Molecular Dynamics Simulations of Adsorption of Organic Compounds at the Clay Mineral/Aqueous Solution Interface

Abstract

Computational studies of the sorption of organic compounds at clay mineral surfaces are described. Molecular dynamics simulations were performed with a recently developed empirical force field for dioctahedral clays. The studies allow the identification of three general mechanisms of adsorption. In the absence of water, organic compounds adsorb to mineral surfaces in such a way that contact area is maximized. In the presence of a sufficient amount of water, some molecules can adsorb via a single functional group, while the bulk of the molecular structure is immersed in the aqueous phase. When many water molecules are present, they form a structured layer, excluding organic adsorbates from the mineral basal plane. A detailed description is given of the characteristic structure found for water layers in the interlayer space of clays. The calculated trends are reasonable, but we also expect that current dynamics simulations may overestimate the extent of the structuring of water because of the absence of polarization terms in the available empirical force fields.

Full text

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
Structural Chemistry, Vol. 14, No. 2, April 2003 (C°2003)
Molecular Dynamics Simulations of Adsorption of Organic
Compounds at the Clay Mineral/Aqueous Solution Interface1
Ching-Hsing Yu,2Susan Q. Newton,2Mya A. Norman,2
Lothar Sch¨afer,2and David M. Miller3
Received November 29, 2001; revised January 4, 2002; accepted January 14, 2002
Computational studies of the sorption of organic compounds at clay mineral surfaces are described.
Molecular dynamics simulations were performed with a recently developed empirical force field for
dioctahedral clays. The studies allow the identification of three general mechanisms of adsorption. In
the absence of water, organic compounds adsorb to mineral surfaces in such a way that contact area is
maximized. In the presence of a sufficient amount of water, some molecules can adsorb via a single
functional group, while the bulk of the molecular structure is immersed in the aqueous phase. When
many water molecules are present, they form a structured layer, excluding organic adsorbates from
the mineral basal plane. A detailed description is given of the characteristic structure found for water
layers in the interlayer space of clays. The calculated trends are reasonable, but we also expect that
current dynamics simulations may overestimate the extent of the structuring of water because of the
absence of polarization terms in the available empirical force fields.
KEY WORDS: Molecular dynamics simulations; clay minerals; adsorption on clay surfaces; interactions of
organic compounds with clays; soil pollutants.
INTRODUCTION
In 1808 Gay-Lussac predicted [1] that “we are per-
haps not far removed from the time when we shall be able
to submit the bulk of chemical phenomena to calculation.”
The great French chemist could not have anticipated the
computational revolutions that occurred in the mean time
but, nearly 200 years later, we have come very close to
realizing his vision.
In the current paper we will describe some fairly
novel applications of molecular dynamics simulations in
the area of soil chemistry. In soil chemistry, detailed de-
scriptions of the interactions of molecules with mineral
surfaces are of great general interest. During the past
decades soil chemists have increasingly employed spec-
troscopic techniques for this purpose. In support of these
1Dedicated to the memory of Barbara Starck.
2Department of Chemistry University of Arkansas, Fayetteville,
Arkansas 72701.
3Department of Crop, Soil, and Environmental Science, University of
Arkansas, Fayetteville, Arkansas 72701; email: dmmiller @uark.edu
efforts we are exploring the possibilities of atomic scale
computer simulations as a means to help interpret the em-
pirical data.
Withintheframework of aquasi-classical formalism,
molecular dynamics (MD) simulations are an effective
tool for providing information on the properties of large
molecularsystems. Simulationsof thiskind typicallygen-
erate collections of molecular configurations, which can
serve as approximate statistical ensembles. MD simula-
tions are characterized by the fact that the equilibrium
state of a system is not approached in a static way, but
all the atoms and molecules are allowed to move without
any restraints other than those imposed by the force field
employed and the state variables, such as temperature and
pressure. Standard procedures from statistical thermody-
namicscan beusedto calculateensembleaverages,which,
in turn, can be related to the macroscopic properties of
interest.
In the current paper we will describe some general
trends emerging from ongoing MD simulations of the
sorption of organic compounds, specifically pesticides, to
mineralsurfaces. Specialattention will begiven to therole
of water in the adsorption process.
175
1040-0400/03/0400-0175/0 C
°2003 Plenum Publishing Corporation

