Steam methane reforming is an important industrial reaction for the conversion of methane with steam to synthesis gas, a mixture of carbon monoxide and hydrogen. Hydrogen is used in many applications, e.g. for hydrogenation purposes. It is also used for the production of bulk chemicals such as ammonia and in combination with carbon monoxide for the manufacture of methanol and other oxygenates. Fischer-Tropsch synthesis is rapidly becoming a more important source of clean transportation fuels and potentially of chemicals and is also based on the use of synthesis gas. The most frequently used metal for steam methane reforming (SMR) catalysts is nickel. Typical reaction conditions for SMR are 800-900 °C to obtain high conversion of this endothermic reaction. However, in a number of applications, e.g. in pre-reforming and especially in future membrane steam reformers, the use of noble metals with a higher catalytic activity may be interesting. Especially in separation-enhanced reformers, which are typically run at lower temperatures, it is important to maintain high catalytic activity by shifting to noble metals such as Ru or Rh. Noble metals also show a higher resistance against carbon formation, which is one of the primary causes of deactivation, especially under carbon-rich conditions as can occur in a separation-enhanced reformer. The main goal of separation-enhanced steam reformers is to obtain pure streams of H2 and CO2 (CO is converted by the water-gas shift reaction to CO2), the former to be used for electricity generation, the latter for storage. As such, this concept is a potential technology for carbon capture and sequestration. As cost is a crucial factor for electricity generation, it is important to decrease the catalyst cost as much as possible, especially when the use of noble metals is required. Thus, we explore here in detail the structure dependence of the SMR reaction for Rh nanoparticles with the aim to guide the design and synthesis of optimal steam reforming catalysts. To investigate structure sensitivity, we study the three candidate rate controlling elementary reaction steps in steam methane reforming, i.e. water dissociation (Chapter 3), CO formation (Chapter 4) and CH4 dissociation (Chapter 5). This is done by density functional theory (DFT), which is a state-of-the-art technique to compute activation barriers of periodic models of transition metal surfaces. The results of these three studies are used as input for a microkinetic simulation model, which aims to understand the importance of the various reaction mechanisms as well as to explain experimentally observed structure dependence (Chapter 6). Figure 1. Representation of two Rh surfaces models employed in the present work. A stepped surface (a) with (211) steps and (111) terraces. The reactivity of the edge atoms as occurring on nanoparticles is simulated using a nanorod model (b). In Chapter 3, the dissociation of water on planar and stepped surfaces (Figure 1a) and the role of oxygen adatoms herein were investigated. It is concluded that the activation of water is not influenced by the coordination number of the surface atoms involved. The activation energies are very close with a value of 63 kJ/mol for the planar Rh(111) and of 61 kJmol for the stepped Rh(221) surface. However, in the presence of a surface oxygen atom, which can act as a hydrogen acceptor, the barrier for the stepped surface (28 kJ/mol) is significantly lower than on the planar surface (53 kJ/mol). Despite this difference, a close inspection of the potential energy diagram indicates that the overall barrier in the latter cases is also around 60 kJ/mol, because one needs to account for the energy cost to bring the oxygen adatom in a favorable position. This process is endothermic because of the lateral interactions. A Brønsted-Evans-Polanyi type correlation of the activation barriers with the metal-hydroxyl bond shows that the transition state of water dissociation has a slightly late character. Another finding is that there are large compensation effect for the oxygen-assisted water dissociation mechanism. Chapter 4 compares three different reaction mechanisms for CO formation on planar Rh(111) and stepped Rh(211) surfaces (Figure 1a) starting from adsorbed CH and O. A general conclusion is that the direct mechanism via C+O recombination competes with the one going through a formyl intermediate (HCO), whereas the pathway via an alcoholate intermediate (COH) is unfavorable. On the planar surface the barrier for CH dissociation is ~100 kJ/mol, but the overall barrier for CO formation is 180 kJ/mol. The formation of the formyl intermediate on the planar surface has an activation barrier of 180 kJ/mol and its subsequent decomposition towards CO and H is easy. Overall barriers for these two processes are very similar. The presence of a step-edge site is favorable for formyl formation and decreases the barrier to 93 kJ/mol. Its influence of CH dissociation is minor and as a result the overall barrier for the formyl pathway is slightly preferred on the stepped surface. The consequences of these differences in reactivity are discussed for particle size dependence of the steam methane reforming reaction. Because this reaction is typically applied at relatively high temperature, the activation free energy of dissociative methane adsorption will be significantly higher than that of the surface