Surface phenomena associated to transition metals nanocatalizersa computational study

Zusammenfassung

Nanotechonology, in all its varieties, from pure industrial applications to nanobiomedicine, is one of the most exciting and promising fields in present scientific research. This thesis focus in the use of Theoretical and Quantum Chemistry in specific themes of commercial, social and health interest, and how it can not only answer questions and propose solutions but also probe deeper than the experimental tools at present disposal. The surface phenomena studied in the research body of work in this thesis can be summarized to an expression: mini(aturized) power. Either in portable electronic devices or in in vivo biosensors, sustainable miniaturized fuel cells are a key concern in our technological society. The elusive aspects of the reactions in the electrocatalysts makes the approach of using methodologies of Computational Chemistry not only logic but inescapable. In particular, for this thesis, they are used to explore the particular aspects of pre-activation of a catalytic material or proposing reaction pathways. Therefore, regarding the themes in this thesis, and starting by the reactivity of nano gold clusters and/or surfaces, which allows their use as nanocarriers and electrodes in miniaturized bio fuel cells, the first theme concerns the electrooxidation of glucose in gold surfaces. Since the early sixties, when reactivity of nano gold surfaces was discovered, the prospective use for the glucose electrooxidation with gold as catalyst was immediately. One of such possible application was in bio fuel cells and, at present, there are already several promising prototypes of bio fuel cells using glucose as fuel source and gold in the electrodes. In biomedical context a biofuel cell converts chemical energy stored in, for example, glucose, into electricity. Glucose as fuel source has commercial advantages as it is cost efficient, due to its abundance in the Biosphere and as additional positive feature, is a clean source of power. In simple terms, the power from glucose oxidation/oxygen reduction that characterizes glucose fuel cells can power bioelectronic devices, such as in vivo glucose biosensors. The main challenges for bio fuel cells aimed at bioimplants or biomaterials in general still remain the low power densities, the short lifetimes and operating capacity in neutral pH. In this research crossroad, computational data is fundamental to ensure safe and accurate arrival to destination. Computational approaches and tools can probe possibilities still beyond the experimental capacity, and bring back new hypothesis to be tested in laboratories and industry. In fact, being able to study chemical parameters of a reacting system at micro level or even nanoscale is feasible within computational studies, as long as the choice for model and methodology is validated to the problem and system under scrutiny. Quantum methods depart from the solving of the Schrodinger wave function equation for a given reacting system, but in multiple-particle reacting systems, approximation methodologies need to be applied to this process of determining a solution for the Schrodinger equation. The chosen formalism in the results presented in this thesis is the Density Functional theory (DFT) where a density functional represents the electron density. DFT methods vary with the density functional considered. Though DFT is usually considered an Ab initio method, i.e, a method that solves the Schrodinger equation without inclusion of empirical or semiempirical parameters in the equation, the fact is most common density functional have empirical parameters, combining the density functional exchange term with that derived from the Hartree-Fock approximation, leading to the so called hybrid density functional. The Hartree-Fock approximation considers that each electron in the chemical system is moving under an average potential created by all the others. Therefore, the electronic Coulombic interaction is approximated by a mean field and the electronic exchange interaction is included using a Slater determinant, for including the antisymmetry effect in the system wave function. The Hartree-Fock is not the first or last approximation made in DFT in order for solving the Schrodinger equation in general, an in DFT context in particular. However, DFT is efficient in the treatment of may-electrons systems, which makes the formalism very interesting when studying systems for prospective technological development. The total electronic charge density corresponds to the ’density’ in DFT. departing from the Hohenberg and Kohn theorems, and through the Kohn-Sham formulation, the initial many-electron system is replaced by a fictitious system of non-interacting electrons but which giver the exact electronic density of the original system. With this, the total energy of the system can be expressed as function of the electronic density, and is the sum of several terms, namely the kinetic energy of the Kohn-Sham system, the external