l
QOOl

571009707

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the

thesis entitled

MOLECULAR SIMULATION OF COMPETITION BETWEEN WATER AND
TRICHLOROETHENE FOR ADSORPTION TO MINERAL SURFACES

presented by

Chunhui L1

has been accepted towards fulfillment
of the requirements for

M.S. Jegree in Crop and Soil Sciences

 

TWA gr- T‘W‘x

Major professor

Date 8/1?!"

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

PLACE IN REfURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c'JCIRCJDateDuepss-pAs

MOLECULAR SIMULATION OF COMPETITION BETWEEN WATER AND
TRICHLOROETHENE FOR ADSORPTION To MINERAL SURFACES
By

Chunhui Li

AN ABSTRACT OF A THESIS

Submitted to

Michigan State University

in partial fulfillment of the requirements

for the degree of

MASTER OF SCIENCE

Department of Crop and Soil Sciences

2001

Professor Brian J. Teppen

ABSTRACT
MOLECULAR SIMULATION OF COMPETITION BETWEEN WATER AND
TRICHLOROETHENE FOR ADSORPTION TO MINERAL SURFACES
By

Chunhui Li

Trichloroethene (TCE) is a chlorinated solvent that is a major environmental
contaminant of soils and groundwater, and its interactions with minerals play a key role
in the fate and transport of TCE. This molecular simulation study is complementary to
experimental work on TCE sorption to hydrated zeolites. This research is the first to
perform grand canonical Monte Carlo simulations of competitive TCE and water
adsorption by zeolites. Potential energy parameters for the system were improved, and
the resulting force field reproduces the properties of TCE, water, and aluminosilicates.
For more complex systems, the force field predictions matched experimental data for
water sorption by zeolite Y and for TCE sorption to hydrated zeolite Y. The structural
and energetic features of adsorbates were examined in detail. Water and TCE domains
remain quite separate in the zeolite, with very little mixing of the two species even when
clusters of each are present in the same pore. Water-TCE competition was examined in
zeolites that were empty or pre-equilibrated with either water or TCE. Two commercial
force fields were also tested but our own modeling results were in far better agreement

with the overall experimental data and especially the water sorption data.

ACKNOWLEDGEMENTS

First of all, I would like to express my great gratitude to my advisor, Dr. Brian J.
Teppen. Without his continuing guidance, encouragement, and innovative comments and
suggestions, this research work couldn’t be successfitl. Besides the scientific knowledge
and experience I gained during working with Brian, I also learned that enthusiasm is so
important to the life. All the experience I have had during this period will be a great
treasure to my life.

I also would like to thank my graduate committee, Dr. Stephen A. Boyd and Dr.
Christian M. Lastoskie for offering suggestion, insight to this research work. I also want
to thank Vaneet Aggarwal, Yao-Ying Chien, and Hui Li for their help.

Finally, I would like to thank my families for their love and encouragement.

Especially I want to thank my husband, Jian Yi.

iii

TABLE OF CONTENTS

LIST OF TABLES ................................................................................ iv
LIST OF FIGURES ............................................................................... iv
INTRODUCTION ................................................................................. 1
METHODS ........................................................................................ 12
RESULTS AND DISCUSSION ................................................................ 29
CONCLUSIONS .................................................................................. 58
REFERENCES .................................................................................... 60

iv

LIST OF TABLES

Table 2.1. Part of force field parameters from COMPASS for our adsorption simulation
........................................................................................................ 15

Table 2.2. Part of force field parameters from PCFF for our adsorption simulation 16

Table 2.3. The change of accessible volume and percent of total volume with the radius

of probe sphere .................................................................................... 20
Table 2.4. Atom types and force field functions and parameters of TCE from PCFF force
field .................................................................................................. 26
Table 2.5. Atom type and force field parameters used in our simulation .................. 28

Table 3.1. Different force fields adsorption of TCE and water by zeolite at 308K ....... 3'7

LIST OF FIGURES

Figure 1.1. United States production of trichloroethene (TCE) .............................. 1
Figure 2.1. Zeolite viewed along (011) plane ................................................ 21
Figure 2.2. a. Single TCE molecule from MP2/6-311+G (2d, p) geometry optimization b.

Single water molecule from energy minimization by COMPASS force field ............ 22

Figure 2.3. Equilibrated liquid TCE box with 64 TCE molecules per unit cell after 100ps
of NPT molecular dynamics ..................................................................... 25

Figure 3.1. Experiment data for pure water adsorption by zeoliteHY80 at different
temperatures ........................................................................................ 29

Figure 3.2. Experimental data for water and TCE adsorption by zeolite HY 80 at 308K. 31
Figure 3.3. Energy distribution for pure water adsorbed by zeoliteHY80 .................. 33

Figure 3.4. a. Zeolite before adsorption. b. The snapshot for the pure water adsorption by
zeolite at 308K. c. The mass cloud of sorbate for the water adsorption by zeolite at 308K
........................................................................................................ 34

Figure 3.5. Trajectory for TCE and water adsorption simultaneously by zeolite at 308K
....................................................................................................... 36

Figure 3.6. Comparison of experimental results with molecular simulations for TCE
adsorption by zeolite at 308K with the competition of water at 100% relative humidity. In
each case, the zeolite pores were initially empty ............................................. 37

Figure 3.7. Comparison of experimental results with molecular simulations for H20
adsorption by zeolite at 308K with the competition of TCE at 100% saturation. In each
case, the zeolite pores were initially empty .................................................... 38

Figure 3.8. Energy distribution for TCE and water during simulation of TCE and water
competitive adsorption by zeolite at 308K ..................................................... 40

Figure 3.9. a. Zeolite before adsorption. b. Mass cloud for sorbate after adsorption. Red
stands for TCE and purple stands for water. c. Mass cloud for energy after adsorption.
Blue stand for energy about —18 kcal/mole and yellow stands for about —10 kcal/mol
........................................................................................................ 42

vi

Figure 3.10. Mass cloud of energy for TCE and water adsorption by oven-dried zeolite. a.
From —40 to —25 kcal/mol. b. From —20 to —12 kcal/mol. c. From —12 to O kcal/mol 43

Figure 3.11. a. Zeolite before adsorption b. TCE locations inside zeolite afier TCE and

water coadsorption. c. water locations inside zeolite afier TCE and water coadsorption
......................................................................................................... 44

Figure 3.12. Radial distribution function for hydrogen cation in zeolite and oxygen atom
in water ............................................................................................... 46

Figure 3. 13. Radial distribution function for chlorine atoms in TCE and oxygen atoms in
the zeolite ........................................................................................... 46

Figure 3.14. Trajectory for water and TCE adsorption by zeolite pre-equilibrated with
100% humidity water at 308K .................................................................. 49

Figure 3.15. a. Zeolite before adsorption. b. Zeolite afier equilibration with water at 308K
and 100% humidity. c. Zeolite after TCE and water coadsorption onto system b 50

Figure 3.16. Trajectory for competitive TCE and water adsorption by zeolite after pre-
equilibration with TCE alone .................................................................... 51

Figure 3.17. Energy change for the adsorption of TCE and water by zeolite that was pre-
equilibrated with TCE ........................................................................... 53

Figure 3.18. Energy distribution for pure TCE adsorbed by zeolite HY 80 in the absence
of competition from water ....................................................................... 53

Figure 3.19. Radial distribution function for the hydrogen cation in zeolite and chlorine
atoms in TCE in the absence of competition by water ....................................... 55

Figure 3.20. Radial distribution function for the hydrogen cation in zeolite and oxygen
atoms in water after the entire simulation summarized in Figure 3.16 ..................... 55

Figure 3.21. Energy distribution for TCE and water adsorbed by zeolite HY80 pre-
equilibrated with 100% TCE .................................................................... 56

Figure 3.22. The mass cloud of sorbate for TCE and water adsorption to zeolite HY80. a.
Empty zeolite. b. TCE and water simultaneously sorbed by zeolite. c. TCE and water
sorbed by zeolite pre-equilibrated with 100% TCE ........................................... 57

vii

INTRODUCTION

Trichloroethene (C2HC13, TCE), a chlorinated solvent, was widely used for metal
degreasing, for organic extractions, and as a chemical intermediate due to its
effectiveness, noncorrosivity, and nonflammability. Figure 1.1 shows how the United
States production of TCE changed over time (Doherty 2000). It has been produced since
the early 19208. By the 19605, people began to notice its role in contaminating drinking
water wells, as exemplified by the popularized case (Harr 1996) in Wobum,
Massachusetts. After the 19705, TCE production began to decrease due to several
regulations and economic factors, but then during the late 1980s, TCE production
increased again because it is a suggested replacement for other solvents, which were

banned in 1990 amendments.

 

 

g 700
3 600
E 500
n.

g 400
:1 300
a

c 200
o

g 100
E 0

1930 1940 1950 1960 1970 1980 1990 2000
Year

 

 

 

Figure 1.1. United States production of trichloroethene (T CE). (Doherty 2000)

TCE has commonly been found beneath landfills and industrial parks, and it is
unfortunately difficult to degrade and persistent in subsurface enviromnents. TCE is
identified in at least 852 of the 1416 sites proposed for inclusion on the US. E.P.A.
National Priorities list (Doherty 2000). TCE can enter surface waters via direct
discharges or by seepage and can enter groundwater through leaching from disposal
operations. The water solubility of TCE is ~1385 mg/L (Sabatini and Knox 1992). The
density of TCE is 1.46 g/cm3. So it is moderately soluble in water and can leach from
soils into groundwater, while it is also dense and can sink through groundwater as a dense
non-aqueous phase liquid (NDAPL). The upper limit for TCE in drinking water is 0.005
mg/L, (Code of Federal Regulations 1993) which is just 0.00036% of its water
solubility. In 1994 alone, 42 million pounds of TCE were released into the US.
enviromnent, as reported to the US. E.P.A. toxic release inventory (Doherty 2000).
Suppose all of this TCE entered the groundwater at the legal upper limit of 0.005 mg/L.
Since there are approximately 42*E17 kg groundwater from O to 250 m depth on earth,
(Sigg and Stumm 1994), a single year’s TCE release in the US. would place about 0.9%
of the world’s ground water in danger of contamination.

TCE is a probable carcinogen to human health. The National Academy of
Sciences has described the risk assessmerii of TCE (Scott and Cogliano 2000). The
hazard identification, dose response, and exposure have been addressed. TCE has been
shown to cause liver and kidney cancer in experimental animals (Wartenberg, Reyner et
al. 2000). Kidney cancer, liver cancer, non-hodgkin’s lymphoma, cervical cancer,

hodgkin’s disease and multiple myeloma have been found to have an excess cancer

incidence among people with the most rigorous occupational exposure (Wartenberg,
Reyner et a1. 2000). Although these effects may be caused by several solvents together,
TCE is likely one of these active agents.

