UC Berkeley

The selective oxidation and ammoxidation of light alkenes underpins the 10 billion pounds per year acrylics industry, and is among the essential processes for preparing monomers and intermediates. The most effective catalysts known are multi-component oxides based on bismuth molybdate, and were first patented in the 1950s. A significant body of research has been done since then to understand the mechanism, yet a surprising number of questions remain unanswered. The aim of this thesis is to identify "descriptors" relating the activity of i) different catalysts to their material properties and ii) different reactants to their physical properties in order to systematically guide catalyst design for greater activity for oxidation reactions.

A systematic investigation of the kinetics of propene oxidation to acrolein was carried out over Bi1-x/3V1-xMoxO4 (x = 0 - 1). X-ray absorption near edge spectroscopy (XANES) was used to determine the oxidation state of Bi, Mo, and V before and after exposure of the catalyst to propene at 713 K. It was established that, contrary to previous discussions of the mechanism of propene oxidation on Bi1-x/3V1-xMoxO4, Bi remains in the 3+ state and only V and Mo undergo reduction and oxidation during reaction. The kinetics of propene oxidation were investigated to establish the activation barrier for acrolein formation and determine how the partial pressure dependences on propene and oxygen change with the value of x. The data obtained from this study were then used to propose a generalized model for the kinetics of propene oxidation over Bi1-x/3V1-xMoxO4. This model is fully consistent with our findings regarding the reducibility of the metallic elements in the oxide. According to this model, vanadium and molybdenum are randomly distributed to form three types of sites each associated with its own rate parameters. Mo-V sites are found to exhibit the highest activity. The proposed model provides a good description of the experimental data for all catalyst formulations examined, for a range of propene and oxygen partial pressures, and for temperatures above 653 K.

UV-Vis spectroscopy revealed that the rate of propene oxidation to acrolein correlates with the rate of catalyst reduction, suggesting that absorption edge energy or catalyst band gap might be a good descriptor of activity. Further work demonstrated that the apparent activation energy for the oxidation of propene over scheelite-structured, multicomponent mixed metal oxides (Bi3FeMo2O12, Bi2Mo2.5W0.5O12, and Bi1-x/3V1-xMoxO4) correlates with the band gap of the catalyst measured at reaction temperature. The relationship between band gap of the oxide and the activation energy for propene activation can be rationalized by examining a Born-Haber cycle relating the electronic excitation observed during the band gap measurement to an electronic excitation taking place in the transition state for propene activation. We also found that the change in band-gap energy with composition arises from the interplay between the sizes and energies of the V 3d, Fe 3d, Mo 4d, and W 5d orbitals, which give rise to the lowest unoccupied crystal orbitals. Subsequent work showed that the correlation of the apparent activation energy with the catalyst band gap also holds for metal oxides having an aurivillius structure, Bi4V2-xMoxO11+x/2 (x = 0 - 1). The band gap was also found to be a good descriptor of acrolein selectivity. Catalysts with band gaps above ~ 2.1 eV exhibit intrinsic selectivities of ~ 75%, whereas the intrinsic acrolein selectivity rapidly decreases for catalysts with band gaps below ~ 2.1 eV. Catalysts with band gaps below ~ 2.1 eV also promote acrolein combustion, whereas catalysts with higher band gaps do not.

A systematic investigation of the oxidative dehydrogenation of propane to propene, 1- and 2-butene to 1,3-butadiene and the selective oxidation of isobutene to methacrolein was carried out over Bi1-x/3V1-xMoxO4 (x = 0-1) with the aim of defining the effects of catalyst and reactant composition on the reaction kinetics. This work revealed that the reaction kinetics can differ significantly depending on the state of catalyst oxidation, which in turn depends on the catalyst composition and the reaction conditions. When the catalyst is fully oxidized, the kinetics for the oxidation of propene to acrolein and isobutene to methacrolein, and the oxidative dehydrogenation of propane to propene, 1-butene and trans-2-butene to butadiene are very similar - first order in the partial pressure of the alkane or alkene and zero order in the partial pressure of oxygen. These observations, together with XANES and UV-Vis data, suggest that all these reactions proceed via a Mars van Krevelen mechanism involving oxygen atoms in the catalysts and that the rate-limiting step involves cleavage of the weakest C-H bond in the reactant. Consistent with these findings, the apparent activation energy and preexponential factor for both oxidative dehydrogenation and selective oxidation correlate with the dissociation energy of the weakest C-H bond in the reactant. As the reaction temperature is lowered, catalyst reoxidation becomes rate-limiting, and the transition to this regime depends on the ease of catalyst reduction and effectiveness of the reacting hydrocarbon as a reducing agent. A third regime is observed for isobutene oxidation at lower temperatures, in which the catalyst is more severely reduced and oxidation now proceeds via reaction of molecular oxygen, rather than catalyst lattice oxygen, with the reactant.