Key points are not available for this paper at this time.
The mechanism of CO oxidation over supported gold catalysts has long been debated, with two prevailing mechanisms dominating the discussion: a water-assisted mechanism and a mechanism involving O-defect sites. In this study, we directly address this debate through a kinetic and mechanistic investigation of the role of water in CO oxidation over Au/TiO2 and Au/Al2O3 catalysts; the results clearly indicate a common water-assisted mechanism to be at work. Water adsorption isotherms were determined with infrared spectroscopy; the extracted equilibrium constant was essentially the same for both catalysts. Added water decreases CO adsorption on Au/TiO2, likely by blocking CO binding sites at the metal-support interface. Reaction kinetics (CO, O2, and H2O reaction orders) were essentially the same for both catalysts, as were measured O-H(D) kinetic isotope effects. These data indicate that the two catalysts operate by essentially the same mechanism under the conditions of these experiments (ambient temperature, significant amounts of water available). A reaction mechanism incorporating the kinetic and thermodynamic data and accounting for different CO and O2/COOH binding sites is proposed. The mechanism and kinetic data are treated with an active site (Michaelis-Menten) approach. This indicated that water adsorption does not significantly affect reaction rate constants, only the number of active sites available at a given water pressure. Extracted water and O2 binding constants are similar on both catalysts and consistent with previous DFT calculations. Water adsorption constants are also similar to independently determined equilibrium constants measured by IR spectroscopy. The likely roles of water, surface carbonates, and oxygen vacancies at the metal-support interface are discussed. The results definitively show that, at least in the presence of added water, O vacancies cannot play an important role in the room-temperature catalysis, and that the water-assisted mechanism is far more consistent with the preponderance of the kinetic data.
Lopez et al. (Sun,) studied this question.