The transport and fate of TCE are largely controlled by the interaction with
minerals and organic constituents of soil and sediments. Even though sorption of
organic contaminants to surface soils is dominated by a partitioning into the organic
fraction of the soil, (Chiou, Peters et al. 1979; Green and Karickhoff 1990) the sorption of
TCE by minerals also important to TCE fate and transport in soil, since the TCE can be
sorbed by subsurface minerals and can leach to the groundwater from surface soil. Many
aquifer materials contain only traces of organic matter and sorption to mineral phases
must be invoked in order to explain observed transport results (Curtiss, Roberts et al.
1986; Stauffer, MacIntyre et al. 1989; Jackson, Lesage et al. 1992; Sabatini and Knox
1992). Sorption of TCE on fractions of whole soils has been studied experimentally
(Pignatello 1990). Sorption correlated with organic carbon (OC) content, but an
important contribution of the mineral fraction was also shown, since some fractions
contributed more to sorption than they should have based on DC content. The literature
is quite diverse onthis point, with from 10 to 104 mg/kg TCE sorbs to low-organic-matter
minerals (Stauffer and MacIntyre 1986; Estes, Shah et al. 1988; Stauffer, MacIntyre et al.
1989; Farrell and Reinhard 1994; Farrell and Reinhard 1994; Xing and Pignatello 1996;
Werth and Reinhard 1997; Li and Werth 2001; Pires, Carvalho et al. 2001). Adsorption
onto porous silicated minerals is often found to be high. For example, at 100% relative
humidity, the montrnorilonite Swy-l sorbed 1300 mg/kg TCE while a silica gel sorbed up

to 2000 mg/kg (Farrell and Reinhard 1994). Micropores, defined by IUPAC as pores

smaller that 20 A in diameter, may be responsible for adsorption isotherm nonlinearity
and slow desorption phenomena.

Zeolites are 3-dimension porous aluminosilicates that form naturally in volcanic
deposits and are also widely synthesized. Approximately 121 different microporous
structures have structure codes in the International Zeolite Commission (IZC) (Wagner
and Davis 2000). Zeolites are composed of $0.; and A104 tetrahedra, which are
connected by comer-sharing oxygen atoms. The arrangement of these tetrahedra results
in a variety of structures containing cavities and channels. Their unique structures
provide many functions in industry, where zeolites are mainly used as adsorbents,
catalysts, molecular sieves, and ion-exchangers in detergents.

TCE adsorption in three fauj asite type zeolites, NaY, NaX, and siliceous faujasite
F AU, has been studied by calorimetric experiments, inelastic neutron scattering and MR-
F T Raman spectroscopy (Mellot, Cheetham et a1. 1998; Davidson, Mellot et al. 2000).
These zeolites all have the same basic structure but differ in their charges, with cation-
exchange capacities (CBC) of 418, 654, and 0 cmol/kg zeolite respectively. The
calorimetric study (Mellot, Cheetham et a1. 1998) showed that adsorption heat increases
with increasing polarity of the zeolite host. At room temperature and in the zero loading
limit, the adsorption heat of TCE is about 45, 55, 80 kJ/mol for dry and degassed
siliceous Y, NaY, NaX zeolite, respectively. Up to 30-40 molecules per unit cell, the
adsorption heat for these three types of zeolites increased to 55, 72, and 95 kJ/mol,
respectively, in which indicates that favorable sorbate-sorbate interactions stabilize the
system. The spectra (Davidson, Mellot et al. 2000) showed that in siliceous F AU, TCE

interacts with zeolite by hydrogen bonds between hydrogen atoms in the sorbate and

oxygen atoms in the framework, and by van der Waals interactions between chlorine
atoms in sorbate and oxygen atoms in fiamework. In dry zeolites NaX and NaY, there _
are also electrostatic interactions between chlorine atoms in the sorbate and sodium
cations in the zeolite pores. The adsorbate-adsorbate interaction and the dipolar nature of
the adsorbate are important to the adsorption. The sorbate/zeolite interactions are related
to the basicity of the zeolite. Pires et al. (Pires, Carvalho et al. 2001) also measured TCE
adsorption onto dried and degassed zeolites and pillared clays. For zeolite Y, about 0.44
g TCE/ g zeolite were adsorbed to either the Na- or H-exchanged mineral.

The studies summarized above dealt with evacuated zeolites and provide useful
background data, but the addition of water to the system is necessary to give
environmental relevance to these model minerals. The thermodynamic properties of TCE
adsorption by zeolite NaX in the presence of water have been studied by high-pressure
liquid chromatography methods (Farrell, Hauck et a1. 1999). Zeolite NaX is completely
hydrophilic because 'of its high charge density, and the TCE adsorption enthalpy is about
20.3 i 12.1 kJ/mol. Endothermic heats of adsorption are known to occur for solutes that
are less strongly adsorbed than the accompanying solvent. Thus TCE adsorption on the
zeolite is endothermic because it requires the displacement of water from hydrophilic
pores. The competitive sorption mechanisms of TCE and tetrachloroethene (PCE) by
hydrated zeolites have also been studied (Li and Werth 2001). Sorption of TCE from 30
mg/L solution onto four zeolites ranged from 1 to 100 mg/ g, with lower sorption to the
higher-charged zeolties. Competition between TCE (less favored) and PCB (more
favored) adsorption by zeolites increases with increasing hydrophobicity and decreasing

micropore width. The contribution to the adsorption potential is mainly from van der

Waals forces and electrostatic interactions. The authors reasoned that since PCE is more
polarizable than TCE, PCE has a higher sorption potential and can displace TCE in
strongly hydrophobic micropores. Since PCE is also slightly larger than TCE, PCE can
displace TCE in moderately hydrophobic micropores with decreasing pore size. For the
unsubstituted, siliceous zeolite silicalite, 200 mg TCE/g zeolite was sorbed rapidly form
aqueous solution (Alvarez-Cohen, McCarty et al. 1993).

The study of water adsorption by zeolites showed that the adsorption of water by
zeolites with similar structures largely depends on the chemical compositions of the
framework, in which the Si/Al ratio plays an important role. The study of water
adsorption by the zeolite mordenite showed that when the Si/Al ratio is less than 20, the
volumes of adsorbed water and hydrocarbons were comparable, and when the SiOz/A1203
ratio is greater than 40, there is almost no water adsorbed. By contrast, hydrocarbon
adsorption does not depend on the ratio between 40 and 90 (Chen 1976). Comparing
with mordenite, the adsorption of water by dealuminated faujasite typically shows an
inverse relationship with the Si /Al ratio (Li, Annen et al. 1991). The study of water
adsorption isotherms on zeolite 4A, Y, ZSM-20 (Na and H form) at 298K and P/Po = 0.8
showed the adsorption amount is 0.25, 0.25, 0.28 g/ g zeolite respectively (Pires and De
Carvalho 1997). Water adsorption by zeolites 4A, 5A, B, X, Y, erionite, offretite,
mordenite (Landolt 1971) has been studied at 298 K, and the adsorption amount of water
by these zeolites is 0.25, 0.25, 0.20, 0.32, 0.28, 0.11, 0.16, 0.13 g/g zeolite respectively at
P/Po = 0.5.

Computer simulation as a powerful research tool has been developed rapidly in

recent decades with the proliferation of computer resources. Simulation saves time and

money compared with wet-lab experiments, especially when the experiment is difficult to
perform. Simulations provide insight from the molecular level to macro-scale system
properties. Nowadays more and more computational software is available to the
researcher, so that both empirical and quantum ab initio simulations have become routine
in many labs. The ab iiritio calculation, which only uses some basic physical constants,
can provide electronic information about the system. The capacity and speed of
computers limit electronic methods to smaller systems compared with empirical
simulations. The advantages of the empirical molecular simulation are the speed to study
large systems of many thousand atoms and consideration of solvent effects. Electronic
information is not computed in empirical simulations, so bond information or other
electron transfers may not be modeled. Still, empirical modeling can be a useful method
to study mineral systems in general, including adsorption isotherms in particular. This is
because many adsorption interactions are dominated by electrostatics and van der Waals
forces that do not involve electron transfer. In addition to computing the adsorbate
loading as a function of chemical potential, one can observe precisely which sites on the
adsorbent are loaded first and the computation yields adsorption energy as a function of
loading. A limitation on empirical modeling is the lack of effective potential energy
parameters, especially when there are not many experimental data available, or when the
system is too large to do first principles calculations, from which we can derive the
potential functions and parameters for empirical modeling.

Monte Carlo simulation, one of the empirical molecular simulation techniques,
has been successfully conducted on the sorption of zeolites for different species, such as

alkanes, chlorocarbons, aromatic organic compounds, water and inorganic gases (Smit

and Siepmann 1994; Smit and Siepmann 1994; Smit 1995; Smit and Maesen 1995;
Douguet, Pellenq et al. 1996; Bremard, Buntinx et a1. 1997; Channon, Catlow et al. 1998;
Klemm, Wang et al. 1998; Nascimento 1999; Shen, Jale et al. 1999). This technique has
also been used to study the structures and properties of clay systems. TCE sorption by
layer clay was found (Teppen, Yu et al. 1998) to occur by three distinct mechanisms: The
most stable mode of sorption was by full molecular contact, coplanar with the clay
surface. This kind of interaction is suppressed by increasing water loads. A second,
more reversible interaction involves adsorption through single-atom contact between one
chlorine atom and the mineral. In a third mechanism, TCE never contacts the clay
directly but sorbs onto the external air/water interface. For TCE desorption (Teppen,
Arands et al. 1998; Teppen, Yu et al. 1998) from kaolinite, TCE desorption kinetics are
much faster from the hydroxylated aluminol surface than from the siloxane surface.
Simulations of TCE adsorption by the dry zeolites NaX, NaY, and siliceous fauj asite
have been studied (Mellot, Cheetham et al. 1998). The calculated isosteric adsorption
heats are in good agreement with the calorimetric experimental data. The adsorption heat
was decomposed into short range and electrostatic component. For siliceous Y at “zero”
loading the short range interactions accounts for 90% of the total adsorption energy.
With the increase of TCE loading, the electrostatic component remains same at 4.2
kJ/mol. All the increase in adsorption heat is from dispersion component. For NaY at
zero loading, the electrostatic contribution accounts for 14.2 kJ/mol, which give the
higher affinity of TCE for charged zeolite compared with siliceous zeolite. For NaX at
zero loading, the electrostatic contribution consists of41.3 kJ/mol, which is 50%iof the

total adsorption energy. At fixed loading, TCE heats of adsorption increase in the

sequence of host basicity and cation content for these dry zeolites. The pair distribution
functions also showed the central elements to the interaction between TCE and zeolites:
1) C1 atoms and O atoms van der Waals interactions, 2) Cl atoms and Na cation
electrostatic interactions, and 3) H atoms and O atoms hydrogen bonding. The
simulation of water adsorption isotherm by siliceous and aluminum containing
Heulandite-type zeolites showed that the cation displacement is important to the polar
molecule adsorption by zeolite (Channon, Catlow et al. 1998). The sorbed water formed
water clusters by hydrogen bonding. At 300K for siliceous zeolite, there are 1.8 water
molecules per simulation box at 2 atm. For aluminum containing zeolite, there are about
70 water molecules per simulation box at 0.5 atm, which is a loading rate of 0.072 g
water/ g zeolite. Molecular simulations of adsorption to zeolites were recently reviewed
(Fuchs and Cheetham 2001). To date, Monte Carlo simulations have been mostly used
for single component adsorption systems. There have been a few computational studies
of binary competition for adsorption sites in zeolites, mostly focused on organic-organic
competition. For water —organic competition, which is critical to understanding organic
sorption in soils, we know of no previous studies. Even though simulation of binary
mixture adsorption can be more complicated and time-consuming than the single
component adsorption, it can give us more information on the adsorption processes, as
they would be expected to occur in situ in a hydrated environment.
There are three reasons to study the system of TCE, H20 competitive adsorption
by zeolite in this work: First, people have proposed to use zeolites as TCE adsorbent
because of their special physical and chemical properties (Alvarez-Cohen, McCarty et al.