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
176 Yu, Newton, Norman, Sch¨afer, and Miller
COMPUTATIONAL PROCEDURES
Aluminosilicate clays typically consist of sheets of
aluminum oxide, in which the metal is octahedrally co-
ordinated, bonded to sheets of tetrahedral silicon oxide.
The sheets can be combined in various ways. Pure alu-
minum oxide sheets exist in gibbsite, whose crystal struc-
ture is given in Ref. [2]. Layers of aluminum oxide sheets
bonded to silicon oxide sheets exist in the 1:1 type clay
kaolinite, whose crystal structure is given in Ref. [3]. The
2:1 clay pyrophyllite contains layers of aluminum oxide
sandwiched between two silicon oxide sheets; its crystal
structure is given in Ref. [4].
In order to perform MD simulations of dioctahedral
clays, we have developed [5] potential energy functions
that can be used in combination with the force field sup-
plied with the former MSI/Insight/Discover (now Accel-
rys) software suite [6]. Among various options, the latter
contains the cff91 parameters [7] for standard organic and
inorganic compounds, augmented with parameters for sil-
icates and zeolites developed by Hill and Sauer. [8,9] For
octahedrally coordinated aluminum, it was necessary to
derive[5]additional potentialparameters thatmakeit pos-
sible to perform computations of phyllosilicates. Specifi-
cally,itwasnecessary toderiveanangle-bending potential
for octahedral O Al O angles, which displays dual min-
ima at 90 and 180◦. Furthermore, since the force field by
Hill and Sauer [8,9] was based in part on the results of
ab initio calculations that did not include electron correla-
tion,a new setofnonbonded parameters andpartialatomic
charges was derived from electron-correlated ab initio ge-
ometry optimizations of molecular structures represent-
ing fragments of phyllosilicates. The resulting potential
parameters [5] were further refined with the help of the
X-ray crystal structures of some oxides and phyllosilicate
minerals.
Our MD simulations are characterized by the fact
that all the atoms of a given system, including those in the
mineral lattice, are allowed to move subject only to the
constraints of the force field. This is in contrast to those
previous procedures, which were only partially dynamic
in that only some of the atoms of a system were allowed
to move, while others were held rigid.
General information on the procedural details in-
volved in performing molecular dynamics simulations
can be found in a number of standard texts, such as the
one by Allen and Tildesley [10]. Some characteristic re-
sults of previous modeling efforts by others can be found
for simple oxides [11–14], mixed oxides [15–18], mica
[12,19–21], and smectite clay minerals [22]. References
[23–55]provide asurveyof therich spectrumof topics ad-
dressed and techniques applied, in computer simulations
of minerals. The excellent book by Cygan and Kubicky
[56] illustrate the current importance of this field.
Detailed descriptions of specific technical aspects of
ourcalculations havebeen given [57–60] andonly some of
the main points shall be repeated here. In all our studies,
we construct mineral surfaces from unit cells of repre-
sentative systems obtained from X-ray crystal structures.
For example, the experimental coordinates determined by
Bish [3] for the 1:1 clay mineral kaolinite can be used to
construct a surface for simulations by fusing several unit
cells together to form a supercell. A supercell with com-
position Al32Si32O80(OH)64, for example, offers a nearly
rectangular repeat unit base of 20.61 by 17.88 ˚
Ainthe
crystallographic ab-plane [57].
Unlike smectites, kaolinite does not typically swell
along the c-axis upon hydration. The layers can never-
theless be separated in the simulations, increasing the c
axis from 7.4 ˚
A, the crystal spacing [3], to some 20.0
˚
A or larger, in order to create an interlayer space of de-
sireddimension,where thebehaviorofwaterand adsorbed
molecules can be modeled. Kaolinite is an interesting
model mineral, since it presents two very different types
of surfaces to aqueous solutions. The artificial expansion
of the interlayer space creates a pore, which provides both
typesof basal externalsurfacefoundon akaolinitemineral
grain. Slit pores of this type are found in kaolinite books
[61] and perhaps at interfaces between silicate grains and
aluminum oxide coatings.
Systems with an artificially expanded interlayer
space are modeled under NVT (constant mass, volume
and temperature) conditions, if the intention is to keep
the d(001) spacing constant. In order to equilibrate an
expanded system, simulations are subjected to NPT (con-
stant mass, pressure, and temperature) conditions. During
that process, the separated layers spontaneously anneal,
restoring the equilibrium interlayer spacing characteristic
of a given system.
In other MD simulations [59], we have used the crys-
tallographic structure [4] of pyrophyllite, an uncharged,
2:1, dioctahedral phyllosilicate, to construct a model sys-
tem for studying adsorption. For example, fusing six unit
cells of pyrophyllite will yield an Al24Si48O120(OH)24 su-
percell of a neutral, idealized 2:1 clay. Pyrophyllite is well
suited for study because it has the same structure as the
smectites, but it is a neutral clay and the interlayer space
is thus devoid of any hydrated counter ions. When Si ions
in the tetrahedral sheet of a pyrophyllite supercell are iso-
morphically substituted with Al, an idealized beidellite
system results with a cation-exchange capacity dependent
on the chosen extent of substitution. Alternatively, model
montmorillonite systems can be constructed by isomor-
phic substitutions of Al in the octahedral sheet by Mg.

