AmericanChemicalSociety.com

Modeling amorphous silica-alumina

       A remaining frontier for theory in catalysis is to understand the molecular level properties of amorphous supports like silica and silica-alumina (Si/Al).  In collaboration with the Scott and Chmelka labs at UCSB, my group is studying olefin metathesis by CH3ReO3 on amorphous Si/Al.  This catalyst is highly active on amorphous Si/Al, but inactive on a zeolite with the same composition.  Experimental evidence (EXAFS, TPD, Bronsted acid titration) suggests that the Lewis acid sites in amorphous Si/Al are responsible for its unique activity.  Additionally, the reactivity as a function of the CH3ReO3 loading suggest that the strongest Lewis acid sites yield the most active sites from CH3ReO3.  These observations all suggest that the details of coordination and structure around aluminum atoms (Lewis acid sites) at the amorphous Si/Al surface are key determinants of catalyst activity.[1]  

       Amorphous materials are not at equilibrium; modeling them requires difficult averages over the quenched disorder from their preparation history.  Because of this enormous challenge, theory and simulation remain embarrassingly unable to provide insight on amorphous catalyst supports.  However, for amorphous Si/Al with a high Si:Al ratio, some molecular insights may be gained through the following considerations.  

(1) At the surface of a dense silica-alumina, the number of ways (topologies) for Al to bond with neighboring SiO₄-tetrahedra is limited by excluded volume within the first coordination shell of aluminum.  

(2) For an isolated surface Al-site, the arrangement of Si-atoms in the first coordination shell is influenced both by the preferred geometries of bonds to the Al-atom and also by the structure of the surrounding amorphous silica.

(3) Experimental structure factors for the amorphous material combined with liquid structure models (like Ornstein-Zernike) can provide boundary distributions for the terminal Si atoms in cluster models.  

Using Gaussian approximations to these new boundary distributions, we are ‘analytically sampling’ clusters from the quenched distribution that have properties of interest (e.g. high Lewis acidity or low activation barrier for a particular reaction).  This will help understand how highly active or highly acidic sites differ from typical aluminum sites.  The ability to use cluster models for amorphous Si/Al would then enable studies like those that transformed our understanding of catalysis in zeolites.  
       The strategy outlined above is a change from the strategy in our original proposal.  In fact, the strategy outlined here is our third strategy.  The first (brute force mapping of the acidity landscape) was astronomically expensive and difficult to compare with key data like TPD experiments.  A second strategy used a Monte Carlo importance sampling scheme on the quenched distribution.  The Monte Carlo strategy, despite efficient importance sampling, also proved too expensive because of ab initio calculations that were needed for each step.  The new strategy is mathematically complex, but computationally feasible even for large cluster models of amorphous Si/Al.  Additionally, a talented PhD student (Anthony Fong) has joined the project (I did much of the early work myself).    The Si/Al work is the main thrust of our ACS-PRF effort, but the work is still in progress.  I anticipate publishable results in the next few months.  

Modeling reactions between CH₃ReO₃ and H₂O₂

       We have nearly completed a study of the reactions between CH3ReO3 and H2O2 which yield  peroxo versions of MTO that is active for olefin oxidation, Baeyer-Villiger oxidations, and epoxidation reactions.[2, 3]   Previous computational investigations[4] gave activation parameters (and also equilibria) that were inconsistent with the experiment.  By using an improved model chemistry we obtained better agreement with the observed equilibria.  By investigating alternative mechanisms, we found that a new water assisted mechanism brings theory and experimental kinetics into approximate agreement.  This work is the subject of ‘nugget 1’.   Our analysis computed the Gibbs free energy landscape for unassisted and water-assisted pathways from MTO (CH3ReO3) to an 18O labeled MTO as observed in experiments by Espenson and coworkers, to A = CH₃ReO₂(eta²-O₂), and B = CH3ReO(eta²-O₂)₂.H₂O  after successive additions of H₂O₂.   The water assisted pathways discovered in our work are dominant, which explains observed dependence of the rates on water concentration in work by Wang and Espenson.[2]  The theoretical study of reactions between CH₃ReO₃ and H₂O₂ has inspired a set of corroborating experiments by Susannah Scott’s group to validate a predicted change in the rate law when nitrogen bases are also in the solution.   This manuscript is in preparation and should be submitted within a month.  

Peripheral efforts

    With the ACS-PRF support, I wrote a review article that was invited as part of a special “Foundations of Molecular Modeling and Simulation” issue of the journal Molecular Simulation.  That article is currently in press.[5]   
       Also with the help of ACS-PRF support, I analyzed the rate promoting vibrations (RPV) model of enzyme catalysis.  That model has recently emerged at the center of a debate over the importance of dynamics in enzyme catalysis.  See for example, open letters between Karplus and Warshel in PNAS.  Unusual dynamics in the RPV model are cited as evidence that dynamics may be extremely important in enzymes.  However, the unusual dynamics result only from a complete lack of coupling between the promoting vibration and the bath modes.  No vibrational mode can be truly isolated from the environment in a condensed phase system, so the unusual dynamics in the RPV model are not likely a feature of real enzymes.  This work is summarized in ‘nugget 2’.   This article was recently published.[6]  

[1] Moses, A. W.; Raab, C.; Nelson, R. C.; Leifeste, H. D.; Ramsahye, N. A.; Chattopadhyay, S. K.; Eckert, J.; Chmelka, B. F.; Scott, S. L., J. Am. Chem. Soc. 2007, 129, 8912.
[2] Wang, W.-D.; Espenson, J. H., Inorg. Chem. 1997, 36, 5069.
[3] Espenson, J. H., Chem. Comm. 1999, 479.
[4] Gonzales, J. M.; Distasio, R.; Periana, R. A.; Goddard, W. A.; Oxgaard, J., J. Am. Chem. Soc. 2007, 129, 15794.
[5] Peters, B.; Molecular Simulation, in press [invited, special issue]
[6] Peters, B.; J. Chem. Theory and Comp, 2010, 6, 1447.