1993; Weber, Bertrand et al. 1996). Their fimctions are similar to activated carbon for

sequestration and conversion of TCE. There is a need to understand T CE adsorption by
zeolites at the molecular level. Meanwhile, the functional groups are the same as those of
clay minerals. The understanding of adsorption mechanism by zeolite may provide insight
into the mechanisms of TCE adsorption by the clay minerals, which play an important role
in the transportation and fate of pollutants in the soil and ground water.

Second, technically, simulations of zeolites are somewhat easier than simulations
of clay minerals. Zeolites have fixed cavity diameters, but for clays with the layer
structure, the distance between layers changes with the water content. Zeolites can have a
wider range of charge than the clay minerals, but pore-size distributions of zeolite are
much less dependent on charge. Considering these factors, simulation of adsorption by
zeolites is easier to control due to fewer variables.

Third, good force field parameters can be transferred to other mineral systems.
This work will validate and improve the force field parameters of the zeolite and the
organic solvent, which will increase the parameter database of the mineral systems and
help in the study of more complex systems.

The objective of this work is to use grand canonical ensemble Monte Carlo
(GCMC) simulations to study the competitive adsorption of TCE and water by zeolite.
This goal can be separated into two subgoals:

The first goal is to use all available experimental and quantum mechanical data to
produce a new and improved set of potential energy parameters for molecular mechanics
simulations of TCE and H20 adsorption by zeolite.

The second goal is to use simulations to determine the adsorption mechanisms

and to study this adsorption at the molecular level. This mechanistic information will be

10

used to help interpret the macroscopic adsorption data for the competition between water
and TCE at mineral surfaces.

Images in this thesis are presented in color.

 

ll

METHODS

1.Force Field Methods

Force field methods are empirical computational methods, which use a set of
mathematical functions and parameters to analytically describe the potential energy
surface of a system as a function of the atomic coordinates. Force field simulations can
handle large systems, such as condensed matter, macromolecules, inorganic crystals and
organic interphases, because they bypass solving the electronic SchrOdinger equation. In
other words, empirical methods do not calculate the electronic energy for a given nuclear
configuration. The basis of force field methods is that molecules tend to be composed of
units that are structurally similar in different molecules, a small set of molecules are used
to parameterize the force field, then these force field parameters are transferred to study
other systems. The force field energy is decomposed into several parts:
E F}: = E (bond stretch) + E (angle bend) + E (torsion) + E (inversion) + B (Cross term) +
E (van der Waals) + E (electrostatic)
The first five terms on the right of above equation are the bond terms. The last two terms
are the non-bonded terms. Some parts, especially the cross terms, are commonly ignored.
Given this energy, the relative energy of a system can be calculated and the optimized
geometry can be found by minimizing this energy. Since each atom has a unique atom
type in the force field, the energy contribution of the individual atoms or classes of atoms
can be analyzed.

Many kinds of force fields have been developed (Weiner, Kollman et a1. 1986;

Jorgensen and Tirado-Rives 1988; Allinger, Yuh et al. 1989; Halgren 1996; Halgren

12

1996; Halgren 1996; Halgren 1996; Halgren and Nachbar 1996; Sun 1998; Sun 1998;
Sun, Ren et a1. 1998; Newsam, Sun et al. 1999). Different force fields use different
functions and parameters and are suitable for studying different systems. In general,
force fields can be classified into three categories: 1. Broadly applicable force fields,
which cover a broad range of the periodic table by using simple rules to create energy
parameters. This type of force field can ofien predict molecular structure reasonably
well, but the prediction of the other molecular properties is often limited. 2. Classical
force fields: This type of force field generally has relatively simple functional forms and
focuses on the application of certain areas, mainly biochemistry. 3. Second-generation
force fields: This type of force fields can achieve higher accuracy in predicting various
molecular properties; meanwhile they can still cover a broad range of the periodic table.
In this research work, we will mainly talk about two commercial force fields, COMPASS
and PCFF, which are both second-generation force fields. PCF F (Sun, Mumby et al.
1994) stands for Polymer Consistent Force Field and is an ab initio force field. Most
parameters are derived from ab initio data using a least-squares fitting technique. The
non-bond parameters were derived by fitting to molecular crystal data based on energy
minimization calculations. COMPASS (Sun 1998) stands for Condensed-phase
Optimized Molecular Potential for Atomistic Simulation Studies and was developed on
the basis of PCF F . The valence parameters for molecular classes in the PCF F were
transferred to the COMPASS. The nonbond parameters were reparameterized.
COMPASS was also expanded to include the most common organic and polymer

materials and inorganic materials.

13

During Grand Canonical Monte Carlo simulation of adsorption (GCMC), we only
consider nonbonded interactions, which include the Lennard-J ones 6-9 function for the
van der Waals interaction and the Coulombic function for the electrostatic interaction.

E = Eelec + Evdw

Eelec = ‘11 *qj /(8 * R ij)

Where qi and q, are the partial charges for atom i and j, Rij is the distance between atoms
i. and j., and e is the dielectric constant which here is 1.0 since we are working at the
molecular scale.

E... = Do (2(Ro/ R) 9 — 3 (Rn/R) 6)

Where Do is the well depth in kcal/mol, and R0 is the equilibrium distance in angstroms.
Tables 2.1 and 2.2 show the force field parameters of the zeolite-water-TCE system for

both the COMPASS and PCFF force fields. (M81 1999)

14

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Force field type Atoms in model Partial Charge VDW
Re Do (A)
(kcal/mol)
Si4z Si in zeolite 0.89 4.29 0.105
022 0 between two Si -0.4450 3.3 0.08
A142 Al in zeolite 0.7343 4.4 0.2
022 0 between A1 and -0.4578 3.3 0.08
Si
032 0 between H,Al,Si -0.3348 3.3 0.08
H10 hydrogen 0.0839 1.087 0.008
Hlo H in H20 0.42 1.087 0.008
02* 0 in H20 -O.84 3.695 0.21
C3= C in TCE 0.204 3.92 0.08
C3= C in TCE -0.0248 3.92 0.08
C11 C1 in TCE -0. 102 3.823 0.286
H1 H in TCE 0.1268 2.878 0.023

 

 

 

 

 

 

Table 2.1. Part of force field parameters from COMPASS for our adsorption simulation.

(M81 1999)

15

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Force field type Atoms in model Partial VDW

Charge R0 Do (31)

(kcal/mol)

52 Si in zeolite 0.5236 0.001 0.00
oss 0 between two Si -0.2618 3.4506 0.1622
a2 Al in zeolite 0.5366 0.0001 0.00
oas 0 between A1 and Si -0.2959 5.2591 0.0129
Hb hydrogen 0.0839 1.2149 5.2302
0b 0 between H,Al,Si -0.2515 5.2191 0.0135
Hw H in H20 0.3991 1.098 0.013
0* 0 in H20 -0.7982 3.608 0.274
C: C in TCE 0.2040 3.9 0.064
C1 C1 in TCE -0.1020 3.92 0.2247
Hc H in TCE 0.1268 2.995 0.020

 

Table 2.2. Part of force field parameters from PCFF for our adsorption simulation (M51

1999)

2. MC:

The Monte Carlo technique as applied to molecular simulation was first

introduced by Metropolis et al (Metropolis, Rosenbluth et al. 1953). It is a stochastic

modeling method for obtaining optimized molecular structures and configurations based

on the analysis of possible equilibrium configurations from a random sampling of the

16

 

potential energy surface. Monte Carlo methods can calculate the average property for
any ensemble. In this work, we are using grand canonical ensemble Monte Carlo
(GCMC) methods, in which the chemical potential, volume, and temperature are fixed
and the numbers of molecules in the system are allowed to change during a simulation.
GCMC is particularly useful for studying multilayer adsorption near a surface. For a
GCMC simulation, there are two kinds of energy terms, Coulomb electrostatics
interactions and van der Waals interactions.

In the GCMC adsorption system, there is a given crystal framework, which is the
adsorbent, and one or more gas phases, which are adsorbates. Here we treat the adsorbate
as an ideal gas whose chemical potential is determined only by the temperature and
partial pressure. During the simulation, the chemical potential of the gas molecules and
temperature are fixed while the number of molecules fluctuates. The volume of the
adsorbent framework is also fixed and the fi'amework and gas molecules are treated as
rigid bodies.

At the beginning of a simulation, an initial configuration for the system is defined
and the total potential energy is determined. There are four kinds of moves to generate
the next configuration: Create, destroy, translate or rotate a gas molecule in the pores of
the fiamework. The adsorbate model is randomly displaced to a new configuration in the
adsorbent and a new energy is calculated. The new configuration, based on the energy
change, is accepted or rejected by the Metropolis rules. If the energy of a new
configuration is lower than the original, this configuration is accepted and the next
displacement is continued. If the new configuration energy is greater than the original,

the configuration is accepted or rejected depending on the comparison of a Boltzmann

l7

distribution to a random number. If the value of Boltzmann distribution is less than the
random number between 0 and 1, the configuration is accepted and used as the basis for a
new displacement. Otherwise, the new configuration is rejected and the previous
configuration is used for the next displacement. The chemical potential is translated into
the partial pressure of each component until equilibrium is achieved. At this time, the
temperature and chemical potential of the gas are equal to each other inside and outside
of the framework.

The result of this simulation is a set of configurations that converge towards the
specified chemical potential and temperature. A large number of configurations are
generated after the equilibration period. The simulation takes several steps to equilibrate
from its original random position. For accurate statistical results, the steps made prior to
equilibration should be excluded in the analysis, and a large number of configurations
should be generated after the equilibration period. These configurations can be sampled
and saved as a trajectory. From a trajectory file, the adsorption energy, conformational
structure, mass distribution, and loading amount can be analyzed as a function of
adsorbate chemical potential, allowing construction of an isotherm along with the
structural and energetic features underlying the isotherm.

3.MD:

MD is a deterministic molecular modeling tool that evaluates the forces on
individual atoms using the same energy force field. Newton’s differential equations of
motion are used to compute new atomic positions after a short time interval (on the order
of a femtosecond), then evaluation for a large number of time steps provides a time-

dependent trajectory of all atomic motions.

