UC Berkeley

The field of heterogeneous catalysis has long been interested in understanding the role of site structure on reactivity and selectivity for the rational design of catalysts. Vanadia is of particular interest because of its potential to be highly active and selective for a variety of reactions, such as oxidative dehydrogenation of alkanes to alkenes, or the oxidation of n-butane to maleic anhydride.

When supported, vanadia can exist in several environments depending on its surface coverage. At the lowest loadings, below 2 V/nm², the vanadium exists predominantly in well dispersed, tetrahedral structures with 3 V-O-support bonds and 1 V=O bond. At higher loadings, above 2 V/nm², V-O-V bonds form on the surface, and at loadings above 7 V/nm², the vanadia begins to form 3 dimensional domains of V₂O₅. Methanol oxidation rates over catalysts with varying vanadia loadings have shown no significant effect of the V surface density on the formaldehyde formation rate. However, significant differences in the formaldehyde production rates are observed for different supports. Changing the support from silica to titania or zirconia, will result in increases in the production of formaldehyde from methanol by over an order of magnitude for similar vanadia surface coverages. These differences in rate are observed even though the reaction mechanism is believed to be the same regardless of the support. The mechanism is thought to proceed as follows. First methanol dissociatively adsorbs across a V-O-support bond, producing V-OCH₃ and M-OH (M = Si, Ti, Zr, Ce) in a quasi-equilibrated step. Next a surface oxygen abstracts hydrogen from the methoxy group in the rate determining step, and formaldehyde desorbs. The final steps are fast and involve the production of H₂O from neighboring hydroxyls and the reoxidation of the catalyst by gas-phase oxygen.

When vanadia is supported on bulk TiO₂, ZrO₂, or CeO₂, the support surface area is relatively small (~200 m²/g at its highest), and the bulk support causes side reactions which make it difficult to understand the role of the vanadia. Furthermore, by using a bulk support, only the vanadia surface coverage can be varied, such that the effect of different V structures can be elucidated, but not that of V-O-M bonds. Therefore, high surface area silica with a variable coverage of two-dimensional TiO₂, ZrO₂, and CeO₂ layers are used to support isolated vanadate structures to vary the quantity of vanadia bound to the modifying layer. These bilayered catalysts can be used to determine the effect of the V-O-support bonds on the formaldehyde production rate.

Three mesoporous silica supports, MCM-48 (1550 m²/g), MCM-41 (1353 m²/g), and SBA-15 (700 m²/g) were used as the high surface area silica. Ti was grafted to the MCM-48 surface using Ti(OiPr)₄ and a maximum surface coverage of 2.8 Ti/nm² was obtained after 3 graftings. The grafting of zirconium was performed using Zr-2-methyl-2-butoxide on MCM-41, and a maximum loading of 2.1 Zr/nm² was achieved after 3 graftings. The final modifying element, cerium, was grafted onto SBA-15 using Ce(OtBu)₄ for a maximum surface coverage of 0.9 Ce nm^-2. After treating the MO_x/SiO₂ (M = Ti, Zr, Ce) samples in air to remove any organic ligands, OV(OiPr)₃ was grafted onto the MO_x/SiO₂ (M = Ti, Zr, Ce) support to achieve the desired V surface coverage of approx. 0.7 V/nm².

The resulting catalysts contain amorphous two-dimensional layers of TiO₂, ZrO₂, or CeO₂ with V existing in a pseudo-tetrahedral structure on the surface. As the surface density of the modifying element layer increases, the quantity of vanadia bound to TiO₂, ZrO₂, or CeO₂ increases. For the VO_x/ZrO₂/SiO₂ catalysts, the fraction of vanadia bound to the zirconia layer was able to be quantified and determined to be 35% of all V for a Zr surface density of 2.8 Zr nm^-2.

Even for small quantities of modifying elements (0.2 M nm^-2), the apparent rate constant for formaldehyde production on VO_x/MO₂/SiO₂ (M = Ti, Zr, Ce) is an order of magnitude higher than for VO_x/SiO₂ catalysts at the same V surface density. Regardless of the modifying element used, the increase in apparent rate constant is comparable for all catalysts. As the modifying element surface density is increased, the apparent rate constant also increases, which is a result of an increasing fraction of V bound to the MO_x layer. Each of these bilayered catalysts can be described using a two-site model of VO_x/SiO₂ and VO_x/MO₂ (M = Ti, Zr, Ce) with the latter being responsible for the increased apparent rate constant.

This higher activity for the VO_x/MO₂ site is due to a lower apparent activation energy. For VO_x/SiO₂, the apparent activation energy is 23 kcal mol^-1, but is approximately 17 kcal mol^-1 for VO_x/MO₂ sites. The apparent activation energy can be expressed as the sum of the heat of methanol adsorption and the activation energy for H-abstraction. My results indicate that the lower apparent activation energy observed for the bilayered catalysts is a result of a decrease in the activation energy for H-abstraction. This lower energy pathway occurs because the MOx layer can abstract H from surface methoxy groups. For VO_x/SiO₂, however, the vanadyl oxygen abstracts H in a higher energy step.