potential acting on the interacting system, which has nuclei-electron interaction, electronic Coulombic interaction and an additional term, the exchange correlation functional, which includes the non-classical electron-electron interaction energy and the difference between the exact kinetic energy and the kinetic energy of a non-interacting electron gas. In order to determine the unknown exact exchange-correlation functional, several approximations are made, suc as the Local Density Approximation (LDA), which considers that in each point, the exchange-correlation energy only depends of the local density of the point. This was the first approximation allowing DFT to be feasible in computational research but other followed, such as General Gradient Approximation (GGA), which add to LDA the gradient in each point. The most classical exchange-correlation fucntionals come after these approximations: Perdew-Burke-Ernzerhof (PBE), Perdew-Wang 91 (PW91), Becke3-Lee-Yang-Parr (B3LYP) or even the Revised Perdew-Burke- Ernzerhof (RPBE). Other approximations followed, other exchange-correlation functional were developed, and the computational tools evolved. At present, there is little question to the importance or advantages of using DFT in research in subjects such as the determination of reaction pathways or the nano geometric parameters in catalysts. More in-depth explanation regarding DFT and Computational approaches is presented in Section A, specifically in Chapter A2. In the same section, but in the previous chapter, A1, attention is given to models. Methodologies are related with accuracy while models are related with realism. In the path of research for this thesis, the possibilities of nano gold as nanocarrier, namely in transdermal diffusion, were also taken in consideration and Chapter A1 starts with a description of the mathematical models regarding dermic diffusion through stratum corneum, as basic research in that context. Due to the applications in health sciences, transdermal technologies are an increasing field of interest, and in the particular dermis outermost layer, the stratum corneum, there are several mathematical models available to explore in different contexts and for a more inclusive perception of the experimental data. The models range from the ones based on Fick’s First Law, which uses the concept of passive diffusion to describe drug delivery through skin and is more adequate in contexts where the Scaled Particle Theory Models may not be as ideal, and vice-versa. In fact, the mathematical modeling of the skin has made remarkable progresses in the last decades and the accuracy of the models can be tested with new computer software and capacities. The skin permeability, itself, can be described through three parameters, (1) partition, which involves partitioning between multiple phases, (2) the partitioning coefficient, also called the permeability coefficient and can be used to predict the mass of drug penetrating the skin, and (3) the path length. Though it would be easier to use a single theory applicable to any type of drug delivery system, the fact is there is a broad spectrum of models, each applicable to not only specific delivery devices but skin models. Chapter A1 summarizes and explains these models, alongside the cluster and periodic slab models used in DFT calculations. In Section B, results by means of DFT based methods regarding studies of the interaction between several ionic and radical adsorbates and low-index Miller gold surfaces are presented and explored. One can divide these adsorbates in three categories: (1) Radical oxigen and hydrogen; (2) ions O-, O2-, H+ and H-; and (3) the radical hydroxyl and the anion hydroxide. The studied adsorbates are crucial in many electrochemically and heterogeneously catalyzed reactions, as in the case of electrooxidation of glucose in gold surfaces. Category (1) of adsorbates was also use to validate the DFT methodology regarding the study of the adsorbing surface + adsorbate, as it compared same methodology in different physical realities, i.e., low coverage surface versus high coverage surface (using the same exchange-correlation functional, PW91, for equal adsorbate interaction in two different models, cluster versus periodic slab model) and same model with different methodology (when studied oxygen and hydrogen adsorption in the cluster model with different exchange-correlation functional, B3LYP and PW91). The computational results and the adequate choice of the model were then validated by experimental results and coherence between the obtained theoretical data. The study of interaction of Au(hkl) and the adsorbates of categories (2) and (3) ensued. Regarding the most relevant results of this study, there are several aspects to note. When considering the three studied gold surfaces, Au(111), Au(110) and Au(100), the results of Au(110) and (100) indicate little to null sensitivity to the change represented by a different physical reality (low coverage cluster model versus high coverage periodic slab model), as their