18

4.Comparison of MC and MD:

The prime purpose of both Monte Carlo and molecular dynamics program is to
compute equilibrium properties of classical many-body systems and MD simulations can
also be used to follow nonequilibrium kinetics. Since all the atoms are labeled in the
simulations, the exact atomistic behavior that causes the averaged data can be followed in
order to determine the molecular mechanisms. Classical molecular dynamics and Monte
Carlo simulations cannot provide any electron information, but instead need a set of
empirical parameters to calculate the potential surface.

In MD simulations, all particle coordinates are updated simultaneously. In
conventional MC simulations, only a few coordinates are changed in a trial move. As a
consequence, collective molecular motions are not well represented by MC, and this may
adversely affect the rate of equilibration. An advantage of MC is that, unlike MD,
unphysical moves can be carried out. Moreover, in MC the system is not constrained to
move on a hypersurface where some Hamiltonian is Conserved. The time step in MD is
limited by the need to conserve energy. Clearly, no such constraint applies to MC. The
computation of time-average quantities distinguishes molecular dynamics from Monte

Carlo simulation.

5. Model
In this research, we are going to study the zeolite HY80, which belong to the large
class of faujasite-type zeolites. Figure 2.1 is one supercell zeolite (M81 1999) with the

composition of H58i379A150763. Internal pores are mostly siloxane surfaces with elliptical

l9

cross-section about 8 by 12 A. Total surface area of the model is about 1240 m2/ g. The
accessible volume to a sphere with radius of 1.2 A is about ~40% of total volume.
Spheres with different radii give different estimates for accessible volume, and Table 2.3
gives the accessible volume and its percentage of total volume versus radius r of the
probe molecule. The cell parameters for this supercell are: a = 50.06A, b =25.03A, c
=25.03A, a =90°, B = 90°, 7 = 90°. Si atoms were substituted by Al atoms randomly by
Loewenstein’s rule, which asserts that no two aluminum atoms share a common oxygen
atom. The model Si/Al ratio is 75.8, which is the closest approximation to the ratio 80
used in experiment while we can still keep the model size amenable to calculation. The
substitution charges were balanced by exchangeable H+. The figure 2.1 is the zeolite
viewing along the crystal direction (011) plane. The dominant pores are 12-member ring

channels and 6-member ring cages.

 

 

 

 

Radius of probe Occupiable volume Percent of total
sphere (A) (A3/unit cell) volume

1.0 13799 44

1.2 12541 40

1.4 1 1480 37

 

 

 

 

 

Table 2.3. The change of accessible volume and percent of total volume with the radius

of probe sphere. The total volume is 31355 A3/unit cell.

20

 

Figure 2.2. a. Single TCE molecule from MP2/6-311+G (2d, p) geometry optimization b.

Single water molecule from energy minimization by COMPASS force field.

6. Research work method

In this research work, the two commercial force fields COMPASS and PCFF
were first tested for TCE and water adsorption isotherm by zeolite HY80. Tables 2.1 and
2.2 contain the relevant force field parameters for COMPASS and PCFF respectively.
Based on the analysis of these force field functions and parameters, and also based on the
results of the first few simulations, we refined the parameters for the whole system and
then created a new force field. Using this force field, the adsorption isotherms of TCE
and water by zeolite were also studied.

The first step of this work is to refine the charges for chlorine atoms in the TCE
using ab initio calculations including the solvent effect. The chlorine atoms in the TCE
are highly polarizable, as are the oxygen atoms inside the zeolite framework. We

expected that their interactions are important to the adsorption isotherm, especially when

22

we include solvent effects in each simulation. The effect of a solvent may be divided into
two major parts: 1) Specific solvation or short-range effects, concentrated in the first
solvation sphere, and 2) Macroscopic or long-range effects, involving screening of
charges (solvent polarization), which requires a large number of solvent molecules. To
consider the effect of solvent water in the simulation, the quantum calculation was
performed under the condition that a polarized continuum of water was added. The
polarizable continuum model (PCM) (Cossi, Barone et al. 1996) employs a van der
Waals surface type cavity, a detailed description of the electrostatic potential, and
pararneterizes the cavity/dispersion contributions based on the surface area. Charges for
TCE are from an MP2/6-311+G (2d, p) (Moller and Plesset 1934; Hehre, Radom et al.
1986) geometry optimization, which applies the perturbation theory to calculate the
approximate electron correlation energy, followed by a ChelpG analysis, (Chirlian and
Francl 1987; Breneman and Wiberg 1990; Francl, Carey et al. 1996) which finds partial
atomic charges that correspond to the ab initio molecular electrostatic potential. The
final charges for TCE are given in Table 2.4. Since in our force field, we did not separate
the two chlorine atoms in the same side of carbon double bond, the partial charges for
these two chlorine atoms were averaged to the same number. The charges we used
(Table 2.4) were very close to those used in another successful study (Mellot, Cheetham
et al. 1998) of TCE adsorption to dry zeolites. Notably, our charges on C1 atoms of TCE
were somewhat more negative due to our inclusion of the polarizeable continuum in the
quantum calculations.

The second step is to refine the van der Waals parameters for TCE by molecular

dynamics fitting. After fixing the TCE charges, the van der Waals potential were refined

23

by molecular dynamics simulation at room temperature. A liquid TCE box with 64 TCE
molecules was built. According to the following approximate thermodynamic
calculation, the liquid TCE potential energy of one unit cell should be about -247 kJ/mol.

E = U = PE+KE

H = U+PV

A Hvap= A Hgas - A Hliq

A H up = (U+PV) gas — (U+PV)1,q

= (PE+KE +PV) gas - (PE +KE+PV) “q

Assume that KEh-q = KBgals at the same temperature

Pan=0, PVgas = nRT

Then A Hvap = -PE1,cl +PE 8,, + nRT

PE],q = -A H V3,, + PE 8,, + nRT

At 25°C A H vap = 34.54 KJ/mol (Lide 1995)

PE],q = -34.54/4.18 +8.312* 298/4.18*1000 +PE gas (kcal/mol TCE)

PE 335 = 3.81 kcal/mol TCE

PEnq= -3.86 kcal/mol TCE

PEqu /unit cell = 64 *(-3.86) = -247.04 kcal/mol

The equilibrated TCE liquid box containing 64 unique TCE molecules is shown in
Figure 2.3. Using molecular dynamic simulation on this TCE liquid box, the van der
Waals potential parameters for TCE were refined by trial and error each time until the
final density of the TCE box was 1.466 g/cm3, which is in agreement with the

experimental data of 1.46 g/cm3 at 293 K. The final potential energy is —249.97

24

kcal/mol, which is also in good agreement with the results of the above calculation. -
247.04 kcal/mol. The mathematical functions and parameters of bond-terms of TCE for
molecular dynamic simulation are shown on Table 2.4, which are all from the PCFF force
field. The charges for TCE are from the above ab initio calculation. Our van der Waals
terms for chlorine atoms are quite similar to the parameters in PCFF, which means the

solvent effect does not change the van der Waals terms for chlorine atoms too much.

 

Figure 2.3. Equilibrated liquid TCE box with 64 TCE molecules per unit cell after lOOps

of NPT molecular dynamics.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Bond stretch Kb R0 (A) d
(kcal/molAZ)

C= C=1 Quartic 1090.5 1.4 -1.8

H1 =2 Quartic 731.5 1.1 -1.9

C11 C=1 harmonic 557.0 1.7

C12 C= harmonic 557.0 1.7

Angle bend K0 (kc al/molAz) 00 (°) (1

H1 =2 C=1 Theta quar 70.6 124.9 -0.50

H1 C=2 C12 Theta harrna 75.0 120.0

C12 C=2 C=1 Theta harrna 72.4 120.0

C11 C=1 C11 Theta harma 120.0 120.0

C11 C=1 C= Theta harrna 72.4 120.0

Torsions Barrier (kcal/mol) Period

H1 =2 C=1 C11 Dihedral 8.2 2.0

C12 C= C=1 C11 Dihedral 8.2 2.0

Inversions K0 (kcal/mol)

C=1 C11 C=2 C11 Wilson avg 16.0

C=2 C12 C=1 H1 Wilson avg 15.0

 

 

 

 

 

 

 

 

 

Table 2.4. Atom types and force field functions and parameters of TCE fi'om PCF F force
field. C=1 and C=2 stands for the carbon atoms, C11 and C12 stands for the chlorine

atoms, and H1 stands for hydrogen atoms.

26

The partial charges for the zeolite atoms are from previous work by Teppen et al.
(Teppen, Rasmussen et al. 1997). The charges were derived from the CHELPG method
using MP2-optimized geometries of several small aluminosilicate molecules. The partial
charges for silicon and oxygen of zeolite in PCFF and COMPASS are from Hartree-Fock
calculations, (Hill and Sauer 1994; Hill and Sauer 1995) which do not include electron
correlation effects. The charges in PCFF are also rather artificial because only half of the
ab initio calculation charges were used in the force field. This 0.5 factor was given (Hill
and Sauer 1994; Hill and Sauer 1995)because organic force fields generally use low
charges and the authors believed that better structure prediction could be achieved by also
decreasing the zeolite charges.

After refining these parameters, our field force was used to simulate the
adsorption isotherms. The sorption module in the Cerius2 software fiom MSI was used
for all the sorption simulations. Ewald summation methods were used for calculating
non-bond interaction at 0.025 kcal /mol accuracy (Karasawa and Goddard 1989).
Periodic boundary conditions were used to remove the surface effects, resulting in

simulation of adsorption by an infinitely large zeolite crystal.

27

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Force field Atoms in model Partial VDW
type Charge R0 Do
sz Si in zeolite 1.4 3.8 0.05
055 0 between two Si 07 3.85 0.26
a2 A1 in zeolite 1.2 3.9 0.06
oas 0 between A1, Si -0.9 3.8 0.26
H+ hydrogen cation 1.0 ignore ignore
hw H in H20 0.41 1.145 0.015
0* 0 in H20 082 3.7 0.26
=1 C in TCE 0.0367 3.9 0.064
=2 C in TCE -0.0441 3.9 0.064
C11 C1 in TCE -0.0374 3.92 0.236
C12 C1 in TCE -0.0966 3.92 0.236
H1 H in TCE 0.1788 2.995 0.02

 

 

 

 

 

28

Table 2.4. Atom type and force field parameters used in our simulation.

 

RESULTS AND DISCUSSION

Experimental description:

These are unpublished data from Dr. James Farrell’s group at the University of Arizona
and their methods generally follow their previous studies of TCE sorption and desorption
(Farrell and Reinhard 1994; Farrell and Reinhard 1994; Farrell, Grassian et al. 1999;
Farrell, Hauck et al. 1999). The experimental results will be very briefly presented here
and then discussed in more detail along with the simulation results.

a. Farrell’s group measured the water adsorption isotherm by zeolite both by continuous

gas chromatography (GC) and gravimetric methods at 278, 283, 293, and 320 K.