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
Molecular Dynamics Simulations of Adsorption of Organic Compounds 177
The mineral surfaces constructed in this way are usu-
ally hydrated at various levels, producing various layers
of water, as needed. Subsequently, the adsorbates to be
studied are added to the interlayer space or pore surface
and subjected to MD simulations for time periods of, typ-
ically, several hundred picoseconds, using a time step of
0.5 fs.
RESULTS AND DISCUSSION
Three Mechanisms of Adsorption of Organic
Compounds on Mineral Surfaces
The structures of two pesticides, 2-chloro-4-
ethylamino-6-isopropylamino-S-triazine, commonly
called atrazine (ATZ), and of 2-methyl-4,6-dinitro
phenol, commonly called DNOC, are shown in Fig. 1. We
are currently studying the adsorption of both systems on
a model montmorillonite with a cation-exchange capacity
of105 meq/100g, usingK+ascounter ions.Other counter
ions are also considered because they affect the transport
properties of pesticides on soils [62]. The behavior of
DNOC and ATZ on clays follows trends generally found
for the sorption of organic compounds [57–60].
When organic compounds like ATZ and DNOC are
placed with random orientations in the clay interlayer
space, as shown for DNOC in Fig. 2, they will move spon-
taneously during MD simulations to the mineral surface
and adsorb coplanar with the mineral basal plane (Fig. 2).
During this process, the counter ions will perform a simi-
lar movement, sometimes forming ion bridges, as seen for
Fig. 1. The molecular structures of 2-chloro-4-ethylamino-6-isopropylamino-S-triazine, commonly called atrazine (ATZ, on
the left), and of 2-methyl-4,6-dinitro phenol, commonly called DNOC (on the right).
DNOC in Fig. 2. Interestingly, no ion bridging was found
for ATZ.
The tendency to maximize contact area is a general
characteristic for organic compounds on dry mineral sur-
faces, when no water is present. In the case of peptides
and proteins [60] it is the cause of significant denaturing,
i.e., large changes in φ,ψ-torsional angles [60]. Details
of how adsorbed species in this state interact with silox-
ane surfaces are shown in Fig. 3. In a characteristic way,
adsorbing species tend to point a functional group to the
inside of a hexagonal siloxane cavity, but not to the ex-
act center. Off-center adsorption is particularly clear for
single ions (Fig. 3).
In the presence of water, the adsorption mechanism
can change significantly because water can displace or-
ganic compounds from surface sites, and because polar
compounds show an affinity of their own for aqueous so-
lutions. Basically, two different processes are possible in
the presence of water.
When the amount of water in the interlayer space
is sufficiently high for forming several water layers (see
below), organic compounds can attach to a mineral sur-
face via a single functional group, while its bulk is im-
mersed in the aqueous phase. Such a case is shown in
Fig. 4, in which DNOC is seen to sorb via one of its NO2
groups, while the main part of its body remains immersed
in the aqueous phase. Similarly, ATZ was found to inter-
act with a mineral surface by penetrating a water layer
with its C-Cl bond. The same mechanism was previously
obtained for trichloroethene [57]. In the single-group ad-
sorption mode, the adsorbates are typically mobile and

