Mechanism and Design of Mixed Metal Oxide Catalysts for Selective Hydrocarbon Oxidation and Functionalization
Catalysts performing selective oxidation and ammoxidation of propene and propane using multi-metal-oxide (MMO) catalysts are of major commercial importance. However, this process it is far from optimal and even small improvements in the catalytic efficiency or reaction conditions can have a major effect upon the energy requirements and environmental impact. Furthermore, direct selective oxidation or ammoxidation of the most abundant and cheapest hydrocarbons - methane, ethane, or propane, would be greatly preferred to the current use of propene feedstocks.
A breakthrough by Mitsubishi and by BP reported direct conversion from propane to acrylonitrile using very complex mixed-metal oxides (involving oxides of Mo, V, Nb, Ta, Te, plus alkali and other metals). Although the ability of these MMO catalysts to convert propane directly to acrylonitrile and/or acrolein is impressive, the selectively and reactivity are not adequate. The MMO catalysts which successfully ammoxidize propane appear to have very complex structures with 100ís of atoms per cell, which is additionally complicated by reconstruction and non-stoichiometry at the surface. As a result of this enormous complexity, experimental techniques have provided little information about the mechanism and progress in improving these catalysts has been very empirical and ineffective.
We are using computational techniques to investigate the mechanism for the current MMO catalysts and to seek ways to improve the selectivity for activating propane to form acrylonitrile. The size of this system is making it difficult to apply modern quantum mechanical (QM) methods. We utilize a multiscale chemistry strategy, in which we use our recently developed ReaxFF reactive force field in molecular dynamics (MD) calculations to predict the structures for these complex oxides of Mo, V, Nb, Ta, Te, and other metals both under equilibrium conditions at the temperatures and pressures of interest and the dynamical structures as the reactions proceed. The parameters in ReaxFF are derived directly from QM and we have demonstrated that it gives reasonable accuracy for a wide variety of reactions. In addition with ReaxFF any new reactions discovered are checked by using QM on simplified configurations and where necessary iterating by refitting the ReaxFF parameters using the new reactions and then reexamining with ReaxFF the full mechanism.
Personnel: Dr. Kimberly Chenoweth, Sanja Pudar, Mu-Jeng Cheng and Dr. Yun-Hee Jang.
This project is co-directed by Dr. Jonas Oxgaard and Dr. Adri van Duin.
A file repository can be found here.