Multiscale Modelling for Automotive Exhaust-Gas Aftertreatment – From the Quantum Chemistry to the Engineering Level

In this work the behaviour of a novel automotive catalytic converter is theoretically investigated using a multiscale-modelling approach. Processes ranging from the nano-scale (chemical reactions) to the macro-scale (transient conversions in a single monolithic channel of a catalyst) are addressed by different modelling techniques. These independent techniques are combined in a hierarchical multiscale-modelling approach to yield a comprehensive model of an automotive catalytic converter. The investigated rhodium-based exhaust-gas aftertreatment system converts poisonous nitrogen oxides (NOx, x = 1,2) into nitrogen (N2) in an oxygen-rich environment with short, oxygen-lean fuel-rich pulses (0.1 s – 5 s). Experimental investigations of NOx and oxygen on stepped as well as low-index rhodium surfaces indicate that rhodium is inactive towards NOx decomposition. Since oxygen decomposes much faster on any rhodium surface and is present in a vast excess, the rhodium surface should be blocked by oxygen rather instantaneously and, hence, inactive towards NOx decomposition. In order to achieve a more detailed comprehension of the relevant surface processes, surface reactions and diffusion of surface species were studied by means of quantumchemical density functional theory (DFT) calculations. It was shown that the predominant plane in real fcc metal catalyst particles (111) is rather inactive towards NO dissociation. This is even more distinct when the surface is covered with oxygen to a high extent. Furthermore, it was shown that oxygen initially dissociates fast on this facet of rhodium, while this process is self-inhibiting; the activation barrier increases with increasing coverage. Charge analysis supports that these deactivation effects are due to the electron withdrawing effect of the electronegative oxygen precoverage in case of either decomposition reaction (O2 and NO). DFT investigations of monatomic steps, the most common defects on rhodium catalyst particles, showed that decomposition is more likely to occur here, while it is also deactivated by electron withdrawing coadsorbants. Furthermore, it was shown that even though electronic effects influence the probability of dissociation, steric effects are more important. The qualitative knowledge gained by DFT investigation of surface processes was incorporated into time-dependent simulations of reactive flows using the computational tool DETCHEM. DETCHEM, a module of DETCHEM (O. Deutschmann et al.), which was developed as part of the presented work, simulates the transient behaviour of reactive flows based on a hierarchical modelling approach. Time-dependent conversions calculated by DETCHEM based on elementary-step reaction mechanisms, which were improved significantly by knowledge gained from DFT calculations, could reproduce experimentally determined conversions. It can be summarised that this thesis presents a significant step towards a detailed multiscale-modelling of automotive catalytic converters. In a comprehensive approach processes have to be described on different relevant levels from the microscopic to the macroscopic level, i.e., from the quantum chemistry to the continuum engineering level. In particular, insight gained on the quantum-chemistry level can aid the understanding of processes on much higher length and time scales, if properly incorporated within a multiscale modelling approach.