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
178 Yu, Newton, Norman, Sch¨afer, and Miller
Fig. 2. When DNOC molecules are placed with random orientations in the interlayer space of a dry montmorillonite (top), they will move
spontaneously during MD simulations to the mineral surface plane and adsorb coplanar with the mineral basal plane (bottom).

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
Molecular Dynamics Simulations of Adsorption of Organic Compounds 179
Fig. 3. The positions of adsorbed ATZ (left), DNOC (right), and counter ions (K+, isolated spheres in both figures) relative to the hexagonal siloxane
cavity of montmorillonite.
Fig. 4. In the presence of several layers of water, an organic molecule can adsorb at a mineral surface via a single group, while
the bulk of its structure is immersed in the aqueous phase, as shown here for DNOC on montmorillonite (snapshot after 250 ps
of MD simulations). For graphic clarity, water molecules are rendered in the line-drawing mode, while the ball and stick mode
was chosen for DNOC and the mineral lattice.

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
180 Yu, Newton, Norman, Sch¨afer, and Miller
Fig. 5. Histogram constructed from 5000 time steps of MD simulations of an equilibrated ensemble of
DNOC, water, and K+ions on montmorillonite. The frequencies (vertical axis) are shown of distances (˚
A,
horizontal axis) between DNOC–oxygen atoms and K+ions.
Fig. 6. Rubredoxin with six DNOC molecules and 2070 water molecules in an artificially enlarged interlayer space of
pyrophyllite. DNOC can interact both with the mineral surface and the protein. After 150 ps of NPT dynamics, the protein
retained its globular shape and remained inside the aqueous solution phase. For reasons of graphic clarity, water molecules
are rendered in the line-drawing mode.

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
Molecular Dynamics Simulations of Adsorption of Organic Compounds 181
Fig. 7. When the interlayer space of a clay mineral is randomly soaked with water (top), MD simulations will lead spontaneously, within
a few picoseconds, to the formation of several structured water layers (bottom).

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
182 Yu, Newton, Norman, Sch¨afer, and Miller
able to diffuse significantly, jumping from one surface site
to another. In contrast, coplanar adsorbates are effectively
immobilized.
The ability of pesticides to form complexes with
counter ions is an important factor in their transport prop-
erties in soils [62]. To investigate whether such complexes
are being formed to a statistically significant extent be-
tween DNOC and K+, the histogram of Fig. 5 was gen-
erated, involving distances between DNOC–oxygen and
K+on montmorillonite. From that analysis (Fig. 5), short-
rangeinteractions betweenDNOC–NO2andK+appear as
relatively infrequent. Further analyses with different ions,
such as Ca2+, are currently under investigation.
Fig. 8. The structure of the aqueous phase (∼2000 water molecules) of Fig. 6, after 150 ps of MD simulations. Note the distinct layers of water
immediately adjacent to the mineral surfaces and separated from the bulk phase in the interior of the interlayer space.
When the interlayer space is soaked with a large
amount of water, DNOC and ATZ show a tendency to
remain completely in the aqueous phase without mak-
ing contact with the mineral surface. We have obtained
the same result before for TCE [57] and methylene blue
[59]. For proteins in proximity to mineral surfaces, large
amountsofwater allow for retainingtheglobularstructure,
which dry surfaces destroy. Figure 6 shows the protein,
rubredoxin, in the presence of DNOC and some 2000 wa-
ter molecules on pyrophyllite. It is seen that, after 150 ps
of dynamics, the essential globular structure has been pre-
served.In asystem of thiskind, agivenpesticide caninter-
actwith boththe surfacesofthe mineraland ofthe protein.

