Impact of Water Management on Local Potential Evolutions during PEM Fuel Cell Operation with Dead-Ended Anode

The operation of proton exchange membrane (PEM) with a dead-ended anode may lead to local fuel starvation due to the excessive accumulation of liquid water and possibly nitrogen (due to crossover through the membrane) in the anode compartment. Such fuel-starvation events are localized and may remain undetected in terms of overall cell performance. However, these events can lead to significant rise of the anode (and thus cathode) local potentials and accelerate carbon corrosion and catalyst degradation, which can be particularly damaging to the cell performance [1-3].

In previous works we used a segmented linear cell with reference electrodes to monitor simultaneously the local anode and cathode potentials (i.e. the difference between the metal and electrolyte potential) and current densities during operation while the anode outlet remaining closed between several hundreds and a few thousands of seconds [4]. Apart from the addition of the reference electrodes, this cell is similar to that described in [5]. Such dead-ended anode operation sequences were repeated during 216 hours, which led to non-uniform ElectroChemical Surface Area (ECSA) losses, strong carbon corrosion rates (estimated by measuring CO₂ concentration at the cathode exhaust) and performance degradation along the cell area. The damage was more pronounced in the regions suffering the longest from fuel starvation: i.e., close to the anode outlet.

In a first step, several parameters like membrane thickness, air relative humidity and average current density were varied and the results indicated that water transport from the cathode to the anode as well as nitrogen crossover through the membrane [6] were most probably the two key factors governing fuel starvation.

Then, we evaluated with more details the contributions of nitrogen crossover and water transport to hydrogen starvation. Based on the recent works of Thomas et al. [7, 8], the cell was operated by using three thermal configurations with different cooling temperature on the cathode and anode sides (while the cell average temperature was kept constant) in order to drive water toward the colder side: 

1. T_anode = T_cathode = 65°C;
2. T_anode = 62.5°C  < T_cathode = 67.5°C;
3. T_anode = 67.5°C  > T_cathode = 62.5°C; 

Figure 1 shows the cathode potential evolutions for the six first cycles of dead-end/purge for each thermal configurations (i), (ii) and (iii). For the second configuration (ii), i.e. Ta < Tc, water accumulation in the anode side was favored. As a consequence, the average closing time was 35% faster than for configuration (i). For the third configuration (iii), Ta > Tc, the average cycle duration was seven times greater than that of configuration (i) and ten times greater than that of configuration (ii). For the third configuration, water accumulation in the anode compartment was limited – in practice, no liquid water can be observed at the anode outlet following the purge –. Nitrogen crossover through the membrane was probably the main reason for hydrogen starvation.  In order to assess the impact of the thermal configuration on carbon corrosion and performance decrease, the three aging protocols (for each thermal configuration) were repeated but by setting a fixed time span between two successive purges (and by this way the total number of purges performed in 216 hours). The results allowed us to state that the configuration (iii), T_anode>T_cathode, was less damaging than the two other configurations.

Eventually, such results should help optimizing purge protocols of fuel cell stack operated with a dead-end mode.