Faster Model Calibration for Aged Diesel Oxidation Catalysts and NOx Trap Catalysts
The fuel consumption of vehicles containing NOx-trap catalysts strongly depends on the required regeneration frequency. Since the regeneration frequency increases with catalyst deactivation, aging effects have to be considered for simulation-supported optimisation of operation parameters. Within an FVV project, a strategy has been developed at the Karlsruhe Institute of Technology and the University of Stuttgart which enables deriving the deteriorated conversion behaviour of aged catalytically from a few measurements. Additionally, a comprehensive experimental database for the correlation between light-off temperature and catalytic active surface has been established for the diesel oxidation catalyst.
The demand on automotive exhaust-gas aftertreatment systems of complying with the European emission standards during the whole life time requires knowledge about the aging behaviour of the particular abatement units. In the following, efforts are focused on the diesel oxidation catalyst (DOC) and the NOx storage/reduction catalyst (NSC). The DOC is mostly the first unit of the exhaust aftertreatment system. Its task is to promote the oxidation of unburned hydrocarbons and CO on the one hand, and to provide a sufficiently high NO2/NOx ratio for the subsequent catalysts (NSC, SCR, CRT-DPF) on the other hand. Therefore, the activity of the DOC has a strong influence on the whole aftertreatment system. The NSC is the most sensitive exhaust-gas converter regarding thermal aging and deactivation. Aged NSCs have to be regenerated more frequently by rich exhaust gas due to the degraded NOx storage ability. This directly affects the fuel consumption and the CO2 emission.
In order to ensure optimal operation conditions, the current (aging dependent) catalyst state during vehicle operation has to be considered when developing operation strategies and engine control concepts. The development of such concepts and the design of the exhaust system are complex processes which require a simulation-supported approach to limit the costs and the development time. By now, the adjustment of simulation models to aged catalyst states demands comprehensive experimental investigations and model recalibration. In the FVV project “Accelerated model recalibration for aged catalysts” (KASPar), a simplified model reparameterisation strategy has been developed which enables determining changed model parameters just by performing a few experiments.
In order to determine the influence of aging on the conversion and storage behaviour and the morphology of the catalyst, DOCs and NSCs have been aged under various hydrothermal conditions. Subsequently, the catalyst has been characterised and changes of the conversion behaviour have been investigated by kinetic measurements, ➊. Transmission electron microscopy (TEM) was used to investigate the Pt particle size distribution and the spatial distribution of the washcoat components, whereas X-ray diffraction (XRD) was applied to identify crystalline components. The specific surface area and the noble metal dispersion, defined as the ratio of the number of surface atoms to the overall number of atoms, have been determined by physisorption and chemisorption measurements. A macrokinetic model was used for simulation which has been first adapted to the conversion behaviour of a fresh catalyst. For the aged samples, correlations between the physical and chemical properties of the catalyst and the model parameters have been investigated.
2 Experimental Investigations
A commercial platinum-containing DOC and a commercial NSC with platinum, palladium and rhodium as catalytically active components and barium and cerium-zirconium-oxide as storage components for NOx and oxygen, respectively, have been investigated. A schematic illustration of the catalysts is shown in ➋.
lean, hydrothermal furnace aging: compressed air with 10 % H2O, 6 h
reductive posttreatment: 5 % H2, 10 % H2O in N2, 2 h at 350 °C.
3 Diesel Oxidation Catalyst
For diesel oxidation catalysts, it could be shown in a preceding FVV project (“catalyst simulation“) that it is possible to simplify the reparameterisation significantly since the reaction rates merely depend on the number of accessible platinum surface sites . Within the framework of KASPar, a comprehensive experimental database has been established for that correlation. For this purpose, the noble metal dispersion and the catalytic activity of commercial DOC samples have been investigated subsequent to the aging treatments mentioned above.
For all investigated oxygen-rich aging atmospheres, the catalytic activity could be correlated with the accessible noble metal surface. Independent from the aging temperature and the detailed composition of the lean aging atmosphere, the same trend for the correlation between these two characteristic quantities has been observed. However, samples aged in N2 behaved differently which is assigned to a deviant sintering mechanism.
4 NOx Storage Catalyst
Compared to the DOC, the chemical structure of the NSC is significantly more complex. It contains palladium and rhodium as noble metal components next to platinum as well as the NOx storage component barium and the oxygen storage component cerium. Correspondingly, the aging behaviour is more complex: Besides the changes of the noble metal dispersion, there are also changes of the specific surface of the washcoat and phase changes. Furthermore, mixed oxide formation from the NOx and oxygen storage material and the support material is expected [2, 3, 4, 5].