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
Molecular Dynamics Simulations of Adsorption of Organic Compounds 183
Fig. 9. Distances ( ˚
A) between water–oxygen atoms and the crystallo-
graphic ab plane (vertical axis), and the bc plane (horizontal axis) taken
from a snapshot of a random starting configuration (top) and of an MD
equilibrated configuration (bottom) of water in a montmorillonite inter-
layer space.
The Structure of Water in Mineral Interlayer Spaces
In our studies of hydrated clay surfaces, we noticed
thatdynamics relaxationwill leadina shorttime fromper-
fectly disordered starting configurations to highly struc-
tured arrangements of water. The phenomenon is illus-
trated in Fig. 7, where three distinct water layers are seen
in a typical snapshot of hydrated DNOC on montmoril-
lonite. When the loading is very high, as in the system
shown in Fig. 8, narrow water layers will form adjacent to
a surface and are visibly separated from the bulk phase in
the interior of the interlayer space.
To study the dependence of water structures on state
variables, a model beidellite system, with a 4 ×2×1 su-
percell, 178 water molecules, and six charges compen-
sated by Ca2+, was subjected to MD equilibrations at a
pressure of 1 atm and temperatures of 300, 350, 400, 450,
and 500 K; and at a temperature of 300 K and external
pressures of 1, 2, 3, 5, and 10 atm. In the first series,
d(001) spacings were found at 20.46, 20.66, 20.89, 21.20,
and 21.75 ˚
A, respectively (averages of batch averages of
50 steps taken during the last 5 ps of 30 ps runs). As the
temperature increased, the layer structure was destroyed
by the high kinetic energy. In contrast, in the second se-
ries (80 ps runs), changes in pressure had no effect on the
water layers, and also the d(001) spacing was practically
unaffected (at 20.46, 20.40, 20.47, 20.45, and 20.40 ˚
A,
respectively).
The ordering of water in interlayer spaces is an in-
teresting phenomenon [27–29]. Plots of the distances be-
tween water oxygen atoms and mineral surfaces (Fig. 9)
show that the water arranges not only in layers along the
c-axis, but also along the other axes. These trends are
apparent in single snapshots of equilibrated systems, as
well as in ensemble averages (Figs. 10 and 11). We think
that the extent of the ordering is perhaps exaggerated by
the calculations. Some ordering of the kind revealed by
Figs. 9–11 must be expected, [27–29], but the absence of
polarization functions in the current empirical force fields
may very well lead to a higher order than is realistic.
CONCLUSIONS
Molecules are more strongly adsorbed on dry sur-
faces than on hydrated ones. The siloxane cavity of clay
minerals plays a major role in neutral molecule adsorp-
tion as it does in the adsorption of ions. Ions as well
as polar functional groups (NO2in DNOC and C Cl
in ATZ) position themselves in a characteristically off-
center arrangement relative to the hexagonal siloxane
cavity.
In the presence of water, the organic compounds
can make contact with clay surfaces by single functional
Fig.10. Histogram constructed from 1000 time steps of MD simulations
of water equilibrated in a montmorillonite interlayer space. Frequencies
of occurrence are shown (vertical axis) of distances ( ˚
A, horizontal axis)
between water–oxygen atoms and the crystallographic ab plane.