 

 

 

 

”05°C ’

110°C

;A20°c

I 0.6 I 1*" 9
1-0.5..fi____4-._. L__,, w .- .-_. _-. _fir
:s 0..----1-.-_--._: _ _ _ _ _ __ T
EOBITTILT‘A‘ --=~ 22+. .22.
I ,3 0.2 rmfij _ ————- A 4K— A ~ a ——
i 2 0:) :fi-TT 11“?“ ' __ “T" *‘ * “T"

0 0.00001 0.00002 0.00003 0.00004 0.00005

Vapor Pressure (gImL)

Figure 3.1. Experiment data for pure water adsorption by zeoliteHY80 at different
temperatures. (J arnes Farrell, unpublished)

b. Farrell’s group measured desorption isotherms for TCE interacting with hydrated
zeolites by two methods. In the first, oven-dried zeolite was allowed to equilibrate

simultaneously with both water and TCE, each at saturation vapor pressures. The

29

relative humidity was then maintained at 100% in the GC gas stream, while the TCE
content of the gas stream was gradually lowered. The quantities of sorbed TCE in
equilibrium with various partial pressures of TCE gas and 100% relative humidity are
shown in the “oven-dried” series of Fig 3.2. Adsorption and capillary condensation
of water and TCE on the outside of the particles is a problem for experiment. There
is an inflection point (i.e. change in sign of the second derivative) at about 20% of
Psat (loading = 0.45 g/ g) in the experimental data, Figure 3.2. This inflection point is
where capillarity condensation begins. Most likely the capillary condensation that
begins at this point is on dry spots on the zeolite surface.

c. The second type of measurement of TCE—water competition by Farrell’s group was
gathered by pre-equilibrating the zeolite with a stream of 100% relative humidity and
no TCE over night. Then, desorption isotherms in the presence of 100% humidity
and various TCE contents were gathered as above, and the results are plotted as the “
pre-equilibrated with water” series in Fig 3.2. On this graph, no inflection point
appear until about 90% of Psat, indicating approximately 0.38 g/ g TCE sorption
within the zeolite. Thus, pre-equilibrating the mineral with water seems to reduce

TCE sorption by about 10 to 20%, although TCE sorption remains quite substantial.

30

 

+ pro-equilibrated with

water
+ oven-dried

 

gTCE/gzoollte

 

 

 

 

Figure 3.2. Experimental data for water and TCE adsorption by zeolite HY80 at 308K

Simulation results and discussion
1. At 308K, our simulation results for pure water adsorption by the model for zeolite
HY80 is 0.27 ng zeolite at 100% humidity. The pore space in the HY80 is
enough to hold about 0.33 yg zeolite water according to the free volume
calculation. Since the sphere with 1.2 A radius was used to measure the pore
space in zeolite and the water molecule is larger than this sphere, this 0.33 ng
amount should be an upper limit amount for water. Our results are in good
agreement with measured experimental data for water adsorption, 0.28 gg zeolite
Y (Landolt 1971) or 0.25 g/g zeoliteY (Pires and De Carvalho 1997) at 298K.
Comparing with Farrell’s experimental water adsorption by zeolite HY80 at
different temperatures (discussed above), our results are also in the right range of

the data between 293K and 320K.

31

Experimentally, water adsorption to a zeolite powder is somewhat difficult
to compare with simulation results, because the real powder has external surfaces
while the simulated zeolite has only internal porosity. Thus, the highest
experimental points for each temperature in Fig.3.1 include both internal sorption
of water and capillary condensation of water on external surfaces of the zeolite
powder. Indeed the zeolite powder will take up water ahnost indefinitely, beyond
1 g/g zeolite. The amount sorbed to internal surfaces alone is therefore
experimentally uncertain. Farrell’s group tried to address this uncertainty by
measuring water sorption at 303 K to two different zeolite columns (data not
shown) in a stream of 100% relative humidity. The two duplicated columns show
the same water uptake until 0.38 g/g, which may be regarded as an upper limit on
water sorption to zeolite internal surfaces. Beyond that point (i.e. afier longer
equilibration) water condensation depends on the packing density of the columns
and the two uptake profiles diverge. In Figure 3.1, the water sorption data at 278
and 283K clearly show inflection points at 0.2 to 0.25 g/g zeolite that may
indicate the extent of internal pore filling of zeolite HY 80 by water and would
agree with both our simulations and other experimental data (Landolt 1971; Pires
and De Carvalho 1997).

Figure 3.4 shows a) zeolite before adsorption b) the snapshot of the final
configuration (as representative of the equilibrium state), and 0) mass cloud of
sorbate in zeolite. These figures show that water occupied all the 12-member
rings channels and 6-member ring channels. It is apparent that water occupies

nearly all of the available porosity.

32

 

 

4.E+04
4.E+04
3.E+04 7
3.E+04 .
2.E+04
2.E+04
1 .E+04

5.E+03 ‘
O.E+OO I #1 r

Energy (kcal/moi)

 

 

 

 

 

 

Configurations

 

 

 

 

 

 

 

 

 

 

 

Figure 3.3. Energy distribution for pure water adsorbed by zeoliteHY80.

Figure 3.3 shows that there are two peaks for the energy distribution for pure water
adsorbed by zeolite HY 80. The energy for the high peak is about —15 kcal/mol, for the
small peak is about —38 kcal/mol. The former represents the bulk of the pore water in the

zeolite, while the latter represents the few water molecules that are directly coordinated to

exchangeable cations.

33

3
‘v
/
1"“
06‘
3“
I5
,‘1/

 

34

Figure 3.4. a. Zeolite before adsorption. b. The snapshot for the pure water adsorption by

zeolite at 308K. c. The mass cloud of sorbate for the water adsorption by zeolite at 308K.

The COMPASS and PCFF force fields have also been tested for water adsorption.
Even though there are no TCE competing with water in these simulations, there are only
0.00076 g/g zeolite and 0.0035 g/g zeolite water sorbed by the zeolite for the two force
fields, respectively. These results are consistent with previous results (Channon, Catlow
et al. 1998) using the PCFF force field, in which little or no water sorbed to siliceous
pores in a Heulandite zeolite. For each one unit cell, there is no more than one water
sorbed by the zeolite. This means that the zeolite in these two force fields is much more
hydrophobic than the zeolite as modeled by our force field. Both PCF F and COMPASS
use lower charges than our force field. Since electrostatic interaction in this system
accounts for the large part of the whole interaction, these lower charges apparently result
in the greater hydrophobicity of the zeolite. In this respect, our force field agrees much
better with the experimental data and is superior to either of the commercially available
force fields.

2. The experiment (b) by Farrell’s group (described above) was also simulated.
ZeoliteHY80, initially vacant, was allowed to equilibrate simultaneously with
both water and TCE gas molecules. The relative humidity was maintained at
100%, while the pressure of TCE was gradually lowered. For each TCE chemical
potential, a sorption run of 2"‘107 steps was completed. Figure 3.5 is the trajectory
for TCE, water both with 100% saturation pressure adsorbed by oven-dried
zeoliteHY 80 at 308K, which shows that after about 1"‘107 steps, the adsorption
achieved equilibrium. For each chemical potential, the data from 1"‘107 to 2"‘107

steps were used as the ensemble for statistical mechanical analysis.

35

 

 

— H20 loading
80 —TCE loading
70
60
50
40
30
20
1 0

loading amount per unit cell

 

o
0.00E+00 5.00E+06 1.00E+O7 1.50E+07 2.00E+07
simulation steps

 

 

 

Figure 3.5. Trajectory for TCE and water adsorption simultaneously by zeolite at 308K.

36

pressure ~ SAT%

(kPa)

 

Table 3.1. Different force fields for adsorptions of TCE and water by zeolite at 308K.

 

 

 

 

*pcff
+oornpass

,3 *ourlorcefleld

§ *Exp. data

N

U

El

0

'—

3

or

E

'U

3

I.“

O

l-

0 20 40 60 80 100
TCE SAT ( %)

 

Figure 3.6. Comparison of experimental results with molecular simulations for TCE
adsorption by zeolite at 308K with the competition of water at 100% relative humidity.

In each case, the zeolite pores were initially empty.

37

 

 

*pcfl

_-_
w-uyuoo

p
8

 

 

.0
o
or

 

our force
field

 

 

.0
O
A

 

 

9
o
co

 

.0
0
re

 

_O
O
_A

 

 

 

water loading (g/gzeolite)

 

 

 

0 L‘s—8...; ~...~ Vanua— ‘J
0 I 20 40 60 80 1 00
TCE SAT(%)

 

 

 

Figure 3.7. Comparison of experimental results with molecular simulations for H2O
adsorption by zeolite at 308K with the competition of TCE at 100% saturation. In each

case, the zeolite pores were initially empty.

Table 3.1 and Figures 3.6 and 3.7 represent the loading of TCE and water through
simultaneous adsorption to oven-dried zeolite at 308K. From Table 3.1 and Figure 3.6,
we can see that for TCE loading, all force fields predicted approximately the amount of
TCE at 308K and high TCE vapor pressure; the results of our force field, 0.42 gjg zeolite
are closest to the experimental data. At low TCE chemical potential, our force field and
COMPASS somewhat overpredictcd the amount of TCE sorption, while the PCFF force
field strongly underpredicted sorption below about 50% of the TCE saturation vapor
pressure. At high TCE pressures, the experimental data are again difficult to interpret
because of condensation on external surfaces as discussed above, but 0.45 g TCE/g

zeolite is the experimental estimate based on the inflection point in Farrell’s data (Figure

38

3.2). Furthermore, this value agrees well with the measurements of another group (Pires,
Carvalho et a1. 2001) who found 0.44 g TCE/g zeolite adsorbed to either H-exchanged or
Na-exchanged zeolite Y.

The pore-space in the zeolite is enough to hold about 0.48 gTCE/g zeolite, which
is the upper limit of the adsorption amount since the pore space was measured using a
1.2A sphere. Since T CE is much larger, the volume accessible to TCE is much smaller.
Comparing these three force fields, both COMPASS and PCFF under-predicted the
amount of water adsorption by zeolite, which also happened in pure water adsorption by
zeolite as described above. Even with our force field, competition fi'om TCE suppressed
water sorption by 80 to 90%. Notice that the partial charges on atoms of the zeolite
lattice have much less impact on the amount of TCE sorbed than on the amount of water
sorbed. This is to be expected since the partial charges on TCE itself are small. This
may be the reason that PCFF and COMPASS are widely used even though the previous
section shows they poorly predict water sorption to the zeolite. While there are no
reported measurements of water sorption in the presence of TCE, our prediction of 3 to
4% by weight is plausible given that the pores are filled and that the TCE sorption agrees

with experimental data.

39

 

2.E+04
2.E+04
2.E+04
1.E+04
1.E+04

8. E+03

Configuratlons
i"

99’
r+nrn
as

2.E+03
0.E+00

 

-60 -5O -40 -30 -20 ~10 0
Energy (keel/moi)

 

 

 

Figure 3.8. Energy distribution for TCE and water during simulation of TCE and water

competitive adsorption by zeolite at 308K.

