What is the status of this request? Ray On 12/10/10 5:05 PM, "[email protected]" <[email protected]> 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