December 2010 - Allocations-admins - mailman3.cels.anl.gov

Re: [allocations-admins] [LCRC Accounts] Project Request: Dissolution
by John Valdes 21 Dec '10

21 Dec '10

On Tue, Dec 21, 2010 at 10:35:40AM -0600, Bair, Raymond A. wrote: > What is the status of this request? This was approved at the alloc meeting. I'll be creating the project latter today. John

1 0

Re: [allocations-admins] [LCRC Accounts] Project Request: Dissolution
by Bair, Raymond A. 21 Dec '10

21 Dec '10

What is the status of this request? Ray On 12/10/10 5:05 PM, "accounts(a)lcrc.anl.gov" <accounts(a)lcrc.anl.gov> wrote: > Hello, > > A new project on the LCRC cluster has been requested. Please forward > the information on to the LCRC Allocation sub-committee. > > Applicant's name: Roy Benedek > Applicant's institution: ANL > Applicant's division: CSE > Project Name: Dissolution > Project title: Dynamical simulation of lithium manganate > dissolution in acid > Associated funding: Office of FreedomCar and Vehicle Technologies > (Batteries for Advanced Transportation Technologies (BATT) Program), U. S. > Dept. of Energy > Other Systems: NERSC > Science: The acid-promoted dissolution of lithium manganate, > LiMn2O4, is an obstacle to its application as a cathode material in > lithium-ion batteries. Although dissolution-inhibition strategies are > available (such as coating the surface with a thin protective layer), a better > fundamental understanding of the dissolution process could help focus > battery-design efforts. In previous work, we have modeled the thermodynamics > of the proton-promoted dissolution of lithium manganate. The dependence of > dissolution on surface orientation and termination for flat surfaces, and on > surface coordination numbers, was addressed. Some limitations of the > thermodynamic analysis, however, were that (a) only the proton-promoted > dissolution could be readily addressed, and the corrosive effect of the anion > was neglected, (b) of all possible dissolving species, dissolution only of > neutral molecular units (such as MnO and Li2O), could be addressed, and (c) > activation energies are o > utside the scope of thermodynamics. A strong motivation therefore exists to > address kinetic modeling, which would hold the prospect of surmounting these > limitations, and even the possibility of predicting absolute dissolution > rates. Dissolution kinetics of transition metal oxides, however, has largely > eluded first-principles atomic scale modeling, partly because of a lack of > suitable techniques to simulate ³rare events² (phenomena whose time scale is > much longer than practical molecular dynamics time scales of tens, or perhaps > hundreds, of picoseconds.) In recent years, techniques such as constrained > dynamics, umbrella dynamics and metadynamics have proven effective in modeling > gas and liquid phase reaction dynamics. We believe the application of such > rare-event dynamical modeling methods, in conjunction with first principles > molecular dynamics, can lead to substantial progress in understanding > fundamental mechanisms of aqueous dissolution of transition metal oxides, and > , in particular, acid-promoted aqueous dissolution. The practical goal of > this work would be to apply the resultant knowledge and predictive capability > to engineer lithium manganate-based electrodes with longer service life by > optimization, for example, of electrode chemistry, protective surface coating, > and electrode geometry. > Project description: Experiments Planned > > The objective of this work is to gain insight into atomic-scale mechanisms > involved in the acid promoted dissolution of lithium manganate, and, > eventually, to develop a capability to predict absolute dissolution rates > (e.g., moles per unit area per unit time) as a function of such parameters as > temperature and pH. To do so, we intend to apply constrained (³Blue Moon²) > dynamics, and perhaps other simulation techniques, while utilizing first > principles density functional theory codes, particularly VASP, to perform the > atomic-scale dynamics. The coordination number of the dissolving unit with the > lithium manganate substrate will be employed as the reaction coordinate. > Calculations are done of the free energy as a function of the reaction > coordinate along the path of the dissolving unit. Candidate dissolving units > include Mn2+, Li+, MnO, LiO-, and Li2O. The resultant free energy curve > provides a prediction of the activation energy for a given dissolving unit. In > addition to t > he activation energy, the detailed histories of the dissolution simulations > may be mined by analysis and graphical methods to elucidate the ligand > exchange process that occurs as the binding of the dissolving unit to the > substrate diminishes. > > Simulations will be done of dissolution of relevant dissolving units from flat > lithium manganate surfaces, and, if time permits, from surfaces with lower > coordination (such as wires). The dissolution from defect sites, with lower > coordination, are expected to have lower activation energies, and are the most > typical dissolution events, according to currently employed phenomenological > descriptions of the dissolution process. An important benchmark will be the > activation energies for a