Figure 3.8 shows the simulated TCE and water energy distributions in zeolite HY80. The
small peak between —30 and —40 kcal/moi corresponds to the nonbonded energy for water
molecules adsorbed directly to the HI cations in the zeolite. Figure 3.8 shows that 12-
member ring channels with diameter about lO-llA can hold both TCE and water, while
the 6-member ring channels with diameter about 5-6A can only adsorb water molecules.
This remained the same in the whole framework. Water molecules were occasionally
sorbed in all the 12-member ring channels, but were far more common as small cluster in
those 12-member ring channels that contained Al-substitution sites and H+ cations.

The energy distribution in Figure 3.8 can be subdivided to show the locations of
sites of a specific energy (Figures 3.9 and 3.10). As indicated above, the most strongly

bound water molecules are coordinated directly to H+ cations (Figure 3.10 a). The least

40

strongly bound water sorb at 6-9 kcal/mol and seem to occupy the 6-member ring
channels (green areas in Figure 3.10c). Less frequently, this loosely bound water is
found in the 12-member rings. The next-most weakly bound water (red-orange areas of
Figure 3.10c and orange-green areas of Figure 3.9c) at —10 to —13 kcal/mol sorbs in a set
of 6-member ring channels that are distinct from the most weakly-bound water then there
is a broad group of water sorbing from about —20 to —10 kcal/moi that are mostly found
in the water clusters in the large pores near substitution sites (blue-green areas of Figure
3.90). These water molecules seem to be found in the second and third coordination
spheres around the H+ cations.

TCE molecules sorb at a single broad energy from —13 to —22 kcal/moi (Figure
3.8), centered at about —18 kcal/mol. This value agrees very well with experimental
calorimetric measurements (Mellot, Cheetham et al. 1998) of 17.2 kcal/moi for the heat
of TCE sorption to dry zeolite NaY. Comparison of all TCE (red areas of Figure 3%)
versus the more weakly-bound TCE (blue areas of Figure 3.9c) show little trend other
than TCE molecules near water cluster having somewhat lower energies, TCE seems to
adopt a rather smooth distribution of energies regardless of their location in the large

pores of the zeolite.

41

.36:an 2158a c8 mega 30:0» was 68:80— 3188a 36:0 he 8.83. SE

doqumuu coca $.55 new “520 v.32 .o .533 Be €53 2&3 Ea mop. c8 meg—Sm

com 50:95.8.“ Baa 835m c8 use—o 3&2 .o .5388? 883 BESN d dd 2=wE

 

42

BEE o 9 NT seem a 6:580. NT 9 81 52m a ._oe\_§_ mm. 2 91 52m
d .oEooN actuégo .3 5:983 583 28 mob he 3.20:0 we use—o v.32 .2 .m 25mm

 

43

 

 

Figure 3.11. a. Zeolite before adsorption b. TCE locations inside zeolite after TCE and
water coadsorption. c. water locations inside zeolite after TCE and water coadsorption

C

 

Note from Figure 3.9 b that there is little mixing of TCE and water in the zeolite
structure, despite their simultaneous coadsorption. This is consistent with hypotheses
(Corley, Hong et al. 1996; Farrell, Hauck et al. 1999) of capillary-phase separation as a
factor controlling sorption of dissimilar solvents in microporous media. Figure 3.11 is
another representation of the localized separation of TCE and water in this zeolite.
Figure 3.12 is the radial distribution function for chlorine atom in TCE and oxygen atom
in zeolite. The closest distance from chlorine atoms to oxygen atoms is about 3.0
angstrom, which is in agreement with the sum of the van der Waals radii of chlorine and
oxygen. The positions of the hydrogen cations in the zeolite have a considerable
influence on the TCE and water adsorption. Figure 3.13 is the radial distribution function
for H+ in zeolite and oxygen atom in water. The first maximum for the distance between
them is about 1.6 to 1.8 angstrom, which agrees with the van der Waals radii of hydrogen
and oxygen. The oxygen in water can form a hydrogen bond with H+ in zeolite, then
other water molecules can interact with these water molecules via more hydrogen bonds
and may form water clusters. Comparing with Figure 3.11, we can see that chlorine
atoms do not interact with H+ closely because water is preferentially adsorbed near the

cation.

45

 

 

 

0.E+00 2.E+00 4.E+00 6.E+00 8.E+OO ,1 .E+01
r (angstrom)

 

 

 

Figure 3.12. Radial distribution function for hydrogen cation in zeolite and oxygen atom

in water.

 

 

0.E+OO 2.E+00 4.E+00 6.E+00 8.E+OO 1.E+01
r (angstrom)

 

 

 

Figure 3. 13. Radial distribution function for chlorine atoms in TCE and oxygen atoms in

the zeolite.

3. In order to simulate experimental system c of Farrell’s group, we pre-
equilibrated the model zeolite HY 80 with water at 308K and 100% relative humidity for
4"‘107 steps. After equilibration after about 2"‘107 steps, the zeolite crystal had adsorbed
about 370 water molecules, or 0.27 g/g zeolite (Figure 3.14). This is the same endpoint
reached in our previous result for pure water adsorption. Beginning with this equilibrated
system, the simulation conditions were changed so that the zeolite was exposed to both
TCE and water at 308K and their respective saturation pressures. From 2"‘107 steps to
1"‘108 steps, water was slowly displaced by TCE (Figure 3.14). From 1*108 steps to
2.5"‘108 steps, the displacement was faster, and after 2.5"‘108 steps the equilibrium was
achieved again. The final equilibrium adsorption amount for water is 0.038 g/g zeolite,
and for TCE is 0.44 g/g zeolite. Figure 3.15 compares the pure zeolite before adsorption,
a snapshot at 4"‘107 steps when the zeolite is firll of water, and a snapshot at 3"‘108 steps,
when water has mostly been replaced by TCE. Comparing these three pictures, we can
see that water was initially sorbed into both 12-member ring channels and 6-member ring
channels. As TCE entered, all the water inside 12-member ring channels was displaced
by TCE except those water molecules close to Hi cations. This means the water has a
stronger interaction with the cations than TCE molecules, but in the large cavities the
TCE interacted more strongly with the siliceous zeolite framework than did water. The
water molecules can form strong hydrogen bonds with H+ cations, then form small water
clusters, which are shown on Figure 3.15. The water inside 6-member ring channels
cannot be replaced by T CE because it is difficult for TCE with its larger size to enter into

these channels.

47

The final result for this simulation, with 0.44 g TCE/g zeolite and 0.04 g H20/g
zeolite adsorbed, agrees very closely with the equilibrium achieved when TCE and water
competed for the empty zeolite as reported in section 2 above. Thus, preequilibration of
the mineral with water had no effect on the simulation equilibrium that was eventually
attained. This is in contrast to the experimental result (Figure 3.2) in which pre-
equilibration with water reduced TCE sorption as compared with TCE-water competition
for sites in the degassed zeolite. The experimental system is hysteretic but the simulated
system does not seem to be. Possible reasons for the deviation of the simulation results
from experimental data might include a) In the experimental system, preadsorption of
water may have blocked external sites at pore entrances and thus inhibited the uptake of
TCE by the zeolite. No such kinetic artifacts exist in the simulation system. b) The force
field methodology that we have created may still over —predict TCE-zeolite binding and
under-predict water-zeolite binding, despite the fact that TCE and water sorption agreed
quite well with experimental data in sections 1 and 2 above. c) The simulation was at
100% TCE vapor pressure, while the experiment ended at 95% of saturation. At that
point the experimental adsorption was 0.44 gTCE/g zeolite, which would agree perfectly
with the simulation result, but the experimental value probably included some sorption to
zeolite external surfaces or water films.

Investigations to further test the force field simulations of the system should
include GCEMC modeling of the entire TCE adsorption isotherm for comparison with
Figure 3.2. A potential difficulty is that simulation at lower TCE pressures may take even

greater computational effort before equilibrium is reached (see Figure 3.14).

48

 

 

 

‘ —water

450 —TCE
400
350
300
250
200
150
100
50
0
0.0E+00 5.0E+07 1.0E+08 1.5E+08 2.0E+08 2.5E+08 3.0E+08

simulation steps

 

 

 

loading per unit cell

 

 

 

 

Figure 3.14. Trajectory for water and TCE adsorption by zeolite pre-equilibrated with

100% humidity water at 308K.

49

 

 

 

50

Figure 3.15. a. Zeolite before adsorption. b. Zeolite after equilibration with water at 308K and

100% humidity. c. Zeolite after TCE and water coadsorption onto system b.

4. TCE and water competition for a zeolite that was pre-equilibrated with TCE.
In order to investigate another approach to equilibrium, we also modeled pre-
equilibration of the zeolite with TCE alone, followed by competition between TCE

and water.

 

—TCE
—water

90
80
70
60
50
40
30
20
10

0
0.E+Oo 2.E+07 4.E+07 6.E+07 8.E+O7

configurations

loading/unit cell

 

 

 

 

Figure 3.16. Trajectory for competitive TCE and water adsorption by zeolite after pre-

equilibration with TCE alone.

For‘the pre-equilibration phase, 2*107 GCEMC steps were run at 308K with a TCE
pressure of 15.75 kPa, or 100% of its saturation vapor pressure. The TCE loading of the
zeolite during this phase of the simulation was 0.45 g TCEJg zeolite, which is just 2 to
7% higher than the TCE loadings we observed previously in competition with water.
After this pre-equilibration, the GCEMC simulation conditions were changed to a

competition between water and TCE, both at 100% of their saturation vapor pressures.

51

The simulation was run for a further 6"‘107 steps (for a total simulation of 8*107) and the
overall trajectory is depicted in Figure 3.16. After the competition between TCE and
water had resolved to equilibrium, the zeolite contained 0.029 g H20/ g zeolite and 0.46 g
TCE/g zeolite. The water adsorption for this system was thus about 20% less than that
for the other systems, where TCE and water competed for either an empty zeolite or a
zeolite that had been pre-equilibrated with water. Conversely, TCE sorption for this
system was 5-10% higher than it was in the other systems. Pre-equilibration of the
zeolite with TCE has therefore introduced some hysteresis into the competitive
interaction between water and TCE.

The system energy trajectory (Figure 3.17) shows that the 79 TCE molecules
adsorbed during the pre-equilibration phase had the total adsorption energy (for the whole
unit cell) of about —1400 kcal/moi, for an average of about —17.5 kcal/moi TCE. Figure
3.18 shows that this average energy was very smoothly distributed over a continuum of
sites, indication that there were no especially strong interactions between TCE and the Hi
cations in the zeolite. This is plausible since the partial charges on TCE atoms are so
low. (Table 2.4)

The radial distribution function for this interaction between C] atoms of TCE and

H+ cations in the zeolite is given in Figure 3.19.

