You are on page 1of 405

A Molecular Modeling Exploration of Smectite Interlayers as Adsorption Sites for

Inorganic and Organic Molecules


by
Rebecca Ann Sutton
B.S. (University of California, Davis) 1997
A dissertation submitted in partial satisfaction of the
requirements for the degree of
Doctor of Philosophy
in
Environmental Science, Policy and Management
in the
GRADUATE DIVISION
of the
UNIVERSITY OF CALIFORNIA, BERKELEY
Committee in charge:
Professor Garrison Sposito, Chair
Professor Harvey E. Doner
Professor David L. Sedlak
Dr. Margaret S. Tom
Spring 2004
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3147022
Copyright 2004 by
Sutton, Rebecca Ann
All rights reserved.
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy
submitted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.

UMI
UMI Microform 3147022
Copyright 2004 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A Molecular Modeling Exploration of Smectite Interlayers as Adsorption Sites for
Inorganic and Organic Molecules
Copyright 2004
by
Rebecca Ann Sutton
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract
A Molecular Modeling Exploration of Smectite Interlayers as Adsorption Sites for
Inorganic and Organic Molecules
by
Rebecca Ann Sutton
Doctor of Philosophy in Environmental Science, Policy and Management
University of California, Berkeley
, Professor Garrison Sposito, Chair
The abundance and reactivity of smectite minerals have prompted considerable
scientific and engineering interest in these negatively charged, shrinking and swelling,
layer type aluminosilicates. Molecular modeling techniques provide a uniquely
molecular scale view of the complex interactions that occur within a smectite interlayer
region between interlayer species and surrounding mineral surfaces. Atomistic
simulations describing two hydrated smectite interlayers illustrate the effects of mineral
structure and solution properties on adsorption mechanisms available to inorganic and
organic species.
Detailed understanding of Cs-smectites is essential to accurate prediction of clay
liner permeability to radiocesium cations at nuclear waste containment facilities. Monte
Carlo (MC) and molecular dynamics (MD) calculations performed on three Cs-smectites
suggested that Cs-saturated interlayers contained less than a monolayer of water
molecules, organized into partial hydration shells around the cations. Cs+ formed strong
inner sphere complexes with the mineral surface and displayed jump diffusional motion.
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Water molecules exhibited typical diffusional movements that accelerated with increased
water content. Animation of the MD trajectories of interlayer species revealed the
phenomenon of water sharing, when a single water molecule hydrates two cations
simultaneously for hundreds of picoseconds. Differences in smectite mineral structure,
including charge site distribution and hydroxyl group orientation, affected the prevalence
of water sharing interactions, as well as the location and mobility of Cs+.
To manage the landscape to sequester carbon and reduce the severity of climate
change, we must expand our knowledge of the organo-mineral interactions responsible
for long-term recalcitrance of humic substances. A model humic molecule, subjected to
energy minimization (EM) and MD calculations for different hydration and protonation
states, mimicked experimental properties of humic substances, including bulk density,
infrared spectrum, pseudomicellar reorientation with hydrated, and relative complexation
of Na+and Ca2+. When placed within a hydrated Ca-montmorillonite interlayer, the
deprotonated, Ca-saturated version of this model humic molecule adsorbed to the mineral
surface via numerous cation bridges, a few water bridges, and indirect H-bonding
interactions mediated by water molecules. The protonated version of this organic
molecule formed direct hydrophobic and H-bonding interactions with the mineral. Both
studies highlight the complex molecular scale surface chemistry associated with smectite
surfaces.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table of Contents
List of Figures vi
List of Tables xiv
Acknowledgments xvi
Chapter One - Introduction
1.1 Smectites as Adsorbents 1
1.2 Cs-smectite Systems 11
1.3 Organo-smectite Systems 25
1.4 Atomistic Simulations Exploring the Smectite Interlayer Region 49
1.4.1 Molecular Modeling Techniques 49
1.4.2 The Structure of Smectite Minerals 56
1.5 Hypotheses and Research Objectives 61
1.6 References 63
Chapter Two - Monte Carlo and molecular dynamics studies of Cs-smectite
hydrates
2.1 Introduction 78
2.2 Methods 91
2.2.1 Model Cs-smectite Hydrates 91
2.2.2 Monte Carlo Simulations 93
2.2.3 Molecular Dynamics Simulations 95
2.3 Results 97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.3.1 Monte Carlo Simulations 97
2.3.2 Molecular Dynamics Simulations 106
2.4 Discussion 113
2.4.1 Water Content of 12.4 A Cs-smectites 113
2.4.2 Molecular Structure of the Interlayer 115
2.4.3 Molecular Dynamics of the Interlayer 117
2.4.4 Comparison to NMR and EXAFS Studies of Interlayer 119
Speciation
2.5 Conclusions 122
2.6 Appendices 123
2.6.1 Appendix A: Conversion of LJ Cs-0 Parameters to MCY 123
Parameters
2.6.2 Appendix B: Addition of Atom Type and MCY Parameters to 124
MONTE
2.7 References 126
Chapter Three - Visualization of molecular dynamics simulations of Cs-smectite
hydrates
3.1 Introduction 133
3.2 Methods 142
3.2.1 Simulation of Cs-smectite Hydrates 142
3.2.2 Constructing Animations from MD Trajectory Data 145
3.3 Results and Discussion 146
3.4 Conclusions 161
3.5 Appendix: Sample File Formats for Trajectory and Lattice Coordinates 163
ii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.6 References 163
Chapter Four - Molecular simulation of a DOM molecule: Effects of hydration
and deprotonation
4.1 Introduction 169
4.1.1 Humic Substances as Supramolecular Associations Displaying 171
Pseudomicellar Structure
4.1.2 Nonpolar Functional Groups of Humic Substances 178
4.1.3 Oxygen Functional Groups of Humic Substances 182
4.1.4 Nitrogen Functional Groups of Humic Substances 190
4.1.5 Sulfur Functional Groups of Humic Substances 197
4.1.6 Humic Molecular Fragments 201
4.1.7 Molecular Models of Humic Substances 209
4.2 Methods 224
4.2.1 The COMPASS Force Field 224
4.2.2 Simulation of Bulk Water 228
4.2.3 Simulation of the Schulten DOM Molecule 232
4.2.4 Simulation of the Hydrated DOM Molecule 236
4.2.5 Simulation of the Deprotonated, Na-saturated DOM Hydrate 238
4.2.6 Simulation of the Deprotonated, Ca-saturated DOM Hydrate 241
4.3 Results and Discussion 243
4.3.1 The Bulk Water System 243
4.3.2 Energetic and Structural Properties of the DOM Molecule 250
4.3.3 Energetic and Structural Properties of the Hydrated DOM 255
Molecule
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3.4 Energetic and Structural Properties of the Deprotonated, Na- and 261
Ca-saturated DOM Complexes
4.3.5 Model Infrared Spectra of Protonated and Na- and Ca-saturated 275
DOM Hydrates
4.4 Conclusions 281
4.5 References 283
Chapter Five - Molecular modeling of a DOM-montmorillonite system
5.1 Introduction 300
5.2 Methods 306
5.2.1 Dry Ca-montmorillonite System 306
5.2.2 Hydrated Ca-montmorillonite System 308
5.2.3 The DOM-montmorillonite Systems 310
5.3 Results 315
5.3.1 Simulations of Ca-montmorillonite 315
5.3.2 The DOM-montmorillonite Systems 324
5.4 Discussion 340
5.4.1 Ca-montmorillonite and the COMPASS Force Field 340
5.4.2 Hydrophobic, Hydrogen Bonding, and Water Bridging Organo- 345
mineral Interactions within Protonated DOM-montmorillonite
5.4.3 Indirect Hydrogen Bonding and Cation Bridging Organo- 353
mineral Interactions within Ca-saturated DOM-montmorillonite
5.5 Conclusions 357
5.6 References 359
iv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter Six - Conclusion
6.1 Molecular Modeling as a Means of Examining Complex Systems
6.2 References
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Figures
Chapter One
1.1 The structure of montmorillonite, a typical smectite mineral. A cross- 2
section of the clay, a) illustrates the composition of each layer as two sheets
of Si tetrahedra (grey spheres are Si, red spheres are O) surrounding a sheet
of A1 octahedra (blue-green spheres are Al). Structural hydroxyl groups
(white spheres are H) appear at octahedral vertices not attached to
tetrahedra. The Si tetrahedra are tilted slightly, creating corrugated
surfaces. The space between the clay layers, or the interlayer region,
features a cation (blue) that balances the charge caused by isomorphous
substitution of Mg (dark blue) for Al at an octahedral site. Water
molecules hydrate the interlayer cation and fill the interlayer space. The
layer spacing of this montmorillonite is 10.3 A, calculated as the distance
between octahedral sheets. A closer, top-down view of a cluster of Si
tetrahedra making up the mineral surface, in which sphere size has been
adjusted to reflect the relative size of Si and O atoms, b) shows surface
features including oxygen triads and the distorted hexagonal (ditrigonal)
cavity formed by the arrangement of these triads. Mineral and interlayer
cation coordinates for this illustration are taken from quantum chemical
calculations performed by S.-H. Park (personal communication, 2003).
1.2 Diagram of the global C cycle, after Figures 3.1 in IPCC (2001). The sizes 26
of all major C pools are listed in parentheses in units of Pg C, using values
presented by Schnitzer (1997), with additions from Australian Greenhouse
Office (2002), and an updated atmospheric pool value from Keeling and
Whorf (2004). Thick lines indicate the largest and therefore most
important fluxes for the contemporary atmosphere, and purple text
describes anthropogenic fluxes. The abbreviation DOM near the center of
the figure stands for dissolved organic matter.
1.3 Typical structure of a rigid, idealized montmorillonite simulation cell 57
suitable for molecular modeling. Colors encode elemental identity as
described in Figure 1.1, with the exception that the interlayer cations shown
here are large brown spheres. This 8 crystalline unit simulation cell is
outlined in grey in the cross-sectional view provided in a), and features a
layer spacing of 15.2 A. Note the uniform, rigid structural hydroxyl
orientations in a), as well as the flat mineral surfaces in a) and the perfectly
hexagonal six-oxygen cavities in the top-down view of the mineral surface
seen in b). Compare this structure to the more realistic montmorillonite
structure seen in Figure 1.1.
1.4 A top-down view of clusters of alumina (and magnesia) octahedra within 59
the octahedral layers of cfr-vacant (left) and trans-vacant (right)
dioctahedral smectites, using a polyhedral display style to emphasize
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
geometry, after Sainz-Diaz et al. (2001a). The letters C and T mark the
location of mineral hydroxyl groups within cis-vacant and
trans- vacant octahedral sheets. Colors are linked to atom identity as
described in Figure 1.1, and polyhedra adopt the color of the atom at the
center of the shape (the blue-green polyhedra are alumina octahedra, while
the blue polyhedron represents a magnesium octahedron, a negative charge
site).
Chapter Two
2.1 Interlayer density profiles for water O. The z axis is normal to clay 100
surfaces, and the profiles are plotted with the interlayer midplane as the
origin.
2.2 Interlayer distribution of the angle (0) measured clockwise between the 100
dipole moment vector of a water molecule and a line normal to the siloxane
surface nearest to the molecule (Allen and Tildesley, 1987; Chang et al.,
1997). This distribution [P (0)] is not normalized to the solid angle through
division by sin 0.
2.3 Radial distribution functions for interlayer water 0 - 0 correlations. The 102
RDF for MCY liquid water is shown in blue.
2.4 Radial distribution functions for interlayer water O-H correlations. The 102
RDF for MCY liquid water is shown in blue.
2.5 Interlayer density profiles for Cs+. The z axis is normal to clay surfaces, 104
and the profiles are plotted with the interlayer midplane as the origin.
2.6 Radial distribution functions for Cs-water (Cs-O, Cs-H) and Cs-clay 105
(Cs-O*, Cs-H*) for the three smectites with 1/3 water monolayer.
2.7 Radial distribution functions for Cs-water (Cs-O, Cs-H) and Cs-clay 105
(Cs-O*, Cs-H*) for the two smectites with 2/3 water monolayer.
2.8 Interlayer water MSD vs. elapsed time for Cs-montmorillonite with 1/3 108
water monolayer. A 400 ps elapsed time plot required 800 ps of data.
2.9 Trajectory plots (xy plane, 400 ps) for Cs-hectorite with (a) 1/3 water 109
monolayer and (b) 2/3 water monolayer. Motions of Cs+ are shown in grey,
while blue depicts the trajectory of a representative water molecule. Light
grey dots correspond to surface O of one clay layer. The gridlines are
spaced 2 A apart. If the trajectory of a molecule exceeds the boundary of
the simulation cell, it reappears on the opposite side of the cell from
whence it came (seen in b), an artifact of periodic boundary conditions.
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.10 Trajectory plots (xy plane, 400 ps) for (a) Cs-beidellite with 1/3 water 111
monolayer and (b) Cs-montmorillonite with 1/3 water monolayer. Same
color scheme as in Figure 2.9.
2.11 The be coordinates of Cs+(400 ps MD simulation) in smectites with 1/3 112
water monolayer. Surrounding clay layers are located at 1.52, 1.44 A,
and 1.54 A along the z-axis for hectorite, beidellite, and montmorillonite.
Beside plots are images describing average locations of Cs+in relation to
clay layers. Orange spheres represent Cs+, purple spheres O, white spheres
H, grey spheres Si, and green spheres Al.
2.12 Molecular dynamics x-y trajectory plots (400 ps) for Cs-montmorillonite 113
with 2/3 water monolayer. Movements of four less mobile Cs+ are shown
in grey, while movements of two more mobile Cs+ are shown in green and
brown, respectively. Black dots represent surface oxygens of one clay
layer, and the gridlines are spaced 2 A apart. If the trajectory of a molecule
exceeds the boundary of the simulation cell, it reappears on the opposite
side of the cell from whence it came, an artifact of three-dimensional
periodic boundary conditions.
2.13 Monte Carlo snapshot of two Cs+(orange) sandwiched between beidellite 118
layers enclosing 1/3 water monolayer. The purple spheres represent
mineral O, the grey spheres represent mineral Si, and the green spheres
represent tetrahedral charge substitution sites containing Al instead of Si.
Water O are blue and water H are white. The grey lines connect to O
within the coordination sphere of Cs+, which for Cs-Ociay has a radius of 3.2
A, and for Cs-Owater a radius of 4.7 A.
Chapter Three
3.1 A view of the simulation cell (outlined in grey) for Cs-montmorillonite 135
with 1/3 water monolayer. Interlayer Cs+ are represented by orange
spheres, while interlayer water O are represented by blue spheres and water
H by white spheres. The surrounding mineral layers are displayed in ball
and stick style, with the greyish purple spheres representing the mineral O
atoms, white spheres representing H, grey spheres representing Si, and
green spheres representing Al.
3.2 Snapshots taken from an animation of a MD simulation of Cs-hectorite 148
with 1/3 water monolayer. Interlayer Cs+are represented by orange
spheres, while interlayer water molecules are represented by blue spheres.
The greyish purple spheres symbolize the surface oxygen atoms of the
lower clay layer surrounding this interlayer region. The rest of the atoms
making up this lower clay layer, as well as all the atoms making up the
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
upper clay layer, have been removed to improve the visibility of the
interlayer region. Due to the periodic nature of the simulations, interlayer
species which exit from one side of the simulation cell reappear on the
opposite side. In this series of images, the highlighted Cs+ (red) displays
jump diffusion as it hops between two sites above neighboring surface
oxygen triads. The highlighted water molecule (lighter blue) positioned in
the center of the simulation cell in frame a) displays continuous diffusional
behavior as it moves from one Cs+ partial hydration shell to another.
Another highlighted water molecule positioned to the left of the first in
frame a) can be seen while visiting a surface cavity site for a short time in
frame c).
3.3 A view in the xy plane of the two Cs+ - two water complex seen in an 149
animation of Cs-montmorillonite with 1/3 water monolayer.
3.4 Snapshots taken from the animation of Cs-montmorillonite with 1/3 water 151
monolayer. Interlayer species are represented as described in Figure 3.2.
The two highlighted Cs+(red) share the two highlighted water molecules
(lighter blue). The molecules making up this two Cs+- two water complex
are so strongly bound that they are observed to rotate as a unit during the
animation. Interlayer water molecules did not enter six-oxygen cavities in
this portion of the simulation.
3.5 Snapshots taken from the animation of Cs-beidellite with 1/3 water 152
monolayer. Interlayer species are represented as described in Figure 3.2.
The three highlighted water molecules (lighter blue) reside within six-
oxygen cavities of the clay surface for extended time periods.
3.6 Representations of preferred Cs+positions above six-oxygen cavities for 154
a) hectorite, b) beidellite, and c) montmorillonite. Interlayer species are
represented as in Figure 3.1.
3.7 A series of snapshots taken from the Cs-montmorillonite, 2/3 water 157
monolayer MD simulation animation. Interlayer species are represented as
described in Figure 3.2. The four Cs+toward the right of the animation are
bound together by extended water sharing arrangements into a structure
that is stable over the 800 ps timespan of the simulation. The two Cs+
toward the left of the simulation cell (red and brown) display continuous
diffusional motion and participate in an exchange reaction, with first one
and then the other taking up a position near an octahedral charge site in the
clay layer. In image a), the cation near the charge site is highlighted with a
pale color. When this cation moves away from the charge site in image b),
the pale highlight color is replaced by red. In image c), the red cation has
moved beyond the boundary of one side of the simulation cell, and has
reappeared on the opposite side, closer to the viewer, as a result of the
periodic boundary conditions imposed on the calculations. Meanwhile, the
ix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
brown cation has begun to move toward the charge site in the upper left
comer of these images. The arrival of the brown Cs+at this charge site in
image d) is official when its color pales.
3.8 Side view of a water molecule nestling into a six-oxygen cavity in the Cs- 158
beidellite system with 1/3 water monolayer.
3.9 Snapshots of the Cs-hectorite, 2/3 water monolayer MD simulation 161
animation. Interlayer species are represented as described in Figure 3.2.
The highlighted Cs+(red) can be seen to compete with surrounding water
molecules for favorable positions over a six-oxygen cavity, when it moves
forward to occupy a surface cavity site closer to the viewer, then returns to
the oxygen triad site it prefers. A water molecule can be seen hovering
over the cavity in question both before and after its occupation by Cs+.
Chapter Four
4.1 Image of a model humic molecule similar to that described by Schulten 221
(1999a), including associated trisaccharide, polypeptide, and water
molecules. Grey spheres represent C, red spheres represent O, white
spheres represent H, and N and S are blue and yellow, respectively.
4.2 Radial distribution functions for O-O, 0-H, and H-H atom pairs of 247
COMPASS bulk water at T = 300 K (MD simulation).
4.3 Model IR spectrum of COMPASS bulk water system. 250
4.4 Model IR spectrum of dry DOM molecule. 253
4.5 Histograms of organic C, O, and H distances relative to the geometric 258
center of the DOM molecule in dry (Section 4.3.2) and hydrated
configurations. Values along the x axes represent the upper boundaries of
each 1 A distance bin.
4.6 Dry (left) and hydrated (right) DOM molecules (waters removed from 260
views) in the XY (top), YZ (middle), and ZX (bottom) planes. Grey
represents C, red represents O, green represents H (for visibility), and N
and S are blue and yellow, respectively. In the lower views, light blue
highlights hydrophobic regions occupying more protected positions with
hydration.
4.7 Views of Na- (left) and Ca-DOM (right) molecules (water removed from 266
images) as seen in Figure 4.6. Brown symbols represent cation positions.
4.8 Radial distribution functions between the cation and total O (grey), 269
x
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
carboxylate O (black), and water O (blue) for Na- (top) and Ca-DOM
(bottom) complexes.
4.9 A portion of the Na-DOM structure featuring an outer sphere Na- 273
carboxylate complex (left), a doubly inner sphere Na-carboxylate complex
(top), and a singly inner sphere Na-carboxylate complex, which also
happens to possess inner sphere coordination with an alcohol group (right).
4.10 Model IR spectra for the dry and hydrated DOM molecules, as well as the 276
Na-DOM (top) and Ca-DOM (bottom) systems, normalized using the
intensity of the nitrile peak.
4.11 Model IR spectra for the Ca-DOM complex in dry and partially hydrated 279
states (no normalization).
Chapter Five
5.1 The dry Ca-montmorillonite system with layer spacing 10.26 A, considered 319
to occupy the global potential energy minimum. Ca2+are represented by
large brown spheres. Red spheres represent O, white spheres represent H,
grey spheres represent Si, blue-green spheres represent Al, and blue spheres
represent Mg. The Ca2+ to the right and left of the cell are adsorbed to
oxygen triads near tetrahedral charge sites, while the Ca2+ in the center is
adsorbed between six-oxygen cavities near an octahedral charge site.
5.2 Average distribution of water O along the z axis, normal to the clay layer, 320
for the six hydrated Ca-montmorillonite systems (layer spacing 15.2 A)
subjected to an EM - annealing MD - EM - MD (T = 300 K) - EM cycle.
The origin represents the midplane of the interlayer.
5.3 Ca-montmorillonite hydrate system with lowest potential energy (layer 322
spacing 15.2 A). Elements are color-coded as described in Figure 5.1.
5.4 Model IR spectra of dry and hydrated Ca-montmorillonite (top, no 324
normalization), as well as bulk water (bottom, Section 4.3.1).
5.5 Views of protonated (left) and Ca-saturated (right) DOM molecules within 329
interlayers of hydrated Ca-montmorillonite (waters removed from images)
in the XZ (top), YZ (middle), and XY (bottom) planes. Grey denotes C,
red denotes O, green denotes H (for visibility), N and S are blue and
yellow, respectively, and Ca2+is brown. The clay layer is color-coded as
described in Figure 5.1. The clay layer and associated Ca2+ were removed
from the XY plane view to improve visibility of organic structures.
5.6 The illustration to the left is an example of a direct organo-mineral H-bond 332
xi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
taken from the protonated system ( 0 - 0 distance 2.58 A, O-H-O angle
149.3), while the illustration to the right is an example of an indirect
organo-mineral H-bonded structure taken from the Ca-saturated system
( 0 - 0 distances 2.53 A and 2.68 A, O-H-O angles 161.4 and 157.3 for
upper and lower H-bonds, respectively). Atoms are color-coded as
described in Figure 5.5, except that H is now white instead of green.
5.7 A portion of the protonated DOM-montmorillonite system displaying a 334
typical hydrophobic organo-mineral interaction as well as the attraction of
an organic O to a six-oxygen cavity. Atoms are color-coded as described in
Figure 5.6.
5.8 Ca-ORDFs for the protonated DOM-montmorillonite system. 335
5.9 Ca-0 RDFs for the Ca-saturated DOM-montmorillonite system. 336
5.10 A water bridge within the protonated DOM-montmorillonite system (left), 337
featuring outer sphere complexation between Ca2+and a carboxyl O atom
marked by a black line, and inner sphere coordination with the mineral
surface, and a cation bridge within the Ca-saturated DOM-montmorillonite
system (right), featuring inner sphere complexation between Ca2+ and
carboxylate groups marked by black lines, and outer sphere coordination to
the mineral surface. Atoms are color-coded as described in
Figure 5.6.
5.11 Model IR spectra of the protonated, periodic DOM molecule extracted 338
from the interlayer of a Ca-montmorillonite hydrate (top), as well as the
dry, densely packed DOM molecule (Section 4.3.2), and the DOM hydrate
after water molecules were removed from the structure (Section 4.3.5)
(bottom). The absorbance values for all spectra were normalized via the
absorbance of the nitrile peak of the dry, densely packed DOM molecule.
5.12 Model IR spectra of the periodic Ca-saturated DOM molecule extracted 339
from the interlayer of a Ca-montmorillonite hydrate (top), as well as the
dry, densely packed DOM molecule (Section 4.3.2), and the Ca-DOM
hydrate after water molecules were removed from the structure (Section
4.3.5) (bottom). The absorbance values for all spectra were normalized via
the absorbance of the nitrile peak of the dry, densely packed DOM
molecule.
5.13 Model IR spectra of the periodic Ca-saturated DOM molecule extracted 341
from the interlayer of a Ca-montmorillonite hydrate and including water
molecules associated with Ca2+hydrates (top), as well as the Ca-DOM
complex including water molecules associated with Ca2+ hydrates (Section
4.3.5), and bulk water (Section 4.3.1) (bottom). These spectra are not
normalized.
xii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.14 Views of the protonated (top) and Ca-saturated (bottom) DOM-
montmorillonite systems. Water molecules are represented using cylinders
or thin sticks to reduce visual clutter, while DOM and clay molecules are
represented in the ball and stick style. The O atoms of a few water
molecules near mineral surfaces are directed toward surface cavities.
Atoms are color-coded by element as described in Figure 5.5, except that H
is now white instead of green.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Tables
Chapter One
1.1 Surface Chemistry of Organo-mineral Interactions
Chapter Two
2.1 Layer Spacing and Molar Potential Energy of Water in Cs-smectite
Hydrates (MC Simulations)
2.2 Coordination Numbers (CN) for Interlayer Water Oxygens and
Surrounding Water Oxygens and Hydrogens
2.3 Coordination Number (CN) for Nearest-Neighbor Oxygens around
Interlayer Cesium
2.4 Self-Diffusion Coefficients (Dw) for Interlayer Water in Cs-smectite
Hydrates
Chapter Four
4.1 Structural and Dynamic Behavior of COMPASS Water (MD, T = 300 K)
as Compared to Well-Established Water Models
4.2 Energy Distribution of the COMPASS Bulk Water System
4.3 Energy Distribution of the Minimized, Periodic COMPASS DOM
Structure
4.4 Energy Distributions of the DOM Hydrate, the Isolated DOM Molecule,
and its Water Shell
4.5 Energy Distributions of Cation Hydrates
4.6 Energy Distributions of Na-DOM and Ca-DOM with and without
Associated Water Molecules, and of Water Molecules without Associated
Organic Complexes
4.7 Water Molecule Characteristics in Hydrated DOM Systems
4.8 Cation Coordination Numbers for Na- and Ca-DOM Complexes
4-5
98
103
106
107
245
246
251
256
262
263
268
270
xiv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter Five
5.1 Energy Distributions for Dry Ca-montmorillonite System 319
5.2 Energy Distribution of Ca-montmorillonite Hydrate 323
5.3 The Potential Energy Distributions of the Protonated DOM- 327
montmorillonite Interlayer Complex, its Individual Components, and their
Combinations
5.4 The Potential Energy Distributions of the Ca-saturated DOM- 328
montmorillonite Interlayer Complex, its Individual Components, and their
Combinations
5.5 Hydrogen Bonding within the DOM-montmorillonite Interlayer Complexes 331
5.6 Cation Coordination Numbers for Protonated and Ca-saturated Systems, 334
Using a Maximum Radius of 2.75 A
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgments
First, I must thank Gary Sposito for initiating me into the world of research
science. My experiences within his laboratory have taught me about science and about
myself. As part of my dissertation committee, Harvey Doner, David Sedlak, and
Margaret Tom have provided constant support for my research and helpful comments on
the document that follows, despite its ridiculous length. As well, I owe my introduction
to geochemistry to friend and mentor Bill Casey.
My work on Cs-smectite systems would not have been possible without
considerable help from scientist-programmers Neal Skipper and Keith Refson, many
consultations with Jeff Greathouse and Sung-Ho Park, and useful discussions with David
Smith. Grants of supercomputer time from the National Energy Resources Scientific
Computing (NERSC) Center were essential for this research. Wes Bethel of NERSCs
Visualization Group provided invaluable visualization and programming support for the
Cs-smectite animations, as well as continued friendship. This research was supported in
part by the Director, Office of Energy Research, Office of Basic Energy Sciences,
Geosciences Division, of the U.S. Department of Energy under Contract No. DE-AC03-
76SF00098, with an additional contribution from the Clay Minerals Society.
My work on organo-smectite systems would not have been possible without
Hans-Rolf Schulten, creator of the DOM molecule, and Mamadou Diallo, architect of the
simulation strategy I employed. Paul Kung and Amitesh Maiti of Accelrys Inc. provided
substantial technical support for this project. Three supercomputer facilities, and their
hardworking sysadmins, deserve special thanks - Kathy Durkin and UC Berkeleys
xvi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
College of Chemistry Molecular Graphics Facility, Dodi Heryadi and University of
Illinois at Urbana-Champaigns National Center for Supercomputing Applications
(NCSA), and Agnieszka Szymanska-Kwiecien and Polands Wroclaw Centre for
Networking and Supercomputing (WCSS). I am also grateful to Jim Kubicki, Cliff
Johnston, David Bish, Yuji Arai, and an informal network of Californian soil carbon
scientists for helpful discussions on molecular modeling or experimental techniques, as
well as to Xiangyun Song for thermal analysis training and time. This research was
supported in part by the Keamey Foundation of Soil Science, with an additional
contribution from the Clay Minerals Society.
Finally, I must thank my friends and family for their support and understanding
over the last several years. I owe special thanks to Josh Mehlman, my biggest fan, and
creator of the atomicDistance program, and to Claire Boudreaux, my best friend for 25
years, who helped me with many of my figures. Finally, I d like to acknowledge many of
my Sposito labmates, in no particular order, for creating such a supportive working
environment: Brandy Toner, Stacey Cooper, Wei-Cheng Lo, Ian Bourg, Javiera Cervini-
Silva, Mario Villalobos, Mary OConnor, Stephan Kraemer, Terry Burton, Jasquelin
Pena, Owen Duckworth, Tanya Peretyazhko, Caterina Tommaseo, Zach Struyk, Ludmilla
Aristilde, Gordon Vrdoljak, and Andy Yang.
xvii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter One
Introduction
1.1 Smectites as Adsorbents
Widely distributed within soils of temperate regions, smectite clay minerals have
long fascinated chemists and soil scientists. The extraordinary capacity of these tiny
minerals, in small amounts, to effect dramatic changes in both the structure and the water
and nutrient retention of soils, as well as their remarkable ability, when present in larger
concentrations, to topple man-made constructions through the action of shrinking and
swelling, has persuaded many scientists to investigate their chemical composition and
functional properties. Renowned chemist Linus Pauling (1930) determined the atomic
structure of mica, a mineral cousin of the smectites. Like mica, smectites are 2:1 layer
type clay minerals (Figure 1.1). Each layer consists of an alumina (dioctahedral) or
magnesia (trioctahedral) sheet, sandwiched between two silica tetrahedral sheets. The
arrangement of silica tetrahedra in sheets results in a pattern of oxygen triads across the
siloxane mineral surface that forms a network of six-oxygen rings, at the center of each a
distorted hexagonal (or ditrigonal) cavity.
Work by Hofmann et al. (1933), modified by Marshall (1935) and Hendricks
(1942), established the structure of smectites and elucidated their charge properties.
Smectite minerals are defined by the amount of structural charge they contain, an excess
electron charge of 0.5 - 1.2 moles per crystalline unit (Sposito, 1989), created by
isomorphic substitutions within the neutral unit formulae given below:
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[Sig](Al4)0 2 o(OH) 4 (dioctahedral)
[Si8](Mg6)02o(OH)4 (trioctahedral)
Isomorphic substitution of Mg for Al, or Li for Mg, results in negative octahedral charge
sites, while substitution of Al for Si creates negative tetrahedral charge sites. The
distribution of charge sites divides the smectite group into specific minerals. A
trioctahedral smectite with charge located primarily in the octahedral sheet is known as a
hectorite. A dioctahedral smectite with more octahedral charge than tetrahedral charge is
called a montmorillonite; with the reverse charge distribution, the mineral is classified as
a beidellite (Greathouse and Sposito, 1998).
Figure 1.1. The structure of montmorillonite, a
typical smectite mineral. A cross-section of the
clay, a) illustrates the composition of each layer as
two sheets of Si tetrahedra (grey spheres are Si, red
spheres are O) surrounding a sheet of Al octahedra
(blue-green spheres are Al). Structural hydroxyl
groups (white spheres are H) appear at octahedral
vertices not attached to tetrahedra. The Si tetrahedra
are tilted slightly, creating corrugated surfaces. The
space between the clay layers, or the interlayer
region, features a cation (blue) that balances the
charge caused by isomorphous substitution of Mg
(dark blue) for Al at an octahedral site. Water
molecules hydrate the interlayer cation and fill the
inter layer space. The layer spacing of this
montmorillonite is 10.3 A, calculated as the distance
between octahedral sheets. A closer, top-down view
of a cluster of Si tetrahedra making up the mineral
surface, in which sphere size has been adjusted to
reflect the relative size of Si and O atoms, b) shows
surface features including oxygen triads and the
distorted hexagonal (ditrigonal) cavity formed by the
arrangement of these triads. Mineral and interlayer
cation coordinates for this illustration are taken from
quantum chemical calculations performed by S.-H.
Park (personal communication, 2003).
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In order to compensate for this negative charge, positive counterions and
associated waters of hydration move into interlayer spaces between the smectite layers.
The cations attracted to smectites form a nutrient reservoir for plants and soil microbes,
measured as the cation exchange capacity (CEC) of the mineral. These counterions also
cause the interlayer region in which they reside to expand or contract, depending on the
amount of interlayer water they have drawn into the region. Smectite charge is therefore
responsible for both the nutrient retention and the shrinking and swelling properties of
these minerals, and provides access to the minerals enormous internal surface areas, up
to 8 x 105m2 kg'1(Borchardt, 1989), for a variety of adsorption interactions.
The expansive surface area of smectite minerals is by no means chemically
homogeneous. Broken mineral edges provide pH-dependent charge sites, typically
positive under soil pH conditions. And while the net charge of smectite minerals is
negative, recent studies reveal that the basal planes of smectites also contain hydrophobic
regions distant from isomorphic charge sites. These hydrophobic microsites can adsorb
nonpolar molecules through van der Waals interactions (Jaynes and Boyd, 1991a, 1991b;
Boyd and Jaynes, 1992; van Oss and Giese, 1995; Giese et al., 1996; Laird, 1996; Chiou
and Rutherford, 1997).
Due to their ubiquitous distribution and large, moderately charged surface area,
smectite clay minerals influence many biogeochemical cycles. The attraction of
inorganic cations to numerous structural charge sites can result in previously mentioned
cation retention in soils, or rapid transport through aquifers via the movement of
suspended smectitic colloids. Organic substances interact with smectite surfaces through
a broad range of organo-mineral interactions (Table 1.1). The complexes that form often
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
R
e
p
r
o
d
u
c
e
d

w
i
t
h

p
e
r
m
i
s
s
i
o
n

o
f

t
h
e

c
o
p
y
r
i
g
h
t

o
w
n
e
r
.

