AmericanChemicalSociety.com

Copper can be added to promote the activity of iron oxide catalysts for the high temperature water-gas shift reaction. There are two hypotheses regarding the manner by which copper acts. One hypothesis is that copper cations substitute for iron cations in the oxide, and the associated changes in the electronic structure of the catalyst affect its catalytic behavior. The other hypothesis is that under reaction conditions, the added copper forms small metallic particles on the iron oxide surface, and these particles are involved in the catalysis. In previous work performed under this grant, computational chemistry was used to model the active sites for water-gas shift and to estimate the strength of bonding of various species on the catalyst surface. The species that were considered include molecularly adsorbed water, dissociatively adsorbed water, an oxidized site, a carbonate and a formate. The results indicate that when copper cations substitute for iron cations at sites near, but below, the surface, the strength of bonding decreases by less than 20 kJ mol^-1, whereas when copper cations substitute for iron cations that are in the surface, the strength of bonding decreases by 60 to 80 kJ mol^-1.

The mechanism of water-gas shift over a copper-free ferrochrome catalyst has not been determined unequivocally. Both redox and formate/carboxylate mechanisms have been advocated. Starting with a redox mechanism, we found that a simple four-step mechanism provided an accurate description of the water-gas shift kinetics. The heats of adsorption of surface intermediates were among the parameters in the model that were determined as a result of fitting it to kinetic data. The same mechanistic model also was found to accurately model experimental kinetics data for a copper-promoted ferrochrome catalyst. The fit of the model to the data for the promoted catalyst was not sensitive to the surface bond strength of adsorbed carbon dioxide, so an accurate value was not obtained for that species. However, for the other surface intermediates, the strength of bonding on the promoted and unpromoted catalysts differed by only 20 to 30 kJ mol^-1. This is in very good agreement with the predictions from the computational chemistry models.

A four-step mechanism involving a formate intermediate was also examined, and the results were similar. The formate mechanism was able to model the experimental kinetics data for both the copper-free and the copper-promoted ferrochrome catalysts with accuracy comparable to that of the redox mechanism. In the case of the formate mechanism, the differences in the surface bond strengths obtained via kinetic modeling showed differences between 5 and 65 kJ mol^-1 when comparing the promoted and unpromoted catalysts. These differences are greater than those for the redox mechanism, but they are still consistent with the predictions from computational chemistry.

It is not entirely surprising that mechanistic kinetic modeling did not discriminate between the redox and formate mechanisms. Good experimental data exist in support of each mechanism, and the question of mechanism has remained unresolved for decades. The kinetic modeling does prove more useful with respect to the role of copper as a promoter, however. In both mechanisms studied, the same mechanism was capable of accurately describing the kinetics of both the promoted and unpromoted catalysts. Doing so required relatively small changes in surface bond strengths, and the magnitude of those changes was consistent with predictions from computational chemistry that assumed copper cations to substitute for iron cations within the iron oxide. This strongly suggests that metallic copper particles are not necessary to explain the promotional effect of copper. Furthermore, recently reported power-law kinetic models for water-gas shift over metallic copper differ markedly from models for the reaction over iron oxide, whereas the present results show only small differences in the kinetic expressions.

The addition of copper to ferrochrome catalysts has been reported to improve performance at low steam to CO ratios by reducing the co-production of methane, to increase activity at lower temperatures by decreasing the apparent activation energy and to help stabilize the catalyst surface area. Our kinetics studies revealed that adding copper also lessens the degree of product inhibition by carbon dioxide. The apparent reaction order with respect to carbon dioxide is -0.6 for a ferrochrome catalyst. When 5 wt% copper is added to the catalyst, the reaction order with respect to carbon dioxide increases to -0.26. Further increase in the copper loading does not lead to any additional lessening of the inhibition by carbon dioxide.

Finally, we have also examined the effect on water-gas shift activity of adding gold nanoparticles to the ferrochrome catalyst. As reported by many others, the activity of the catalysts is very sensitive to the method of preparation; gold particles smaller then 5 nm are necessary to observe enhanced activity. However, when properly prepared, the resulting catalysts are active at temperatures as low as 160 Celsius. We have studied the kinetics of water-gas shift using these catalysts in the range from 160 to 180 Celsius. The catalysts deactivate significantly during the course of the kinetics measurements so that bracketing techniques are needed to account for the deactivation. When this is done, the reaction orders are found to be 0.39 for CO, 0.98 for H₂O, -0.28 for CO₂ and -0.26 for H₂, and the activation energy is 88.2 kJ mol^-1. There does not appear to be a single cause for the deactivation of the catalyst. Part, but not all, of the deactivation can be attributed to growth in the size of the gold nanoparticles. The deactivation also correlates with the carbon activity of the gas phase, but the reason for the correlation remains to be determined.