52

 

energy (kcal/moi)

-2000

 

-2500
0.E+00 1.E+07 2.E+O7 3.E+07 4.E+07 5.E+O7 6.E+O7 7.E+07 8.E+07
configurations

 

 

 

Figure 3.17 . Energy change for the adsorption of TCE and water by zeolite that was pre-

equilibrated with TCE.

 

2.E+04
2.E+04

ages
Essa

8.E+03
6.E+03
4.E+03
2.E+03
O.E+00

configurations

 

s

-20 -10 0 10 20
Energy (kcal/moi)

 

 

 

Figure 3.18. Energy distribution for pure TCE adsorbed by zeolite HY80 in the absence

of competition from water.

The radial distribution function in Figure 3.19 shows that when there is no water
competing with TCE, the closest distance between chlorine atom in TCE and hydrogen
cation is about 2.0 A. After equilibration with water, this distance changed to 4.0 A.
This means that the TCE molecules close to the cation were displaced by water (Figure
3.20). The energy distributions of both TCE and water after competition (Figure 3.21)
again show that water forms the stronger bond (about —35 kcal/moi) with the cation than
the TCE interaction with the cation. But unlike the results discussed in section 3, since
TCE molecules enter into the zeolite first, these molecules tend to occupy all the sites
near the cations. Even though water can replace the TCE close to cations, the larger
water clusters are not able to form, so there is somewhat less water adsorption and
slightly more TCE adsorption than the situation in which water entered the zeolite lattice
first. Water cannot displace the TCE inside 12-member ring channels at locations far
from hydrogen cations. Water molecules also enter into the 6-member ring channels,
which are not available to TCE molecules. The spatial distribution of water and TCE
after their competitive sorption to this zeolite system is shown in Figure 3.22.
Comparison of Figures 3.22 and 3.21 with the results from previous simulations (Figure
3.8 and Figure 3.9) show clearly the locations of the hysteresis. Pre-equilibration of the
zeolite with TCE results in more TCE filling the large pores near HI cation sites at
energies near —20 kcal/moi. Once water is introduced, some of these TCE molecules
prevent the expansion of water clusters (at water sorption energies of —10 to —20

kcal/moi) that were able to form in the absence of pre-equilibration with TCE.

54

 

6.E+00
5.E+00
4.E+00

RDF

3.E+00
2.E+00
1.E+00

0.E+OO
0.E+OO 2.E+00 4.E+00 6.E+00 8.E+OO 1.E+O1

r (angstrom)

 

 

 

 

Figure 3.19. Radial distribution function for the hydrogen cation in zeolite and chlorine

atoms in TCE in the absence of competition by water.

 

3.E+02
3.E+02
2. E+02

2. E+02

RDF

1 .E+02

5.E+01

 

0. E+00
0. E+OO 2. E+00 4.E+00 6.E+00 8. E+00 1.E+01

r (angstrom)

 

 

 

Figure 3.20. Radial distribution function for the hydrogen cation in zeolite and oxygen

atoms in water after the entire simulation summarized in Figure 3.16.

 

—water
—TCE

1 .E+05
1 .E+05
1 .E+05
8.E+04
6.E+04
4.E+O4
2.E+04

configurations

 

0. E+00
-50 -40 ~30 -2O -1 0 0

Energy (kcal/moi)

 

 

 

Figure 3.21. Energy distribution for TCE and water adsorbed by zeolite HY80 pre-

equilibrated with 100% TCE.

56

 

 

.n 52: Ewen 808 2 o E 580 83.5w 05 worm a a can we...— m_ 0 c8 3on couflsfia 2: 85m
.mUH $03 5:3 6035:869an 8:03 .3 388 .833 23 HUB .o .oEoou 3 @088 images—:65 can?

can mug. .n .0508 mem d .8; 8:03 2 8:883 has? 28 MOE c8 23.5w we “520 $2: 05. .Nm.m Esmfi

 

57

CONCLUSIONS

This study demonstrates the feasibility and utility of grand canonical Monte Carlo
simulations of binary TCE and water adsorption by zeolite. Potential energy parameters

for the system were improved using both quantum mechanical results and empirical

adjustment for internal self-consistency. The resulting force field is able to reproduce the ’
densities and heats of vaporization for bulk TCE and bulk water phases, as well as the 7
structures of numerous aluminosilicates. When applied to more complex systems, the L

 

force field predictions matched experimental data for water sorption by zeolite Y and for
TCE sorption to hydrated zeolite Y. These data indicate that the zeolite HY with Si/Al
ratio 80 is moderately hydrophilic in that the zeolite adsorbs 0.27 g H2O/ g zeolites from
the vapor phase. When TCE and water are simultaneously adsorbed into the zeolite
HY80, the TCE outcompetes water and occupies most 12-member channels. Many water
molecules are relegated to the 6-member channels, which TCE cannot enter due to its
larger molecular size. Some water molecules are adsorbed strongly near the Hi cations in
the zeolite, and other waters interact with them to form small water clusters in the 12-
member channels. Water and TCE domains remain quite separate in the zeolite, with
very little mixing of the two species even when clusters of each are present in the same
pore.

The same equilibrium is reached in the molecular simulations whether TCE and
water compete for an empty zeolite or for a zeolite that has been pre-equilibrated with
pure water. Pre-equilibration of the zeolite with pure TCE results in a slightly different

final equilibrium in which somewhat more TCE and somewhat less water is sorbed.

58

These simulation findings are somewhat in contrast to the experimental results, in which
pre-adsorption of water reduces TCE adsorption by about 10%. The experimental data
are somewhat difficult to compare with the simulations owing to capillary condensation
on external surfaces in the experimental system, while the simulated zeolite has no
external surfaces.

Two commercial force fields (PCFF and COMPASS) were tested and compared
with our force field. All force fields predicted the amount of TCE adsorption by the
zeolite in approximate agreement with experiment at high TCE pressures, but PCFF
predicted sorption that was much too low at low TCE pressures. The zeolite as modeled
by COMPASS and PCFF is too hydrophobic overall because their predictions of water
sorption are much smaller than the experimental data. Our simulation results are in far
better agreement with the overall experimental data and especially for the water sorption
data.

Suggested projects for further validation and improvement of our force field
include simulations of zeolites with other exchangeable cations that might be allowed to
be mobile within the zeolite pores, and simulations of broader parts of the isotherm in the
case of TCE-water competition for zeolites that have been pre-equilibration with water.
The latter might be expected to be the most relevant GCEMC molecular modeling
systems for studying TCE sorption to zeolites and other minerals in the environment. In
this respect, simulation and experimental studies of sorption to minerals at high water

activity and low TCE activity should be pursued.

59

l'

REFERENCES

Allinger, N. L., Y. H. Yuh, et al. (1989). “Molecular mechanics. The MM3 force field for
hydrocarbons. 1.” Journal of the American Chemical Society 111(23): 8551—8565.

Alvarez-Cohen, L., P. L. McCarty, et al. (1993). “Sorption of trichloroethylene onto a
zeolite accompanied by methanotrophic biotransformation.” Environmental
Science and Technology 27: 2141-2148.

 

Bremard, C., G. Buntinx, et al. (1997). “Vibrational studies and Monte Carlo simulations
of the sorption of aromatic carbonyls in faujasitic zeolites.” J. Mol. Struct. 410-
411: 379-382.

 

Breneman, C. M. and K. B. Wiberg (1990). “Determining atom-centered monopoles from
molecular electrostatic potentials. Tire need for high sampling density in
forrnamide conformational analysis.” Journal of Computational Chemistry 11(3):
361 -3 73.

CF R (Code of Federal Regulations) (1993). National primary drinking water regulations.
40 CFR, Part 141, pp. 592-731. US. Gov't. Printing Office, Washington, DC.

Channon, Y. M., C. R. A. Catlow, et al. (1998). Grand Canonical Monte Carlo
Investigation of Water Adsorption in Heulandite-type Zeolites. J. Phys. Chem. B.
102: 4045-4048.

 

Chen, N. Y. (1976). Hydrophobic properties of zeolites. J. Phys. Chem. 80: 60-4.

 

Chiou, C. T., L. J. Peters, et al. (1979). “A physical concept of soil-water equilibria for
nonionic organic compounds.” Science 206: 831-832.

Chirlian, L. E. and M. M. F rancl (1987). “Atomic charges derived from electrostatic
potentials: A detailed study.” Journal of Computational Chemisty 8(6): 894-905.

 

Corley, T. L., B. Hong, et al. (1996). “VOC Accumulation and Pore Filling in
Unsaturated Porous Media.” Environ. Sci. Technol. 30(10): 2884-2891.

 

Cossi, M., V. Barone, et al. (1996). “Ab initio study of solvated molecules: a new
implementation of the polarizable continuum model.” Chem. Phys. Lett.
255(4,5,6): 327-335.

Curtiss, G. P., P. V. Roberts, et al. (1986). “A natural gradient experiment on solute
transport in a sand aquifer. 4. Sorption of organic solutes and its influence on
mobility.” Water Resources Research 22: 2059.

 

60

 

Davidson, A. M., C. F. Mellot, et al. (2000). “An Inelastic Neutron Scattering and NIR-
FT Raman Spectroscopy Study of Chloroform and Trichloroethylene in
Faujasites.” J. Phys. Chem. B 104(3): 432—438.

 

Doherty, R. E. (2000). A history of the production and use of carbon tetrachloride,
techachloroethylene, trichloroethylene and 1,1,1-trichloroethane in the United
States. Journal of Environmental Forensics. 1: 83-93.

 

Douguet, D., R. J. M. Pellenq, et al. (1996). “The adsorption of argon and nitrogen in
silicalite-l zeolite: a grand canonical Monte-Carlo study.” Moi. Simul. 17(4-6):
255-288.

Estes, T. J., R. V. Shah, et al. (1988). “Adsorption of low molecular weight halocarbons
by montmorillonite.” Environmental Science and Technology 22(4): 377-381.

 

Farrell, J ., D. Grassian, et al. (1999). “Investigation of Mechanisms Contributing to Slow
Desorption of Hydrophobic Organic Compounds fiom Mineral Solids.” Environ.
Sci. Technol. 33(8): 1237-1243.

 

Farrell, J., B. Hauck, et al. (1999). “Thermodynamic investigation of trichloroethylene
adsorption in water-saturated microporous adsorbents.” Environmental
Toxicology and Chemistry 18(8): 1637-1642.

 

 

Farrell, J. and M. Reinhard (1994). “Desorption of halogenated organics fiom model
solids, sedimants, and soil under unsaturated conditions. 1. Isotherms.”
Environmental Science and Technology 28: 53-62.

 

Farrell, J. and M. Reinhard (1994). “Desorption of halogenated organics from model
solids, sedimants, and soil under unsaturated conditions. 2. Kinetics.”
Environmental Science and Technology 28: 63-72.

 

Francl, M. M., C. Carey, et al. (1996). “Charges fit to electrostatic potentials. II. Can
atomic charges be unambiguously fit to electrostatic potentials?” Journal of
Cormmtational Chemistry 17: 367-383.