F
u
r
t
h
e
r

r
e
p
r
o
d
u
c
t
i
o
n

p
r
o
h
i
b
i
t
e
d

w
i
t
h
o
u
t

p
e
r
m
i
s
s
i
o
n
.
Table 1.1. Surface Chemistry of Organo-mineral Interactions3
Adsorption
Mechanisms Illustration
Some Relevant
Minerals
Principal Organic
Functional Groups
Hydrophobic
Interactions11
(also known as
physical bonding
or van der Waals
interactions)
Hydrogen
Bondingc
Si l
II O'
II
I
II
Any minerals with Nonpolar groups
neutral microsites, such as aromatic or
including smectite
and kaolinite
surfaces
alkyl C
Any minerals with
oxygen surfaces,
for example
kaolinite
Amine, carbonyl,
carboxyl, phenol,
alcohol,
heterocyclic N
Protonation
C l a y ] " - 1 1,0
(A.
; . c - r
Aluminosilicate
edge sites,
Fe and A1 oxides,
allophane and
imogolite
Amine, heterocyclic
N, carbonyl,
carboxylate
Ligand Exchange
o,
no'
Clay] 011+ _R -* Clay 1 ' ) R 11;0
Aluminosilicate
edge sites,
Fe and A1 oxides,
allophane and
imogolite
Carboxylate,
phenolate
A Few Significant
Soil Properties
Solution ionic
strength,
amount SOM
attached to
minerals
Water content
pH,
cations present,
water content
Structural metal
cation,
water content
Table 1.1, continued. Surface Chemistry of Organo-mineral Interactions3
Adsorption
Mechanisms Illustration
Some Relevant
Minerals
Principal Organic
Functional Groups
Ion Exchange
permanent
charge sites
(cation exchange)
pH-dependent
charge sites'1
(anion exchange
usually, cation
exchange rarely)
C o lloid ( o l i o i d
pa r t i c l e j
o f soil 1)1 soil
I I H, H ,
^llillyl ll ,n
ii n .. . 1,
0 : - O Al (3:,,: O ^ r r t A I ^ T i Q ; , ,
\ / / \ \ / / \ \ /
AI('!~{)'--V(> O J :
Smectite, illite,
vermiculite
Aluminosilicate
edge sites,
Fe and A1 oxides,
allophane and
imogolite
Amine,
heterocyclic N
Carboxylate for
anion exchange,
amine,
heterocyclic N for
cation exchange
Cation Bridging
Water Bridging
i g G , ;
' c o r !
H
OH- UM
il
: - o - h , |
O - l I O M"
Smectite, illite,
vermiculite
Smectite, illite,
vermiculite
ncgal i v
c h ar u e s
Carboxylate, amine,
carbonyl, alcohol,
phenol
Amine, carboxylate,
carbonyl, alcohol,
phenol
compiled from Sposito (1989) and Mortland (1986).
billustration adapted from Theng (1979).
illustration adapted from Thurman (1986).
dillustration adapted from Hayes and Himes (1986).
A Few Significant
Soil Properties
Cations present
pH,
solution ionic
strength
Cations present
Cations present,
water content
act to protect the organic molecules from microbial decomposition, dramatically
lengthening the lifespan of these carbon-rich substances in soils.
Abundant geological deposits of smectite minerals, or the primarily smectitic
mineral mixtures known as bentonites, have allowed humans to develop a number of
applications for their unique chemical properties. Smectite liners have been used for
many years to seal landfills and other waste containment sites, in order to isolate
hazardous materials (Koch, 2002). Of particular concern are radioactive and water-
soluble contaminant cations such as the unstable isotopes of Cs+, given their potential
mobility in many environments. Smectite minerals show great promise as liners for
radioactive waste containment facilities, spurring an international effort to investigate
their potential advantages and limitations in this capacity (Pushkareva et al., 2002).
Smectites and other soil minerals have also been singled out for another, more subtle
form of waste disposal, based on their ability to sequester organic materials in soils.
Land management has been suggested as a means of mitigating a portion of the
anthropogenic emissions of CO2 responsible for current climate change (Bruce et al.,
1999; Lai et al., 1999). Careful study of the mechanisms responsible for mineral
stabilization o f organic matter may allow us to harness the power of smectites to expand
the soil C sink and begin to alleviate the effects of global warming.
In order to design suitable uses for these versatile minerals, we must understand
their mechanisms of adsorption. Experimental work on the adsorptive properties of
smectites traditionally takes the form of either bulk chemical methods or spectroscopic
measurements. Here, bulk chemical methods may be defined as those designed to
measure the partitioning of adsorbates between supernatant solutions and mineral
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
surfaces. Adsorption and desorption isotherms, for example, are plots of the amount of a
particular adsorbate associated with a mineral surface relative to its concentration in bulk
solution, under fixed temperature and pressure conditions. Selectivity studies use the
same techniques to examine competition between two adsorbates for adsorption sites, by
monitoring the concentrations of each species associated with the mineral surface relative
to its proportion in the bulk solution. Diffusion experiments assess dynamic aspects of
the adsorption process, by monitoring the rates at which adsorbate species move through
clay mineral samples, rates that are reduced by adsorption reactions. These bulk
chemical methods provide limited insight regarding mechanisms on a molecular scale.
X-ray diffraction (XRD) data collected during the course of experiments
exploring the bulk chemistry of adsorption to smectites can provide information about the
structure of the clay interlayer, offering clues relevant to adsorption mechanisms. The
periodic atomic structure of clay minerals scatters x-rays into patterns that represent
distances between atoms and can be used to establish the average layer spacing present in
the sample (Moore and Reynolds, 1997). The hydrophilicity or hydrophobicity of
interlayer cations and other adsorbed interlayer species determines the amount of water
present in the interlayer at a particular relative humidity, which in turn determines the
layer spacing of the smectite documented by XRD. Thus, XRD data, in combination
with bulk chemical methods, can be used to suggest the presence or absence of
significant sites of adsorption within the interlayer region for the adsorbate species under
investigation (Theng, 1979).
A more direct look at molecular scale adsorption interactions is possible with
spectroscopy, or the study of the interaction of electromagnetic radiation with matter.
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Some of the most useful spectroscopic methods for examination of surface chemistry are
infrared (IR), nuclear magnetic resonance (NMR), and extended x-ray absorption fine
structure (EXAFS) spectroscopies. Infrared spectroscopy involves irradiation of a
sample with infrared light, typically using a range of wavelengths from 2.5 to 25 pm.
Sample molecules will absorb those frequencies of light that match the natural vibrational
frequencies of the fluctuating dipole moments specific to different molecular bonds. By
monitoring the intensity of the same infrared frequencies after they have passed through a
sample, and transforming this output into an IR spectrum, it is possible to quantify the
degree to which different wavelengths of light have been absorbed by the sample, and
therefore to assess the presence and relative quantity of various bonds or functional
groups within the sample. What makes IR spectroscopy especially useful in terms of
surface chemistry is that many adsorption reactions result in subtle changes to molecular
bond lengths and angles, as a result of electrostatic attraction, conformational
rearrangement, the formation of H-bonds or new chemical bonds, and other interactions
between adsorbate and mineral. These differences translate to changes in the infrared
frequencies associated with the bonds, or shifts in peak position and shape in the IR
spectrum. A detailed discussion of IR spectroscopy in relation to soil and surface
chemistry is provided by Piccolo and Conte (1998), while Johnston and Aochi (1996)
provide comprehensive information regarding the technique itself.
Nuclear magnetic resonance spectroscopy involves the exposure of a sample to a
strong magnetic field while exciting specific nuclei present in the sample using pulses of
radiofrequency radiation. Certain nuclei, for example 133Cs, 13C, or !H, feature magnetic
spins that can be excited to a high energy state through the adsorption of radiofrequency
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
photons under these conditions. After excitation, the spin states of these nuclei relax and
emit radiofrequency radiation. Nuclear spin states are sensitive to the surrounding
chemical environment because neighboring electrons respond to magnetic fields by
reducing their magnitude, or shielding the nucleus (Blearn, 1991). This chemical
shielding affects the frequency of the energy required to force the nuclei into higher
energy spin states and, therefore, the energy of the photons emitted when they relax.
Detection of the photons emitted by these nuclei after elimination of the radiofrequency
pulse, and subsequent mathematical processing, provides a NMR spectrum characterized
by signals indicating different levels of chemical shielding affecting the nuclei examined.
In terms of adsorption, the chemical shift information can be used to identify and
quantify relative amounts of functional groups present in the sample, and can provide
information concerning the local coordination environment experienced by the nuclei
(Bleam, 1991).
Most solid-state NMR spectra are collected under conditions of magic angle
spinning (MAS), in which the sample is spun at an angle of 54.74 relative to the static
magnetic field (Bleam, 1991). This spinning state increases the sensitivity of NMR
spectroscopy by reducing line broadening caused by chemical shift anisotropy, nuclear
dipole-dipole interactions, and quadrupolar coupling interactions, although it also
produces spinning side bands, satellite peaks that diminish the relative area of the central
peak. While a straightforward NMR experiment directly excites the nuclei of interest, as
described above, a cross polarization (CP) experiment excites nontarget nuclei, such as
1 13
H, which are in close proximity to less abundant nuclei of interest, for example C. A
specific combination of radiofrequency pulses transfers the spin energy of the nontarget
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
nuclei to target nuclei, enhancing NMR sensitivity for specific nuclei with low natural
abundance. Further information regarding the use of NMR spectroscopy to examine
adsorption interactions may be found in Earl and Johnston (1998), while a theoretical
description of NMR techniques, including CP/MAS methods, is provided by Bleam
(1991).
Extended x-ray absorption fine structure spectroscopy is one of a host of powerful
new synchrotron-based x-ray absorption spectroscopies. The EXAFS signal occurs as a
result of the absorption of an x-ray photon with energy sufficient to cause a particular
element to expel one of its electrons. The ejected photoelectron experiences coherent
scattering from neighboring atoms, and this interference modifies the x-ray frequencies
detected and mathematically manipulated to provide an EXAFS spectrum (Brown and
Sturchio, 2002). Data concerning the elemental identity and number of atoms
surrounding the EXAFS target atom can be extracted from the EXAFS spectrum,
information extremely relevant to determination of mechanisms of adsorption.
Additional discussion of the theory of EXAFS, as well as its use in explorations of
surface chemistry, is provided by Brown and Sturchio (2002).
All forms of spectroscopy share two general properties. First, spectroscopy
provides spatially integrated data, combining all signals provided by the sample. Spectra
cannot always distinguish the effects of molecular-scale heterogeneity, and instead must
be deconvoluted so that signals from many chemical species can be recognized
independently. Second, each spectroscopic technique probes systems over a specific time
scale associated with the type of energetic transition excited by the electromagnetic
radiation. Infrared spectroscopy probes a time scale in the range of 10' 15 to 10' 12 s, and is
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
therefore capable of observing the vibrational motions of molecular structures. Nuclear
magnetic resonance spectroscopic measurements occur in the range of 1 0 ' 9 to 1 0 ' 7 s,
corresponding to states averaged in terms of vibration, rotation, and even translation.
Low temperature NMR can lengthen rotational and translational time scales and diminish
these forms of averaging in the NMR spectrum (Weiss et al., 1990a). The time scale for
EXAFS measurements is in the range of 10' 20 to 10' 17 s, such that the technique probes
the instantaneous molecular structure of the sample. In combination with the bulk
chemical methods and XRD, described previously, these spectroscopies have provided
much insight regarding interlayer adsorption processes active in two important smectite
systems, those involving the inorganic counterion Cs+, and those housing complex
natural organic molecules known as humic substances.
1.2 Cs-smectite Systems
Byproducts of nuclear power production and nuclear weapons testing, Cs
radioisotopes are among the most abundant components of all radionuclides present in
radioactive waste (Bradbury and Baeyens, 2000), making the transport and fate of
unstable Cs isotopes an immediate concern. 134Cs and 137Cs have half-lives of 2.065 and
30.2 years, respectively, a time scale relevant to human life, while 135Cs represents a
long-term threat, with a half-life of 2.3 x 106 years (Winter, 2003). Cesium has no known
biological role, but, if ingested, it can replace the essential nutrient potassium in the body
due to the chemical similarity of the two alkali elements (Winter, 2003).
Large scale Cs radionuclide contamination has occurred as a result of atmospheric
fallout from both nuclear weapons testing and the explosion at the Chernobyl nuclear
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
power plant in 1986 (Avery, 1996). Once the radioactive debris from these incidents
reaches the earths surface, the mobility and bioavailability of Cs radionuclides is
determined by the properties of the Cs+ ion, which has unlimited solubility in aqueous
solutions, and participates in enduring electrostatic interactions with negatively charged
surfaces. Careful inventories reveal that much of the 137Cs reaching terrestrial
ecosystems is adsorbed and immobilized by soil minerals like smectites, but that
significant portions are incorporated into plants and microbes through substitution for
potassium, and bioaccumulate in animals further along the food chain (Avery, 1996; Zhu
and Smolders, 2000; Fesenko, 2001). The partitioning of Cs radioisotopes among these
pools depends on the mineral content (Hird et al., 1996; Avery, 1996), mineralogy
(Cremers et al., 1988; Shaban and Mikulaj, 1996; Paasikallio, 1999; Facchinelli et al.,
2001), pH (Dumat et al., 2000; Sanchez et al., 2002), abundance of other monovalent
cations (Avery, 1996; Zhu and Smolders, 2000), and the amount of natural organic matter
(Barber, 1964; Cremers et al., 1988; Dumat et al., 2000; Gri et al., 2000) present at each
site. With broad assumptions of locally uniform fallout distribution and rapid adsorption
of Cs+to soil particles, geomorphic processes of erosion and deposition can be measured
using the signature of this 137Cs fallout (He and Walling, 1997; Di Stefano et al., 1999;
Mabit et al., 1999; Quine, 1999; Ritchie and Rasmussen, 2000; Collins et al., 2001; Li
and Lindstrom, 2001; Vanden Bygaart and Protz, 2001; Wallbrink et al., 2002).
In addition to atmospheric fallout, Cs radionuclides enter the environment in more
localized emissions, through intentional releases of low-level aqueous radioactive wastes
that can include 137Cs and 134Cs at nuclear reactors or nuclear fuel-reprocessing plants
(Avery, 1996), or through leaks from sites storing nuclear materials, such as the Hanford
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Site in Washington (McKinley et al., 2001). Adsorption of Cs+ onto charged mineral
particles abundant in both aquatic and terrestrial systems soon follows, and transport via
association with suspended colloidal materials becomes a mechanism for rapid
dissemination into downstream environments. Kersting et al. (1999) found evidence of
colloidal transport of 137Cs by smectite and other clay mineral particles in groundwater
below the Nevada Test Site, the location of 828 underground nuclear tests from 1956 to
1992.
To prevent further Cs contamination, the use of smectitic liners at nuclear waste
containment facilities has been suggested as a cost-effective means of limiting the
movement of escaping radionuclides. Layers of bentonite, a mineral mixture that is
primarily smectite, are already widely used to line and cap municipal and industrial waste
sites (Koch, 2002). In a typical engineered scenario for high level nuclear waste
containment (Cho et al., 1993; Chun et al., 2001), compacted bentonite backfill material
is used to create an environmental buffer separating waste containers from surrounding
bedrock. This engineered attenuation strategy makes use of the adsorption and
sequestration of Cs+ by the interlayer region of smectite clay minerals (Evans et al., 1983;
Jensen and Radke, 1988; Cho et al., 1993; Oscarson et al., 1994; Iwasaki and Onodera,
1995; Onodera et al., 1998; Kozaki et al., 1999), and assumes a very low mobility of Cs+
in the interlayer region after adsorption.
An understanding of Cs-smectite systems is imperative for the development of
methods to predict the mobility of radioactive Cs+ in terms of large-scale soil
contamination, colloidal transport, and liner lifespan. The majority of research on Cs-
smectites has been performed using montmorillonites, the most common of the smectite
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
minerals, and many experiments have been conducted on suspensions rather than on
samples under typical soil or clay liner moisture conditions. Adsorption and desorption
isotherms of Cs+ on a variety of montmorillonite suspensions indicate that the adsorption
process is fast, on the order of a few hours to a few days (Staunton and Roubaud, 1997;
Tsai et al., 2001; Atun and Bodur, 2002; Shahwan and Erten, 2002), is reversible upon
exposure to a sufficient quantity of competing cations (Staunton and Roubaud, 1997;
Atun and Bodur, 2002), and that the amount of cation adsorbed is approximately equal to
the mineral CEC, typical of cation exchange reactions (Krumhansl et al., 2001; Atun and
Bodur, 2002).
Selectivity studies performed under a variety of conditions produce distribution
coefficients (Kd = amount Cs+in 1 g solid / amount Cs+ in 1 mL solution equilibrated
with solid) that indicate Cs+ is adsorbed preferentially over monovalent and divalent
cations like Li+, Na+, K+, Ca2+, Mg2+, and Sr2+(Iwasaki and Onodera, 1995; Staunton and
Roubard, 1997). Cs+ is particularly attracted to adsorption sites because its low charge to
radius ratio causes it to have a low hydration energy and, therefore, to be poorly hydrated
as compared to other cations (Ohtaki and Radnai, 1993). This encourages the formation
of strong inner sphere Cs+complexes with mineral surfaces, complexes in which no
water molecules intercede between the cation and the clay (Sposito et al., 1999). Before
diffusional movements shift the ion away from the adsorption site, inner sphere
complexes typically exist for 1 0 ' 7 to 1 0 ' 9 s, much longer than outer sphere complexes,
which feature water molecules between cation and clay players, and generally last only
10 10 to 10' 11 s (Sposito et al., 1999). In contrast, diffuse ions are surrounded by water
molecules, but spend even less time associated with a mineral surface. Although K+has
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a hydration energy low enough to facilitate the formation of some inner sphere
complexes with the clay surface, its presence does not appear to affect substantially the
selectivity in favor of Cs+ (Staunton and Roubard, 1997).
The value of Kd for Cs+changes little for low concentrations of Cs+, but decreases
with higher concentrations (Staunton and Roubard, 1997; Atun and Bodur, 2002).
Preference for trace amounts of Cs+increases with decreasing ionic strength (Staunton
and Roubard, 1997; Atun and Bodur, 2002), perhaps due to the larger proportion of Cs+
relative to other exchange cations, or to changes to clay stacking caused by changes in
ionic strength. Selectivity for trace amounts of Cs+also increases with decreasing
temperature over a range of 25 - 200 C (Iwasaki and Onodera, 1995; Shahwan and
Erten, 2002), possibly because higher temperatures allow other competing ions to
dehydrate and form more inner sphere complexes that better compete with Cs+.
Comparison of different adsorbent minerals provides further insight into
adsorption selectivity for Cs+. Iwasaki and Onodera (1995) compared the Kjs of several
smectite minerals and found that selectivity for Cs+correlated positively with the amount
of octahedral charge present in the clays. This dependence on the site of isomorphous
substitution is unusual, though it may be rationalized using the Hard and Soft Acid and
Base (HSAB) model of Xu and Harsh (1992). According to this model, octahedral
charge sites may be considered soft bases, because due to their position deep within the
clay layer, their charge is delocalized over a number of connected atoms, and is capable
of greater polarization and participation in more covalent interactions. On the other hand,
tetrahedral charge sites are hard bases, because their location closer to the surface means
their charge is more localized and less polarizable. The HSAB model suggests that soft
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
acids prefer to interact with soft bases, while hard acids prefer to interact with hard bases
(Xu and Harsh, 1992). Because Cs+is a much softer acid than any of the other ions it is
paired with in selectivity studies, a preference for interaction of Cs+with octahedral
charge sites would be predicted by the HSAB model.
Further studies involving a Ba2+-exchange treatment designed to remove any Cs+
not tightly bound to the surfaces of several Cs-smectites also provided strong correlations
between Cs+ retention and octahedral charge, and showed that trioctahedral minerals
retained less Cs+ than dioctahedral minerals (Onodera et al., 1998). The dimensions of
Cs+ are similar to that of the ditrigonal surface cavity, so it is often suggested that as the
cation adsorbs to the surface of a hydrated smectite, it sinks partway into this cavity when
attracted to an octahedral charge site located below (Onodera et al., 1998). A structural
hydroxyl group is located within the clay layer, between the octahedral site and the cavity
(Figure 1.1), but in dioctahedral clay minerals it is tipped at a 23 angle relative to the
surface (Giese, 1979), and does not interfere substantially with the electrostatic
interaction between cation and mineral charge. However, in trioctahedral minerals, the
structural hydroxyl group located at the bottom of the cavity is oriented perpendicularly
to the clay surface, and therefore can act to repel the cation away from a location within
the cavity (Onodera et al., 1998). Cations attracted to tetrahedral charge sites prefer
interlayer positions near the oxygen triads directly bonded to the substitution site, and
therefore do not enter surface cavities (Onodera et al., 1998). The latter two adsorption
interactions may be more easily exchanged with Ba2+, as the cations involved are in more
accessible interlayer positions, resulting in high Cs+ retention for dioctahedral clays with
a large amount of octahedral substitution. In the process of Cs+ fixation, cations are
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
forced into ditrigonal cavities by a heat treatment that causes dehydration and collapse of
the interlayer region, preventing subsequent attempts to remove the Cs+ via exchange
(Iwasaki and Onodera, 1995). The amount of Cs+ fixed in this manner increases with
increasing CEC and especially octahedral charge, with the use of hydrothermal
conditions, and with longer treatment times (Iwasaki and Onodera, 1995). A standard
procedure has been developed that uses the Cs+ fixation process for determining the
amount of structural charge in smectite minerals (Anderson and Sposito, 1991; Schroth
and Sposito, 1997).
The studies mentioned above have all been conducted using clay suspensions,
hydration conditions that are infrequent in the environmental context of the clay liners
used at waste containment facilities. Muurinen et al. (1987) and Oscarson et al. (1994)
examined the selectivity for Cs+ in compacted bentonite samples using different
techniques, and reported contradictory results. Muurinen et al. (1987) examined a Na-
bentonite plug compacted to 1.75 Mg m'3, which was exposed to NaCl solutions
containing 134Cs tracer on one side, and free of tracer on the other side. After
establishment of steady state diffusion of Cs+, selectivity coefficients were calculated for
the high concentration sides of the clay plugs, and were compared to more typical
compacted clay batch experiments. Both methods provided Kd values higher by a factor
of 2 to 3 when compared to those found for unconsolidated bentonite. Oscarson et al.
(1994) performed batch experiments on smectitic plugs with densities ranging from 0.5 to
1.5 Mg m'3. Equilibration times of up to 600 days were required for these compacted
mineral samples to reach steady state with the 137Cs+ concentration of 1 pmol m'3. The
resulting Kds for compacted clays were 1/2 to 1/3 of those of comparable suspensions,
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
perhaps because Cs+was unable to access small and occluded pores in the compact
structure (Oscarson et al., 1994). These results suggest that the selectivity for Cs+as
measured in suspensions must be adjusted for the porosity conditions found in clay liners
in order to produce accurate predictions of adsorption and diffusion. However, the
magnitude and direction of an appropriate adjustment is not yet clear.
Apparent self-diffusion coefficients (Da) [which receive the designation
apparent because they are calculated via measurements of solute flux and concentration
gradient within the adsorbent material, such that solute concentrations are based on the
volume of both adsorbent and solution, rather than the volume of the solution alone (Cho
et al., 1993)] have been measured for several compacted smectites and smectite-rich
bentonites. Values in the range of 10' 12 to 10' 13 m2 s' 1 are common, and increase
dramatically with decreasing density or CEC, and more modestly with increasing ionic
strength (Muurinen et al., 1987; Cho et al., 1993; Wanner et al., 1996; Tsai et al., 2001;
Cormenzana et al., 2003). A value of 6 x 10' 16 m2 s' 1 was reported for a 12 A Cs-
montmorillonite sample featuring a low water content (Calvet, 1973). These Das are
orders of magnitude smaller than the value of 2 x 1 0 ' 9 m2 s' 1 measured for the cation in
bulk solution (Nye, 1979), evidence of significant retardation resulting from Cs+
adsorption to mineral charge sites. A few lines of reasoning suggest that a substantial
portion of the diffusion present occurs within the interlayer regions of these compacted
mineral samples, including the observation of a discontinuity in Da after sample
compaction beyond a density calculated to remove most external pore space (Bourg et al.,
2003), and the observation that Na-saturated montmorillonite features a higher Da than
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the Ca-saturated version, though the latter possesses more non-interlayer pore space due
to differences in stacking of the layered smectite mineral (Kozaki et al., 1999).
Research on the properties of the water molecules that share the Cs-smectite
interlayer reveals behavior quite different from that found in smectite systems saturated
with well-hydrated cations like Li+. A number of XRD studies show that the layer
spacing of Cs-smectites remains in the range 1 1.9 - 12.5 A after exposure to water vapor
at any relative humidity (Mooney et al., 1952; Calvet, 1972; Prost, 1976; MacEwan and
Wilson, 1980; Berend et al., 1995; Chiou and Rutherford, 1997), or even after immersion
in an aqueous solution (MacEwan and Wilson, 1980). These values are consistent with
the presence of a single layer of water molecules between the two clay layers, while
similar measurements on Li-smectites reveal spacings consistent with one layer of water
for relative humidities of < 40%, and two to three layers at higher humidity conditions
(Berend et al., 1995). The high charge to radius ratio of Li+ is so attractive to water
molecules that, in dilute suspension, Li-smectite layers dissociate completely, allowing
maximum hydration of the mineral surfaces, while Cs-smectite layers remain bound
together in small stacks (Cebula et al., 1980). After removal of water, the Cs-smectite
system produces layer spacings close to 11 A (Berend et al., 1995; Chiou and Rutherford,
1997), perhaps averaged from spacings resulting from both symmetrical collapse of the
mineral layer, in which ditrigonal cavities encapsulate Cs+cations from opposing sides,
and asymmetrical collapse, in which the Cs+is nestled within a cavity on one side, and
rests against an oxygen triad on the opposite side (Berend et al., 1995).
Water vapor adsorption and desorption isotherms (Mooney et al., 1952; Calvet,
1972; Prost, 1976; Berend et al., 1995; Chiou and Rutherford, 1997) indicate increasing
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
hydration of Cs-smectites with increasing relative humidity. However, after examination
of adsorption data using the Brunauer-Emmett-Teller (BET) adsorption model (Brunauer
et al., 1938), Calvet (1972) argued that most of the water adsorbed resides on external
surfaces or in micropores, not in the interlayer region. Measurements of the micropore
volume in homoionic montmorillonites (Prost, 1975; Chiou and Rutherford, 1997;
Rutherford et al., 1997) have confirmed the extensive non-interlayer microporosity of Cs-
smectites as compared to smectites saturated with other alkali and alkaline earth cations.
The organization of dry Cs-smectite mineral structures into stable stacks 8-11 layers
thick, smaller than the stacks formed by other alkali cation-saturated smectites (Berend et
al., 1995), may contribute to this microporosity. Calvet (1972) suggested that the 12.4 A
layer spacing in Cs-smectite was achieved and stabilized after the adsorption of only
about 1.4 H20 per unit cell of the clay mineral, well below the nominal monolayer water
content of about 4 H20 per unit cell observed for smectites containing the strongly-
hydrating Li+ counterion (Calvet, 1972; Berend et al., 1995).
Spectroscopic techniques may elucidate the nature of Cs+and associated water
molecules within smectite interlayers. Infrared spectra of smectite minerals feature a
sharp peak associated with the stretching of structural hydroxyl groups at wavenumbers
of 3700-3600 cm'1, a broad form consisting of several overlapping peaks representing the
OH stretch of adsorbed water molecules and contained between wavenumbers 3800 and
2800 cm'1, followed by an adsorbed water HOH bend signal near 1600 cm'1, intense Si-O
vibrations around 1100 cm'1, several distinct structural hydroxyl bend bands from 920 to
800 cm' 1 representing hydroxyl groups attached to different octahedral atoms, and peaks
due to Al-O and Al-O-Si vibrations at lower wavenumbers. The influence of interlayer
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cs+ may be seen in the far infrared range, on peaks between 115 and 50 cm' 1 evident in
highly charged or tetrahedrally charged smectites substituted with poorly hydrated
cations (Fripiat, 1982; Schroeder, 1992), and assigned alternately to cation translational
vibrations (Ishii et al., 1969; Fripiat, 1982) or to Si-0 torsional or other structural
movements (Velde and Couty, 1985; Schroeder, 1992). The lack of these Cs-related
peaks in spectra of smectites with less charge, or mainly octahedral charge, may indicate
that these interlayer cations do not have equilibrium positions that encourage significant
interaction with surface oxygens (Schroeder, 1992). Although adsorbed water molecules
provide a direct measure of their presence on IR spectra, an indirect assessment of the
effect of water on structural hydroxyl vibrations can provide more specific chemical
information. On the basis of a detailed IR study of Cs-hectorite, Prost (1975) assigned a
band at 3685 cm' 1 to structural hydroxyl stretch vibrations occurring in the absence of
cations or water molecules, and a band at 3695 cm' 1 to the same movement occurring
when water molecules are close enough to enter surface cavities. By measuring the
relative areas of these two bands, he determined an interlayer coverage of water
molecules of 1.2 H2 O per unit cell for Cs-hectorite under relative humidity conditions of
less than 50%, a value similar to that postulated by Calvet (1972). The remaining water
molecules were not in close contact with the clay surface, and must therefore be adsorbed
in micropores (Prost, 1976).
The NMR-active 133Cs nucleus can be used to assess the structure and dynamics
of Cs+within smectite interlayers. Weiss et al. (1990a, b) examined Cs-hectorite systems
under a variety of different relative humidity and temperature conditions, and were able
to identify multiple peaks in the resulting spectra. Cs-hectorite examined at room
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
temperature and room humidity featured two adsorbed Cs+ peaks. A low, broad, more
shielded peak was assigned to Cs+ adsorbed within dehydrated interlayers, while an
intense, narrow, less shielded peak was assigned to motionally-averaged interlayer
counterions in Stem (inner and outer sphere) and Gouy (diffuse swarm) layers. By
reducing temperature or hydration, Weiss et al. (1990a, b) were able to observe the
narrow, motionally averaged peak split into two peaks: a narrow signal that was
unaffected by the concentration of Cs+ in solution, and therefore was thought to represent
inner and outer sphere Cs+, and a broader peak representing a more heterogeneous
environment, which exhibited decreased shielding with decreased water content as does
Cs+ in bulk solution, and was therefore associated with diffuse swarm Cs+. Subsequent
model calculations suggested a rate constant for exchange between these two species of
~1.0 x 104 s' 1 at room temperature and 100% relative humidity (Weiss et al., 1990a).
Motional averaging between the adsorbed Cs+peaks and another peak representing
cations in bulk solution did not take place, providing no evidence of exchange between
these species faster than -100 s' 1 (Weiss et al., 1990b). Upon heating to 500 C, the
spectmm of dehydrated Cs-hectorite provides evidence for the existence of two Cs+
species, a less symmetrical and possibly less coordinated species, and a more
symmetrical and possibly more coordinated species (Weiss et al., 1990a). Weiss et al.
(1990a) suggest that these signals may represent the asymmetric, 9 coordinated interlayer
cation and the symmetric, 1 2 coordinated interlayer cation, respectively.
Examination of other Cs-smectite systems adds further complexity to this
interpretation of the interlayer. Several Cs-montmorillonites investigated at room
temperature and humidity showed similar adsorbed Cs+ peaks (Weiss et al., 1990b; Kim
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
et al., 1995,1996), but varying temperature and hydration state did not induce the
distinctive peak split seen in the hectorite system, and thought to represent two interlayer
species capable of exchange reactions (Weiss et al., 1990b; Kim et al., 1996). It is
possible that the higher iron content of the montmorillonites induced paramagnetic
broadening sufficient to hide such a division. Spectra of Cs-beidellite samples under all
environmental conditions yielded two peaks representing adsorbed cations, assigned to
two Cs+interlayer species prevented from motional averaging by compositional
heterogeneity occurring over large spatial scales within the mineral (Weiss et al., 1990b).
Montmorillonite and beidellite spectra featured two adsorbed Cs+peaks upon heating to
450 C, but while the signals from montmorillonite samples appeared similar to those of
hectorite, the less shielded positions and broad shapes of the beidellite peaks indicated the
presence of two spatially separated, asymmetric, 9 coordinated interlayer species (Weiss
et al., 1990b).
Analysis of spectra provided by a second spectroscopic method builds on the
ideas presented above. Bostick et al. (2002) interpreted Lm-edge EXAFS spectra of Cs-
montmorillonite to suggest the presence of two types of Cs+ surface complex,
distinguished by average Cs- 0 distances of 3.2 A and 4.25 A, respectively. They
considered the complex with the shorter Cs-0 distance to be an outer sphere complex,
fully hydrated and therefore separated from mineral surfaces by water molecules, while
the complex with the longer Cs-0 distance was thought to be an inner sphere complex,
partially dehydrated and directly associated with a mineral surface. Their analysis
suggests that in a fully-saturated Cs-montmorillonite, the majority of cations exist as
outer sphere complexes directly coordinated by 2-3 water molecules. Exchange reactions
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
using Na+displaced much of the outer sphere Cs+, leaving behind only inner sphere
complexes coordinated to as many as 13 O atoms.
Nakano et al. (2003) recorded K-edge EXAFS spectra of Cs-bentonite samples
experiencing a range of pH conditions. No significant variation with pH was detected,
and spectral analysis identified a Cs-0 distance for the first coordination shell of
3.15 - 3.16 A away from Cs+, in agreement with Bostick et al. (2002). Nakano et al.
(2003) suggested that an average of 7.0 - 7.4 water molecules surround an outer sphere
Cs+complex, a number far greater than that suggested by Bostick et al. (2002). This
deviation may stem from use of large Debye-Waller constants (Nakano et al. (2003) used
values -0.04 A2, as opposed to the value of 0.015 A2used by Bostick et al. (2 0 0 2 ))
during spectral fitting (ODay et al., 1994). A second coordination shell was detected
3.59 - 3.62 A from the hydrated cation and featured 5.7 - 6 . 1 O, identified as members of
a six-oxygen cavity on the mineral surface (Nakano et al., 2003). The interpretation of
Nakano et al. (2003) favors the presence of outer sphere Cs+complexes within Cs-
smectite interlayers.
Bulk chemical and spectroscopic analyses of Cs-smectites suggest a complex
system sensitive to differences in mineral charge distribution and water content. The
high selectivity for, and reduced diffusion of, Cs+within smectites is thought to be due to
the formation of inner sphere complexes with interlayer positions near isomorphic
substitution sites, especially the softer, more polarizable octahedral sites, due to the low
hydration energy of the cation. Consistent with this interpretation are water adsorption,
XRD, and IR measurements indicating that the interlayer space of Cs-smectites contains
less water relative to smectites saturated with cations of high hydration energy, because
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cs+does not draw enough water into the region to become fully hydrated. Additional
water molecules adsorbed by Cs-smectite hydrates were thought to reside near external
surfaces and in abundant micropores present in the media. However, recent NMR and
EXAFS spectroscopic examinations of adsorbed Cs+indicate that the interlayer region
contains cations in both inner and outer sphere complexes. There is an urgent need to
reconcile these conflicting models of the Cs-smectite interlayer, in order to understand
cation-mineral interactions relevant to both the use of smectitic barriers at nuclear waste
storage facilities, and the movement of radioactive Cs+contaminants via colloidal
transport through aquifers.
1.3 Organo-smectite Systems
The global C cycle can be represented by a collection of C pools, interconnected
by a series of fluxes into and out of each pool (Figure 1.2). Soils contain a sizeable pool
of about 1500 Pg of C (1 Pg = 1015 g), more than the atmosphere and terrestrial
vegetation pools combined (Schlesinger, 1997). Most soil C exists as soil organic matter
(SOM), a diverse collection of substances that may be defined as consisting of all
biologically derived organic material located within or on the surface of soils (Baldock
and Skjemstad, 2000). Soil organic matter serves as energy and nutrient sources for a
host of microbes, which rework constituent molecules as they scavenge for food. The
resulting mixture of organic materials spans a gradient of molecular complexity, ranging
from monomers and simple acids to complex biopolymers, and a gradient of
decomposition, ranging from unaltered, easily-identified structures to molecules that have
undergone extensive microbial processing, or humification (Baldock and Skjemstad,
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
physical air-sea
exchange &
V biological
\ activity
Atmosphere (780)
primary
production
& respiration
modified by
land use change
carbonate
weathering
Land Biota (560)
vulcanism
Soil (1500)
DOM
export
Surface Ocean (1000)
river
transport
Deep Ocean (38,000)
fossil ization
carbonate
weathering
Fossil Organic C (5000)
/ burial
af sediment
& rock formation
Figure 1.2. Diagram of the global C cycle, after Figures 3.1 in IPCC (2001). The sizes
of all major C pools are listed in parentheses in units of Pg C, using values presented by
Schnitzer (1997), with additions from Australian Greenhouse Office (2002), and an
updated atmospheric pool value from Keeling and Whorf (2004). Thick lines indicate the
largest and therefore most important fluxes for the contemporary atmosphere, and purple
text describes anthropogenic fluxes. The abbreviation DOM near the center of the figure
stands for dissolved organic matter.
2000). Although all components of SOM display their own chemical properties, largely a
result of charge and polarity, size and conformation, all are constructed of the same basic
functional groups, including short and long-chain alkyl, aromatic, carboxyl and
carboxylate, alcohol and phenol, carbonyl, ether and ester groups (Stevenson, 1994).
Chemical separation schemes designed to divide SOM into different operational
groups have been used to study the most fascinating and chemically challenging form of
SOM, humic substances. Humic substances may be described as a series of complex and
26
permission of the copyright owner. Further reproduction prohibited without permission.
heterogeneous compounds produced through microbial action, as a by-product of
metabolic activity rather than in direct support of it (Sposito, 1989). The extensive
microbial action that leads to humification of a portion of SOM results in the creation of
diverse molecules that collectively share several chemical characteristics, such as narrow
C.N ratio, relatively high apparent molecular mass, and enrichment in oxygenated
functional groups (Stevenson, 1994). Humic substances are easily distinguished from
source biopolymers, such as lipids, lignin, proteins, and carbohydrates, by both
differences in molecular structure and their long-term persistence in soils (Sposito, 1989).
Humic substances may be chemically separated into three fractions, based on
solubility under acidic or alkaline conditions (Stevenson, 1994). Humin is the fraction of
humic substances that is insoluble under both conditions, either because it is tightly
bound to mineral components of the soil, or because it is made up of highly cross-linked
and condensed C compounds. Humic acid (HA) is the fraction that is soluble under
dilute basic conditions but not under acidic conditions, while fulvic acid (FA) is the
fraction that is soluble under both conditions. Differences in chemical properties underlie
these differences in solubility; FA molecules tend to be smaller in size, and to contain
more carboxyl and other polar functional groups, as compared to HA molecules
(Stevenson, 1994). However, it must be remembered that the separation of humic
substances into humin, HAs, and FAs is operational only, and does not indicate the
existence of three distinct types of organic molecule.
Sposito (1989) identifies four structural characteristics of humic substances that
contribute to their chemical reactivity in soils. The first of these is polyfunctionality, due
both to the existence of a variety of functional groups, and to their close proximity to
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
each other, which modifies the reactivity of each group. The second noteworthy property
is macromolecular charge, as most humic substances feature deprotonated functional
groups, such as carboxylates, under typical soil pH conditions. The third quality is
hydrophilicity, or the tendency to form strong H-bonds with water molecules solvating
polar or charged functional groups. The fourth humic characteristic is structural lability,
or the capacity to change molecular conformation in response to changes in pH, redox,
electrolyte, or solvent conditions. As a direct result of these four properties, humic
substances play an important role in soil systems, affecting the structure, water retention,
cation exchange, pH, mineral weathering, and fate of xenobiotics in soils (Stevenson,
1994). But because they do not have an identifiable molecular structure, traditional
chemical methods for determining functionality and reactivity have yet to provide a
comprehensive model of the interaction of humic molecules with each other and with soil
minerals at the molecular scale.
One means of studying these poorly-defined materials is by assembling virtual
humic molecules from well-defined biological precursors or appropriate distributions of
organic functional groups. These model humic substances represent heuristic tools used
to explore the effects of structure and composition on functional qualities like those listed
above. An examination of a range of virtual humic molecules and associated chemical
properties can provide clues concerning the makeup of actual humic molecules, and
generate hypotheses concerning their interactions with other soil components. More
information on model humic molecules is presented in Section 4.1.7.
Humic substances typically represent the oldest form of organic matter in soil
systems, indicating extreme resistance to microbial degradation (Stevenson, 1994).
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Through radiocarbon dating, soil C compounds can be divided into different groups
based on the mean residence time of the C, or the time an average C atom spends in the
soil medium before being oxidized by microbial action into CO2 and released into the
atmosphere. Many of the most stable soil C compounds, those with the longest residence
times, are closely associated with soil minerals (Oades, 1988; Stevenson, 1994; Meredith,
1997; Baldock and Skjemstad, 2000; Gaudinski, 2000; Wattel-Koekkoek et al., 2003).
Further evidence to support the critical role of minerals for protection of humic materials
lies in the rapid degradation of humic substances removed from their mineral matrices
and exposed to microbial cultures (Gramss et al., 1999), as well as the vulnerability of
organic materials with few mineral associations to decay after disturbances such as tillage
(Hassink, 1995; Golchin et al., 1997; Parfitt et al., 1997; Solomon et al., 2000). Organo-
mineral interactions appear to be essential for development of the recalcitrance of humic
substances (Schulten and Leinweber, 1996; Zech et al., 1997; Baldock and Skjemstad,
2000).
The longevity of soil humic materials becomes a matter of key importance when
considering the recent impacts of human expansion and industrialization on the C cycle
(Figure 1.2). Anthropogenic release of CO2 and other greenhouse gases to the
atmosphere through fossil fuel burning and land use change has caused the global
average surface temperature of our planet to increase by 0.6 0.2 C since the late 19th
century (IPCC, 2001). The effects of this rapid change are increasingly apparent in
ecosystems all over the world. Earlier springs and later autumns have resulted in changes
in the timing of seasonal activities of numerous plant and animal populations (Walther et
al., 2002). Warming trends have shifted low temperature climate regimes towards poles
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
or higher altitudes, and resulted in range shifts for individual species, leading to changes
in community structure and establishment of new invasive species (Walther et al., 2002).
The melting of polar ice, combined with the expansion caused by the warming of ocean
th
waters, has resulted in a rate of increase in global mean sea level during the 2 0 century
of 1.0 to 2.0 mm/yr (IPCC, 2001).
Atmosphere-ocean general circulation models predict that these changes will
become more extreme over the next century, even under relatively optimistic CO2
emission scenarios (IPCC, 2001). Increasing surface temperatures, melting snow and
glaciers, and warming and rising oceans will increase the frequency and intensity of
extreme weather events such as storms, droughts, and the El Nino-Southern Oscillation
(ENSO) (IPCC, 2001). Such dramatic incidents may lead to the collapse of many
essential and economically valuable natural and managed ecosystems, and may trigger
famine and disease outbreaks among human populations (IPCC, 2001). Without an
appropriate international socio-economic framework, climate change will exacerbate the
already wide division between rich and poor regions, as nations with greater economic
power will work to find technology-based solutions for problems resulting from climatic
shifts, while countries without such scientific resources may be unable to cope.
The initial international response to climate change took shape with the 1992
adoption of the United Nations Framework Convention on Climate Change (UNFCCC), a
framework for efforts to stabilize atmospheric concentrations of greenhouse gases at
levels that would prevent human-induced actions from leading to "dangerous
interference" with the climate system (UNFCCC, 2002). In December of 1997, delegates
from nations party to this framework drafted the Kyoto Protocol, which would commit
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
developed countries, as well as countries making the transition to a market economy, to
reduce their overall emissions of six greenhouse gases by at least 5% below 1990 levels
over the period between 2008 and 2 0 1 2 , with specific targets varying from country to
country (UNFCCC, 1997). So far, the U. S. government has not adopted the Kyoto
Protocol (UNFCCC, 2003).
Given that the U. S. is currently responsible for over 20% of the worlds CO2
emissions (UNFCCC, 2002), ecological and international pressures will force this nation
to take steps to curb these releases in the near future. One strategy the U. S. is likely to
employ is the use of land management as a means of emissions mitigation through
enhancement of existing C sinks in many agricultural and natural ecosystems
(Showstack, 2003). Several different land management techniques expected to increase
soil C have been suggested, including conservation tillage and management of
agricultural residues, intensification of cropping systems, reduction of soil erosion,
conversion of marginal agricultural land to non-agricultural, restorative uses, and
restoration of degraded lands (Bruce et al., 1999; Lai et al., 1999). A better
understanding of the C cycle as it relates to land use in temperate regions is necessary to
design an effective approach for sequestration of C over the diverse geography of the
U. S.
To learn how to promote long-lasting soil C storage, we must understand the
formation and properties of recalcitrant organo-mineral complexes. The individual and
collective strengths of the interactions binding organic matter to mineral surfaces, along
with the spatial organization of these two components, likely affects the capacity of soils
to sequester C, and determines the ease with which microbial degradation releases this
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
stored material (Oades, 1988; Baldock and Skjemstad, 2000). Improved understanding of
organo-mineral interactions at the molecular scale may provide us with the ability to
predict both the storage capacity of different landscapes, as well as the approximate
residence time of this stored C, fostering land management efforts able to expand soil C
pools accurately and efficiently, and reducing the need for intensive measurements to
monitor soil C reservoirs. A mechanistic understanding of these interactions is necessary
to generate sophisticated and successful soil C sequestration methods.
One means of investigating organo-mineral interactions involves the particle size
fractionation of soil samples. Such studies are nearly unanimous in finding increasing
amounts of organic materials in the finer, silt and/or clay-sized fractions of soils
dominated by well-crystallized minerals (Oades et al., 1987; Oades, 1988; Baldock et al.,
1992; Hassink, 1997; Hassink et al., 1997; Feller and Beare, 1997; Parfitt et al., 1997;
Mahieu et al., 1999; Nelson et al., 1999). As the particle size decreases, the C:N ratio
narrows (Baldock et al., 1992; Stevenson, 1994; Feller and Beare, 1997; Parfitt et al.,
1997), indicating that the associated SOM has received more microbial processing.
Biological indicators such as the (galactose + mannose):(arabinose + xylose) sugar ratio
also suggest a greater contribution of microbial rather than plant components to the clay
sized fraction (Feller and Beare, 1997). In general, the mean residence time of C
increases, or its rate of decomposition, decreases as particle size decreases between sand,
silt, and clay (Hassink, 1995; Feller and Beare, 1997; Monreal et al., 1997), suggesting
that more long-term C storage is occurring in organo-mineral associations present in the
clay-sized fraction. Eusterhues et al. (2003) used HF mineral dissolution to determine
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that organic matter associated with mineral particles was much older than that which was
not.
Moving from comparisons of the relative amounts of C contained within particle
size fractions of a single soil, to comparisons of the C present within a broad range of soil
landscapes, it is clear that temperature and moisture regimes exert significant control over
the amount of organic matter within a given soil (Oades, 1988, Amundson, 2001).
However, the influence of clay-sized particles on soil C storage capacity is reflected in
widely observed correlations between SOM content and soil texture in well-drained soils
containing crystalline minerals and experiencing a temperate climate (Anderson, 1987;
Oades, 1988; Stevenson, 1994; Feller and Beare, 1997). Such correlations do not provide
mechanistic information regarding the chemical interactions occurring between minerals
and organic matter. In fact, the minerals that make up the clay-sized particle fraction
vary widely from soil to soil, and feature different chemical and physical properties such
as charge and surface area.
Studies comparing the organic matter of soils dominated by smectite minerals, a
common constituent of the clay-sized fraction of temperate soils, with soils dominated by
other minerals, present a complicated picture of the significance of the mineral involved.
Some datasets indicate that the general correlation of the percentage of silt- and clay
sized particles with either total SOM (Feller and Beare, 1997) or SOM associated with
these particles (Hassink, 1997), is indistinguishable for soils influenced by smectite or by
their low charge, low surface area, 1:1 aluminosilicate cousin, kaolinite. Another dataset
(Schulten and Leinweber, 2000) indicates smectitic particle size fractions feature a wide
range of organic C contents, often higher than those from kaolinitic particle size
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
fractions. However, smaller scale decomposition studies indicate that the presence of
smectite greatly reduces the mineralization of organic materials (Allison et al., 1949;
Saggar et al., 1996), and radiocarbon dating reveals mean residence times of thousands of
years for the C within organic material associated with smectitic soils, as compared to
mean residence times of only hundreds of years for that associated with kaolinitic soils
(Wattel-Koekkoek et al., 2003). Along the same lines, the turnover time associated with
soil organic materials is positively correlated with the amount of smectite associated with
these materials in some soils (Monreal et al., 1997). Evidence from extraction (Wattel-
Koekkoek et al., 2001) and adsorption (Chorover and Amistadi, 2001) experiments
suggest that the functional composition of the retained organic material is influenced by
the presence of smectite surfaces. Both studies indicate substantial adsorption of alkyl
and carboxylate functional groups, among others.
Smectite minerals often form from parent materials rich in polyvalent base cations
like Ca2+(Borchardt, 1989), so it is important to determine to what extent these cations
may affect the ability of smectitic soils to store organic matter. There is substantial
evidence, ranging from additions (Sokoloff, 1938; Muneer and Oades, 1989a-c) or
replacements (Gaiffe et al., 1984; Muneer and Oades, 1989a) of Ca2+in soil samples, to
landscape level trends (Oades, 1988), that an abundance of this polyvalent cation
facilitates the retention of soil organic matter. Holder and Griffith (1983) compared
acidic and calcareous versions of smectite-rich Vertisols, and determined that strong
organo-mineral interactions occurred in both types of soil, but that the organic matter
present in the calcareous soils was especially difficult to extract. Based on the studies
presented above, it would appear that, while the presence of smectite minerals in soils
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
may not result necessarily in larger reservoirs of soil C, it may reduce significantly the
rate of decomposition of organic materials through the formation of organo-mineral
associations, especially when in the company of Ca2+. More information is needed to
characterize fully the organo-smectite interactions occurring within soils.
How might smectites interact with organic matter on a molecular level? A simple
comparison of the overall negative charge present in both smectites and humic substances
under typical soil pH conditions might lead one to assume that the two substances would
repel each other. However, on the more chemically relevant scale of organic functional
groups and mineral reaction microsites, it becomes apparent that a number of chemical
reactions can take place (Table 1.1). These interactions may lead to relative protection of
the organic molecule from microbial attack, and it is this protection that can result in
long-term C sequestration in the soil. Preservation of organic molecules likely takes the
form of a combination of chemical protection, or specific chemical interactions that make
the molecules unappealing to scavenging microbes, and physical protection, in which the
location and structure of adsorbing minerals makes it difficult for microbes to gain access
to the organic molecules (Schulten and Leinweber, 1996).
The extensive basal surfaces of smectite minerals, described in Section 1.1,
contain a variety of microsites and chemical features allowing adsorption of organic
molecules, including negatively charged structural sites, inorganic cations attracted to
these charge sites, and recently revealed hydrophobic microsites (Jaynes and Boyd,
1991a, 1991b; Boyd and Jaynes, 1992; van Oss and Giese, 1995; Giese et al., 1996;
Laird, 1996; Chiou and Rutherford, 1997). The negative structural charge sites can
attract positively charged organic moieties or molecules to positions at or near smectite
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
surfaces (Table 1.1). Inorganic cations balancing this negative charge can attract
negatively charged organic moieties or molecules, creating a cation bridge between the
mineral surface and the organic functional group (Sposito, 1989). The presence of a
water molecule between cation and organic material results in an organo-mineral
interaction known as a water bridge (Sposito, 1989). Polar organic functional groups can
participate in hydrogen bonding (H-bonding) interactions with the plane of O atoms at
the basal mineral surface, or with adsorbed water molecules. Finally, hydrophobic
organic moieties or molecules feel entropic encouragement from surrounding water
molecules to adsorb to hydrophobic mineral microsites (Theng, 1979, 1982). A
polyfunctional organic molecule may become adsorbed to a smectite surface via multiple
interactions, discouraging desorption.
The organic matter associated with smectite minerals in most soils is adsorbed to
external smectite surfaces (Baldock and Skjemstad, 2000). Through organo-mineral
adsorption interactions, smectite particles may partially encapsulate organic molecules,
preventing microbes and their enzymes from easily approaching this C source, and
therefore inhibiting its decomposition (Baldock and Skjemstad, 2000). However,
smectite interlayer regions should be far more effective locations for long-term protection
of organic molecules, for a number of physical and chemical reasons. An important
physical feature of the interlayer region is its narrow geometry, typically less than a
nanometer in width. This restricted dimension prevents the entry of soil microbes,
usually near a micrometer in length, into the area. Organic matter that finds its way into
this region may therefore receive substantial long-term physical protection from hungry
microorganisms. Although the interlayer remains accessible to several extracellular
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
enzymes, the activity of many of these enzymes is substantially impaired, perhaps as a
result of adsorption to mineral surfaces or changes in conformation caused by the
confining geometry (Pinck, 1962; Theng, 1979; Bums, 1986).
A few studies of natural soil systems have documented the existence of organo-
smectite interlayer complexes. Kodama and Schnitzer (1971) found evidence of an
interlayer organic complex within the fine clay fraction of a Spodosol Ae horizon. Some
of this organic matter resisted a mild treatment with H2 O2 , and an XRD study of the
smectitic minerals present revealed basal spacings that collapsed upon heating to
550 C, indicating removal of the C material. Similarities between differential thermal
analysis (DTA) thermograms provided by the natural, mildly oxidized organo-mineral
sample and a synthesized FA-montmorillonite interlayer complex (Kodama and
Schnitzer, 1969) supplied further evidence that a portion of the organic matter resided
within smectite interlayers. Later characterization of the SOM associated with this fine
clay fraction using IR and solid-state 13C NMR spectroscopies indicated the prominence
of long-chain alkyl functional groups (Schnitzer et al., 1988). Extraction with 0.5 M
NaOH revealed the presence of substantial amounts of FA in the associated organic
matter, while extraction with n-hexane and chloroform removed many simple, and
frequently water-insoluble, long-chain alkyl molecules such as n-alkanes and alcohols
from the sample (Schnitzer et al., 1988).
A study of two New Zealand soil A horizons by Theng et al. (1986) gave solid-
state 13C NMR and XRD evidence suggesting the existence of alkyl humic substances
