Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Bin Liu Project Name: dft_biomass Division: CNM Project title: DFT studies of catalytic biomass conversion Associated funding: IACT (EFRC) Other Systems: Carbon cluster (Center for Nanoscale Materials) Science: Periodic, self-consistent density functional theory (DFT) calculations provide extremely valuable insights in heterogeneous catalysis involving biomass-derived compounds, such as, furan, furfural, catalyzed on metal surfaces. These molecules can undergo various processes, such as, hydrogenation, decarbonylation, and become higher commodity value products. Typically, many concurrent reaction pathways compete against each other. Hence, the selectivity to the desired products is one of the top priorities in catalyst materials selection and catalyst design. Interactions between reaction species and catalyst surfaces undoubtedly play a crucial role in determining the reaction selectivity and sometimes can be significantly influenced by dispersion effects, a type of van der Waals interactions. Using the DFT+D method, our calculations attempt to accurately describe the thermochemistry and kinetics of the reactions in order to establish a fundamental trend that governs the selectivity and activity for furan/furfural hydrogenation, decarbonylation, on different single metals. Research efforts are also engaged in developing robust correlation relationships to facilitate an efficient evaluation of the complex reaction networks. This work also has a long-term goal of investigating multi-component, multi-functional catalyst materials. The outcome of this project will have a broad and significant impact on many surface science topics. Project description: Periodic DFT calculations are performed using the Vienna ab initio Simulation Package (VASP) for this project. Standard generalized gradient approximation functionals, such as, PW91 and PBE, tend to underestimate the binding energies of furfural on Pd and Cu surfaces. The underestimated binding energy is largely due to the inability of standard PW91, PBE functionals to capture the dispersion effects on metal surfaces, especially the relatively more inert metals. This may create potentially serious problems for the analyses on the adsorption configurations, thermochemistry and kinetics. Fortunately, functionals with corrections to the dispersion effects, developed by Grimme [1] and Klimes et. al [2, 3], are available in the existing computational framework. In FY2014, we plan to perform an extensive, systematic study on this effect using the DFT+D method on the thermochemistry and kinetics of furfural hydrogenation on Cu, Pd, Pt, and Ni surfaces. It is apparent that predictions of using DFT+D will increase the binding energy of a surface adsorbate. One objective of the proposed work is to conduct a comparative study on the impact of different types of dispersion functionals (Grimme, optPBE, revPBE, optB88, optB86b, and so on) on key energetic values for different metal surfaces. Initial study will focus on the terraces sites, step sites will also be considered subsequently. Another key aspect of this study is to elucidate whether DFT+D method will affect the transition states and energy barriers. The strategy adopted in this project is to perform calculations on the hydrogenation, dehydrogenation, and decarbonylation reactions with standard GGA, and DFT+D methods, and evaluate the functionals based on the Brønsted-Evans-Polanyi (BEP) relationships. The thermochemistry and kinetics will then be used to revise the reaction mechanism and compare the revised analyses with experimental results. To complete the proposed tasks, the necessary computation will approximately involve 60% of optimizations, and 40% of transition state search (primarily using NEB, combined with the Dimer algorithm). A typical optimization job using standard GGA functionals uses 8 nodes (64 processor), and the job convergence approximately takes 24 hours. Usually, DFT+D functionals will run moderately longer per ionic step for the same system. For NEB calculations, usually 10 or 14 nodes are used, depending on the number of interpolated images. Each NEB calculation typically takes 40~50 hours, plus 100 CPU hours for vibrational frequency calculations. For FY2014, Dr. Bin Liu of CNM (Dr. William Parker) will be the primary investigator of this project with Dr. Jeff Greeley as the co-investigator. To complete the above work, we are requesting 500,000 core hours for FY2014, with 125,000 for each quarter. This is an estimate based on our preliminary work running on Fusion in the previous FYs. We anticipate that our need for Fusion resources will be relatively steady in the coming FY. Project URL: http://www.anl.gov/catalysis-science/ Current FY Hours Used: undetermined amount New FY Requested allocation: 500000 Q1: 125000 Q2: 125000 Q3: 125000 Q4: 125000 Justification: Recent improvements in the parallelization of VASP combined with the improved hardware capabilities (e.g. the expanded memory) allow for reasonably efficient near-linear scaling for jobs running on up to 80 cores (10 nodes). However, biomass compounds like furan or furfural will require tremendous computational resources. Typically, a DFT job uses 4~8 nodes at a timescale of 48 hours. Future usage will require larger number of nodes (8~10 nodes) due to a shift of our focus on transition state calculations. A large number of allocations will go into transition state searches and vibrational frequency analyses, which is still a computational challenge for the first-principles method. A Large allocation is preferred to guarantee the completion of this investigation. In the mean time, we will efficiently use the requested allocation by carefully selecting the key reactions to produce outstanding science. We hope to continue to use Fusion cluster to satisfy our sci entific interest. Thank You, The LCRC Accounts System