recombination reactions that lead to CO formation. The particle size dependence observed experimentally, therefore, follows the changes in the activation of methane and this should be due to the increase of edge and corner sites with decreasing particle size. When C-O bond formation would have been rate controlling, a maximum in the rate of the methane steam reforming reaction as a function of decreasing particle size would have been predicted, because smaller particles will have fewer step-edge sites for CO recombination. Based on the metal-carbon and metal-oxygen binding energies the periodic trends of transition metals for the elementary reaction steps of the steam methane reforming reaction are compared. For highest catalytic performance both carbon and oxygen intermediates are required. The activation of methane and water can be related to the metal-carbon and metal-oxygen binding energies. Because of the requirement of optimal O coverage, the metal with the lowest barrier for methane activation (Ir) is not the metal with the highest reactivity in the methane steam reforming reaction (Ru). In Chapter 5, we compare the energetics of the dehydrogenation of CH4 to C on extended Rh(111) and Rh(211) surfaces and a planar (111) surface and the edge atoms that are shared between two (111) facets of a nanorod model (Figure 1b). We found that similar surfaces on the periodic and nanorod model have almost comparable adsorption and reaction energies. This is not the case for the adsorption of the carbon adatom, which adsorbs stronger on the (111) surface of the nanorod than on the periodic planar surface. This is attributed to the binding of the adsorbate to the more reactive edge atoms in the nanorod model. The reaction energies only differ slightly due to small geometrical differences. The dissociation of CH was on all surfaces the reaction with the highest barrier. However, due to the contribution of the entropic loss during dissociative methane adsorption, this step is the most likely rate controlling step. In accordance with the earlier comparison of CO recombination on different surfaces, the barrier for CH4 dissociation is strongly structure sensitive. The rate is much higher for surfaces that contain low-coordinated surface atoms. A microkinetic model of SMR has been constructed and is described Chapter 6. This model is based on the elementary reaction steps and allows us to determined macroscopic properties such as the overall reaction rate, the rates of the individual elementary reaction steps, surface coverages and rate control parameters. The reaction rate constants are computed from the DFT computed barriers and entropic contributions. Based on this model, we conclude that for both planar and stepped surfaces, the rate of dissociative methane adsorption is rate limiting. The rate on the stepped surface is higher than on the planar. It becomes obvious that to maintain high activity a balance in the rates of dissociative adsorption of CH4 adsorption, dissociation of H2O and formation of CO is required. Although methane dissociation is rate limiting, it is necessary to also generate sufficient O adatoms to remove the carbon-containing surface intermediates. For the planar surface, the surface contains sufficient Oads to remove Cads under typical reaction conditions. However, because the structure dependence of water adsorption is less strong than for methane dissociation, it is found that at high temperature on the stepped surface the reaction rate becomes very low due to Oads depletion leading to Cads poisoning. The most important finding is that the experimentally observed structure dependence is consistent with the proposal of the dissociative methane adsorption on low coordinated surface atoms as the rate limiting step. However that may be, to maintain high activity one needs step-edge sites to remove Cads and Oads with low barrier. The step-edge density, however, is not critical as the activation free energy barrier for CO recombination on these sites is many times higher than the activation free energy barrier for dissociative CH4 adsorption. This work provides a complete picture on Rh-catalyzed steam methane reforming in which all potential rate controlling steps, namely CH4 and H2O dissociation and CO formation, have been studied with the same accuracy and methods. It is also clear that a microkinetic model is required to understand the exact implications of these reaction energy parameters. The present data only take into account lateral interactions between adsorbed CO. Moreover, the model has only assumed the presence of one type of sites. In this respect, it would be important to carry out similar simulations by kinetic Monte Carlo modeling with the advantage of (i) easily incorporating lateral interactions, (ii) diffusion, which may be important since the site for Cads generation is different from Cads removal and, related, (iii) the presence of different sites in one particle, their proportion depending on the particle size. Another research direction could be to use Brønsted-Evans-Polanyi relations to predict periodic trends as well as the behaviour of alloys. Coking due to carbon-carbon coupling reactions, likely occurring on terrace facets, is another topic of future interest.
|Qualification||Doctor of Philosophy|
|Award date||24 Jan 2012|
|Place of Publication||Eindhoven|
|Publication status||Published - 2012|