Hello,
A yearly allocation for the LCRC cluster has been requested with the
following updated information:
Submitter/PI: Valerie Vallet
Project Name: MDcounterions
Division: CSE
Project title: Competing solvent and counter ion coordination and its effect on metal hydration properties: a polarizable molecular dynamics simulation
Associated funding: DOE, OBES, Chemical Sciences
Other Systems: We currently have access to the French GENCI CINES computer center, with 200,000 core-hours for calendar year 2013.
Science: Understanding the speciation of ions at liquid interfaces is the first step toward building a predictive understanding of the partitioning and transport of metal ions under a wide range of biotic and abiotic processes. In FY2013 we initiated a joint experimental-theoretical study of molecular-scale interactions that play a fundamental role in the chemistry and physical properties of solute-solvent interactions in aqueous solutions. Our objective is to improve the basic understanding of ion interactions both in the bulk and at interfaces, and to explore how the chemistry occurring at an aqueous interface differs from that seen in the bulk. Amongst their relevance, these questions are key issues in the context of liquid-liquid separation processes of spent fuel materials, in particular as they pertain to outstanding issues of lanthanide/lanthanide and lanthanide/actinide separations germane to the international energy portfolio. From the experimental side, the ANL’s
Heavy Elements group has developed an expertise in using synchrotron-generated, high-energy (> 60 keV) X-rays in scattering (HEXS) experiments to probe atom-atom correlations in solutions out to distances of 1 nm or more. Recently they have built upon this effort (Luo et al. accepted in J. Phys. Chem. C), by demonstrating that X-ray scattering (reflectivity measurements) from a liquid surface of aqueous ErCl3 can reveal unprecedented detail of a non-monotonic electron density profile perpendicular to the interface. These new and unexpected experimental results call for the development of an accurate and predictive theoretical model to describe the metal-ion speciation at the interface, how it is different from the bulk, and the forces driving this difference. We propose to do this using classical force fields for molecular dynamics simulations. The classical force field we use to describe all interactions present in the solution (water-water, solute-water, water-counter-io
n, solute-counterion) is more sophisticated than common polarizable force fields as it accounts for all the subtle many-body interactions taking place in a solute-solvent-anion system, including hydrogen bonding and covalent interactions. This is the level of sophistication needed to reach a unifying molecular picture of the lanthanide(actinide) chemistry from the gas-phase, into the bulk solution and at its interface. Such a unified description offers a potentially break-through not just in separations chemistry but in such diverse areas as atmospheric chemistry and biophysics.
Project description: Work plan and methodology
In an attempt to explain, on a molecular level, the heuristic ordering of a solution in which metal cations, solvent molecules and anions are present, we will employ molecular dynamics modeling to compare to HEXS and X-Ray reflexivity measurements. This will complement our new experimental efforts to develop surface-derived pair-distribution functions (PDF)s from APS synchrotron data. The computational project proposed for FY2014 will focus on simulations of lanthanide and actinide solutions with chloride and bromides as counter-ions. We will build up upon the work achieved in FY2013. In particular, as described in our project report, we have devoted significant effort towards improving the description of water-water interactions (TCPEP2013 water model), and now have a force field able to depict any water system, from gas phase clusters to bulk liquid water, to air/liquid water interfaces, and to describe them over a wide range of temperatures and pressures. This model is rea
dy for broad application to the solvation of cations in the bulk or at interfaces.
1/ for a given Ln/An element, we will use the new TCPEP2013 water model to parametrize, Ln/An-water and Ln/An-halide interactions. We wish to investigate the effect of the counter ions on Ln/An bulk properties by comparing the results of simulations conducted without and with counter ions present in the simulation box. In the latter, the counter ion will be placed in the “second coordination shell”, as experiments predict, to start from a structural configuration close to what is observed in solution.
2/ for a given Ln/An-anion (bromide) system, we will explore the changes of free energy as a function of the Ln/An-bromide distance (potential of mean-force estimated from the results 10 trajectories at fixed Ln/An-bromide distances), to see if and where it reaches a minimum, which would correspond to the formation of contact-ion pairs.
3/ We wish to explore the behavior of halides (F-, Cl-, Br- and I-) at an air-water interface, to see if there is a correlation between polarizability and adsorption propensity at the surface, as previously suggested by molecular dynamics simulations and electrospray ionization mass spectrometry data. These data will serve to validate our interface model.
3/ for lanthanide ions and in particular Er(III), we will investigate liquid-air interfaces of aqueous ErCl3 to compare to X-Ray reflectivity data, in particular focusing on the origin of the non-monotonic variation of ion concentration as a function of depth from the liquid surface.
The MD simulations will be used to generate atomistic bulk pair-correlations or electron density profiles that can be directly compared with our newly developing experimental capabilities. Code to transform the MD simulations, obtained from our LCRC project in 2013, into HEXS PDF spectra has been written and is currently available (Y.J Hu et al, manuscript in preparation). The fundamentally new aspect of this work is that these comparisons can be made, with significant accuracies, out to distances of about 0.8 nm in real space. Such comparisons will play a critical role in assessing and refining the force fields used in simulations.
Job characteristics and performance of the POLARIS(MD) code
Each trajectory described above should be run on a 5 ns time scales to reach system equilibrium. Typically for a simulation box size of a cation/anion system solvated by 1000 water molecules, the calculation requests 60 hours (walltime) on 32 processors, that is about 2000 core hours. The POLARIS(MD) code is currently parallelized with MPI scheme. In its current version it runs well up to about 32 processors, but its scalability drops to 50 % when reaching about 100 CPUs. This is why we ideally run these calculations on 32 processors on the BLUES platform. However, with the help of Intel scientists and the French Exascale Research Laboratory (ERL), we have rewritten parts of the code in order to prepare for forthcoming exascale multicore computers. A new version of our code will be ready by the end of this year.
