a catalyst suspension on the membrane or a decal transfer foil. Understanding the rheology of catalyst inks is crucial to optimize the printed microstructure of the catalyst layer and hence the performance of the fuel cell. The suspension consists of platinum on carbon powder, ionomer dispersion (perfluorosulfonic acid ionomer) and solvents. Its rheology is directly affected by various material properties of all ink components. Therefore, measuring the flow curves represents a suitable tool for quality control at the beginning of the production line. In this study, we investigate the influence of carbon to solvent ratios with and without ionomer as well as different types of ionomers (equivalent weight variations) on the inks flow behavior. Moreover, we compare low surface carbon (VulcanXC72R) with high surface carbon (Ketjenblack EC300J) for all ink variations. A rotational rheometer with TwinDrive technology is used to analyze a total of 20 different ink recipes. Four measuring profiles are applied to characterize the dynamic viscosity and yield stress in rotational mode and the viscoelastic behavior by amplitude sweep. Further, three interval thixotropy tests are conducted to investigate the transient flow behavior during oscillation. The collected data is fitted by an evolutionary algorithm to common models for Non-Newtonian flow, like Power Law, Cross Power Law, Bird-Carreau and Herschel-Bulkley models. Thereby, determined parameters are then linked to the ink component and their material properties. The resulting correlations and database represent the first step towards designing desired flow behaviors for coating processes, hence optimizing the production process, as well as predicting the ink microstructure by only knowing material properties and proportions of the catalyst ink ingredients.