respective order of adsorption preferences for the different adsorbates is unaffected. Au(111) surface, however, evidenced more sensitivity, as in the case of hydroxyl, for example, presenting other adsorption site as the preferred (hcp site in the cluster model versus the bridge site in the periodic slab model). Nevertheless, the order of more favorable adsorption energetics for the hydroxide and hydroxyl, for both periodic and cluster calculations, on the three low-index Miller gold surfaces is equal and in accordance with experimental data. The order, from the most favorable to the least favorable is Au(110) > Au (100) > Au(111), is in concordance with the decreasing of the coordination number of the gold atoms in each surface. The short-bridge site of the Au(110) is consistently predicted as the most stable adsorption site for all studied adsorbates and within both models, with the exception of the oxygen radical in the periodic slab approach. However, in the Au(110) the difference of approximately 0,08 eV between the short-bridge site and the preferential site (hollow-3-fold site) for the oxygen adsorption in periodic slab approach indicates both sites to be evenly populated at low coverage. More interestingly, the most energetically favorable surface, Au(110), presents a variability in the order of adsorption sites preferences for the different adsorbates that is inexistent in the lest energetically favorable surface, Au(111), and only partially present in the Au(100) surface. Taking the research further with analysis of the crystal orbital overlap populations (COOPs) for the several studied sites of the radical and ionic adsorbates, relation between the specific site bond and population distribution indicate a geometric factor as possible cause for the variability in Au(110). In fact, as concluding hypothesis, it is considered the variability in the adsorption order of preferences of the Au(110) and Au(100) may not only result from a geometric factor but indeed also be function of the site’s coordination number and the adsorbate charge. Within the theme of Electrochemical Reactions For Fuel Cells, the research then continued from the exploration of the glucose electrooxidation in gold nano surfaces to the quest for determining the reaction pathways of the electrooxidation of methanol in direct methanol fuel cells context. DMFC are among the strongest options under current development for adequate power sources in the market of portable devices. Though DMFC catalysts are mostly platinum-based, as this metal outperforms in the key areas of activity, selectivity and stability, within methanol oxidation framework, it nevertheless, presents a problem, as platinum is poisoned with by-products of the methanol oxidation. Ruthenium-platinum alloys are the preferred catalysts at present to face this situation, and indeed, solo ruthenium also presents to be a viable catalyst for this reaction. Section C is exclusively devoted to the analytical and computational research regarding methanol electrooxidation in DMFC context and with ruthenium as electrocatalyst in particular. The fact is the reaction pathway still is elusive in many aspects and this section approaches the problem with the following strategy: (1) reviewing of significant experimental data, regarding methanol electrooxidation with platinum-ruthenium and ruthenium catalysts; (2) computational calculations and studies regarding methanol electrooxidation in solo Ru(0001); (3) reviewing significant computational data regarding ruthenium-platinum, ruthenium and platinum, in transversal analysis with the data from (1) and (2). The main goal of proposing reaction pathways for the methanol electrooxidaton in DMFCs context is achieved, as integration of experimental data adjusts the computational results and provides a fluid and likely explanation for the mechanism of methanol decomposition. Regarding the original computational calculations, the several possible pathways for the methanol decomposition in clean Ru(0001) are examined by means of DFT methodologies, departing from three initial and mutually exclusive bond scissions, i.e, the initial boond scission could be either C-H, C-O and O-H bond scission. Data regarding adsorption and coadsorption of the several reactants, intermediates and products of all possible pathways was obtained. The termochemistry and the reaction barriers of the elementary steps was also studied. The four possible pathways under studied (arbitrarily named Blue Route, Red Route, Black Route and Green Route) were constructed from the experimental data reviewed. Among several conclusions, the pre step of the four possible pathways, methanol adsorption was found to be of weak nature, implying reasonable rate of sufficient desorption, and thus compromising the effective methanol decomposition. Also, the adsorption interactions seem related with the geometry of the studied chemical species and three categories could be defined: (1) planar structures interact with ruthenium through the C−O bond; (2) tetrahedral type structures