within the interlayers of smectitic clay minerals in one of these soils, also a Spodosol.
The organic matter concentrated in the clay-sized fractions of the two soils examined was
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
resistant to peroxidation, and to dithionite/citrate/bicarbonate and pyrophosphate
extractions. Upon heating of the Spodosol clay fraction to 400 C for 24 hours, this
resistant organic matter mineralized and the interlayer spacing of the clay minerals
decreased by 4 A. Theng et al. (1986) calculated that i f the alkyl molecules were
arranged so that the C backbones and other functional groups lay parallel to the silicate
layer, the van der Waals radii of the resulting organic molecules would require a 4 A
width to fit within the interlayer. Given this arrangement, they also calculated that only
about half the resistant SOM would fit in the interlayer. The rest of the organic carbon
must be adsorbed to external clay surfaces. Subsequent radiocarbon dating of the
peroxide resistant C revealed this material to be substantially older than the C in both the
whole soil and the clay-sized fraction (Theng et al., 1992). Multiple pyrolysis
techniques, combined with methylation (Schulten et al., 1996), confirmed the primarily
long-chain alkyl composition of the complexed organic matter indicated by the NMR
data (Theng et al., 1986). The authors suggested a combination of soil factors that may be
necessary for the formation of these interlayer complexes: the presence of smectitic clay,
an accumulation of organic matter with low microbial activity, and an acidic soil
environment needed to counteract pH-dependent interparticle repulsive forces (Theng et
al., 1986).
Like Spodosols, soils derived from volcanic ash are frequently acidic, and may
therefore be appropriate environments for the formation of organo-smectite interlayer
complexes. Satoh and Yamane (1971) examined a clay fraction (0.5 - 0.1 pm) of the
predominantly montmorillonitic A horizon of an acidic, volcanic ash soil, via XRD.
Samples dried over P2 O5 exhibited basal spacings of 14 A, indicating the presence of an
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
interlayer species with an average width of 4 A. A strong peroxidation treatment resulted
in P2(>5-dehydrated samples featuring basal spacings of 12 A. Satoh and Yamane (1971)
deduced that the interlayer constituents consist of a combination of organic matter, which
was destroyed by the peroxidation treatment, and hydroxy-Al ions, which remained. A
complete collapse of the interlayer did not take place until the mineral was heated to
600 C, and this collapse occurred more quickly and completely with previous
peroxidation of the sample, or in the presence of air, as opposed to an inert atmosphere
such as N2 . The role of oxidation in the removal of the interlayer material, either through
treatment with peroxide or through heating in the presence of air, provides further
evidence supporting the existence of organic interlayer species.
An acidic Inceptisol profile examined by Righi et al. (1995) provides another
example of natural organo-smectite interlayer complexes. Characterization of clay
fractions of the soil horizons established a pathway of mineral weathering of the less
acidic (pH 5.3) C horizon, containing mica, vermiculite, chlorite, and various
interstratified minerals, into the more acidic (pH 4.6), largely smectitic Al and A2
horizons. Mica minerals were the dominant source of potassium in the soil, and the
amount of K2O in the minerals of each horizon decreased with increasing presence of
smectite and vermiculite, and decreasing mica content. The amount of C left in samples
after a mild peroxidation treatment, followed by citrate/bicarbonate/dithionite extraction,
was inversely correlated with K2O content, suggesting the resistant C was associated with
smectite and vermiculite minerals. Stepwise heating of clay fractions from the A and B
horizons to 550 C resulted in gradual collapse of many of the 14 A basal spacing
minerals to 1 0 A, indicating removal of an interlayer species. Pyrolysis data revealed the
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
composition of the resistant organic material in clay fractions of the A and B horizons to
be predominantly alkylaromatic, with few of the signals linked to the presence of
carbohydrates, phenols, lignins, and lipids, and abundant in C horizon pyrolysates.
While considerable experimental evidence indicates that strong external or
interlayer organo-smectite adsorption interactions take place within soil systems, the
complexity of these natural samples prevents a clear view of the specific adsorption
mechanisms involved. Laboratory experiments involving the reaction of well
characterized humic fractions with pure smectite minerals have provided further
information regarding relevant organo-mineral interactions. Solution conditions
including pH regime and counterion identity, exert substantial influence on the adsorption
of organic materials. Properties of the humic fraction and smectite mineral involved, and
the addition o f washing or drying phases to the adsorption scheme, also modify the
amount of adsorption observed.
The pH condition under which these adsorption reactions take place affects the
availability of interlayer adsorption sites to humic molecules. Laboratory experiments
designed to evaluate external versus interlayer adsorption of humic substances with Na-
montmorillonite, as assessed by increases in basal spacing upon adsorption, report no
substantial inter layer adsorption above pH 5 (Schnitzer and Kodama, 1966; Martin
Martinez and Perez Rodriguez, 1969). This is consistent with natural observations of
organo-smectite interlayer complexes in acidic soils (Kodama and Schnitzer, 1971; Satoh
and Yamane, 1971; Theng et al., 1986; Righi et al., 1995). Accessible interlayer regions
provide more mineral surface area for adsorption reactions, leading to the retention of far
more organic material within low pH systems (Schnitzer and Kodama, 1966; Martin
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Martinez and Perez Rodriguez, 1969). Evans and Russell (1959) show similar trends in
C retention with pH, while finding no evidence of interlayer adsorption. Schnitzer and
Kodama (1966) noted that the FA they used for their adsorption experiments had an
apparent pKa of 4.5, indicating that the deprotonation of carboxyl groups hindered entry
into the interlayer region. Increases in pH result in the negative charge of FAs through
deprotonation of acidic functional groups, which lead to significant electrostatic repulsion
of these organic molecules, given the overall negative charge of smectite minerals
(Theng, 1979). Humic substances adsorbed in interlayer regions under acidic pH regimes
were not substantially desorbed after washing with water or salt solutions, but were
desorbed when treated with NaOH, suggesting that deprotonation of carboxyl functional
groups leads to rapid repulsion of organic and mineral particles (Schnitzer and Khan,
1972). Ion-dipole cation bridging is suggested as the dominant adsorption mechanism
operating within Na-montmorillonite interlayers, given the lack of displacement of
adsorbed humic molecules when leached with 1 M NaCl, and the inflection point
observed in the adsorption-pH curve near the pH value corresponding to the apparent pKa
of the humic sample used (Schnitzer, 1986).
The acidic conditions that encourage the entry of humic materials into the
interlayer also encourage dissolution of the smectite mineral and consequent release of
hydroxy-aluminum ions into solution (Theng, 1979). It is likely that the formation of
organo-smectite interlayer complexes at such low pH values is strongly influenced by the
presence of these ions, which can attract negatively charged organic materials and then
become complexed themselves by clay charge sites within the interlayer (Singer and
Huang, 1988; Buondonno et al., 1989; Violante et al., 1999; De Cristofaro and Violante,
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2001). In fact, through additions of hydroxy-aluminum ions, humic materials can be
introduced into smectite inter layers even under neutral pH conditions (Violante et al.,
1999). Nevertheless, numerous experiments contrasting the effects of different
counterions on the adsorption of humic substances under low pH conditions indicate that
hydroxy-aluminum ions are not the only cations significant to interlayer adsorption
processes.
For example, Mirabella et al. (1996) assessed Na- and Ca-montmorillonite
complexed at pH 3.5 and 4.5 to a FA and noted basal spacings for the Na-saturated
system dropped from 20.9 A to 14.6 A with increasing pH, though the basal spacings for
the Ca-saturated system remained at 17.3 A under both conditions. These results suggest
that the hydroxy-Al FA adsorption mechanism plays a stronger role for the Na+ systems
than for the Ca2+ ones. A small portion of the adsorbed humic material could be removed
through water washes, and was thought to be adsorbed originally through water bridges.
The majority of humic molecules, which resisted desorption via water washes, were
assumed to be adsorbed through stronger cation bridges.
A more extensive comparison of the effects of saturating cation on FA-
montmorillonite complexes produced at pH -2.5, as summarized by Schnitzer and Khan
(1972), revealed extensive variation in FA adsorption and layer spacing with different
24"
saturating cations (Kodama and Schnitzer, 1968). Forty mg of Pb -montmorillonite
adsorbed the largest amount of organic matter, 25 mg, and produced the highest basal
spacing (19.2 A), closely followed by Na+ (25 mg and 18 A), while Ca2+ and Mg2+
adsorbed only 2 0 mg of the FA and produced basal spacings near 16 A. Trends in
organic adsorption and interlayer expansion with cation identity provide a cation
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sequence unlike that obtained through comparisons of metal-FA stability constants or
ionic potential (Kodama and Schnitzer, 1968). Thermal analysis of Pb2+, Cu2+ and Na+,
the three cations that encourage exceptional C retention and interlayer expansion, within
inorganic, hydrated montmorillonite samples, indicates these cations are relatively poorly
hydrated near smectite surfaces (Kodama and Schnitzer, 1968). Perhaps this reduced
hydration allows humic molecules to form cation bridging complexes more readily with
these cations, leading to greater adsorption of organic matter. Theng (1979) suggests that
many factors such as the presence of hydroxy-aluminum ions, the occurrence of specific
metal-FA interactions, and the variation in layer spacing with saturating cation, all exert
an influence on the formation of humic interlayer complexes within these smectite
systems.
Under neutral pFI conditions, adsorption of humic materials is primarily limited to
sites on external smectite surfaces, avoiding the complications of interlayer fixation.
Theng and Scharpenseel (1975) and Theng (1976) observed linear adsorption isotherms
describing the interaction of humic materials with montmorillonite saturated with various
cations. Adsorption of organic molecules within pores, followed by intercrystalline
expansion to expose fresh adsorption sites, was invoked as a mechanism to explain the
linearity of these isotherms (Theng and Scharpenseel, 1975; Theng, 1976). Comparison
of the isotherms revealed that for polyvalent cations, there was a positive correlation
between the logarithm of the adsorption isotherm slope and the ionic potential of the
saturating cation, while for monovalent cations, there was a negative correlation.
Consistent with these trends in ionic potential is the formation of water bridging
complexes involving polyvalent cations, and cation bridging complexes involving
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
monovalent cations. Al3+, Fe3+, and possibly Cu2+cations were also suspected of forming
ligand exchange interactions with humic molecules. Comparison of HA and FA behavior
indicated that secondary interactions between adsorbed organic molecules contributed to
the overall affinity of the humic substance for available mineral surfaces (Theng, 1976).
These results may be compared to those provided in an earlier study by Evans and
Russell (1959). Montmorillonites saturated with Ca2+and Mg2+adsorbed the most HA in
this case, followed by montmorillonites containing H+, then K+and Na+. However, when
the same experiments were repeated using the FA fraction, the H-saturated mineral
adsorbed the most organic matter, followed by those containing Ca2+, Mg2+, and K+and
Na+. Linear adsorption isotherms were observed, as described above. Adsorption
experiments were conducted over a broad range of pH conditions, and while no
indication of interlayer complexation was recorded, in the absence of Ca2+ the decrease in
C retention with increasing pH was especially dramatic. Theng (1979) suggests that the
release of hydroxy-Al ions is responsible for the increased adsorption of organic matter
by H-montmorillonites relative to montmorillonites saturated with other monovalent
cations. The lack of adsorption was observed for the native montmorillonite, which had
not experienced acid treatment as part of the purification process (Evans and Russell,
1959), provides further evidence favoring the importance of hydroxy-Al in organo-
smectite complexation. Theng (1979) concludes that ion-dipole interactions in the form
of cation or water bridges must be the mechanism responsible for most organo-mineral
adsorption in these systems.
Varadachari et al. (1991) performed similar adsorption experiments, but subjected
their samples to a drying regime after the adsorption process. They noted that the cation
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sequence defined by the amount of humic material adsorbed by montmorillonite is
similar to the sequence described by Theng and Scharpenseel (1975) for those cations
included in both studies. However, further analysis reveals that the cation sequence
provided by the dried samples does not correspond to that provided by comparisons of
ionic potential or hydrated or dehydrated ionic radius. Instead, the sequence is similar to
that provided by comparisons of cationic bonding energies (lyotrope series). Varadachari
et al. (1991) infer that the extent of fixation of humic materials is partially determined by
the attraction of the cation for the clay surface. That K+encourages greater adsorption of
humic molecules than Mg2+, a reversal of the trend predicted by cationic bonding
energies, is explained by the promotion of K+ fixation to clays through drying, as well as
the reduced affinity of Mg2+for HA. Thus, Varadachari et al. (1991) conclude that while
cations are responsible for the majority of humic adsorption to montmorillonite, both the
bonding strength of cation to clay and cation to humic molecule determine the amount of
organo-mineral adsorption possible. The hydration properties of each cation influence
each of these bonding strengths indirectly.
Based on the bulk chemical information presented above, many researchers
(Schnitzer and Khan, 1972; Theng and Scharpenseel, 1975; Theng, 1976; Theng, 1979)
suggested that ion-dipole cation or water bridging interactions were the most probable
adsorption mechanism active under acidic pH conditions, when many carboxyl groups
are fully protonated. However, spectroscopic investigations show that deprotonated
organic functional groups play a strong role in the adsorption of humic fractions. In fact,
IR spectra of synthesized complexes between humic materials and smectite minerals
indicate that carboxylate groups are dominant under alkaline, neutral (Tan, 1976), and
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
acidic (Mirabella et al., 1996; Chorover and Amistadi, 2001) pH conditions. The
adsorption process appears to select for humic molecules containing carboxylate groups,
suggesting that these anionic components may be involved in particularly strong organo-
mineral interactions favored over similar ion-dipole ones, despite the potential for
electrostatic repulsion given the negative charge of the clay mineral. Although it has
been suggested that organic anions could adsorb predominantly on the positively charged
edge sites of smectites, experiments in which these edges sites are blocked through the
adsorption of specific organic compounds show little change in the amount of humic
material adsorbed, suggesting that edge sites are not a major site for C fixation
(Varadachari et al., 1995). Further inspection of spectra indicates that alkyl and
carbohydrate functional groups are also preferentially adsorbed by montmorillonite
(Chorover and Amistadi, 2001).
The synthesized complexes described above were all formed over the space of
hours or days, a time scale that may prove less relevant to soil systems. Long term
(10 week) adsorption experiments conducted by Scharpenseel (1966) and involving
14C-labeled humic fractions and a variety of minerals including montmorillonite in
neutral to alkaline suspensions, provided evidence that the presence of NFL+, Al3+, Ca2+
and Fe3+ cations, increasingly in this order, led to substantial retention of humic materials
within the smectite mineral matrix, while the presence of H4, Na+, K+and Mg2+ did not.
Interestingly, after elution with NaOH, the resulting organo-mineral samples contained
similar quantities of humic material regardless of which cation had been used, indicating
that the higher valency cations formed less stable adsorption complexes. A portion of the
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
material that remained adsorbed to the montmorillonite samples after elution was
suspected of being involved in hydrophobic interactions with mineral surfaces.
Further evidence indicating a hydrophobic component to the adsorption of humic
materials by smectite minerals is provided by Specht et al. (2000), who use size exclusion
chromatography and zetapotential analysis to show that large and hydrophobic humic
molecules are preferentially adsorbed, at least over a 2 day period, by a bentonite mixture
containing 45% swelling smectite minerals, while small and carboxylate-rich molecules
tend to remain in solution for a range of acidic to near neutral pH conditions. This
finding contradicts those described by Tan (1976) and Chorover and Amistadi (2001),
who document substantial adsorption of humic molecules with low molecular masses and
high carboxylate content. Perhaps the impurity of the mineral phase used by Specht et al.
(2 0 0 0 ) influenced the results.
Nayak et al. (1990) examined the adsorption of a number of humic materials with
different acidities, as well as a single humic sample experiencing various degrees of
methylation to remove carboxyl groups, to determine that organic fractions with fewer
acidic functional groups achieved greater levels of adsorption to Ca-montmorillonite over
a period of several hours. Humic materials with lower acidity must be less hydrophilic
and feel less electrostatic repulsion from negatively charged clay minerals under the
neutral pH conditions of the organo-mineral suspension, encouraging adsorption. The
drying regime employed after complexation in this study may favor the formation of
hydrophobic interactions within these systems as well. Varadachari et al. (1991)
adsorbed humic materials to a range of reduced-charge montmorillonites, and found that
reduced charge led to increased adsorption, until mineral charge was diminished so much
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that a loss of swelling ensued. Although this trend could be considered evidence for the
importance of hydrophobic interactions on mineral surfaces made more hydrophobic
through charge reduction, Varadachari et al. (1991) suggest it is evidence that adsorption
occurs through the prying open of interlayer regions, a process that is hindered by
electrostatic repulsion when mineral and humic molecule are fully charged, and which
results in the disruption of the stacking of clay layers.
Hydrophobic and H-bonding interactions are assumed to be less stable under high
temperature conditions (Varadachari et al., 1994), so one means of assessing the
importance of this adsorption mechanism is to compare the amount of fixation of humic
materials achieved under different temperature regimes. Numerous studies indicate that
uptake of organic matter is both rapid and insensitive to temperature over a range of
suspension conditions and adsorption methods (Evans and Russell, 1959; Schnitzer and
Khan, 1972; Varadachari et al., 1994). This may indicate that hydrophobic interactions,
if present, may act only on humic molecules already adsorbed to smectite surfaces via
stronger cation or water bridges.
The data described in the preceding paragraph suggest that hydrophobic
interactions are unimportant in synthesized organo-mineral complexes. At the same time,
soil studies show that alkyl functional groups are preferentially preserved in the presence
of smectites (Theng et al., 1986; Schnitzer et al., 1988; Schulten et al., 1996; Righi et al.,
1995; Chorover and Amistadi, 2001; Wattel-Koekkoek et al., 2001), a process that can be
explained most easily through the activity of hydrophobic interactions. This is just one
example of the apparent contradictions present within the diverse array of data
concerning the complexes that form between humic materials and smectite minerals.
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
While it is readily apparent that organo-mineral interactions within soils can protect
humic substances from microbial degradation (Oades, 1988; Stevenson, 1994; Meredith,
1997; Baldock and Skjemstad, 2000; Eusterhues et al., 2003), and that smectites
specifically are capable of maintaining this protection over long time scales (Monreal et
al., 1997; Wattel-Koekkoek et al., 2003), adsorption information at the molecular scale is
lacking. In order to understand soil C processes relevant to management decisions
designed to sequester organic matter, it is necessary to resolve questions concerning the
relative roles of different organo-mineral adsorption mechanisms over a range of pH
conditions relevant to soil systems.
1.4 Atomistic Simulations Exploring the Smectite Interlayer Region
1.4.1 Molecular Modeling Techniques
Although we lack experimental tools capable of direct exploration of individual
molecular interactions occurring at mineral surfaces, we do possess sufficient empirical
and quantum chemical knowledge to create virtual molecular systems and observe
interactions that occur within them. These molecular scale computer models can provide
insights that lead to improved analysis of experimental data, or suggest the use of new
analytical techniques and provide predictions of potential results. In turn, improved
experimental, quantum chemical, and computational information leads to refinements in
both molecular modeling methods and parameters. Molecular modeling is a useful, even
essential, adjunct to experimental science, offering a mechanistic view of chemical
interactions at a dimensional scale as yet inaccessible via empirical means.
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Atomistic simulations predict molecular interactions based on a force field, an
expression of energy terms and associated parameters that provide each atom with
electrostatic and van der Waals properties, valence characteristics associated with bond
energies, and describe the way each atom interacts with every other atom in the system.
The use of a force field, with its inherent atomic scale resolution, distinguishes atomistic
simulations from quantum chemical ones, which account for the interactions of
subatomic particles, and which can be applied only to systems containing relatively few
atoms given the complex quantum calculations required. Numerous force fields are
described in molecular modeling literature (Park and Sposito, in press). Force fields
derived from quantum chemical information taken from simulations on smaller molecular
systems are known as ab initio force fields, while those designed to produce behavior
consistent with experimental data are known as empirical force fields. Selection of an
appropriate force field for the system under study is perhaps the most important step in
any simulation exercise, as a poor choice can result in unrealistic behavior. For example,
while empirical force fields perform quite well for the specific molecules and
experimental conditions present in the experimental datasets on which they are based,
they may not provide realistic simulations of systems containing different types of
molecules or experiencing different conditions (Skipper et al., 1995).
After force field selection comes the choice of which computational algorithm or
algorithms to employ to obtain results of interest. The three atomistic simulation
algorithms in common use are energy minimization (EM), Monte Carlo (MC), and
molecular dynamics (MD). Energy minimization, also known as geometry optimization
or molecular mechanics, involves repeated adjustment of atomic coordinates leading to a
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reduction in the potential energy of the system, as calculated by summing all relevant
interaction terms described in the force fields energy expression. Several specific EM
algorithms are available, and some EM methods involve the use of different algorithms at
different stages in the minimization (Accelrys Inc., 2001). Most EM algorithms rely on
gradient optimization, in which atoms are nudged into different positions in order to
reduce the net forces acting upon them (Park and Sposito, in press). Minimization
algorithms are independent of temperature, and 0 K conditions are assumed. The
potential energy of the molecular system drops after each EM step, until no further
decrease is possible. The specific conformation adopted after minimization is considered
to occupy a stable state known as a local energy minimum, an energy state lower than
any that would be produced from systems featuring the same general organization but
many small differences.
However, a molecular system has the potential to express many different
conformations representing local energy minima, and there is no way of knowing how
similar a single minimized structure is to that which would produce the lower global
energy minimum, representative of a well-equilibrated virtual system that is expected to
correspond structurally to its counterpart in the real world. In order to avoid entrapment
within a local, not global, energy minimum, a modeler may perform several EM
calculations using different initial states, then select the minimized system that features
the lowest potential energy, or best fits experimental measurements of interest (Section
5.2.1). Alternatively, a modeler may use the final configuration of a minimized system as
the initial configuration for calculations via MD or another form of atomistic simulation,
with the expectation that the shift to an algorithm that allows the system to adopt more
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
conformations will encourage the molecules to equilibrate at their most stable state, the
global energy minimum (Sections 4.2, 5.2.2, and 5.2.3). Despite the potential for
structural entrapment within local energy minima, EM is a valuable molecular modeling
method, because it is computationally inexpensive, can provide useful information on its
own regarding molecular interactions, especially if several initial states are examined,
and can be used in conjunction with more computationally intensive modeling methods in
seeking out configurations with low potential energies.
The Metropolis MC algorithm (Metropolis et al., 1953) is a computationally
intensive method that can be used to identify a systems global energy minimum,
typically starting from a single initial structure. Each step in a MC calculation begins
with the random movement of an atom or molecule (Allen and Tildesley, 1987). The
potential energies of the system before and after this random movement are compared. If
the potential energy is lower after the change, the change is accepted; if the potential
energy is higher, the change is accepted or rejected by comparing a randomly selected
number x (0 < x < 1) to the Boltzmann factor e~AUIkB'! , where AU is the change in energy
of the system, ka is the Boltzmann constant, and T is the absolute temperature of the
system (K). I f x is less than the Boltzmann factor, the change is accepted, while if x is
equal to or greater than the Boltzmann factor, it is rejected. Because some moves that
increase the potential energy of the system are accepted using the MC algorithm, the
system avoids entrapment within local energy minima. The temperature variable within
the Boltzmann distribution above indicates that temperature is an adjustable parameter
within MC calculations, another improvement over EM. After several thousand or a few
million MC steps, system potential energy will stabilize, oscillating around an average
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
value that typically represents the global energy minimum. Average properties of the
molecular system under these equilibrated conditions provide the same thermodynamic
and structural information obtained from EM, with the added statistical dimensions of
standard deviation or confidence interval.
Molecular dynamics, as the name suggests, is an appropriate method for obtaining
dynamic information from a model system. Molecular dynamics algorithms solve
Newton-Euler equations of motion at an atomic scale, essentially simulating the motion
of atoms or molecules as they interact with each other over time (Allen and Tildesley,
1987). Particle velocities are assigned initially using a Maxwell-Boltzmann distribution
centered at a particular temperature. By means of one of several algorithms (Allen and
Tildesley, 1987), the trajectory of each atom is calculated over a timestep on the order of
1 fs ( 1 0 ' 15 s), using the initial velocity as modified by forces imposed on the atom from
surrounding particles. The velocity associated with each atom is adjusted to reflect
acceleration caused by these forces, and then calculation of trajectories for the next MD
timestep begins. Molecular dynamics output provides time-dependent measurements
such as rates of diffusion and exchange. Static or animated depictions of the coordinates
sampled by model atoms during MD simulations can provide information concerning
multibody interactions occurring within the virtual system. It is also possible to compute
thermodynamic properties from MD data via time-averaged integrations of energetic or
structural values (Allen and Tildesley, 1987).
After selection of a force field and a simulation algorithm, it becomes necessary
to make a few other choices concerning the model system. The first is the choice of
ensemble, relevant to MC and MD calculations, which identifies system variables to be
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
held constant during the course of the simulation. The most common ensembles include
the canonical (NVT) ensemble, which fixes the number of particles (N), the volume (V),
and the temperature in the system; the microcanonical (NVE) ensemble, which fixes the
number of particles, the volume, and the energy (E) in the system; the isothermal-isobaric
(NPT) ensemble, which fixes the number of particles, the pressure (P), and the
temperature in the system; and the grand canonical (pPT) ensemble, which fixes the
chemical potential (p), the pressure, and the temperature in the system.
The next choice, relevant to all three molecular modeling methods described
above, is that of a periodic or nonperiodic molecular system. A typical periodic system is
surrounded in three dimensions by copies of itself, such that atoms near one edge of the
simulation cell interact with replicates of atoms near the opposite edge. Periodic
boundary conditions are useful for simulation of densely packed systems, or systems with
regular mineral structures. A periodic mineral structure appears to extend infinitely in all
directions, alleviating problems that stem from the high proportion of broken edge sites
present in the tiny mineral fragments that fit into simulation cells, as compared to natural
minerals. A nonperiodic system, on the other hand, is surrounded only by vacuum.
Nonperiodic systems can be useful for simulation of large molecules in dilute solutions.
Finally, decisions concerning the summation of energy terms must be made. It is
only feasible to incorporate the energy terms associated with every single atom-atom
nonbond (electrostatic and van der Waals) interaction into the calculated potential
energies of small nonperiodic systems of perhaps a hundred atoms or less. For larger
systems, several methods can be used to limit the number of energy terms collected to
those acting over smaller distances, and therefore most relevant to the energetics of the
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
molecules under study. Appropriate limitations are essential for periodic systems in
order to avoid edge effects, or artifacts related to the size of the simulation cell and the
use of periodic boundary conditions (Skipper et al., 1995). The three-dimensional Ewald
sum, involving a two-step summation (one in real space and one in reciprocal space), is
regarded as the most accurate means of computing long-range electrostatic contributions
within periodic systems (Allen and Tildesley, 1987). Van der Waals and nonperiodic
electrostatic terms may be summed using a direct cutoff method, in which energy
contributions produced by atom-atom interactions are ignored if the distance between
atoms is greater than the specified cutoff distance. Alternatively, these terms may be
summed using a spline switching method, which uses a spline function to gradually
reduce the interaction energy with distance experienced by each atom, from that
calculated directly for atom-atom pairs separated by less than the spline-on distance, to
zero at the spline-off distance (Accelrys Inc., 2001). This spline switching method, while
more computationally intensive, avoids the discontinuities in potential energy and its
derivatives inherent in the direct cutoff method.
The results of carefully constructed atomistic simulations, employing appropriate
force fields, algorithms, and modeling conditions, can provide considerable insight
concerning molecular interactions occurring in natural systems. Experimental subjects
that produce either conflicting data when analyzed with different techniques, as in the
case of Cs-smectite hydrates (Section 1.2), or insufficient data due to the complexity of
the materials under study, as in the case of humic substances adsorbed to smectite
surfaces (Section 1.3), are excellent candidates for investigation via molecular modeling.
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.4.2 The Structure o f Smectite Minerals
The desire to understand and predict the adsorptive properties of smectite
interlayers has resulted in numerous studies of the region using molecular modeling
techniques. The development of suitable atomic structures for smectite minerals is a
prerequisite for this research. Idealized coordinates for each atom in the crystalline cells
of clay minerals are adapted from extensive XRD studies of smectites or related minerals
(Figure 1.3). Most atomistic simulations of smectite systems rely on rigid mineral
structures, in which the positions of the atoms making up the smectite layers are fixed
relative to one another (Delville, 1991; Skipper et al., 1991,1995; Sato et al., 1992;
Keldsen et al., 1994). The use of rigid smectite layers reduces the number of calculations
required in each study, making them less computationally intensive. However, this
simplification also eliminates the ability of the simulated clay to alter its conformation
subtly in response to different conditions. Studies of fully flexible smectite minerals
(Breu and Catlow, 1995; Teppen et al., 1997; Sainz-Diaz et al., 2001a, b; Cygan, 2003;
Cygan et al., 2004) require the use of specially designed force fields that properly
describe the behavior of octahedrally coordinated Al, and as a result are quite
computationally costly.
Early simulations of smectite interlayers further reduced the number of
calculations necessary by neglecting portions of the clay structure distant from the region
of interest. For example, the clay structure created by Delville (1991) is periodic in two
(ab) dimensions only, and lacks the Si tetrahedral sheets farthest from the interlayer,
while early work by Skipper et al. (1991) describes a 3D periodic clay structure with
interaction sites on Si, surface O, and structural hydroxyl groups only, for a total of 28
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
interaction sites per crystalline unit of smectite. With continued advances in computer
hardware and software, however, a complete description of the clay mineral is now the
norm.
Another shortcut employed in conjunction with rigid smectite structures is the use
of a perfectly flat mineral surface featuring hexagonal arrangements of Si tetrahedra
(Figure 1.3), rather than a more realistic corrugated mineral surface, featuring ridges
created through tipping of the Si tetrahedra, which are organized around dioctahedral,
rather than hexagonal, surface cavity sites (Figure 1.1). Rigid clay structures also feature
fixed structural hydroxyl groups, and the selection of the angle of orientation of these
Figure 1.3. Typical structure of a rigid,
idealized montmorillonite simulation cell
suitable for molecular modeling. Colors
encode elemental identity as described in
Figure 1 . 1 , with the exception that the
interlayer cations shown here are large brown
spheres. This 8 crystalline unit simulation
cell is outlined in grey in the cross-sectional
view provided in a), and features a layer
spacing of 15.2 A. Note the uniform, rigid
structural hydroxyl orientations in a), as well
as the flat mineral surfaces in a) and the
perfectly hexagonal six-oxygen cavities in the
top-down view of the mineral surface seen in
b). Compare this structure to the more
realistic montmorillonite structure seen in
Figure 1.1.
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
groups with respect to the mineral surface may affect resulting interactions with
interlayer species. The orientation of hydroxyl groups cannot be determined from XRD
studies (Giese, 1979), as the H atom has too few electrons to induce significant scattering
of x-rays. Electrostatic energy calculations suggest that the hydroxyl groups of
trioctahedral smectites have a perpendicular orientation with respect to the surface, while
the hydroxyl groups of dioctahedral smectites are tipped at a shallow angle relative to this
surface (Giese, 1979) (Figure 1.1). A few early calculations employed mineral structures
inconsistent with these calculations (Delville, 1991; Skipper et al., 1991), but later
versions of smectites constructed by Skipper and co-workers (Greathouse and Sposito,
1998), and commonly used by the molecular modeling community, feature more
appropriate hydroxyl angles. Orientations of structural hydroxyl groups measured in
simulations of flexible smectite minerals indicate that the angle is influenced by the
distribution of charge (Sainz-Diaz et al., 2001a; Cygan, 2003), as well as by the
organization of hydroxyls in cis- or tram-vacant configurations (Sainz-Diaz et al.,
2 0 0 1 a), as described below.
In dioctahedral smectites, a pair of hydroxyl groups points toward each vacancy
in the octahedral layer. In cis-vacant smectites, the pair of hydroxyls is located on the
same side of the vacancy, while in tram-vacant smectites, the pair is located on opposite
sides of the vacancy (Figure 1.4). Experimental studies indicate that smectites are
typically cis-vacant (Drits et al., 1998), and one set of simulations suggests that cis-
vacant smectites have slightly lower potential energies than tram-vacant smectites under
many conditions (Sainz-Diaz et al., 2001a), but the most common atomic structure of a
dioctahedral smectite features tram-vacant hydroxyl groups (Skipper et al., 1995).
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 1.4. A top-down view of clusters of alumina (and magnesia) octahedra within the
octahedral layers of cA-vacant (left) and tram-vacant (right) dioctahedral smectites,
using a polyhedral display style to emphasize geometry, after Sainz-Diaz et al. (2001a).
The letters C and T mark the location of mineral hydroxyl groups within cA-vacant and
trans-vacant octahedral sheets. Colors are linked to atom identity as described in Figure
1 . 1 , and polyhedra adopt the color of the atom at the center of the shape (the blue-green
polyhedra are alumina octahedra, while the blue polyhedron represents a magnesium
octahedron, a negative charge site).
After careful selection of parameters related to the crystalline structure of a
smectite, a modeler must specify the number of crystalline units in each simulation cell,
large enough to capture interactions of interest, while small enough to make the required
calculations practical. A typical simulation cell features 8 unit cells of smectite,
measuring 2 1 . 1 2 A x 18.28 A (Skipper et al., 1995) (Figure 1.3), although simulations
have been performed with model minerals containing as many as 24 unit cells (Delville,
1991), or as few as 1 unit cell (Sainz-Diaz et al., 2001a, b). Smith (1998) advocates the
use of two layers of smectite mineral per simulation cell, such that two different
interlayer regions can be modeled simultaneously, in order to better explore the effects of
clay layer registry on interactions of interest.
A discussion of force fields, and particularly electrostatic charges, appropriate to
the atoms making up smectite minerals must follow. It is possible to model smectite
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
interlayer systems adequately using several different force fields, either adapted or
designed specifically to model these minerals, though many of these force fields lack the
ability to describe octahedrally coordinated Al, and therefore require that the mineral
structure be fixed, as mentioned above. When creating or adjusting a force field for the
task of modeling smectite interlayer regions, great care must be taken in assigning
appropriate charges to each atom in the mineral. A variety of methods have been
employed to disperse electrostatic charge over the smectite structure, from use of formal
charges for all but a few atoms in the mineral (Breu and Catlow, 1995; Sainz-Diaz et al.,
2 0 0 1 a, b), to derivations of partial charges for all atoms based on quantum calculations
with smaller atom clusters (Delville, 1991; Teppen et al., 1997). One popular set of
smectite charges evolved from a combination of assumptions, including the use of
octahedral cation charges equivalent to the formal charge, hydroxyl charges borrowed
from an ab initio force field describing water molecules, and surface O atom charges
limited by quantum chemical estimates of charge distributions in silicate clusters
(Skipper et al., 1995). After electrostatic parameters for the uncharged mineral are
developed, changes to these parameters that mimic the effect of isomorphic substitution
must be implemented. Most smectite charge distributions simply reduce the partial
charge associated with a tetrahedral or octahedral charge site by le (Delville, 1991;
Skipper et al., 1991, 1995; Sato et al., 1992; Breu and Catlow, 1995; Sainz-Diaz et al.,
2001a, b). However, the beidellite modeled by Teppen et al. (1997) features a smearing
of the negative charge at each tetrahedral substitution site, such that the substituted cation
and the O atoms in the surrounding tetrahedron all experience a portion of the charge
reduction.
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
After initial development, these model smectites and their variants have been used
in the examination of numerous clay interlayer systems. Summaries of key findings for
inorganic and organic smectite interlayer systems may be found in Sections 2.1 and 5.1,
respectively. It is important to remember that the structure of these model smectites, as
well as the force fields and charge distributions associated with them, undergo frequent
refinement with improved knowledge of the systems under study. Recent advances in
quantum chemical methods now make feasible quantum simulations of small smectite
systems (Boek and Sprik, 2003). Further quantum chemical study of smectite structure
will no doubt lead to the development of more accurate structures, force fields, and
charge distributions for these important mineral systems.
1.5 Hypotheses and Research Objectives
Molecular modeling calculations can provide novel insights regarding the
complex surface chemistry associated with smectite minerals. Synthesis of the
information presented in Section 1.2 highlights a need for molecular simulations designed
to assess interpretations of Cs-smectite experimental data and to suggest further ideas
regarding cation speciation and mobility within the interlayer. In addition, the convenient
periodic molecular system provided by models of the smectite interlayer region can be
used to elucidate possible mechanisms involved in the adsorption of large, complex
organic molecules to soil smectites, furthering our understanding of the C-stabilizing
environments described in Section 1.3. Atomistic simulations addressing these two very
different smectite systems may provide us with broader appreciation of the utility of these
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
computational techniques when applied to investigations of surface chemistry relevant to
natural and engineered mineral structures.
Guiding the research concerning Cs-smectite hydrates presented in this thesis is
the hypothesis that interlayer regions of these systems contain less than a monolayer of
water, and feature cations that form both inner and outer sphere complexes with mineral
surfaces near charge sites, resulting in the reduced mobility of Cs+. This view of the Cs-
smectite interlayer is consistent with the experimental interpretations summarized in
Section 1.2. To test this hypothesis, molecular modeling techniques were used to
determine equilibrium water content, cation speciation, and self-diffusion coefficients of
model Cs-smectite systems, as described in Chapter 2. Further analysis using animated
MD data brought to light potential chemical mechanisms behind cation hydration and
speciation trends observed in the Cs-smectite hydrates, as described in Chapter 3.
Exploration of the organo-mineral interactions that take place between humic
molecules and smectite minerals is guided by the hypothesis that adsorption of such
complex organic molecules is the result of a combination of individual adsorption
interactions between organic functional groups and the mineral surface, which tend to
exclude water from near-surface locations, and are dominated by cation bridging
interactions between carboxylate functional groups and the mineral surface when the
humic material is deprotonated. A model humic molecule is needed to test this
hypothesis; the DOM molecule designed by Schulten (1999) proved to mimic the
properties of natural humic materials under a variety of hydration and protonation states,
as documented in Chapter 4. A thorough examination of the adsorption of this model
molecule to hydrated Ca-montmorillonite surfaces, suggesting the availability of an array
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of adsorption mechanisms sensitive to pH conditions, is provided in Chapter 5. Overall
conclusions regarding the utility of molecular modeling in the exploration of molecular
scale chemical interactions taking place on heterogeneous environmental surfaces are
presented in Chapter 6 .
1.6 References
. Accelrys Inc. (2001) Cerius2 Release 4.6. Accelrys Inc.
Allen M. P. and Tildesley D. J. (1987) Computer Simulation of Liquids. Oxford
University Press.
Allison F. E., Sherman M. S., and Pinck L. A. (1949) Maintenance of soil organic matter:
I. Inorganic soil colloid as a factor in retention of carbon during formation of humus. Soil
Sci. 6 8 , 463-478.
Amundson R. (2001) The carbon budget in soils. Ann. Rev. Earth Planet. Sci. 29, 535-
562.
Anderson D. W. (1987) Pedogenesis in the grassland and adjacent forests of the great
plains. Adv. Soil Sci. 7, 53-93.
Anderson S. J. and Sposito G. (1991) Cesium-adsorption method for measuring
accessible structural surface charge. Soil Sci. Soc. Am. J. 5 5 , 1569-1576.
Atun G. and Bodur N. (2002) Retention of Cs on zeolite, bentonite and their mixtures. J.
Radioanal. Nucl. Chem. 235, 275-279.
Australian Greenhouse Office. (2002) Understanding Greenhouse Science: Frequently
Asked Questions. Australian Greenhouse Office.
Avery S. V. (1996) Fate of caesium in the environment: Distribution between the abiotic
and biotic components of aquatic and terrestrial ecosystems. J. Environ. Radioact. 30,
139-171.
Baldock J. A., Oades J. M., Waters A. G., Peng X., Vassallo A. M., and Wilson M. A.
(1992) Aspects of the chemical structure of soil organic materials as revealed by solid-
state 13C NMR spectroscopy. Biogeochem. 16, 1-42.
Baldock J. A. and Skjemstad J. O. (2000) Role of the soil matrix and minerals in
protecting natural organic materials against biological attack. Org. Geochem. 31, 697-
710.
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Barber D. A. (1964) Influence of soil organic matter on the entry of caesium-137 into
plants. Nature 204, 1326-1327.
Berend I., Cases J. M., Francis M., Uriot J. P., Michot L., Masion A., and Thomas F.
(1995) Mechanism of adsorption and desorption of water vapor by homoionic
montmorillonites. 2. The Li+, Na+, K+, Rb+ and Cs+-exchanged forms. Clays Clay Miner.
43, 324-336.
Bleam W. F. (1991) Soil science applications of nuclear magnetic resonance
spectroscopy. Adv. Agron. 46, 91-155.
Boek E. S. and Sprik M. (2003) Ab initio molecular dynamics study of the hydration of a
sodium smectite clay. J. Phys. Chem. B 107, 3251-3256.
Borchardt G. (1989) Smectites. In Minerals in Soil Environments, 2nd Ed. (ed. J. B.
Dixon and S. B. Weed), pp. 675-727. Soil Science Society of America.
Bostick B. C., Vairavamurthy M. A., Karthikeyan K. G., and Chorover J. (2002) Cesium
adsorption on clay minerals: An EXAFS spectroscopic investigation. Environ. Sci.
Technol. 36, 2670-2676.
Bourg I. C., Bourg A. C. M., and Sposito G. (2003) Modeling diffusion and adsorption in
compacted bentonite: A critical review. J. Contam. Hydrol. 61, 293-302.
Boyd S. A. and Jaynes W. F. (1992) Role of layer charge in organic contaminant sorption
by organo-clays. Layer charge characteristics of clays. Pre-Meeting Workshop CMS and
SSSA, 48-77.
Bradbury M. H. and Baeyens B. (2000) A generalised sorption model for the
concentration dependent uptake of caesium by argillaceous rocks. J. Contam. Hydrol. 42,
141-163.
Breu J. and Catlow C. R. A. (1995) Chiral recognition among tris diimine-metal
complexes. 4. Atomistic computer modeling of a monolayer of [Ru(bpy)3]2+ intercalated
into a smectite clay. Inorg. Chem. 34, 4504-4510.
Brown G. E. and Sturchio N. C. (2002) An overview of synchrotron radiation
applications to low temperature geochemistry and environmental science. In Applications
of Synchrotron Radiation in Low-Temperature Geochemistry and Environmental Science
(ed. P. A. Fenter, M. Rivers, N. C. Sturchio, and S. R. Sutton) pp. 1-119. Mineralogical
Society of America.
Bruce J. P., Frome M., Haites E., Janzen H., Lai R., and Paustian K. (1999) Carbon
sequestration in soils. J. Soil Water Conserv. 54, 382-389.
Brunauer S., Emmett P. H., and Teller E. (1938) Adsorption of gases in multi-molecular
layers. J. Am. Chem. Soc. 60:309-319.
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Buondonno A., Felleca D., and Violante A. (1989) Properties of organo-mineral
complexes formed by different addition sequences of hydroxy-Al, montmorillonite, and
tannic acid. Clays Clay Miner. 37, 235-242.
Bums R. G. (1986) Interaction of enzymes with soil mineral and organic colloids. In
Interactions of Soil Minerals with Natural Organics and Microbes (ed. P. M. Huang and
M. Schnitzer), pp. 429-451. Soil Science Society of America, Inc.
Calvet R. (1972) Water adsorption on clays: Study of the hydration of montmorillonite.
Bull. Soc. Chim. Fr. 8, 3097-3104.
Calvet R. (1973) Hydration of montmorillonite and diffusion of exchangeable cations.
Part II. Study of exchangeable cation diffusion in montmorillonite. Ann. Agrono. 24, 135-
217.
Cebula D. J., Thomas R. K., and White J. W. (1980) Small angle neutron scattering from
dilute aqueous dispersions of clay. J. Chem. Soc. Faraday 176, 314-321.
Chiou C. T. and Rutherford D. W. (1997) Effects of exchanged cation and layer charge
on the sorption of water and EGME vapors on montmorillonite clays. Clays Clay Miner.
45, 867-880.
Cho W. J., Oscarson D. W., Gray M. N., and Cheung S. C. H. (1993) Influence of
diffusant concentration on diffusion coefficients in clay. Radiochim. Acta 60, 159-163.
Chorover J. and Amistadi M. K. (2001) Reaction of forest floor organic matter at
goethite, bimessite, and smectite surfaces. Geochim. Cosmochim. Acta 65, 95-109.
Chun K. S., Kim S. S., and Kang C. H. (2001) Release of boron and cesium or uranium
from simulated borosilicate waste glasses through a compacted Ca-bentonite layer. J.
Nucl. Mat. 298, 150-154.
Collins A. L., Walling D. E., Sichingabula H. M., and Leeks G. J. L. (2001) Using 137Cs
measurements to quantify soil erosion and redistribution rates for areas under different
land use in the Upper Kaleya River basin, southern Zambia. Geoderma 104, 299-323.
Cormenzana J. L., Garcia-Gutierrez M., Missana T., and Junghanns A. (2003)
Simultaneous esimation of effective and apparent diffusion coefficients in compacted
bentonite. J. Contam. Hydrol. 61, 63-72.
Cremers A., Elsen A., Preter P. D., and Maes A. (1988) Quantitative analysis of
radiocaesium retention in soils. Nature 335, 247-249.
Cygan R. T. (2003) Molecular models of metal sorption on clay minerals. In Molecular
Modeling of Clays and Mineral Surfaces (ed. J. D. Kubicki and W. F. Bleam), pp. 143-
194. Clay Minerals Society.
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cygan R. T., Liang J.-J., and Kalinichev A. G. (2004) Molecular models of hydroxide,
oxyhydroxide, and clay phases and the development of a general force field. J. Phys.
Chem. BI OS, 1255-1266.
De Cristofaro A. and Violante A. (2001) Effect of hydroxy-aluminium species on the
sorption and interlayering of albumin onto monmorillonite. Appl. Clay Sci. 19, 59-67.
Delville A. (1991) Modeling the clay water interface. Langmuir 7, 547-555.
Di Stefano C., Ferro V., and Porto P. (1999) Linking sediment yield and caesium-137
spatial distribution at basin scale. J. Agric. Engng. Res. 74, 41-62.
Drits V. A., Lindgreen H., Salyn A. L., Ylagan R., and McCarty D. K. (1998)
Semiquantitative determination of trans-vacant and cis-vacant 2:1 layers in illites and
illite-smectites by thermal analysis and X-ray diffraction. Am. Miner. 83, 1188-1198.
Dumat C., Quiquampoix H., and Staunton S. (2000) Adsorption of cesium by synthetic
clay-organic matter complexes: Effect of the nature of organic polymers. Environ. Sci.
Technol. 34, 2985-2989.
Earl W. L. and Johnston C. T. (1998) Applications of NMR spectroscopy to the study of
the chemistry of environmental interfaces. In Structure and Surface Reactions o f Soil
Particles (ed. P. M. Huang, N. Senesi, and J. Buffle), pp. 251-280. John Wiley & Sons,
Ltd.
Eusterhues K., Rumpel C., Kleber M., and Kogel-Knabner I. (2003) Stabilisation of soil
organic matter by interactions with minerals as revealed by mineral dissolution and
oxidative degradation. Org. Geochem. 34, 1591-1600.
Evans D. W., Alberts J. J., and Clark R. A. (1983) Reversible ion-exchange fixation of
cesium-137 leading to mobilization from reservoir sediments. Geochim. Cosmochim.
Acta 47, 1041-1049.
Evans L. T. and Russell E. W. (1959) The adsorption of humic and fulvic acids by clays.
J. Soil Sci. 10, 119-132.
Facchinelli A., Gallini L., Barberis E., Magnoni M., and Hursthouse A. S. (2001) The
influence of clay mineralogy on the mobility of radiocaesium in upland soils of NW Italy.
J. Environ. Radioact. 56, 299-307.
Feller C. and Beare M. H. (1997) Physical control of soil organic matter dynamics in the
tropics. Geoderma 79, 69-116.
Fesenko S. V., SoukhovaN. V., SanzharovaN. I., Avila R., Spiridonov S. I., Klein D.,
Lucot E., and Badot P.-M. (2001) Identification of the processes governing long-term
accumulation of 137Cs by forest trees following the Chernobyl accident. Radiat. Environ.
Biophys. 40, 105-113.
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fripiat J. J. (1982) Application of far infrared spectroscopy to the study of clay minerals
and zeolites. In Advanced Techniques for Clay Mineral Analysis (ed. J. J. Fripiat), pp.
191-210. Elsevier Scientific Publishing Co.
Gaudinski J. B., Trumbore S. E., Davidson E. A., and Zheng S. (2000) Soil carbon
cycling in a temperate forest: Radiocarbon-based estimates of residence times,
sequestration rates and partitioning of fluxes. Biogeochem. 51, 33-69.
Giese R. F. (1979) Hydroxyl orientations in 2:1 phyllosilicates. Clays Clay Miner. 27,
213-223.
Giese R. F., Wu W., and van Oss C. J. (1996) Surface and electrokinetic properties of
clays and other mineral particles, untreated and treated wtih organic or inorganic cations.
J. Dispers. Sci. Technol. 17, 527-547.
Golchin A., Baldock J. A., and Oades J. M. (1997) A model linking organic matter
decomposition, chemistry, and aggregate dynamics. In Soil Processes and the Carbon
Cycle (ed. R. Lai, J. M. Kimble, R. F. Follett, and B. A. Stewart), pp. 245-266. CRC
Press.
Gramss G., Ziegenhange D., and Sorge S. (1999) Degradation of soil humic extract by
wood- and soil-associated fungi, bacteria, and commercial enzymes. Microb. Ecol. 37,
140-151.
Greathouse J. and Sposito G. (1998) Monte Carlo and molecular dynamics studies of
interlayer structure in Li H20 3 -smectites. J. Phys. Chem. B 102, 2406-2414.
Gri N., Stammose D., Guillou P., and Genet M. (2000) Mobility of 137Cs related to
speciation studies in contaminated soils of the Chernobyl area. J. Radioanal. Nucl. Chem.
246, 403-409.
Hassink J. (1995) Decomposition rate constants of size and density fractions of soil
organic matter. Soil Sci. Soc. Am. J. 59, 1631-1635.
Hassink J. (1997) The capacity of soils to preserve organic C and N by their association
with clay and silt particles. Plant Soil 191, 77-87.
Hassink J., Whitmore A. P., and Kubat J. (1997) Size and density fractionation of soil
organic matter and the physical capacity of soils to protect organic matter. Euro. J.
Agron. 7, 189-199.
Hayes M. H. B. and Himes F. L. (1986) Nature and properties of humus-mineral
complexes. In Interactions o f Soil Minerals with Natural Organics and Microbes (ed. P.
M. Huang and M. Schnitzer), pp. 103-158. Soil Science Society of America, Inc.
He Q. and Walling D. E. (1997) The distribution of fallout 137Cs and 210Pb in undisturbed
and cultivated soils. Appl. Radiat. Isot. 48, 677-690.
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hendricks S. B. (1942) Lattice structure of clay minerals and some properties of clays. J.
Geol. 50, 276-290.
Hird A. B., Rimmer D. L., and Livens F. R. (1996) Factors affecting the sorption and
fixation of caesium in acid organic soil. Euro. J. Soil Sci. 47, 97-104.
Hofmann U. K., Endell K., and Wilm D. (1933) Kristallstruktur und quellung von
montmorillonit. Z. Krystallogr. 86, 340-348.
Holder M. B. and Griffith S. M. (1983) Some characteristics of humic materials in
Caribbean Vertisols. Can. J. Soil Sci. 63, 151-159.
IPCC (2001) Climate Change 2001: The Scientific Basis. Contribution o f Working Group
I to the Third Assessment Report o f the Intergovernmental Panel on Climate Change (ed.
J. T. Houghton, Y. Ding, D. J. Griggs, M. Noguer, P. J. van den Linden, X. Dai, K.
Maskell, and C. A. Johnson). Cambridge University Press.
Ishii M., Nakahira M., and Takeda H. (1969) Far infrared absorption spectra of micas. In
Proceedings o f the Interanational Clay Conference, 1969 (ed. L. Heller), pp. 247-259.
Israel Universities Press.
Iwasaki T. and Onodera Y. (1995) Sorption behaviour of caesium ions in smectites. In
Proceedings o f the 10th International Clay Conference, 1993, pp. 67-73. CSIRO
Publishing.
Jaynes W. F. and Boyd S. A. (1991a) Clay mineral type and organic compound sorption
by hexadecyltrimethylammonium-exchanged clays. Soil Sci. Soc. Am. J. 55, 43-48.
Jaynes W. F. and Boyd S. A. (1991b) Hydrophobicity of siloxane surfaces in smectites as
revealed by aromatic hydrocarbon adsorption from water. Clays Clay Miner. 39, 428-
436.
Jensen D. J. and Radke C. J. (1988) Cesium and strontium diffusion through sodium
montmorillonite at elevated temperature. J. Soil Sci. 39, 53-64.
Johnston C. T. and Aochi Y. O. (1996) Fourier transform infrared and raman
spectroscopy. In Methods o f Soil Analysis. Part 3. Chemical Methods (ed. D. L. Sparks),
pp. 269-321. Soil Science Society of America.
Keeling C. D. and Whorf T. P. (2004) Atmospheric CO2 records from sites in the SIO air
sampling network. In Trends: A Compendium o f Data on Global Change. Carbon
Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department
of Energy, Oak Ridge, Tenn., U.S.A.
Keldsen G. L., Nicholas J. B., Carrado K. A., and Winans R. E. (1994) Molecular
modeling of the enthalpies of adsorption of hydrocarbons on smectite clay. J. Phys.
Chem. 98, 279-284.
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Kersting A. B., Efurd D. W., Finnegan D. L., Rokop D. J., Smith D. K., and Thompson J.
L. (1999) Migration of plutonium in ground water at the Nevada Test Site. Nature 397,
56-59.
Kim Y., Cygan R. T., and Kirkpatrick R. J. (1996) 133Cs NMR and XPS investigation of
cesium adsorbed on clay minerals and related phases. Geochim. Cosmochim. Acta 60,
1041-1052.
Kim Y., Kirkpatrick R. J., and Cygan R. T. (1995) 133Cs NMR study of Cs reaction with
clay minerals. In Environmental Issues and Waste Management Techniques in the
Ceramic and Nuclear Industries (ed. Y. Jain and R. Palmer), pp. 629-636. American
Ceramic Society.
Koch D. (2002) Bentonites as a basic material for technical base liners and site
encapsulation cut-off walls. Appl. Clay Sci. 21, 1-11.
Kodama H. and Schnitzer M. (1968) Effects of interlayer cations on the adsorption of a
soil humic compound by montmorillonite. Soil Sci. 106, 73-74.
Kodama H. and Schnitzer M. (1969) Thermal analysis of a fulvic acid-montmorillonite
complex. International Clay Conference, 165-11 A.
Kodama H. and Schnitzer M. (1971) Evidence for interlamellar adsorption of organic
matter by clay in a podzol soil. Can. J. Soil Sci. 51, 509-512.
Kozaki T., Sato H., Sato S., and Ohashi H. (1999) Diffusion mechanism of cesium ions in
compacted montmorillonite. Eng. Geol. 54, 223-230.
Krumhansl J. L., Brady P. V., and Anderson H. L. (2001) Reactive barriers for 137Cs
retention. J. Contam. Hydrol. 47, 233-240.
Laird D. A. (1996) Interactions between atrazine and smectite surfaces. In Herbicide
Metabolites in Surface Water and Groundwater (ed. M. T. Meyer and E. M. Thurman),
pp. 86-100. American Chemical Society.
Lai R., Follett R. F., Kimble J., and Vole C. V. (1999) Managing U.S. cropland to
sequester carbon in soil. J. Soil Water Conserv. 54, 374-381.
Li Y. and Lindstrom M. J. (2001) Evaluating soil quality-soil redistribution relationship
on terraces and sleep hillslopes. Soil Sci. Soc. Am. J. 65, 1500-1508.
Mabit L., Bernard C., Laverdiere M. R., and Wicherek S. (1999) Assessment of soil
erosion in a small agricultural basin of the St. Lawrence River watershed. Hydrobiologia
410, 263-268.
MacEwan D. M. C. and Wilson M. J. (1980) Interlayer and intercalation complexes of
