Methane combustion is a very active area of current research since natural gas is expected to be used more and more as a source of energy in the future. Methane is also attractive because it emits the least amount of CO2 per unit energy of all the hydrocarbons. Many researchers have found palladium based catalysts to be the most promising for the methane combustion reaction. However, many aspects of this reaction, such as the reaction mechanism and the activation of the C-H bond, are not yet well understood. This research focuses on using density functional theory (DFT) to help answer these questions.
We have conducted periodic DFT calculations, utilizing the VASP software, to study the adsorption of 11 different species on two PdO surfaces (100 and 110) with and without oxygen defects. The 11 species were chosen as likely surface intermediates involved in the actual reaction mechanism. These adsorption calculations are then used as starting points for examining different reaction pathways and performing transition state analyses.
The first reaction pathway studied is perhaps the most important from an experimental point of view, namely, the activation of methane. The cleavage of the first C-H bond is considered to be the rate limiting step in the methane combustion process. Three possible pathways on the (100) surface have been discovered for the activation step through the use of constrained optimization calculations. Two of the pathways gave similar activation barriers of about 30 kcal/mol, which is well within the range seen by various experimental researchers. The third pathway predicts a barrier of 23 kcal/mol which is significantly lower than the other two. This pathway occurs with oxygen vacancies on the surface. However, further thermodynamic calculations showed that under reaction conditions that oxygen vacancies rarely exist.
Nudged elastic band (NEB) calculations were performed in order to fine tune the constrained optimization results and determine transition state geometries. Attempts at such calculations failed due to convergence problems. To overcome this difficultly, we have developed a cluster model of the PdO(100) surface in order to utilize the more robust algorithms contained within the Gaussian 03 software package. The results from the cluster model are then compared with those from the periodic model to validate the use of a cluster.