interact with the metallic surface through the oxygen atom; and (3) chemical species presenting angular and linear geometry adsorb on Ru(0001). These categories apply in the co adsorption studied scenarios. Further, in co adsorptions, there seems to be a general preference in positions pairing hollow-type sites and for larger fragments to interact through their carbon atoms. This kind of co pairing maximizes the adsorption interaction of the chemical species with the surface by creating a larger amount of adsorbate-surface bonds, and, therefore, even considering diffusion, the results indicate high probability of the reaction occurring in these sites. The studied co adsorbed pairs of big fragment + hydrogen atom reached in general very expressive negative adsorption energies. Either considering solo or co adsorption data, the highest interaction energy for adsorbate(s) + surface corresponds to the shortest distance between the adsorbate and the nearest superficial metallic atom. The Blue Route, considering the computational results for Ru(001), is the most likely pathway for the methanol decomposition, because: (1) includes an experimentally detected intermediate byproduct of the reaction on ruthenium, the radical methoxy; (2) with exception for the step between methanol adsorption on ruthenium and initial methanol O-H bond scission, all the steps of this route consistently have higher adsorption energies in comparison with the forward activation energy barriers, thus making this route the most energetically favorable; (3) computational data explain experimental detection for methoxy dissociation into formaldehyde and lack of experimental detection of formaldehyde dissociation products; (4) it is the most kinetically favorable of all four possible pathways, namely for temperatures of 220 and 340 K, and even includes the fastest step of the several routes. Other relevant results are the indication that, in a certain temperature range, the Black Route (with initial methanol C-H bond scission) could be active as a minority reaction pathway and finally, the possible hydrogen formation from the hydrogen atoms obtained in the methanol decomposition on clean ruthenium was also studied and the reaction presents a quite low activation energy barrier, which implies molecular hydrogen formation as possible in these conditions. Following the original computational study, then the transversal analysis with other computational studies and the already reviewed experimental data was made. As the aim was analytical integration, rather than exhaustive enumeration, the transversal analysis was surgical rather than extensive. In result, a sound scientific and coherent proposal for the reaction mechanism was achieved for ruthenium-platinum, ruthenium and platinum catalysts. There are two mutually exclusive possibilities, an initial O-H bond scission via ruthenium catalysis, from which the Blue Route would follow as the likely pathway, and an initial C-H bond scission via platinum catalysis, from which the three other pathways (Red Route, Black Route and Green Route) are possible. In fact, experimental data regarding studies in clean Ru(0001), for temperatures between 160 and 180 K, support a methanol electrooxidation via ruthenium with initial O-H bond scission, while the computational data accounts not only for the detected and undetected intermediates but also indicates a very favorable kinetics. The platinum possibility, with initial C-H bond scission, concurs with experimental and computational data for clean Pt(111) surfaces, and in fact there is even a microkinetic model for two initial C-H bond scissions when the surface for catalysis is platinum. Finally, there is also the possibility of the mechanism in platinum-ruthenium surfaces to be bufunctional and further studies have to be made in order to determine the prevailing pathways when both metals are present in the electrocatalysts. To conclude this summary, it should be noted the approach made in methodological choices. In Section B, regarding adsorbate – gold surface interaction, within a glucose electrooxidation context, the choice was comparison of methodologies, i.e., the pre-phase of validating the chosen model for the study of the interaction between ionic adsorbates and gold surfaces, explored effects of model and DFT methodologies, sustaining the obtained data for the ions. In the methanol electrooxidation, future works implies not only the transversal analysis of experimental versus computational data, as was made, but also a conjugation of methodologies, departing from the computational results of methanol decomposition on ruthenium surface, specifically, from the optimized structures of the several steps and use them in new studies and theoretical approaches, which can explore the stage of bond formation and breaking during the reaction pathways within the reacting system and with the ruthenium surface. Another avenue is using those results in microkinetic models, ensuing a more in-depth study regarding the kinetics of the reacting system.