 

Fuchs, A. H. and A. K. Cheetham (2001). “Adsorption of guest molecules in zeolitic
materials: Computational aspects.” J. Phys. Chem. B 105(31): 73 75-73 83.

 

Green, R. E. and S. W. Karickhoff (1990). Sorption estimates for modeling. Pesticides in
the soil environment: Processes, impacts, and modeling. H. H. Cheng. Madison,
WI, Soil Science Society of America: 79-101.

 

 

Halgren, T. A. (1996). “Merck molecular force field. 1. Basis, form, scope,
parameterization, and performance of MMFF94.” Journal of Computational
Chemistry 17: 490-519.

61

 

Halgren, T. A. (1996). “Meer molecular force field. H. MMF F94 van der Waals and
electrostatic parameters for intermolecular interactions.” Journal of
Computational Chemistry 17: 520-552.

 

Halgren, T. A. (1996). “Merck molecular force field. III. Molecular geometries and
vibrational frequencies for MMFF94.” Journal of Computational Chemistry 17:
553-586.

Halgren, T. A. (1996). “Meer molecular force field. V. Extension of MMFF 94 using
experimental data, additional computational data, and empirical rules.” Journal of
Computational Chemistry 17 : 616-641.

 

Halgren, T. A. and R. B. Nachbar (1996). “Merck molecular force field. IV.
Conformational energies and geometries for MMFF94.” Journal of Computational
Chemist_ry 17: 587-615.

 

Harr, J. (1996). A Civil Action, Random House, Inc., New York.

 

Hehre, W. J., L. Radom, et al. (1986). Ab initio molecular orbital theory. New York, NY,
John Wiley & Sons.

Hill, J .-R. and J. Sauer (1994). “Molecular mechanics potential for silica and zeolite
catalysts based on ab initio calculations. 1. Dense and microporous silica.”
Journal of Physical Chemistry 98(4): 1238-1244.

 

Hill, J .-R. and J. Sauer (1995). “Molecular mechanics potential for silica and zeolite
catalysts based on ab initio calculations. 2. Aluminosilicates.” Journal of Physical
Chemisfl 99: 9536-9550.

 

Jackson, R. E., S. Lesage, et al. (1992). Estimating the fate and mobility of CF C-1 13 in
groundwater: Results from the Glouchester landfill project. Groundwater
contamination and analysis at hazardous waste sites. S. Lesage and R. E. Jackson.
New York, NY, Marcel Dekker: 511-526.

 

 

J orgensen, W. L. and J. Tirado-Rives (1988). “The OPLS potential function for proteins.
Energy minimizations for crystals of cyclic peptides and crambin.” Journal of the
American Chemical Sociefl 110(6): 1657-1666.

 

 

Karasawa, N. and W. A. Goddard, III (1989). “Acceleration of convergence for lattice
sums.” Journal of Physical Chemistry 93: 7320-7327.

 

Klemm, E., J. Wang, et al. (1998). “A comparative study of the sorption of benzene and
phenol in silicalite, HAlZSM-S, and NaAlZSM-5 by computer simulation.”
Microporous Mesoporous Mater. 26(1-3): 11-21.

 

62

 

Landolt, G. R. (1971). Method for rapid determination of adsorption properties of
molecular sieves. Anal. Chem. 43: 613-15.

 

Li, H. X., M. J. Annen, et a1. (1991). “Dealumination of hexagonal polytype of faujasite
by treatments with silicon tetrachloride vapor.” J. Mater. Chem. 1(1): 79-85.

Li, J. and C. J. Werth (2001). Evaluating competitive sorption mechanisms of volatile
organic compounds in soils and sediments using polymers and zeolites.
Environmental Science & Technology. United States. 35: 568-74.

 

Li, J. and C. J. Werth (2001). “Evaluating competitive sorption mechanisms of volatile
organic compounds in soils and sediments using polymers and zeolites.”
Environmental Science & Technology 35(3): 568-574.

 

Lide, D. R. (1995). Handbook of Chemistry and Physics, 76th Edition.

Mellot, C. F ., A. K. Cheetham, et a1. (1998). Halocarbon Adsorption in Nanoporous
Materials: A Combined Calorimetric and Monte Carlo Study of Trichloroethylene
(TCE) in Faujasite-Type Zeolites. Langmuir. 14: 6728-6733.

Mellot, C. F ., A. K. Cheetham, et al. (1998). “Halocarbon Adsorption in Nanoporous
Materials: A Combined Calorimetric and Monte Carlo Study of Trichloroethylene
(TCE) in Faujasite—Type Zeolites.” Langmuir 14(23): 6728-6733.

Metropolis, N., A. W. Rosenbluth, et al. (1953). “Equation of state calculations by fast
computing machines.” The Journal of Chemical Physics 21(6): 1087-1092.

 

Moller, C. and M. S. Plesset (1934). “Note on an approximate treatment for many-
electron systems.” Physical Review 46: 618-622.

 

MSI (1999). Cerius2 Molecular Simulation Software, Release 4.0. San Diego, CA,
Molecular Simulations, Inc.

Nascimento, M. A. C. (1999). “Computer simulations of the adsorption process of light
alkanes in high-silica zeolites.” Journal of Molecular Structure-Theochem 464(1-
3): 239-247.

Newsam, J. M., H. Sun, et al. (1999). “Force-field technology suitable for calculating
thermodynamic parameters of condensed phases.” Book of Abstracts, 217th ACS
National Meeting, Anaheim, Calif, March 21-25: COMP-096.

 

Pignatello, J. J. (1990). “Slowly reversible sorption of aliphatic halocarbons in soils. II.
Mechanistic aspects.” Environmental Toxicology and Chemistry 9: 1117-1126.

Pires, J., A. Carvalho, et al. (2001). “Adsorption of volatile organic compounds in Y
zeolites and pillared clays.” Microporous Mesoporous Mater. 43(3): 277-287.

63

 

Pires, J. and M. B. De Carvalho (1997). Water adsorption in aluminum pillared clays and
zeolites. J. Mater. Chem. 7: 1901-1904. .

 

Sabatini, D. A. and R. C. Knox (1992). Transport and remediation of subsurface
contaminants: Review and future directions. Transport and remediation of
subsurface contaminants. Colloidal, interfacial, and surfactant phenomena. D. A.
Sabatini and R. C. Knox. Washington, DC, American Chemical Society: 234-240.

Scott, C. S. and V. J. Cogliano (2000). “Trichloroethylene health risks-~state of the
science.” Environ. Health Perspect. 108 Suppl 2: 159-60.

 

Shen, D., S. R. Jale, et al. (1999). “Sorption thermodynamics of nitrogen and oxygen on
CaA zeolite.” Stud. Surf. Sci. Catal. 125(Porous Materials in Environmentally
Friendly Processes): 667-674.

 

Sigg, L. and W. Stumm (1994). Aquatic Chemistry: An Introduction to the Chemistry of
Aqueous Solutions and Natural Waters.

 

Smit, B. (1995). “Simulating the adsorption isotherms of methane, ethane, and propane in
the zeolite silicalite.” Journal of Physical Chemistry 99: 5597-5603.

 

Smit, B. and T. L. M. Maesen (1995). “Commensurate ‘freezing' of alkanes in the
channels of a zeolite.” Science 267: 42-44.

Smit, B. and J. I. Siepmann (1994). “Computer-Simulations of the Energetics and Siting
of N-Alkanes in Zeolites.” Journal of Physical Chemistry 98(34): 8442-8452.

 

Smit, B. and J. I. Siepmann (1994). “Simulating the Adsorption of Alkanes in Zeolites.”
Science 264(5162): 1118-1120. '

Stauffer, T. B. and W. G. MacIntyre (1986). “Sorption of low-polarity organic
compounds on oxide minerals and aquifer material.” Environmental Toxicology
and Chemistry 5: 949-955.

 

 

Stauffer, T. B., W. G. MacIntyre, et al. (1989). “Sorption of nonpolar organic chemicals
on low-carbon-content aquifer materials.” Environmental Toxicology and
Chemistry 8: 845-852.

 

Sun, H. (1998). “COMPASS: An ab initio F orce-Field Optimized for Condensed-Phase
Applications - Overview with Details on Alkane and Benzene Compounds.” ._l_.
Phys. Chem. B 102(38): 7338-7364.

 

Sun, H. (1998). “Integrated force field development for applications in material science.”
Book of Abstracts, 216th ACS National Meeting, Boston, August 23-27: COMP-
159.

64

Sun, H., S. J. Mumby, et al. (1994). “An ab initio cff93 all-atom force field for
polycarbonates.” Journal of the American Chemical Society 116: 2978-2987.

Sun, H., P. Ren, et al. (1998). “The compass force field: parameterization and validation
for phosphazenes.” Comput. Theor. Polym. Sci. 8(1/2): 229-246.

Teppen, B. J ., R. R. Arands, et al. (1998). “TCE interactions with mineral surfaces:
Correlating molecular simulations with experimental data.” In Agronomy
Abstracts. American Society of Agronomy, Madison, WI.

Teppen, B. J ., K. Rasmussen, et al. (1997). “Molecular dynamics modeling of clay
minerals. 1. Gibbsite, kaolinite, pyrophyllite, and beidellite.” Journal of Physical
Chemist_ry B101: 1579-1587.

Teppen, B. J ., C.-H. Yu, et al. (1998). “Molecular dynamics simulations of sorption of
organic compounds at the clay mineral / aqueous solution interface.” Journal of
Computational Chemistry 19: 144-153.

 

Teppen, B. J ., C.-H. Yu, et al. (1998). “Molecular modeling studies of TCE
sorption/desorption on clays.” p. 39 In Meeting program and abstracts, 35th
annual meeting. Clay Minerals Society, Boulder, CO.

 

Wagner, P. and M. E. Davis (2000). Introduction to zeolite science. Photofunct. Zeolites:
1-38.

Wartenberg, D., D. Reyner, et al. (2000). “Trichloroethylene and cancer: epidemiologic
evidence.” Environ. Health Perspect. Suppl. 108(2): 161-176.

 

Weber, G., O. Bertrand, et a1. (1996). “TCE adsorption on hydrophobic Y and MFI
zeolites. Possibilities of using such materials for solvent recovery.” J. Chim. Phys.
Phys-Chim. Biol. 93(7-8): 1412-1425.

 

Weiner, S. J., P. A. Kollrnan, et al. (1986). “An all atom force field for simulations of
proteins and nucleic acids.” Journal of Computational Chemistry 7(2): 230-252.

Werth, C. J. and M. Reinhard (1997). “Effects of temperature on trichloroethylene
desorption from silica gel and natural sediments .2. Kinetics.” Environmental
Science & Technology 31(3): 697-703.

 

Xing, B. S. and J. J. Pignatello (1996). “Time-dependent isotherm shape of organic
compounds in soil organic matter: Implications for sorption mechanism.”
Environmental Toxicology and Chemistry 15(8): 1282-1288.

65