surface in contact with neutral water. The rate of > dissolution of lithium manganate in neutral water, however, is virtually nil, > and the phenomenon of interest is the acid-promoted dissolution, particularly > for hydrofluoric acid. As a first step to understand the effect of acid, > protons will be added to the simulation cell, and their influence on the > dissolution process assessed. At a later stage the fluorine anions, which are > known experimentally to have an additional corrosive effect to that of > the protons, will be included. > > Computational Methods and Software > > The density functional theory simulation code VASP, previously employed in our > analyses of the thermodynamics of lithium manganate dissolution, will be used > in the dynamical simulations now being proposed. The most recent release of > VASP (version 5.2.11, December 2010) implements constrained molecular dynamics > in a form referred to as ³Blue Moon² sampling, the primary technique to be > used in this work. This approach yields the free energy of activation for a > dissolving ion or molecular complex. Other forms of constrained molecular > dynamics implemented in VASP are also available for tests. The electronic > structure of lithium manganate is realistically described within the GGA+U > level of approximation, which is implemented in VASP. The application of Blue > Moon sampling is presently not documented in the VASP manual, and we have > benefited from the generous assistance of Prof. Tomas Bucko (Bratislava), who > has helped us run test problems and acquaint us with the input data s > tructure. > > Visualization > > Visualization may be crucial in helping to identify ligand-exchange > mechanisms. The public domain graphics packages VMD and VESTA are convenient > for depicting three dimensional atomic configurations, and for making movies > of dynamical processes. > > Scalability Issues > > The smallest cells for which the effects of interest can realistically be > simulated contain about 200 atoms (for slabs of (001) orientation), and 700 > electrons. For orientations other than (001), cells with 300 or more atoms > will be required. For cells of this size (particularly when spin-polarized > calculations are required) the VASP code scales well up to about 100 > processors, which may be extended somewhat by tuning parameters such as NPAR. > > Efficiency and calculation size > > Because of the large CPU-time requirements for first-principles molecular > dynamics of multi-hundred atom cells, computational efficiency is of paramount > importance. The following measures have been found to greatly enhance > computational efficiency: (i) using an optimal setting of the parameter NPAR; > (ii) running the gamma-point version of VASP for which electronic orbitals are > real; (iii) using a soft pseudopotential for oxygen; (iv) considering > tritiated water, for which a relatively long 1 fs MD timestep can be applied > without significant loss of accuracy. With these measures, the CPU time > requirement for 16 processors is 30 sec. per 1 fs MD timestep. A full > dynamical simulation for one dissolution event (which involves MD runs for > several discrete values of the constraint variable) may require about 100 > picoseconds, or 100,000 MD time steps. This would correspond, roughly to a > maximum of 10,000 core-hours per dissolution event. > > Further tuning of the molecular dynamics will allow us to reduce the number of > steps required for converged forces from the Blue-Moon Ensemble. Our initial > long runs may show that fewer steps are required for converged results. > Possible approaches to accelerating convergence may include pre-equilibration > with classical molecular dynamics, simulated annealing or advanced data > analysis. > > We plan to model the following 27 reactions in this project. The dissolution > of Lithium, Manganese and Oxygen from the (001), (110) and (111) surfaces of > LiMn2O4 in water with neutral pH, low pH and in the presence of fluoride ion. > We will extend these models to surfaces with defects which are probably the > sites with the fastest dissolution rates and dominate the overall rate in real > systems. Our request of 300,000 core hours will cover the reactions on low > index surfaces and allow for the study of surfaces with defect sites. > > Project Members > > This project will primarily be a collaboration between Roy Benedek (CSE) and > John Low (MCS). > > Project URL: > Requested allocation: 300,000 > Justification: > > > The requester has used 0 hours of their initial startup project. > > In addition to approving an initial amount, please specify a Category > and Subcategory for this project. For a list of the current selection > of approved categories, please see: > > https://wiki.lcrc.anl.gov/wiki/Processes/Categories > > Once the Allocation committee has approved the project, please go to > the Project Management page to create it: > > https://accounts.lcrc.anl.gov/projects.php > > Thank You, > The LCRC Accounts System ----------------------------------------- Ray Bair Computing, Environment, and Life Sciences Argonne National Laboratory and the University of Chicago TCS Building 240, Room 4126 9700 South Cass Avenue Argonne, IL 60439 email: rbair(at)anl.gov Phone: (630)252-5751