Requested computer time.
Altogether we wish to study several metal ions, with various numbers of counter-ions, in bulk and at interfaces, summing up to about 100 trajectories. The PMF calculations require 10 trajectories per metal-bromide couple, and various numbers of counter ions, adding up to about 200 trajectories. Altogether the requested time sums up to 500,000 core hours.
To test the accuracy of the quantum chemical reference data (MP2-level), we would like to perform a couple of benchmark calculations using the latest coupled-cluster implementations in the NWChem code. Given the size of the system, we expect to run these calculations across several nodes (probably 4). Each calculation may request up 24-48 hours walltime. To be conduct, we request 50,000 core hours.
In total the project sums up to 550,000 core hours.
Participants to the project:
PI: Lynda Soderholm (ANL, CSE)
Valérie Vallet, CNRS research scientist (France) and visiting scientist (ANL), and collaborator of Lynda Soderholm.
Project URL:
Current FY Hours Used: undetermined amount
New FY Requested allocation: 550000
Q1: 150000
Q2: 150000
Q3: 150000
Q4: 100000
Justification: Work plan and methodology
In an attempt to explain, on a molecular level, the heuristic ordering of a solution in which metal cations, solvent molecules and anions are present, we will employ molecular dynamics modeling to compare to HEXS and X-Ray reflexivity measurements. This will complement our new experimental efforts to develop surface-derived pair-distribution functions (PDF)s from APS synchrotron data. The computational project proposed for FY2014 will focus on simulations of lanthanide and actinide solutions with chloride and bromides as counter-ions. We will build up upon the work achieved in FY2013. In particular, as described in our project report, we have devoted significant effort towards improving the description of water-water interactions (TCPEP2013 water model), and now have a force field able to depict any water system, from gas phase clusters to bulk liquid water, to air/liquid water interfaces, and to describe them over a wide range of temperatures and pressures. This model is rea
dy for broad application to the solvation of cations in the bulk or at interfaces.
1/ for a given Ln/An element, we will use the new TCPEP2013 water model to parametrize, Ln/An-water and Ln/An-halide interactions. We wish to investigate the effect of the counter ions on Ln/An bulk properties by comparing the results of simulations conducted without and with counter ions present in the simulation box. In the latter, the counter ion will be placed in the “second coordination shell”, as experiments predict, to start from a structural configuration close to what is observed in solution.
2/ for a given Ln/An-anion (bromide) system, we will explore the changes of free energy as a function of the Ln/An-bromide distance (potential of mean-force estimated from the results 10 trajectories at fixed Ln/An-bromide distances), to see if and where it reaches a minimum, which would correspond to the formation of contact-ion pairs.
3/ We wish to explore the behavior of halides (F-, Cl-, Br- and I-) at an air-water interface, to see if there is a correlation between polarizability and adsorption propensity at the surface, as previously suggested by molecular dynamics simulations and electrospray ionization mass spectrometry data. These data will serve to validate our interface model.
3/ for lanthanide ions and in particular Er(III), we will investigate liquid-air interfaces of aqueous ErCl3 to compare to X-Ray reflectivity data, in particular focusing on the origin of the non-monotonic variation of ion concentration as a function of depth from the liquid surface.
The MD simulations will be used to generate atomistic bulk pair-correlations or electron density profiles that can be directly compared with our newly developing experimental capabilities. Code to transform the MD simulations, obtained from our LCRC project in 2013, into HEXS PDF spectra has been written and is currently available (Y.J Hu et al, manuscript in preparation). The fundamentally new aspect of this work is that these comparisons can be made, with significant accuracies, out to distances of about 0.8 nm in real space. Such comparisons will play a critical role in assessing and refining the force fields used in simulations.
Job characteristics and performance of the POLARIS(MD) code
Each trajectory described above should be run on a 5 ns time scales to reach system equilibrium. Typically for a simulation box size of a cation/anion system solvated by 1000 water molecules, the calculation requests 60 hours (walltime) on 32 processors, that is about 2000 core hours. The POLARIS(MD) code is currently parallelized with MPI scheme. In its current version it runs well up to about 32 processors, but its scalability drops to 50 % when reaching about 100 CPUs. This is why we ideally run these calculations on 32 processors on the BLUES platform. However, with the help of Intel scientists and the French Exascale Research Laboratory (ERL), we have rewritten parts of the code in order to prepare for forthcoming exascale multicore computers. A new version of our code will be ready by the end of this year.
Requested computer time.
Altogether we wish to study several metal ions, with various numbers of counter-ions, in bulk and at interfaces, summing up to about 100 trajectories. The PMF calculations require 10 trajectories per metal-bromide couple, and various numbers of counter ions, adding up to about 200 trajectories. Altogether the requested time sums up to 500,000 core hours.
To test the accuracy of the quantum chemical reference data (MP2-level), we would like to perform a couple of benchmark calculations using the latest coupled-cluster implementations in the NWChem code. Given the size of the system, we expect to run these calculations across several nodes (probably 4). Each calculation may request up 24-48 hours walltime. To be conduct, we request 50,000 core hours.
In total the project sums up to 550,000 core hours.
Participants to the project:
PI: Lynda Soderholm (ANL, CSE)
Valérie Vallet, CNRS research scientist (France) and visiting scientist (ANL), and collaborator of Lynda Soderholm.
Thank You,
The LCRC Accounts System