clay minerals. In Crystal Structures o f Clay Minerals and their X-ray Identification (ed.
G. W. Brindley and G. Brown), pp. 197-248. Mineralogical Society.
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Mahieu N., Powlson D. S., and Randall E. W. (1999) Statistical analysis of published
carbon-13 CPMAS NMR spectra of soil organic matter. Soil Sci. Soc. Am. J. 63, 307-319.
Marshall C. E. (1935) Layer lattices and the base exchange clays. Z. Krystallogr. 91, 433-
449.
Martin Martinez F. and Perez Rodriguez J. L. (1969) Interlamellar adsorption of a
blackearth humic acid on Na-montmorillonite. Z. Pflanzenerna.hr. Bodenk. 124, 52-57.
McKinley J. P., Zeissler C. J., Zachara J. M., Seme R. J., Lindstrom R. M., Schaef H. T.,
and Orr R. D. (2001) Distribution and retention of 137Cs in sediments at the Hanford
Site, Washington. Environ. Sci. Technol. 35, 3433-3441.
Meredith J. A. (1997) Soil organic matter: Does physical or chemical stabilization
predominate? In Humic Substances in Soils, Peats and Waters: Health and
Environmental Aspects (ed. M. H. B. Hayes and W. S. Wilson), pp. 121-135. Royal
Society of Chemistry.
Metropolis N., Rosenbluth A. W., Rosenbluth M. N., Teller A. H., and Teller E. (1953)
Equations of state calculations by fast computing machines. J. Chem. Phys. 21, 1087-
1092.
Mirabella A., Piccolo A., and Pietramellara G. (1996) Intercalation between a well-
characterized Andisol fulvic acid and montmorillonite. Fresenius Environ. Bull. 5, 430-
435.
Monreal C. M., Schulten H.-R., and Kodama H. (1997) Age, turnover and molecular
diversity of soil organic matter in aggregates of a Gleysol. Can. J. Soil Sci. 77, 379-388.
Mooney R. W., Keenan A. G., and Wood L. A. (1952) Adsorption of water vapor by
montmorillonite. II. Effect of exchangeable ions and lattice swelling as measured by X-
ray diffraction. J. Am. Chem. Soc. 74, 1371-1374.
Moore D. M. and Robert C Reynolds J. (1997) X-ray Diffraction and the Identification
and Analysis o f Clay Minerals. Oxford University Press.
Mortland M. M. (1986) Mechanisms of adsorption of nonhumic organic species by clays.
In Interactions o f Soil Minerals with Natural Organics and Microbes (ed. P. M. Huang
and M. Schnitzer), pp. 59-76. Soil Science Society of America, Inc.
Muneer M. and Oades J. M. (1989a) The role of Ca-organic interactions in soil aggregate
stability. III. Mechanisms and models. Aust. J. Soil Res. 27,411-423.
Muneer M. and Oades J. M. (1989b) The role of Ca-organic interactions in soil aggregate
stability. I. Laboratory studies with 14C-glucose, CaC0 3 and CaS04 2 H2O. Aust. J. Soil
Res. 27, 389-399.
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Muneer M. and Oades J. M. (1989c) The role of Ca-organic interactions in soil aggregate
stability. II. Field studies with 14C-labelled straw, CaC0 3 and CaSCL 2 H2 O. Aust. J. Soil
Res. 27, 401-409.
Muurinen A., Pentti'la-Hiltunen P., and Rantanen J. (1987) Diffusion mechanisms of
strontium and cesium in compacted sodium bentonite. In Scientific Basis f o r Nuclear
Waste Management X{ed. J. K. Bates and W. B. Seefeldt), pp. 803-812. Materials
Research Society.
Nakano M., Kawamura K., and Ichikawa Y. (2003) Local structural information of Cs in
smectite hydrates by means of an EXAFS study and molecular dynamics simulations.
Appl. Clay Sci. 23, 15-23.
Nayak D. C., Varadachari C., and Ghosh K. (1990) Influence of organic acidic functional
groups of humic substances in complexation with clay minerals. Soil Sci. 149, 268-271.
Nelson P. N., Baldock J. A., Clarke P., Oades J. M., and Churchman G. J. (1999)
Dispersed clay and organic matter in soil: Their nature and associations. Aust. J. Soil Res.
37, 289-315.
Nye P. H. (1979) Diffusion of ions and uncharged solutes in soils and soil clays. Adv.
Agron. 31, 225-272.
Oades J. M. (1988) The retention of organic matter in soils. Biogeochem. 5, 35-70.
Oades J. M., Vassallo A. M., Waters A. G., and Wilson M. A. (1987) Characterization of
organic matter in particle size and density fractions from a red-brown earth by solid-state
13C N.M.R. Aust. J. Soil. Res. 25, 71-82.
O'Day P. A., Rehr J. J., Zabinsky S. I., and Brown G. E. J. (1994) Extended x-ray
absorption fine structure (EXAFS) analysis of disorder and multiple-scattering in
complex crystalline solids. J. Am. Chem. Soc. 116, 2938-2949.
Ohtaki H. and Radnai T. (1993) Structure and dynamics of hydrated ions. Chem. Rev. 93,
1157-1204.
Onodera Y., Iwasaki T., Ebina T., Flayashi H., Torii K., Chatterjee A., and Mimura H.
(1998) Effect of layer charge on fixation of cesium ions in smectites. J. Contam. Hydrol.
35, 131-140.
Oscarson D. W., Hume H. B., and King F. (1994) Sorption of cesium on compacted
bentonite. Clays Clay Miner. 42, 731-736.
Paasikallio A. (1999) Effect of biotite, zeolite, heavy clay, bentonite and apatite on the
uptake of radiocesium by grass from peat soil. Plant Soil 206, 213-222.
Parfitt R. L., Theng B. K. G., Whitton J. S., and Shepherd T. G. (1997) Effects of clay
minerals and land use on organic matter pools. Geoderma 75, 1-12.
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Park S.-H. and Sposito G. (in press) Molecular modeling of clay mineral structure and
surface chemistry. In Layered Materials f o r Catalytic and Other Applications (ed. S. A.
Auerbach, K. A. Carrado, and P. K. Dutta), Marcel Dekker, Inc.
Pauling L. (1930) The structure of micas and related minerals. Proc. Natl. Acad. Sci. USA
16, 123-129.
Piccolo A. and Conte P. (1998) Advances in nuclear magnetic resonance and infrared
spectroscopies of soil organic particles. In Structure and Surface Reactions o f Soil
Particles (ed. P. M. Huang, N. Senesi, and J. Buffle), pp. 183-250. John Wiley & Sons,
Ltd.
Pinck L. A. (1962) Adsorption of proteins, enzymes, and antibiotics by montmorillonite.
Clays Clay Miner. 9, 520-529.
Prost R. (1975) Study of the hydration of clays: Water-mineral interactions and
mechanism of water retention. II. Study of a smectite (hectorite). Ann. Agrono. 26, 463-
535.
Prost R. (1976) Interactions between adsorbed water molecules and the structure of clay
minerals: Hydration mechanism of smectites. In Proceedings o f the International Clay
Conference, 1975, Mexico City (ed. S. W. Bailey), pp. 353. Applied Publishing Ltd.
Pushkareva R., Kalinichenko E., Lytovchenko A., Pushkarev A., Kadochnikov V., and
Plastynina M. (2002) Irradiation effect on physico-chemical properties of clay minerals.
Appl. Clay Sci. 21,117-123.
Quine T. A. (1999) Use of caesium-137 data for validation of spatially distributed erosion
models: the implications of tillage erosion. Catena 37, 415-430.
Righi D., Dinel H., Schulten H.-R., and Schnitzer M. (1995) Characterization of clay-
organic-matter complexes resistant to oxidation by peroxide. Euro. J. Soil Sci. 46,423-
429.
Ritchie J. C. and Rasmussen P. E. (2000) Application of 137cesium to estimate erosion
rates for understanding soil carbon loss on long-term experiments at Pendleton, Oregon.
Land Degr. Devel. 11, 75-81.
Rutherford D. W., Chiou C. T., and Eberl D. D. (1997) Effects of exchanged cation on
the microporosity of montmorillonite. Clays Clay Miner. 45, 534-543.
Saggar S., Parshotam A., Sparling G. P., Feltham C. W., and Hart P. B. S. (1996) 14C-
labelled ryegrass turnover and residence time in soils varying in clay content and
mineralogy. Soil Biol. Biochem. 28, 1677-1686.
Sainz-Diaz C. I., Hemandez-Laguna A., and Dove M. T. (2001a) Theoretical modelling
of c'E-vacant and trans- vacant configurations in the octahedral sheet of illites and
smectites. Phys. Chem. Miner. 28, 322-331.
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Sainz-Diaz C. I., Hemandez-Laguna A., and Dove M. T. (2001b) Modeling of
dioctahedral 2:1 phyllosilicates by means of transferable empirical potentials. Phys.
Chem. Miner. 28, 130-141.
Sanchez A. L., Smolders E., van den Brande K., Merckx R., Wright S. M., and Naylor C.
(2002) Predictions of in situ solid/liquid distribution of radiocaesium in soils. J. Environ.
Radioact. 63, 35-47.
Sato H., Yamagishi A., and Kato S. (1992) Theoretical study on the interactions between
a metal chelate and a clay - Monte-Carlo simulations. J. Phys. Chem. 96, 9377-9382.
Satoh T. and Yamane I. (1971) On the interlamellar complex between montmorillonite
and organic substance in certain soil. Soil Sci. Plant Nutr. 17, 181-185.
Scharpenseel H. W. (1966) Tracer investigations on synthesis and radiometric
combination of soil organo-mineral complexes. Soil Chemistry and Fertility: Meeting o f
Commissions II and IV o f the International Society o f Soil Science, 41-52.
Schlesinger W. H. (1997) Biogeochemistry: An Analysis o f Global Change, 2nd Ed.
Academic Press.
Schnitzer M. and Khan S. U. (1972) Humic Substances in the Environment. Marcel
Dekker, Inc.
Schnitzer M. and Kodama H. (1966) Montmorillonite: Effect of pH on its adsorption of a
soil humic compound. Science 153, 70-71.
Schnitzer M., Ripmeester J. A., and Kodama H. (1988) Characterization of the organic
matter associated with a soil clay. Soil Sci. 145, 448-454.
Schroeder P. A. (1992) Far-infrared study of the interlayer torsional-vibrational mode of
mixed-layer illite/smectites. Clays Clay Miner. 40, 81-91.
Schroth B. K. and Sposito G. (1997) Surface charge properties of kaolinite. Clays Clay
Miner. 45, 85-91.
Schulten H.-R. (1999) Analytical pyrolysis and computational chemistry of aquatic
humic substances and dissolved organic matter. J. Anal. Appl. Pyrolysis 49, 385-415.
Schulten H.-R. and Leinweber P. (1996) Characterization of humic and soil particles by
analytical pyrolysis and computer modeling. J. Anal. Appl. Pyrolysis 38, 1-53.
Schulten H.-R. and Leinweber P. (2000) New insights into organic-mineral particles:
Composition, properties and models of molecular structure. Biol. Fertil. Soils 30, 399-
432.
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Schulten H.-R., Leinweber P., and Theng B. K. G. (1996) Characterization of organic
matter in an interlayer clay-organic complex from soil by pyrolysis methylation-mass
spectrometry. Geoderma 69, 105-118.
Shaban I. S. and Mikulaj V. (1996) Sorption-desorption of radiocesium on various
sorbents in presence of humic acid. J. Radioanal. Nucl. Chem. 208, 593-603.
Shahwan T. and Erten H. N. (2002) Thermodynamic parameters of Cs+sorption on
natural clays. J. Radioanal. Nucl. Chem. 253, 115-120.
Showstack R. (2003) Agricultural sequestration called useful stop-gap mitigation
measure for reducing atmospheric carbon. Eos, Trans., Am. Geophys. Union 84, 270.
Singer A. and Huang P. M. (1988) Thermal analysis of AlOH
polymer/montmorillonite/humic acid complexes. Thermochim. Acta 135, 307-312.
Skipper N. T., Chang F.-R. C., and Sposito G. (1995) Monte Carlo simulation of
interlayer molecular structure in swelling clay minerals. 1. Methodology. Clays Clay
Miner. 43, 285-293.
Skipper N. T., Refson K., and McConnell J. D. C. (1991) Computer simulation of
interlayer water in 2:1 clays. J. Chem. Phys. 94, 7434-7445.
Smith D. E. (1998) Molecular computer simulations of the swelling properties and
interlayer structure of cesium montmorillonite. Langmuir 14, 5959-5967.
Sokoloff V. P. (1938) Effect of neutral salts of sodium and calcium on carbon and
nitrogen of soils. J. Agr. Res. 57, 201-216.
Solomon D., Lehmann J., and Zech W. (2000) Land use effects on soil organic matter
properties of chromic luvisols in semi-arid northern Tanzania: Carbon, nitrogen, lignin
and carbohydrates. Agr. Ecosys. Env. 78, 203-213.
Specht C. H., Kumke M. U., and Frimmel F. H. (2000) Characterization of NOM
adsorption to clay minerals by size exclusion chromatography. Water Res. 34, 4063-
4069.
Sposito G. (1989) The Chemistry o f Soils. Oxford University Press.
Sposito G., Skipper N. T., Sutton R., Park S.-H., Soper A. K., and Greathouse J. A.
(1999) Surface geochemistry of the clay minerals. Proc. Natl. Acad. Sci. USA 96, 3358-
3364.
Staunton S. and Roubaud M. (1997) Adsorption of 137Cs on montmorillonite and illite:
Effect of charge compensating cation, ionic strength, concentration of Cs, K and fulvic
acid. Clays Clay Miner. 45, 251-260.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Stevenson F. J. (1994) Humus Chemistry: Genesis, Composition, Reactions, 2nd Ed. John
Wiley & Sons, Ltd.
Tan K. H. (1976) Complex formation between humic acid and clays as revealted by gel
filtration and infrared spectroscopy. Soil Biol. Biochem. 8, 235-239.
Teppen B. J., Rasmussen K., Bertsch P. M., Miller D. M., and Schafer L. (1997)
Molecular dynamics modeling of clay minerals. 1. Gibbsite, kaolinite, pyrophyllite, and
beidellite. J. Phys. Chem. B 101, 1579-1587.
Theng B. K. G. (1976) Interactions between montmorillonite and fulvic acid. Geoderma
15, 243-251.
Theng B. K. G. (1979) Formation and Properties o f Clay-Polymer Complexes. Elsevier
Scientific Publishing Company.
Theng B. K. G. (1982) Clay-polymer interactions: Summary and perspectives. Clays
Clay Miner. 30, 1-10.
Theng B. K. G., Churchman G. J., and Newman R. H. (1986) The occurrence of
interlayer clay-organic complexes in two New Zealand soils. Soil Sci. 142, 262-266.
Theng B. K. G. and Scharpenseel H. W. (1975) The adsorption of 14C-labelled humic
acid by montmorillonite. International Clay Conference, 643-653.
Theng B. K. G., Tate K. R., and Becker-Heidmann P. (1992) Towards establishing the
age, location, and identity of the inert soil organic matter of a spodosol. Z.
Pflanzenernahr. Bodenk. 155, 181-184.
Thurman E. M. (1986) Organic Geochemistry o f Natural Waters. Nijhoff/Junk
Publishers.
Tsai S.-C., Ouyang S., and Hsu C.-N. (2001) Sorption and diffusion behavior of Cs and
Sr on Jih-Hsing bentonite. Appl. Radiat. Isot. 54, 209-215.
UNFCCC. (2002) United Nations Framework Convention on Climate Change. United
Nations, accessed 12 February 2004 from:
http ://unfccc. int/resource/docs/convkp/conveng.pdf.
UNFCCC. (2003) Kyoto Protocol Status o f Ratification. United Nations, accessed 17
February 2004 from: http://unfccc.int/resource/kpstats.pdf
Vanden Bygaart A. J. and Protz R. (2001) Bomb-fallout 137Cs as a marker of geomorphic
stability in dune sands and soils, Pinery Provincial Park, Ontario, Canada. Earth Surf.
Process. Landforms 26, 689-700.
Van Oss C. J. and Giese R. F. (1995) The hydrophilicity and hydrophobicity of clay
minerals. Clays Clay Miner. 43, 474-477.
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Varadachari C., Mondal A. H., and Ghosh K. (1991) Some aspects of clay-humus
complexation: Effect of exchangeable cations and lattice charge. Soil Sci. 151, 220-227.
Varadachari C., Mondal A. H., and Ghosh K. (1995) The influence of crystal edges on
clay humus complexation. Soil Sci. 159, 185-190.
Varadachari C., Mondal A. H., Nayak D. C., and Ghosh K. (1994) Clay-humus
complexation: Effect of pH and the nature of bonding. Soil Biol. Biochem. 26, 1145-
1149.
Velde B. and Couty R. (1985) Far infrared spectra of hydrous layer silicates. Phys. Chem.
Miner. 12, 347-352.
Violante A., Arienzo M., Sannino F., Colombo C., Piccolo A., and Gianfreda L. (1999)
Formation and characterization of OH-Al-humate-montmorillonite complexes. Org.
Geochem. 30, 461-468.
Wallbrink P. J., Roddy B. P., and Olley J. M. (2002) A tracer budget quantifying soil
redistribution on hillslopes after forest harvesting. Catena 47, 179-201.
Walther G.-R., Post E., Convey P., Menzel A., Parmesan C., Beebee T. J. C., Fromentin
J.-M., Hoegh-Guldberg O., and Bairlein F. (2002) Ecological responses to recent climate
change. Nature 416, 389-395.
Wanner H., Albinsson Y., and Wieland E. (1996) A thermodynamic surface model for
caesium sorption on bentonite. Fresenius J. Anal. Chem. 354, 763-769.
Wattel-Koekkoek E. J. W., Buurman P., van der Plicht J., Wattel E., and van Breemen N.
(2003) Mean residence time of soil organic matter associated with kaolinite and smectite.
Euro. J. Soil Sci. 54, 269-278.
Wattel-Koekkoek E. J. W., van Genuchten P. P. L., Buurman P., and van Lagen B.
(2001) Amount and composition of clay-associated soil organic matter in a range of
kaolinitic and smectitic soils. Geoderma 99, 27-49.
Weiss C. A., Kirkpatrick R. J., and Altaner S. P. (1990a) The structural environments of
cations adsorbed onto clays - 133Cs variable-temperature MAS NMR spectroscopic study
ofhectorite. Geochim. Cosmochim. Acta 54, 1655-1669.
Weiss C. A., Kirkpatrick R. J., and Altaner S. P. (1990b) Variations in interlayer cation
sites of clay minerals as studied by 133Cs MAS nuclear magnetic resonance spectroscopy.
Am. Miner. 75, 970-982.
Winter M. (2003) WebElements. The University of Sheffield and WebElements Ltd.,
UK, accessed 27 January 2003 from: http://www.webelements.com.
Xu S. and Harsh J. B. (1992) Alkali cation selectivity and surface charge of 2:1 clay
minerals. Clays Clay Miner. 40, 567-574.
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Zech W., Senesi N., Guggenberger G., Kaiser K., Lehmann J., Miano T. M., Miltner A., -
and Schroth G. (1997) Factors controlling humification and mineralization of soil organic
matter in the tropics. Geoderma 79,117-161.
Zhu Y.-G. and Smolders E. (2000) Plant uptake of radiocaesium: A review of
mechanisms, regulation and application. J. Exper. Bot. 51, 1635-1645.
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 2
Monte Carlo and molecular dynamics studies of Cs-smectite hydrates
2.1 Introduction
Radioactive isotopes of Cs pose a significant environmental hazard because of
their long half-lives and high bioavailability (Evans et al., 1983; Cho et al., 1993; Smith,
1998; Winter, 2003). Both natural and engineered attenuation strategies for these
radionuclides make use of their adsorption and sequestration by the interlayer region of
smectite clay minerals (Evans et al., 1983; Jensen and Radke, 1988; Cho et al., 1993;
Oscarson et al., 1994; Iwasaki and Onodera, 1995; Onodera et al., 1998). In the typical
engineered scenario for nuclear waste containment (Cho et al., 1993; Chun et al., 2001),
compacted smectite-based backfill material is used as an environmental buffer separating
waste containers from a host rock formation. This use of smectite presupposes a very
low mobility of Cs+in the interlayer region after adsorption.
Smectites are expandable 2:1 layer type clay minerals containing negatively
charged isomorphic substitution sites within their octahedral and tetrahedral sheets
(Sposito, 1989). This negative mineral charge is balanced by cations (counterions)
adsorbed between the clay layers in the interlayer region. Hydration of the interlayer
region can cause the layer spacing to increase, hence the expandable nature of these clay
minerals. In addition, hydration may increase the mobility of interlayer ions by solvating
them, transforming them from inner to outer sphere complexes. In order for Cs+
sequestration to be effective, Cs+ adsorption in smectite interlayers must be accompanied
by a significant decrease in ionic mobility through fixation. As pointed out by Onodera
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
et al. (1998), the extent to which this desired effect is realized depends on the nature of
the layer charge distribution between the tetrahedral and octahedral sheets.
Although Cs-smectite hydration has been investigated for many years (MacEwan
and Wilson, 1980), a comprehensive understanding of the mechanisms involved remains
elusive. A number of XRD studies show that the layer spacing of Cs-smectites remains
in the range 11.9 - 12.5 A after exposure to water vapor at any relative humidity
(Mooney et al., 1952; Calvet, 1972; Prost, 1976; MacEwan and Wilson, 1980; Berend et
al., 1995; Chiou and Rutherford, 1997), or even after immersion in an aqueous solution
(MacEwan and Wilson, 1980). Water vapor adsorption isotherms (Mooney et al., 1952;
Calvet, 1972; Prost, 1976; Berend et al., 1995; Chiou and Rutherford, 1997) indicate an
increasing hydration of Cs-smectite with increasing relative humidity, but Calvet (1972)
and Prost (1976) have argued that most of the water adsorbed resides in micropores, not
in the interlayer region. Recent measurements of the micropore volume in homoionic
montmorillonites (Chiou and Rutherford, 1997; Rutherford et al., 1997) have confirmed
the extensive microporosity of Cs-smectites. Calvet (1972) and Prost (1975) suggested
that the 12.4 A layer spacing in Cs-smectite was achieved and stabilized after the
adsorption of only about 1.2 - 1.4 H2 O per unit cell of the clay mineral, well below the
nominal monolayer water content of about 4 H2O per unit cell observed for smectites
containing the strongly-hydrating Li+counterion (Calvet, 1972; Berend et al., 1995).
Reduced water content within Cs-smectite interlayers can be rationalized given the large
size and low hydration energy of Cs+, resulting in a more crowded and less hydrophilic
interlayer region.
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Recent spectroscopic studies provide further insight regarding interlayer structure.
Studies of hydrated Cs-smectites using 133Cs NMR have been performed by Weiss et al.
(1990a, b) and Kim et al. (1995, 1996). The NMR spectra of a Cs-hectorite sample at
25% relative humidity (1990b), corresponding to a 12 A layer spacing, revealed two
peaks. The authors (Weiss et al., 1990b) attributed the narrower peak to motionally-
averaged Stem (inner and outer sphere) and Gouy (diffuse swarm) layer Cs+counterions
within hydrated interlayers, and the broader peak to adsorbed Cs+ in dehydrated
interlayers. This conclusion is based on the separation of the narrow spectral peak into
two peaks at lower water content or lower temperature (Weiss et al., 1990a, b). Spectra
of Cs-beidellite at room humidity and under all other conditions examined (Weiss et al.,
1990b) revealed two peaks, thought to result from two Cs+species which were prevented
from motional averaging by compositional heterogeneity occurring on large spatial scales
within the sample. The NMR spectra of several Cs-montmorillonite samples did not
yield consistent results for adsorbed Cs+ speciation at 25% relative humidity (Weiss et
al., 1990b; Kim et al., 1995, 1996). One montmorillonite sample produced a spectrum
very similar to those of Cs-hectorite (Weiss et al., 1990b), suggesting the possible
presence of hydrated and dehydrated interlayer Cs+ species. Other montmorillonite
samples produced spectra with just one interlayer species peak which, on examination,
broadened at either lower temperature or water content, but did not split into two separate
peaks as observed with Cs-hectorite (Kim et al., 1995, 1996). The authors (Kim et al.,
1995,1996) hypothesized that paramagnetic broadening blurred the clear separation of
one motionally-averaged peak into the two separate peaks seen for Cs-hectorite hydrates.
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Informed by the NMR analyses described above, Bostick et al. (2002) recently
interpreted Lra-edge EXAFS spectra of Cs-montmorillonite to suggest the presence of
two types of Cs+ surface complex, distinguished by average Cs-0 distances of 3.2 A and
4.25 A, respectively. They consider the complex with the shorter Cs-O distance to be a
fully hydrated outer sphere complex, directly coordinated by 2 - 3 water molecules,
while the complex with the longer Cs-0 distance is thought to be an inner sphere
complex, partially dehydrated and directly associated with 7 - 1 3 0 atoms of the mineral
surface. Nakano et al. (2003) examined Cs+ adsorption to bentonite pastes over a range
of pH conditions using K-edge EXAFS spectroscopy. The resulting spectra showed no
significant variation with pH, and indicated that the first shell of O atoms was located
3.15 - 3.16 A away from Cs+, in agreement with Bostick et al. (2002). Although Nakano
et al. (2003) attributed these O to water molecules, as did Bostick et al. (2002), they
suggest a CN of 7.0 - 7.4 for this outer sphere complex. The greater CN calculated by
Nakano et al. (2003) may be influenced by use of large Debye-Waller constants
(~0.04 A2, as opposed to the value of 0.015 A2used by Bostick et al. (2002)) during
spectral fitting, as these constants are highly correlated to CN (ODay et al., 1994). A
second coordination shell was detected 3.59 - 3.62 A from the hydrated cation, and
featured 5.7 - 6.1 O, identified as members of a six-oxygen cavity on the mineral surface
(Nakano et al., 2003). Both analyses suggest that in a fully-saturated Cs-montmorillonite,
the majority of cations exist as outer sphere complexes. An abundance of outer sphere
complexes within Cs-smectite interlayers would appear inconsistent with the low
hydration energy of the cation, resulting from its large size and low charge. A meticulous
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
study of Cs-smectite at the molecular scale is needed in order to examine issues
surrounding water content and cation adsorption in the interlayer region.
Molecular simulations based on Monte Carlo (MC) and molecular dynamics
(MD) techniques have proven to be a useful adjunct to experimental studies of smectite
interlayer hydration (Sposito et al., 1999a). A network of scientists, known informally as
the Sposito simulation team, have developed software and methodologies designed to
produce realistic models of smectite interlayer structure and dynamics consistent with
experimental measurements. Most simulations conducted by team members utilize MC
and/or MD algorithms, take place under standard temperature and pressure conditions,
and feature varying numbers of rigid water molecules described by the Matsuoka-
Clementi-Yoshimine (MCY) water potential function (Matsuoka et al., 1976). The
simulation cells consist of a single clay interlayer bound on opposing sides by 8
crystalline units of a rigid smectite mineral (Skipper et al., 1995a). Typical MC
calculations are performed using an isothermal-isobaric (NaT) ensemble, which
maintains a constant number of molecules, pressure normal to the clay surface (a), and
temperature, while typical MD calculations are performed using a microcanonical (NVE)
ensemble, which maintains a constant number of molecules, cell volume, and energy.
Simulations of smectite interlayers containing the monovalent counterions Li+,
Na+, and K+, and the divalent counterions Mg2+ and Ca2+, reproduce experimentally
determined values and trends for bulk properties such as layer spacing and diffusion
coefficients (Skipper et al., 1995a, b; Chang et al., 1995, 1997, 1998, 1999; Greathouse
and Sposito, 1998; Greathouse et al., 2000; Greathouse and Storm, 2002). Examination
of simulated interlayer water indicates it has a layered structure (Skipper et al., 1995b;
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chang et al., 1995, 1997, 1998, 1999) that is substantially altered from that of bulk water
for those one or two layers closest to the mineral surfaces (Sposito et al., 1999b; Park and
Sposito, 2000, 2002), as suggested by neutron diffraction data (Powell et al., 1997, 1998).
Many water molecules in these model interlayer systems adopt orientations reflecting a
balance between the electrostatic fields of the mineral layers, such that the hydrogen pairs
are directed towards opposing clay layers, while other water molecules are oriented so as
to allow H-bonding interactions with structural hydroxyl groups (Skipper et al., 1995b;
Chang et al., 1995,1997). Model water molecules within clay interlayers have reduced
coefficients of diffusion as compared to bulk water (Chang et al., 1995, 1997, 1998,
1999), consistent with experimental measurements (Cebula et al., 1981; Tuck et al.,
1985).
The organization of water into partial or full hydration spheres is determined by
the hydration energy of the interlayer cations, which scales with their charge to radius
ratios. Li+ and Na+are thus capable of creating fully hydrated, outer sphere complexes
with the clay surface near recessed, octahedral mineral charge sites (Skipper et al., 1995b;
Chang et al., 1995, 1997, 1999), while K+ is more likely to form only partially hydrated,
inner sphere complexes under the same conditions (Chang et al., 1998, 1999; Sposito et
al., 1999a, b). All three ions tend to form inner sphere complexes near tetrahedral charge
sites located just below the surface plane of oxygen atoms (Skipper et al., 1995b; Chang
et al., 1995,1997, 1998, 1999). Though strongly attracted to charge sites, these ions
often experience diffusional motion over the timescale of MD simulations, with
associated coefficients of diffusion for ions and water molecules increasing with
increasing water content (Chang et al., 1995,1997, 1998, 1999). Further simulations of
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Li-hectorite, -beidellite, and -montmorillonite with low water contents (3 H2 O / Li+)
showed no signs of diffusional motion over the 200 ps MD period for both cations and
water molecules, though outer and inner sphere complexes formed near octahedral and
tetrahedral charge sites, respectively (Greathouse and Sposito, 1998). Model interlayer
systems containing divalent cations showed different patterns of complexation, with
Mg2+ forming outer sphere complexes near the tetrahedral charge sites of a beidellite
2 b
interlayer containing two layers of water molecules (Greathouse et al., 2000), while Ca
within montmorillonite transitioned from inner sphere to outer sphere complexation as
water content increased from one to two layers (Greathouse and Storm, 2002).
Monte Carlo and MD work by researchers using alternative software packages,
ensembles, potential functions, and simulation cells shows general agreement with both
experimental work and the modeling results summarized above, including layer spacing
values and trends associated with counterion (Boek et al., 1995a, b; Karabomi et al.,
1996; Teppen et al., 1997; Chavez-Paez et al., 2001a, b; Marry et al., 2002; Cygan,
2003), the step-wise expansion of clay layers with adsorption of water molecules (Boek
et al., 1995a, b; Karabomi et al., 1996; Chavez-Paez et al., 2001a, b; Hensen et al., 2001;
Hensen and Smit, 2002; Marry et al., 2002), reduced diffusion of cations and water
molecules relative to bulk solution values (Marry et al., 2002), interlayer water structure
(Delville, 1991, 1992; Delville and Sokolowski, 1993; Boek et al., 1995a, b; Karabomi et
al., 1996; Chavez-Paez et al., 2001a, b; Hensen et al., 2001; Hensen and Smit, 2002;
Marry et al., 2002; Boek and Sprik, 2003) and orientation (Boek et al., 1995a, b; Marry et
al., 2002), and the greater capacity of strongly hydrating cations like Li+and Na+to form
complete hydration spheres and outer sphere complexes with mineral surfaces, relative to
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
weakly hydrating cations like K+(Boek et al., 1995a, b; Chavez-Paez et al., 2001a, b;
Hensen et al., 2001; Marry et al., 2002). Recent publications describing flexible smectite
models (Teppen et al., 1997), rather than the rigid ones in common use, or application of
density functional theory-based MD to smectite systems (Boek and Sprik, 2003), a
quantum chemical modeling method which accounts for the motions of electrons, suggest
exciting future directions for clay simulation research.
Though the body of work outlined above indicates the utility of molecular
modeling for determination of structural and dynamic properties of smectite interlayer
systems, only recently has this tool been directed toward Cs-smectite hydrates. Smith
and co-workers (Smith, 1998; Shroll and Smith, 1999; Young and Smith, 2000) have
performed MD calculations to examine the effects of hydration on a few variations of the
Cs-montmorillonite system. Cs-montmorillonites containing solely octahedral charge
(Young and Smith, 2000), as well as 1/3 tetrahedral and 2/3 octahedral charge (Smith,
1998), were simulated under water contents ranging from about one-half of a monolayer
(16 water molecules per interlayer) to more than four monolayers using a constant
number of molecules, pressure, and temperature (NPT) ensemble. Another set of MD
simulations which held potential energy, volume, and temperature constant (pVT), while
varying the number of water molecules within the interlayer, was calculated using the
octahedrally charged montmorillonite (Shroll and Smith, 1999). Interlayer water
molecules were simulated with the extended simple point charge (SPC/E) water potential
function (Berendsen et al., 1987). The clay coordinates employed were the same as those
developed by Skipper et al. (1995a), but modifications to some of the electrostatic and
van der Waals parameters associated with clay atoms were made in order to match
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
structural hydroxyl parameters with those of the SPC/E water molecule. Two clay
interlayer regions, mirror-images of each other, were modeled within each simulation
cell. Simulations of dry Cs-montmorillonites required a special MC annealing procedure,
in which the systems were simulated at very high temperatures and then rapidly cooled,
in order to increase the efficiency of the search for global potential energy minima
(Smith, 1998).
The three sets of simulations yielded consistent results. Plots of layer spacing
versus number of water molecules residing within an interlayer region featured a plateau
at layer spacings of ~12.5 A, corresponding to ~40 water molecules per interlayer
(0.13 g H20 / g clay, or 1.3 water monolayer) (Smith, 1998; Shroll and Smith, 1999;
Young and Smith, 2000). Calculations of energetic quantities indicated a global minima
at the water content of the 12.5 A hydrate, as well as a local minima for a system
containing ~80 water molecules and a layer spacing of 15.7 - 16.3 A, identified as a two-
layer hydrate of Cs-montmorillonite (Smith, 1998; Shroll and Smith, 1999; Young and
Smith, 2000). The water content required to produce Cs-montmorillonites with layer
spacings near 12.5 A, according to these simulations, is higher than that measured for
smectites containing the highly hydrated Na+and Li+ ions (Calvet, 1972; Berend et al.,
1995). In addition, the energetic indications of a stable two-layer Cs-montmorillonite,
not found in nature, suggest that the model systems may not be realistic. The Cs+ within
the interlayer formed primarily inner sphere complexes with the clay surface, although
outer sphere complexes were also observed (Smith, 1998; Young and Smith, 2000).
Simulations of the dry clay revealed two stable states as predicted from experimental
studies (Berend et al., 1995), a symmetric configuration in which cations were
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
surrounded by opposing surface cavities, and an asymmetric arrangement in which
cations were enclosed by a cavity on one side, and adjacent to an oxygen triad on the
other (Smith, 1998).
Marry et al. (2002) performed a set of MC and MD simulations on a similar
octahedrally charged, SPC/E water hydrated, two layer Cs-montmorillonite system.
Monte Carlo (NPT) calculations at a variety of water contents pointed toward a stable Cs-
montmorillonite hydrate possessing a layer spacing of 12.7 A and containing 36 water
molecules in each interlayer region (0.11 g H2 O / g clay, or 1.1 water monolayer), nearly
as much as suggested by Smith and co-workers (Smith, 1998; Shroll and Smith, 1999;
Young and Smith, 2000). Structural analysis of the equilibrated interlayer region
indicated that Cs+ preferred positions close to clay surfaces, regardless of water content.
Total Cs-0 coordination numbers (CNs) were greater than those found in bulk solution
simulations, as mineral oxygens in combination with water oxygens occupied the
hydration sphere surrounding the ion. Examination of water structure indicated that
interlayer water was more disordered than that of simulated Na-montmorillonite or bulk
solution, and that many water molecules were oriented such that each hydrogen tended to
point toward opposite clay layers. Molecular dynamics (NVE) calculations, performed
on the MC-equilibrated systems to evaluate the mobility of water and ions, were used to
calculate diffusion coefficients of 3.9 x 10'10and 1.0 x 10'10m2 s'1, respectively, which
may be compared to the experimentally determined diffusion coefficients in bulk
solution, which for both species is 2 x 109 m2 s'1(Nye, 1979; Ohtaki and Radnai, 1993).
Nakano et al. (2003) simulated a Cs-beidellite system containing three layers of
water molecules using an MD (NPT, T = 293 K) algorithm and a custom-built force field
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
capable of modeling flexible mineral systems (Kawamura and Ichikawa, 2001). This
model Cs-beidellite featured a layer spacing of 19.367 A, a value much greater than any
found in nature. Both inner and outer sphere Cs+ roamed freely over the mineral surface
and throughout the interlayer region, showing no significant attraction for the tetrahedral
charge sites present. Cations were surrounded by an average of 6.79 water molecules, at
a mean distance of 3.12 A, and were located an average of 3.70 - 3.80 A away from 0.9
surface O atoms.
A key discrepancy between experimental analyses of Cs-smectite systems and the
modeling results outlined above is the stable water content appropriate to clay hydrates
with -12 A layer spacings. The mathematical description of water molecule behavior, or
the water potential function, used in these simulations may be to blame. Three water
potential functions are in common use for the modeling of clay interlayer regions. The
MCY water potential function (Matsuoka et al., 1976) is based on a rigid water structure
with an O-H bond distance of 0.9572 A, and a H-O-H angle of 104.52. The MCY water
molecule contains four interaction sites used to define the pair potential interactions
which govern its properties. Centered on each of the H atoms are sites with a charge of
0.71748e, capable of both electrostatic and van der Waals interactions. Centered on the
O atom is a van der Waals active site, while located 0.2677 A away from this point, along
the water axis and away from the Hs, is a point charge of-1.43496e. The location of this
charge site away from the center of the atom is thought to account for the effect of the
free electron pairs of water O. The MCY potential function was developed by fitting
quantum chemical calculations of the potential energy of numerous water dimer
configurations to a mathematical model, and as such may be considered ab initio, or
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
based on first principles of quantum chemistry. Perhaps due to its derivation from first
principles and water dimers, MCY water does not reproduce the density of bulk water at
standard temperature and pressure, a property derived from the synergistic interactions of
a larger number of molecules.
The four point transferable intermolecular potential (TIP4P) model of water
(Jorgensen et al., 1983) has many of the same basic properties as the MCY potential
function. The TIP4P water molecule is also a rigid, four interaction site model, and it has
the same geometry as the MCY water molecule, though the electrostatic charges
associated with H and O atoms are 0.52e and -1.04e, respectively. However, its
interaction parameters have been fitted so that simulations of TIP4P bulk water feature
properties such as density which match experimental measurements made at standard
temperature and pressure. This makes TIP4P an empirical potential function, unlike
MCY. The mathematical expression employed by TIP4P to calculate van der Waals
interactions is in the Lennard-Jones (LJ) 6-12 form, which is less accurate than that used
in the MCY expression, but has the advantage of being less computationally costly.
An even more abbreviated version of water may be found in the empirical SPC/E
potential function (Berendsen et al., 1987). The SPC/E water molecule is rigid, and
features O-H bond distances of 1.0 A and H-O-H bond angles of 109.47. The model
contains just three reaction sites, located at the centers of the three atoms, and the
electrostatic charges associated with H and O atoms are 0.4238e and -0.8476e,
respectively. By using only three reaction sites, and the easily calculated LJ expression
to represent van der Waals interactions, the SPC/E potential function is the least accurate
and least computationally demanding of the water models described above.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Each of these water potential functions is capable of reproducing experimental
values for bulk water energetics, H-bonded structure, and dynamics (Matsuoka et al.,
1976; Beveridge et al., 1983; Jorgensen et al., 1983; Berendsen et al., 1987; Robinson et
al., 1996; Soper et al., 1997). However, the advantages and drawbacks of each model
indicate they are not equally appropriate for particular modeling applications. Within
smectite interlayer systems, the MCY potential function has been shown to model
accurately experimental quantities such as layer spacing and interlayer species
configuration and mobility (Skipper et al., 1995a, b; Chang et al., 1997, 1999; Greathouse
and Sposito, 1998). Because the MCY potential function for the water-water interaction
is optimized on dimer structures (Matsuoka et al., 1976), it does not enforce the
tetrahedral configuration found in bulk water as fervently as do other empirical water
potential functions such as TIP4P and SPC/E. This fact may explain its enhanced ability
to model the constrained geometry of the interlayer region (Skipper et al., 1995a, b;
Chang et al., 1997; Chang et al., 1999). The improved accuracy of the potential energies
generated using the more rigorous MCY calculations compensates for its added
computational expense.
In this chapter, I present the results of MC and MD simulations of interlayer
behavior in three model Cs-smectite hydrates at low water contents. The three smectites
differed in respect to their layer charge distribution, allowing an examination of the effect
of this property on Cs+and water molecule interlayer configurations and mobility. Their
water content was varied in order to determine, within the context of MC simulation,
whether a stable 12.4 A Cs-smectite hydrate can exist with less than one water
monolayer, as proposed by Calvet (1972) and Prost (1975, 1976), and not investigated
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
thoroughly in previous studies (Smith, 1998; Shroll and Smith, 1999; Young and Smith,
2000; Marry et al., 2002; Nakano et al., 2003). Comparisons also were made with Cs+
mobility measurements in montmorillonite (Calvet, 1973; Jensen and Radke, 1988; Cho
et al., 1993; Oscarson et al., 1994) and with NMR- and EXAFS-based interpretations of
Cs+ interlayer speciation (Weiss et al., 1990a, b; Kim et al., 1995, 1996; Bostick et al.,
2002; Nakano et al., 2003) within smectite interlayers.
2.2 Methods
2.2.1 Model Cs-smectite Hydrates
Given the accuracy with which the MCY water model simulates the structure and
dynamics of water molecules found in the confined, high ionic strength solutions present
within smectite interlayers, this model was chosen to represent interlayer water
molecules. All other interactions between interlayer species, as well as between them
and atoms in the clay layers, were represented parametrically by a MCY-type potential
function (Matsuoka et al., 1976):
N N
U = I
i=l j>i
M i _ A . e B,*j +C e v?
1J 1J
[2.1]
where U is the potential energy of the system, i and j are sites on atoms within the
simulation cell, pj is the distance between sites i and j, q represents atom charge, and A,
B, C, and D are van der Waals interaction parameters. Van der Waals parameters for the
Cs-0 interaction were obtained through conversion of LJ (6-12) potential function
parameters taken from Smith and Dang (1994). I matched the two types of potential
function by constraining them to be equal for the r-value at which the potential energy
equals zero, and to be equal and have the same curvature for the r-value at which the
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
potential energy is a minimum (see Appendix A). When combined with the LJ
parameters from Smith and Dang (1994), these constraints provided the MCY Cs-0
interaction parameters: A = 51.0 kJ mol'1, B = 1.0850 A'1, C = 3.72 x 106 kJ mol1,
D = 4.2758 A'1. The steps required to insert the identity of Cs+ and its associated van der
Waals interaction parameters into the Monte Carlo program MONTE (Skipper, 1996)
may be found in Appendix B. Van der Waals interaction parameters for Cs-H, Cs-Al,
and Cs-Si atom pairs were assumed to be negligible.
Several model Cs-smectite hydrates were simulated in order to examine the
effects of charge site and hydration state on the behavior of inter layer species. The
simulation cells used consisted of two opposing halves of a rigid 2:1 clay layer
surrounding an interlayer region that contains rigid MCY water molecules and Cs+ions.
The unit cell formulae of the Cs-smectites simulated are:
Hectorite: CsojstSisKMgs 2sLio.75)02o(OH)4
Beidellite: Cso.7s[Si7 25Alo.75](Al4)02o(OH)4
Montmorillonite: Cso.75[Si7.75Alo.25](Al3.5Mgo.5)02o(OH)4
Hectorite is a trioctahedral 2:1 clay mineral, whereas montmorillonite and beidellite are
dioctahedral (Greathouse and Sposito, 1998). In addition, hectorite shows charge
substitution in the octahedral sheet only, whereas beidellite contains only tetrahedral
charge substitution. Montmorillonite exhibits both types of charge substitution, with
twice as much octahedral as tetrahedral layer charge. The simulation cells contained
eight unit cells, and therefore had the lateral (ab) dimensions 18.28 x 21.12 A2. Each half
of the clay mineral layer was 4.9 A thick. Movement of the clay layers along a direction
normal to their basal planes was permitted during MC simulation, making the c-axis
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
dimension variable. Mineral movement along a direction parallel to basal planes was
permitted as well, allowing changes in the registration of the clay layers to occur.
Three-dimensional periodic boundary conditions were applied to the simulation
cells (Allen and Tildesley, 1987). This eliminated potential artifacts due to small cell size
by replication of the cells in all three dimensions, resulting in an infinite stack of clay
layers. A real-space cutoff of 9 A was employed for the summation of short-range van
der Waals interactions, while the three-dimensional Ewald summation method with a
reciprocal space cutoff k of 3 A'1was used for the calculation of long-range electrostatic
interactions. The Cs-smectite hydrates were simulated at 300 K and 100 kPa normal
stress (a).
2.2.2 Monte Carlo Simulations
The program MONTE, written by Skipper and Refson (Skipper, 1996), was used
to run MC simulations on a variety of Cs-smectite hydrates. MONTE is a Fortran-77
computer program designed to perform Metropolis Monte Carlo simulations for systems
consisting of molecular liquids located between solid surfaces. The program simulates
arrays of rigid molecules, atoms, or ions under periodic boundary conditions, and
calculates interactions derived from continuous pairwise potential functions acting
between specific sites found on each chemical species. As such, MONTE is uniquely
suited for studies of the smectite interlayer region, though it may require substantial
modifications in order to model systems with more diverse requirements, such as those
containing flexible molecules.
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Using MONTE, Skipper and co-workers (Skipper et al., 1995a, b), Chang et al.
(1997, 1999), and Greathouse and Sposito (1998) have developed simulation
methodologies specifically tailored to explore the interlayer region of clay mineral
hydrates using the isothermal-isobaric (NoT) ensemble. This ensemble provides an
effective means of determining the appropriate layer spacing for a model system. A key
procedure for achieving an efficient MC equilibration of a smectite system is that of
phase-space sampling optimization, as developed by Chang et al. (1997,1999). This
optimization involves a preliminary set of MC steps in which only water molecules are
allowed to move, followed by a second set in which water molecule movements and
adjustments to the layer spacing (c-axis) are permitted. With this initial organization of
interlayer water structure and clay layer spacing, we can provide conditions which are
well-suited to the cations in their initial positions, and avoid situations under which
poorly structured water molecules force ions to flee the interlayer and become trapped in
unrealistic positions near smectite surfaces.
For these simulations, six Cs+were positioned initially at the midplane of the
interlayer region, and surrounded by a fixed number of randomly-placed and -oriented
water molecules. The initial layer spacing was 14 A, and several different water contents
were investigated. Phase-space sampling optimization consisted of 20,000 MC steps of
water-only movements, followed by another 20,000 steps in which both water molecules
could move and interlayer spacing could vary. Optimization calculations were neglected
for dehydrated Cs-smectite systems. After optimization, all molecules were allowed to
move in any direction, although the clay layer was permitted to move only one step for
roughly every five interlayer species movements.
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Information about total potential energy, layer spacing, orientation of water
molecules, density profiles of interlayer species, and radial distribution functions were
collected every 500 steps. Systems were considered equilibrated when convergence
profiles of total potential energy and layer spacing provided minimum average values.
The standard deviation of the average layer spacing was further required to be consistent
with the precision of available experimental data on layer spacing. Lack of equilibration,
especially after ten million steps, was considered an indication that the interlayer water
content modeled was not appropriate for the particular Cs-smectite system. Quantitative
results reported from these convergent simulations are averages of 500,000 MC steps.
The radial distribution functions [RDFs; denoted g(r)] provided by MONTE may be
defined as:
dna p = 47t^ g a p ( r) r 2dr [2.2]
where dnap is the average number of p species atoms within a spherical shell of radius r
and thickness dr centered on an a species atom (Sposito et al., 1999b). The program also
supplies running integrations of RDFs, which can be used to obtain CN information.
2.2.3 Molecular Dynamics Simulations
The program MOLDY v2.12g, written by Refson (Refson, 2000), was used to
perform MD simulations of the Cs-smectite hydrates which had converged under MC
simulation. MOLDY is a C program designed to perform MD calculations on solids and
liquids. The atoms, molecules, or ions that MOLDY models must be rigid, and are
examined using periodic boundary conditions, as in MONTE. Interactions are calculated
via pairwise potential functions and point-charge electrostatics. The Newton-Euler
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
equations of motion are integrated using a modified form of the Beeman algorithm (Allen
and Tildesley, 1987). The program may be compiled to run for serial or parallel
computing systems. MOLDY is able to model a diverse range of molecular systems
using a variety of computer systems, though the use of the rigid molecule approximation
makes it unsuitable for some applications.
Monte Carlo-equilibrated configurations in which the layer spacing was within
one standard deviation of its average value were used as initial coordinates for MD
simulations using the microcanonical (NVE) ensemble. Because the MC simulations
provide an equilibrium volume parameter, this ensemble is an efficient means of
assessing mobility of species within the system. The clay layers were fixed in position,
while initial velocities were assigned randomly to interlayer molecules following a
Maxwell-Boltzmann distribution. In order to stabilize the temperature at 300 K, the MD
simulations began with 20 ps of temperature scaling (Refson, 2000) using a 0.5 fs
timestep. Temperature scaling was performed for each molecular species individually,
with rotational and translational kinetic energies scaled separately. After initial scaling,
the MD simulations proceeded for another 800 ps.
System configurations were saved every 200 timesteps (0.1 ps) to provide the
trajectory data necessary to calculate interlayer species self-diffusion coefficients. The
self-diffusion coefficients (D) were determined using a three-dimensional Einstein
relation (Allen and Tildesley, 1987):
diffusing species (Allen and Tildesley, 1987), and t is the time elapsed during a MD
[2.3]
where ^r2^ is the center-of-mass mean-square displacement (MSD) averaged over all
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
simulation. The MSD utility within MOLDY (Refson, 2000) calculated the mean-square
displacement for each interlayer species using trajectory data. In a model diffusional
system, a plot of ^r2^ vs. t is linear with a slope of 6D. The smallest value of D that can
be determined from these 800 ps MD simulations is about 1O'11 m2 s'1, or the D required
for a water molecule to be displaced by the length of its radius during 400 ps of elapsed
time. Molecules that exhibit alternative types of motion on the timescale of the MD
simulation will not display a linear plot.
2.3 Results
2.3.1 Monte Carlo Simulations
Cs-smectite hydrates at several water contents were simulated using MC, but only
a few of these systems converged. Besides the clay hydrates described in Table 2.1, Cs-
smectites were modeled with 1.0 monolayer of water within the interlayer (4 H2 O per
unit cell) and with 1.31, 0.844, and 0.75 monolayer. These systems were simulated for
up to 10 million MC steps, but continued to oscillate between layer spacings > 12.4 A
that differed by 1 A or more. On the other hand, the systems that contained either 1/3 or
2/3 monolayer (11 or 21 H2 O per simulation cell) did equilibrate, producing layer
spacings of 12 - 13 A (Table 2.1). Cs-beidellite with 2/3 water monolayer was an
exception to this pattern; after more than 10 million MC steps, the layer spacing for this
hydrate stabilized reasonably well, while potential energy did not. The dehydrated clay
systems equilibrated rapidly at layer spacings of 12.08 0.10, 12.14 0.11, and
11.94 0.09 A for Cs-hectorite, -beidellite, and -montmorillonite, respectively. These
large layer spacings indicate that Cs+were not captured between cavities found at the
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
center of six-oxygen rings located on both clay surfaces. No attempts were made to force
Cs+into surface cavities, or to bias the registration of clay layers in order to sequester the
cations between opposing cavities.
Table 2.1. Layer Spacing and Molar Potential Energy of Water in Cs-smectite Hydrates
(MC Simulations)
Hydrate MC Stepsa Layer Spacing Potential Energy
(A) (kJ mol'1)
Interlayer Water
Density (g cm3)c
1/3 water monolayer
Cs-hectorite 106 12.37 0.08 -47.43 3.53 0.37
Cs-beidellited
4 x 106 12.31 0.10 -42.98 4.58
0.41
Cs-montmorillonite 106 12.46 0.09 -32.33 4.04 0.38
2/3 water monolayer
Cs-hectorite 3.0 x 106 12.41 0.07 -46.56 2.00
0.71
Cs-montmorillonite 3.5 x 106 12.68 0.10 -36.66 2.12
0.64
aMonte Carlo steps required for convergence.
b(Total potential energy of hydrate - total potential energy of clay mineral) -s- moles of
water per simulation cell (Skipper et al., 1995a, b).
cThe ratio of water mass per simulation cell to interlayer volume (not including the
volume inhabited by 6 Cs+).
dThis hydrate exhibited MC convergence only at the lower water content.
The average potential energy per mole of water [the difference between the total
potential energy per mole of the hydrated and dry clay, divided by the number of moles
of water per mole of simulation cells (Skipper et al., 1995a, b; Smith, 1998)] may be
compared to the MCY-simulated bulk water molar internal energy o f -35.4 0.2 kJ/mole
(Sposito et al., 1999b). It is noteworthy that interlayer water in the two montmorillonite
hydrates exhibits a potential energy similar to that in MCY bulk water. This implies that
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the interlayer water in montmorillonite is stabilized because it possesses a higher entropy
than the bulk liquid.
Interlayer O density profiles (Figure 2.1) indicate that most of the water O are
clustered near the midplane in a single peak, irrespective of the type of smectite or its
water content. In contrast, Smith (1998) and Marry et al. (2002) show two peaks of water
density near the midplane for Cs-montmorillonite with a 12 A layer spacing. This
segregation of water molecules into two layers may be a result of the tetrahedral
structural bias present in the water models used. Examination of the orientation of the
water dipole moment, defined by a vector originating from the water oxygen and
bisecting the H-O-H angle, relative to a vector normal to the clay surface (P(0)), indicates
differences among the clay mineral hydrates (Figure 2.2). Most of the water molecules in
Cs-hectorite and Cs-montmorillonite with 1/3 water monolayer are oriented so that their
dipole moments are parallel to the clay mineral surfaces [P(0) peak at 90 (Chang et al.,
1997)]. Some of the water molecules within Cs-hectorite tend to tilt as well so that their
oxygens point directly toward the clay surface [P(0) peak at 180], an orientation which
increases greatly in probability at the higher water content. Cs-beidellite with 1/3 water
monolayer displays a bimodal distribution, having P(0) peaks around 30 and 100. This
means that the water molecules position themselves so that one of their H atoms points
toward the clay mineral surface (Chang et al., 1997). In the case of Cs-montmorillonite
with 1/3 water monolayer, a much smaller fraction of water molecules orient their dipole
moments toward the clay mineral surface and, at 2/3 water monolayer, a bimodal
distribution develops much like that in the Cs-beidellite system, with peaks around 30
and 90.
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cs-hectorite (1/3)
Cs-beidellite (1/3)
Cs-montmorillonite (1/3)
Cs-hectorite (2/3)
Cs-montmorillonite (2/3)
C/3
a
<u
Q
a>
i
3 2 0 1 3 2 1
z(A)
Figure 2.1. Interlayer density profiles for water O. The z axis is normal to clay surfaces,
and the profiles are plotted with the interlayer midplane as the origin.
0.30
Cs-hectorite (1/3)
Cs-beidellite (1/3)
Cs-montmorillonite (1/3)
Cs-hectorite (2/3)
Cs-montmorillonite (2/3)
0.25
0.20
0.10
0.05
0.00
180 120 150 30 60 90 0
0
Figure 2.2. Interlayer distribution of the angle (0) measured clockwise between the
dipole moment vector of a water molecule and a line normal to the siloxane surface
nearest to the molecule (Allen and Tildesley, 1987; Chang et al., 1997). This distribution
[P (0)] is not normalized to the solid angle through division by sin 0.
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Interlayer water structure may be examined using RDFs for water 0 - 0 and 0-H
spatial correlations. These RDFs (Figures 2.3 and 2.4) illustrate disorder in the
configuration of water molecules in the interlayer region of Cs-smectite hydrates, as
compared to MCY bulk liquid water (MC simulation at 300 K, NVT ensemble). Based on
first peak positions, Cs-montmorillonite with 2/3 water monolayer shows the most bulk
water-like structure, followed by Cs-beidellite with 1/3 water monolayer. The first peaks
in goo(r) and goH(r) for the other three hydrates are shifted to larger values of r relative to
bulk water, and all of the interlayer water RDFs are far less smooth. The variation in
peak heights in the clay hydrates is a result of normalizations in the RDF equation.
Coordination numbers for water 0 - 0 and 0-H intermolecular pairs ranged from 0.2 to
2.0 and 0.04 to 0.6, respectively (Table 2.2), which are much lower than those for bulk
water [experimental values of 5.4 and 2.0, respectively (Soper et al., 1997)].
Counterion density profiles, presented in Figure 2.5, depict three trends in the
interlayer Cs+ distribution. The density profiles for Cs-hectorite with 1/3 or 2/3 water
monolayer reveal a high concentration of counterions at the midplane, whereas the
density profile for Cs-beidellite with 1/3 water monolayer is bimodal, showing Cs+
density peaks close to each clay layer. The Cs-montmorillonite profiles display
intermediate behavior: at 1/3 monolayer water content, there is a wide midplane density
peak and significant shoulder features near each clay layer under the beidellite peaks,
whereas at 2/3 water monolayer, there is an asymmetric density profile peaked at the
midplane but without shoulders.
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
Cs-hectorite (1/3)
ao
Cs-beidellite (1/3)
Cs-montmorillonite (1/3)
Cs-hectorite (2/3)
Cs-montmorillonite (2/3)
bulk water
5
4
3
2
1
0
5 6 3 4 1 2 0
r(A)
Figure 2.3. Radial distribution functions for interlayer water 0 - 0 correlations. The RDF
for MCY liquid water is shown in blue.
3.0
Cs-hectorite (1/3)
X
o
so
Cs-beidellite (1/3)
Cs-montmorillonite (1/3)
Cs-hectorite (2/3)
Cs-montmorillonite (2/3)
bulk water r \
2.5
2.0
1.5
1.0
0.5
0.0
3 5 6 1 2 4 0
r(A)
Figure 2.4. Radial distribution functions for interlayer water O-Fl correlations. The RDF
for MCY liquid water is shown in blue.
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2.2. Coordination Numbers (CN) for Interlayer Water Oxygens and Surrounding
Water Oxygens and Hydrogens
Hydrate OO
r(A)a CNb
OH
r(A>" CNb
1/3 water monolayer
Cs-hectorite 3.3 0.32 2.3 0.08
Cs-beidellite 3.3 0.29 2.8 0.15
Cs-montmorillonite 3.3 0.18 2.3 0.04
2/3 water monolayer
Cs-hectorite 3.4 0.81 2.5 0.28
Cs-montmorillonite 3.8 1.93 2.5 0.57
bulk water simulation
MCY waterc 3.4 5.2 2.5 2.1
aValue of r at the minimum in gooOO or goH(r) following its first peak or peak pair
(Figures 2.3 and 2.4).
bCalculated with the conventional formula (Greathouse and Sposito, 1998):
= gox( r ) r 2dr
where Nx is the number of water oxygens or hydrogens in a simulation cell of
volume V.
cSposito et al. (1999b).
Radial distribution functions for Cs-0 and Cs-H spatial correlations involving
both water molecules and clay mineral O and H (denoted by starred O or H) are shown in
Figures 2.6 and 2.7. The five hydrates, with H2 O/CS molar ratios of either 1.8 or 3.6,
feature Cs-O peaks at 3.0 to 3.1 A. This peak position is supported by XRD data on
concentrated CsF solutions, which provide an average interatomic distance of
3.10 + 0.03 A between water O and Cs+at a H2 O/CS molar ratio of 2.3 (Bertagnolli et al.,
1974). Despite this similarity, the RDFs for Cs-H spatial correlations differ considerably
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cs-hectorite (1/3)
Cs-beidellite (1/3)
Cs-montmorillonite (1/3)
Cs-hectorite (2/3)
Cs-montmorillonite (2/3)
J/5
s
Q
<u
0.4 0.6 0.8 1.0 - 0. 2 0.0 0.2 0.8 - 0.6 -0.4 1.0
z(A)
Figure 2.5. Interlayer density profiles for Cs+. The z axis is normal to clay surfaces, and
the profiles are plotted with the interlayer midplane as the origin.
among the three hydrates. The RDF for Cs-montmorillonite with 1/3 water monolayer
displays a shoulder at r = 2.8 A and a peak at r = 3.6 A, whereas that for Cs-
montmorillonite with 2/3 water monolayer contains only one broad peak centered at
r = 3.5 A. The RDF g . (r) for Cs-beidellite is unique, in that the first peak is centered
at r = 2.9 A, whereas for the other hydrates this peak occurs at r > 3 A. Both of the Cs-
hectorite RDFs have significant gCsH*(r) peaks at r = 5.1 A, quite unlike the other two
hydrates. Coordination numbers for oxygen atoms nearest to Cs+ are listed in Table 2.3.
Those for O in the clay minerals are near 6 for hectorite and montmorilIonite and near 3
for beidellite, while those for water O are near 2 or 4, depending on the water content.
Coordination numbers of 6 to 8 are typical for Cs-0 in aqueous solution (Ohtaki and
Radnai, 1993).
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
g
(
r
)