1 0

[LCRC Accounts] Project Request: Dissolution
by accounts＠lcrc.anl.gov 10 Dec '10

10 Dec '10

Hello, A new project on the LCRC cluster has been requested. Please forward the information on to the LCRC Allocation sub-committee. Applicant's name: Roy Benedek Applicant's institution: ANL Applicant's division: CSE Project Name: Dissolution Project title: Dynamical simulation of lithium manganate dissolution in acid Associated funding: Office of FreedomCar and Vehicle Technologies (Batteries for Advanced Transportation Technologies (BATT) Program), U. S. Dept. of Energy Other Systems: NERSC Science: The acid-promoted dissolution of lithium manganate, LiMn2O4, is an obstacle to its application as a cathode material in lithium-ion batteries. Although dissolution-inhibition strategies are available (such as coating the surface with a thin protective layer), a better fundamental understanding of the dissolution process could help focus battery-design efforts. In previous work, we have modeled the thermodynamics of the proton-promoted dissolution of lithium manganate. The dependence of dissolution on surface orientation and termination for flat surfaces, and on surface coordination numbers, was addressed. Some limitations of the thermodynamic analysis, however, were that (a) only the proton-promoted dissolution could be readily addressed, and the corrosive effect of the anion was neglected, (b) of all possible dissolving species, dissolution only of neutral molecular units (such as MnO and Li2O), could be addressed, and (c) activation energies are o utside the scope of thermodynamics. A strong motivation therefore exists to address kinetic modeling, which would hold the prospect of surmounting these limitations, and even the possibility of predicting absolute dissolution rates. Dissolution kinetics of transition metal oxides, however, has largely eluded first-principles atomic scale modeling, partly because of a lack of suitable techniques to simulate “rare events” (phenomena whose time scale is much longer than practical molecular dynamics time scales of tens, or perhaps hundreds, of picoseconds.) In recent years, techniques such as constrained dynamics, umbrella dynamics and metadynamics have proven effective in modeling gas and liquid phase reaction dynamics. We believe the application of such rare-event dynamical modeling methods, in conjunction with first principles molecular dynamics, can lead to substantial progress in understanding fundamental mechanisms of aqueous dissolution of transition metal oxides, and , in particular, acid-promoted aqueous dissolution. The practical goal of this work would be to apply the resultant knowledge and predictive capability to engineer lithium manganate-based electrodes with longer service life by optimization, for example, of electrode chemistry, protective surface coating, and electrode geometry. Project description: Experiments Planned The objective of this work is to gain insight into atomic-scale mechanisms involved in the acid promoted dissolution of lithium manganate, and, eventually, to develop a capability to predict absolute dissolution rates (e.g., moles per unit area per unit time) as a function of such parameters as temperature and pH. To do so, we intend to apply constrained (“Blue Moon”) dynamics, and perhaps other simulation techniques, while utilizing first principles density functional theory codes, particularly VASP, to perform the atomic-scale dynamics. The coordination number of the dissolving unit with the lithium manganate substrate will be employed as the reaction coordinate. Calculations are done of the free energy as a function of the reaction coordinate along the path of the dissolving unit. Candidate dissolving units include Mn2+, Li+, MnO, LiO-, and Li2O. The resultant free energy curve provides a prediction of the activation energy for a given dissolving unit. In addition to t he activation energy, the detailed histories of the dissolution simulations may be mined by analysis and graphical methods to elucidate the ligand exchange process that occurs as the binding of the dissolving unit to the substrate diminishes. Simulations will be done of dissolution of relevant dissolving units from flat lithium manganate surfaces, and, if time permits, from surfaces with lower coordination (such as wires). The dissolution from defect sites, with lower coordination, are expected to have lower activation energies, and are the most typical dissolution events, according to currently employed phenomenological descriptions of the dissolution process. An important benchmark will be the activation energies for a surface in contact with neutral water. The rate of dissolution of lithium manganate in neutral