Heterogeneous catalyst preparation by liquid-phase atomic layer deposition

Heterogeneous catalyst preparation has been performed for close to a century but has been rapidly evolving spurred by the evolution in synthetic techniques and growing needs in energy, chemicals and materials. An increasing number of catalyst synthesis strategies and tools have emerged in the past decades, including Atomic Layer Deposition (ALD). This technique allows the deposition of thin films with unmatched accuracy. Originally designed for flat substrates, this method has recently begun to be used for the preparation of heterogeneous catalysts. The development of ALD, both in academic research and industrial processing, is inhibited by in-herent high investment and operating costs of the technique. The global aim of this doctoral project was to design new ALD processes specifically designed for heterogeneous catalyst preparation. This aim is achieved in three steps: transposing gas-phase layer-by-layer growth to the liquid phase, matching the quality and versatility of gas-phase ALD in the liquid phase, and finally going beyond what is technically possible with gas-phase ALD set-ups. The first chapter of this manuscript describes current state-of-the-art methods used both in hetero-geneous catalyst preparation and atomic layer deposition on particles. This chapter does not aim to be a comprehensive review but to present the main research developments necessary to understand this thesis work in a broader context. The second chapter is based on a published article in which I describe a first step toward liquid-phase ALD. A procedure was developed for stoichiometrically limited layer-by-layer deposition of alumina in solution onto a copper-based catalyst. A conformal porous layer was obtained which al-lowed the diffusion of chemicals to the copper below. Additionally, the deposited layer was an ef-ficient barrier against irreversible deactivation during the catalytic hydrogenation of a biosourced molecule in a continuous flow reactor. The third chapter, based on a published communication, presents the fully optimized method for liquid-phase ALD. In this work, I demonstrate the self-limiting nature of the studied chemical reac-tions and use this process to deposit various type of inorganic layers (oxide, phosphate and sulfide) on a broad range of substrates relevant for catalysis. This technique also allows a direct monitoring of surface reaction, highlighting phenomena such as nucleation of clusters and their coalescence to form films. The materials were extensively characterized by electronic microscopy to confirm that the quality of the coating matched that of coatings obtained via gas-phase ALD. The fourth chapter describes soon to be submitted work where deposition is used during the early stages of nucleation on a surface to form targeted atomic clusters for catalysis. The goal was to build a catalytic cluster atom-by-atom to carefuly tune its chemical environment. Starting from a single Zn atom deposited on silica, I demonstrate the positive effect of Al on dehydrogenation catalytic activ-ity and the enhancement of stability brought by Si anchoring. These effects demonstrate that liquid-phase ALD can be more than just a tool for overcoating catalysts but also a method for the rational design of new catalysts. Finally, the thesis concludes with my perspective on new challenges and opportunities in the further development of liquid-phase ALD.

Développement d'un réacteur microstructuré basé sur des filaments métalliques catalytiques

