The durability of proton exchange membrane fuel cell stacks is a key limiting factor in transportation application due to severe conditions, where startup-shutdown and local fuel starvation most severely threaten the structure stability. In this paper, potentiostatic test at the average of 1.4 V is conducted to accelerate the carbon corrosion and structure damage in the cathode in a seven-cell commercial stack. A complete diagnosis framework based on micro-current excitation is established to systematically and in-situ evaluate the state-of-health of the cathode and the membrane, the mass transfer state, and the cell inconsistency in stack-level. Afterwards, the structure damage of cathode catalyst layer is characterized with scanning electron microscopy and energy disperse spectroscopy to provide direct evidences. In the early stage, the electrochemical surface area attenuates severely attributed to detachment and agglomeration of Pt particles, and the catalyst layer framework becomes more porous and hydrophilic, thus raising the double-layer capacitance. The weak catalyst layer structure tends to collapse starting from the lower layer close to the membrane, thereby leaving the dense lower layer and the porous upper layer in all cells. The damage mode is accurately manifested by the stepwise increase in the hydrogen transfer coefficient in the cathode that is identified in electrochemical testing. Meanwhile, the performance decay during operation exhibits typical characteristics of water flooding deterioration, as indicated by the three stages of steady voltage decay, drastic and periodic voltage fluctuation, and cell reversal. Therefore, the increase in water sensitivity and its dominant role in startup-shutdown-induced degradation are confirmed, which are also supported by the carbon-corrosion-induced ionomer agglomeration surrounding the enlarged pores in the upper layer. The enlarged pores are prone to be flooded, and the collapsed partial structure inherently limits the mass transfer. The aging process, evolution mechanisms of internal damage, and dominant factors during startup-shutdown are therefore identified and clearly revealed in stack-level.
Read full abstract