4.1 Physical and Chemical Characterisation
The investigation of the aged samples by nitrogen physisorption revealed that the impact of aging on the washcoat structure depends significantly on the aging atmosphere, ➍. While lean/rich cycling barely has any influence on the washcoat porosity, in the case of lean hydrothermal aging, the specific surface area decreases markedly with aging temperature. Pore size distribution analysis clearly showed that smaller pores gradually collapse in consequence of lean aging. Especially, the proportion of pores with a pore radius below the maximum of 8 nm is substantially reduced.
The CO chemisorption results showed that the noble metal dispersion decreases linearly with increasing aging temperature, \w4. Furthermore, it is shown that elongation of the aging time only has a minor impact on the catalytic surface area while lean/rich cycling at elevated temperature appears to be detrimental to the noble metal dispersion.
For lean aging atmospheres, TEM measurements revealed that the noble metals remained highly dispersed even after aging at high temperatures. The mean diameter of the Pt nanoparticles only increased from 1 nm for NSC500 to 2 nm for NSC800, ➎. However, lean/rich aging conditions caused remarkable particle sintering and therefore led to a broad distribution of Pt crystallite sizes (2 to 20 nm). This is ascribed to the temperature maxima during the rich periods. Furthermore, a significant growth of the ceria agglomerates has been observed. It could be shown that the ceria agglomerates of NSC700LR exhibit a broad size distribution between 30 nm and 800 nm whereas the pore diameters of NSC800 are in the range of 20 to 30 nm, ➏.
XRD data revealed ceria, alumina and BaCO3 as the main crystalline washcoat components of the NSC. While the amount of crystalline BaCO3 is hardly influenced by variation of the aging temperature, the increase of aging time from 6 to 24 h caused a complete conversion of the crystalline BaCO3 phase into an amorphous phase. Aging using lean/rich cycling conditions also resulted in a remarkable decrease of the crystallinity of the barium component. Independent from the aging treatment, no formation of mixed oxides was observed. This indicates that mixed oxides are present as amorphous phase.
4.2 Conversion Behaviour
The commercial catalysts have been investigated in an isothermal flat-bed reactor with a synthetic off-gas under realistic flow conditions, \w1. Due to the small sample height, the reaction conditions in the reactor are nearly isothermal by which reaction kinetics can be studied in the absence of disturbing axial and radial temperature gradients. The segmentation of the catalyst into five sequential slices with a length of 40 mm and width of 30 mm, respectively, enables the withdrawal and analysis of a partial flow downstream of each slice. Gas analysis was accomplished with SI-mass spectrometry and FTIR. Hereby, concentration profiles can be measured along the catalyst length under steady-state and transient conditions.
Generally — as expected — the light-off temperature of the hydrothermally aged catalysts increases with aging temperature and the O2- and NOx storage capacity decreases. Regarding the conversion behaviour, the sample aged at 700 °C under lean/rich cycling conditions is mostly in-between the samples aged at 700 and 800 °C under lean conditions. In contrary, the oxidation activity is reduced more markedly in comparison to the lean aged samples. This is consistent with the TEM results which revealed that this NSC exhibits significantly larger noble metal particles.
The increase of aging time from 6 to 24 h at 700 °C causes a shift of the light-off temperature for the CO and propene oxidation to the level of aging at 800 °C. The O2 and NOx storage capacity further decreases with elongated aging time.
4.3 Computer Simulation
Within the FVV project DeNOx Modell III, a model for the NOx-storage catalyst was developed which is able to correctly describe even dynamic processes during NOx storage and regeneration for a broad range of operation conditions . The applied model stands out by a detailed display of the processes occurring inside the NOx storage component (barium), ➐. However, the NSC model requires a large number of model parameters for the description of the chemical conversion processes at the noble metal surfaces, the transport processes and the chemical conversion processes within the storage materials. Overall, 84 model parameters are needed: oxidation reactions: 32, reduction reactions: 23, water-gas shift: 8, oxygen storage: 7, NOx storage: 14.
At first, all model parameters had been determined for the pre-conditioned reference catalyst (NSC500). Subsequently, it has been investigated, whether it is possible to reduce the experimental effort when determining the kinetic parameters for the aged catalysts. Hereby, the kinetic of the fresh catalyst was used as a starting-point.
scaling of oxidation reactions with one scaling factor fOxi
scaling of NO oxidation’s frequency factor by fNO-Oxi
scaling of all reduction reactions with one scaling factor fRed
scaling of water-gas shift reaction’s frequency factor by fWGS
adaptation of oxygen storage capacity according to measurement
scaling of the accessible barium surface proportional to the measured NOx storage capacity
scaling of barium nitrate regeneration with fNit-Reg (reduction of the barium nitrate by H2, CO, and propene).