Chrystèle Horny

The aim of this work is to develop a microstructured reactor based on filamentous catalysts for the Oxidative Steam-Reforming of Methanol (OSRM), to produce hydrogen as feed for a fuel cell, in an autothermal way. Hydrogen is produced by the methanol Steam-Reforming (SR) reaction. This endothermic reaction requires an external heat source which is, in our case, generated by methanol oxidation. The coupling of these two reactions – SR and oxidation, called oxidative steam-reforming of methanol – is performed in a single reactor. As the oxidation is much faster than SR, it occurs in the first part of the reactor, the SR takes place in the second part. If a conventional fixed bed reactor is used, pronounced axial temperature profiles are developed: a hot–spot due to the exothermicity of the oxidation is generated at the entrance followed by a cold–spot due to the SR. The high temperature may damage the catalyst and the low temperature diminishes the rate of reforming reaction leading to poor reactor performance. Thus the temperature control is crucial. Consequently, a microstructured reactor is used. This kind of reactor has multiple parallel channels with a diameter ranging from ten to several hundreds micrometers. These submillimetric dimensions lead to a high surface to volume ratio and a much higher heat transfer coefficient than in the traditional heat exchangers. These characteristics allow to increase heat exchange between reactions and to avoid hot–spot formation. In this work, brass wires introduced into a macro tubular reactor parallel to the walls are used to create the microstructure. Brass is chosen because of its composition – it contains copper and zinc catalyzing the reforming/oxidation of methanol – and for its high heat conductivity which ensures heat exchange improvements. Moreover the small diameter of reactor channels ensures narrow residence time distribution, leading to high selectivity, and a short residence time, improving reactors dynamic. The characteristics mentioned above are verified in chapter 4 for the reactor developed during this study. The hydrodynamic of this reactor is presented under the influence of wire diameter and catalyst preparation treatment. The flow in the microstructure is close to a plug flow: a Bodenstein number of 105 is obtained for brass wires with a diameter of 480μm. Concerning catalytic treatment, it doesn't appear to modify the hydrodynamic. Compared to a fixed bed reactor, the measured residence time distribution for our microstructured reactor is found to be much narrower; the pressure drops are also smaller. Chapter 5 focuses on the reaction conditions for SR of methanol, the reaction that generates hydrogen. These conditions are determined by using an industrial catalyst (based on copper –zinc – aluminium) and by comparing the catalytic activity measured in our conditions with the ones found in the literature. A molar water to methanol ratio of 1.2 is chosen in order to avoid carbon monoxide production, which is a poison for fuel cells. Chapter 6 deals with the development, the optimisation and the characterisation of brass based catalyst. Brass grids are first tested for SR: alloy composition, type and time of leaching are studied. The optimal catalyst is found to be a CuZn37 grid incorporated with aluminium and treated by an acid leaching during 20 minutes in order to increase its surface area (30m2/g). In order to test brass activity in presence of oxygen, partial oxidation of methanol is first carried out over this catalyst. Then the oxidative steam-reforming of methanol is studied: it is found that the modification of the support and hence of the treatment applied decreases the activity and essentially the stability. Deactivation is attributed to copper particles sintering and to oxidation of the catalyst which is not active for hydrogen production. A screening of additives is performed and it is shown that brass wires incorporated with aluminium and doped with chromium by an impregnation method is active, stable and selective: methanol conversion of 25.3% with a hydrogen selectivity of 42.6% are maintained during more than ten hours. Analyses indicate that this catalyst is not easily oxidised and reduced, due to the presence of spinels on the surface which modify electronic properties of copper. In chapter 7, the OSRM is analysed with focus on reaction mechanism and heat exchange. The reaction mechanism is quite complex due to reactions involved (partial oxidations, total oxidation, decomposition) and to their strong temperature dependence. A production of formaldehyde is observed at low temperature (T < 240°C), whereas at higher temperatures secondary reactions disappear and hydrogen production is initiated. Finally, measurements of the axial temperature profile allow to verify the isothermicity of our microstructured reactor: for a methanol conversion of 43% at T = 262°C a hot–spot of less than 3.5°C is measured and no cold–spot is observed. However, it is shown that the temperature profile is highly influenced by methanol conversion and by the oxygen quantity introduced in the reactor. Temperature variations in the entire reactor are nevertheless much lower than those developed in a fixed bed reactor, which is in agreement with our expectations and corresponds to our pursued objectives.

EPFL2005

Novel catalytic applications of carbon nanofibers on sintered metal fibers filters as structured supports

Marina Ruta

Supported metal catalysts are important from both an industrial and a scientific point of view. They are used, amongst others, in large-scale processes such as catalytic reforming, hydrotreating, polymerization reactions and hydrogenations. Often, these catalysts consist of nanosized metal particles deposited on a suitable support, which acts as an anchor for the active phase and, in several cases, contributes to improve the overall catalyst performances. The growth of carbon nanofibers on sintered metal fibers filters (CNF/SMF) by hydrocarbons catalytic decomposition results in a structured composite material presenting all the suitable characteristics for its application as catalyst support. The main objective of this thesis was to develop novel catalytic systems based on CNF/SMF as catalytic support for continuous gas-phase hydrogenations. At first, the synthesis of CNF/SMF was optimized to tailor the properties of the material for further applications as catalytic support. The use of ethylene as carbon precursor and high synthesis temperature (973K) provided high yields of well-ordered CNF with increased specific surface area (SSA), up to 516 m2/g. On the other hand, the synthesis of CNF by ethane decomposition, which is less reactive than ethylene, resulted in a support with higher permeability, minimizing the pressure drop and being more suitable in a continuous-flow reactor. After the synthesis, the CNF surface was activated by treatment with O3 or H2O2. A novel technique, which consist of XPS analysis after step-wise TPD in UHV conditions, was developed and applied for the surface characterization. It allowed to assess the nature of the functional groups created on the CNF surface. The O3-activation was found the most effective for increasing the acidity of the CNF/SMF surface, yielding a relatively high amount of carboxylic functional groups. These groups are important because involved in the chemical anchoring and stabilization of the active metal deposited on the support. Industrially, many hydrogenations are performed over supported metal catalyst. The majority of these reactions are known to be structure sensitive reactions, so the control of the metal particle size is crucial in order to study the catalyst activity and selectivity. We prepared monodisperse-sized Pd nanoparticles (8, 11 and 13 nm) via the reverse microemulsion method and deposited on CNF/SMF and activated-CNF/SMF for a two-fold study: the effect of the nanoparticles size and the effect of the support nature on the selective hydrogenation of acetylene. The antipathetic size dependence of the TOF disappeared at particle size bigger than 11 nm. The initial selectivity to ethylene (∼60%) was found size-independent. The structure-sensitivity relations have been discussed in terms of "geometric" and "electronic" nature of the size-effect and rationalized assuming a Pd-Cx phase formation which is known to be size-dependent. CNF/SMF supports with increased acidity diminished the formation of coke and changed the by-products distribution, diminishing the catalyst deactivation. Subsequently, CNF/SMF was applied to develop a new catalytic system: the Structured Supported Ionic Liquid Phase (SSILP) catalyst, embedding an active Rh complex. The selective gas-phase hydrogenation of 1,3-cyclohexadiene to cyclohexene was used as model reaction. The SSILP based on CNF/SMF supports showed a high turnover frequency, up to 250 h-1 and selectivity > 96%. Multiple advantages of supporting a homogeneous active phase, dissolved in IL, on CNF/SMF have been demonstrated: the thin layer of IL was homogeneously distributed over the mesoporous support, so mass transfer limitations could be avoided, furthermore the chemical inertness of the CNF prevented any interaction between the active phase dissolved in IL ensuring an efficient use of the Rh complex. The SSILP concept can be applied not only for the "heterogenization" of a transition metal catalyst, but also to obtain metal nanoparticles with monodispersed size via the reduction of a metal precursor dissolved in IL. In this case, IL provides a suitable environment for the nucleation and growth of the nanoparticles. By applying the SSILP concept to the selective hydrogenation of acetylene, the influence of IL on the stability of the nanoparticles was investigated. Moreover, the different solubilities of reactants and products in the IL phase allowed to improve the selectivity toward ethylene. Monodispersed Pd nanoparticles of 5 and 10 nm in SILP on CNF/SMF showed excellent long-term stability. The lower solubility of ethylene than acetylene in the IL had the beneficial effect of allowing high selectivity to ethylene up to 85%, due to the inhibition of its consecutive hydrogenation to ethane. In conclusion, a novel catalytic system for continuous-flow, gas-phase hydrogenations was conceived, taking advantage of the suitable characteristics of CNF/SMF supports. The SSILP concept, involving both homogeneous and heterogeneous active phases has been demonstrated, indicating a great potential and flexibility for continuous-flow industrial applications.

