Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Glen Ferguson Project Name: metal_clusters Division: MSD Project title: Catalytic Activity of Subnanometer Metal Clusters Associated funding: Competitive LDRD Other Systems: CNM Carbon Cluster 200,000 core-hours, ALCF Intrepid ~1,000,000 core-hours, EMSL Chinook ~200,000 core-hours Science: The production of classical and next generation fuels using highly active, selective and environmentally safe catalysts is a major challenge that, if met, would provide one approach to meeting future energy needs. This challenge, along with others, is a strong driving force for developing catalysts that are more effective. Toward this goal, there is a desire for the rational design of catalysts with a high degree of selectivity and activity. An important, recently published, advancement is the use of monofunctional surface supported subnanometer metal clusters that catalyze several important reactions of small hydrocarbons, i.e., the oxidative dehydrogen of propane or separately the epoxidation of propene. These advancements open the door for new class of catalysts, constructed by rational design, with high activity and selectivity at relatively low temperatures. Two important directions to explore that may improve and extend the usefulness of these catalysts are th e design of bifunctional catalysts and the effect of the surface support on the catalyst's reactivity. Our work in these areas will focus on the use of both density functional and ab initio calculations to predict the reactivity of bimetallic subnanometer alloys and how the surface support affects their catalytic activity and selectivity. One problem is the prediction of catalysts to perform the stepwise dehydrogenation and partial oxidation of propane to propylene oxide. Currently, there is no bifunctional catalyst that forms partially oxidized propene starting from propane. The possibility of a bifunctional metal alloy cluster catalyst gives a unique opportunity to accomplish this goal in a stepwise fashion exploiting the reactivity of the two catalytic sites of the alloy. The alumina (Al2O3) surface was shown to have a central role in the reactivity of supported Ag3 clusters with propene to form propylene oxide. Specifically, the surface was a key player in the dissociation of oxygen, which is the first step in the partial oxidation of propene. This surface reactivity indic ates the possibility the surface support is important in other reactions. We will therefore investigate other surface supports to understand their effects on the catalytic activity of subnanometer bimetallic alloys. These calculations will be closely coupled with experimental work to produce and test the predicted catalysts and surface supports. Project description: New Results and Goals: We have investigated the use of Co4O2 on alumina as a model system for the production of cyclohexene from cyclohexane using periodic density functional calculations. During the course of the investigation it was determined that the oxidation state of the cluster on the surface is Co4O4 and not Co4O2. This required a change in focus to the cluster with the new oxidation state. The previous mechanism of oxidative dehydrogenation of cyclohexane to cyclohexene was reexamined and found to be unsupported for this oxidation state. The higher oxidized clusters are not as reactive as the Co4O2 cluster but consequently are not as easily poisoned by the reaction products. Using this information we have calculated a new pathway for cyclohexane dehydrogenation. The reaction can be represent by the following scheme. Cyclohexene formation Co4O4 + H2O + C6H12 → Co4O3 + C6H10 + 2H2O Oxygen defect passivation Co4O3 + ½O2 → Co4O4 This pathway is distinct from others calculated in that the surface is required as a reactant. The inclusion of water in the reaction mechanism is supported by the strongly exothermic nature of water reacting with oxide surfaces. This work is currently being prepared for submission. While a highly active dehydrogenation catalyst is a major discovery these results can be extended using catalyst design methods. While these methods were generally used only for catalytic surfaces our group has recently extended them for use with subnanometer cluster, with the results presented in an invited review in the journal Topics in Catalysis. Our method requires calculating binding energies and estimating activation barriers for various clusters develop volcano type relationships for the same reaction on new subnanometer clusters. The primary codes used in these calculations are VASP and Gaussian 09. Both of which are currently available on Fusion. While it is often acceptable to omit the surface significantly reducing the computational cost, as is indicated above, this is not possible for our systems, therefore, the resources needed for each binding energy is ~ 1000 core hours with over 200 binding energies needed. To generate the necessary data to estimate barrier a nother 100,000 hours are necessary. One of the essential tasks of using catalyst design methods for subnanometer clusters is finding the minimum energy subnanometer clusters that correspond to the experimentally observed clusters. Previously we have attempted to use the genetic algorithm of Johnston and coworkers. However, after implementation this algorithm was found to be insufficient to locate the global minima of high spin clusters used in this work. We have instead moved to the new method, called the coalescence kick method of Boldyrev and coworkers (Utah State University). In this method the clusters are randomly generated then formed into fragments and coalesced followed by optimization. The optimization is performed using a standard code but the first steps are performed using novel code that forms and coalesces the clusters. Preliminary tests using the lithium peroxide trimer and Co4O4 as test molecules indicate that the code is currently not suitable for use in our applications but indicate which imp rovements are necessary to make the method useful. For the case of lithium peroxide trimer the minima found via coalescence kick differed from the lowest minimum found by brute force by the interchange of two atoms. This immediately suggested that adding functionality to the code to perform atom interchanges would fix this problem. While the coalescence kick program is not computationally demanding the structures produced are optimized to the nearest geometric local minimum using the Gaussian 09 suite of programs. To test the additional functionality to the code would require an estimated 50,000 core hours. Another problem with this approach, revealed by attempts to find the global minimum for Co4 and Co4O4 clusters, is that random starting configurations of metal oxide and metal clusters cannot be efficiently optimized using Gaussian 09. To alleviate this problem the code will be modified to use the VASP program. Preliminary calculations performed by modifying the output of the coalescence kick program to run using VASP optimized efficiently showing the global minimum produced by brute force calculations. To validate the addition of using VASP as the optimizer for the clusters will require 100,000 core hours. Project URL: Current FY Hours Used: undetermined amount New FY Requested allocation: 400000 Q1: 100000 Q2: 100000 Q3: 100000 Q4: 100000 Justification: Thank You, The LCRC Accounts System