water, however, is virtually nil, and the phenomenon of interest is the acid-promoted dissolution, particularly for hydrofluoric acid. As a first step to understand the effect of acid, protons will be added to the simulation cell, and their influence on the dissolution process assessed. At a later stage the fluorine anions, which are known experimentally to have an additional corrosive effect to that of the protons, will be included. Computational Methods and Software The density functional theory simulation code VASP, previously employed in our analyses of the thermodynamics of lithium manganate dissolution, will be used in the dynamical simulations now being proposed. The most recent release of VASP (version 5.2.11, December 2010) implements constrained molecular dynamics in a form referred to as “Blue Moon” sampling, the primary technique to be used in this work. This approach yields the free energy of activation for a dissolving ion or molecular complex. Other forms of constrained molecular dynamics implemented in VASP are also available for tests. The electronic structure of lithium manganate is realistically described within the GGA+U level of approximation, which is implemented in VASP. The application of Blue Moon sampling is presently not documented in the VASP manual, and we have benefited from the generous assistance of Prof. Tomas Bucko (Bratislava), who has helped us run test problems and acquaint us with the input data s tructure. Visualization Visualization may be crucial in helping to identify ligand-exchange mechanisms. The public domain graphics packages VMD and VESTA are convenient for depicting three dimensional atomic configurations, and for making movies of dynamical processes. Scalability Issues The smallest cells for which the effects of interest can realistically be simulated contain about 200 atoms (for slabs of (001) orientation), and 700 electrons. For orientations other than (001), cells with 300 or more atoms will be required. For cells of this size (particularly when spin-polarized calculations are required) the VASP code scales well up to about 100 processors, which may be extended somewhat by tuning parameters such as NPAR. Efficiency and calculation size Because of the large CPU-time requirements for first-principles molecular dynamics of multi-hundred atom cells, computational efficiency is of paramount importance. The following measures have been found to greatly enhance computational efficiency: (i) using an optimal setting of the parameter NPAR; (ii) running the gamma-point version of VASP for which electronic orbitals are real; (iii) using a soft pseudopotential for oxygen; (iv) considering tritiated water, for which a relatively long 1 fs MD timestep can be applied without significant loss of accuracy. With these measures, the CPU time requirement for 16 processors is 30 sec. per 1 fs MD timestep. A full dynamical simulation for one dissolution event (which involves MD runs for several discrete values of the constraint variable) may require about 100 picoseconds, or 100,000 MD time steps. This would correspond, roughly to a maximum of 10,000 core-hours per dissolution event. Further tuning of the molecular dynamics will allow us to reduce the number of steps required for converged forces from the Blue-Moon Ensemble. Our initial long runs may show that fewer steps are required for converged results. Possible approaches to accelerating convergence may include pre-equilibration with classical molecular dynamics, simulated annealing or advanced data analysis. We plan to model the following 27 reactions in this project. The dissolution of Lithium, Manganese and Oxygen from the (001), (110) and (111) surfaces of LiMn2O4 in water with neutral pH, low pH and in the presence of fluoride ion. We will extend these models to surfaces with defects which are probably the sites with the fastest dissolution rates and dominate the overall rate in real systems. Our request of 300,000 core hours will cover the reactions on low index surfaces and allow for the study of surfaces with defect sites. Project Members This project will primarily be a collaboration between Roy Benedek (CSE) and John Low (MCS). Project URL: Requested allocation: 300,000 Justification: The requester has used 0 hours of their initial startup project. In addition to approving an initial amount, please specify a Category and Subcategory for this project. For a list of the current selection of approved categories, please see: https://wiki.lcrc.anl.gov/wiki/Processes/Categories Once the Allocation committee has approved the project, please go to the Project Management page to create it: https://accounts.lcrc.anl.gov/projects.php Thank You, The LCRC Accounts System