Scaling 1 to 4 account for the sintering of the noble metal particles (Pt, Pd, Rh) leading to a decreased catalytic active surface and a slowdown of the reactions. Scaling 5 to 7 are required due to the aging induced alteration of the storage components underwent (conversion of barium and cerium into inactive mixed oxides). All activation energies of the reactions and parameters accounting for inhibition effects on the catalytic reactions can be directly transferred to the aged catalysts. The experimental effort for the determination of the scaling factors is reduced to the five experiments listed in ➑.
In ➒ it can be seen by means of simulation of lean/rich cycles for the reference catalyst NSC500 and for NSC700LR, that the changed conversion behaviour of the aged NSC is reproduced very well by using the reparameterisation strategy described above. In ➓, the applied scaling factors are shown for the lean aged catalysts.
Based on a successful correlation of the physical and chemical properties of thermal aged NOx storage catalysts and kinetic measurements, the relevant catalyst properties for the change of the catalytic activity and the storage capacity could be identified.
It became apparent that the changes of the catalyst properties strongly depend on the aging atmosphere. The main changes of the catalyst induced by aging are sintering of the noble metal particles, reduction of the specific surface area of the washcoat, and in case of NOx storage catalyst a decrease of the crystallinity of the barium component as well as the growth of ceria crystallites and agglomerates have been observed.
The simplification of the reparameterisation of the kinetic model for various aging states was successful: In case of DOC, only one parameter has to be adjusted, which accounts for the direct correlation of the changes in the catalytic activity with the noble metal surface.
For NSC, only seven model parameters have to be adapted. Thus, the parameterisation effort is reduced significantly and moreover the required experimental effort for the investigation of aged catalysts is markedly diminished.
The authors gratefully acknowledge the Forschungsvereinigung Verbrennungskraftmaschinen (FVV e.V.) for the financial support and Umicore AG&Co.KG for providing the catalysts. The project has been accompanied by a workgroup of the FVV under the direction of Dr. Volker Schmeißer, Daimler AG. Special thanks goes to him for the invaluable support. Furthermore, the authors thank Karlsruhe Nano Micro Facility http://www.kit.edu/knmf for carrying out the TEM measurements.
- Hauff, K.; Boll, W.; Tischer, S.; Chan, Ch.; Tuttlies, U.; Eigenberger, G.; Deutschmann, O.; Nieken, U.: Macro- and Microkinetic Simulation of Diesel Oxidation Catalyst: Effect of Aging, Noble Metal Loading and Platinum Oxidation. In: Chemie Ingenieur Technik (2013), pp. 673–685Google Scholar
- Fekete, N.; Kemmler, R.; Voigtländer, D.; Krutzsch, B.; Zimmer, E.; Wenninger, G.; Strehlau, W.; Tillaart, J. v. d.; Leyrer, J.; Lox, E. S.; Müller, W.: Evaluation of NOx Storage Catalysts for Lean Burn Gasoline Fueled Passenger Cars. In: SAE Technical Paper (1997), pp. 249–264Google Scholar
- Toops, T.; Bunting, B. G.; Nguyen, K.; Gopinath, A.: Effect of engine-based thermal aging on surface morphology and performance of Lean NOx traps. In: Catalysis Today (2007), pp. 285–292Google Scholar
- Casapu, M.; Grunwaldt, J. D.; Maciejewski, M.; Baiker, A.; Eckhoff, S.; Gobel, U.; Wittrock, M.: The fate of platinum in Pt/Ba/CeO2 and Pt/Ba/Al2O3 catalysts during thermal aging. In: Journal of Catalysis (2007) pp. 28–38Google Scholar
- Eberhardt, M.; Riedel, R.; Göbel, U.; Theis, J.; Lox, L.: Fundamental investigations of thermal aging phenomena of model NOx storage systems. In: Topics in Catalysis (2004), pp. 135–142Google Scholar
- Schmeißer, V.; Koop, J.: Modellierung und Simulation der NOx-Minderung an Speicherkatalysatoren in sauerstoffreichen Abgasen. In: Final report of the project No. 833 of Forschungsvereinigung Verbrennungskraftmaschinen (2008), Issue 848Google Scholar