EPFL2008

Nitrous Oxide Abatement over Fe-Zeolite

Bryan Bromley

The use of zeolites for catalytic reactions is a field in continuous development. Since it was demonstrated that metal-zeolites are efficient catalysts for NOx abatement, decomposition of nitrous oxide (N2O) over Fe-containing zeolites has attracted great interest by the scientific community mainly due: N2O is the third most important gas contributing to global warming, its concentration in atmosphere is still on the rise and direct N2O decomposition to N2 and O2 is a sustainable alternative. N2O interaction with Fe-zeolite leads to the formation of "special" adsorbed oxygen (often called α-oxygen) with great potential for selective oxidation. Although the focus of a number of studies, the nature of the actives sites for the formation of this adsorbed oxygen is still unclear. The main objectives of this work are: 1) study the decomposition mechanism of N2O over isomorpously substituted Fe-ZSM-5 zeolite aiming on catalyst optimization and 2) optimization and further use of structured Fe-zeolite in novel micro-structured reactor. A number of refined transient kinetics methods were intensively used such as transient response, Temperature Programmed Desorption (TPD) and Temporal Analysis of Product (TAP). Density Functional Theory (DFT) and in situ IR spectroscopy studies were conducted in collaboration and our results support the following: In chapter 3, quantitative measurements have been carrier out to analyze the effect of temperature on the activation (under helium flow) and the irreversible deactivations that can happen during the pretreatment. The results obtained indicate that high temperature treatment (1273 K) results in the formation of a high amount of α-sites. The addition of a very low fraction of oxygen (2%) in the feed demonstrates that the catalyst is sensitive to its presence. Certain sites can be irreversibly destroyed, a result ascribed to the oxidation of active Fe(II) to Fe2O3 clusters which contain inactive Fe(III). Catalyst treatment at high temperature (1273 K) in He for up to 3 hours, also reduced irreversibly the concentration of α-sites. TPD measurements have shown desorption of oxygen at 3 different temperatures. Desorption at the lowest temperature (650 K) was assign to the α-oxygen recombination. The oxygen which desorbs at 700 K was associated to the oxygen from the deactivated sites. Oxygen which desorbs at the highest temperature (730 K) is originated from the surface NOx that desorbed as NO and O2. After a comparison of the concentration of the molecules accumulated, we propose that NO2 is the adsorbed specie. The mechanism of N2O decomposition on the binuclear oxo-hydroxo bridged extraframework iron core site [FeII(μ-O)(μ-OH)FeII]+ inside the ZSM-5 zeolite has been studied by combining theoretical and experimental approaches (see chapter 4). Rate parameters computed using standard statistical mechanics and transition state theory reveal that elementary catalytic steps involved into N2O decomposition are strongly dependent on the temperature. This theoretical result was contrasted with the experimentally observed steady state kinetics of the N2O decomposition and TPD experiments. As predicted by the theoretical study, the switch of the reaction order with respect to N2O pressure (from zero to one) occurs at around 800 K which confirms a change of the rate determining step from the α-oxygen recombination to α-oxygen formation. The proposed mechanism was also confirmed by O2-TD which confirmed the viability of the binuclear complex. The α-oxygen recombination at low temperature (

EPFL2010