Thrust Area B: Molecular
Thermodynamics and Computer Simulations
Accomplishments for
11/1/99-7/31/00
Tasks during the last nine months include such preliminary work as recruiting graduate students and post-docs, ordering and installing equipment, developing computer facilities, developing codes for computer simulations, etc. Most of the research programs have been initiated. Currently there are too few interactions between the research programs. During the next year, considerable effort will be made to coordinate the research activities of Thrust Area B. Common materials will be identified for studies in all the programs. The materials will be selected according to the needs of the other thrust areas. The principal investigators, graduate students, and post-docs of all the programs will meet regularly to exchange ideas and to ensure maximum cooperation between programs. These efforts are expected to yield positive results as far as the progress of research in this thrust area as well as other thrust areas. The current progress of each of the six research programs is reported below.
Molecular
Dynamics Simulations of Multicomponent Systems in CO2 (6) The Beowulf Cluster of 17 Alpha-DEC
workstations (described in detail under Shared Experimental Facilities Section)
was purchased, assembled and tested.
Molecular dynamics suite of programs DL_POLY was adjusted for
simulations of both sub-critical and super-critical CO2 and tested. Molecular dynamics of CO2 represented by
different potential models was performed.
Preliminary simulations of alkanes and fluoroalkanes in low density CO2
were performed
Molecular Structure and
Dynamics at the CO2/Water Interface (7) During the first nine
months, Rossky has made considerable progress on the initial project of
characterizing the supercritical CO2/liquid water interface at the
molecular level. Simulations are
executed with ~1000 molecules with a number of important technical elements
implemented, including full treatment of long range electrostatic forces via
Ewald sums and equilibration of the interfacial system at constant
pressure. Simulations are executed at
the equilibrated volume with molecular dynamics, so that dynamical information
is accessible in addition to structure.
Simulations have revealed a realistic interface between a water-rich
phase and an essentially pure CO2 phase. Preliminary values for the interfacial tension support the
usefulness of the pair of solvent molecular models, each of which is a good
model for the respective neat fluid. A
key element in the analysis is the characterization of the interface both in
terms of the large length scale dividing surface, corresponding to the
macroscopic interface, and a local, molecular scale, interface. The latter analysis reveals the molecular
character on the scale relevant to the interactions with surfactant
molecules. Of most interest, results
reveal a distinct difference in “wetting” behavior when the interface is viewed
from the water-rich or CO2 sides.
Significant effects of electrostatic interactions, associated with the
polarity of a realistic model of CO2, are evident for water that is
in contact with CO2 molecules at the interface and in the bulk
aqueous phase.
Molecular
Simulations and Thermodynamics for CO2-Surfactant Systems (8) Gubbins and
Hall have focussed on developing computer simulation and analytical models that
enhance our understanding of and ability to predict phase equilibria and
properties of binary and ternary mixtures containing CO2, including
the self assembly and transformations of micellar structures. Over the past
nine months work has proceeded on three fronts. First, we have been working on adapting the SAFT (Statistical
Associating Fluid Theory) equation of state to polyFOA/CO2
systems. This has involved the
development of a code for these systems, parameter sets for the species
involved, and tests of the predictions vs. experiment. Our initial tests show that good agreement
with experiment will be achieved.
Second, we have completed preliminary runs of a discontinuous molecular
dynamics (DMD) simulation of a CO2/surfactant system modeled as
square well chains (surfactant molecules) and square well spheres (CO2
molecules). Because DMD is extremely
fast, we are able to access time scales that are long enough to allow us to
monitor the formation of model micelles and the exchange of surfactant
molecules between them. Third, we have
been trying to invent ways to use SAFT (or any equation of state approach) to
treat the self-assembly of micelles. While equations of state can handle
liquid/vapor and liquid/liquid transitions they cannot handle transitions which
involve a change in symmetry such as liquid /solid transitions or
liquid/micelle transitions. We are
hopeful that a combination of the SAFT equation with density functional theory
(DFT) will provide a route to solving this problem, since DFT provides an
accurate description of highly inhomogeneous systems.
Thermodynamic Measurements and Models for CO2 containing Systems (9) Kabadi’s efforts have been in three areas: (1) Thermodynamic data compilation for CO2 systems, (2) Design and construction of apparatus for VLE measurements of CO2 systems, (3) Design and construction of apparatus for enthalpy measurements for CO2 systems. Data compilation includes VLE, enthalpy, and density data. Excellent literature reviews for VLE data are available until year 1992. A complete literature search has been carried out for data published after 1992. Complete tables of all literature data with references will be available at the end of summer, 2000. Literature searches have also been conducted for enthalpy and density data. Tables for these data will be completed by the end of year 2000.
Two
different VLE apparatuses have been designed.
The first one is a semi-continuous flow apparatus that is for
measurements of equilibria of CO2 with relatively heavy
liquids. A known amount of liquid is
placed in the VLE cell, and CO2 is bubbled through the liquid at a
low rate. Temperature and pressure
inside the cell are maintained at desired values. When equilibrium is attained, samples of liquid and vapor are
analyzed by an on-line gas chromatographic system. The second apparatus is a continuous flow and feed
recirculation apparatus. This is
designed for measurement of VLE of CO2 with components of comparable
volatility. A liquid mixture of CO2
and components of known composition is maintained in a feed tank under helium
pressure. The mixture is pumped into
the VLE cell where desired conditions of temperature and pressure are
maintained such that the mixture separates into vapor and liquid phases. Vapor and liquid exit the VLE cell through
separate ports. Pressure in the cell is
controlled by manipulating vapor flow rate, and a constant liquid level is
maintained by manipulating the liquid flow rate. The vapor and the liquid streams are combined and pumped back
into the feed tank. When equilibrium conditions
are attained and maintained for sufficient time, vapor and liquid streams are
sampled and analyzed by an on-line gas chromatographic system. Design of both the apparatuses has been
completed. Equipment lists are being
finalized. All equipment will be
installed by September 2000.
