We have investigated the mechanism of increase in oxygen transport resistance (Rm) due to aging of electrodes with low-loaded de-alloyed PtCo cathode catalysts supported on high surface area carbon (d-PtCo/C). Commercially available membrane electrode assemblies with 0.05, 0.1 and 0.15 mg/cm2 Pt loadings in d-PtCo/C cathode catalysts were assembled into cells and subjected to accelerated stress tests (ASTs) that consisted of 0.6-0.95 V square wave potentials with 3-s hold at upper and lower potential limits. Catalyst degradation was characterized by measuring mass activity for the oxygen reduction reaction, electrochemically active surface area (ECSA) and polarization performance in H2/air at beginning of test (BOT) and after 15k and 30k (EOT) potential cycles in H2/N2. Concurrently, limiting current densities (iL) were measured for different pressures (1.3, 1.8, 2.3, 2.8 and 3.2 atm) and oxygen concentrations (0.01, 0.02 and 0.04 O2 mole fraction in dry air) at 80oC and 90% relative humidity (RH). We observed that the O2 transport resistance inferred from the measured limiting current densities depends not only on pressure (P), catalyst loading and aging, but also on the current density itself. The observed increase in Rm with iL is almost linear and independent of pressure, suggesting the following new representation for Rm: Rm = {Rg P/Pr + Rd P/Pr} + RKn + {Rf + Ri i/ir}, Rf = RO2/SPt In this representation, the various terms denote the resistance for O2 transport across the boundary layer in the gas channel (Rg), across the gas diffusion layer (Rd), across the microporous layer and secondary pores in cathode electrode (RKn), and across the ionomer film on Pt particles supported on carbon (Rf). Note that Rf is inversely proportional to roughness (SPt, cmPt 2/cm2), and Ri is the portion of Rm that depends linearly on the normalized current density (ir = 1 A/cm2). As in earlier formulations, Rg and Rd are pressure dependent (Pr = 1 atm) as they are related to molecular diffusivity of O2 in gas channel (GC) and in gas diffusion layer (DM), whereas RKn and Rf are pressure-independent as they are related to Knudsen diffusion in <50-nm pores and O2 permeability across the ionomer film. Using the new representation for the three cells with different loadings, we determined Ri from the current density dependence of Rm at BOT as 0.306 ± 0.1 s/cm; Ri was found to be nearly independent of pressure. We estimated Rg + Rd from the pressure dependence of Rm at BOT as 0.409 ± 0.032 s/cm. Knowing Rg, Rd, and Ri, we calculated RKn + Rf from Rm, and plotted it as a function of 1/ SPt to determine RKn from the intercept and Rf from the slope. Using all the available data at BOT and after 15k and 30k potential cycles, the estimated values are 0.07 s/cm for RKn and 9.15 s/cm for RO2. This value of RO2 compares favorably with the reported literature data for fresh electrodes. To explain the current density dependence of Rm, we conducted O2 transport simulations on an electrode microstructure determined from nano-computed X-ray tomography at an APS (Advanced Photo Source) beamline. The microstructure was digitized at a resolution of 2.5 nm to reconstruct the connectivity and heterogeneous size and spatial distributions of secondary pores, primary pores, carbon, catalyst particles, and ionomer phases. Assuming that the secondary pores are dry at 90% RH, the primary pores were progressively filled with liquid water to match the measured increase in electrode resistance at higher limiting current densities. The water-filling algorithm follows the concept of capillary condensation whereby the smallest primary pores are allowed to preferentially saturate first. In these simulations, changing the primary pore saturation from 0 (dry) to 1 (flooded) results in the limiting current density increasing from 0 to 1 A/cm2 and the electrode resistance from 0.2 to 0.65 s/cm. Figure 1 summarizes the effects of Pt loading and catalyst aging on O2 transport resistance at one set of reference conditions: 1.5 atm operating pressure, 4% O2 mole fraction, 80oC and 90% RH. At BOT, reducing Pt loading to 0.05 from 0.15 mg/cm2 causes a 31% increase in Rm and a 23% drop in iL. After 30k cycles, Rm increases by 12% for the 0.15 mg/cm2 loading and by 43% for the 0.05 mg/cm2 loading, and is 66% higher in the lower loaded cell even though iL is 40% smaller. More telling is the increase in the electrode resistance (Rf + Ri i/ir) after 30 k cycles: 35% for 0.15 mg/cm2 loading and 115% for 0.05 mg/cm2 loading. Figure 1
Read full abstract