TOWARDS THE DESIGN OF MIXED METAL (OXIDE) CATALYSTS AND LIGHT-RESPONSIVE LIGANDS USING QUANTUM CHEMISTRY CALCULATIONS

Abstract

Traditional chemical engineering approach of designing materials and chemical processes involved heuristic design rules and rigorous experimentation. In the last few decades, however, there has been a paradigm shift in this approach and numerical experiments performed on computers are drastically reducing the number of actual experiments that need to be performed to obtain materials of desired properties and optimal process conditions. More importantly, such numerical experiments may often provide high resolution insight of the system under consideration that is otherwise difficult to achieve in actual experiments. As a case in point, computational fluid dynamics (CFD) simulations of a continuous stirred tank reactor (CSTR) now easily provides the detailed velocity profile inside the reactor that may otherwise be only attained using highly sophisticated and expensive flow visualization techniques. Such insight may then be used to design the optimal size, shape, and operating conditions of the CSTR. It is also now possible to combine CFD simulations with mass transfer/heat transfer/chemical kinetics models to study coupled, ‘multiphysics’ problems typical of chemical engineering applications. Though highly powerful, above mentioned approach still require experimental determination of model parameters, e.g., fluid viscosity, density, diffusion coefficient, etc. Also, they typically involve the continuum approximation and therefore does not provide molecular-level insights. Atomistic modelling techniques, e.g., Monte Carlo (MC) and Molecular Dynamics (MD) simulations bridges this gap to a substantial extent, as they capture the molecular configurations/trajectories of individual atoms in a molecular system and use it to derive thermodynamic and dynamic properties of the system. For instance, viscosity, density, and diffusion coefficient of a material may be obtained using MC/MD simulations on the basis of the knowledge of their molecular structures alone, without the need of experiments. Extensive work has been done in last few years to use MC and MD to obtain the properties of materials as a surrogate to actual experiments. Atomistic simulations do not invoke the continuum approximation and employ ‘force fields’ that incorporates various forces between atoms of molecules. However, there are two problems of interest to chemical engineers where atomistic simulations are also insufficient. First, since atomistic simulations consider atom as the smallest entity and chemical reactions involve electrons, chemical reactions cannot be typically captured in atomistic simulations. With the exception of some recent developments (e.g., those using ReaxFF force field), the chemical identity of a molecule cannot be changed in the course of an atomistic simulation and only a configurational change in the molecular structure is possible Second, ‘force fields’ are not available for many molecules of interest, especially in diverse solvent environments and under the influence of external fields. Further, the development of force fields for atomistic simulations entail either the use of experimental data or the use of quantum chemistry calculations. In this thesis, I have worked on two representative case studies on quantum chemistry calculations motivated by experimental studies on (1) mixed metal and mixed metal oxide catalysts formed by Cu, Zn, and ZnO and their use in glycerol hydrogenolysis, (2) self-assembly of nanoparticles functionalized with light-responsive azobenzene dithiol (ADT) ligands. My primary focus is on first case study that is further divided into three objectives. The second case study forms the fourth objective of this thesis. The specific objectives of this thesis are the following: 1. To study the structure, stability, and glycerol/hydrogen adsorption behavior of Cu, Cu-Zn and Cu-ZnO clusters. 2. To analyze glycerol hydrogenolysis (glycerol to 1,2-Propanediol) reaction mechanism over Cu/Cu-Zn/Cu-ZnO clusters. 3. To understand the role played by MgO support on the catalytic activity of Cu/Cu-Zn/Cu-ZnO clusters for the glycerol hydrogenolysis reaction. 4. To obtain the intermolecular interactions that drive the light-induced self-assembly of ADT molecules in methanol-toluene mixture. In recent years, a large amount of glycerol (>1 million tons per year) is being produced by the biodiesel industry, which has given rise to the pressing need to utilize glycerol. One possible strategy is to convert glycerol into value-added products such as 1,2 propanediol (1,2-PDO) that is useful for the production of pharmaceuticals, cosmetics, polyester resin, antifreeze agent and paints, etc. In recently reported experiments, mixed metal and mixed metal oxide catalysts of copper (Cu) and zinc (Zn) have been shown to be effective as a catalyst for glycerol to 1,2-PDO. Conversion of glycerol to 1,2-PDO using these catalysts has been hypothesized to occur in two steps: (1) condensation (water removal) of glycerol to form acetol, and (2) hydrogenation (hydrogen addition) of acetol to form 1,2-PDO. An alternative mechanism is that the glycerol is first hydrogenated to form ethylene glycol. Interestingly, while pure Zn is much less effective than pure Cu as a catalyst for glycerol