Dynamic modeling of membrane swelling in fuel cell manufacturing
Silverman, Timothy J.
MetadataShow full item record
Fuel cells are promising energy conversion devices, but they have not been widely adopted because of their very high cost. The most expensive component of a fuel cell is the membrane electrode assembly, a polymer film coated with catalyst material. The catalyst layer is fabricated by depositing and drying a liquid ink on the membrane. The membrane can rapidly absorb water from the ink, causing swelling strain sufficient for wrinkling, which can cause defects in the finished product. These challenges limit most catalyst layer fabrication to individual preparation by hand. We seek to understand the swelling phenomenon in a way that enables the control of defects during mass production. Membrane swelling is a transient, three-dimensional, coupled mass transfer, heat transfer and solid mechanics problem. Existing models describe the membrane in fuel cell operating conditions, making use of simplifying assumptions that are not valid for predicting manufacturing defects. We present a new model incorporating effects that are missing from other models. We present simulation results for scenarios relevant to the control of defects. Simple spatial variations in water content can cause curl and wrinkling; we establish criteria for the formation of these defects by simulating the membrane's response when subjected to the full pre-swollen coating and drying process. We investigate the sensitivity of wrinkling to nonuniformity in the coating and to processing conditions in the coating line. We propose a rationale for controlling wrinkling caused by these effects and for diagnosing coating defects using the membrane's wrinkling response. We show how the membrane behaves differently depending on whether the coating is applied to one side or to both sides simultaneously. We have designed and constructed a machine to pre-swell the membrane, apply a coating and then dry the coating under controlled tension, speed, temperature and humidity. We present the design and discuss how the machine may be used, together with the membrane model, to predict and control defects in catalyst-coated membranes.