(
s
r
4
)
g
(
r
)

(
s
r
4
)
g
(
r
)

(
s
r
4
)
Cs-hectorite, 1/3 water monolayer
20
Cs-O
Cs-H
Cs-H*
6 4 5 3
r( A)
Cs-beidellite, 1/3 water monolayer
20
C s- 0
Cs
Cs-H*
Cs-O*
5 6 3 4
r( A)
Cs-montmorillonite, 1/3 water monolayer
20
Cs-O
Cs-H1
Cs-H*
6 5 3 4
Figure 2.6. Radial distribution functions
for Cs-water (Cs-O, Cs-H) and Cs-clay
(Cs-O*, Cs-H*) for the three smectites
with 1/3 water monolayer.
Cs-hectorite, 2/3 water monolayer
20
5 -
Cs-O
Cs-H
Cs-H*
6 5 3 4
r (A)
Cs-montmorillonite, 2/3 water monolayer
20
Cs-O
Cs-H
Cs-H"
6 4 5 3
(A)
Figure 2.7. Radial distribution functions
for Cs-water (Cs-O, Cs-H) and Cs-clay
(Cs-O*, Cs-H*) for the two smectites
with 2/3 water monolayer.
r (A)
105
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Table 2.3. Coordination Number (CN) for Nearest-Neighbor Oxygens around Interlayer
Cesium
Hydrate Cs-water O Cs-clay O* Total CN
____________________ p(A)a CNb p(A)a CNb
1/3 water monolayer
Cs-hectorite 4.4 2.32 4.0 6.03 8.4
Cs-beidellite 4.7 2.77 3.2 3.14 5.9
Cs-montmorillonite 4.3 2.70 4.1 6.22 8.9
2/3 water monolayer
Cs-hectorite 4.8 4.15 4.0 6.01 10.2
Cs-montmorillonite 4.0 4.27 4.2 6.45 10.7
aValue of r at the minimum in gcso(r) or gcso*(r) following its first peak (Figures 2.6 and
2.7).
bCalculated with the conventional formula (Greathouse and Sposito, 1998):
CN= 4^ gcs0(r)r3dr
where No is the number of oxygens (water O or clay O*) in a simulation cell of
volume V.
2.3.2 Molecular Dynamics Simulations
Self-diffusion coefficients for interlayer water molecules (Dw) moving over
400 ps elapsed time are summarized in Table 2.4. All values found were smaller than the
experimentally-determined Dw for bulk liquid water (Sposito, 1981), or the Dwfor MCY
water (Lie and Clementi, 1986). The plot of MSD vs. elapsed time for Cs-
montmorillonite with 1/3 water monolayer was nonlinear and, therefore, cannot provide a
value of Dw. Examination of the component MSDs (Figure 2.8) reveals water motion in
the y-direction to be diffusive, in that the MSD can be fitted to a straight line with a small
nonzero intercept from just pass zero to about 200 ps, but, beyond 200 ps and about
30 A2, the MSD decreases. Motion in the x-direction yields a linear MSD only for about
106
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
50 ps, then slows substantially thereafter, ending in a plateau at about 24 A2that indicates
entrapment. It appears that water molecules are able to diffuse more or less continuously
in this hydrate over distances less than 3 - 5 A, but that diffusion over larger distances is
hindered. Similar behavior has been found in MD simulations of water motions in the
two-layer hydrate of Mg-beidellite (Greathouse et al., 2000). (Motion in the z-direction
is negligible because of the constrained geometry of the interlayer.)
Table 2.4. Self-Diffusion Coefficients (Dw) for Interlayer Water in Cs-smectite Hydrates
Hydrate Water Content
(H20 monolayers)
Dwa
(10-9 m2 s'1)
Cs-hectorite 1/3 0.229 0.001
Cs-hectorite 2/3 0.685 0.006
Cs-beidellite 1/3 0.342 0.002
Cs-montmorillonite 2/3 1.168 0.006
Bulk liquid waterb 2.3C
Value calculated from linear regression of MSD and elapsed time (Equation 2.3), with
P = 0.05 confidence interval.
bMCY water yields Dw = 2 x 10'9 m2 s'1(Park, unpublished data).
"Sposito (1981).
Figure 2.9 displays trajectories (400 ps elapsed time) of Cs+ and a representative
water molecule in the ab plane (i.e., the clay mineral surface) for Cs-hectorite. While the
water molecule shows solvating, diffusive behavior, the Cs+do not display such wide-
ranging motions. Instead, the counterions tend to hover near the layer charge sites, and
occasionally hop from one attractive spot to another. This type of motion is known as
jump diffusion, and is differentiated from typical diffusion only by the amount of time a
subject in motion spends inhabiting particular sites relative to the time over which
107
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
80
total
60
40
20
0
200 400 100 300 0
Elapsed Time (ps)
Figure 2.8. Interlayer water MSD vs. elapsed time for Cs-montmorillonite with 1/3 water
monolayer. A 400 ps elapsed time plot required 800 ps of data.
movement is measured. Examined over time scales longer than 800 ps, the motions of
the Cs+ would display typical diffusional behavior, but at the time scale of these
simulations, it is possible to identify stationary and mobile intervals for individual
cations. Both H20 diffusion and Cs+jump diffusion occur more extensively in the wetter
clay system.
The behavior of Cs+ is consistent with strongly adsorbed counterions, as
suggested by the low values of experimentally determined apparent Cs+ self-diffusion
coefficients, typically in the range of 10'12to 10'13 mV1(Muurinen et al., 1987; Cho et
al., 1993; Wanner et al., 1996; Tsai et al., 2001; Cormenzana et al., 2003) for compacted
Cs-montmorillonites and -bentonites, or 6 x 1016m2 s'1for 12 A Cs-montmorillonite
containing little water (Calvet, 1973) [the bulk solution value being 2 x 10'9 m2 s'1(Nye,
108
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
Figure 2.9. Trajectory plots (xy plane, 400 ps) for Cs-hectorite with (a) 1/3 water
monolayer and (b) 2/3 water monolayer. Motions of Cs+ are shown in grey, while blue
depicts the trajectory of a representative water molecule. Light grey dots correspond to
surface O of one clay layer. The gridlines are spaced 2 A apart. If the trajectory of a
molecule exceeds the boundary of the simulation cell, it reappears on the opposite side of
the cell from whence it came (seen in b), an artifact of periodic boundary conditions.
109
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
1979), essentially the same as for H2 O (Ohtaki and Radnai, 1993)]. Figure 2.10, showing
Cs+trajectory data (ab plane, 400 ps) for Cs-beidellite and Cs-montmorillonite with 1/3
water monolayer, is consistent with this interpretation, although the counterions within
montmorillonite appear to be more mobile than those within hectorite or beidellite. This
latter difference is emphasized with a change in perspective from the ab plane to the be
plane. Figure 2.11 shows 400 ps Cs+ be plane trajectories for all three low water content
systems. In the octahedral-charge Cs-hectorite, the counterions prefer the midplane,
whereas in the tetrahedral-charge Cs-beidellite, they stay closer to the clay layers.
Similar behavior is evident in the MC counterion density profiles (Figure 2.6). The Cs+
in montmorillonite do not display intermediate behavior, but instead show an exaggerated
vertical movement within the interlayer region. Figure 2.12 reveals an interesting
development in Cs+ motions (400 ps) in the 2/3 water monolayer hydrate of
montmorillonite. Four Cs+, hovering near the two tetrahedral charge sites, have
extremely limited motions, while the other two Cs+ display more diffusional movement.
Plots of MSD vs. elapsed time for the two more mobile Cs+(data not shown) are
nonlinear, similar to those illustrated in Figure 2.8, and therefore may indicate
entrapment.
110
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
X (A)
X (A)
Figure 2.10. Trajectory plots (xy plane, 400 ps) for (a) Cs-beidellite with 1/3 water
monolayer and (b) Cs-montmorillonite with 1/3 water monolayer. Same color scheme as
in Figure 2.9.
Il l
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Cs-hectorite
N

