Solid oxide fuel cells (SOFC) provide an opportunity for fuel-flexible fuel cells that operate at higher efficiencies than other types of fuel cells. These advantages arise from the high temperature of SOFC operation, 800–1000 °C, which facilitates both direct oxidation and reforming of hydrocarbon fuels. The two reactions are distinguished as follows. In direct oxidation the fuel oxidizes on the anode to produce electricity and CO₂ and H₂O products. In reforming fuel is converted to a mixture of CO and H₂, which in turn is subject to water gas shift equilibrium. The reforming products then undergo electrooxidation to produce electricity, CO₂ (from CO) and H₂O (from H₂).

Understanding the surface electrochemistry in hydrocarbon oxidation requires an analysis of the extents of direct oxidation and reforming. We have developed a SOFC mounted in a vacuum system with facilities for accurate control of fuel and oxygen partial pressures and measurement of reaction products by a calibrated mass spectrometer. A variety of SOFC configurations have been studied, including Pt/GDC/Pt, Pt/GDC/LSC, PtPd/GDC/LSC, Au/GDC/Pt and Pt-GDC/YSZ/LSC, where the designation a/b/c represents anode/electrolyte/cathode, GDC is Gd-doped ceria, and LSC is LaSrCoO_x. These configurations probe the effect of Pt and PtPd as anode catalysts and the influence of a mixed conducting electrolyte (GDC) vs. an ionically conducting electrolyte (YSZ) on the oxidation reactions. Reactions were carried out at 800–1100 K with CH₄ at 5 Torr at the anode and O₂ at 100 Torr at the cathode.

As we have reported previously, methane and higher hydrocarbons exhibit oscillatory reaction behavior in a Pt/GDC/Pt fuel cell. The oscillatory response occurs for a wide range of temperature (750–900 K), but a narrow range of fuel/oxygen partial pressure ratios. Tests with the other fuel cell configurations showed that the oscillatory response occurred only in the presence of Pt as an anode catalyst, as a Au anode did not elicit the response. The response occurred for both GDC and YSZ electrolytes, demonstrating that electronic conduction through the bulk of the electrolyte does not play a role.

The oscillatory response was followed throughout the polarization curve measured from open circuit to short circuit conditions. The oscillatory response was detected as a potential oscillation under galvanostatic control. Oscillations occurred throughout a wide range of the polarization curve, becoming more pronounced with increasing current. Peaks in CO and H₂ production occurred in concert with the oscillatory peaks in current. Under a controlled current of 3 mA (compared with a short circuit current of 7 mA) the potential oscillated between –0.3 and +0.18 V for a PtPd-GDC/YSZ/LSC fuel cell operating at 1100 K. The period of oscillation in this case was 7 hours.

When fed a mixture of methane and water at the anode, the fuel cell did not exhibit oscillations and instead yielded a polarization curve similar to what could be constructed from the peak of the oscillatory response for operation with pure fuel. That is, for a given current, the potential of the methane/water fed cell agreed well with the potential at the peak of the oscillatory response for the methane fed cell. This result, coupled with the observation of CO and H₂ during the peak, indicates that the oscillatory response is due to in situ reforming at the anode in the case of the pure methane fed cell. The methane/water fuel cell exhibits reforming at all times, thus explaining its higher performance.

The absolute rates of reactant disappearance and product appearance during the peak were followed with the calibrated mass spectrometer. These resulting data provide a detailed description of the surface processes and enable direct oxidation to be distinguished from in situ reforming. The results show that direct oxidation is dominant during the long baseline phase and that a peak is triggered by an increase in water production, which leads to in-situ reforming and consequent production of CO and H₂, which react more readily than CH₄, thus leading to the peak in electrical output. Following the peak, CO and H₂ decay, and the CO₂ signal goes through a minimum prior to returning to the baseline condition. This response suggests that a decrease in carbon coverage on the anode allows for increased direct oxidation, thereby initiating the peak, after which reestablishment of surface coverage causes a minimum in CO₂ production and a return to the baseline condition.

We gratefully acknowledge support of this work by the Office of Naval Research.