Global kinetic model for lean NOx traps

Modeling and flow reactor experiments are used to investigate the global kinetics for lean NO, traps (LNTs). Experiments were conducted with a Pt/Rh/BaO/Al2O3 model catalyst, and the inlet feed gas was switched between lean and rich periods. It has previously been observed that NO oxidation to NO2 is important for NOx storage, and therefore a global mechanism for NO oxidation on Pt/Al2O3 is developed. This is then used in the NOx trap model, after the parameters had been adjusted to match the NO and NO2 concentrations from experiments on the Pt/Rh/BaO/Al2O3 catalyst. The mass transport of NO and NO2 inside the particles is described by a shrinking-core model. Further, it is found that two global reaction steps are needed for storage in order to explain the experimental observations: one step for the formation of barium nitrates and the other step for the formation of loosely bound barium nitrites. Reaction steps were added to the model for regeneration of the trap with C3H6. The model is tuned based on six experiments at three different temperatures and two different NO concentration levels. The model is able to adequately describe NOx storage during the lean period, the NO reduction during the regeneration period, the NOx breakthrough peaks observed initially in the rich period, and the relation between the measured NO and NO2 concentrations. Experimentally, we have observed that only a fraction of the barium is used for storage in our model catalysts. In the simulations, only 7% of the barium is used for NOx storage. In addition, TEM experiments have shown that our barium particles are large, and therefore a model is evaluated using an inert core in the center of the particle, which resulted in an equally good fit. However, when using catalysts with small particles, which probably is the case in commercial catalysts, a model without an inert core in barium particles seems to be the most realistic one. The model with an inert core is validated with three additional experiments not included in the fitting procedure. In these experiments the oxygen concentration was lowered to 4% during the lean period, compared to 8% O-2 in the experiments used when adjusting the kinetic parameters. The model can simulate the experimental features of these experiments well.