Effect of heat treatment on membrane electrode assembly performance

Abstract

Polymer electrolyte membrane fuel cells (PMFCs) are promising power converters which have received attention in microelectronics, automotive and stationary applications. The heart of a PEMFC is its catalyst layers where hydrogen oxidation and oxygen reduction occur. The constitutive materials of the catalyst layers are the platinum catalyst dispersed on a carbon support and a proton conducting material (the ionomer). These solid materials interact and form active sites where electrochemical reactions take place. The interactions between the supported catalyst and the ionomer are initiated during the catalyst slurry preparation. The catalyst slurries are coated onto membranes to form membrane electrode assemblies (MEAs). The current catalyst slurry preparation method produces MEAs with benchmark performance yet requires a thermal treatment step which is time and energy consuming. Therefore, this study aimed to optimize the manufacturing time by determining the impact of thermal drying on the ionomer-supported catalyst interactions. Herein, the catalyst slurry preparation methods were varied as well as the catalyst support (Graphitized Vulcan and Vulcan) to determine how these variables impact the interactions and the resulting performance. Four methods of catalyst slurry preparation were used: thermal drying, no drying, freeze-drying and air-spray drying. The freeze-drying method isolated the impact of heat during catalyst slurry preparation and the air-spray drying method aimed to decrease the process time. Physical characterizations such as Raman spectroscopy was used to study the surface properties of the supported catalysts, scanning electron microscopy to determine the thicknesses of the catalyst layers, mercury intrusion porosimetry and atomic force microscopy to study the pore structure and the aggregates in the catalyst layers, respectively. MEA performance tests were done in-situ and consisted of current-potential polarization curve, cyclic voltammetry to calculate the electrochemical surface area, and electrochemical impedance spectroscopy to determine the charge transfer resistance. Accelerated stress tests to investigate carbon corrosion were conducted ex-situ in a half-MEA cell. The mechanism of ionomer attachment onto the supported catalyst was independent on the catalyst slurry preparation method and highly depended on the surface properties of the catalyst support. The hydrophilic side-chain of the ionomer showed preference to the platinum catalyst and the less hydrophobic support (Vulcan) and the hydrophobic backbone of the ionomer interacted with the more hydrophobic support (graphitized Vulcan (GV)). The pore structure of the catalyst layers and the ionomer distribution varied with the catalyst slurry preparation methods. Slow-drying rate methods such as thermal drying and freeze-drying methods produced thin GV40 (40% of Pt supported on graphitized Vulcan) catalyst layers with large intrusion volume pores and optimal ionomer distribution which improved the mass transport properties and increased ionomer-supported catalyst interfaces, respectively. These resulted in benchmark performance, larger electrochemical surface areas and lower charge transfer resistances. While the no drying and the air-spray drying methods produced thick GV40 catalyst layers with small ionomer-supported catalyst interfaces which decreased the protonic connectivity in the catalyst layers. These GV40 MEAs showed high potential losses, low electrochemical surface areas and high charge transfer resistances. Similar catalyst layer structure and electrochemical properties were observed between V40 (40% of Pt supported on Vulcan) MEAs prepared by the thermal drying and no drying methods, this was likely due to the surface properties of Vulcan which facilitated the interactions with the ionomer to form large interfaces. However, the air spray drying method resulted in small ionomer-supported catalyst interfaces which lowered the performance of the MEAs. It was concluded that optimal GV40 catalyst layer structure and performance were functions of the drying rate (and not heat) during the catalyst slurry preparation. And no drying step was required during V40 catalyst slurry preparation to achieve similar performance