Durability of PEM Fuel Cells and the Relevance of Accelerated Stress Tests

Durability and cost have been identified by the US-DOE as the critical barriers to commercialization of polymer electrolyte membrane fuel cells (PEMFCs).¹The reduction of cost and improvement of durability are intimately connected as the development of low cost components invariably influences the durability of that component in an operating PEMFC. Reliable accelerated stress tests (ASTs) are required to rapidly evaluate the durability of components in a PEMFC and facilitate component materials development. The need for ASTs is apparent given the target lives for fuel cell systems: 5500 hours (~ 7.5 months) for automotive, and 60,000 hrs (~ 6.8 years) for stationary systems. In this presentation we will discuss various ASTs that have been proposed for catalyst layers and membranes in PEMFCs and will present data from ASTs, simulated drive cycles and field testing.

The ASTs recommended by the DOE fuel cell technical team were performed on commercial MEAs from W. L. Gore and Associates Inc., MEAs used by Ballard Power Systems in fuel cell bus stacks, and MEAs provided by Ion Power Inc. using various catalysts purchased from Tanaka Kikinzoku Kogyo.^2,3 The MEAs were also subjected to drive cycle testing using the U.S. DRIVE Fuel cell Tech Team recommended protocol.²In addition to this wet/dry drive cycle, MEAs were also subjected to a wet drive cycle where only the high RH portion of the protocol was repeated. Extensive in-situ electrochemical and ex-situ structural characterization was performed on the MEAs to track the degradation of components.

The electro-catalyst degradation observed in the wet/dry drive cycle test showed a direct correlation with the electro-catalyst AST performed on an identical catalyst. For example, the % ECSA loss observed in the wet/drive drive cycle and the 0.6 to 1.0V cycling AST is plotted in Fig. 1 and shows excellent agreement in terms of ECSA loss with potential cycles. Similar results were also obtained for the wet drive cycle testing where the ECSA loss was similar during the first 20,000 cycles. Post-mortem testing illustrated an increase in Pt crystallite size with decreasing ECSA consistent with field test data from buses. In addition to the Pt-growth, significant electrode thinning was also observed after long-term drive cycle testing especially in high surface area carbon (HSAC)-based MEAs. This is consistent with NDIR measurements showing evolved CO₂ from carbon corrosion of the cathode at voltages as low as 0.9V.

The drive cycle testing also initiated membrane degradation resulting in increased H₂ cross over and eventual failure of the cells. This is illustrated in Fig 2 where the crossover of various MEAs tested under both the wet and wet/dry drive cycles is plotted. The wet/dry drive cycle resulted in faster failure than the wet drive cycle and the non-stabilized (Sample A) membrane resulted in faster failure than the chemically and mechanically stabilized membrane (Sample B). Ex-situ characterization of beginning of life (BoL) and end of life (EoL) MEAs revealed little change in thickness of the membrane up to 2500 hours of the drive cycle testing. These results are consistent with the field data from buses where no membrane thinning was observed in the field at EoL and failure was due to cracks in the membrane mainly at the inlet and outlet of cells. A combined mechanical/chemical AST which involves RH cycling under OCV at 80^oC was able to reproduce the failure mode observed in the field data and will be studied further to serve as a membrane AST capable of predicting lifetimes.

References

1. R. Borup, et al., “Scientific Aspects of Polymer Electrolyte Fuel Cell Durability and Degradation”, Chemical Reviews, V. 107, No. 10,3904-3951 (2007).

2. Cell Component AST and polarization curve Protocols for PEM Fuel Cells (Electrocatalysts, Supports, Membranes and MEAs), Revised December 16, 2010.

3. N. L. Garland, T.G. Benjamin, J. P. Kopasz, ECS Trans., V. 11 No. 1, 923 (2007).

Acknowledgements

The authors wish to acknowledge the financial support of the Fuel Cell Technologies Program and the Technology Development Manager Nancy Garland. The authors also wish to acknowledge Ion Power, Inc. for supplying the MEAs used in this study.