Reactor simulation and kinetic modeling of monolith catalysts for lean nitrogen oxide traps

Lei Cao, Purdue University

Abstract

The reduction of nitrogen oxides (NO and NO2) in the exhaust of lean burn gasoline engines and diesel engines are a challenge with the main reason being the presence of excess oxygen. Starting from 1996, monolith lean NOx trap (LNT) catalysts containing both noble motel sites (usually Pt) and NOx adsorption sites (BaO or K2O) are used industrially for NOx abatement. This technology requires the engine run in cyclic mode between a long fuel lean phase and a short fuel rich phase. Under lean condition, NO oxidation and NO/NO2 adsorption take place to form nitrites and nitrates on the catalyst. The surface NOx are subsequently reduced to N2 by reductants under rich condition. In spite of the industrial success, detailed mechanisms of NOx storage and reduction are not fully understood and robust kinetic models are sparse in the literature. At the same time, experimental approaches in studying NOx traps face difficulties in interpreting the data due to the dynamics of the system and the coupling between simultaneous reactions happening on different sites. In this work, a kinetic simulation by combining a reactor model (heat and mass balance) with global reaction mechanisms was carried out to better understand the performance of the Pt/BaO/Al2O3 catalyst. The following major assumptions were made: (1) radial gradients are negligible compared to axial gradients; (2) axial diffusion/conduction is negligible compared to convection, and (3) fully developed laminar flow along the monolith channel. The resulting time dependent differential equation set is in hyperbolic type, characterized by the absence of the second order derivatives. Finite difference method with upwinding scheme is found effective in solving the partial differential equations without oscillation. Numerical diffusive error caused by upwinding can be minimized by using finer grids and higher order schemes. By solving the NO oxidation model, differences in spatial NO conversion curve were found between NO2 inhibition and non-inhibition cases. First, with no NO2 in the feed, the inhibition model predicts a fast build-up of NO2 close to the inlet than non-inhibition models. Second, upon the addition of NO2 in the inlet NO/O2 mixture, the conversion of NO was decreased significantly while non-inhibition models are not affected. The above model predictions fitted well with experimental observations. An adsorption model assuming NO2 disproportionation and direct NO surface reaction was used for NOx storage with NO2/O 2 and NO/O2 as the inlet. The two time scale mechanism was found necessary to describe NO2 adsorption on the 20wt% BaO catalyst. The model and parameters required to fit the NOx breakthrough curves suggest that CO2 and H2O in the feed reduce the number of sites for NO adsorption. The rate constants for both fast and slow NO2 uptake are decreased in the presence of CO2 and H2O, but the total capacity remains the same. NO2 inhibition in NO oxidation makes the interaction between NO oxidation and NO/NO2 adsorption profound. Under reaction conditions, H2 reduction of surface NOx is reductant supply limited with NH3 as the reducing intermediate. The confined reduction front moving along the channel localizes the heat generation leading to a surface temperature in the reduction front about 35°C higher than the inlet gas temperature at our reaction conditions.

Degree

Ph.D.

Advisors

Caruthers, Purdue University.

Subject Area

Chemical engineering

COinS