Docstoc

0428774c

Document Sample
0428774c Powered By Docstoc
					Virtual Laboratory for Earth and Planetary Materials, VLab
 Renata Wenztcovitch, Yousef Saad, Ilja Siepmann, Don Truhlar, Dave Yuen (Minnesota), Philip Allen
(Stony Brook), Gordon Erlebacher (Florida), Bijaya Karki (Louisiana) , M arlon Pierce (Indiana) , Frank
         Spera (Santa Barbara), ITR 0428774, 0425059, 0427264, 0426601, 0426867, 0426757

              Simulating Fluid Phase Equilibria from First Principles

Phase diagrams play a central role in
thermodynamics and water holds a unique role
among fluids, not only because of its ubiquity
and importance on Earth, but more so because
of its anomalous liquid properties. This work
presents the first calculation of the vapor-liquid
coexistence curve for water using a first
principles representation and efficient Monte
Carlo algorithms. The BLYP density functional
representation yields a saturated liquid density
and normal boiling and critical temperatures
which are somewhat underestimated.

M.J. McGrath, J.I. Siepmann, I-F.W. Kuo, C.J. Mundy, J.
VandeVondele J. Hutter, F. Mohamed, and M. Krack,
Simulating Fluid Phase Equilibria of Water from First
Principles , in press, J. Phys. Chem. A 2005, in press
(Truhlar Festschrift).
Computer resources were provided by LLNL and MSI



                                                            Colored according to molecular dipole
                       http://vlab.msi.umn.edu            3.2 D < purple < 3.3 D 1.8 D < red < 1.9 D
                Virtual Laboratory for Earth and Planetary Materials, VLab
             Renata Wenztcovitch, Yousef Saad, Ilja Siepmann, Don Truhlar, Dave Yuen (Minnesota), Philip Allen
            (Stony Brook), Gordon Erlebacher (Florida), Bijaya Karki (Louisiana) , M arlon Pierce (Indiana) , Frank
                     Spera (Santa Barbara), ITR 0428774, 0425059, 0427264, 0426601, 0426867, 0426757

                                               Improved Density Functionals for Water
   B1B95
   PWB6K

 PBE1KCIS

 MPW1B95
  MPWB1K

  PW6B95

    B97-2
 MPW3LYP

  PBE1PBE

mPW1PW91                                                                                    The ubiquitous nature of water has prompted great interest
   MPW1K

    B3LYP                                                                                   in using simulation techniques, such as molecular
    X3LYP

    B97-1
                                                                                            dynamics and Monte Carlo methods, to better understand
      B98
    BB95
                                                                                            the structure and properties of liquid water and ice. Recent
  TPSSLYP                                                                                   attention has focused on using density functional theory
     TPSS
 PBELYP1W                                                                                   (DFT) as a means to study these systems, however the high
TPSSLYP1W

MPWLYP1W                                                                                    expense of these simulations has restricted the choice of
   PBE1W
     OLYP
                                                                                            available density functionals to less costly, and often, less
    HCTH

   PBELYP
                                                                                            accurate, methods.
     BLYP

      PBE                                                                                   We have compared a series of 25 density functional
  mPWLYP
   SVWN5                                                                                    methods to highly accurate ab initio “data” for a set of 28
            0      0.5    1              1.5              2                 2.5   3   3.5   water dimers and 8 water trimers, in order to assess the
                              Mean Unsigned Error per Molecule (kcal/mol)
                                                                                            accuracy of existing density functional methods.
      The newly parameterized density functionals (purple) perform                          Furthermore, we have gone on to optimize a new functional,
      better than those currently in use in solid state simulation (yellow                  PBE1W, that achieves accuracy equal to that of the best
      and green) and as well as, or better than, the more expensive
                                                                                            existing (and more costly DFT methods) while retaining the
      functionals commonly used in quantum chemistry (blue, red, and
                                                                                            cost-effectiveness of those methods currently in use.
      grey). All errors in the figure are relative to binding energies
      obtained using the Weizmann-1 level of theory.
                                                                                            Dahlke, E. E.; Truhlar, D. G., Improved density functionals for water,
                                                                                            J. Phys. Chem. B, in press (2005).

                                      http://vlab.msi.umn.edu
   Virtual Laboratory for Earth and Planetary Materials, VLab
   Renata Wenztcovitch, Yousef Saad, Ilja Siepmann, Don Truhlar, Dave Yuen (Minnesota), Philip Allen
(Stony Brook), Gordon Erlebacher (Florida), Bijaya Karki (Louisiana), Marlon Pierce (Indiana), Frank Spera
               (Santa Barbara), ITR 0428774, 0425059, 0427264, 0426601, 0426867, 0426757

                      Low ↔ High Density Transformations in H 2O Ice
                                                    Structural change upon compression (upper figures) of low density ice,
 Ice XI and VIII are prototypical forms of
                                                    consisting of a single hydrogen-bond network, into high density ice, consisting
 low- and high-density ice. We observed
 direct reconstructions between these               of two interpenetrating hydrogen-bond networks. The opposite is observed
 phases in both directions. These                   under decompression (lower figures).
 transformations are not observed in                                                  compression
 practice, instead both phases                                Ice XI                                            Ice VIII-like phase
 amorphize. From this perspective, the
 amorphous can be viewed as an
 intermediate step in the low  high
 density transformations. The
 reconstruction paths we observed
 bypass the intermediate amorphous
 phases and preserve the style of dipole
 ordering of the parent structure leading
 to metastable phases and large
 hystereses.
 Umemoto, K., Wentzcovitch, R.,
 Theoretical reinvestigation of the
 isostructural transformation in ice VIII, Phys.
 Rev. B 71, 012102 (2005).
 Umemoto, K., Wentzcovitch, R., Low 
 high density transformations in ice, Chem.
 Phys. Lett. 405, 59 (2005).




