(583an) Oxidation of Propene Over Bismuth Vanadate-Molybdates: Insights From Density Functional Theory

For over 60 years, catalysts based on bismuth molybdate have been used for the selective oxidation and ammoxidation of propene to acrolein and acrylonitrile. However, until recently, relatively little was known about the mechanism by which propene oxidation takes place on bismuth molybdates. Previous researchers have established that the initial and rate-determining step (RDS) involves abstraction of hydrogen from the methyl group of propene, and that lattice oxygen in the catalyst is responsible for both hydrogen abstraction and oxygen insertion steps. [1] However, the nature of the active site, the identities of the post-RDS reactive intermediates, and the roles of bismuth and molybdenum in enabling catalysis have until now not been clearly established. In addition, catalysts used in industrial practice contain a variety of promoter elements that are known to improve activity or selectivity; however, there is not yet a clear understanding of the means by which these promoter elements influence catalytic performance.

We have addressed these questions by applying a variety of methods from density functional theory to calculate the mechanism and active site requirements for propene oxidation in bismuth molybdate [2], vanadium-promoted bismuth molybdate, and bismuth vanadate. In agreement with experiment [3], the apparent activation barriers calculated on bismuth vanadate are lower than those on bismuth molybdate. On the mixed vanadate-molybdate phase, vanadate sites are found to be more active than molybdate sites, with an activation barrier between those determined for bismuth vanadate and bismuth molybdate. In all cases, lattice oxygen initiates abstraction of a hydrogen radical from propene, yielding an allyl radical physisorbed to the reduced catalyst. This allyl radical then attaches to a lattice oxygen site to form an allyl alcohol-like intermediate, which undergoes a second hydrogen abstraction to produce the final acrolein product. The calculated electronic structure of the reduced catalyst is consistent with experimentally measured UV-Vis and XANES data, which show that bismuth is not reduced, while reduction is delocalized across multiple vanadium and molybdenum centers. Calculations suggest that intrinsic activation barriers can be anticipated from the electronic structure of the reduced catalyst. The conclusions drawn here are expected to have broader implications for understanding the role of dopants in tuning the reducibility of mixed metal oxide catalysts.

[1] Snyder T.P.; Hill Jr.; C.G. Cat. Rev.-Sci.Eng., 1989, 31(1&2), 43-95.

[2] Getsoian A.; Shapovalov V.; Bell A.T. J. Phys. Chem. C, 2013, 117, 7123-7137.

[3] Zhai Z.; Getsoian A.; Bell A.T. J. Catal. 2013, just accepted.