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
184 Yu, Newton, Norman, Sch¨afer, and Miller
Fig.11. Histogram constructed from 1000 time steps of MD simulations
of water equilibrated in a montmorillonite interlayer space. Frequencies
of occurrence are shown (vertical axis) of distances ( ˚
A, horizontal axis)
between water–oxygen atoms and the crystallographic bc plane.
groups and can then be expected to move rapidly through
soils. When adsorbed by full molecular contact, copla-
nar with the basal surface, the adsorbates are essentially
immobilized. From that position desorption by water can
occur, with different rates, depending on surface type and
type of compound adsorbed.
The current work was based entirely on empirical
computational techniques. In the future, quantum molec-
ular dynamics simulations, using techniques such as those
implemented by program CASTEP [63] can be expected
to be of increasing importance.
ACKNOWLEDGMENTS
The authors gratefully acknowledge support by
USDA CSREES grant 99-35107-7782 and by the IBM
Shared University Research Program. Special thanks are
due to Prof. Collis Geren. Vice Chancellor for Research,
University of Arkansas, and Dr. Jamie Coffin, IBM.
REFERENCES
1. Gay-Lussac, J. L. Mem. Soc. d Arcueil 1808, 2, 207.
2. Saalfeld, H.; Wedde, M. Z. Kristallogr. 1974, 139, 129–135.
3. Bish, D. L. Clays Clay Miner. 1993, 41, 738–744.
4. Lee, J. H.; Guggenheim, S. Amer. Miner. 1981, 66, 350–357.
5. Teppen, B. J. et al. J. Phys. Chem. 1997, 101, 1579–158.
6. MSI, Discover User Guide (1997) MSI Corp. (now Accelrys), San
Diego, California.
7. Maple, J. R. et al. J. Comp. Chem. 1994, 15, 162–182.
8. Hill, J.-R.; Sauer, J. J. Phys. Chem. 1994, 98, 1238–1244.
9. Hill, J.-R.; Sauer, J. J. Phys. Chem. 1995, 99, 9536–9550.
10. Allen,M.P.;Tildesley, D. J. Computer Simulation of Liquids; Oxford
University Press, Oxford, 1987.
11. Sanders, M. J.; Leslie, M.; Catlow, C. R. A. J. Chem. Soc. Chem.
Commun. 1984, pp. 1271–1273.
12. Collins, D. R.; Catlow, C. R. A. Molec. Simul. 1990, 4, 341–346.
13. Gale, J. D.; Catlow, C. R. A.; Mackrodt, W. C. Modeling Simul.
Mater. Sci. Eng. 1992, 1, 73–81.
14. Purton, J. et al. Phys. Chem. Miner. 1993, 19, 392–400.
15. Bush, T. S. et al. J. Mater. Chem. 1992, 2, 309–316.
16. Bush, T. S. et al. J. Mater. Chem. 1994, 4, 831–837.
17. Battle, P. D.; Bush, T. S.; Catlow, C. R. A. J. Amer. Chem. Soc. 1995,
117, 6292–6296.
18. Lewis, D. W. et al. J. Phys. Chem. 1995, 99, 2377–2383.
19. Collins, D. R.; Catlow, C. R. A. Acta Crystall. 1991, B 47, 678–682.
20. Collins, D. R.; Catlow, C. R. A. Amer. Miner. 1992, 77, 1172–1181.
21. Collins, D. R. et al. Phys. Chem. Miner. 1993, 19, 520–527.
22. Breu, J.; Catlow, C. R. A. Inorg. Chem.,1995, 34, 4504–4510.
23. Rustad,J. R.; Hay,B.P.Geochim. Cosmochim. Acta 1995,59,1251–
1257.
24. Rustad, J. R.; Hay, B. P.; Halley, J. W. J. Chem. Phys. 1995, 102,
427–431.
25. Rustad, J. R.; Felmy, A. R.; Hay, B. P. Geochim. Cosmochim. Acta
1996, 60, 1553–1562.
26. Rustad, J. R.; Felmy, A. R.; Hay, B. P. Geochim. Cosmochim. Acta
1996, 60, 1563–1576.
27. Delville, A. Langmuir 1991, 7, 547–555.
28. Delville, A. Langmuir,1992, 8, 1796–1805.
29. Delville, A. J. Phys. Chem. 1993, 97, 9703–9712.
30. Delville, A.; Sokolowski, S. J. Phys. Chem. 1993, 97, 6261–6271.
31. Skipper, N. T. Mineralogical Mag. 1998, 62, 657–667.
32. Skipper,N.T.;Refson,K.;McConnell, J. D. C.J.Chem.Phys. 1991,
94, 7434–7445.
33. Skipper, N. T.; Chang, F.-R. C.; Sposito, G. Clays Clay Miner. 1995,
43, 285–293.
34. Skipper, N. T.; Sposito, G.; Chang, F.-R. C. Clays Clay Miner. 1995,
43, 294–303.
35. Chang, F.-R. C. et al. In Mineral-Water Interfacial Reactions: Ki-
netics and Mechanisms, (Sparks, D. L. and Grunell, T. J., American
Chemical Society, Washington, D.C., 1999) pp. 88–106.
36. Chang, F.-R. C.; Skipper, N. T.; Sposito, G. Langmuir 1995, 11,
2734–2741.
37. Chang, F.-R. C.; Skipper, N. T.; Sposito, G. Langmuir,1997, 13,
2074–2082.
38. Boek, E. S.; Coveney, P. V.; Skipper, N. T. Langmuir 1995, 11,
4629–4631.
39. Boek, E. S.; Coveney, P. V.; Skipper, N. T. J. Amer. Chem. Soc. 1995,
117, 12608–12617.
40. Karaborni, S. et al. Science 1996, 271, 1102–1104.
41. Greathouse, J.; Sposito, G. J. Phys. Chem. B 1998, 102, 2406–2414.
42. Cygan, R. T. In Molecular Modeling of Clays and Mineral Surfaces;
Kubicki, J. D.; Bleam, W. F. Ed. Clay Minerals Society: Boulder,
CO, 1998, pp. 127–176.
43. Cygan, R. T.; Nagy, K. L.; Brady, P. V. In Adsorption of Metals by
Geomedia;Jenne,E. A., Ed.; Academic Press,SanDiego,CA,1998;
pp. 383–399.
44. Smith, D. E. Langmuir 1998, 14, 5959–5967.
45. Capkov´a, P. et al. Clays Clay Miner. 1998, 46, 232–239.
46. Capkov´a, P. et al. Clays Clay Miner. 1998, 46, 240–244.
47. Capkov´a, P. et al. J. Molec. Model. 1997, 3, 467–472.
48. Janeba, D.; Capkov´a, P.; Weiss, Z. J. Molec. Model. 1998, 4, 176–
182.
49. Sato, H.; Yamagishi, A.; Kato, S. J. Phys. Chem. 1992, 96, 9377–
9382.
50. Sato, H.; Yamagishi, A.; Kato, S. J. Phys. Chem. 1992, 96, 9382–
9387.
51. Sato, H. et al. J. Phys. Chem. 1996, 100, 1711–1717.
52. Breu, J.; Raj, N.; Catlow, C. R. A. J. Chem. Soc. Dalton 1999,
pp. 835–845.
53. Park,S.;Fitch,A.;Wang,Y.J.Phys.Chem.B1997,101,4889–4896.
54. Skipper,N.T.InMolecularModelingofClays and Mineral Surfaces.
Kubicki, J. D.; Bleam, W. F. Eds.; Clay Minerals Society, Boulder,
Colorado, 1998, pp. 90–126.
55. Keldsen, G. L. et al. J. Phys. Chem. 1994, 98, 279–284.