                               http://vlab.msi.umn.edu       Ice VIII          decompression & annealing         Ice XI-like phase
   Virtual Laboratory for Earth and Planetary Materials, VLab
   Renata Wenztcovitch, Yousef Saad, Ilja Siepmann, Don Truhlar, Dave Yuen (Minnesota), Philip Allen
(Stony Brook), Gordon Erlebacher (Florida), Bijaya Karki (Louisiana), Marlon Pierce (Indiana), Frank Spera
               (Santa Barbara), ITR 0428774, 0425059, 0427264, 0426601, 0426867, 0426757
   Thermal-Chemical Convection Simulations Including Materials Properties
                       Obtained by First Principles

 Today it is possible to model thermal-chemical
 convection and mantle dynamics taking into account
 thermodynamics and rheological properties of multi-
 phase aggregates. First principles calculations are
 nowadays offering input for these simulations that
 cannot be obtained by other means. Particularly
 important is the knowledge of phase boundaries and
 changes in properties resulting from phase
 transitions. Contrast in properties across the post-
 perovskite transition in the deep mantle computed
 by first principles by Tsuchiya et al., were used to
 perform novel simulations. This transition appears to
 inhibit the formation of superplumes unless some
 form of radiative heat transport is included in these
 simulations.
 Matyska , C. and Yuen, D.A., The importance of
 radiative heat transfer on superplumes in the lower
 mantle with the new post-perovskite phase change, Earth
 Planet. Sci. Lett. 234, 72 (2005).

                                                      Comparison between two-dimensional upwelling developments for constant thermal
                                                      conductivity (top panel) and for radiative thermal conductivity (bottom panel) in
                                                      thermal-chemical convection including both, the post-spinel transition in the
                                                      transition zone, and the post-perovskite transition above the core mantle boundary.

                            http://vlab.msi.umn.edu
   Virtual Laboratory for Earth and Planetary Materials, VLab
   Renata Wenztcovitch, Yousef Saad, Ilja Siepmann, Don Truhlar, Dave Yuen (Minnesota), Philip Allen
(Stony Brook), Gordon Erlebacher (Florida), Bijaya Karki (Louisiana), Marlon Pierce (Indiana), Frank Spera
               (Santa Barbara), ITR 0428774, 0425059, 0427264, 0426601, 0426867, 0426757

      Automatic Workflow for First Principles Equations of State Generation

                                                 Obtaining equation of state or elasticity from first principles (FP) at
                                                 finite temperatures is a time consuming procedure that involves
                                                 hundreds of job submissions and runs. Results from some runs must
                                                 be used as parameters in the subsequent ones. Input preparation
                                                 from run to run is usually made by the user. However, data
                                                 manipulation between runs are well defined and decisions to be
                                                 made during execution of the workflow are also easily predictable.
                                                 This permits these workflows to be programmed and scheduled for
                                                 automatic execution.

                                                 We designed an algorithm necessary to run automatically workflows
                                                 for equation of state computation. Human involvement is restricted to
                                                 the elaboration of a single initial input, which is not much more
                                                 complex than the typical input for a single execution of an FP code.
                                                 The present implementation makes extensive use of awk, shell
                                                 scripting, and auxiliary C programs. The usual workflow was
                                                 restructured to achieve high degree of parallelism and is well suited
 Block diagram for automatic generation of       for implementation in distributed environments.
 equations of state. Action blocks embody high   da Silva, C. R. S., Automatic Workflow for First Principles Equations of State
 degree of parallelism.                          Generation.



                       http//:vlab.msi.umn.edu
   Virtual Laboratory for Earth and Planetary Materials, VLab
   Renata Wenztcovitch, Yousef Saad, Ilja Siepmann, Don Truhlar, Dave Yuen (Minnesota), Philip Allen
(Stony Brook), Gordon Erlebacher (Florida), Bijaya Karki (Louisiana), Marlon Pierce (Indiana), Frank Spera
                                     (Santa Barbara), ITR 0428774

        Automatic Workflow for First Principles Equations of State Generation

                                                                     Most repetitive work is due to
                                                                     parameter space sampling (pressure,
                                                                     strains, phonon q-points). For efficiency
                                                                     sake structures are usually optimized
                                                                     in increasing order of pressures
                                                                     (output Pn-1  input Pn ). This implies
                                                                     in sequential runs.
                                                                     Solution:
   Flow diagram for the
                                                                     Generate an approximate Equation of
   action blocks labeled
   “Long Refine” and                                                 State (EOS), VxP  Acceptable initial
   “Lagrangean Strains” in                                           configuration at any pressure is
   previous page.                                                    available  Calculations at various
                                                                     pressures is decoupled and can run in
                                                                     parallel. Other parameters sampling
                                                                     (strains, q-points) are naturally
                                                                     parallelizable.
                                                                     da Silva, C. R. S., Automatic workflow for first
                                                                     principles equations of state generation.




                         http://vlab.msi.umn.edu

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:0
posted:2/15/2012
language:
pages:6