hydrogenolysis, some doping of zinc in copper is found to improve the catalyst performance. In this study, quantum chemistry calculations are performed to understand the reason behind this nontrivial behaviour and to further determine the optimal Cu/Zn composition for this reaction. The computational study of Cu/Cu-Zn/Cu-ZnO catalysts using quantum chemistry calculations are performed in four steps. In the first step, small clusters of Cu, Cu-Zn, and Cu-ZnO (up to 10 atoms) are simulated using Gaussian 09 software. Since the number of atoms that can be accurately simulated within reasonable computer time is limited, calculations performed on small clusters provide a feasible alternative to understand the effect of Zn/ZnO doping. The minimum energy structures of Cu-Zn and Cu-ZnO clusters are found by doping minimum energy pure Cu clusters with Zn atom(s) and ZnO molecule(s), respectively, followed by energy minimization of the resultant clusters. The physical stability of the clusters of different compositions are inferred from an analysis of the average bond length, binding energy and dissociation energy, and the second-order difference of the total energy. Chemical stability and the chemical activity of the clusters are also evaluated using the HOMO (highest occupied molecular orbital)-LUMO (lowest unoccupied molecular orbital) energy gap, and chemical hardness defined as the difference between the ionization potential and electron affinity. These properties are determined for bare clusters without any chemical reaction. Odd-even alteration in properties that determine the cluster stability/activity is observed with cluster size, which may be attributed to the presence/absence of unpaired electrons. The difference in behaviour between Zn/ZnO doping can be interpreted in terms of charge transfer between atoms. Charge transfers from Zn to Cu in the Cu-Zn clusters and from Cu and Zn atoms to O atom in Cu-ZnO clusters, which implies that the Cu atom acts as an electron acceptor in the Cu-Zn clusters but not in the Cu-ZnO clusters. In the second step, the adsorption energies of glycerol and hydrogen on Cu-Zn/Cu-ZnO clusters are computed in the context of the use of Cu-Zn/Cu-ZnO catalysts in glycerol hydrogenolysis. Glycerol adsorption is generally found to be more energetically favourable than hydrogen adsorption. Dual-site glycerol adsorption is also observed in some of the planar clusters. In the third step, the reaction path for glycerol to 1,2-PDO conversion is studied by finding the transition states and activation barriers of individual reaction steps for different compositions of the catalyst. Calculations are first performed for reactants in gas phase (non-catalyzed system) and reactants in gas phase with 3-atom Cu cluster (catalyzed system). We demonstrate that glycerol to ethylene glycol conversion is preferred in the non-catalyzed system, but glycerol conversion to 1,2-propanediol via 2-acetol intermediate is preferred in the catalyzed system. We next analyze the adsorption energy of the reactant and product species involved in the glycerol to 1,2-PDO reaction on 8-atom Cu cluster and Cu cluster doped with a Zn atom or a ZnO molecule. Finally, we study the effects of Zn or ZnO doping on the activation barriers of the two steps of the glycerol to 1,2-PDO reaction. In the fourth step, plane wave calculations are performed in Quantum Espresso with Cu-Zn/Cu-ZnO clusters on MgO surface, in order to decipher the role played by MgO support. Fundamental insights obtained in this study can be useful in the design of Cu-Zn/Cu-ZnO catalysts. Photo responsive materials can switch their magnetic, optical and structural properties when exposed to visible/UV light. One promising example is the azobenzene dithiol (ADT) functionalized gold nanoparticles (NPs) in different compositions of methanol-toluene solution. When ADT ligand is exposed to visible/UV light of different wavelengths, the molecular configuration changes from cis to trans or vice versa. ADT possess a net dipole moment in the cis state that results in the self-assembly of functionalized NPs due to dipole-dipole attraction. The dipole moment vanishes in the trans configuration, resulting in a reversible nature of this self-assembly. The self-assembly behavior further depends on the composition of the solution, as the changes in dielectric behavior of solution affect the dipole-dipole interactions and other solvophobic interactions. In this study, we use quantum chemistry calculations to study the behavior of isolated ADT molecule(s) in different compositions of methanol-toluene solution to gain further insights into this self-assembly phenomenon. We performed DFT calculations to study the ligand-ligand interactions that drive this self-assembly process. The dipole moment of individual ADT molecules and interaction energy between a pair of ADT molecules are computed for various compositions of methanol-toluene mixture used as the solvent. Dipole moment of the mixture increases with methanol % up to 20% methanol, followed by lesser change on further addition of methanol. Trans-cis isomerization results in a shift of the position of energy minimum on the interaction energy against separation curve. The scaling of effective interaction between nanoparticles against separation for different ADT loading are also estimated from the results of DFT calculations.