M M
w
*
Y (A)
<
N
Cs-beidellite
1.00
0.75
Cs-montmorillonite
N
Figure 2.11. The be coordinates of Cs+ (400 ps MD simulation) in smectites with 1/3
water monolayer. Surrounding clay layers are located at 1.52, 1.44 A, and 1.54 A
along the z-axis for hectorite, beidellite, and montmorillonite. Beside plots are images
describing average locations of Cs+ in relation to clay layers. Orange spheres represent
Cs+, purple spheres O, white spheres H, grey spheres Si, and green spheres Al.
112
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
. ' / - 4
*
ilk
* :ii
X (A)
Figure 2.12. Molecular dynamics x-y trajectory plots (400 ps) for Cs-montmorillonite
with 2/3 water monolayer. Movements of four less mobile Cs+ are shown in grey, while
movements of two more mobile Cs+are shown in light and dark orange, respectively.
Black dots represent surface oxygens of one clay layer, and the gridlines are spaced 2 A
apart. If the trajectory of a molecule exceeds the boundary of the simulation cell, it
reappears on the opposite side of the cell from whence it came, an artifact of three-
dimensional periodic boundary conditions.
2.4 Discussion
2.4.1 Water Content o f 12.4 A Cs-smectites
Our MC simulation results are consistent with an interlayer water content of 1/3
monolayer (H20 / Cs+ = 1.8) for 12.4 A Cs-smectite hydrates (Table 2.1). The
monolayer water content data reported by Mooney et al. (1952), Calvet (1972), and Prost
(1976) for Cs-montmorillonite and Cs-hectorite with a 12.4 A layer spacing indicate
1.6 - 1.8 H20 per Cs+ after correction for non-interlayer water according to methods
suggested by the last two cited authors. Although the 2/3 monolayer hydrates of Cs-
113
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
hectorite and Cs-montmorillonite (H2 O / Cs+= 2.4) were found to be stable states by our
MC convergence criteria, the absence of such convergence for Cs-beidellite suggests that
octahedral charge substitution is a necessary condition for the existence of the wetter
hydrate.
That Smith and co-workers (Smith, 1998; Shroll and Smith, 1999; Young and
Smith, 2000) and Marry et al. (2002) were able to obtain stable Cs-montmorillonite
hydrates at higher water contents may indicate the inappropriateness of their methods for
the purpose of simulating a realistic interlayer water content. The SPC/E water model
used in these investigations contains a strong tetrahedral structural bias typical of
empirical water potentials designed using bulk water data (Berendsen et al., 1987), which
may be unsuitable for simulations under the constrained geometry of the clay interlayer
(Skipper et al., 1995a, b; Chang et al., 1997,1999). In addition, the MD simulations
performed by Smith and co-workers (Smith, 1998; Shroll and Smith, 1999; Young and
Smith, 2000) cannot provide data on equilibrium properties. In contrast, the MC
annealing simulations of dehydrated Cs-montmorillonite performed by Smith (1998)
achieved equilibrium positions in good agreement with experimental data (Berend et al.,
1995). This may be taken as further indication that the SPC/E water model is not capable
of realistic description of the properties of interlayer water molecules. The dry clay
simulations reported here equilibrated with larger layer spacings because the simulation
method was held consistent for all Cs-smectites, rather than tailored to the specific needs
of the dehydrated systems.
114
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
2.4.2 Molecular Structure o f the Interlayer
Water RDF data reflect the greatly distorted structure of interlayer water in Cs-
smectite hydrates, relative to bulk water simulations. The goo(r) and goH(r) peaks have
shifted to higher r values, contain significant shoulders at higher r values, or both, for Cs-
montmorillonite with 1/3 water monolayer and both Cs-hectorite systems, which
indicates the presence of longer H-bonds. Such dramatic changes have not been seen in
model smectite simulations containing different cations and more water molecules (e.g.
Sposito et al., 1999b). Plots of all the clay hydrates are much noisier than those of bulk
water, a result of the strained organization of water in the clay interlayer as well as the
reduced abundance of water molecules over which to average the RDF information. Cs-
montmorillonite with 2/3 water monolayer appears to contain molecules with a structure
closest to that of the bulk liquid, and presents features similar to those provided by Marry
et al. (2002).
The clay surface appears to have a significant influence on water structure
through its ability to affect water dipolar orientation. We see from Figure 2.2 that many
water molecules are oriented with their dipole moments parallel to the clay mineral
surfaces in the five hydrates (0 ~ 90), delicately balanced in the electrostatic fields
exerted by the two negatively-charged clay layers. Marry et al. (2002) also noted an
abundance of water molecules in similar orientations within their Cs-montmorillonite
simulations. It is likely that the water molecules residing at the midplane of the interlayer
have this parallel orientation. Water molecules closer to the clay surface are responsible
for the other peaks in Figure 2.2. Cs-hectorite with 1/3 water monolayer contains water
molecules oriented at 120 - 150, whereas Cs-hectorite with 2/3 water monolayer shows
115
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
orientation of 0 = 180. Water molecules oriented in the latter way are tilted so that their
oxygen atoms are close to the clay surface, attracted to the proton of the clay mineral OH
group that is found at the bottom of a surface cavity. In the trioctahedral hectorite, this
OH group is oriented perpendicularly to the surface of the clay mineral (see Figure 2.11).
With greater water content, more water molecules interact with the clay structure in this
way. In Cs-beidellite with 1/3 water monolayer, a significant number of water dipoles
oriented at 0 = 0 - 30 occur, with the positive end of the dipole pointed toward a clay
mineral surface. As a dioctahedral clay mineral, beidellite contains OH groups that lie
nearly parallel to the clay surface and, therefore, are farther away from and less attractive
to interlayer water molecules. Beidellite contains tetrahedral charge which is very close
to its surface, and it is for sites near this negative charge that both the dipolar water and
the interlayer Cs+compete. In Cs-montmorillonite with 1/3 or 2/3 water monolayer, P(0)
peaks at or near 30 also occur. Like beidellite, montmorillonite is dioctahedral and,
therefore, also contains OH groups directed almost parallel to the mineral surface.
Montmorillonite contains some tetrahedral charge, which in this case creates water
orientation behavior similar to that observed for beidellite.
The simulations presented here indicate that only inner sphere surface complexes
of Cs+form within smectite interlayers, in contrast to results of simulations of Li-smectite
hydrates with 3.0 H2 O per Li+ (Greathouse and Sposito, 1998), in which inner sphere
complexes were found near tetrahedral charge sites, while outer sphere complexes were
associated with octahedral charge sites. The ability of Li+to form outer sphere
complexes may be attributed to its high ratio of charge to radius relative to Cs+. Despite
this difference, the influence of mineral charge distribution on the location of Cs+ within
116
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
the interlayer follows patterns similar to those reported for the Li-smectite systems
(Greathouse and Sposito, 1998). Within the octahedrally charged hectorite, Cs+ remain at
the midplane of the interlayer because the charge sites to which they are attracted are
deeply recessed in the clay layer. Within beidellite, on the other hand, Cs+are drawn to
the less-recessed tetrahedral charge sites and, therefore, remain near the clay mineral
surface. The occurrence of both types of charge site in montmorillonite creates disorder
in the behavior of the interlayer Cs+.
The RDF results (Figures 2.6 and 2.7) show interlayer cations do organize nearby
water molecules into partial hydration shells, despite their low charge to radius ratio and
the influence of surrounding clay layers. Our coordination number data suggest that the
Cs+in hectorite and montmorillonite utilize clay surface O to achieve a coordination
greater than the stable sixfold coordination suggested by XRD data on concentrated CsF
solutions (Bertagnolli et al., 1974). Marry et al. (2002) also noted a greater Cs-O
coordination within simulated smectite interlayers relative to solution values, due to the
contribution of mineral oxygen atoms. The Cs+ in beidellite are located close to one clay
surface and so must use water O to attain a sixfold coordination, as they are unable to
achieve significant coordination with the opposite clay surface (Figure 2.13).
2.4.3 Molecular Dynamics o f the Interlayer
The self-diffusion coefficients for interlayer water molecules (Dw) are a fraction
of the bulk water value (Table 2.4). However, water molecules do display continuous
diffusional behavior, unlike interlayer Cs+, which exhibit only jump diffusion on the
simulation time scale. The MD behavior of Cs+is consistent with strongly adsorbed
117
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Figure 2.13. Monte Carlo snapshot of two Cs+ (orange) sandwiched between beidellite
layers enclosing 1/3 water monolayer. The purple spheres represent mineral O, the grey
spheres represent mineral Si, and the green spheres represent tetrahedral charge
substitution sites containing Al instead of Si. Water O are blue and water H are white.
The grey lines connect to O within the coordination sphere of Cs+, which for Cs-Ociay has
a radius of 3.2 A, and for Cs-Owater a radius of 4.7 A.
counterions, as suggested also by apparent Cs+ self-diffusion coefficients in the range of
10'12to 10'13mV1(Muurinen et al., 1987; Cho et al., 1993; Wanner et al., 1996; Tsai et
al., 2001; Cormenzana et al., 2003) for compacted Cs-montmorillonites and -bentonites,
or 6 x 10"16m2s'1for a low water content Cs-montmorillonite (Calvet, 1973), as
compared to the bulk solution value, 2 x 10'9 m2 s'1(Nye, 1979). The volume in which a
Cs+ will oscillate between jumps is not the same for all Cs-smectite hydrates. The
counterions in Cs-montmorillonite with 1/3 water monolayer sample the largest volume
as they oscillate (Figure 2.11), for reasons discussed in Chapter 3. Spectroscopic data on
Li-smectite hydrates at similar water contents suggest that interlayer Li+ oscillate
vertically over as much as 1 A (Greathouse and Sposito, 1998). In Cs-montmorillonite
118
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
with 2/3 water monolayer, another extreme is present: Four Cs+are tightly bound, while
two others exhibit continuous diffusion behavior. The four tightly-bound Cs+ are near
two tetrahedral-charge sites, which implies that these sites have a dominating influence
on interlayer species in a clay mineral with both tetrahedral and octahedral charge. The
two mobile Cs+are seen to be attracted to some of the same locations on the clay surface,
though at different times during the simulation, and they display paths which overlap so
much that they cannot be distinguished easily (Figure 2.12).
2.4.4 Comparison to NMR and EXAFS Studies o f Interlayer Speciation
Although Weiss et al. (1990a, b) found NMR evidence to suggest the existence of
both inner and outer sphere complexes within the interlayer of Cs-hectorite, and even
estimated an exchange rate for the two species, our simulation data reveal only inner-
sphere surface coordination of interlayer cations. The large size of Cs+is inconsistent
with the large energy of hydration needed to maintain an outer sphere complex near a
charged mineral surface, as in the case of Li-montmorillonite (Greathouse and Sposito,
1998). It is possible that the solvated species identified by NMR may actually exist in
micropores, which are ubiquitous within Cs-smectite samples (Prost, 1975; Chiou and
Rutherford, 1997; Rutherford et al., 1997). Weiss et al. (1990a) argued against
assignment of either adsorbed Cs+peak to cations in micropores on the basis of the
calculated exchange rate between the two species. In fact, the exchange rate presented
for inner and outer sphere complexes is a maximum exchange rate only, and therefore
may not preclude exchange between internal and external Cs+. It is also possible that
other changes to the adsorbed cations occurring as a result of reduced temperature or
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
hydration state, such as entrapment within collapsed interlayers, have produced the
pattern of spectroscopic signals thought to indicate the process of exchange.
The NMR spectra of Cs-beidellite (Weiss et al., 1990b) and Cs-montmorillonite
(Weiss et al., 1990b; Kim et al., 1995, 1996) did not produce the same trends which
resulted in identification of exchanging populations of inner and outer sphere complexes.
The Cs-beidellite sample examined featured two Cs +species, which were thought to be
prevented from motional averaging by compositional heterogeneity occurring on large
spatial scales within the sample (Weiss et al., 1990b). Our simulations indicate that only
inner-sphere surface complexation is the stable state of Cs+ within 12 A beidellite.
Perhaps due to paramagnetic broadening, spectra for several montmorillonite samples did
not feature the same peak splitting phenomenon with reduced temperature and hydration
consistent with the existence of both inner and outer sphere complexes in the interlayer
region (Kim et al., 1995, 1996). Our MD simulation of Cs-montmorillonite with 2/3
water monolayer does reveal two distinct Cs+ species, one which shows oscillatory
movement and one which exhibits diffusional behavior. It is possible that the distinction
between these two species persists at NMR timescales, as suggested by the broad density
profile of Cs+ within montmorillonite (Figure 2.5). It is possible also that outer sphere
surface complexes in micropores contributes to the broad signals representing adsorbed
species in the NMR studies.
The EXAFS interpretations of Bostick et al. (2002), largely informed by the NMR
work discussed above, also suggest the existence of both inner and outer sphere complexes
within Cs-montmorillonite, while the EXAFS analysis of Nakano et al. (2003) provides
evidence for the presence of outer sphere complexes only. The simulations presented here
120
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
feature cations forming exclusively inner sphere complexes with the clay mineral surface
and surrounded by at least six oxygen atoms, while maintaining average Cs-O distances of
around 3 A, similar to the distance Bostick et al. (2002) and Nakano et al. (2003) assign to
an outer sphere complex. These models do not support the existence of an alternative
adsorbed Cs+species featuring Cs-O distances of 4.25 A.
A highly variable Cs coordination environment caused by sample preparation may
account for the discrepancies between our results and the NMR and EXAFS
interpretations. While simulations can focus specifically on the interlayer environment,
spectroscopic analysis of Cs-smectites inevitably averages signals from Cs+ in the
interlayer, adsorbed to edge sites, and present in micropores (Prost, 1975). Careful sample
preparation after Cs+ saturation, involving a drying step designed to encapsulate the ion
between surface cavities in the interlayer, followed by a LiCl rinsing step to remove less
tightly bound ions (Anderson and Sposito, 1991; Schroth and Sposito, 1997), would result
in a sample containing primarily inner sphere Cs+complexes residing in the interlayer of
the smectite. A sample produced with this method would have a far less variable Cs+
coordination environment, and could be useful as a reference NMR sample, or as an end
member describing inner sphere coordination for linear combination fitting of EXAFS
spectra (Bostick et al., 2002). Reinterpretation of the NMR and EXAFS data using this
approach may result in a view of interlayer structure similar to that provided by our
simulations.
121
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
2.5 Conclusions
Our MC simulations are consistent with a water content for 12 A Cs-smectite
hydrates that is less than one monolayer of water, as indicated by previous experimental
work. A 1/3 monolayer water content (1.33 H2 O per unit cell) seems to be the more
realistic, although two Cs-smectite systems successfully equilibrated while containing 2/3
monolayer. Simulated thermodynamic properties such as average layer spacing are in
agreement with experimentally derived properties. Radial distribution functions for
interlayer water molecules indicate the influence of the clay surface in creating a highly
distorted interlayer water structure. Analogous data indicate that Cs+is capable of
organizing a partial hydration shell, despite its low charge to radius ratio, and the
hindrance of surrounding mineral electrostatic fields.
Interlayer water motions correspond to self-diffusion at a much lower rate than in
bulk water. The Cs+cations within Cs-hectorite and Cs-beidellite interlayers hover close
to the octahedral and tetrahedral charge sites, respectively, displaying jump diffusion
over 800 ps, especially in wetter systems. This behavior is consistent with the low values
of experimentally determined bulk diffusion coefficients for Cs+. The cations in Cs-
montmorillonite interlayers exhibit more varied behavior. At 1/3 water monolayer, the
cations are still tightly bound, but display more movement normal and parallel to the clay
surface. At 2/3 monolayer, four cations are tightly bound, while two cations engage in
diffusional motion. Comparisons to 133Cs NMR and EXAFS analyses suggest the need
for further spectroscopic studies which control for layer spacing and take into account the
effects of microporosity in Cs-smectites.
122
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
2.6 Appendices
2.6.1 Appendix A : Conversion o f U Cs-O parameters to MCYparameters
The MCY parameters for the Cs-O interaction were obtained through conversion
of LJ potential function parameters for the same interaction provided by Smith and Dang
(1994). The LJ potential function may be mathematically described as:
N N
i=l j>i
m i
+ Sy
/ A12
mij
- 2
/ A6
r
mij
J
[2.4]
where e and rmare LJ parameters specific to atom pairs. Comparison with the MCY
potential function (Equation 2.1) indicates both the electrostatic contribution and the
summation over all atoms are equivalent, and therefore can be ignored during the
following manipulations. In order to transfer the LJ interaction parameters to a MCY
format, I constrain them such that the r values for which the potential energy equals 0,
and is at a minimum, are equal:
U(c) = 0 [2.5]
U(rm) = - e [2.6]
where a = 2'1/6rm. Examination of the MCY potential function reveals that an energy
minimum exists at rffl if and only if the following is true:
ln(CD/AB)
r =
D - B
[2.7]
I further constrain the equations such that the curvatures of the potential functions at rn
are identical:
^d2U A
V d r J
v ' r = r m
72s
[2.8]
123
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
It is then possible to evaluate the MCY parameters through the following equalities,
where (I and 8 are unitless parameters:
These equalities may be inserted into the portion of the MCY potential function
concerning van der Waals interactions to form the following equation:
These constraints yield values of 8 = 16.842 and p = 4.275, which when combined with
the LJ parameters of Smith and Dang (1994), provide the following MCY Cs-O
interaction parameters: A = 51.0 kJ mol'1, B = 1.0850 A"1, C = 3.72 x 106 kJ mol'1, and
D = 4.2758 A'1.
2.6.2 Appendix B: Addition o f Atom Type and MCY Parameters to MONTE
In order to insert Cs into the MONTE program as atom type 22, the text in bold
was added to lines 111 - 127 of the reader.f file, while the numbers in italics were
reduced from original values by 1 (Skipper, 1996):
C Molar masses
3 0.0, 1.0079, 15.9994, 0.0,
4 28.086, 26.919, 24.305, 17.0073, 40.08, 22.990,
A = d ePs
8 - p
[2.9]
[2.10]
[2.11]
[2.12]
[2.13]
124
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
5 39.1, 35.5, 40.1, 6.9, 12.0, 16.0, 14.0,
6 1 . 0 0 7 9 , 1 5 . 9 9 9 4 , 1 6 . 0 , 1 . 0 0 7 9 , 1 5 . 9 9 9 4 , 132.9, 18*0 . 0 /
C
C Potential is in units of electronic charge, Kcal mol-1, A, ps
C
data apot /maxint*0.0/, bpot /maxint*0.0/, cpot /maxint*0.0/
1 ,dpot /maxint*0.0/, epot /maxint*0.0/, fpot /maxint*0.0/
2 ,hpot /maxint*0.0/
data stype/'Vac','H ','0 ','Ch',
1 'Si', 'Al', 'Mg', 'OH', 'Ni', 'Na', 'K', 'Cl ',
2 'Ca','Li','C ','Me','N ','H_N','01','LJ','H_C','02'
3 , ' C s ' , 19* ' ' /
data debug/0/
end
In order to add the MCY Cs-O parameters to MONTE, the text in bold was added
to the i_potentials_mcy_srm_v3.0 file (Skipper, 1996):
$phys
nwat(l) =4,
mz=0.2677, qhwat=0.717484,
$end
$potp
bpotz = 0.0, 1088213.2,
12.1838741, 888054.41,
0 . 0 , 0 . 0 ,
0 . 0 , 0 . 0 ,
cpotz = 0.0, 5.152712,
1.0849875, 4.2747953,
0 . 0 , 0 . 0 ,
0 . 0 , 0 . 0 ,
ipot(l,22) = 19,
ipot(2,22)
ipot(4,22)
ipot(5,22)
ipot(17,22)
ipot(18,22)
The first line of bold text contains Cs-O A and C parameters in units of kcal mol1, while
the fourth line of bold text contains Cs-O B and D parameters. Each of the lines
125
= i s ,
= 2 0 ,
= 2 0 ,
= 1 9 ,
= 18,
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
beginning with ipot links an atom type pair to its MCY parameters. Cs-O pairs
represented by (2,22) and (18,22) are linked to the 18th set of MCY parameters, while Cs
paired with H (atom types 1 and 17), Si (atom type 4), and Al (atom type 5) are linked to
the 19th and 20th sets of MCY parameters, all of which are valued at 0.
2.7 References
Allen M. P. and Tildesley D. J. (1987) Computer Simulation o f Liquids. Oxford
University Press.
Anderson S. J. and Sposito G. (1991) Cesium-adsorption method for measuring
accessible structural surface charge. Soil Sci. Soc. Am. J. 55, 1569-1576.
Berend I., Cases J. M., Franfois M., Uriot J. P., Michot L., Masion A., and Thomas F.
(1995) Mechanism of adsorption and desorption of water vapor by homoionic
montmorillonites. 2. The Li+, Na+, K+, Rb+ and Cs+-exchanged forms. Clays Clay Miner.
43, 324-336.
Berendsen H. J. C., Grigera J. R., and Straatsma T. P. (1987) The missing term in
effective pair potentials. J. Phys. Chem. 91, 6269-6271.
Bertagnolli V. H., Weidner J.-U., and Zimmerman H. W. (1974)
Rontgenstrukturuntersuchung wasriger Caesiumfluorid-Losungen. Ber. Bunseges. Phys.
Chem. 78, 2-19.
Beveridge D. L., Mezei M., Mehrotra P. K., Marchese F. T., Ravi-Shanker G., Vasu T.,
and Swaminathan S. (1983) Monte Carlo computer simulation studies of the equilibrium
properties and structure of liquid water. In Molecular-Based Study o f Fluids (ed. J. M.
Haile and G. A. Mansoori), pp. 297-351. American Chemical Society.
Boek E. S., Coveney P. V., and Skipper N. T. (1995) Molecular modeling of clay
hydration: A study of hysteresis loops in the swelling curves of sodium montmorillonites.
Langmuir 11, 4629-4631.
Boek E. S., Coveney P. V., and Skipper N. T. (1995) Monte Carlo molecular modeling
studies of hydrated Li-, Na-, and K-smectites: Understanding the role of potassium as a
clay swelling inhibitor. J. Am. Chem. Soc. 117, 12608-12617.
Boek E. S. and Sprik M. (2003) Ab initio molecular dynamics study of the hydration of a
sodium smectite clay. J. Phys. Chem. B 107, 3251-3256.
126
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Bostick B. C., Vairavamurthy M. A., Karthikeyan K. G., and Chorover J. (2002) Cesium
adsorption on clay minerals: An EXAFS spectroscopic investigation. Environ. Sci.
Technol. 36, 2670-2676.
Calvet R. (1972) Water adsorption on clays: Study of the hydration of montmorillonite.
Bull. Soc. Chim. France 8, 3097-3104.
Calvet R. (1973) Hydration of montmorillonite and diffusion of exchangeable cations.
Part II. Study of exchangeable cation diffusion in montmorillonite. Ann. Agrono. 24, 135-
217.
Cebula D. J., Thomas R. K., and White J. W. (1981) Diffusion of water in lithium-
montmorillonite studied by quasi-elastic neutron scattering. Clays Clay Miner. 29, 241-
248.
Chang F.-R. C., Skipper N. T., and Sposito G. (1995) Computer simulation of interlayer
molecular structure in sodium montmorillonite hydrates. Langmuir 11, 2734-2741.
Chang F.-R. C., Skipper N. T., and Sposito G. (1997) Monte Carlo and molecular
dynamics simulations of interfacial structure in lithium-montmorillonite hydrates.
Langmuir 13, 2074-2082.
Chang F.-R. C., Skipper N. T., and Sposito G. (1998) Monte Carlo and molecular
dynamics simulations of electrical double-layer structure in potassium-montmorillonite
hydrates. Langmuir 14, 1201-1207.
Chang F.-R. C., Skipper N. T., Refson K., Greathouse J. A., and Sposito G. (1999)
Interlayer molecular structure and dynamics in Li-, Na-, and K-montmorillonite-water
systems. In Mineral-Water Interfacial Reactions: Kinetics and mechanisms (ed. D. L.
Sparks and T. Grundl), pp. 88-106. American Chemical Society.
Chavez-Paez M., dePablo L., and dePablo J. J. (2001) Monte Carlo simulations of Ca-
montmorillonite hydrates. J. Chem. Phys. 114,10948-10953.
Chavez-Paez M., van Workum K., dePablo L., and dePablo J. J. (2001) Monte Carlo
simulations of Wyoming sodium montmorillonite hydrates. J. Chem. Phys. 114, 1405-
1413.
Chiou C. T. and Rutherford D. W. (1997) Effects of exchanged cation and layer charge
on the sorption of water and EGME vapors on montmorillonite clays. Clays Clay Miner.
45, 867-880.
Cho W. J., Oscarson D. W., Gray M. N., and Cheung S. C. H. (1993) Influence of
Diffusant Concentration On Diffusion Coefficients in Clay. Radiochim. Acta 60, 159-
163.
127
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Chun K. S., Kim S. S., and Kang C. H. (2001) Release of boron and cesium or uranium
from simulated borosilicate waste glasses through a compacted Ca-bentonite layer. J.
Nucl. Mat. 298, 150-154.
Cormenzana J. L., Garcia-Gutierrez M., Missana T., and Junghanns A. (2003)
Simultaneous estimation of effective and apparent diffusion coefficients in compacted
bentonite. J. Contam. Hydrol. 61, 63-72.
Delville A. (1991) Modeling the clay water interface. Langmuir 7(3), 547-555.
Delville A. (1992) Structure of liquids at a solid interface - An application to the swelling
of clay by Water. Langmuir 8, 1796-1805.
Delville A. and Sokolowski S. (1993) Adsorption of vapor at a solid interface - A
molecular model of clay wetting. J. Phys. Chem. 97, 6261-6271.
Evans D. W., Alberts J. J., and Clark R. A. (1983) Reversible ion-exchange fixation of
cesium-137 leading to mobilization from reservoir sediments. Geochim. Cosmochim.
Acta 47,1041-1049.
Greathouse J. and Sposito G. (1998) Monte Carlo and molecular dynamics studies of
interlayer structure in Li(H20)3 -smectites. J. Phys. Chem. B 102, 2406-2414.
Greathouse J. A., Refson K., and Sposito G. (2000) Molecular dynamics simulation of
water mobility in magnesium-smectite hydrates. J. Am. Chem. Soc. 122, 11459-11464.
Greathouse J. A. and Storm E. W. (2002) Calcium hydration on montmorillonite clay
surfaces studied by Monte Carlo simulation. Mol. Sim. 28, 633-647.
Hensen E. J. M. and Smit B. (2002) Why clays swell. J. Phys. Chem. B 106, 12664-
12667.
Hensen E. J. M., Tambach T. J., Bliek A., and Smit B. (2001) Adsorption isotherms of
water in Li-, Na-, and K-montmorillonite by molecular simulation. J. Chem. Phys. 115,
3322-3329.
Iwasaki T. and Onodera Y. (1995) Sorption behaviour of caesium ions in smectites. In
Proceedings o f the 10th International Clay Conference, 1993, pp. 67-73. CSIRO
Publishing.
Jensen D. J. and Radke C. J. (1988) Cesium and strontium diffusion through sodium
montmorillonite at elevated temperature. J. Soil Sci. 39, 53-64.
Jorgensen W. L., Chandrasekhar J., Madura J. D., Impey R. W., and Klein M. L. (1983)
Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 70,
926-935.
128
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Karabomi S., Smit B., Heidug W., Urai J., and van Oort E. (1996) The swelling of clays -
Molecular simulations of the hydration of montmorillonite. Science 271, 1102-1104.
Kawamura K. and Ichikawa Y. (2001) Physical properties of clay minerals and water -
by means molecular dynamics simulations. Bull. Earthq. Res. Inst. Univ. Tokyo 76, 311-
320.
Kim Y., Cygan R. T., and Kirkpatrick R. J. (1996) 133Cs NMR and XPS investigation of
cesium adsorbed on clay minerals and related phases. Geochim. Cosmochim. Acta 60,
1041-1052.
Kim Y., Kirkpatrick R. J., and Cygan R. T. (1995) 133Cs NMR study of Cs reaction with
clay minerals. In Environmental Issues and Waste Management Techniques in the
Ceramic and Nuclear Industries, (ed. V. Jain and R. Palmer), pp. 629-636. American
Ceramic Society.
Lie G. C. and Clementi E. (1986) Molecular-dynamics simulation of liquid water with an
ab initio flexible water-water interaction potential. Phys. Rev. A 33, 2679-2693.
MacEwan D. M. C. and Wilson M. J. (1980) Interlayer and intercalation complexes of
clay minerals. In Crystal Structures o f Clay Minerals and their X-ray Identification (ed.
G. W. Brindley and G. Brown), pp. 197-248. Mineralogical Society.
Marry V., Turq P., Cartailler T., and Levesque D. (2002) Microscopic simulation of
structure and dynamics of water and counterions in a monohydrated montmorillonite. J.
Chem. Phys. 117, 3454-3463.
Matsuoka O., Clementi E., and Yoshimine M. (1976) Cl study of the water dimer
potential surface. J. Chem. Phys. 64, 1351-1361.
Mooney R. W., Keenan A. G., and Wood L. A. (1952) Adsorption of water vapor by
montmorillonite. II. Effect of exchangeable ions and lattice swelling as measured by X-
ray diffraction. J. Amer. Chem. Soc. 74, 1371-1374.
Muurinen A., Penttila-Hiltunen P., and Rantanen J. (1987) Diffusion mechanisms of
strontium and cesium in compacted sodium bentonite. In Materials Research Society
Symposium Proceedings: Scientific Basis for Nuclear Waste Management X, (ed. J. K.
Bates and W. B. Seefeldt), pp. 803-812. Materials Research Society.
Nakano M., Kawamura K., and Ichikawa Y. (2003) Local structural information of Cs in
smectite hydrates by means of an EXAFS study and molecular dynamics simulations.
Appl. Clay Sci. 23, 15-23.
Nye P. H. (1979) Diffusion of ions and uncharged solutes in soils and soil clays. Adv.
Agron. 31, 225-272.
129
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
ODay P. A., Rehr J. J., Zabinsky S. I., and Brown G. E. Jr. (1994) Extended x-ray
absorption fine structure (EXAFS) analysis of disorder and multiple-scattering in
complex crystalline solids. J. Am. Chem. Soc. 116, 2938-2949.
Ohtaki H. and Radnai T. (1993) Structure and dynamics of hydrated ions. Chem. Rev. 93,
1157-1204.
Onodera Y., Iwasaki T., Ebina T., Hayashi H., Torii K., Chatterjee A., and Mimura H.
(1998) Effect of layer charge on fixation of cesium ions in smectites. J. Contam. Hydrol.
35, 131-140.
Oscarson D. W., Hume H. B., and King F. (1994) Sorption of cesium on compacted
bentonite. Clays Clay Miner. 42, 731-736.
Park S.-H. and Sposito G. (2000) Monte Carlo simulation of total radial distribution
functions for interlayer water in Li-, Na- and K-montmorillonite hydrates. J. Phys. Chem.
B 104, 4642-4648.
Park S.-H. and Sposito G. (2002) Structure of water adsorbed on a mica surface. Phys.
Rev. Lett. 89, 085501.
Powell D. H., Tongkhao K., Kennedy S. J., and Slade P. G. (1997) A neutron diffraction
study of interlayer water in sodium Wyoming montmorillonite using a novel difference
method. Clays Clay Miner. 45, 290-294.
Powell D. H., Tongkhao K., Kennedy S. J., and Slade P. G. (1998) Interlayer water
structure in Na- and Li-montmorillonite clays. Physica B 241-243, 387-389.
Prost R. (1975) Study of the hydration of clays: Water-mineral interactions and
mechanism of water retention. II. Study of a smectite (hectorite). Ann. Agronom. 26, 463-
535.
Prost R. (1976) Interactions between adsorbed water molecules and the structure of clay
minerals: Hydration mechanism of smectites. In Proceedings o f the International Clay
Conference, 1975, Mexico City (ed. S. W. Bailey), pp. 353. Applied Publishing Ltd.
Refson K. (2000) MOLDY: A portable molecular dynamics simulation program for serial
and parallel computers. Computer Phys. Commun. 126, 310-329.
Robinson G. W., Zhu S.-B., Singh S., and Evans M. W. (1996) Water in Biology,
Chemistry and Physics: Experimental Overviews and Computational Methodologies.
World Scientific.
Rutherford D. W., Chiou C. T., and Eberl D. D. (1997) Effects of exchanged cation on
the microporosity of montmorillonite. Clays Clay Miner. 45, 534-543.
130
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Shroll R. M. and Smith D. E. (1999) Molecular dynamics simulations in the grand
canonical ensemble: Formulation of a bias potential for umbrella sampling. J. Chem.
Phys. 110, 8295-8302.
Shroth B. K. and Sposito G. (1997) Surface charge properties of kaolinite. Clays Clay
Miner. 45, 85-91.
Skipper N. T. (1996) MONTE User's Manual. Department of Physics and Astronomy,
University College London, England.
Skipper N. T., Chang F.-R. C., and Sposito G. (1995) Monte Carlo simulation of
interlayer molecular structure in swelling clay minerals. 1. Methodology. Clays Clay
Miner. 43, 285-293.
Skipper N. T., Sposito G., and Chang F.-R. C. (1995) Monte Carlo simulation of
interlayer molecular structure in swelling clay minerals. 2. Monolayer hydrates. Clays
Clay Miner. 43, 294-303.
Smith D. E. (1998) Molecular computer simulations of the swelling properties and
interlayer structure of cesium montmorillonite. Langmuir 14, 5959-5967.
Smith D. E. and Dang L. X. (1994) Computer simulations of cesium water clusters - Do
ion water clusters form gas-phase clathrates? J. Chem. Phys. 101, 7873-7881.
Soper A. K., Bruni F., and Ricci M. A. (1997) Site-site pair correlation functions of water
from 25 to 400 degrees C: Revised analysis of new and old diffraction data. J. Chem.
Phys. 106, 247-254.
Sposito G. (1981) Single-particle motions in liquid water. II. The hydrodynamic model.
J. Chem. Phys. 74, 6943-6949.
Sposito G. (1989) The Chemistry o f Soils. Oxford University Press.
Sposito G., Park S.-H., and Sutton R. (1999) Monte Carlo simulation of the total radial
distribution function for interlayer water in sodium and potassium montmorillonites.
Clays Clay Miner. 47,192-200.
Sposito G., Skipper N. T., Sutton R., Park S.-H., Soper A. K., and Greathouse J. A.
(1999) Surface geochemistry of the clay minerals. Proc. Natl. Acad. Sci. USA 96, 3358-
3364.
Teppen B. J., Rasmussen K., Bertsch P. M., Miller D. M., and Schafer L. (1997)
Molecular dynamics modeling of clay minerals . 1. Gibbsite, kaolinite, pyrophyllite, and
beidellite. J. Phys. Chem. B 101, 1579-1587.
Tsai S.-C., Ouyang S., and Hsu C.-N. (2001) Sorption and diffusion behavior of Cs and
Sr on Jih-Hsing bentonite. Appl. Rad. Isotopes 54, 209-215.
131
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Tuck J. J., Hall P. L., Hayes M. H. B., Ross D. K., and Hayter J. B. (1985) Quasielastic
neutron-scattering studies of intercalated molecules in charge-deficient layer silicates.
Part 2. High-resolution measurements of the diffusion of water in montmorillonite and
vermiculite. J. Chem. Soc. Far. Trans. 1 81, 833-846.
Wanner H., Albinsson Y., and Wieland E. (1996) A thermodynamic surface model for
caesium sorption on bentonite. Fresenius J. Anal. Chem. 354, 763-769.
Weiss C. A., Kirkpatrick R. J., and Altaner S. P. (1990a) The structural environments of
cations adsorbed onto clays: 133Cs variable-temperature MAS NMR spectroscopic study
ofhectorite. Geochim. Cosmochim. Acta 54, 1655-1669.
Weiss C. A., Kirkpatrick R. J., and Altaner S. P. (1990b) Variations in interlayer cation
sites of clay minerals as studied by 133Cs MAS nuclear magnetic resonance spectroscopy.
Amer. Miner. 75, 970-982.
Winter M. (2003) WebElements, The University of Sheffield and WebElements Ltd.,
UK., accessed 27 January 2003 from: http://www.webelements.com.
Young D. A. and Smith D. E. (2000) Simulations of clay mineral swelling and hydration:
Dependence upon interlayer ion size and charge. J. Phys. Chem. B 104, 9163-9170.
132
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Chapter Three
Visualization of molecular dynamics simulations of Cs-smectite hydrates
3.1 Introduction
The human eye and brain have evolved into a sensitive pattern recognition
instrument designed to probe both three-dimensional space and time for predictive
associations and causal relationships. Yet, until recently, when analyzing and presenting
scientific data, researchers have been constrained to the two dimensions of a sheet of
paper. While many creative methods of escaping flatland have been developed to
capture multiple dimensions of data within 2D illustrations (Tufte, 1990), a true escape
has only become possible with the advent of powerful graphics and personal computers
and visualization software. Modem researchers can manipulate, even animate,
projections of complex 3D structures as part of routine analysis of numerous multi
dimensional datasets. The scientific visualizations produced, and the insights gained
from them, may be shared through the use of newly established electronic journals
(Tomney and Burton, 1998) or laboratory websites.
Molecular modeling data are uniquely suited for 3D visualization, given the
complex spatial relationships typical of interacting molecules, as well as the large volume
of spatial information collected from a simulation. Powerful graphics programs have
allowed scientists to examine images representing electrostatic potential surfaces
(Gillilan and Ripoll, 1995; Leonard, 1995; Mackay, 1995) and molecular orbital
configurations (Smith, 1995); to identify stable molecular conformations (Mackay, 1995;
Murray and Daage, 1995) as well as subtle differences in shape among similar substances
133
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
(Blaney et al., 1995; Richards et al., 1995; de Groot and Grubmuller, 2001); to uncover
adsorption interactions (Richards et al., 1995; Lindan et al., 1996; Brickmann, 1999;
Bailey et al., 2001; Wang et al., 2001) and enzyme docking sites (Brickmann, 1999); and
to follow complex chemical reactions on a molecular scale (Gillilan and Ripoll, 1995;
Ranck, 1995; Brickmann, 1999; Ludemann et al., 2001; Gruia et al., 2003). Three-
dimensional images and animations allow careful analysis of the effects of molecular
structure on properties such as reactivity, and can provide considerable insight
concerning interactions involving more than two molecules, common in complex
systems, yet often difficult to examine with traditional 2D methods.
Molecular dynamics (MD) simulations produce datasets consisting of atomic or
molecular motions over time, perfect for conversion into 3D animations. However, most
MD research is published without the benefit of this visualization technique. Images
reflecting the system configuration at a single point in simulation time are often presented
in MD publications. These snapshots may be strictly two dimensional, as in Figure 3.1,
or may feature visual cues, such as perspective and lighting (Ware, 2000), which allow
the viewer to extrapolate three-dimensional form (Figure 2.13). Electronic media offer
an opportunity to publish 3D snapshots, allowing the viewer to rotate the simulation cell
and inspect the spatial relationships presented (see Figure 6 in Sherman and Collings
(2002)). Two-dimensional plots of the positions visited by a particular atom or molecule
during a MD calculation provide data beyond that of a single instant in time. These
positions may be represented by a collection of points, as in Figure 2.11, or by a line
which provides limited information concerning the order of occupation of various
positions (Figures 2.9, 2.10, and 2.12). Given the likelihood that atoms or molecules
134
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
within a simulation cell will occupy some of the same sites at different times during the
calculation, only a few different trajectories may be displayed on any particular plot
without extensive entanglement. This limitation substantially reduces the likelihood of
spotting multi-body interactions taking place within a simulation.
Figure 3.1. A view of the simulation cell (outlined in grey) for Cs-montmorillonite with
1/3 water monolayer. Interlayer Cs+ are represented by orange spheres, while interlayer
water O are represented by blue spheres and water H by white spheres. The surrounding
mineral layers are displayed in ball and stick style, with the greyish purple spheres
representing the mineral O atoms, white spheres representing H, grey spheres
representing Si, and green spheres representing Al.
While the tangled mess of a traditional 2D plot of the path of a few molecules
over time (see Figures 2.9, 2.10, and 2.12) is often difficult to interpret, animated motion
has been shown to enhance the identification of multi-body spatial relationships (Ware
and Franck, 1996). The presence of motion also vastly improves perception of the three-
dimensional nature of a view under observation (Hubona et al., 1997). In addition,
motion provides invaluable information concerning changes which occur over time
135
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
(Ware, 2000). Animations thus can be used to reveal complex and dynamic three
dimensional relationships hidden within MD coordinate data.
Use of this visualization technique to supplement analysis of MD output produced
for a variety of aqueous systems is becoming more common, though publication of the
results in accessible movie formats remains infrequent. Animations of biomolecular
trajectories can provide substantial insight into the mechanisms operating within many
complex biological systems. Piana et al. (2002) made available a movie of their MD
simulation of an aqueous system containing the enzyme HIV-1 protease (a major target
of anti-AIDS drug therapies) bound to a peptide substrate. The enzyme is visualized as a
long thin tube which bends and twists to follow the backbone of the molecule. This
simplified structure, and the absence of any representation of surrounding water
molecules, are common features of animations of complex aqueous organic systems. The
animated molecule displays large-scale motions that modify the geometry of the
enzymes cleavage site. Subsequent quantum chemical calculations, using quantum
chemical MD methods, confirm that these motions are crucial to the catalytic activity of
the enzyme (Piana et al., 2002).
Animation of MD data describing motions of a version of Staphylococcal
nuclease reveals between a pair of amino acids the presence of a salt bridge, which
hinders the unfolding of the enzyme from its crystalline to aqueous conformation (Gruia
et al., 2003). In this animation, a ribbon symbolizing the enzymes backbone structure is
amended with representations of the functional groups involved in the salt bridge as thin
cylinders, or sticks, which indicate atom position and bonding and are color coded to
indicate elemental composition. The added detail allows a meticulous inspection of the
136
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
salt bridge interaction. A movie of an aqueous antibody-buckyball MD calculation
features similarly detailed representations of the aromatic functional groups of a common
mouse antibody with high binding affinity and specificity for C6o buckyballs (Noon et al.,
2002). The visualization demonstrates that the buckyball, which may have powerful
applications in nanotechnology, is almost completely desolvated by these antibodies.
Animations aided the analysis of MD simulations of a cytochrome P450 enzyme
ejecting substrate and products (Ludemann et al., 2001). To probe the accessibility of the
enzymes buried active sites (important for synthesis and degradation of numerous
physiological compounds) over the short time scale accessible to MD methods, substrates
and products were inserted into these sites, then forced out using a randomly-oriented
force, a technique known as random expulsion MD (Ludemann et al., 2001). Animations
of the simulations illustrate the three access pathways identified, presenting the enzyme
as a ribbon and the substrate and product molecules as three-dimensional shapes made up
of sticks or lines depicting atoms and their bonds (Ludemann et al., 2001).
74-
Another set of animations describes calmodulin (a Ca binding protein important
for enzyme regulation) modeled in three aqueous states: calcium-activated (Yang et al.,
2001), calcium-free (Yang and Kuczera, 2002a), and complexed with a target peptide
(Yang and Kuczera, 2002b). Separate animations displaying the protein as a simple line
drawing and as a ribbon are provided for the first two systems. Color is used to
differentiate three segments of the protein ribbon, marking three distinct domains, while
spheres represent Ca2+. Animations of both systems reflect a pattern of large-scale
structural fluctuations between protein domains, but little movement within domains
(Yang et al., 2001; Yang and Kuczera, 2002a). The calmodulin-peptide complex, on the
137
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
other hand, shows relatively little motion (Yang and Kuczera, 2002b), and only one
animation, featuring ribbon representations of protein and peptide, is offered. Laaksonen
and Rosenholm (1993) display another way to use color. They modeled, then animated, a
Na+ octanoate micelle in solution, mapping over time the relative force experienced by
each atom in the ions using color to illustrate the fluctuations observed over the course of
the MD simulation.
Calculations concerning an estrogen receptor protein complexed with an estrogen
response element of DNA were animated by Harris et al. (1996). Color was used here to
differentiate DNA nucleotides, as well as the relative polarities of different amino acids
in the protein. Molecules were represented by animated line drawings, and several
significant water molecules were included in the animations. Two movies were provided,
one which featured movements of the entire protein-DNA complex, and a close-up which
illustrated the abundance of direct and water-mediated H-bonds, represented by dashed
lines, between the two organic molecules.
Representations of water molecules are also essential to animations of MD
simulations of aquaporins, proteins of the cell membrane which act as highly selective
and efficient water channels. Data from large scale simulations of two aquaporin
tetramers embedded in fully solvated bilayer membranes were used to produce a complex
animation of a sideview of the cell membrane featuring proteins represented by ribbons,
lipids with hydrophilic spherical heads and hydrophobic cylindrical tails, and water
molecules made of joined spheres (de Groot and Grubmuller, 2001). Also provided is a
movie illustrating the permeation of a water molecule through a single aquaporin channel
over a 2 ns time scale (de Groot and Grubmuller, 2001). Here the protein ribbon is
138
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
supplemented with ball and stick structures representing specific functional groups which
protrude into the pores interior, while water molecules are represented by joined spheres,
yellow highlighting being used to emphasize the one which completes the passage
through the protein.
Molecular dynamics studies of inorganic aqueous systems can benefit from
animation as well. Small systems consisting of only a few simple molecules may be
studied using extremely accurate quantum chemical molecular dynamics methods (Tse,
2002). Resulting animations typically represent atoms as spheres of different sizes and
colors, depending on their elemental identity, and display interactions such as the
dynamic nature of the water molecule solvent shell around V02(0H2)+(Buhl and
Parrinello, 2001), the near hydrophilic, far hydrophobic behavior of dimethyl sulfoxide
toward surrounding water molecules (Kirchner and Hutter, 2002), and the temperature
sensitive oscillation of a proton between two nearby water molecules (Cheng and Krause,
1997). The last animation was produced using a virtual reality tool which allows the
viewer to move and rotate the system while the movie plays. A set of movies taken from
quantum MD simulations of a mineral system, a TiCb (110) surface (Lindan et al., 1996),
illustrate a water molecule dissociating on the mineral surface, or experiencing repulsion.
Simulation of a larger inorganic system involving more numerous and complex
molecules, such as a layered mineral structure containing water molecules and
counterions, necessitates a return to classical MD techniques. Wang et al. (2001)
modeled three layers of hydrated Mg/Al hydrotalcite with C1 counterions, and animated
the results. The mineral layers were illustrated using a mixture of styles. Mineral oxygen
and hydrogen were represented by connected, color-coded sticks, either surrounding
139
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
spheres representing Mg and Al, or emerging from octahedra whose vertices correspond
to the oxygen atoms coordinating the mineral cations. Polyhedral representations of
minerals emphasized their regular crystalline structure and, in this case, were color-coded
to allow identification of the cation present in the center of each polyhedron. The
animation demonstrated the distortion experienced by the octahedral mineral structure
due to direct and indirect interactions with adsorbed anions and water molecules during
the course of the simulation (Wang et al., 2001).
Bridging the gap between organic and inorganic chemistry, Bailey et al. (2001)
published a series of animations of annealed MD simulations of aqueous organic and
organo-mineral systems, including movements of a model humic molecule in solution,
and the adsorption of this organic molecule or a group of atrazine molecules to a hydrated
mica layer surrounded by K+ counterions. The molecules were represented in ball-and-
stick style, and water molecules were removed to allow the viewer to focus on organo-
mineral interactions. Animations indicated that the model humic substance undergoes
numerous adjustments to facilitate adsorption to the mineral, while atrazine appeared to
possess little affinity for the hydrated mica surface, preferring to adsorb to broken edge
sites (Bailey et al., 2001).
These examples illustrate the power of animation as a tool for identification of
complex multi-body interactions occurring in three-dimensional environments over time.
Adsorption processes within smectite mineral interlayers are prime candidates for this
method of analysis. The smectite mineral provides two types of adsorption site for
interlayer species, the surface oxygen triads coordinated to each tetrahedral cation, and
the six-oxygen surface cavities formed as a result of the arrangement of these triads in the
140
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
smectite layer. The affinity of water molecules and cations for these surface locations is
modified by the presence of negative charge sites found in the tetrahedral and octahedral
sheets within the mineral. It is clear that an understanding of the spatial relationships
between surface adsorption sites and interior charge sites must precede an understanding
of adsorption interactions in the smectite interlayer region. And while the information in
a mineral structure may be considered fairly static, trajectories of interlayer species
commonly indicate a variety of complex responses to this organization, as well as to the
presence of neighboring ions and water molecules (Chang et al., 1995; Skipper et al.,
1995; Greathouse and Sposito, 1998; Smith, 1998; Chang et al., 1999; Marry et al.,
2002).
Monte Carlo and MD simulations of Cs-smectite hydrates have been described in
detail in Chapter 2. The MD results were analyzed for average properties, such as
interlayer water self-diffusion coefficients, as well as for individual molecular properties,
such as 2D plots of xyz coordinates sampled by specific Cs+ions or by water molecules.
These traditional simulation outputs could not characterize the full complexity of
interactions among cations, water molecules, and mineral surface sites, nor could they
suggest an explanation for the unusually mobile behavior of cations within
montmorillonite interlayers. Further exploration of the data using a different set of
techniques was advisable.
The Cs-smectite hydrate systems are especially suited to animation as a means of
observation of surface adsorption processes. The presence of relatively few water
molecules which reside in a single layer across the surface of the mineral greatly
simplifies the number of moving elements in the system, facilitating identification of
141
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
interactions of interest. Therefore, to continue analysis of the MD results, I constructed
animations of molecular motions in the five Cs-smectite hydrate systems described in
Chapter 2 to examine the effects of charge site and hydration state on the behavior of
interlayer species. Spheres of different size and color were used to represent the species
visualized, a common display style intuitive to most viewers. Color was further used to
highlight species involved in processes of interest. These animations provided substantial
insight into the complex three dimensional relationships among interacting interlayer
molecules over time.
3.2 Methods
3.2.1 Simulation o f Cs-smectite Hydrates
A detailed description of the Cs-smectite systems examined and the modeling
methods employed is provided in Section 2.2. Briefly, the simulation cell used consists
of two opposing halves of a rigid 2:1 clay layer surrounding an interlayer region filled
with rigid water molecules and Cs+ions (Figure 3.1). The unit cell formulae of the three
Cs-smectites modeled are:
Hectorite: Cso.75[Si8](Mg5.25Lio.75)02o(OH)4
Beidellite: Cso.7 5 [Si7 25Alo.75](Al4)02o(OH)4
Montmorillonite: Cso.75[Si7.75Alo.25](Al3.5Mgo.5)02o(OH)4
Hectorite is an octahedrally charged, trioctahedral smectite, while beidellite is a
tetrahedrally charged, dioctahedral smectite, and montmorillonite is a dioctahedral
smectite possessing both types of charge site (Greathouse and Sposito, 1998). The
simulation cell contains eight unit cells, and has lateral (ab) dimensions 21.12 x 18.28 A2.
142
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Layer spacing is determined for each simulation system individually during MC
equilibration.
Interactions between the interlayer species, and between them and atoms in the
clay layers, are represented parametrically by a MCY-type potential function (Equation
2.1) (Matsuoka et al., 1976). This form of potential function was used because it has led
to accurate modeling of experimentally accessible quantities such as layer spacing and
interlayer species structure and mobility (Chang et al., 1995; Skipper et al., 1995;
Greathouse and Sposito, 1998; Chang et al., 1999). The MCY potential function for the
water-water interaction is optimized on dimer structures and, therefore, does not enforce
the tetrahedral configuration of bulk water as strongly as do other, more empirically-
based water-water potential functions (Matsuoka et al., 1976). This fact may explain its
enhanced ability to model water molecules in the especially constrained geometry of the
clay interlayer region (Chang et al., 1995; Skipper et al., 1995; Chang et al., 1999). Van
der Waals parameters for the Cs-0 interaction were obtained through conversion of
Lennard-Jones potential function parameters taken from Smith and Dang (1994) (see
Section 2.6.1). The resulting MCY Cs-O interaction parameters were: A = 51.0 kJ mol'1,
B = 1.0850 A"1, C = 3.72 x 106kJ mol'1, D = 4.2758 A'1. Van der Waals interactions for
Cs-H, Cs-Al, and Cs-Si atom pairs were assumed to be negligible.
Allen and Tildesley (1987) have outlined the general principles of MC and MD
simulations of liquids, whereas Chang et al. (1995, 1999) and Greathouse and Sposito
(1998) have developed simulation methodologies customized for exploration of the
interlayer region of 2:1 clay mineral hydrates. The program MONTE (Skipper, 1996)
was used to run MC (NgT ensemble) simulations on Cs-smectite hydrates at 300 K and
143
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
100 kPa normal stress (a). Monte Carlo calculations began with a simulation cell that
had a 14 A layer spacing and contained 6 Cs+ in the midplane of the inter layer, along
with several randomly placed water molecules. Three-dimensional periodic boundary
conditions were applied (Allen and Tildesley, 1987). Simulation started with an
optimization developed by Chang et al. (1995, 1999), consisting of 20,000 MC steps in
which only water molecules were allowed to move, followed by another 20,000 steps in
which water molecules could move and interlayer spacing could vary. After this
optimization, all molecules were allowed to move in any direction, provided that the clay
layer only moved about once for every five interlayer molecule movements. Information
about total potential energy and layer spacing was collected every 500 steps. Systems
were considered MC-equilibrated when convergence profiles of total potential energy and
layer spacing provided minimum average values. The standard deviation of the average
layer spacing was further required to be consistent with the precision of available
experimental data on layer spacing.
The program MOLDY (Refson, 2000) was used to perform MD (NVE ensemble)
simulations of the Cs-smectite hydrates which had converged under MC simulation.
Monte Carlo-equilibrated configurations in which the layer spacing was held constant
were used as initial coordinates for the MD calculations. With the rigid clay layers fixed
in position, initial velocities were assigned randomly to interlayer molecules following a
Maxwell-Boltzmann distribution. In order to stabilize the temperature at 300 K, the
simulations began with 20 ps of temperature scaling (Refson, 2000) using a 0.5 fs
timestep. Temperature scaling was performed for each molecular species individually,
144
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
with rotational and translational kinetic energies scaled separately. After initial scaling,
the MD simulations proceeded for 800 ps.
3.2.2 Constructing Animations from MD Trajectory Data
Molecular center-of-mass coordinates collected at 0.1 ps intervals during the MD
simulations were used as the input data for these animations, providing a total of 8000
sets of coordinates for the interlayer region of each Cs-smectite hydrate. Input data files
containing the trajectories of the interlayer species were constructed as follows. The first
row of these files assigned a number and species identity to each set of columns
describing the motion of a particular cation or water molecule. The first three columns in
a set indicated the progression of x, y, and z values, respectively, occupied by the moving
particle, while the fourth column provided a numerical identity for the subject, and the
fifth simply described the radius (A) of the species. Each water molecule was assigned
its own number, starting with 0, while all cations were assigned the same value, 11 or 21,
depending on the number of water molecules present in the interlayer. A radius of 1.67 A
was used for Cs+ (Atkins, 1994), and a radius of 1.44 A (or half the average distance
between pairs of water O observed experimentally (Soper and Phillips, 1986)) was used
for water molecules. A separate file was created to hold coordinates for the stationary
plane of mineral surface oxygen atoms. The mdshak utility in MOLDY (Refson, 2000)
provided clay mineral coordinates, which were converted from fractional to Cartesian
format, then further adjusted to provide the same center of origin as the trajectory data.
Stationary coordinates were assembled into a lattice file, containing columns for the x, y,
and z coordinates and atomic radius (1.40 A (Atkins, 1994)) for all surface oxygens, all
145
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
of which was preceded by a line stating the total number of atoms on the lattice (48).
Abbreviated examples of the file formats employed may be found in the Appendix.
I used a custom animation tool to create a series of JPEG images, one for each set
of coordinates, based on the center-of-mass information. The animation tool, created for
this project by the Lawrence Berkeley National Laboratory/National Energy Research
Scientific Computing Center (LBNL/NERSC) Visualization Group (http://vis.lbl.gov/),
runs on Windows or Unix/Linux platforms, and is based on OpenRM Scene Graph, an
Open Source, cross-platform scene graph library (http://openrm.sourceforge.net/). Color
was used to differentiate interlayer species and surface oxygens, and further color
adjustments allowed the highlighting of particular ions or molecules. These images were
then animated in MPEG-1 or Quicktime format using the MediaConvert tool and
executed on a Silicon Graphics visualization server located in the Visualization Group
Laboratory at LBNL.
3.3 Results and Discussion
Molecular dynamics simulation and subsequent animation revealed complex and
fascinating interactions between Cs+ions, water molecules, and the charged clay surface.
Short MPEG-1 animations of twenty to fifty picoseconds (~l-2 MB) illustrating features
relevant to the discussion below may be viewed in Sutton and Sposito (2002), whereas
full-length (800 ps) Quicktime animations (-80-120 MB) may be found on the Sposito
simulation team website (http://esd.lbl.gov/sposito/). Significant findings from these
animations are illustrated here with a series of snapshots (Figures 3.2, 3.4, 3.5, 3.7, and
3.9).
146
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
Animations of the five systems share some basic similarities of molecular
behavior, as described in Chapter 2. For example, the Cs+tend not to roam, but instead to
hover near charge sites recessed within the clay layer. When the ions do move, they
display jump diffusional behavior, in which the ions vibrate in one place for some time,
then hop to new locations (Figure 3.2). These motions tend to be confined to different
locations around the charge sites attracting the cations. In contrast, water molecules
display continuous diffusion on the timescale of these animations. Both ions and water
molecules move more frequently in the 2/3 monolayer systems, a trend indicated for
water molecules by an increased self-diffusion coefficient (Table 2.4). However, Cs+and
water molecule movements in the interlayer are substantially slower than in bulk ionic
solution, where these two species have self-diffusion coefficients of 2 x 10'9m2 s'1(Nye,
1979; Ohtaki and Radnai, 1993).
Animations allowed us to see changes in the structure of hydrating water
molecules around interlayer Cs+ over time. Water molecules move from one Cs+ to
another, but are rarely far from an interlayer ion (Figure 3.2). Thus, despite the weak
electrostatic field of Cs+ and its large size, these ions are able to organize water molecules
into partial hydration shells, defined as those water molecules within a radial distance of
< 5 A from the cation (obtained from MC radial distribution function calculations
(Figures 2.6 and 2.7, Table 2.3)). The number of water molecules within each shell
varies substantially with time for each cation. The exchange behavior seen among Cs+
waters of hydration contrasts with the results of MD calculations of similarly low water
content Li-smectites (Greathouse and Sposito, 1998), in which waters of hydration
remained near the same ion for as long as 200 ps. The difference may be attributed to the
147
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Figure 3.2. Snapshots taken from an animation of a MD simulation of Cs-hectorite with
1/3 water monolayer. Interlayer Cs+are represented by orange spheres, while interlayer
water molecules are represented by blue spheres. The greyish purple spheres symbolize
the surface oxygen atoms of the lower clay layer surrounding this interlayer region. The
rest of the atoms making up this lower clay layer, as well as all the atoms making up the
upper clay layer, have been removed to improve the visibility of the interlayer region.
Due to the periodic nature of the simulations, interlayer species which exit from one side
of the simulation cell reappear on the opposite side. In this series of images, the
highlighted Cs+(red) displays jump diffusion as it hops between two sites above
neighboring surface oxygen triads. The highlighted water molecule (lighter blue)
positioned in the center of the simulation cell in frame a) displays continuous diffusional
behavior as it moves from one Cs+partial hydration shell to another. Another highlighted
water molecule positioned to the left of the first in frame a) can be seen while visiting a
surface cavity site for a short time in frame c).
148
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
smaller radius of the Li+ ions, which results in stronger electrostatic attraction for water
molecules, as well as to the shorter timespan of the Li-smectite simulations (Greathouse
and Sposito, 1998), which may not have been long enough to capture water exchange
processes.
The animations bring to light the phenomenon of water molecule sharing between
two Cs+ions, an effect seen in all five simulations to varying degrees (Figure 3.3). A
shared water molecule may be defined as one which lies within the < 5 A hydration shells
of two Cs+ simultaneously for a period of hundreds of picoseconds or more. Shared
water molecules remain in a fairly fixed position relative to the two Cs+they hydrate
during this period of sharing. Our animations appear to be the first evidence for such an
interaction, which likely requires the conditions of confined geometry, presence of
surface charge, and close proximity of ions that are found in clay interlayers.
Figure 3.3. A view in the xy plane of the two Cs+ - two water complex seen in an
animation of Cs-montmorillonite with 1/3 water monolayer.
Cs-hectorite with 1/3 water monolayer (Figure 3.2) exhibits the least water
sharing, with two Cs+ pairs (the top left and bottom right Cs+ are one pair, and the middle
149
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
two Cs+on either side of the simulation cell are another), each sharing one water
molecule for hundreds of picoseconds. In contrast, Cs-montmorillonite with 1/3 water
monolayer (Figure 3.4) contains two pairs of Cs+that share two water molecules per pair
over the entire simulation (two Cs+at the bottom of the simulation cell form one pair,
while the other is made up of the middle two Cs+ on either side of the simulation cell).
The molecules making up the two Cs+ - two water complex near the bottom of the Cs-
montmorillonite, 1/3 water monolayer animation are so strongly bound that they can be
seen to rotate as a unit (Figure 3.4). In addition, other Cs+pairs in this system share one
or two water molecules, but only for a few hundred picoseconds or less. Such
entrapment of the water molecules could underlie the nonlinear relationship between their
mean-square displacement (MSD) and elapsed time noted for this system, which signaled
the absence of self-diffusion in the interlayer water (Table 2.4). Cs-beidellite with 1/3
water monolayer (Figure 3.5) features three pairs of cations which participate in water
sharing of one or two water molecules for substantial portions of the 800 ps simulation
(cations to the left and right of the animation are one pair, cations to the top and bottom
of the animation are another, and the third consists of the top cation and its neighbor to
the lower left). Cs-beidellite with 1/3 water monolayer may be considered to have an
intermediate level of water sharing, as compared to the hectorite and montmorillonite
systems, as it features more water sharing cation pairs than hectorite, but a reduced
number of shared water molecules and duration of water sharing per pair, as compared to
montmorillonite.
150
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
Figure 3.4. Snapshots taken from the animation of Cs-montmorillonite with 1/3 water
monolayer. Interlayer species are represented as described in Figure 3.2. The two
highlighted Cs+ (red) share the two highlighted water molecules (lighter blue). The
molecules making up this two Cs+ - two water complex are so strongly bound that they
are observed to rotate as a unit during the animation. Interlayer water molecules did not
enter six-oxygen cavities in this portion of the simulation.
151
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Figure 3.5. Snapshots taken from the animation of Cs-beidellite with 1/3 water
monolayer. Interlayer species are represented as described in Figure 3.2. The three
highlighted water molecules (lighter blue) reside within six-oxygen cavities of the clay
surface for extended time periods.
A difference in cation location on the clay surface, caused by both the orientation
of structural hydroxyl groups and the relative proportion of tetrahedral charge sites, may
account for the degree of water sharing present in each clay system. Figures 2.10 and
2.11 provided the first indications that the Cs+within montmorillonite interlayers
experienced a larger range of motion as compared to the cations within other smectite
hydrates. Careful comparison of the Cs-montmorillonite system with 1/3 water
monolayer (Figure 3.4) to the corresponding hectorite and beidellite systems (Figures 3.2
and 3.5) reveals the specific cation motions which result in this greater variation in
location. The cations within montmorillonite frequently hover over and dip into the
cavities of the mineral surface, whereas the cations within hectorite and beidellite
interlayers rarely stray from positions directly over oxygen triads found at the edges of
the cavities. Smith (1998) and Marry et al. (2002) noted the attraction of Cs+both to
oxygen triads and to surface cavities within their simulated montmorillonite systems, but
152
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
they did not examine contrasting minerals. Nakano et al. (2003) documented a
preference for cation adsorption near oxygen triads within their simulated Cs-beidellite
hydrate.
The availability of oxygen triad sites to cation adsorption may be affected by clay
layer registration, as suggested by Marry et al. (2002). Entrapment of Cs+ between
opposing surface cavities is a natural process that can be induced in the laboratory
through alternate wetting and drying procedures, and is the basis for standard methods
used to determine structural surface charge (Anderson and Sposito, 1991; Schroth and
Sposito, 1997). Our MC calculations, which provided the coordinates used in the MD
simulations and established the fact that the water content is less than a monolayer, are
not constrained to achieve the layer registration necessary to encapsulate a cation
between opposing surface cavities, nor can they be used to simulate the process of
wetting and drying. As a result, our simulations do not lead to cation fixation within
surface cavities.
The difference in Cs+ behavior within the interlayer of the three clay minerals,
which has an electrostatic explanation, may be contrasted with the behavior of the much
smaller Li+ ion in similar low water content systems (Greathouse and Sposito, 1998).
The structural hydroxyl groups located at the bottom of the six-oxygen cavities of
hectorite are oriented perpendicularly to the clay surface, such that the positive portion of
this polar group repels the interlayer cation and discourages it from entering the cavity
(Figure 3.6). It is likely that this electrostatic repulsion is exaggerated in our simulations
as a result of imposing a rigid clay structure, which prevents structural hydroxyl groups
from relaxing into non-perpendicular orientations. The octahedral charge sites of
153
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
hectorite, submerged within the clay layer, are too far away to counteract this repulsion
effectively. Cs+thus maintain midplane positions in the interlayer as a result of the
distant octahedral charge and structural hydroxyl repulsion (Figure 3.6), as did Li+
(Greathouse and Sposito, 1998). The smaller size of Li+ provides it with sufficient
energy of hydration to form an outer sphere complex within hectorite (Greathouse and
Sposito, 1998), while the larger Cs+ forms an inner sphere complex with the clay surface.
Figure 3.6. Representations of preferred Cs+positions above six-oxygen cavities for a)
hectorite, b) beidellite, and c) montmorillonite. Interlayer species are represented as in
Figure 3.1.
Though the structural hydroxyls of beidellite are oriented at a shallow angle
relative to the ab plane (Giese, 1979), thus reducing electrostatic repulsion, the
tetrahedral charge sites present in the mineral are located directly beneath oxygen triads,
making them attractive locations for interlayer cations (Figure 3.6). Molecular modeling
of Li-beidellite resulted in Li+positioned directly over oxygen triads (Greathouse and
Sposito, 1998). In contrast, cations within Cs-montmorillonite interlayers can inhabit
surface cavities as well as positions near oxygen triads. Only montmorillonite contains
structural hydroxyl groups with shallow orientations relative to the surface (Giese, 1979)
and a small proportion of tetrahedral charge, conditions that allow Cs+to occupy
154
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
positions within or near six-oxygen cavities and close to oxygen triads (Figure 3.6).
Although Li+ions within montmorillonite also displayed greater variation in position than
their counterparts in hectorite and beidellite, they did not enter surface cavities, because
these ions existed either as inner sphere surface complexes near oxygen triads, tightly
bound to substitution sites, or as fully hydrated outer sphere complexes too large to enter
a cavity (Greathouse and Sposito, 1998). The difference in solvation of Cs+ and Li+, a
result of their contrasting ionic radii, thus modifies the electrostatic effect of differing
smectite minerals.
It is possible that the ability to inhabit sites near oxygen triads as well as near or
within six-oxygen cavities (Figure 2.11), promotes the optimized geometry required for
water sharing interactions to develop and perpetuate within the Cs-montmorillonite
interlayer over the timescale of our MD simulations. While Cs+ within beidellite does not
show this flexibility, the cations show another type of spatial variation, in that they are
segregated into two layers along the z axis (Figure 2.11), one containing ions attracted to
charge sites in the upper mineral layer, the other containing ions attracted to charge sites
in the lower mineral layer. Although limited to positions close to oxygen triads, this
vertical separation of cations may enhance the potential for sharing water molecules. In
contrast, the charge of hectorite is located solely in the octahedral sheet, which compels
Cs+ions to remain in stable, midplane positions near oxygen triads (Figure 2.11),
possibly inhibiting the formation of sharing configurations.
Extensive water sharing occurs in Cs-montmorillonite with 2/3 water monolayer
(Figure 3.7). In the lower right of the animation we see a complex consisting of four Cs+
155
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
and five shared water molecules. This complex is stable over the 800 ps timespan of the
simulation. The ions are located in close proximity to one tetrahedral and one octahedral
charge site, and are near other tetrahedral and octahedral charge sites. In addition, we see
the two Cs+not involved in the shared configuration roaming over the clay surface in a
typical diffusional manner, unlike the hovering motions seen in the other systems. Plots
of center-of-mass mean-square displacement vs. elapsed time for the two more mobile
Cs+are nonlinear (see Section 2.3.2), which may indicate some form of entrapment.
These Cs+ can be seen to engage in ion exchange, as first one and then the other is
attracted to an octahedral charge site submerged within the clay structure, located in the
upper left comer of the simulation cell (Figure 3.7).
Cs+are not the only interlayer species attracted to specific sites within the clay
mineral surface. Water molecules in Cs-hectorite with 1/3 water monolayer (Figure 3.2)
frequently dip into the six-oxygen cavities, but rarely remain for more than a hundred
picoseconds. The negative dipole of this interlayer species is attracted to the structural
hydroxyl group of hectorite which rests at the bottom of each cavity (Figure 3.6), creating
a population of water molecules with their dipole moments pointing away from the
mineral surface. Such a water orientation was found in MC simulations of Cs-hectorite
(Figure 2.2). However, because the charge of the positive pole of a water molecule is
smaller than that of Cs+, the interaction is not very strong. As a result, water molecules
visit the cavities for tens of picoseconds or less. Simulations of Li-hectorite with low
water content featured water molecules which inhabited cavities for the entire length of
the 200 ps MD calculation, likely due to entrapment caused by the substantially smaller
layer spacing of the mineral (Greathouse and Sposito, 1998).
156
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Figure 3.7. A series of snapshots taken from the Cs-montmorillonite, 2/3 water
monolayer MD simulation animation. Interlayer species are represented as described in
Figure 3.2. The four Cs+toward the right of the animation are bound together by
extended water sharing arrangements into a structure that is stable over the 800 ps
timespan of the simulation. The two Cs+toward the left of the simulation cell (red and
brown) display continuous diffusional motion and participate in an exchange reaction,
with first one and then the other taking up a position near an octahedral charge site in the
clay layer. In image a), the cation near the charge site is highlighted with a pale color.
When this cation moves away from the charge site in image b), the pale highlight color is
replaced by red. In image c), the red cation has moved beyond the boundary of one side
of the simulation cell, and has reappeared on the opposite side, closer to the viewer, as a
result of the periodic boundary conditions imposed on the calculations. Meanwhile, the
brown cation has begun to move toward the charge site in the upper left comer of these
images. The arrival of the brown Cs+at this charge site in image d) is official when its
color pales.
157
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Surface cavities receive more attention in Cs-beidellite (Figure 3.5) because the
negative charge sites just below the mineral surface can attract the positive dipole of a
water molecule just as they attract Cs+(Figure 3.8). Water molecules so attracted arrange
themselves with a hydrogen ion directed toward the clay surface, as seen in distributions
of water dipolar orientation from MC calculations (Figure 2.2). The orientation of
hydroxyl groups at a low angle relative to the plane of the clay surface (Giese, 1979)
allows some water molecules in the Cs-beidellite simulation to remain in the surface
cavities for hundreds of picoseconds. Simulations of Li-beidellite with a similarly low
water content did not contain water molecules which sample the six-oxygen cavities
(Greathouse and Sposito, 1998). The Li-beidellite system had a larger layer spacing than
that of Li-hectorite which allowed all water molecules to coordinate around the strongly
hydrating Li+ion (Greathouse and Sposito, 1998).
Figure 3.8. Side view of a water molecule nestling into a six-oxygen cavity in the Cs-
beidellite system with 1/3 water monolayer.
158
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
In Cs-montmorillonite with 1/3 water monolayer, we see relatively little water
entry into surface cavities (Figure 3.4). Because montmorillonite contains 1/3 as much
tetrahedral charge as beidellite, and its hydroxyl groups are oriented at a low angle with
respect to the clay surface (Giese, 1979), the surface does not strongly attract either the
positive or the negative poles of water molecules. The observation that Cs+ within
montmorillonite can enter six-oxygen cavities, unlike Cs+ within hectorite and beidellite,
requires that cations feel a stronger attraction than water molecules for these locations.
However, MC data presented in Chapter 2 indicate that a portion of the water molecules
within montmorillonite have a similar orientation to the fraction of water molecules in
beidellite that are attracted to the mineral surface (Figure 2.2). Like Li-beidellite, Li-
montmorillonite MD calculations did not show water molecules within six-oxygen
cavities (Greathouse and Sposito, 1998). This system also featured a larger layer spacing
relative to Li-hectorite, allowing for more complete solvation of the strongly hydrating
ions.
The amount of interlayer water modifies the extent of interaction of water
molecules with the mineral surface. Animations of both Cs-montmorillonite and
hectorite with 2/3 water monolayer (Figures 3.7 and 3.9) show a crowded interlayer
featuring more extensive and lengthier sampling of the surface cavities by water
molecules. As noted above, water movement in these systems is faster and more
diffusional, and ions exhibit more rapid jump diffusion. Why do the 2/3 water monolayer
systems, though more tightly packed than the 1/3 water monolayer systems, exhibit more
movement among their interlayer species? Competition between ions and water
molecules for surface sites may be the answer. An example of such competitive action
159
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
takes place in the animation of the Cs-hectorite system (Figure 3.9). A Cs+displaces a
water molecule from a surface cavity near which it has hovered for tens of picoseconds.
