Mechanism and Kinetics of Carbon Corrosion in Polymer Electrolyte Fuel Cells during Drive Cycles
Catalyst support-carbon can also corrode below the open-circuit potentials typical of normal operation of automotive fuel cells. Figure 1a is a typical trace of the measured carbon corrosion rate during a simulated drive cycle in which the potential was varied between 0.4 V and 0.95 V. The hold times were 0.5 min at the low potential and 5 min at the high potential. The data are for an Ion Power supplied MEA with 0.15 mg/cm2 Pt loading in cathode, 0.23 Pt/C ratio, and high surface area Ketzenblack (E-type) carbon support. The data are for H2-air at fixed flow rates, 80oC cell temperature, and 100% relative humidity (RH). The corrosion trace in Fig. 1a has a few particularly interesting features. The larger spike in corrosion rate occurs as the cell potential is raised from 0.4 V to 0.95 V. However, the corrosion rate declines rapidly and approaches zero as the cell is held at the high potential for a long time. The smaller spike in corrosion rate occurs as the cell potential is reduced from 0.95 V to 0.4 V. The corrosion rate also declines as the cell is held at the low potential but does not approach zero.
Similar features in carbon corrosion rates were observed in other MEAs with Vulcan XC-72 (V-type) and graphitized Ketzenblack (EA-type) carbons as catalyst supports. We have formulated a simple carbon corrosion model to quantitatively capture the essential features of the data. As indicated in Fig. 1a, the surface carbon sites are represented as vacant (C#), active carbon oxides (C#OH), and passive carbon oxides (C#Ox). The active carbon oxides are formed at low potentials. The spike in carbon corrosion after a step increase in potential is associated with the direct oxidation of C#OH to CO2 through its reaction with H2O.
Holding the cell at a high potential leads to the formation of C#Ox and displacement of C#OH. The corrosion rate goes down as C#Ox passivates the carbon surface. Also, at potentials higher than 0.4 V, Pt sites begin to convert to PtOH, and at still higher potentials, PtOH converts to PtO.
In the model shown in Fig. 1a, a step decrease in potential results in the sites occupied by passive C#Ox being vacated. The vacated sites are available to be occupied by active C#OH which reacts with the OH-like species on adjacent catalyst surface to produce a spike in carbon corrosion. With time, PtOH gradually reduces to Pt, and the carbon corrosion rate decreases to a lower, non-zero value.
As partial validation, Fig. 1b shows the modeled steady-state corrosion rate and compares it with the asymptotic rates measured in the experiments when the cell with E-type carbon is held at high potentials for 5 min. Both the model and the data exhibit peak steady-state corrosion rates for E-type carbon at about 0.6 V. As seen in Fig. 1a, the transient corrosion rates can be much higher under potential cycling.
Acknowledgments: This work is supported by DOE-EERE-FCTO with Dr. Nancy Garland as the Technology Development Manager.