P1: GXB
Structural Chemistry (STUC) pp709-stuc-456242 January 14, 2003 17:16 Style file version Nov. 07, 2000
Molecular Dynamics Simulations of Adsorption of Organic Compounds 185
56. Cygan, R. T.; Kubicky, J. D. Eds., In Reviews in Mineralogy and
Geochemistry, Vol. 42; Geochemical Society, Rosso, J. J., Series
Ed.; Mineralogical Society of America: Ribbe, P. H., Series Editor;
Washington, 2001.
57. Teppen, B. J. et al. J. Comp. Chem. 1998, 19, 144–153.
58. Teppen, B. J. et al. Procedures of ECCOMAS, European Congress
on Computational Methods in Applied Sciences and Engineering,
Barcelona, 2000.
59. Yu, C.-H. et al. Clay Clay Miner. 2000, 48, 665–681.
60. Yu, C.-H. et al. J. Mol. Struct. 2000, 556, 95–103.
61. Dixon, J. B. In Minerals in Soil Environments, 2nd ed Dixon, J.
B.; Weed, S. B., Eds.; Soil Science Society of America: Madison,
Wisconsin, 1989, pp. 468–526.
62. Sheng, G. et al. J. Agr. Food Chem. 2001, 49, 2899–2907.
63. Payne, M. C. et al. Rev. Mod. Phys.1992, 64, 1045–
1097.