After only a few picoseconds, this cation is displaced in turn by another water molecule,
as might be predicted given cation preference for locations over oxygen triads, rather than
six-oxygen cavities, within hectorite. Similar and more subtle interactions across the clay
surface would lead to greater movement of both water and Cs+. The electrostatic
attraction between Cs+and the polar water molecules, responsible for the organization of
the partial hydration shell, further encourages these competitive interactions by
maintaining the two species in close proximity to each other as well as to the surface sites
to which they are attracted.
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
160
Figure 3.9. Snapshots of the Cs-hectorite, 2/3 water monolayer MD simulation
animation. Interlayer species are represented as described in Figure 3.2. The highlighted
Cs+ (red) can be seen to compete with surrounding water molecules for favorable
positions over a six-oxygen cavity, when it moves forward to occupy a surface cavity site
closer to the viewer, then returns to the oxygen triad site it prefers. A water molecule can
be seen hovering over the cavity in question both before and after its occupation by Cs+.
3.4 Conclusions
Animations of 800 ps MD simulations provided new information concerning the
structure and dynamics of the hydrated Cs-smectite interlayer region. A comparison of
the five animations reveals the behavior of interlayer species as modified by water
content and charge site location. Cs-smectite systems feature ions as inner sphere
161
Re pr oduc e d with per mi ssi on of t he copyright owner. Fur t her reproduct i on prohibited without per mi ssi on.
complexes, hovering near charge sites and occasionally using jump diffusional
movements to relocate. Higher water contents encourage more rapid movements for both
Cs+ and the water molecules which solvate them. Different mineral charge patterns and
structural hydroxyl group orientations create interlayer dynamics with different
propensities for water molecule sharing between two Cs+. A low orientation angle for
structural hydroxyl groups relative to the mineral surface (Giese, 1979), combined with a
reduced amount of tetrahedral charge, as in montmorillonite, permits interlayer cations to
occupy positions above and within six-oxygen cavities with frequency. This may allow
ions a greater flexibility in placement, and therefore increased ability to enter into water-
sharing arrangements. Tetrahedrally-charged clays like beidellite provide a different
form of flexibility in ion position, due to the separation of ions into two layers along the
z axis. This segregation may encourage formation of water sharing arrangements as well.
The combination of the octahedral charge and the perpendicular orientation of hydroxyl
groups relative to the hectorite surface results in Cs+ held in stable, midplane locations
above oxygen triads on the clay surface, hindering cations from entering into water
sharing arrangements. The water sharing phenomenon predicted by our animations
reaches an extreme state in the Cs-montmorillonite hydrate with 2/3 water monolayer in
which four Cs+ are held together throughout the 800 ps simulation period in an extended
complex of ions and shared water molecules. Our animations also document the
interaction of water molecules with the clay mineral surface, which increases with higher
water content. Animation reveals competition between water molecules and Cs+ for sites
on the mineral surface. Thus, animation of our MD data provided substantial information
not available from traditional simulation methods.
162
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
3.5 Appendix: Sample File Formats for Trajectory and Lattice Coordinates
As described in Section 3.2.2, a trajectory input file had the following format:
0 water
x y z <
0.36131 1.08528 0.28944
0.05723
5 1.11682 0.10385
0.49193 0.64485 0.02563
3 5 7
0.25037
1 1.33298 0.05067
0.09124
2 1.26203 0.10204
1 water
x y z a
0.00577
1.44 7.81352 1.57747 2
0.14468
1.44 7.89855 1.38648 2
0.05088
1.44 7 .85525 1.59731 5
1.44 7.84451 1.94947 0.02985
0.13979
1.44 7.94765 2.30518 3
Coordinates for the stationary surface oxygen plane were inserted using this format:
48
-8.919820.318986-2.93049 1.44
-7.59982-1.96108-2.93049 1.44
-10.2398-1.96108-2.93049 1.44
-6.27982-4.25101-2.93049 1.44
-10.2398-6.53108-2.93049 1.44
3.6 References
Allen M. P. and Tildesley D. J. (1987) Computer Simulation o f Liquids. Oxford
University Press.
Anderson S. J. and Sposito G. (1991) Cesium-adsorption method for measuring
accessible structural surface charge. Soil Sci. Soc. Am. J. 5 5, 1569-1576.
Atkins P. W. (1994) Physical Chemistry. 5th ed. W. H. Freeman and Company.
Bailey G. W., Akim L. G., and Shevchenko S. M. (2001) Predicting chemical reactivity
of humic substances for minerals and xenobiotics. Use of computational chemistry,
scanning probe microscopy and virtual reality. In Humic Substances and Chemical
Contaminants (ed. C. E. Clapp, M. H. B. Hayes, N. Senesi, P. R. Bloom, and P. M.
Jardine), pp. 41-72. Soil Science Society of America, Inc.
Blaney F., Edge C., Phippen R., and Burt C. (1995) Molecular surface comparison using
Anaconda. In Data Visualization in Molecular Science: Tools for Insight and Innovation
(ed. J. E. Bowie), pp. 99-129. Addison-Wesley Publishing Company, Inc.
163
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
Brickmann J. (1999) Movies, accessed 4 June 2003 from:
http: //www.pc.chemie .tu-darmstadt. de/research/molcad/movie, shtml.
Buhl M. and Parrinello M. (2001) Medium effects on 51V NMR chemical shifts: A
density functional study. Chem. Eur. J. 7, 4487-4494. Supplemental animation: Buhl M.
(2002) Michael Buhl - Current research interests, accessed 4 June 2003 from:
http://www.mpi-muelheim.mpg.de/kofo/institut/arbeitsbereiche/buehl/buehl-
research.html.
Chang F.-R. C., Skipper N. T., Refson K., Greathouse J. A., and Sposito G. (1999)
Interlayer molecular structure and dynamics in Li-, Na-, and K-montmorillonite-water
systems. In Mineral-Water Interfacial Reactions: Kinetics and Mechanisms (ed. D. L.
Sparks and T. Grundl), pp. 88-106. American Chemical Society.
Chang F.-R. C., Skipper N. T., and Sposito G. (1995) Computer simulation of interlayer
molecular structure in sodium montmorillonite hydrates. Langmuir 11, 2734-2741.
Cheng H.-P. and Krause J. L. (1997) The dynamics of proton transfer in FL0 2 +. J. Chem.
Phys. 107, 8461-8468. Supplemental animation: Choy T. S. (1999) Molecular dynamics:
Animated 3D, accessed 4 June 2003 from:
http://www.phys.ufl.edu/~tschoy/r2d2/MD/index.html.
de Groot B. L. and Grubmuller H. (2001) Water permeation across biological
membranes: Mechanism and dynamics of Aquaporin-1 and GlpF. Science 294, 2353-
2357. Supplemental animation: de Groot B. L. and Grubmuller H. (2001) Image/Movie
Gallery, accessed 4 June 2003 from:
http://www.mpibpc.gwdg.de/abteilungen/071/bgroot/gallery.html.
Giese R. F. (1979) Hydroxyl orientations in 2:1 phyllosilicates. Clays Clay Miner. 27,
213-223.
Gillilan R. E. and Ripoll D. R. (1995) Visualizing enzyme electrostatics with IBM
Visualization Data Explorerm. In Data Visualization in Molecular Science: Tools for
Insight and Innovation (ed. J. E. Bowie), pp. 61-81. Addison-Wesley Publishing
Company, Inc.
Greathouse J. and Sposito G. (1998) Monte Carlo and molecular dynamics studies of
interlayer structure in Li(H2 0 )3-smectites. J. Phys. Chem. B 102, 2406-2414.
Gruia A. D., Fischer S., and Smith J. C. (2003) Molecular dynamics simulation reveals a
surface salt bridge forming a kinetic trap in unfolding of truncated Staphylococcal
nuclease. Proteins 50, 507-515. Supplemental animation: Fischer S. (2002) Stefan
Fischer: Research, accessed 4 June 2003 from:
http://www.iwr.uni-heidelberg.de/groups/biocomp/fischer/research/snase_unfolding.html.
Harris L. F., Sullivan M. R., Popken-Harris P. D., and Hickok D. F. (1996) Molecular
dynamics simulation of the estrogen receptor protein in complex with a non-consensus
164
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
estrogen response element DNA sequence in a 10 Angstrom water layer. WWW J. Biol. 2,
2, accessed 4 June 2003 from: http://www.epress.com/w3jbio/vol2/harris2/.
Hubona G. S., Shirah G. W., and Fout D. G. (1997) The effects of motion and stereopsis
on three-dimensional visualization. Int. J. Human-Computer Studies 47, 609-627.
Kirchner B. and Hutter J. (2002) The structure of a DMSO-water mixture from Car-
Parrinello simulations. Chem. Phys. Lett. 364, 497-502. Supplemental animation:
Kirchner B. (2002) Barbara Kirchner, accessed 4 June 2003 from:
http://pciwww.unizh.ch/pci/hutter/barbara/barbara.html.
Laaksonen L. and Rosenholm J. B. (1993) Molecular dynamics simulations of the
water/octanoate interface in the presence of micelles. Chem. Phys. Lett. 216, 429-434.
Supplemental animation: Laaksonen L., Micelle paper Chem. Phys. Lett. 216 (1993) 42,
accessed 4 June 2003 from: http://www.csc.fi/lul/chem/micelles/micelles.html.
Leonard J. M. (1995) Application Visualization System. In Data Visualization in
Molecular Science: Tools for Insight and Innovation (ed. J. E. Bowie), pp. 11-40.
Addison-Wesley Publishing Company, Inc.
Lindan P. J. D., Harrison N. M., Holender J. M., and Gillan M. J. (1996) First-principles
molecular dynamics simulation of water dissocation on TiO2 ( 1 1 0 ). Chem. Phys. Lett.
261, 246-252. Supplemental animation: Lindan P. J. D. Ti02 Surface chemistry from
first principles, accessed 4 June 2003 from:
http://www.dl.ac.uk/TCSC/MatSci/Projects/Surface_Chemistry/tio2.html.
Ludemann S. K., Gabdoulline R. R., Lounnas V., and Wade R. C. (2001) Substrate
access to cytochrome P450cam investigated by molecular dynamics simulations: An
interactive look at the underlying mechanisms. Internet J. Chem. 4, 6 , accessed 4 June
2003 from: http://ijc.chem.trinity.edU/articles/2001v4/6/.
Mackay D. (1995) Visualizing electrostatic interactions with Insight II. In Data
Visualization in Molecular Science: Tools f o r Insight and Innovation (ed. J. E. Bowie),
pp. 161-182. Addison-Wesley Publishing Company, Inc.
Marry V., Turq P., Cartailler T., and Levesque D. (2002) Microscopic simulation of
structure and dynamics of water and counterions in a monohydrated montmorillonite. J.
Chem. Phys. 117, 3454-3463.
Matsuoka O., Clementi E., and Yoshimine M. (1976) Cl study of the water dimer
potential surface. J. Chem. Phys. 64, 1351-1361.
Murray H. H. and Daage M. (1995) The CAChe WorkSystemand molecular
mechanics: Coordination of DBT on M0 S2 . In Data Visualization in Molecular Science:
Tools for Insight and Innovation (ed. J. E. Bowie), pp. 215-245. Addison-Wesley
Publishing Company, Inc.
165
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
Nakano M., Kawamura K., and Ichikawa Y. (2003) Local structural information of Cs in
smectite hydrates by means of an EXAFS study and molecular dynamics simulations.
Appl. Clay Sci. 23, 15-23.
Noon W. H., Kong Y., and Ma J. (2002) Molecular dynamics analysis of a buckyball-
antibody complex. Proc. Natl. Acad. Sci. USA 99(suppl. 2), 6466-6470, accessed 4 June
2003 from: http://www.pnas.org/cgi/content/fiill/99/suppl_2/6466.
Nye P. H. (1979) Diffusion of ions and uncharged solutes in soils and soil clays. Adv.
Agron. 31, 225-272.
Ohtaki H. and Radnai T. (1993) Structure and dynamics of hydrated ions. Chem. Rev. 93,
1157-1204.
Piana S., Carloni P., and Parrinello M. (2002) Role of conformational fluctuations in the
enzymatic reaction of HIV-1 protease. J. Mol. Biol. 319, 567-583. Supplemental
animation: accessed 4 June 2003 from:
http: //www. sissa. it/sbp/bc/publications/supplement/animazione.MPG.
Ranck J. P. (1995) Visualization and representation in chemical education using
HyperChem. In Data Visualization in Molecular Science: Tools for Insight and
Innovation (ed. J. E. Bowie), pp. 247-264. Addison-Wesley Publishing Company, Inc.
Refson K. (2000) MOLDY: A portable molecular dynamics simulation program for serial
and parallel computers. Computer Phys. Commun. 126(3), 310-329.
Richards A. J., Maginn S. J., Leusen F. J. J., and Bick A. (1995) Cerius2 molecular
modeling software for industrial applications. In Data Visualization in Molecular
Science: Tools for Insight and Innovation (ed. J. E. Bowie), pp. 183-214. Addison-
Wesley Publishing Company, Inc.
Schroth B. K. and Sposito G. (1997) Surface charge properties of kaolinite. Clays Clay
Miner. 45, 85-91.
Sherman D. M. and Collings M. D. (2002) Ion association in concentrated NaCl brines
from ambient to supercritical conditions: Results from classical molecular dynamics
simulations. Geochem. Trans. 3, 102-107, accessed 4 June 2003 from:
http://www.rsc.org/CFmuscat/intermediate_abstract.cfm?FURL=/ej/GT/2002/b208671a.
PDF &T YP=EONL Y.
Skipper N. T. (1996) MONTE User's Manual. Department of Physics and Astronomy,
University College London, England.
Skipper N. T., Sposito G., and Chang F. R. C. (1995) Monte Carlo simulation of
interlayer molecular structure in swelling clay minerals. 2. Monolayer hydrates. Clays
Clay Miner. 43, 294-303.
166
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Smith D. A. (1995) Quantum chemistry visualization using the AVS Chemistry
Viewer. In Data Visualization in Molecular Science: Tools for Insight and Innovation
(ed. J. E. Bowie), pp. 41-60. Addison-Wesley Publishing Company, Inc.
Smith D. E. (1998) Molecular computer simulations of the swelling properties and
interlayer structure of cesium montmorillonite. Langmuir 14, 5959-5967.
Smith D. E. and Dang L. X. (1994) Computer simulations of cesium water clusters - Do
ion water clusters form gas-phase clathrates? J. Chem. Phys. 101, 7873-7881.
Soper A. K. and Phillips M. G. (1986) A new determination of the structure of water at
25 C. Chem. Phys. 107, 47-60.
Sutton R. and Sposito G. (2002) Animated molecular dynamics simulations of hydrated
caesium-smectite interlayers. Geochem. Trans. 3, 73-80, accessed 4 June 2003 from:
http://www.rsc.org/CFmuscat/intermediate_abstract.cfm?FURL=/ej/GT/2002/b204973m.
PDF&TYP=EONL Y.
Tomney H. and Burton P. F. (1998) Electronic journals: A study of usage and attitudes
among academics. J. Info. Sci. 24,419-429.
Tse J. S. (2002) Ab initio molecular dynamics with density functional theory. Annu. Rev.
Phys. Chem. 53, 249-290.
Tufte E. R. (1990) Envisioning Information. Graphics Press.
Wang J., Kalinichev A. G., Kirkpatrick R. J., and Hou X. (2001) Molecular modeling of
the structure and energetics of hydrotalcite hydration. Chem. Mater. 13, 145-150.
Supplemental animation: Wang J. (2002) Movies from MD, accessed 4 June 2003 from:
http://www.geology.uiuc.edu/~jwang7/RD/movie.html.
Ware C. (2000) Information Visualization: Perception f o r Design. Academic Press.
Ware C. and Franck G. (1996) Evaluating stereo and motion cues for visualizing
information nets in three dimensions. ACM Trans. Graphics 15, 121-140.
Yang C., Jas G. S., and Kuczera K. (2001) Structure and dynamics of calcium-activated
calmodulin in solution. J. Biomol. Struc. Dyn. 19, 247-271. Supplemental animation:
Yang C. and Kuczera K., Images and animations from MD simulations of calmodulin in
solution, accessed 4 June 2003 from:
http://oolung.chem.ukans.edu/~kuczera/lcll/lcll.html.
Yang C. and Kuczera K. (2002a) Molecular dynamics simulations of calcium-free
calmodulin in solution. J. Biomol. Struc. Dyn. 19, 801-820. Supplemental animation:
Yang C. and Kuczera K., Images and animations from MD simulations of calmodulin in
solution, accessed 4 June 2003 from:
http://oolung.chem.ukans.edU/~kuczera/lcfd/l cfd.html.
167
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Yang C. and Kuczera K. (2002b) Molecular dynamics simulations of a calmodulin-
peptide complex in solution. J. Biomol. Struc. Dyn. 20, 179-197. Supplemental
animation: Yang C. and Kuczera K., Images and animations from MD simulations of
calmodulin in solution, accessed 4 June 2003 from:
http://oolung.chem.ukans.edu/~kuczera/lcdl/lcdl.html.
Re pr oduc e d with per mi ssi on of t he copyright owner. Fur t her reproduct i on prohibited without per mi ssi on.
Chapter Four
Molecular simulation of a DOM molecule: Effects of hydration and deprotonation
4.1 Introduction
Molecular modeling techniques may prove useful in elucidating soil organo-
mineral interactions on a molecular scale. To simulate interactions between minerals and
the complex, heterogeneous natural organic fractions known as humic substances, a
model organic molecule which can reproduce many of the properties exhibited by humic
matter is needed. Presented in Chapter 4 is a review of current experimental data
regarding the molecular structure of humic substances, followed by a series of
simulations providing evidence that a model organic molecule developed by Schulten
(1999a) is capable of mimicking many humic properties under appropriate modeling
conditions. Chapter 5 contains descriptions of simulations of two forms of this model
organic molecule within the interlayer region of hydrated Ca-montmorillonite.
Humic substances are among the most widely distributed organic materials on the
planet (Stevenson, 1994). This complex and poorly understood assembly of organic
leftovers contributes many vital properties to soils, including the capacity for cation
exchange and buffering, which stabilizes nutrient and pH levels; the ability to form
organo-mineral complexes, which improves soil structure and water retention and
reduces erosion; as well as the ability to decompose soil minerals, which results in
releases of nutrients and creation of secondary minerals (Stevenson, 1994). In addition,
humic substances are capable of sequestration, transportation, and oxidative or reductive
transformation of numerous organic xenobiotic molecules, trace gases, and trace metal
169
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
contaminants in soil and water systems (Stevenson, 1994). Given the ubiquitous nature
of humic matter, as well as its essential role in numerous interactions of environmental
relevance, efforts to characterize this complex material began centuries ago, and continue
to this day (Stevenson, 1994).
Early concepts of humic matter, informed by the developing field of polymer
science, posited that humic substances were randomly coiled, macromolecular polymers,
which formed elongated shapes in basic or low ionic strength solutions, and coils in
acidic or high ionic strength solutions (Piccolo, 2001). Faithful adherents to the polymer
model of humic materials emphasize several pieces of evidence favoring this view (Swift,
1999), including the careful ultracentrifugation study of a Sapric Histosol HA by
Cameron et al. (1972), in which the sedimentation velocity technique was used to
monitor the moving solute-sedimentation boundary of several HA subfractions
containing humic molecules of reduced polydispersity, or similar molecular mass, during
high speed centrifugation. The results of this study suggested that an average humic
molecule had a molecular mass of 20,000 - 50,000 Da, a radius of gyration of 4 - 10 nm,
and a random coil conformation.
However, recent information gathered using a variety of modem spectroscopic,
microscopic, and soft ionization techniques is inconsistent with the macromolecular
model of humic substances. Critical analyses of the soil processes active in formation
and preservation of humic matter also cast doubt on the polymer model (Burdon, 2001;
Piccolo, 2001). A new model of humic molecules has emerged, that of the
supramolecular association, in which many small and diverse organic molecules form
clusters linked by H-bonds and weak hydrophobic interactions (Piccolo, 2001). Often
170
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
considered a corollary to this model is the concept of pseudomicellar structure, in which
an association of organic molecules organizes itself in the presence of water to form a
hydrophilic exterior region in contact with the solvent, and a hydrophobic interior region
isolated from water molecules (von Wandruszka et al., 1999). Recent experimental work
on the functional groups and larger scale molecular fragments making up these humic
molecules has resulted in more complete characterization of humic substances, that
recognizes a) the importance of both aromatic rings and alkyl chains to humic backbone
structure, b) the diversity of reactive, oxygenated functional groups, c) the prevalence of
amide N, and the presence of cyclic N functional groups, d) the existence of both highly
oxidized and highly reduced forms of S, and e) the occurrence of humic fragments
derived from lipids, lignin, proteins and carbohydrates, apparently in relatively
independent forms, and exhibiting different levels of mobility or rigidity. A summary of
experimental results concerning each of these conclusions is presented below, followed
by a review of current molecular models of humic substances, with particular emphasis
on the model of dissolved organic matter (DOM) designed by Schulten (1999).
4.1.1 Humic Substances as Supramolecular Associations Displaying Pseudomicellar
Structure
Measuring molecular mass is not a simple task when the heterogeneous collection
of molecules making up a humic fraction is the subject of investigation. For example,
average molecular masses of HAs differ by several orders of magnitude, depending on
the methods used, and are often confounded by the polydisperse nature of the material
(Stevenson, 1994). Values for FAs show less variability, and generally fall in the range
171
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
of hundreds to a few thousand Da. The inconsistency seen in measurements of HA
molecular mass suggests that experimental conditions may cause substantial alteration of
humic molecules, affecting experimental data. The influence of solution pH, cation type
and amount, and C concentration on the size and shape of HAs and FAs adsorbed to
muscovite from surrounding aqueous solutions can be seen in atomic force microscopy
(AFM) images (Maurice and Namjesnik-Dejanovic, 1999; Plaschke et al., 1999;
Namjesnik-Dejanovic and Maurice, 2000). The 1 0 - 2 0 nm diameters of the humic
building blocks, shaped like flattened spheres, which coalesce to form the rings, chains,
and networks seen in these studies under varying solvent conditions, are not inconsistent
with the polymer model of humic substances. However, no evidence is found to support
the presence of coiled structures for individual molecules under acidic or high ionic
strength conditions, or rod-like structures under basic or low ionic strength conditions.
Piccolo (2001 and references therein) and colleagues provide evidence, first from
gel permeation chromatography (GPC), then from high pressure size exclusion
chromatography (HPSEC) in conjunction with a refractive index detector, that the
apparent size of humic molecules changes drastically with additions of simple organic
acids, as compared to HC1. In addition, the absorbances recorded in ultraviolet-visible
spectra of organic acid-treated humic substances were substantially reduced, evidence of
hypochromism, or decreased absorbance due to increased distances between absorbing
structures. Evidently, the sizes of humic molecules are not decreasing due to tight
coiling, as suggested by the polymer model of humic substances, but to the
disaggregation of a cluster of smaller molecules, consistent with the idea of humic
substances as supramolecular associations. Comparison of the effects of different
172
Reproduced with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
substances on humic fractions with different relative hydrophobicities indicates that
aggregate disruption is greatest when more hydrophobic humic groups are combined with
organic molecules containing both hydrophobic and hydrophilic elements. This pattern
provides evidence for the existence of humic associations held together under neutral pH
conditions by hydrophobic interactions, which are more easily disrupted when simple
organic molecules penetrate the hydrophobic humic interior and form higher energy
intermolecular H-bonds, breaking apart the large hydrophobic associations into smaller,
more energy-rich H-bond associations. The low hydrophobicity and high charge of FAs,
relative to HAs, would not support substantial hydrophobic or H-bonded associations,
accounting for the experimental observation of minimal change to the molecular size of
FAs under a variety of solution conditions (Piccolo et al., 2001). However, recent
criticism leveled at this body of work alleges flaws in both technique and interpretation
(Perminova, 1999; Swift, 1999), and recent experiments by Varga et al. (2000) using a
similar HPSEC technique reveal the confounding effect of drastic changes to pH in the
column with addition of organic acids. Piccolo (2001) provides responses to many of
these points.
Perhaps a new method for determination of molecular mass can be used to test the
idea that humic substances are composed of small molecules, rather than polymer-like
macromolecules. Soft desorption ionization techniques such as electrospray ionization
(ESI), or the evaporation of charged droplets (Cech and Enke, 2001), and laser desorption
(LD), or the creation of a precise thermal energy gradient (Brown et al., 1998), volatilize
large ions for subsequent identification with mass spectrometry (MS), and can provide
positive ion mass/charge distributions of organic samples that may represent the mass
173
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
distribution of humic molecules, assuming most cations are singly charged (Brown et al.,
1998). After optimization of experimental conditions to reduce both fragmentation and
aggregation of humic ions, those detected by ESI-MS have masses of well under 10,000
Da, and distributions often feature number or mass averages of around 1,000 (Leenheer et
al., 2001) or 2,000 Da (Brown and Rice, 2000a; Stenson et al., 2002). Laser desorption-
MS provides mass/charge distributions with number averages closer to 500 Da, perhaps
due to fragmentation, decreased ionization efficiency, or poor resolution of signals
beyond 1,000 Da (Brown et al., 1998; Brown and Rice, 2000b). These ESI- and LD-MS
average masses are far lower than those measured by traditional methods (Stevenson,
1994), providing further evidence favoring the model of humic substances as aggregates
of small molecules.
Multidimensional NMR techniques have begun to provide significant insight into
the structure of humic substances (Simpson, 2001). Two-dimensional diffusion ordered
spectroscopy (DOSY) involves use of a pulsed field gradient that modifies the NMR
signals of nuclei such that the intensity of a resonance can be related to its diffusion
coefficient, and therefore to its hydrodynamic radius and approximate molecular mass.
By contrasting 2D DOSY NMR spectra of humic substances dissolved in D2 O and
dimethyl sulfoxide (DMSO) solvents, Simpson et al. (2001a, 2002) have observed
different diffusion coefficients for chemically distinct molecules in numerous samples,
the largest of which have molecular masses of around 1500 Da. The relatively small and
independent molecules observed in these organic samples, which had been extensively
exchanged to remove metal cations, are consistent with the concept of humic substances
as collections of diverse small molecules, perhaps bound together via complexation
174
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
reactions (Simpson et al., 2002). Liquid chromatography of the same samples, combined
with NMR spectroscopy, indicated substantial separation of different structural
groups, a result of aggregate disruption after removal of the metals (Simpson et al.,
2002).
A corollary of the supramolecular association model of humic substance is the
pseudomicellar model, in which intra- and intermolecular organization of humic
molecules produces interior hydrophobic regions separated from aqueous surroundings
by exterior hydrophilic layers. The fluorescence of pyrene in solutions of humic
substances is enhanced within a few hours by the addition of Mg2+, and then decays over
a period of several days (Engebretson and von Wandruszka, 1997, 1998; von
Wandruszka and Engebretson, 2001). The initial increase in fluorescence is thought to
result from cation enhanced formation of pseudomicellar hydrophobic domains due to
charge neutralization and cation bridge formation, resulting in protection of the pyrene
from quenching anions like Br' (von Wandruszka et al., 1999). The later drop in
fluorescence occurs as cations slowly migrate from these positions to higher energy,
inner sphere coordinated positions within the humic substance, allowing the newly
formed hydrophobic regions to dissociate. Measurements of the polarity experienced by
2 "b
the pyrene probe show an initial decrease in polarity with addition of Mg (von
Wandruszka et al., 1999), confirming the existence of temporary hydrophobic
environments within humic substances. Nanny and Kontas (2002) used the fluorescence
of 6-propionyl-2-dimethylaminonaphthalene (Prodan), another probe sensitive to
polarity, to investigate the formation of pseudomicelles in a variety of humic substances
over periods of days to weeks. Prodan experiences a polarity between that of ethanol and
175
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
cyclohexane within humic solutions, a strong indication of the formation of
pseudomicelles. More micellar structures form under neutral to acidic pH conditions, and
FAs form these structures more quickly than FLAs.
A hydrophobic electron spin resonance (ESR) probe known as 5-SASL (stearic
acid with nitroxide free radical in position 5 of the hydrocarbon chain) was used by
Ferreira et al. (2001) to monitor hydrophobic interactions with humic substances as a
function of pH. Electron spin resonance spectroscopy measures the signals produced by
unpaired electrons in paramagnetic substances when placed in a magnetic field and
subjected to microwave radio frequency radiation (Senesi, 1992). Under acidic
conditions (pH < 5), ESR spectra of 5-SASL indicated that it had become immobilized
within the solution in a matter of hours to days, due to hydrophobic interactions with
humic functional groups, while at other pH levels the spectra indicated the molecule
remained mobile. The reversibility of this response with changes to pH further supports
the idea of humic molecules that form pseudomicellar structures, especially under low pH
conditions.
Chien et al. (1997) used HA solutions containing atrazine labeled with a F atom,
and examined with 19F NMR, to assess the effect of paramagnetic probes with different
hydrophobicities on the intensity of the F signal. The hydrophilic Gd-EDTA anion
remained in aqueous solution and caused no paramagnetic relaxation of the signal, while
the hydrophobic 2,2,6,6-tetramethyl-l-piperidinyloxy (TEMPO), which also sorbs within
humic micellar structures, produced rapid relaxation. These data indicate that atrazine
occupies domains accessible only to hydrophobic molecules, consistent with proposed
micellar structure. A more complex NMR spectroscopic technique was utilized by
176
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
Simpson et al. (2001b), who obtained high resolution (HR) MAS NMR solution state
spectra of samples that were not fully soluble, thus selectively recording signals from
those species in contact with the solvent. Humic substances suspended in D2O showed a
reduced aromatic signal as compared both to those suspended in the more highly
penetrating DMSO solvent, and to soluble sample extracts, providing further evidence for
hydrophobic regions of protection. Conte et al. (1997) used a pseudomicellar argument
to explain the reduction in alkyl signal in liquid state 13C NMR spectra relative to solid
state CP/MAS spectra. The hydrophobic interactions of micellar aggregates present in
solution should inhibit alkyl motion, lowering the sensitivity of the liquid state 13C NMR
technique to alkyl moieties, they reasoned (Conte et al., 1997).
Transmission electron microscopy (TEM) observation of abiotic dissolved
organic carbon solutions indicates formation of numerous 0.4 - 0.8 pm micelle-like
colloids over a 10 day period (Kemer et al., 2003). Accompanying 'H HR-MAS NMR
reveals the predominance of mobile CH2 and CH3 signals, and the absence of signals
from sugars and proteins known to be present, indicating significantly constrained
rotational motions. This variation in mobility with structure provides further evidence for
a self-organized aggregate of humic molecules. Addition of EDTA reduces aggregation
by only 2 0 %, suggesting that multivalent cation binding is not a significant force behind
the formation of these micelle-like colloids (Kemer et al., 2003).
In summary, recent evidence supports a view of humic substances as collections
of relatively small, diverse molecules that form large and dynamic associations bound by
hydrophobic interactions and H-bonds, and as a consequence create pseudomicellar
structures in aqueous surroundings. As yet, current theory does not address the changes
177
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
in molecular properties evident with separation of humic molecules into different
apparent size fractions, as mentioned in Sections 4.1.2 - 4.1.5 below.
4.1.2 Nonpolar Functional Groups o f Humic Substances
Humic substances are roughly 50% C by mass (Stevenson, 1994). Nonpolar alkyl
and aromatic groups form the backbone structures of humic molecules, to which are
attached a variety of polar functional groups typically featuring O, N, or S atoms. Early
concepts of the humic backbone, formed in response to results from chemical degradation
experiments, were primarily aromatic in nature (Stevenson, 1994). Alkyl functional
groups were thought to be a minor component of the backbone, necessary only for
connecting aromatic groups.
Beginning in the 1980s, the innovation of solid-state CP/MAS NMR spectroscopy
(Section 1.1) resulted in the detection of substantial signal caused by alkyl functional
groups, often equal to or greater than the signal caused by aromatic groups (Hatcher et
al., 1981; Preston and Ripmeester, 1982; Schnitzer and Preston, 1986; Stevenson, 1994),
forcing a significant revision of ideas concerning the relative proportions of these
nonpolar functional groups in humic substances. Mahieu et al. (1999) conducted a
review of hundreds of published 13C CP/MAS NMR spectra of whole soils and humic
extracts, and found the relative amounts of both alkyl and aromatic groups present in
humic substances to be remarkably similar for samples collected over wide distributions
of climate, land use and management, and experimental conditions. On average, HA C is
made up of around 25% ( 7%) nonpolar alkyl groups (0 - 50 ppm signals) and 30%
( 6 %) aromatic groups ( 1 1 0 - 160 ppm signals, including phenolic contributions), while
178
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
FA C is made up of around 27% nonpolar alkyl groups and a slightly smaller percentage
of aromatic groups (Mahieu et al., 1999). The high proportion of alkyl substituents
remaining after a multi-step chemical degradation process (ultrasonic disruption in
methylene chloride, followed by BF3 methanol transesterification and FH acid treatment
to destroy ether links) suggests that many of these structures are an integral part of the
humic backbone (Almendros et al., 1998).
As indicated by the figures above, it is common for FAs to be somewhat more
alkyl and less aromatic than HAs extracted from the same sediments (Preston and
Ripmeester, 1982; Stevenson, 1994; Christl et al., 2000; Gonzalez-Vila et al., 2001a),
although a few exceptions have been reported (Scheinost et al., 2001; Ussiri and Johnson,
2003). Studies of HAs separated into molecular size fractions show that alkyl content
decreases with decreasing molecular size, while aromatic content increases (Swift et al.,
1992; Shin et al., 1999; Christl et al., 2000). Relatively few studies describe the
properties of humin in comparison to HA and FA fractions, and available 13C CP/MAS
NMR spectra frequently have poor resolution due to interference by paramagnetic cations
like Fe3+. Interaction of local, excited C nuclei with these cations allows the spin of the C
nuclei to equilibrate quite rapidly, leading to line broadening and a reduction in the NMR
signal produced (Bleam, 1991). Nevertheless, published spectra indicate that alkyl C is
prevalent in humin (Preston and Ripmeester, 1982; Almendros et al., 1996; Ussiri and
Johnson, 2003), though this fraction may possess relatively more aromatic groups than
HAs taken from the same sites (Preston and Ripmeester, 1982; Ussiri and Johnson,
2003).
179
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
The preceding analyses are based on the assumption that CP/MAS NMR
spectroscopy is a quantitative technique, a poor assumption given that cross polarization
efficiency, or the transfer of nuclear spin from 'H to 13C nuclei, is reduced for
unprotonated C atoms, such as substituted aromatic C, and for highly mobile C, reducing
signal intensity (Mao et al., 1998). The presence of spinning sidebands, or satellite peaks
resulting from spinning the sample at the magic angle, also reduces the intensity of the
central band, leading to smaller peak areas for some signals. As a result, CP/MAS NMR
spectra can provide only semi-quantitative comparisons of alkyl and aromatic C (Mao et
al., 1998). Alternative direct polarization (DP) MAS NMR spectroscopy, in which the C
nuclei are excited directly, can produce quantitative but time-consuming measurements
of C functional groups (Mao et al., 1998). Studies comparing CP/MAS and DP/MAS
results show an underrepresentation of aromatic groups in the former (Mao et al., 2000;
Smemik and Oades, 2000a and b). Therefore, though alkyl groups can no longer be
considered an insignificant part of the humic backbone, it is possible that they are not as
numerous as CP/MAS data indicate.
Pyrolysis, or the application of heat to volatilize molecular fragments for
identification via MS (Schulten et al. , 1998), has been used to probe the makeup of the
nonpolar humic backbone as well. Analyses of pyrolysates indicate the presence of
numerous alkyl and aromatic groups in humic substances, and suggest the existence of a
substantial number of alkylaromatic moieties (Saiz-Jimenez, 1992, 1996; Almendros et
al., 1996; Fabbri et al., 1996; Leinweber and Schulten, 1998). Pyrolysis also provides
evidence for the existence of small quantities of alkene functional groups within the
nonpolar humic backbone (Saiz-Jimenez, 1992, 1996; Almendros et al.,1996; Lu et al.,
180
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
2000), as do ESI-MS (Kramer et al., 2001) and solvent extraction methods (Jandl et al.,
2002). The trace alkene signal is lost within the overpowering aromatic signal of NMR
spectra. Comparisons of pyrolysis data with NMR work generally confirm that pyrolysis
cannot be considered a quantitative method, and indicate that pyrolysates are enriched in
alkyl and alkylaromatic molecules (Gonzalez-Vila et al., 2001a; Hatcher et al., 2001;
Martin et al., 2001). Secondary reactions during pyrolysis, such as decarboxylation, may
produce a greater number of alkyl molecules than were present in the original humic
substance, while cyclization and aromatization may create most alkylaromatic molecules
detected (Saiz-Jimenez, 1994, 1996; van Heemst et al., 1999; Hatcher et al., 2001). In
addition, detection of pyrolysates via mass spectrometry (MS) is possible only for those
molecules volatile enough to reach the spectrometer and, therefore, selects for lower
molecular weight molecules that may not represent adequately the humic backbone.
Carbon Is (K-edge) x-ray absorption near-edge structure (XANES) spectroscopy
has been employed to examine the C functional group distribution of humic substances
with limited success. This form of x-ray absorption spectroscopy detects multiple
scattering of photoelectrons by both first and higher coordination shells around target
nuclei, providing information on the valence and coordination of a particular element
(Myneni, 2002). Aromatic signals are evident (Scheinost et al., 2001) and even appear to
be enhanced in extracted HAs, as opposed to undisturbed, mineral-associated humic
materials (Myneni et al., 1998), an indication that extraction procedures may selectively
remove a greater proportion of aromatic materials than is typical of the humic substances
in situ. However, alkyl signals are too weak to be reliably determined by this method
(Scheinost et al., 2001).
181
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
In summary, solid-state 13C CP/MAS NMR spectroscopy has revealed the
importance of alkyl functional groups to the humic backbone, although the technique is
semiquantitative and may not detect all aromatic groups present. Pyrolysis and ESI-MS
confirm the presence of substantial quantities of both forms of nonpolar C in humic
substances, while information from XANES is limited.
4.1.3 Oxygen Functional Groups o f Humic Substances
Organic O is the primary source of reactivity in humic substances, because
functional groups containing O can provide polarity or even charge to molecules. The
average molar O to C ratio for HAs is about 0.5, while for FAs it is about 0.7 (Stevenson,
1994). Given the operational basis for separation of HAs and FAs (Section 1.3), these
two values are an indication of the significant impact of O content and functional group
distribution on the chemical properties of humic substances.
Carboxyl groups contribute significant reactivity to humic substances as sources
of protons and as charge sites, and produce intense signals for a variety of spectroscopies.
13C NMR spectra of humic substances feature prominent peaks associated with carboxyl
groups (Stevenson, 1994). Quantification based on these peaks can be problematic,
however, because carboxyl signals overlap with peaks of other carbonyl functional
groups (Monteil-Rivera et al., 2000). Using CP/MAS NMR and x-ray photoelectron
(XPS) spectroscopies, the latter based on the emission of photoelectrons after x-ray
excitation of the sample (Vempati et al., 1996), Monteil-Rivera et al. (2000) found that
almost half the NMR peak in the carboxyl range was produced by amide C in the three
HAs examined. Many scientists simply lump all the signals into a single carbonyl region
182
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
for quantification (Mahieu et al., 1999). In addition, CP/MAS NMR carboxyl signals
produce significant spinning sidebands, reducing by up to 50% the CP/MAS carboxyl
peak area relative to the DP/MAS peak area (Mao et al., 2000; Smemik and Oades, 2000a
and b).
Quantitative measurements of carboxyl content can be performed using XPS
(Monteil-Rivera et al., 2000), as mentioned above, as well as C Is XANES (Myneni et
al., 1998; Scheinost et al., 2001), while more qualitative measurements may be obtained
using IR (Celi et al., 1997). After methylation with tetramethylammonium hydroxide
(TMAH), humic pyrolysates include large quantities of molecules featuring methyl
esters, which represent carboxyl groups protected from decarboxylation reactions under
pyrolysis conditions (Saiz-Jimenez, 1994,1996; Grote et al., 2000; Jandl et al., 2002).
Molecules containing carboxyl groups are detected via ESI-MS analysis as well (Kramer
et al., 2001; Leenheer et al., 2001; Plancque et al., 2001).
When comparing different fractions of humic substances, carboxyl content is
found to be consistently higher in FA relative to HA (Stevenson, 1994; Mahieu et al.,
1999; Christl et al., 2000; Dai et al., 2001; Ritchie and Perdue, 2003), and in smaller size
fractions of HA relative to larger size fractions (Swift et al., 1992; Shin et al., 1999;
Christl et al., 2000). Oik et al. (2000) compared labile, mobile HA fractions (MHA) with
more recalcitrant calcium humates (CaHA), and found that the MHAs had fewer carboxyl
groups. Though few studies compare humin to other humic fractions, some suggest a
reduced carboxyl content relative to HA (Fabbri et al., 1998; Dai et al., 2001; Ussiri and
Johnson, 2003). However, the carboxyl content of humins obtained from NMR spectra
may underestimate dramatically the actual content, given the likelihood that many humin
183
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
carboxyl groups exist as deprotonated carboxylates complexed with paramagnetic iron
cations.
Given the wide range in dissociation constants linked to carboxyl groups attached
to different molecular structures (Leenheer et al., 1995a, b), it is important to consider the
distribution of carboxyl substituted molecular fragments. Leenheer et al. (1995a, b,
2003) noted that a small but significant fraction of humic carboxyl groups possess
dissociation constants (pKas), defined by the pH value for which the activities of
protonated and deprotonated acid are equivalent, as low as 0.5. Using NMR
spectroscopy and a series of techniques including methylation (Leenheer et al., 1995a),
chemical degradation (Leenheer et al., 1995b), and enrichment via pH gradient
fractionation (Leenheer et al., 2003), they demonstrate that the likely identity of such
strongly acidic structures are polycarboxylic acids near ether or ester functional groups,
and surrounded by additional electronegative substituents. They suggest that H-bonds
between adjacent carboxyls stabilize the carboxylate structure, resulting in reduced
dissociation constants (Leenheer et al., 1995b and 2003). More ordinary carboxyl groups
may be associated with alkyl or aromatic backbones, according to 2D NMR spectra (Mao
et al., 2001; Simpson et al., 2001c), ESI-MS (Leenheer et al., 2001; Plancque et al.,
2001), and pyrolysis results (Fabbri et al., 1996; Saiz-Jimenez, 1996; Schulten et al.,
1998). There is also substantial evidence for the presence of di- and tricarboxylic acid
moieties in humic substances (Schulten et al., 1998; Kramer et al., 2001; Simpson et al.,
2001c).
Comparison of humic substances examined with thermochemolysis-MS, or
thermo-chemical degradation and volatilization using TMAH and tetraethylammonium
184
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
acetate (TEAAc) (Grasset et al., 2002), as well as those processed using multistep
chemical degradation techniques and examined with pyrolysis (Grasset and Ambles,
1998) and NMR spectroscopy (Almendros et al., 1998), indicate that alkyl humic
molecules containing carboxyl groups exist as free acids loosely associated with other
moieties, as tightly trapped molecules within the organic matrix, and as acids linked to
the matrix via ester bonds. Heteronuclear 2D NMR spectra of FA confirm the presence
of ester-bound alkyl and aromatic moieties (Simpson et al., 2001c). The metabolites of
13C labelled polyaromatic hydrocarbons also form ester bonds with humic substances
(Kacker et al., 2002), further supporting the importance of this functional group in
connecting humic molecules to the backbone structure.
The presence of another class of carbonyls is suggested by ESR spectra of humic
substances. The ESR splitting factors (g-values) of humic substances, which provide
insight into the orbital state of the unpaired electron, and are calculated based on the
intensity of the magnetic field and the frequencies at which resonances occur, lie in the
range of 2.0023 - 2.0051, consistent with the presence of semiquinone radicals
conjugated to aromatic rings (Senesi, 1992). Electron spin resonance spectra of untreated
humic substances do not feature distinctive patterns of hyperfine structure, likely due to
the presence of a number of superimposed resonances featuring slightly different g-
values. However, processing by oxidation or acid boiling can produce structure
indicating the interaction of unpaired quinone electrons with neighboring H nuclei, and
resembling the spectrum of 2,6-dimethoxy-p-benzosemiquinone (Schnitzer, 1990; Senesi,
1992). Several 2D NMR studies present spectra featuring signals from unsaturated
ketone groups, consistent with the presence of quinones (Kingery et al., 2000; Mao et al.,
185
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
2001; Simpson et al., 2001c, 2002). Ketone and aldehyde C atoms attached to aromatic
rings are also evident in these spectra (Mao et al., 2001; Simpson et al., 2001c), while the
collective signal of ketone groups can be observed using C Is XANES (Myneni et al.,
1998). Excitation-emission matrices produced from fresh and microbially reduced FA
samples using fluorescence spectroscopy show similar responses to electron transfer
reactions as model aromatic quinone molecules (Klapper et al., 2002). Results from a
cyclic voltammetry study of humic substances dissolved in DMSO indicate the presence
of two pairs of redox peaks in a FA fraction enriched in polyphenols, peaks that resemble
those of model para- (1,4) quinones attached to aromatic functional groups (Nurmi and
Tratnyek, 2002). The ESR studies that first suggested the presence of quinone moieties
in humic substances indicate that the population of this functional group is small, in the
range of i o 16"18 spins g"1, and may be divided into stable and transient organic radicals
(Senesi, 1992). Fulvic acid fractions appear to contain slightly more quinones than HA
fractions (Senesi, 1992). It is important to note that the small population of quinone
groups cannot be responsible for the bulk of the electron transfer reactions available to
humic substances (Struyk and Sposito, 2001).
Phenol functional groups provide many weak acid sites to humic substances
(Ritchie and Perdue, 2003). Potentiometric titrations performed by Ritchie and Perdue
(2003) indicate that the average phenol to carboxyl ratio in humic substances is 1:4.
Phenol C provides a NMR signal located near that of aromatic C, and though it may
suffer from the same CP/MAS signal loss processes operating for aromatic C, Monteil-
Rivera et al. (2000) find their quantitative CP/MAS estimate of phenol content is in
agreement with that obtained by chemical means, as the difference between total acidity
186
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
determined using Ba(0H)2 exchange, and carboxyl acidity determined using Ca-acetate
exchange. Kingery et al. (2000) find substantial 2D NMR evidence of ortho- (1,2) and
para- (1,4) substituted phenol groups in the HA they examined. Scheinost et al. (2001)
note that the C Is XANES signal of humic phenol groups occurs below the energy range
of reference phenolic molecules, indicating a slightly different chemical environment
from that of the model compounds. Many phenols recovered through pyrolysis feature
aromatic rings with multiple substitutions, and often include methoxy groups (Lehtonen
et al., 2000), suggesting a lignin origin. Two dimensional NMR spectra of a peat HA
indicate 1/3 of the O attached to aromatics is part of methoxyl groups, while 2/3 is part of
phenol groups (Mao et al., 2001). Fulvic acids tend to contain slightly more phenol
groups than corresponding HAs (Christl et al., 2000), and with decreasing molecular
weight, the proportion of phenols in humic fractions tends to increase (Christl et al.,
2000; Scheinost et al., 2001).
The (alkyl) alcohol functional group is somewhat more challenging to quantify
than phenol. X-ray photoelectron spectroscopy cannot distinguish phenol from alcohol
(Monteil-Rivera et al., 2000), typical ID 13C NMR techniques do not provide an easily
differentiated alcohol peak, and the recovery of alcohols from pyrolysis is hindered
because of their extremely weak acidic character, which requires high energy levels for
deprotonation and MS analysis (Lehtonen et al., 2000). Nevertheless, alcohol groups
have been identified in humic substances using 2D NMR techniques (Simpson et al.,
2001c, 2002). Cook and Langford (1998) suggest that a large portion of weak acid sites
in humic substances may take the form of alkyl/carbohydrate alcohol groups rather than
phenol groups, as the proton dissociation behavior of their samples cannot be explained
187
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
by the low level of phenols detected via their 13C ramped amplitude CP/MAS NMR
technique, which reduces line broadening under rapid spinning conditions designed to
diminish spinning sidebands.
Many alcohols may be associated with carbohydrate-derived moieties, along with
ether functional groups in linear or cyclic alkyl chains. The O-alkyl or carbohydrate
signals found in 13C CP/MAS NMR spectra indicate that this group, made up of
oxygenated C in carbohydrates and cellulose, along with alkyl-amino C and methoxyl C,
is a dominant fraction of humic substances, constituting on average over 1/4 of the 13C
signal (Mahieu et al., 1999). According to Smemik and Oades (2003), even this value
may be an underestimate due to rapid relaxation of in the rotating frame of the
CP/MAS technique. The anomeric signal of cyclic sugars can create a distinct shoulder
structure near 100 ppm on the aromatic 13C NMR peak (Kingery et al., 2000; Mao et al.,
2001). 'H NMR spectra feature signal clusters representing sugars and carbohydrates, as
well as carbohydrate-like groups (Kingery et al., 2000). Various 2D NMR techniques
confirm the presence of alkyl ether and anomeric structures in humic substances (Fan et
al., 2000; Kingery et al., 2000; Mao et al., 2001; Simpson et al., 2001c, 2002). In
addition, C Is XANES can be used to quantify the presence of O-alkyl groups in humic
substances (Scheinost et al., 2001). Pyrolytic study of humic substances produces
molecular fragments that include ether groups within linear and cyclic alkyl chains
(Schulten, 1999a). Pyrolysates or NMR spectra collected after multistep chemical
degradation schemes indicate ether bonds often link humic fragments together
(Almendros et al., 1998; Grasset and Ambles, 1998), as do examinations of 13C labeled
polyaromatic hydrocarbons subjected to biodegradation in the presence of humic
188
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
substances (Kacker et al., 2002). Some of these O-alkyl moieties prove extremely
resistant to chemical degradation (Almendros et al., 1998). With decreasing molecular
size, humic fractions contain decreasing O-alkyl signal (Swift et al., 1992; Haiber et al.,
2001), although there is no consistent trend in O-alkyl content among humic and fulvic
fractions (Preston and Ripmeester, 1982; Cook and Langford, 1998; Christl et al., 2000;
Dai et al., 2001). When contrasting mobile and recalcitrant humic fractions, Oik et al.
(2000) found more carbohydrates in the labile MHAs than in the Ca-humates. Humin
samples often contain substantial O-alkyl signals (Fabbri et al., 1996, 1998; Almendros et
al., 1998; Dai et al., 2001; Ussiri and Johnson, 2003).
To review, a number of oxygenated functional groups contribute to the chemical
reactivity of humic substances. Several types of carboxyls, including unusual alkyl
polycarboxyl moieties near ester or ether groups, determine the deprotonation and charge
behavior of humic substances under low pH conditions, while phenols and possibly
alkyl/carbohydrate alcohols play an important buffering role at higher pH conditions.
Quinone groups, though present, are not sufficiently numerous to account for the redox
behavior of humic substances. Carbohydrate-based O functional groups make up a large
portion of humic substances. Ester and ether bonds contribute numerous links between
different portions of humic molecules. Fulvic acids and humic molecules with smaller
apparent molecular weight tend to be more oxygenated, and therefore to contain more of
these functional groups.
189
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
4.1.4 Nitrogen Functional Groups o f Humic Substances
On average, over 90% of soil N is organic, and soil humic substances frequently
feature C:N ratios close to 10 (Stevenson, 1994). Early chemical methods for
differentiating soil N functional groups relied on acid hydrolysis, followed by procedures
to isolate amino acids, amino sugars, and ammonium from the hydrolysate (Stevenson,
1994). However, large portions of N remained uncharacterized, including fractions
known simply as non-hydrolyzable or acid-insoluble N (20 - 35%) and hydrolyzable
unknown N (> 20%) (Stevenson, 1994). Modern spectroscopy and pyrolysis have
provided considerable insight into the speciation of N within humic substances.
All solid-state 15N CP/MAS NMR spectra indicate that humic N resides
predominantly in the form of amide functional groups, with a smaller contribution from
free amino groups (Knicker et al., 1995; Knicker and Ludemann, 1995; Knicker, 2000,
2002; Mahieu et al., 2000, 2002; Mathers et al., 2000; Zang et al., 2000). Early studies
faced criticism because due to the low abundance of the 15N isotope, a spectrum with
sufficient resolution could only be obtained through 15N fertilizer enrichment and
subsequent incubation (Knicker and Ludemann, 1995). The N signals in such enriched
samples may not reflect the distribution of N functional groups formed over longer time
periods. However, pretreatments to extract humic substances or remove mineral
components (Schmidt et al., 1997), combined with improvements to the NMR technique,
have resulted in natural abundance 15N NMR spectra that support the dominance of
proteinaceous material, represented by the amide signal, in humic substances (Knicker et
al., 1995; Knicker, 2000; Mahieu et al., 2000, 2002; Mathers et al., 2000; Zang et al.,
2000). Studies of non-hydrolyzable N fractions indicate that acid hydrolysis leaves
190
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
behind a substantial proportion of peptidic structures in the soil (Knicker and Ludemann,
1995; Knicker, 2000), perhaps protected by organo-mineral association (Knicker, 2000;
Leinweber and Schulten, 2000) or hydrophobic entrapment (Knicker et al., 1996;
Knicker, 2000; Zang et al., 2000).
Some forms of heterocyclic N, including indoles, pyrroles, and imidazoles, should
provide 15N NMR signals around -180 to -239 ppm (15N NMR reference nitromethane),
while pyridine should provide signals in the range -25 to -90 ppm (Mahieu et al., 2000).
Most spectra do not feature a signal distinguishable from noise in these regions (Knicker
et al., 1995; Knicker, 2000, 2002), implying that less than 10% of the N present in humic
substances is in the form of heterocyclic functional groups (Knicker, 2000). The
CP/MAS NMR technique may be less sensitive to heterocyclic N than to amide N,
because the method depends on the transfer of magnetic spin from *H nuclei to 15N
nuclei, and heterocyclic N is not directly protonated. However, the observation of signals
from nitrate groups, which also lack protonation (Knicker and Ludemann, 1995),
indicates that the CP/MAS NMR technique could be used to measure these functional
groups within humic substances, as does the detection of heterocyclic N in the form of
pyrroles and associated groups in fossil algae samples (Knicker et al., 1996). Longer
CP/MAS contact times do not result in the detection of new peaks (Knicker et al., 1995),
nor do single pulse excitation (SPE) 15N NMR analyses, in which the 15N nuclei are
excited directly (Knicker, 2000).
Dipolar dephasing (DD) solid-state 13C NMR spectra, which can identify C
affected by strong proton dipolar coupling due to presence of a bonded amino group,
have been used to estimate the relative amount of N-substituted alkyl C, and therefore to
191
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
calculate the relative contribution of peptides to total C and N of humic substances,
corroborating the dominant role of peptides in humic N (Knicker, 2000). A 2D NMR
technique, double cross polarization (DCP) MAS 13C 15N NMR spectroscopy, has been
used on 15N enriched humic samples to detect 13C nuclei close to 15N nuclei (Knicker,
2002). These spectra showed significant coupling between amide-C and N-substituted-
alkyl-C, demonstrating again that the amide functional group is the dominant form of N,
and that most of this N is connected to alkyl C. Fan et al. (2000) were able to perform a
variety of solution state ID and 2D NMR analyses on a HA sample treated with Tiron to
remove exchangeable metal ions and increase solubility. Signals from !H NMR spectra
indicated abundant amino acids, while 2D NMR spectra obtained using both H total
correlation spectroscopy (TOCSY), which feature off-diagonal scalar couplings marking
13 1
protons connected by three or fewer covalent bonds, and C- H heteronuclear single
quantum correlation (HSQC), which feature off-diagonal scalar couplings marking
protons directly bonded to C atoms, identified them as components of peptides. The
peptidic chains appear to be mobile components of the humic substance, while aromatic
groups remain relatively rigid, according to interpretations of 2D 'H nuclear Overhauser
effect spectroscopy (NOESY) spectra, which feature crosspeaks identifying protons
located near each other, as well as TOCSY and HSQC data. Further NMR evidence in
the form of DOSY and NOESY spectra provided by Simpson and colleagues (Simpson et
al., 2001a, 2002) supports the presence of peptidic N in humic substances.
A few 15N NMR spectra feature shoulders or peaks in the range of the indole,
pyrrole, and imidazole signals. Mahieu et al. (2000 and 2002) were able to detect a
shoulder representing these heterocyclic N groups from labile MFLAs and recalcitrant Ca-
192
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
humates, accounting for as much as 22% of the total N signal. Heterocyclic N was more
common in CaHA fractions and aerated soils, and was thought to be formed as part of
gradual humification over many years. Mahieu et al. (2000) note that the peak identified
as heterocyclic N does not include signals from all forms of heterocyclic N, so many
other forms may also be present, all at quantities too low for detection. Another study
detected a small signal in a HA spectra, possibly due to pyridinic N (Zang et al., 2000).
Pyrolysis methods provide another means of tackling the question of N speciation
within humic substances. Pyrolysis performed on individual amino acids reveals the
activity of several different decomposition pathways, and provides diagnostic pyrolysis
fragments for most pure amino acids, which include amines, nitriles, imines,
hydrocarbons, N-substituted hydrocarbons, and cyclic N products (Chiavari and Galletti,
1992). It is apparent that even for simple amino acid systems, secondary chemical
reactions under these experimental conditions produce pyrolysates that often show little
resemblance to parent molecules. Therefore, although pyrolysates of humic substances
typically include a suite of molecules representing pyrroles, free and substituted
imidazoles, pyrazoles, pyridines, substituted pyrimidines, pyrazines, indoles, quinolines,
N-derivatives of benzene, alkyl amines, alkyl and aromatic nitriles, and others (Schulten
and Schnitzer, 1998), the relation of these structures to the structure of the parent humic
molecules is unknown.
Arguments in favor of the presence of heterocyclic N within humic substances
include a) the presence of heterocyclic N groups in biological precursors such as proteins
and carbohydrates (Schulten and Schnitzer, 1998), b) the detection of free and substituted
heterocyclic N groups in aquatic humic substances under low temperature (200 - 300 C)
193
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited without per mi ssi on.
pyrolysis conditions (Schulten and Schnitzer, 1998), c) the enrichment of these
components in clay fractions and in managed soils with reduced total and chemically
assessed amino content (Leinweber and Schulten, 1998), d) the detection of heterocyclic
N groups in humic substances via other methods, (Schulten and Schnitzer, 1998), e) the
measurement of consistent intensity of these groups after spiking soils with amino acids
or raising the pyrolysis temperature from 300 to 700 C (Leinweber and Schulten, 1998),
and f) the identification of N-derivatives of benzene in soils and humic substance
pyrolysates but not in plant or microbial pyrolysates, implying formation within soil
(Schulten et al., 1997). Some interpretations of humic structure based on pyrolysis data
conclude that there is nearly as much heterocyclic N (35%) as proteinaceous N (40%)
present (Schulten and Schnitzer, 1998). Others are far more cautious in linking fragment
molecules to the original structure of humic substances, given that the pyrolysates in
question are known secondary products of more defined amino acid and protein samples
(Saiz-Jimenez, 1996), and often do not represent a large proportion of the total N in
humic substances (Knicker et al., 1999), while some interpret the results to be indicative
of primarily proteinaceous materials (Fabbri et al., 1996; Zang et al., 2000). Similar
controversy surrounds the presence of nitriles, especially the long-chain alkyl nitriles that
have been detected as components of soil and humic substance pyrolysates but not of
plant or microbial pyrolysates (Leinweber and Schulten, 1998). Schulten et al. (1997)
cite studies using non-pyrolysis methods that have detected the presence of nitrile groups
within humic substances. However, Nierop et al. (2001) suggest that the presence of
hexadecanitrile in their humic substance pyrolysates may be an artifact due to the
reaction between alkanoic acids and ammonium under pyrolysis conditions.
194
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Pyrolysis of acid hydrolyzates and hydrolysis residues of humic substances
reveals the presence of a host of different cyclic N and nitrile compounds in the two
fractions (Schulten et al., 1997). These results were interpreted to indicate that
heterocyclic N and long-chain alkyl nitrile groups are N species that cannot be
characterized by chemical methods in the two fractions. Further examination of humic
substance hydrolysis residues using a dithionite-citrate-bicarbonate (DCB) treatment to
remove pedogenic oxides and mineral-bound N, followed by hydrolysis and amino acid
analysis or pyrolysis of these DCB extracts, reveals substantial amounts of amino acids
that had been protected from initial hydrolysis by association with minerals, as well as
pyrolyzates indicating the presence of both peptides and heterocyclic N groups
(Leinweber and Schulten, 2000). A pyrolysis examination of the N groups produced
during 15N fertilization experiments similar to those initially used to produce 15N NMR
spectra of humic substances reveals similar heterocyclic N and nitrile pyrolysates seen in
unfertilized samples (Gonzalez-Vila et al., 2001b), indicating that fertilization may not
produce substantial artifacts.
The addition of catalytic chemicals that facilitate molecular fragmentation at
lower temperatures may reduce alteration of the resulting pyrolysates due to secondary
reactions. Amino acids and proteinaceous materials characterized with TMAH
thermochemolysis at 250 C indicate peptide bonds are effectively cleaved under these
conditions, yielding individual amino acid methyl esters, and few secondary reaction
products (Zang et al., 2001). Knicker et al. (2001) confirmed this result, and analyzed the
acid-hydrolyzed humin fraction of an algal sapropel, producing pyrolysates, including
methyl esters of amino acids, which indicated the presence of primarily proteinaceous N
195
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
even after hydrolysis. The HA probed by Zang et al. (2000) generated primarily alkyl
methyl esters of amino acids with thermochemolysis, which may imply a preference for
alkyl amino acids within humic substances. Given the greater yield as compared to
conventional pyrolysis, they suggest that TMAH disrupts the 3D structure of the HA,
releasing entrapped proteinaceous material for hydrolysis and methylation. Further
experiments indicate that additions of proteinaceous materials could become
encapsulated within hydrophobic domains in the HA, and protected from complete
hydrolysis (Zang et al., 2000). Thermochemolytic results support the suggestion that a
large portion of the heterocyclic and nitrile N groups obtained via pyrolysis are the result
of secondary transformations at high temperatures. A primarily peptidic view of humic N
has begun to emerge from application of thermal degradation techniques, in agreement
with NMR work.
Synchrotrons have provided another tool for assessing N in humic substances. A
study on several humic substances using N K-edge XANES verified the dominance of
amide functional groups, based on the peak positions relative to model compounds
(Vairavamurthy and Wang, 2002). Additionally, pyridinic N was measured at 20 - 30%
of the total N content, a portion of which was identified as oxidized derivatives of
pyridine like those with carboxyl groups attached to the ring. However, Myneni (2002)
questions the band assignments used, stating that more work with model molecules is
required before pyridine and a few other functional groups can be assigned
unambiguously.
To summarize, NMR, thermochemolysis, and XANES data indicate that amide N
in the form of peptides, and likely connected to alkyl C, is the dominant form of N in
196
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
humic substances. A substantial portion of this proteinaceous material is protected from
acid hydrolysis by interactions with mineral surfaces or hydrophobic moieties, interfering
with chemical methods for determination of amide content. Free amino groups are
another component of humic N, as end groups of peptidic chains or as individual amino
groups within the larger molecular structure. Heterocyclic N, a category that
encompasses a variety of functional groups, many with individually low concentrations,
also appears to exist within the humic fraction. Limited experimental evidence suggests
that nitrile groups may be a minor component of humic N.
4.1.5 Sulfur Functional Groups o f Humic Substances
Nearly all S in soil is associated with organic matter, as evidenced by consistent
positive correlations between TOC and total S (Stevenson, 1994). Sulfur can exist in a
broad range of oxidation states. Reduced S functional groups include sulfide (oxidation
state -2), thiol (-1), and thiophene (0), moderately oxidized S groups include sulfoxide
(2), sulfite (4), and sulfone (4), and highly oxidized S groups include sulfonate (5) and
sulfate (6). Early methods for assessing S functional groups within humic substances
identified oxidized or ester sulfate S, and reduced or C-bonded S, via chemical
reduction with HI, then Raney Ni (Stevenson, 1994).
X-ray photoelectron spectroscopy provides a more chemically sophisticated
means of assessing the presence of reduced and oxidized S groups within humic
substances. Bubert et al. (2000) used the 2Ss signal in the range of 225 - 237 eV to
analyze a variety of natural and anthropogenic aquatic humic substances, and identified
three peaks. A sulfide peak was found at 228 eV, while two other humic S peaks were
197
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
located at 232 and 234 eV, possibly caused by sulfoxide and/or sulfone groups. A natural
aquatic humic substance so analyzed contained more reduced or sulfide S than oxidized
S, although its spectrum was very close to the detection limit of the technique, while two
anthropogenic wastewater humic substances contained more oxidized S.
A far more promising spectroscopic method for assessing S functional groups is
XANES. The S K-edge can be used to identify multiple organic S oxidation states,
resulting in spectra of humic substances with as many as six peaks or shoulders for
different forms of S (Xia et al., 1998). The quantitative assessment of these peaks, as
well as their assignment to different S functional groups, especially in the reduced range,
is not straightforward. A method used by Xia et al. (1998) to quantify S functional
groups involves an arctangent step function to represent step height (created by the
transition of ejected photoelectrons to the continuum), and Gaussian peaks to represent
the orbital transitions recorded in the spectra, the areas of which can be related to the
percentages of total S at different oxidation states, after scaling them to take into account
the linear peak area vs. peak position correlation. Xia et al. (1998) observe four peaks in
their spectra of humic substances, which they identify as sulfide-thiol S (similar
electronic oxidation states), sulfoxide S, sulfonate S, and sulfate S. In addition they find
two shoulders in the spectra, thought to represent thiophene and sulfone. The
quantitative data extracted from these and other spectra using the same deconvolution
method (Szulczewski et al., 2001; Martinez et al., 2002; Qian et al., 2002) show a
bimodal distribution, with most S either in reduced or highly oxidized forms, and
additionally document the existence of moderately oxidized S functional groups.
Comparison of humic substances formed under different conditions indicates that those
198
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
from aquatic or organic soil environments contain more reduced S, while those from
mineral soil environments contain more oxidized S. Hundal et al. (2000) use XANES in
conjunction with traditional chemical reduction methods to analyze the total hydrophilic
and hydrophobic acid fractions of biosolids-derived FA. Both methods produce data
indicating the hydrophilic fraction is dominated typically by oxidized S, while the
hydrophobic fraction is dominated by reduced S functional groups, as might be predicted
from the polarity and O contents of the two fractions.
The method of quantification described above is valid only if each oxidation state
examined produces a single peak. However, Szulczewski et al. (2001) note that spectra
of disulfide functional groups exhibit doublets. Xia et al. (1998) also note that the areas
of the most reduced peaks, which they associate with thiol-sulfide and thiophene, are
strongly influenced by the breadth and position of the first background step, and that
thiophenic groups have significant post-edge resonance at an energy level that overlaps
with the sulfoxide peak, both of which may affect quantitative analysis (Xia et al., 1998).
Further discussion of the interpretation of XANES spectra focuses on the assignment of
reduced S peaks. In examining the reduction of Cr(VI) by humic substances and other
thiol-containing compounds, thought to take place via a thiol/disulfide redox couple,
Szulczewski et al. (2001) observe a decrease in the most reduced S XANES peak, and an
increase in the next most reduced peak. They conclude that the most reduced peak
represents thiol, and the next most reduced peak represents disulfide and thiophene,
though Xia et al. (1998) assigned them to thiol/sulfide and thiophene, respectively.
Vairavamurthy et al. (1997) used linear combinations of model compound
absorption spectra to fit marine sediment HA spectra, a method with an advantage over
199
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
that used by Xia et al. (1998) when dealing with spectra featuring more than one peak, as
complete model spectra are used to fit data. Vairavamurthy et al. (1997) were able to fit
their HA spectra using five model compound spectra, representing organic di- and
polysulfides, organic sulfides, sulfoxides, sulfonates, and sulfates. They also treated a
HA particularly rich in polysulfides with tributyl phosphine, a reducing agent meant to
cleave di- and polysulfide bonds to generate thiol groups. The thiol reference spectrum
was required for a proper fit of the resulting spectrum. These HA data display the same
bimodal distribution of S oxidation states noted previously, and indicate substantial
concentrations Of sulfides and di- and polysulfides, as might be predicted given their
marine origins. A qualitative investigation by Morra et al. (1997), using a similar method
to interpret spectra of a variety of HAs and FAs, identified many of the same oxidation
states found above. However, they were not able to detect sulfate groups in any of their
FA samples, a result inconsistent with that of Xia et al. (1998) for Suwannee River FA.
A study by Hutchison et al. (2001) used the same spectral fitting technique to assess the
effect of pH and aeration on the oxidation state of reduced S functional groups of a salt
marsh HA. The spectra provided evidence for oxidation only under pH 13 conditions.
At lower pH, spectra showed no signs of S oxidation, indicating either the consumption
of O by non-S functional groups, or oxidation of thiols to produce sulfides, which might
not be evident given the overlapping peaks of the functional groups involved.
Pyrolysis has been used to examine S functional groups, and cyclic S molecules
such as thiophene and benzothiazole have been identified in pyrolysates (Leinweber and
Schulten, 1998; Schulten, 1999a). However, Saiz-Jimenez (1995) discovered that many
of the alkylthiophene products observed are artifacts produced during decarboxylation of
200
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
alkyl carboxyl groups in the presence of S contaminants. Methylation with TMAH prior
to pyrolysis protects carboxyl groups and eliminates most of these artifact products,
though TMAH can react with S to create a few secondary contaminants of its own. Van
Loon et al. (1993) used pyrolysis to identify the presence of sulfonic acid groups in
aquatic humic substances, indicated by release of SO2 over specific temperature ranges.
In summary, recent advances particularly in the field of XANES have resulted in
a broader understanding of S functional groups within humic substances. Sulfur tends to
display a bimodal distribution of oxidation states, with most S either strongly reduced or
strongly oxidized, although intermediate S oxidation states are present as well.
4.1.6 Humic Molecular Fragments
It is now necessary to connect the many humic functional groups described in
Sections 4.1.2 - 4.1.5 into larger structures, or humic fragments, and examine their
relative mobilities. Early oxidation experiments provided evidence for the existence of a
variety of phenols, aromatic aldehydes and carboxylic acids, quinones, and heterocyclic
N compounds, while extractions revealed alkyl constituents such as alkanes and fatty
acids, and acid hydrolysis indicated the presence of amino acids, amino sugars, and
carbohydrates within humic substances (Orlov, 1995). However, the chemical techniques
used are considered harsh, and may result in the destruction of important structures and
the creation of artifact compounds not present in humic substances. In addition, they
provide no indication of rigid or flexible properties of humic moieties. Modem
spectroscopic, pyrolytic, and soft ionization techniques have begun to reveal this
information.
201
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
There is substantial evidence that lipid-derived molecules make up a large portion
of humic substances. Lipid-derived molecules include branched and lineai; alkanes,
alkenes, fatty acids, dicarboxylic acids, and long chain alcohols/ethers,
ketones/aldehydes, and esters. These compounds have been identified as major
components of humic substances in numerous pyrolysis studies, especially when
methylation with TMAH is used to preserve carboxyl groups (Schnitzer, 1990; Fabbri et
al., 1996; del Rio et al., 1998; Schnitzer and Schulten, 1998; Schulten et al., 1998;
Schulten, 1999a; Lu et al., 2000; Gonzalez-Vila et al., 2001a; Chefetz et al., 2002). As
mentioned previously, using both TMAH and TEAAc in combination with pyrolysis,
Grasset et al. (2002) showed that humic substances contain mono- and dicarboxylic acids
that are tightly trapped within the organic matrix, as well as those that are chemically
bound by ester groups to larger structures. The distribution of alkanes and fatty acids
favors an even number of C atoms, and tends to peak at chain lengths of 16 or 18 C
(Schnitzer, 1990; Saiz-Jimenez, 1992; Fabbri et al., 1996; Chefetz et al., 2002), evidence
of a microbial or plant origin of the fragments. Fragments well over 30 C in length,
derived from higher plant waxes, have been identified as well (Grasset et al., 2002).
Electrospray ionization detects the presence of many lipids, including fatty acids with
lengths of over 40 C (Kramer et al., 2001; Stenson et al., 2002).
Simpson et al. (2001c) provides evidence for the existence of similar amounts of
mono- and dicarboxylic acids with average chain lengths of 10 C in one FA sample using
a variety of NMR techniques, including ll (ID), TOCSY (2D), HSQC (2D), 13C
distortionless enhancement by polarization transfer (DEPT) (ID), which allows the
number of protons bonded to a C to be determined, correlation spectroscopy (COSY)
202
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
(2D), used to detect couplings between protons and identify those attached to adjacent C,
heteronuclear multiple quantum coherence - total correlation spectroscopy (HMQC-
TOCSY) (2D), which provides the means to identify all C atoms linked to each other and
to a chosen proton, and heteronuclear multiple bond connectivity (HMBC) (2D), used to
identify C that are two or three bonds distant from a given proton. While fatty acids were
dominant components of this organic matter, smaller amounts (10 - 20% of the acids) of
alkyl esters and alcohols/ethers were also recognized (Simpson et al., 2001c). Diffusion
ordered spectroscopy performed on the same sample indicated the alkyl components were
larger than 7 units in length (Simpson et al., 2001a), and could possess lengths closer to
16 or 18 C. These results held true for a number of other humic substances (Simpson et
al., 2002). Using 3D HMQC-TOCSY, Simpson et al. (2003) were able to identify alkyl
groups containing mid-chain carboxylic acid, hydroxyl, and ester groups, and terminal
carboxylic acid, hydroxyl, and methyl groups, in a forest FA sample. The alkyl material
