We have developed a new reactive force field, ReaxFF, for use in molecular dynamics (MD) simulations to investigate the structures and reactive dynamics of complex metal oxide catalysts. The parameters in ReaxFF are derived directly from QM and have been validated to provide reasonable accuracy for a wide variety of reactions. We report the use of ReaxFF to study the activation and conversion of propene to acrolein by various metal oxide surfaces. Using high-remperature MD-simulations on metal oxides slabs exposed to a propene gas phase we find that (1) Propene is not activated by MoO 3 but it is activated by amorphous Bi 2 O 3 to form allyl which does not get oxidized by the surface; (2) Propene is activated by Bi 2 Mo 3 O 12 to form an allylradical and the hydrogen gets abstracted by a Mo@O bond, which is bridged via an O to a Bi-site; (3) Propene is activated over V 2 O 5 to form an allyl, which is then selectively oxidized on the surface to form acrolein. The propene reations on V 2 O 5 occur at lower temperatures than on Bi 2 O 3 or Bi 2 Mo 3 O 12 . The results are all consistent with experimental observations, encouraging us that such investigations will enhance our mechanistic understanding of catalytic hydrocarbon oxidation sufficiently to suggest modifications for improving efficiency and/or selectivity.

In order to provide a basis for understanding the fundamental chemical mechanisms underlying the selective oxidation of propene to acrolein by bismuth molybdates, we report quantum mechanical studies (at the DFT/B3LYP/LACVP** level) of various reaction steps on bismuth oxide (Bi4O6/Bi4O7) and molybdenum oxide (Mo3O9) cluster models. For CH activation, we find a low-energy pathway on a BiV site with a calculated barrier of ΔH ⧧ = 11.0 kcal/mol (ΔG ⧧ = 30.4 kcal/mol), which is ∼3 kcal/mol lower than the experimentally measured barrier on a pure Bi2O3 condensed phase. We find this process to be not feasible on BiIII (it is highly endothermic, ΔE = 50.9 kcal/mol, ΔG = 41.6 kcal/mol) or on pure molybdenum oxide (prohibitively high barriers, ΔE ⧧ = 32.5 kcal/mol, ΔG ⧧ = 48.1 kcal/mol), suggesting that the CH activation event occurs on (rare) BiV sites on the Bi2O3 surface. The expected low concentration of BiV could explain the 3 kcal/mol discrepancy between our calculated barrier and experiment. We present in detail the allyl oxidation mechanism over Mo3O9, which includes the adsorption of allyl to form the π-allyl and σ-allyl species, the second hydrogen abstraction to form acrolein, and acrolein desorption. The formation of σ-allyl intermediate is reversible, with forward ΔE ⧧ (ΔG ⧧) barriers of 2.7 (9.0 with respect to the π-allyl intermediate) kcal/mol and reverse barriers of 21.6 (23.7) kcal/mol. The second hydrogen abstraction is the rate-determining step for allyl conversion, with a calculated ΔE ⧧ = 35.6 kcal/mol (ΔG ⧧ = 37.5 kcal/mol). Finally, studies of acrolein desorption in presence of gaseous O2 suggest that the reoxidation significantly weakens the coordination of acrolein to the reduced MoIV site, helping drive desorption of acrolein from the surface.

In order to determine the chemical mechanism for the (amm)oxidation of propane and propene on multimetal oxide (MMO) catalysts, we have carried out quantum mechanical (QM) calculations for model reactions on small clusters that we have used to train the parameters for the ReaxFF reactive force field, which enables molecular dynamics (MD) simulations for reactions on the complex reconstructed surfaces of MMO. We report here insights from the QM on the reaction mechanisms of selective (amm)oxidation of propene on bismuth molybdate catalysts and the oxidative dehydrogenation of propane on vanadium oxide catalysts. We also report the application of ReaxFF to predict the stable surfaces of the M1 phases of the MoVTeNbO catalysts.

The hydroxyl radical (•OH) is one of the most attractive reactive oxygen species due to its high oxidation power and its clean (photo)­(electro)­generation from water, leaving no residues and creating new prospects for efficient wastewater treatment and electrosynthesis. Unfortunately, in situ detection of •OH is challenging due to its short lifetime (few ns). Using lifetime-extending spin traps, such as 5,5-dimethyl-1-pyrroline N-oxide (DMPO) to generate the [DMPO–OH]• adduct in combination with electron spin resonance (ESR), allows unambiguous determination of its presence in solution. However, this method is cumbersome and lacks the necessary sensitivity and versatility to explore and quantify •OH generation dynamics at electrode surfaces in real time. Here, we identify that [DMPO–OH]• is redox-active with E 0 = 0.85 V vs Ag|AgCl and can be conveniently detected on Au and C ultramicroelectrodes. Using scanning electrochemical microscopy (SECM), a four-electrode technique capable of collecting the freshly generated [DMPO–OH]• from near the electrode surface, we detected its generation in real time from operating electrodes. We also generated images of [DMPO–OH]• production and estimated and compared its generation efficiency at various electrodes (boron-doped diamond, tin oxide, titanium foil, glassy carbon, platinum, and lead oxide). Density functional calculations, ESR measurements, and bulk calibration using the Fenton reaction helped us unambiguously identify [DMPO–OH]• as the source of redox activity. We hope these findings will encourage the rapid, inexpensive, and quantitative detection of •OH for conducting informed explorations of its role in mediated oxidation processes at electrode surfaces for energy, environmental, and synthetic applications.

We report here first-principles-based predictions of the structures, mechanisms, and activation barriers for propane activation by the M2 phase of the MoVNbTeO multi-metal oxide catalysts capable of the direct conversion of propane to acrylonitrile. Our approach is to combine extensive quantum mechanical (QM) calculations to establish the mechanisms for idealized representations of the surfaces for these catalytic systems and then to modify the parameters in the ReaxFF reactive force field for molecular dynamics (MD) calculations to describe accurately the activation barriers and reaction mechanisms of the chemical reactions over complex mixed metal oxides. The parameters for ReaxFF are derived entirely from QM without the use of empirical data so that it can be applied to novel systems on which there is little or no data. To understand the catalysis in these systems it is essential to determine the surface structures that control the surface chemistry. High quality three-dimensional (3D) Rietveld structures are now available for the M1 and M2 phases of the MoVNbTeO catalysts. However the details of the chemical mechanisms controlling selectivity and activity have remained elusive because the catalytically important sites in these Rietveld structures are occupied by mixtures of Mo and V atoms, obscuring the actual distributions of the metals and oxides at the active sites. To solve this problem we use a supercell of the Rietveld structure sufficiently large that all atoms can be whole, then we use Monte Carlo techniques based on ReaxFF to resolve these partial occupations into the optimum configuration of whole atoms still consistent with the X-ray data. We will report the ReaxFF resolved 3D structures for the M2 phase of the MoVNbTeO system. Using the resolved 3D structures we consider the distribution of sites on the important surfaces and carry out ReaxFF Reactive Dynamics (RD) calculations to follow the initial steps of the reactions. Such studies provide insights into the chemical reaction steps on MMO catalysts that should be useful in designing more selective and more active systems. We find that this suggests the critical role of the Te IV oxo chains for activating propene but not propane in the M2 phase. This suggests a new mechanism for this phase.Keywords Mechanism Á ReaxFF Á Ammoxidation Á Acrylonitrile Á Theory Á M2 phase Á M1 phase Á Mixed metal oxide Á Propane