Proton exchange membrane (PEM) fuel cells may emerge as a dominant clean energy source for heavy duty transport vehicles. Amongst other considerations, the lifetime performance of the fuel cell cathode catalyst layer is a key component of their future viability and a focus in the development of many proposed membrane electrode assemblies (MEAs). Typical targets of the lifetime of a PEMFC stack are on the order of thousands of hours of operation. However, running degradation tests spanning those long timescales is often impractical; and consequently, extensive work has been done developing and implementing accelerated stress tests (AST's) in order to screen proposed cathode catalyst materials for fuel cells.Much value is obtained by accurate and robust lifetime prediction models which are able bridge the gap between degradation from AST's in differential cells, and degradation of the catalyst layer in a fuel cell stack under realistic operating conditions. In other words, by parameterizing a model on short-time, subscale ASTs and extrapolating to a stack after tens of thousands of hours of operation. Commonly in such models, the electrochemically active surface area (ECSA) of platinum is used as the primary state-of-health parameter and predicted using either data-driven models or physics-based models. On one hand, data-driven models may be well suited to utilize large sets of stack data, but offer weaker extrapolation potential, conversely fitting physics-based models on global measurements of stack data is rarely a well-conditioned inverse problem.In this talk we present a scaling law on the long-time behavior of the ECSA obtained from analysis of governing physics equations from the literature. It is assumed that the primary loss mechanism is the coarsening of platinum particles, and the chief technique used is the method of averaging, which allows for the separation of two timescales: the short timescale of platinum oxide formation and dissolution, and the longer timescale over which the particle size distribution changes and a performance loss can be detected. The physics-derived asymptote for the ECSA evolution is then applied to several experiments and found to be in good agreement, allowing for a simple, streamlined lifetime prediction method for PEM fuel cells.
Read full abstract