(269a) Modelling of the Dynamic Behavior of Catalyst Materials in Reacting Conditions: An Application to the Catalytic Partial Oxidation of Methane on Rhodium

Raffaele Cheula¹, Aloysius Soon², Matteo Maestri^1*
1Laboratory of Catalysis and Catalytic Processes, Dipartimento di Energia, Politecnico di Milano, via La Masa 34, Milano, Italy
²Department of Materials Science and Engineering, Yonsei University, 50 Yonsei-ro Seodaemun-gu, Seoul, Republic of Korea
^*Corresponding author: matteo.maestri@polimi.it

1. Introduction

Several experimental studies demonstrated that the activity of heterogeneous catalysts is strongly dependent on their structure [1]. Moreover, the catalyst structure is deeply influenced by the reaction environment. As a result, catalyst materials in heterogeneous catalytic reactors are intrinsically dynamic systems: they change their size and shape in response to the conditions of the reaction environment, which in turn affects the reactivity [2]. Microkinetic modelling is a key tool for the investigation of the behavior of complex chemical reactions, making it suitable for the fundamental understanding of the structure-activity relations of catalytic processes. However, state-of-the-art microkinetic models in heterogeneous catalysis lack in the description of the catalyst structure. In fact, they typically rely on an abstract and “structureless” concept of catalyst active site. This approach leads to a satisfactory description of conversion and selectivity trends and therefore it is a very valuable tool for the description of the macroscopic kinetic behavior of a catalytic process [3]. However, the material gap in the modelling of the active sites intrinsically precludes the understanding of the underlying mechanisms at the atomic-scale level. As a consequence, the incorporation of the catalyst structure in microkinetic modelling becomes of paramount importance for the fundamental understanding of the structure-activity relations, which is widely recognized as one of the main progress area in modelling of catalysis towards the design and the optimization of catalysts based on functional understanding rather empirical testing. In this contribution we couple microkinetic modelling and ab initio thermodynamics for the characterization of catalyst nanoparticles in terms of surfaces morphology and three-dimensional shape as a function of the reaction environment. This allows to calculate the distribution of the catalyst active sites and how it changes during reaction. The methodology is applied in the context of the CH₄ catalytic partial oxidation and reverse water-gas shift reactions on Rh/Al₂O₃.

2. Methods

Multiscale reactor modelling is employed for the calculation of the gaseous composition profiles inside a chemical reactor and for the identification of the most abundant reaction intermediates (MARIs) present on the catalyst surfaces. Density functional theory calculations and ab initio thermodynamics are then exploited to calculate the most stable bulk and surface structures of the catalyst at different conditions of the reaction environment. With this approach we consider the presence of the MARIs in thermodynamic equilibrium with their reservoirs in the gas phase surrounding the catalyst. Atomistic Wulff-Kaishew construction is then applied to calculate the three-dimensional shape of the supported catalyst nanoparticles as function of the chemical environment inside the reactor. Given the particles size distribution from HRTEM images, we accurately calculate the catalyst dispersion and the distribution of the different types of the catalyst active sites and how it changes during reaction.

3. Results and discussion

The multiscale reactor modelling of the CPO on Rh/Al₂O₃ reacting system shows a sharp change in the reaction environment with residence time and temperature. In particular, when oxygen is present in the boundary layer around the catalyst surfaces, the production of CO and H₂ is not observed (zone 1 in Fig. 1a). Once oxygen has been totally depleted, syngas starts producing (zone 2 in Fig. 1a). We investigate the most stable structure of the catalyst in these two different zones of the reactors. First, the most stable bulk phase of the catalyst is calculated, considering the thermodynamic equilibrium with the gas phase. Then, the morphology of different crystal facets of the nanoparticles is analyzed: for each facet, the thermodynamically most stable surface structure in equilibrium with the surrounding gas is calculated. The results are reported in terms of bulk phase diagram in Figure 1b as a function of temperature and partial pressure of O₂. We find that the change from zone 1 to zone 2 is accompanied by a change in the bulk structure and the morphology of the catalyst. In the oxidizing zone of the reactor (zone 1), Rh₂O₃ is the thermodynamically stable bulk phase, and the nanoparticle exposes mainly the (0001), (1-102) and (11-23) crystal facets (Fig. 1b). In zone 2 of the reactor, instead, where O₂ drops to very low values at the catalyst interface, metallic Rh becomes the most stable bulk structure. CO* and H* are the MARIs, and Rh(100) and the Rh(111) are the facets most exposed by the nanoparticle. High Miller index facets – in particular Rh(311) and Rh(331) – become more stable with the increase of coverage of the MARIs, observed during reaction, yielding an important change in shape of the catalyst and in the distribution of the active sites. The change of the oxidation state of Rh as a response of the change in O₂ concentration in the gas phase is in agreement with operando spectroscopy studies [3]. Given the particle size distribution from HRTEM images, we calculate the dispersion of the catalyst by filling Wulff-Kaishew plots with Rh atoms and counting the resulting number of atoms that constitute the catalyst surface. Good agreement is found with experimental values of CO and H₂ chemisorption. The framework is applied also to the reverse water-gas shift reaction on Rh/Al₂O₃. The distribution of the active sites is found to change with the amount of CO in the reacting environment. The three-dimensional map of the active sites of the nanoparticles, is then translated to a two-dimensional grid for kinetic Monte Carlo simulations.

4. Conclusions

Multiscale reactor modelling, ab initio thermodynamics and Wulff-Kaishew construction have been coupled for predicting the morphology of supported catalyst nanoparticles and its variation under reacting conditions. Our results – in agreement with spectroscopic studies – show that Rh can undergo strong variation of bulk phase composition and in the active sites distribution in response to different reaction environment found in the reactor.

Acknowledgments

Financial support from the European Research Council is gratefully acknowledged (ERC project SHAPE, grant n° 677423).

References

[1] G. Ertl: Angew. Chem. Int. Ed., 47 (2008) 3524-3535

[2] M. Maestri. D. G. Vlachos, A. Beretta, G. Groppi, E. Tronconi: Top. Catal., 52 (2009) 1983-1988

[3] J. D. Grunwaldt, S. Hannemann, C. G. Schroer, and A. Baiker: J. Phys. Chem. B, 110 (2006) 8674-8680