1 0

[LCRC Accounts] Project Allocation Request
by accounts＠lcrc.anl.gov 07 Dec '10

07 Dec '10

Hello, A change in allocation has been requested: Requester: wiringa (Robert Wiringa) Project: qmc_for_nuclei Title: Quantum Monte Carlo Calculations of Light Nuclei Description: In FY10 we did not make as much use of Fusion as expected - about 70,000 CPU hours total (counting early FY10 hours on Jazz as half-hours on Fusion), partly because we made extensive use of the MCS SiCortex and ALCF Blue Gene for major parts of our research. We expect to continue using those other resources but the SiCortex time is not guaranteed, and it seems that the machine is getting shaky. In FY11 there will be three staff members and three postdocs working on the project. We would like to do the following: A) GFMC Nucleon-Nucleus Scattering We are continuing our studies of nucleon-nucleus scattering for n+3H, p+3He, n+4He, and p+4He, and for parity-violating (PV) n+4He scattering. Eventually we will continue to larger systems, but there are still some unresolved issues with the 3+1 and 4+1 scattering cases. Based on current experience with Fusion, we will need about 30,000 CPU hours to complete the present studies. B) Transitions in A=8-10 Nuclei We have been studying electromagnetic transitions in 10Be and 10C in collaboration with the PHY low-energy experimental group, funded by a joint LDRD project. These studies show a surprising sensitivity to details of the nuclear Hamiltonian, in particular the three-body forces, which we need to explore further. A new postdoc is arriving this fall who will be studying meson-exchange-current contributions to magnetic moments and transitions in A=8-10 nuclei. We are also starting to provide transition densities in 9Be and 10B to help explain pion inelastic-scattering and new neutrino scattering data. For this work we expect to use 50,000 CPU hours in FY11. C) Unnatural-parity and double-intruder states We have been making GFMC calculations for the unnatural-parity excitations (where one nucleon is excited to the next nuclear shell) in 9Be and VMC calculations for unnatural-parity states in 10Be and 10B. We wish to continue this work in FY11, including double-intruder states in A=10 that have two nucleons excited to the next shell, and the parity-inversion in 11Be, where the 1/2+ intruder state is the ground state. Again, we wish to explore the sensitivity to different nuclear Hamiltonians, and estimate the need for 50,000 CPU hours in FY11. D) Neutron drops We will continue to use Auxiliary Field Diffusion Monte Carlo (AFDMC) and GFMC to compute properties of large systems of neutrons. These will be both neutron drops and neutron matter done in periodic boundary conditions. We have two tasks: 1) to provide guidance to the development of energy density functional theories as part of the UNEDF SciDAC, and 2) to test the effect of three-body forces, fitted to light nuclei, in neutron matter. Based on our work this past year, we estimate 30,000 hours for this work in FY11. The Quantum Monte Carlo methods work very efficiently on parallel processors. We use MPI and see speed-up efficiencies better than 95% for up to 250 CPU's on Fusion, with sustained speeds of 1.8 GFLOPS/CPU. Using ADLB and OpenMP on BG/P we have run efficiently on 130,000 processors. We have already demonstrated that OpenMP also works well for us on Fusion for larger nuclei. This all adds up to 160,000 hours. Current: undetermined amount Justification: Requested: 16000 A specific reason has been given: We received only 80% of our requested allocation in October, and have used about 80% of that through the first two months of the FY. This addition would get us up to our original request and help us get through the next 3.5 months. This needs to be approved and the final allocation amount decided upon. Thank You, The LCRC Accounts System

1 0