A
calorimeter for the enthalpy measurements for CO2 systems between
–40C to +80C has been ordered. The instrument provides for simultaneous
measurement of three samples. Sample
cells are being designed to handle binary systems containing liquid and SCCO2.
Models of Polymeric
Surfactants (10) A molecular dynamics program
for simulation of both sub-critical and super-critical CO2 and one
for simulation of short polymeric chains in CO2 were tested. A Monte Carlo program for simulation of
Lennard-Jones polymeric chains was developed, optimized and tested. And a Monte Carlo program for simulation of
Lennard-Jones polymeric chains in a Lennard-Jones solvent was developed,
optimized and tested. Working with E.
Zhulina, UNC-CH, we have developed a scaling model of block copolymers
associating in a selective solvent.
Special attention was devoted to the stability of micelles of various
shapes (spherical, cylindrical, etc.). Critical parameters controlling the
shape of the micelles (molecular weight and solvent quality of the
corresponding block) were identified. Preliminary results were discussed with
experimentalists M. Adam and S. Wells, working on the complimentary project.
Molecular Simulations of the
Solubility of Organic Compounds in CO2 (11) Excellent thermal, mechanical and gas-transport properties make
fluorine-containing polyimides interesting in the application as gas separation
membranes. We chose to investigate the
two isomeric polyimides formed from hexafluoro dianhydride and para and meta isomers of a hexafluoro diamine: poly-6FDA-6FmDA (meta)
and poly-6FDA-6FpDA (para).
Although chemically identical, they display major differences in
physical properties. For example glass
transition temperature, Tg
= 598 K for the para isomer and 532 K
for the meta isomer. Gas permeation of small molecules through
films of these isomers also differs significantly. Gas permeation is faster at room temperature through the para isomer, even though it has the
higher glass temperature.
Molecular dynamics (MD) simulations were performed
with the Cerius2 package to obtain the cavity size distributions and carbon
dioxide diffusion coefficients in both the meta
and para isomer. In both cases the simulated polymers
consisted of one 20 repeat unit strand.
The applied COMPASS force field models each molecule in an atomistic
way. The initial states were generated
at the experimental densities at 308 K for meta
(1.493 g/cc) and para (1.466 g/cc)
isomer. The Cerius2 package uses an
amorphous state builder, which is capable of folding a polymer onto itself at
these densities by use of its mirror images.
Minimization is required after an amorphous state is formed in order for
MD simulations to be run on an initial state.
The cavity size distributions were determined for each isomer by
generating 50 initial states and equilibrating them over 60 ps, followed by
application of our cavity-sizing algorithm.
Diffusion coefficients were determined at these experimental conditions
by adding 10 carbon dioxide molecules to 20 independent states: 10 states per
isomer. After each state was
equilibrated for 60 ps, the diffusion constants were calculated by center of
mass displacement over a 120 ps interval.
Comparison of experimental with simulated diffusion
constants show excellent agreement.
This is remarkable bearing in mind that an atomistic model without any
adjustable parameters represents polymer and carbon dioxide molecules. Diffusion constant results support the
outcome of the cavity size distribution simulations (Monte Carlo method), which
indicate a higher probability of large cavities in the para isomer (see Figure II.B in
Appendix G). The average cavity size
reflects this trend: 5.0 Å for the meta
and 5.4 Å for the para isomer.
Plans for
8/1/00-7/31/01
We anticipate no major changes in plans for the
coming year. Rossky’s group will
complete analysis of the SC CO2/water interface, including the
effect of pressure on interfacial densities.
They will then focus on the exploration of the impact of surfactant
molecules on the interfacial behavior, emphasizing surfactant-induced
structural effects and the analysis of energetic contributions to surfactant
solvation. Gubbins and Hall will initiate a new project on Monte Carlo
simulation of micellar CO2 solvent systems in collaboration with
Professor A. Z. Panagiotopoulos of Princeton University. The objective of the work will be to map out
the phase diagrams for a range of ternary mixtures involving CO2, a
surfactant, and a solute. Calculations
will be made using a highly successful lattice MC method applied recently by
the Panagiotopoulos group to other micellar systems.
Kabadi will stay the course. He plans to (1) finish the construction of
the VLE and enthalpy apparatuses; (2) initiate VLE and enthalpy measurements
for selected systems; (3) complete data compilation and data tables for VLE,
enthalpy and density data for CO2 systems; and (4) initiate thermodynamic model
review for CO2-containing systems
Rubenstein will systematically investigate
conformations of polymer in a compressible solvent using Monte Carlo computer
simulations and compare the results with scattering data on polymers in CO2. A Monte Carlo program will also be developed
to study a phase diagram of polymer/compressible solvent binary system.
Finally, the Monte Carlo code will be extended to the study self-assembly
properties of block copolymer/compressible solvent binary system. Initially we
will simulate micelles with different aggregation numbers in order to find out
the optimal shape and size of these aggregates. Phase diagrams for block
copolymers will then be constructed as functions of the polymer concentration,
polymer molecular weight, strength of the CO2-phobic and -philic interactions,
polymer architecture and CO2 pressure and temperature.
Outstanding Achievement
The nearly completed simulations of the supercritical CO2/water interface are the first ever carried out for this system, to the best of our knowledge.