strongly resembled the plant biopolymer cutin, especially after oxidation to shorten the
lengths of the alkyl chains.
The degradation of lignin via addition of carboxyl groups, removal of methoxyl
groups, and fragmentation of lignin structures into smaller polymers and monomers, is
likely a key process in the formation of humic substances (Stevenson, 1994). Pyrolysis
studies have identified numerous lignin monomers, including p-hydroxyphenol, guaiacol,
syringol, and vanillin, as well as derivatives featuring methyl or carboxyl substitutions
(Fabbri et al., 1996; Lu et al., 2000; Chefetz et al., 2002; Page et al., 2002). A few of
these compounds have been identified using ESI-MS as well (Kramer et al., 2001).
Pyrolysis also detects lignin dimers and their derivatives within humic substances
203
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
(Schnitzer, 1990; Schulten et al., 1998; Schulten, 1999a). Using a combination of ID and
2D NMR techniques, a number of lignin and lignin-derived backbone structures were
identified, including 1,2-, 1,4-, 1,3,4-substituted benzenes and 2-propenoic acid, despite
the presence of relatively few aromatic groups in the FA sample examined (Simpson et
al., 2001c). Alkyl, carbonyl, carboxyl, methoxyl, and hydroxyl functional groups were
observed to be connected to these rings (Simpson et al., 2001c). A second FA sample
studied using 3D HMQC-TOCSY revealed the same lignin biomarkers (Simpson et al.,
2003). Peat HAs and FAs examined with multiple 1 and 2D NMR methods indicated
prevalence of highly substituted (4 or 5 positions) lignin-derived aromatic rings in both
fractions, though aromatic groups with fewer substitutions were more common in the FA
than in the HA (Hertkom et al., 2002). A peat HA studied using heteronuclear
correlation (HETCOR) (2D) 'H-^C solid-state NMR spectra, which feature off-diagonal
scalar couplings marking atoms connected by covalent bonds, provided evidence that
methoxyl substitutions of aromatic rings, typical of lignin structures, occurred quite
frequently, though at half the level of hydroxyl substitutions (Mao et al., 2001). Ether
and carboxyl substitutions were well-resolved, while methyl substitutions were relatively
rare (Mao et al., 2001). By introducing 'H spin diffusion into HETCOR spectra,
compositional heterogeneity in the form of lignin-like domains was noted for a length
scale of 1 nm, separated by < 3 nm from alkyl regions of the humic sample (Mao et al.,
2001). Detection of heterogeneity was also noted using DOSY NMR spectra, in which
the diffusional properties of the lignin-like portions of a variety of humic substances were
considered to be consistent with lignin polymers containing 2 - 4 units, with chains of
more than 8 units unlikely (Simpson et al., 2002). Suwannee River HA and FA separated
204
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
into molecular size fractions via tangential flow multi-stage ultrafiltration and analyzed
using phase-sensitive HETCOR NMR spectroscopy indicated that as molecule
size increased, decomposition of phenylpropane sidechains was accompanied by
dealkylation of methoxy groups and reaggregation of aromatic rings into more highly
substituted forms (Haiber et al., 2001).
A number of other aromatic units have been identified that may be derived from
lignin or from other sources. Alkylbenzenes and alkylphenols are detected with
regularity in the pyrolysates of humic substances (Schnitzer, 1990; Schnitzer and
Schulten, 1998; Schulten et al., 1998; Schulten, 1999a; Lehtonen et al., 2000; Lu et al.,
2000), and can feature alkyl C chain lengths of over 25 C (Schnitzer and Schulten, 1998).
However, many of these alkylaromatic molecules may have been created by secondary
reactions of precursor substances during pyrolysis (Saiz-Jimenez, 1994, 1996; van
Heemst et al., 1999; Hatcher et al., 2001). Benzene tricarboxylic acids are common
pyrolysates if the heat treatment is preceded by TMAH methylation (Saiz-Jimenez, 1994,
1996; Schulten et al., 1998), and have also been recovered using ESI-MS (Kramer et al.,
2001). By combining ESI with multistage tandem MS (MST-MS), isolated humic ions
can be fragmented gradually through loss of CO2 and H2 O (Leenheer et al., 2001;
Plancque et al., 2001), providing information that can be used to identify more complex
molecules with alkylbenzene backbones and numerous hydroxyl and carboxyl functional
groups.
Derivatives of carbohydrate structures are common constituents of humic
substances identified by pyrolysis (Fabbri et al., 1996; Schulten et al., 1998; Lehtonen et
al., 2000; Lu et al., 2000; Gonzalez-Vila et al., 2001a; Page et al., 2002). Simple sugars
205
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
like ribose and glucose have been identified using ESI-MS spectra (Kramer et al., 2001).
Both ID and 2D NMR techniques can be used to identify chains of glucose (Simpson et
al., 2001c) and other carbohydrate structures (Fan et al., 2000; Mao et al., 2001; Simpson
et al., 2001a, 2002; Hertkom et al., 2002), and DOSY NMR spectra of numerous humic
substances featured slow diffusion of the carbohydrate signals, indicating that
polysaccharide chains of 3 - 8 units (~600 - 1500 Da) are present in the samples
(Simpson et al., 2001a, 2002). The same spectra also provide evidence for the existence
of polypeptides (Hertkom et al., 2002) of up to 10 units in length (Simpson et al., 2001a,
2002) within humic substances. Mobile peptidic chains, and many specific amino acids
within these chains, were identified by a variety of 1 and 2D NMR spectra of a forest soil
humic substance treated with Tiron to improve solubility (Fan et al., 2000).
Pyrolysis, ESI-MS, and NMR techniques all tell a similar story about the
organization of humic fragments. Humic substances appear to be composed primarily of
lipids, lignin monomers and polymers, and other substituted aromatic groups, along with
carbohydrates, polypeptides, and their derivatives. The heterogeneity of organic matter
indicated by NMR spectroscopy provides evidence that all or at least a portion of these
groups may exist separately in small, discrete clusters in close proximity to each other.
A variety of experimental evidence supports the existence of a continuum of
humic domains with flexible and expanded (rubbery) to dense and rigid (glassy)
properties (Pignatello, 1999). The mechanism of sorption of nonionic organic
compounds transitions from that of solid-phase dissolution, or partitioning, to that of site
specific void-filling, along this continuum. The availability of both sorption mechanisms,
known as dual mode sorption, is used to explain nonlinear adsorption isotherms and
206
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
competitive effects observed for sorption of neutral organic species to humic substances
(Pignatello, 1999; Yuan and Xing, 2001; Mao et al., 2002a; Gunasekara et al., 2003), a
result of the limited ability of nanopores within rigid domains to accommodate guest
species (Pignatello, 1999).
To explore the functional makeup of flexible and dense humic domains,
Gunasekara et al. (2001) used NMR to determine spin-spin relaxation values for two HAs
with changes in temperature, and were able to identify mobile and rigid signals, whose
proportions increased and decreased with increasing temperature, respectively. Further
work with bleaching and hydrolysis treatments indicated that aromatic groups may be
associated with the condensed domains, while carbohydrates may be associated with the
flexible domains (Gunasekara et al., 2003). Fan et al. (2000) used several NMR
techniques to establish that peptidic chains were mobile, while aromatic groups were
relatively rigid, in their Tiron-treated HA.
However, other studies indicate that dense, immobile domains within humic
substances are composed predominantly of alkyl groups, not aromatic groups. Wang et
al. (2001) measured spin-lattice relaxation times indicating aromatic groups were more
mobile than alkyl groups in two HAs. Cook and Langford (1998) used NMR to show
that Laurentian FA features largely immobile, unfunctionalized alkyl structures, more
mobile, unfunctionalized aromatic units, and even more mobile carbohydrate groups,
while Laurentian HA is made up of immobile and slightly more mobile aromatic
structures. Hu et al. (2000) identified both rigid (~3 nm thick) and amorphous
poly(methylene) domains in several samples of natural organic matter using solid-state
spin diffusion NMR and wide angle x-ray scattering (WAXS) experiments, and Mao et
207
Re pr oduc e d with per mi ssi on of t he copyright owner. Fur t her reproduct i on prohibited without per mi ssi on.
al. (2002a) used 'H inversion-recovery experiments to establish the presence of ~10 nm
diameter domains of branched alkyl groups in the same humic substances. Mao et al.
(2002a) also observed a positive correlation between the abundance of the amorphous
poly(methylene) domains and the sorption of the hydrophobic organic phenanthrene,
consistent with the dual mode model of sorption of nonionic organic compounds.
Examination of the environment of an adsorbate can provide further evidence for
the existence of multiple sorption domains within humic substances. Khalaf et al. (2003)
used ultrafiltration to separate a HA into molecular size fractions. The largest molecular
size fractions were predominantly alkyl, while the smallest contained greater proportions
of aromatic and carboxyl functional groups. 19F NMR spectra of hexafluorobenzene
sorbed to these fractions suggested the presence of heterogeneous, motionally restricted
sorption environments within alkyl-rich HA fractions, and multiple mobile sorption
domains in aromatic-rich fractions.
A more complex view of the motion of different functional groups can be
achieved using a different set of NMR methods. 'H wideline solid-state NMR spectra
display information about dipolar coupling that can be related to mobility (Mao et al.,
2002b). Analysis of ID *H wideline spectra for HAs measured at 27 - 127 C indicates
humic substances possess a broad continuum of proton mobility little affected by
temperature (Mao et al., 2002b). Separation of proton wideline signals, using associated
13C chemical shifts via the 2D wideline separation (WISE) technique, indicates that alkyl,
carbohydrate, and aromatic functional groups feature at least three different levels of
mobility each, while carboxyl groups exhibit at least two (Mao et al., 2002b). The
diverse motional behavior detected for each functional group adds a new level of
208
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
complexity to spectroscopic evidence favoring existence of rigid and flexible domains
within humic substances.
To summarize, the best available science indicates that humic substances are
composed of molecules and molecular fragments derived primarily from lipids, lignin,
and non-lignin aromatic species, with contributions from carbohydrates and proteins.
These humic constituents may be segregated by location, on a scale of nanometers, as
well as by the motional behavior they exhibit. Both flexible and rigid domains exist
within humic substances, though recent studies present conflicting interpretations
concerning the functional group composition of the two groups.
4.1.7 Molecular Models o f Humic Substances
Several recent attempts have been made to synthesize experimental information
concerning the composition and structure of humic substances into model humic
molecules for use in atomistic simulations. These model molecules can be used to
explore molecular scale interactions such as aggregation (Schulten and Leinweber, 1996,
2000; Schulten et al., 1998; Schnitzer and Schulten, 1998; Schulten, 1999a; Porquet et
al., 2003), sorption of xenobiotic contaminants (Schulten et al., 1998, 2001; Schnitzer and
Schulten, 1998; Kubicki and Apitz, 1999; Schulten, 1999b; Negre et al., 2001; Kubicki
and Trout, 2003), complexation of metal cations (Nantis and Carper, 1998a-c; Schulten
and Leinweber, 2000; Kubicki and Trout, 2003), and formation of organo-mineral
complexes (Schulten and Leinweber, 1996, 2000; Schulten and Schnitzer, 1997; Akim et
al., 1998; Leinweber and Schulten, 1998; Schulten et al., 1998; Shevchenko and Bailey,
1998a, b; Shevchenko et al., 1999; Bailey et al., 2001). An effective model humic
209
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
molecule is not necessarily a molecule that exists within an actual humic sample, but one
that captures functional qualities observed experimentally, and can provide insight
regarding interaction mechanisms of interest. In fact, given recent evidence for a model
of humic substances as an association of diverse and discrete molecules, a system too
large and complex for current hardware and simulation techniques, the model humic
molecules used today possess either an average or partial structural and functional group
makeup as compared to empirical observations.
One method for designing a model humic molecule is to collect an enormous
number of experimental observations regarding a single specific humic substance, then
build a molecule or set of molecules that have the major properties characterizing this
particular sample. Leenheer et al. (1989, 1995b) used this method to create a series of
molecules representing Suwannee River FA. The three basic structures have the identical
molecular formula C 3 3 H 3 2 O 1 9 , and feature nonpolar backbones consisting of two aromatic
rings and several short alkyl chains. Several carboxyl groups are present, most attached
to alkyl units. Numerous phenol groups are also present, as are linear and cyclic ether
and ketones groups. Several variations of these molecules are described as well,
featuring less common constituents including N and S functional groups, quinones,
carboxyl functional groups with especially low dissociation constants, and metal binding
sites.
One of these last, containing a phthalic acid metal binding site, was used to
explore the effects of H-bonding on humic substances (Porquet et al., 2003). A single
model FA molecule, featuring atomic charges and an optimized structure derived from ab
initio quantum mechanical calculations, as well as a cluster of 7 FA molecules, were
210
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
examined under both dry and solvated conditions using a modified version of the
AMBER force field (Cornell et al., 1995) and EM and MD calculations. Trajectory
analyses indicated that phenol groups formed intramolecular H-bonds with adjacent
carbonyls, especially in solvated systems, resulting in creation of more hydrophobic
interior regions. The presence of solvating water molecules did not decrease the number
of intramolecular H-bonds in the single molecule system, and may have encouraged
aggregation in the multiple molecule system through the formation of water bridges.
Hydrogen bonding appeared to play a minor role in aggregation of separate humic
molecules. A series of semi-empirical quantum mechanical PM3 (third parameterization
of the modified neglect of diatomic overlap (MNDO) method) (Stewart, 1989a, b)
explorations of metal cations bound to the phthalic acid metal binding site of the FA
molecule found that calculated standard enthalpies of formation were useful indicators of
complex formation, revealed the ability of the site to accommodate ions of various sizes
as a result of its locational flexibility, and documented the contribution of intermolecular
H-bonds to interactions with hydrated cations (Nantsis and Carper, 1998a-c).
A similar model for the Suwannee River FA, based on the same datasets and
proposed by Leenheer (1994), has a molecular formula of C35H30O20 and is made up of
distinct components representing carbohydrate, lignin, and lipid residues. Kubicki and
Apitz (1999) performed minimizations of this model FA in isolation using the MM+
force field (Allinger, 1977), as well as semi-empirical PM3 (Stewart, 1989a, b) and ab
initio Hartree-Fock (Park and Sposito, in press) quantum mechanical methods. The
molecule was examined in a - 4 charge state, because under typical soil pH conditions the
four carboxyl groups present would be deprotonated. The MM+ force field was unable to
211
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
capture H-bond interactions, and was deemed unsuitable for simulations of this molecule,
while the PM3 and ab initio quantum chemical conformations featured realistic
functional group structures and interactions, as well as similar, open molecular structures.
When benzene and pyridine were added to the PM3 system, hydrophobic and H-bonding
interactions were observed between the model FA and the respective contaminants.
Kubicki and Trout (2003) then performed two sets of simulations exploring the
effects of hydration on this model FA molecule. First, a nonperiodic FA system was
simulated using PM3 MD and EM calculations. The open structure of the neutral,
isolated (without water) FA molecule was contrasted with the more condensed structure
of the deprotonated, isolated FA molecule, produced through extensive intramolecular
H-bonding, and the open structure of the deprotonated, solvated FA molecule, a result of
H-bonding with solvent water molecules rather than organic functional groups.
Minimizations of the solvated system with benzene and pyridine revealed a close
association of the contaminant molecules with the model humic substance, though few
H-bonds formed, due to competition from water molecules for this type of interaction. A
second set of MD simulations was performed on a solvated, deprotonated, periodic FA
polyanion system, featuring Na+ to balance the charge, and using the COMPASS force
field (Sun, 1998). An open structure with few intramolecular H-bonds was observed,
similar to that predicted by the PM3 simulation of the deprotonated, hydrated FA
molecule. Further MD calculations of the system with an added benzene showed little
interaction between the two organic molecules, indicating the inadequacy of the
minimizations described above as a means of finding the most stable configuration of a
system.
212
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Another well known humic substance, the Chelsea Histosol HA has been the
subject of extensive experimental analysis. Diallo et al. (2001, 2003) have integrated the
resulting information to construct several model HA molecules using computer assisted
structure elucidation (CASE). This method involves computer generation of several 3D
model molecules based on an exhaustive set of molecular fragments and interfragmental
bonds derived from experimental data. These model molecules were subjected to a series
of EM and annealing MD (NVT) simulations, under dry, periodic boundary conditions, in
order to purge any chemically unrealistic conformations and pack the molecules into
volumes close to the experimental bulk densities of HAs. Finally, MD (NPT)
calculations were used to evaluate the bulk densities and Hildebrand solubility
parameters (see Section 4.2.3) of each molecule. Those model HAs featuring values far
from experimental measurements were discarded. Two sets of model molecules were
created using this method. A group of 20 large molecules, all with molecular mass
2801 Da and formula C120H111O70N7 S1, was constructed from 18 molecular fragments,
including 8 carboxyl groups attached to alkyl and aromatic backbones, 2 alkyl alcohol
groups, 2 phenols, 5 amino acids, and 1 pyrimidine (Diallo et al., 2001). A group of
18 small molecules, all with molecular mass 1016 Da and formula C45H43O24N1S1, was
constructed from a variety of precursors including 2 lignin degradation products,
2 polyphenols, 1 amino acid, and 1 carbohydrate group (Diallo et al., 2003).
The larger molecules were modeled using the Universal force field (Rappe et al.,
1992), with atomic charges assigned via the Cerius2 Qeq charge equilibration algorithm
(Accelrys Inc., 2001). After extensive simulation, only 6 of the 20 model molecules
achieved bulk densities and Hildebrand solubility parameters within the range of
213
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
experimental values (Diallo et al., 2001). These 6 model molecules were used to
generate thermodynamic input data needed to determine the binding constants of
hydrophobic organic compounds to dissolved humic substances via Flory-Huggins
solution theory (Kopinke et al., 1995; Diallo et al., 2001), a leading framework for
describing hydrophobic interactions with the humic matrix. The magnitudes of
calculated binding constants were consistent with experimental data, indicating that these
model humic substances are useful tools for exploration of hydrophobic sorption
interactions (Diallo et al., 2001).
The smaller molecules were modeled using the Dreiding force field (Mayo et al.,
1990), with atomic charges again assigned via the Cerius2 Qeq charge equilibration
algorithm (Accelrys Inc., 2001). Following the simulation series, only 5 of the 18 model
molecules achieved bulk densities and Flildebrand solubility parameters consistent with
experimental values (Diallo et al., 2003). An equimolar mixture of these 5 molecules
was used to generate a model 13C NMR spectrum comparable to the experimental
spectrum of Chelsea FLA.
Another means of creating a model humic substance is to modify the functional
group composition of precursor biomolecules, simulating natural degradation processes
such as oxidation that are known to occur in soil systems. Shevchenko and Bailey (1996)
used EM and MD annealing calculations to model several possible lignin-carbohydrate
complexes, thought to form within the cell walls of vascular plants and to contribute
significant source material for the formation of humic compounds in soils. The Tripos
force field (Clark et al., 1989) was used to describe each atom, and atomic charges were
assigned using Gasteiger-Huckel calculations (Purcell and Singer, 1967; Gasteiger and
214
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Marsili, 1980). Among the structures investigated was that of a lignin helix, made of 18
guaiacyl units connected by ether bonds, and coiled around an 18 unit cellulose chain.
The lignin component of this inclusion complex forms a protective shell around the
carbohydrate, which could protect it from oxidation reactions in the natural environment.
A hypothetical humic substance was created by mimicking the effect of oxidation
reactions on the lignin fragment, transforming several hydroxyl groups to carboxyl or
carbonyl groups. Simulations performed on the model oxidized lignin-carbohydrate
complex under dry and hydrated conditions indicate that the presence of the carbohydrate
chain substantially reduces conformational changes to the entangled structure, relative to
oxidized lignin alone (Shevchenko and Bailey, 1996; Akim et al., 1998). Oxidation and
the associated rise in hydrophilicity increases the number of H-bonds between the
carbohydrate and the lignin, as well as between the lignin and surrounding water
molecules (Akim et al., 1998).
Recognizing that humic substances are not fully protonated in most natural
environments, further simulations were performed on the model humic molecule with
deprotonated carboxyl groups, using Na+ to compensate for the negative charge of the
organic polyanion (Akim et al., 1998; Shevchenko et al., 1999). Ionization encourages
greater separation between the carbohydrate and lignin derived components of the model
humic complex, but preserves their intertwined nature. This may be contrasted with the
behavior of the oxidized and ionized lignin molecule modeled without a carbohydrate
core, which exhibits drastic conformational changes including the loss of helical structure
(Shevchenko et al., 1999). Simulations of the model humic complex under aqueous
conditions indicate that waters of hydration near the charged carboxylate groups are
215
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
tightly bound, while other water molecules experience unhindered exchange between
inner and outer coordination shells surrounding the organic molecule (Shevchenko et al.,
1999). Explorations of the interaction of this model humic molecule with mineral
surfaces will be discussed in Section 5.1. Animations of some of these MD calculations
are described in Section 3.1 (Bailey et al., 2001).
Another model humic molecule based on the structure of lignin, and informed by
HA elemental composition data, was proposed by Steelink (1985). To create this
structure, 4 phenylpropane units were connected with lignin-like linkages, and substituted
with carboxyl, phenol, quinone, ketone, and hydroxyl groups, resulting in a molecule
with formula C36H36O18, and molecular mass 756 Da. Examination of HA samples using
'H NMR, which detected the presence of numerous amide functional groups, and circular
dichroism, which identified substantial secondary structure within the humic samples via
their differential absorption of left- and right-circularly polarized light, suggested to
Jansen et al. (1996) that the Steelink HA molecule should be modified to include amide
linkages between the tetramers. The modified model humic molecule has an amine group
on one side, and a carboxyl group on the other, resulting in the formula C38H39O16N, and
a molecular mass of 765 Da. It also features 7 chiral centers, or 64 chemically unique
stereoisomers. Each of these was subjected to EM calculations employing the Tripos
force field (Clark et al., 1989). A random conformational search was performed on the
5 minimized structures with the lowest potential energies, in which coordinates of the
molecules were varied randomly within certain tolerances, minimized if the resulting
potential energy was within reasonable limits, and repeated until the structures showed
little change. The conformations with the lowest potential energies were further modeled
216
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
using MD and EM calculations. Common properties of the minimized model HA
molecules included alignment of 2 carboxyl groups, and stacking of aromatic rings. A
HA polymer linked via amide functional groups exhibited a helical structure after
minimization, consistent with the secondary structure observed via circular dichroism.
Further modification of this model humic molecule, to account for the
polysaccaride and protein content observed in HAs via numerous spectroscopies, resulted
in the Temple-Northeastem-Birmingham (TNB) humic building block, with molecular
formula C36H3oOi5N2(Davies et al., 1997). A series of stereoisomers and linked building
blocks for both the TNB and the original Steelink (1985) versions of this model humic
substance were subjected to minimization, random conformational search, and fixed
temperature and annealing MD calculations using Tripos (Clark et al., 1989) and MM+
(Allinger, 1977) force fields, as well as semi-empirical quantum mechanical PM3
(Stewart, 1989a, b) calculations (Sein et al., 1999). Isolated TNB and Steelink molecules
both exhibited the alignment of carboxyl groups and stacking of aromatic rings noted by
Jansen et al. (1996) with the previous version of the model HA (Sein et al., 1999).
Protonated and carboxyl-deprotonated versions of the two compounds within periodic
cells filled with water molecules were substantially unchanged relative to dry systems,
except for a few minor positional adjustments of functional groups involved in H-bonds
(Sein et al., 1999). Linked TNB molecules formed a rod-like helical structure, with two
parallel hydrophobic faces and two parallel hydrophilic faces running along the length of
the polymer, and a hollow center lined with carbonyl and alcohol functional groups
(Davies et al., 1997; Sein et al., 1997). However, Bruccoleri et al. (2001) presents a
critique of these simulations, observing that the MM+ force field (Allinger, 1977) does
217
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
not perform well for polar molecules, and is known to underestimate H-bonding
interactions and overestimate aromatic stacking behavior. The Tripos force field (Clark
et al., 1989) also was not designed to simulate molecules with the complexity of the
Steelink and TNB models, and is therefore likely to produce unacceptable errors
(Bruccoleri et al., 2001). Bruccoleri et al. (2001) performed more detailed PM3
calculations on the TNB molecule with and without associated water molecules, which
provided conformations that differ from those described previously (Davies et al., 1997;
Sein et al., 1997), indicated that solvation can have a strong effect on molecular
geometry, and suggested the possibility of self-assemby of two HA molecules as two
halves of a pseudo-spherical tennis ball, which could provide protected interior
sorption sites for xenobiotic contaminants.
A more inclusive approach to synthesizing data on humic substances was
employed by Stevenson (1982) to create his model HA molecule. This heavily
substituted, largely aromatic structure was devised after examination of chemical
degradation and reactivity data, as well as available spectroscopic insights, for a diverse
collection of humic substances. The molecule features a functional group composition
thought to represent a typical HA, including free and H-bonded phenols, quinone
structures, N and O bridging groups, and aromatic carboxyl groups. Kubicki and Apitz
(1999) performed minimizations on a simplified zwitterion (1 carboxyl deprotonated,
1 amine protonated) version of this molecule in isolation using the MM+ force field
(Allinger, 1977), as well as the semi-empirical quantum chemical PM3 technique
(Stewart, 1989a, b). The structure produced using the MM+ force field was considered to
be inadequate due to poor reproduction of charged functional groups and H-bonds. The
218
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
structure produced using the PM3 method appeared more realistic, and further PM3
simulations of the HA molecule in neutral and fully deprotonated charge states indicated
significant structural differences with protonation state. Later PM3 MD simulations of
the HA molecule with a charge of - 4 (5 carboxyls deprotonated, 1 amine protonated), this
time under solvated conditions, illustrated a trend of an open and extended structure when
multiple negative charge sites were present, and a compact structure when an Al3+was
added to neutralize a portion of the charge (Kubicki and Trout, 2003). When pyridine
was added to these solvated systems, hydrophobic interactions were observed between
this aromatic molecule and the model HA.
Synthesis of modem data on humic substances allowed Schulten and co-workers
(Schulten and Schnitzer, 1993; Schulten, 1999a, b) to create a series of model humic
molecules. Careful consideration of information gained through study of a wide variety
of humic substances via pyrolysis, IR, 13C NMR, XAS, electron microscopy, colloid
chemistry, elemental analysis and ecological theory (Schulten and Schnitzer, 1997),
resulted in a 2D structural concept for humic substances (Schulten and Schnitzer, 1993),
which upon conversion to a 3D molecule suitable for simulation possessed a formula
C308H335O90N5, and a molecular mass 5547.004 Da (Schulten and Leinweber, 1996). The
model humic molecule may be described as consisting of an alkylaromatic backbone,
substituted with numerous carboxyl, phenol and alcohol, ketone, ester, methoxy, ether,
and heterocyclic N functional groups. Its 3D structure contains many voids and clefts
suitable for sorption interactions. Subsequent refinements to the model structure include
a) reduction of alkyl chains to 10 or less C atoms in length, with an average length of
5 C atoms (Schulten, 1999a), b) insertion of trisaccharide (C18H32O16) and hexapeptide
219
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
(C26H46O10N10) molecules into voids within the organic matrix, to mimic association of
carbohydrate and protein materials within humic substances (Schulten and Schnitzer,
1997), c) addition of up to 35 water molecules to positions within voids and near reactive
surfaces, to replicate the strong association of water with humic materials (Schulten,
1999a), d) addition of numerous carboxyl and phenol groups to match atomic C:0 ratios
found in natural humic substances (Schulten, 1999a), and e) addition of cyclic S
functional groups (Schulten, 1999a). Updated model humic complexes (including
associated organic and water molecules) may be made up of as many as 1,262 atoms,
with molecular formula C447H492O306N15S2 , and molecular mass 11034.873 Da (Schulten,
1999a) (Figure 4.1).
A comparison of the generalized model humic molecules designed by Schulten
and co-workers (Schulten and Schnitzer, 1993; Schulten, 1999a, b) to concepts of humic
substances developed through analysis of recent experimental data highlights promising
similarities as well as troubling differences. All of the Schulten model molecules are
significantly larger than the average molecule of a typical soil humic fraction (Section
4.1.1). It would appear that the design of these model molecules was informed by an
earlier concept of humic substances as macromolecular polymers, rather than the current
concept of supramolecular associations (Piccolo, 2001). However, it is quite possible
that this obvious deviation from current theory is not a major disadvantage in terms of
molecular simulation. Such large humic molecules support numerous functional groups
and larger molecular fragments, allowing exploration of more diverse and complex
interactions than is possible with smaller, simpler model molecules. In addition, a large
220
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
model humic compound may behave much like a group of strongly associated smaller
molecules, common components of many soil environments.
* .
.
? *
" - I .
P
# - J *#
' f fc
* -
a * ' V V
S v * . * ,
n
V . - 4; .4 * * ' >
* . S w f . k
' 1" a*V'*k ' * *>
*. ' ** - J* -mV * " *
' i -i v ' . ><: A i * , >*, * - v V -4 *
|T ^ f ' V V / 1 v * '
- ': JW *v. - *
f ^ < ,0
* ? .a % . f <,
i f , - ^ ;- *> y
, 4 * ' < .
* * .#;*vk > v
a. , k t
Figure 4.1. Image of a model humic molecule similar to that described by Schulten
(1999a), including associated trisaccharide, polypeptide, and water molecules. Grey
spheres represent C, red spheres represent O, white spheres represent H, and N and S are
blue and yellow, respectively.
The functional group composition of the more refined versions of these model
molecules is similar to those found experimentally. The humic backbone consists of both
alkyl and aromatic functional groups. Numerous carboxyl, phenol, and alcohol groups
provide polarity perhaps more consistent with a FA than a HA (Stevenson, 1994). A few
221
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited without per mi ssi on.
of the carboxyl groups identified by Leenheer (Leenheer et al., 1995b, 2003) as
dissociating under low pH conditions are also present. A variety of other O functional
groups are scattered throughout the later versions of the humic molecule, including ether,
ester, ketone, and aldehyde groups. The quinone structures found in the model humic
molecules are not conjugated to benzene rings, and are extremely numerous, inconsistent
with ESR spectra of humic substances (Senesi, 1992). Nitrogen functional groups
include amide, amine, and heterocyclic N, as detected via spectroscopic and pyrolytic
means. The cyclic S functional groups added to the Schulten (1999a) dissolved organic
matter (DOM) molecule probably do not represent the dominant form of S found in
humic substances, but the reduced nature of the element is consistent with oxidation state
distributions found most frequently in aquatic or organic humic fractions. At the level of
molecular fragments, the polypeptide and polysaccharide molecules associated with the
humic compound provide proteinaceous and carbohydrate structures, while many of the
alkyl-linked aromatic groups resemble lignin-derived or alkylbenzene moieties. Reliance
on the assumption that the alkylbenzene molecules found in pyrolysates are fragments of
humic molecules, rather than secondary reaction products, may have resulted in an
overabundance of alkylbenzene fragments in these model compounds. Overall, the
modified model humic molecules designed by Schulten and co-workers (Schulten and
Schnitzer, 1993; Schulten, 1999a, b) possess functional group and structural
compositions consistent with current experimental data, excepting their large size.
A variety of systems have been simulated using this series of model molecules,
including isolated and hydrated humic molecules, humic aggregates consisting of as
many as 19 model molecules, and humic substances associated with xenobiotics such as
222
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
pentachlorophenol (Schulten, 1996,1999b; Schulten et al., 1998), atrazine (Schulten and
Leinweber, 1996; Schulten et al., 1998), hydroxyatrazine (Schulten, 1999b), DDT
(Schulten, 1999b), diethyl phthalate (Schulten et al., 2001), and imidazolinone herbicides
(Negre et al., 2001), under wet and dry conditions. Conclusions based on simulation
results include a) the characterization of these humic models as microporous and flexible
(Schulten and Leinweber, 1996), b) the observation of hydrophobic regions near long
alkyl chains that are inaccessible to water molecules (Schulten et al., 1998), c) the
abundance of H-bonds in humic interiors as opposed to external sites (Schulten, 1999a),
and their importance in binding associated polysaccharide and polypeptide molecules
(Schulten and Leinweber, 1996; Schulten and Schnitzer, 1997; Schulten et al., 1998) and
humic aggregates (Schulten et al., 1998), as well as polar xenobiotic compounds such as
atrazine (Schulten and Leinweber, 1996; Schulten et al., 1998), and hydroxyatrazine
(Schulten, 1999b), d) the surprising rarity of H-bonds between humic and water
molecules (Schulten, 1996; Schulten and Schnitzer, 1997; Schulten et al., 1998) and
generally neglible changes noted for humic-xenobiotic interactions taking place in the
presence and absence of water molecules (Schulten, 1996; Schulten et al., 1998), and e)
the prevalence of dipole and hydrophobic interactions, steric hindrance, or aromatic
stacking involved in immobilization of nonpolar xenobiotic compounds such as
pentachlorophenol (Schulten, 1996, 1999b; Schulten et al., 1998), DDT (Schulten,
1999b), diethyl phthalate (Schulten et al., 2001), and imidazolinone herbicides (Negre et
al., 2001).
However, all of these conclusions are based upon EM calculations from a single
initial configuration for each system. Resulting structures occupy local energy minima
223
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
only, and cannot be considered to represent equilibrium configurations (Section 1.4).
The MM+ force field (Allinger, 1977) was used in all of these simulations, a force field
known to produce unrealistic results when applied to molecules with many polar and
aromatic functional groups (Bruccoleri et al., 2001), and was further simplified with
special parameters for C atoms that allowed bonded protons to be ignored (Schulten et
al., 1998), all to expedite specific calculations. Thus, a promising model humic molecule
has yet to be evaluated with advanced molecular modeling methods.
Presented below are simulations of a slightly modified version of the model
molecule provided by Hans-Rolf Schulten and referred to as the DOM molecule, using an
advanced force field and a combination of modeling algorithms better able to produce
equilibrated molecular structures and assess the ability of the humic model to reproduce
experimental properties. These simulations explore the conformation of DOM under dry,
packed conditions and hydrated, dilute conditions, as well as under deprotonated and
hydrated, cation-saturated conditions more common to soil environments. Comparisons
with experimental data indicate that the DOM molecule is a useful tool for studying
humic interactions under aqueous conditions.
4.2 Methods
4.2.1 The COMPASS Force Field
The COMPASS (Condensed-phase Optimized Molecular Potentials for Atomistic
Simulation Studies) force field (Sun, 1998), developed by Accelrys Inc., and available for
use with their Cerius2molecular simulation software package, is a relatively new and
sophisticated all-atom force field suitable for simulation of both organic and inorganic
224
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
molecules. In order to model accurately such a broad variety of substances, the same
elements in different chemical environments are classified as different COMPASS atom
types (Sun, 1998). For example, an unsubstituted, aromatic C is classified as atom type
c3a, while an unsubstituted, alkyl C is classified as atom type c4. Each atom type has its
own set of interaction constants for use in the potential energy equation described below
(Sun, 1998):
= E [ \ ( b - K f + *3 (b b . ) + ' )
b
zr*J(e-0)2+M0-0j+*4(0-0..)4i+
e
- cos 0) + k2(l - cos 20) + k3(l - cos 3^)]
Z k 2Z2+Y , k ( b - b 0) ( b ' - b ' 0) +' k { b - b 0) ( e - & 0)+
X bfi
^ { b - b a)[ki cos0 + k2cos20 + k3cos3^] +
b,0
- 0o) [kxcos 0 + k2cos 20 + k3cos 2>0\ +
0,<t>
2 > ( 0 - 0 o) ( 0 ' - 0 o)+ 2 k ( e - 6o) ( 0 ' - 0 \ ) cos^ +
+
[4.1]
ry uj
( o \ 9
rJL
K r* J
- 3
( o \ b
rJ_
y rv j
COMPASS has the same functional form of energy calculation as the PCFF (Polymer
Consistent Force Field) on which it is based (Sun, 1998). Interaction parameters between
atoms i and j may be divided into two categories. Valence terms represent internal
coordinates of the bond (b), angle (0), torsion angle (<|>), and out-of-plane angle (x), as
well as cross-coupling terms such as bond-bond, bond-angle, and bond-torsion
interactions. Nonbond terms include van der Waals interactions, represented by a
Lennard-Jones (LJ) (6-9) function and LJ variables 8 (kcal/mol) and r (A), and
electrostatic interactions, represented by a Coulombic equation and variables q (partial
225
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
atomic charge, e) and (distance between atoms i and j, A). Nonbond terms are applied
to interactions between pairs of atoms separated by two or more intervening atoms, or
found on separate molecules.
Lennard-Jones variables are defined for like atom type pairs; for interactions
between different atom types, a 6th order combination law is used to calculate off-
diagonal parameters (Sun, 1998). Although the 6-12 LJ equation is more commonly used
to represent van der Waals interactions (see for example Section 2.2.1), it is known to be
too strongly repulsive, or hard, at short distances (Halgren, 1992). The 6-9 LJ equation
used within COMPASS more accurately models van der Waals behavior over short
distances, but may be too attractive, or soft, at longer distances (Halgren, 1992).
Electrostatic interactions rely on partial charges, which for an atom i are constructed
through summation of bond increments (8y) representing the charge separation between i
and covalently bonded j atoms (Sun, 1998).
COMPASS valence parameters and atomic partial charges were derived by fitting
quantum chemical data, while van der Waals parameters were determined by conducting
MD simulations of molecular liquids and fitting the simulated cohesive energies (see
Section 4.2.3) and equilibrium densities to experimental data (Sun, 1998). This novel
combination of parameterization techniques has resulted in a force field uniquely suited
to simulation of condensed phase substances. Validation studies employing isolated
molecules, molecular liquids, and molecular crystals, and representing 28 molecular
classes, indicate the COMPASS force field is capable of accurate and simultaneous
prediction of structural, vibrational, and thermophysical properties for a broad range of
organic and inorganic substances in isolated and condensed states (Sun, 1998). Further
226
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
research supports the accuracy of COMPASS predictions for diverse materials subjected
to a broad range of experimental conditions (Sun et al., 1998; Bunte and Sun, 2000; Yang
et al., 2000; Fried and Li, 2001; Launne et al., 2001; Kubicki and Trout, 2003).
Studies contrasting molecular structural and dynamic properties obtained from
COMPASS with those obtained from other force fields also demonstrate the high quality
of the COMPASS force field. Kubicki (2000) modeled a 683 atom, highly aromatic and
hydrophobic soot molecule with COMPASS and MM+ (Allinger, 1977), and noted that
only MD simulations using the COMPASS force field reproduced the layered structure
seen experimentally. On the other hand, Peters (2000) found calculations using these two
force fields produced very similar structures for C24 - C29 tricyclic terpanes, including an
abrupt conformation change between the C28 and C29 molecules. Skouras et al. (2001)
examined the mean residence times, Henrys constants, and heats of adsorption of several
simple gases (CO, CO2 , O2 , CH4 , Xe) interacting with metal oxide surfaces (Sn0 2 ,
BaTi0 3 , MgO) using both COMPASS and Universal (Rappe et al., 1992) force fields.
Although predictions based on the two force fields were qualitatively similar, comparison
with experimental data indicated that simulation with COMPASS was far more accurate
(Skouras et al., 2001). It is important to note that the accuracy of COMPASS predictions
is computationally expensive, a result of its complex description of energy terms
(Skouras et al., 2001). Kubicki and Trout (2003) produced similarly open and extended
structures for a model FA molecule in a deprotonated, hydrated state through MD
calculations using both the COMPASS force field and the semi-empirical quantum
chemical PM3 method, as described in Section 4.1.7. However, for more sophisticated
interactions including the covalent complexation of free Al3+by carboxylate groups, such
227
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
advanced quantum chemical methods will provide more realistic behavior than is
possible using force fields like COMPASS (Kubicki and Trout, 2003).
4.2.2 Simulation o f Bulk Water
Molecular modeling of the structure and dynamics of bulk water is a useful
9
exploration of the COMPASS force field and the Cerius v4.6 modeling environment,
and can provide information relevant to later work on hydrated organic systems. System
parameters described below were chosen to facilitate comparison with simulations of
MCY water (Sposito et al., 1999), an excellent mimic of experimental measurements of
structure and dynamics under fixed density conditions (see Section 2.1).
Creation of the bulk water system began with a single water molecule that
featured MCY water molecule bond lengths (0.9572 A) and angles (104.52) (Matsuoka
et al., 1976). The Cerius2 Amorphous Builder module was used to clone this molecule,
then build a 3D periodic system featuring these molecules in random positions and
orientations (Accelrys Inc., 2001). The complete system measured 20.000 A on each
side, and contained 267 water molecules, resulting in a density of 0.998 g cm'3. The
COMPASS force field was loaded via the Open Force Field (OFF) Setup module, and
then used to define all atom types and charges. To generate the energy expression for the
bulk water system, a real-space cutoff of 9 A was employed for computation of short-
range van der Waals interactions, while the 3D Ewald summation method with a real-
space cutoff of 9 A and a reciprocal space cutoff k of 3 A'1was used for the calculation
of long-range electrostatic interactions (as described in Section 2.2.1).
228
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The first step in the bulk water simulation procedure is energy minimization. The
OFF Methods module of Cerius2allows the user to minimize structures using a variety of
different algorithms and sets of convergence criteria (Accelrys Inc., 2001). The Smart
Minimizer offered by Cerius2uses a robust but less accurate algorithm during initial
stages of the calculation, then switches to a less robust but highly accurate method
(Accelrys Inc., 2001), and was used for all minimizations described here. The periodic
bulk water system contained only flexible molecules, but Cerius2 minimization can be
performed for any combination of rigid or flexible molecules, and for periodic or
nonperiodic systems. By default, during Cerius2minimization, all cell lengths and angles
are allowed to vary. However, for these bulk water simulations, all cell parameters were
fixed. Trajectory files, updated every 10 steps, were saved for each minimization,
providing a way to monitor the EM run. High level convergence criteria were used to
mark the completion of a calculation, criteria that include an overall system energy
difference between EM steps of less than 1.000 x 10'4kcal mol'1, a maximum atom
displacement of less than 5.000 x 10'5A, and a root mean square atom displacement of
less than 1.000 x 10'5A. The total potential energy, as well as the energy in each of the
terms described in Equation 4.1, was recorded for the final configuration of the bulk
water system after minimization.
As discussed in Section 1.4, EM algorithms are not an efficient means of
sampling a large portion of the possible configurations of a molecular system. It is likely
for a minimization calculation to result in a structure that is associated with a local, rather
than global, potential energy minimum. In order to force the bulk water system out of
any local energy minimum traps, a MD calculation under annealing conditions is
229
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
performed as the second step of this simulation sequence. Annealing may be described
as the process of heating, then cooling a substance in order to change its molecular
structure. In the world of molecular modeling, annealing calculations are used to force
systems into stable energy states more efficiently than would occur if the system were
held at the target temperature throughout the run.
The Cerius2 OFF Methods module provides the ability to perform MD
calculations using the Verlet leapfrog algorithm (Allen and Tildesley, 1987). Simulations
may be performed for constant NVE, NVT, NPH (P = pressure, H = enthalpy), or NPT
ensembles. The final EM configuration of the bulk water system was used as the starting
point for simulations with the canonical (NVT) ensemble, and the Berendsen
(T DAMPING) method of temperature control (relaxation constant = 0.1 ps) (Berendsen
et al., 1984). Initial velocities were assigned using a Maxwell-Boltzmann distribution
centered around 300 K. A 0.5 fs timestep was used, and the trajectory file was updated
with atom coordinates and velocities every 200 steps (0.1 ps). The water molecules
remained flexible for these calculations; in fact, the Cerius2MD algorithm cannot model
molecules defined as rigid, unless they are part of the lattice of a periodic system. A
complete annealing cycle was performed from 300 K to 600 K and back, with
10,000 steps (5 ps) at each 100 K stage. Final atom positions and velocities from each
preceding annealing stage were used as the initial input for the next stage. The complete
annealing cycle ended at 400 K, so a final MD (NVT) simulation of 5 ps at 300 K
brought the bulk water system back to the target temperature.
The annealing MD cycle was followed by another EM calculation and collection
of resulting potential energy data. The final structure of the system was then used as
230
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
input for 25 ps of MD (NVT) at 300 K, in order to examine the dynamic properties of the
bulk water system. As described in Section 2.2.3, trajectory information was used to
provide self-diffusion coefficients (Dw) using a three-dimensional Einstein relation
(Equation 2.3) (Allen and Tildesley, 1987). The mean-square displacement for the water
molecules was calculated using functions accessed via the Analysis card within the OFF
Methods module. The first 2 ps of the MD run were ignored, as the system had not
reached 300 K during this period. The smallest value of Dw that can be determined for a
23 ps simulation is around 0.3 x 10"9 m2 s'1(method of calculation described in Section
2.2.3). Average potential energy data were collected for this range and divided by the
number of water molecules in the simulation cell, to determine energy values per mole of
water. The distribution of bond lengths and angles present was calculated using the
Geometry/Measurements menu. Hydrogen bonds were calculated using a maximum
H - acceptor distance of 1.8 A and a minimum donor - H - acceptor angle of 120
(Schnitzer and Schulten, 1998). Radial distribution functions (Equation 2.2) were
calculated for O-O, O-H, and H-H atom pairs using the Analysis card within the OFF
Methods module, a cutoff distance of 9 A, and a distance interval of 0.1 A. Coordination
numbers were calculated from each RDF with the aid of the trapezoidal area function of
Matlab v5.3 (MathWorks Inc.). The OFF Methods/Analysis card also provided the
means to calculate the dipole moment of the system.
The last step in the bulk water simulation sequence is minimization of the final
configuration of the MD (NVT) calculation described above. The configuration
produced by this calculation is considered to occupy a stable energy state for a
temperature of 0 K, suitable for structural analysis described above, as well as calculation
231
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of the model IR spectrum. The Cerius2 IR-Raman module was used to calculate a model
IR spectrum for bulk water from 400 to 4000 cm'1, in absorbance mode and using default
peak shape (Lorentzian) and peak width (10 cm'1). The IR calculation requires a highly
minimized molecular configuration, and is achieved through a computation to estimate
harmonic vibrational frequencies via the second derivative of the molecules potential
energy surface, followed by an estimation of the intensities of each vibration based on the
movement of the fixed atomic point charges along mode vectors (Diem, 1993).
Simulation data were compared to experimental measurements of bulk water properties,
as well as to molecular modeling results using the MCY (Matsuoka et al., 1976), TIP4P
(Jorgensen et al., 1983), and SPC/E (Berendsen et al., 1987) water models (Section 2.1).
4.2.3 Simulation o f the Schulten DOM Molecule
A slightly modified version of the model DOM molecule described by Schulten
(1999a), was exported from HyperChem (Hypercube Inc.) using the Brookhaven Protein
Data Bank (.pdb/.ent) format, and provided for these simulations. This file could not be
loaded into the Cerius2modeling environment until it was modified by hand to remove all
but the last of the END lines. The Brookhaven format does not contain information on
bond order, so the Cerius2 Build menu was used to calculate the order of all bonds. At
this point, a small adjustment to the structure was made. A cyclic dinitrogen functional
group present in the organic molecule was changed from lH-pyrazole (N nearest
neighbors) to imidazole (N separated by a C atom), as the latter is a common substituent
of organic molecules such as DNA, and therefore more likely to exist in the humic
232
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
fraction of soils. This was the only structural modification made to the molecule
provided by Hans-Rolf Schulten.
The DOM molecule modeled here consists of a Schulten HA molecule covalently
bonded to a hexapeptide and a trisaccharide, is made up of 1157 atoms, and has the
molecular formula C447H421O272N15S2 , resulting in a mass of 10419.3 g mol'1. It also
features 35 associated water molecules, such that the partially hydrated DOM system
contains 1262 atoms, has a molecular formula of C 4 4 7 H 4 9 1 O 3 0 7 N 1 5 S 2 , and a mass of
11049.9 g mol'1. The partially hydrated DOM molecule described by Schulten (1999a) is
also made up of 1262 atoms, but features a slightly altered molecular formula:
C447H492O306N15S2 , with associated mass 11034.9 g mol'1.
After loading the COMPASS force field and using it to assign atom type, mass,
and charge for all atoms within the DOM molecule, an error message indicated that the
force field does not possess bond increments for a few of the bonds present within the
y
DOM molecule. Using the Cerius Discover module (Accelrys Inc., 2001), 13 DOM
bonds were identified as outside the scope of the COMPASS force field: 10 C-0 bonds
between oxygenated functional groups (alcohol, ether, ester) substituted to
p-benzoquinone units, 2 C-N bonds between primary amines bonded to the same C,
which is in turn double bonded to N, and 1 C-C bond between neighboring ketone
groups. Because the force field was not designed to include these bonds, it cannot
determine the charge separation that should exist between the atoms involved, so the
calculation of the partial charges of these 26 atoms is likely inaccurate. Although the
13 structural fragments described above may be considered chemically improbable, they
are a very small percentage of the total DOM molecule, and may be ignored as long as
233
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
they do not play a disproportionate role in any of the interactions explored using this
system.
The simulation procedure employed for molecular modeling of the DOM
molecule is based on that outlined by Diallo et al. (2001), and is designed to determine
whether the model molecule can produce realistic values for bulk density and the
Hildebrand solubility parameter (defined below). The first simulation stage is a lengthy
packing procedure that aims to remove any unrealistic conformations from the DOM
molecule, and free it from local energy minimum traps. It began with EM of the
nonperiodic DOM system, followed by use of the Amorphous Builder module to insert
the DOM into a periodic cell with a density of 0.50 g mol'1. The periodic DOM system
was then subjected to an EM - annealing MD - EM cycle similar to that described for the
bulk water system, except that each annealing stage lasted 15 ps. Using the Crystal
Builder module, the cell dimensions of the periodic DOM system were reduced to
achieve a density of 0.60 g mol1, and a second EM - annealing MD - EM cycle was
performed, with 10 ps at each annealing stage to reduce the length of the calculation.
This packing cycle was repeated for densities 0.70, 0.85, 1.0, 1.15, and 1.25 g mol"1.
Potential energy values collected after each minimization were used to monitor packing
progress. After the packing procedure, MD (NPT) simulation for 50 ps using the
Andersen pressure control method (0.0001 GPa, cell mass prefactor 0.04) (Andersen,
1980) allowed the DOM system to achieve a bulk density appropriate for a temperature
of 300 K through variation of cell parameters, and subsequent EM produced a well-
equilibrated organic structure.
234
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Unusual energy fluctuations during the later stages of this simulation series
indicated a problem with the DOM system. The sources of this behavior were
determined to be both the associated water molecules, and a furan functional group
trapped in an unrealistic, folded configuration. Calculations performed upon removal of
the water molecules and unfolding of the furan resulted in acceptable energy profiles
during MD and EM runs.
Potential energy and density data were collected for the fully minimized DOM
configuration. Using the Cerius2 Geometry menu, free volume and van der Waals and
solvent accessible (solvent radius 1.4 A) Connolly surface areas (Connolly, 1983) were
measured. The dipole moment was calculated using the Analysis card within the OFF
Methods module. The distance distributions of elements with respect to the geometric
center of the DOM molecule were calculated using the custom-built atomicDistance
program (Mehlman, 2003). As described in Section 4.2.2, H-bonds were calculated for
the periodic DOM system as well as the isolated DOM molecule using a maximum
H - acceptor distance of 1.8 A and a minimum donor - H - acceptor angle of 120
(Schnitzer and Schulten, 1998). Determination of the potential energy of the isolated
DOM molecule, or the gas phase strain energy, provides information essential to
1/2 3/2
calculation of the Hildebrand solubility parameter (8, J cm' ):
where Ep is the condensed phase (periodic) potential energy, EnPis the gas phase
(isolated) potential energy, (the value obtained when Enp is subtracted from Ep is defined
as the cohesive energy), Na is Avogadros number, and Vp is the cell volume (Diallo et
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[4.2]
235
al., 2001). Because the Ewald summation method cannot be used for nonperiodic
systems, a different method is necessary to calculate long-range interactions. Here a
cubic spline switching method is used to sum both electrostatic and van der Waals energy
contributions. Interactions occurring over distances longer than the spline-on value (ron),
but shorter than the spline-off value (r0ff), are multiplied by the spline function described
below, creating a gradual reduction in energy contributions, rather than an abrupt
discontinuity (Accelrys Inc., 2001).
isolated DOM molecule. Finally, the IR-Raman module was used to calculate a model IR
spectrum for the DOM molecule. The modeling data were compared to simulation
results published by Schulten (1999), as well as to available experimental measurements.
4.2.4 Simulation o f the Hydrated DOM Molecule
Hydration of the dry DOM molecule in its final configuration was accomplished
using Insightll v2000.1 (Accelrys Inc.). The DOM molecule was imported into this
program using the Biosym (.car) format, then surrounded with a layer o f water
molecules 5 A thick using the Assembly/Soak command. A total of 733 water molecules
were added to the DOM system. The Insightll Builder module was needed to merge the
DOM and its water shell into a single assembly, which was exported using the
1.0
switching function =
0.0
A ron of 20.0 A and a r0ff of 21.0 A were used for calculation of the potential energy of the
236
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Brookhaven file format. Because this format lacks bond order information, after
importing the hydrated DOM system into Cerius2, the Brookhaven DOM molecule was
replaced with the original molecule in the native Cerius2 file format (.msi).
The hydrated DOM system was subjected to minimization, then to four annealing
MD (20 ps / annealing stage) - EM cycles, in order to bring the molecule into
equilibrium with its environment more efficiently. The hydrated DOM system is
nonperiodic, designed to simulate the structure of the organic molecule in a dilute
solution. The cubic spline switching method mentioned in Section 4.2.3 was used to sum
electrostatic and van der Waals energy contributions. Initial calculations were made
using relatively low ron and r0ff of 9.0 and 9.5 A, respectively, to reduce the length of the
computations, while later calculations featured values of 20.0 and 21.0 A, respectively, to
provide more accurate results. After this equilibration series, a MD (NVT) calculation of
50 ps at 300 K was followed by a final minimization, assuring a well-minimized structure
for the DOM hydrate.
The data analyses mentioned in Section 4.2.3 were performed for the final
configuration of the DOM hydrate, with and without associated water molecules. All
H-bonds associated with the hydrated DOM molecule were calculated, as were the
intramolecular H-bonds found in the organic molecule upon removal of the water shell.
The number of intermolecular H-bonds, those between the organic and surrounding water
molecules, was determined by difference. The O-H bond lengths and H-O-H bond angles
of all water molecules were examined as well. Removal of waters of hydration aided
visual inspection of the DOM molecular configuration, as well as structural comparison
with the densely-packed, dry DOM molecule. The IR-Raman module was used to
237
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
calculate a model IR spectrum for the DOM only after removal of water molecules,
because the hydrated system was too large for the module to process. The spectrum of
the (dehydrated) DOM hydrate was compared to that of the dry DOM molecule after
normalizing absorbance values by the absorbance of the signal produced by the single
nitrile functional group present in the molecule (~2200 cm1). Nitrile is assumed to be
relatively unresponsive to environmental change (see Section 4.3.5), and its signal is the
only one present in the spectra that can be associated clearly with a specific functional
group.
4.2.5 Simulation o f the Deprotonated, Na-saturated DOM Hydrate
A fully protonated DOM molecule can exist only under highly acidic pH
conditions uncommon to soils. In order to simulate DOM experiencing a near neutral pH
environment, the molecules carboxyl groups must be deprotonated to form carboxylates,
and the resulting negative charge must be balanced with common metal cations.
Published methods for modeling deprotonated, Na-saturated organic molecules have
ignored the hydration state of Na+(Akim et al., 1998; Sein et al., 1999; Shevchenko et al.,
1999), despite extensive experimental evidence indicating these cations are coordinated
by 6 water molecules in aqueous solution (Ohtaki and Radnai, 1993). Presented here is a
novel simulation technique designed to titrate the DOM molecule with hydrated Na+in
a stepwise fashion. Though this procedure requires an unusual level of direct
manipulation of the system by the modeler, it is believed to account for cation hydration
and to mimic dynamic interaction of solutes with the DOM molecule, thus producing a
final conformation more appropriate to natural systems.
238
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Eleven carboxyl H atoms were removed from locations on the exterior of the
organic molecule in its final dry DOM conformation. The COMPASS force field was
used to retype all atoms, producing atom type and charge adjustments that transformed
the deprotonated carboxyl groups into carboxylate groups. Eleven energy minimized
Na+-6H20 hydrates were placed near these carboxylate groups, such that both Na-
carboxylate O distances were near 5.2 0.2 A (5.2 A being the length of 1 Na+ radius, an
interceding (water) O diameter, and 1 (carboxylate) O radius (Winter, 2003)). The
nonperiodic system was subjected to an EM - annealing MD (NVT) - EM cycle, using
ron and r0ff values of 20.0 and 21.0 A, respectively, and annealing stages of 10 ps. These
steps were repeated 6 times, with deprotonation moving from exterior to interior carboxyl
groups to emulate the process of exchange postulated by von Wandruszka et al. (1999),
until all 77 carboxyl groups were replaced with carboxylates associated with Na-hydrates.
Cramped conditions in the interior region required a more flexible approach to placement
of the Na-hydrates, resulting in shorter initial Na-carboxyl O distances.
Hydration was performed after this titration procedure, so that computations
associated with the previous steps could be performed in a reasonable amount of time.
Insightll was used to coat the Na-DOM complex with a 5 A thick layer of water, as
described in Section 4.2.4. With the addition of the 1272 water molecules making up this
hydration shell, a total of 1734 water molecules were now present in the Na-DOM
complex. A single EM - annealing MD - EM cycle was performed to equilibrate the
system, with just 5 ps for each annealing stage, to reduce the length of these
computations. This was followed by a MD (NVT) simulation of 50 ps at 300 K, and a
final EM run.
239
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Minimized energy distributions were obtained after each titration and hydration
step. Data analyses for the final configuration of the hydrated Na-DOM complex, as well
as for the dehydrated, deprotonated organic polyanion, were performed as described in
Sections 4.2.2 and 4.2.3, and the values measured were compared to those produced by
the protonated DOM molecules. The distribution of distances between each Na+and the
O atoms of the carboxylate group whose charge it balanced initially was plotted. In
addition, RDFs were calculated for Na-0 atom pairs, as well as for Na-water O and Na-
carboxylate O pairs (9 A cutoff, 0.1 A radial length increments). The RDF data were
used to calculate Na-0 CNs as described in Section 4.2.2. A 9 A radius sphere
surrounding each of the 77 cations in this final configuration was then evaluated in order
to determine properties of the local cation environment, such as the type of carboxylate
groups closest to each cation, the type of complexation evident, and the presence of
exchange movement. To be considered part of a cation complex, a carboxylate oxygen
must be either the closest such atom to a particular Na+, or within the distance described
by the first RDF Na-0 peak, and therefore participating in inner sphere coordination with
the cation.
Both the fully hydrated Na-DOM system, and a subsystem containing only the
DOM, the associated cations, and the original waters of cation hydration, were too large
for calculation of a model IR spectrum. A dehydrated Na-DOM system was small
enough for this calculation, and its spectrum was compared to those of previous DOM
systems after normalization using the absorbance value of the nitrile signal ( - 2 2 0 0 cm'1).
240
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
4.2.6 Simulation o f the Deprotonated, Ca-saturated DOM Hydrate
A review of molecular modeling literature provides no examples of simulation of
a natural organic polyanion saturated with divalent metal cations. Given the preference
of carboxylate groups for more highly charged metal cations (Tipping, 2002), this
absence of modeling information must be remedied. Therefore, the titration procedure
outlined in Section 4.2.5 was modified for simulation of the saturation of a DOM
molecule with Ca2+.
A total of 22 exterior carboxyl protons were removed to form carboxylate groups
as the first step of this simulation series. These carboxylates were arranged in 11 pairs,
chosen carefully such that each member of a carboxylate couple was near the other, and
would not isolate currently protonated carboxyl groups from potential partners. Eleven
energy minimized Ca2+-6 H2 0 hydrates were placed near each carboxylate pair, keeping
the shorter of the Ca-carboxylate O distances from each carboxylate at around 5.2 A (the
length of 1 Ca2+ radius, an interceding (water) O diameter, and 1 (carboxylate) O radius
(Winter, 2003)). This was followed by an EM - annealing MD - EM cycle, using the
same parameters mentioned in Section 4.2.5. The titration steps were repeated four
times, beginning with exterior carboxyl groups and ending with interior groups. During
the fourth titration step, only 10 carboxylate groups were created and 5 Ca2+ were added,
for a total of 38 Ca2+. The DOM molecule contains 77 carboxyl groups, so one of these
was left protonated to permit charge balance with the divalent cation. This protonated
carboxyl group was located at the end of an alkyl fragment that may be described as
2-pentenoic acid, and which has the highest dissociation constant (-4.76) of any
molecular fragments of the DOM molecule in isolation (American Chemical Society,
241
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
2001). A higher dissociation constant indicates that this carboxyl group has a statistically
higher chance of being protonated under near neutral pH conditions.
After titration, the Ca-DOM complex was hydrated with 944 water molecules, for
a total of 1222 water molecules in the system, and subjected to calculations as detailed in
Section 4.2.5. Analysis followed as described above, including collection of data
concerning energy distributions, volume, surface area, dipole moment, intra- and
intermolecular H-bonds, Ca-0 RDFs and CNs, the distribution of elements with respect
to the geometric center of the molecule, the distribution of Ca-0 distances from the
original Ca-carboxylate complexes, and the local environment of each cation. Each Ca2+
was considered to have at least two carboxylates coordinated to it. The two carboxylate
groups closest to each cation were classified as participating in complexation reactions,
as were additional carboxylate oxygen atoms within the distance defined by the Ca-0
RDF as characteristic of inner sphere coordination. The values calculated from each
analysis were compared to those produced for the Na-DOM complex and the protonated
DOM hydrate. Model IR calculations were performed for the dehydrated Ca-DOM, and
for the Ca-DOM featuring water molecules originally associated with the Ca2+ ions. The
fully dehydrated Ca-DOM spectrum was compared to previous spectra after
normalization using the nitrile signal. The partially hydrated Ca-DOM spectrum was
compared to those from similar experimental systems (Griffith and Schnitzer, 1975;
Griffith et al., 1984; Alberts and Filip, 1998).
242
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
4.3 Results and Discussion
4.3.1 The Bulk Water System
The structure and dynamics of liquid water are sufficiently complex to have
merited the development of several molecular models designed specifically to emulate
water behavior. Because the COMPASS force field was constructed to simulate a very
broad range of organic and mineral species without careful attention to the unusual
properties of liquid water, it is possible that the behavior of COMPASS water molecules
will not resemble adequately the behavior of their experimental counterparts in hydrated
systems. Before attempting to model such aqueous systems, it is essential to assess the
limitations of predictions based on the COMPASS bulk water system. Comparison of
structural and dynamic properties may be made with experimental data and with the three
most common and effective water models used in simulations of hydrated mineral and
organic systems. These contrasting water models are described in Section 2.1, and
feature rigid molecular structures, unlike the flexible COMPASS water molecule. They
also possess distributions of partial charge on the water molecule that range from nearly
identical (SPC/E) to about 70% greater (MCY) than that of the COMPASS water
molecule (Table 4.1).
The total potential energy of the COMPASS bulk water system is composed of
contributions from nonbond electrostatic and van der Waals interactions, and from
valence bond, angle, and cross-term interactions (Table 4.2). The three rigid water
models feature only nonbond energy terms. The total potential energy of COMPASS
water within the 300 K MD simulation is similar to that of MCY water (Table 4.1).
However, if valence contributions are removed from the energy expression (Table 4.2),
243
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
the COMPASS water model would have a potential energy similar to that of ideal water
and the TIP4P and SPC/E water models (Table 4.1). The COMPASS force field appears
to simulate adequately the potential energy of liquid water at 300 K. The potential
energy of COMPASS water at 300 K may be contrasted with that of fully minimized
COMPASS water (Table 4.2), which is substantially lower than both the ideal potential
energy determined from experimental data, and those calculated for the other water
models (Tables 4.1). The EM algorithm is designed to locate the energy minimum of a
system at 0 K, while the values presented in Table 4.1 are taken from systems
equilibrated at 300 K. Warmer systems have higher potential energies than cooler ones.
The average O-H bond length of the COMPASS water molecule is similar to the
empirical value provided for liquid water at 25 C (Soper and Phillips, 1986). However,
the average COMPASS H-O-H bond angle is smaller than that measured experimentally
or used in established water models (Table 4.1). This indicates that a bias favoring
smaller H-O-H bond angles has been built into the COMPASS o2* force field type
representing sp3 O within water molecules. An average H-O-H bond angle substantially
reduced from that of a strictly tetrahedral conformation (109.5) is likely to prevent the
formation of networks of H-bonds within model water system. Structurally, we can
expect COMPASS bulk water to feature a greater number of H atoms near water
molecules, fewer H-bonds, and a reduced net dipole moment, relative to experimental
studies.
The structural data for COMPASS water at 300 K presented in Table 4.1 and
Figure 4.2 are consistent with the trends outlined above, although the deviations from
established values are not always significant. The O-H RDF shows close agreement with
244
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
Table 4.1. Structural and Dynamic Behavior of COMPASS Water (MD, T = 300 K) as
Compared to Well-Established Water Models
COMPASS MCYa TIP4Pb SPC/EC experimental
O-H bond length (A) 0.975 0.026 0.9572 0.9572 1.000 0.98d
H-O-H bond angle () 99.90 4.66 104.52 104.52 109.47 105.4d
H partial charge (e) 0.42 0.717484 0.52 0.4238 -
Potential energy (kJ mol1) -35.48 0.27 -35.43e -42.13b -41.38 -41.4f
0 - 0 CN8 4.6 5.2 5.1 7.9h 5.4'
peak max. f (A) 2.85 2.8 2.8 2.77 2.8
peak min. i* (A) 3.35 3.4 3.5 3.34 3.5
peak 2 max. r1(A) - 4.3 4.3 4.5 4.6
O-H CN 1.8 2.1 2.0 1.9 2.0
peak max. r (A) 1.95 1.9 1.9 1.79 1.8
peak min. r (A) 2.45 2.5 2.5 2.46 2.45
peak 2 max. r (A) 3.35 3.3 3.3 3.3 3.3
H-H CN 6.6 6.0 - - 5.3
peak max. r (A) 2.60 2.4 2.4 2.4 2.4
peak min. r (A) 3.15 3.0 3.0 3.0 2.95
peak 2 max. r (A) 3.75 3.8 3.8 3.9 3.8
Dw(10 9m2 s 1) 5.92 0.03m 2.54, 2.26 4.5 4.4h 2.4P
aMatsuoka et al. (1976).
bJorgensen et al. (1983).
Berendsen et al. (1987).
dSoper and Phillips (1986).
Potential energy and structural information taken from Sposito et al. (1999).
fBeveridge et al. (1983).
Coordination numbers calculated as described in Table 2.2.
hStructural information taken from Heyes (1994).
Soper et al. (1997).
Value of r at the maximum of the first peak in the g(r) function (Figure 4.2).
Value of r at the minimum of the first peak in the g(r) function, or for the 0 - 0 RDF the value of r at the
inflection point marking the appearance of a distinct shoulder on the first peak (Figure 4.2).
Value of r at the maximum of the second peak in the g(r) function (Figure 4.2).
mSelf-diffusion coefficient calculated as described in Table 2.4.
nS.-H. Park, unpublished data.
Ferrario and Tani (1985).
pRobinson et al. (1996).
245
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
Table 4.2. Energy Distribution of the COMPASS Bulk Water System
Energy Term MD (T = 300 K) EM (T ~ 0 K)
(kJ mol'1) (kJ mol'1)
Total Potential -35.48 0.27 -49.20
Bond 4.09 0.23 1.51
Angle 2.63 0.19 1.59
Cross-term -0.32 0.03 -0.32
Van der Waals 4.34 0.39 6.78
Electrostatic -46.21 0.55 -58.77
experimental measurements as well as those obtained from other water models. The peak
maximum occurs at 1.95 A, indicating fewer or more elongated H-bonds, due to the
smaller H-O-H bond angle of the COMPASS water molecule. The H-H RDF plot
features longer peak maximum and minimum r values for the first peak, probably for the
same reason. The O-H CN of 1.8 is near the experimental value of 2.0. The minimized
water system features an O-H CN of 2.0, consistent with an idealized tetrahedral
structure. The MD-derived H-H CN at 300 K is 6.6, while the EM-derived CN at 0 K is
5.8. Both numbers are larger than those determined for liquid water, a result again of the
smaller H-O-H bond angle. Only 47 H-bonds existed in the final configuration after
MD simulation at 300 K, just 9% of those possible for an idealized tetrahedral network
featuring 267 molecules. The average COMPASS water dipole moment of 0.45 0.15 D
is far lower than experimental or modeled values (Robinson et al., 1996), as expected
given the reduced COMPASS water molecule angle.
The peak maximum of the 0 - 0 RDF occurs at a 2.85 A, a distance similar to
those taken from experimental and modeling literature. However, the shape of the
246
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
2.5
0-0
2
1.5
H-H
0-H
1
0.5
0
4 5 6 2 3 1 0
r (A)
Figure 4.2. Radial distribution functions for 0 - 0 , O-FI, and FI-H atom pairs of
COMPASS bulk water at T = 300 K (MD simulation).
0 - 0 RDF is not typical, in that instead of a clearly defined minimum near 3.5 A,
followed by a second peak, we see an abrupt change in slope indicating the existence of a
broad shoulder (see Figure 2.3 for a more typical 0 - 0 profile). This may indicate either a
less distinct transition between first and second shells of coordination, or the appearance
of a second coordination shell at shorter distances than suggested by experimental data
(Soper et al., 1997). The 0 - 0 CN of 4.6 calculated for COMPASS water is substantially
lower than the experimental CN, perhaps because the peak minimum r distance, defined
by the appearance of the shoulder, is small enough to have eliminated the contributions of
water molecules that are directly coordinated, but at greater distances. Upon
minimization, the water structure adopts an idealized, tetrahedral ice-like configuration,
247
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
with a distinct first shell structure and a CN of 4.0, consistent with experimental
measurements (Robinson et al., 1996).
The COMPASS bulk water structure provides structural parameters resembling
those derived from experiment and from specially designed water models, with
predictable deviations related to the force fields reduced ability to simulate a tetrahedral
H-bonded network. The differences detailed above are not surprising given the fact that
COMPASS was built to model a broad range of organic and mineral molecules, rather
than water alone. However, the COMPASS Dwcalculated from MD data is larger than
the experimental value by a factor of -2.5 (Table 4.1), indicating that diffusional
measurements based on these simulations may be exceptionally inaccurate. The lack of a
substantial H-bonded network likely allows molecules to move more rapidly, creating
increased diffusional motion within the COMPASS water system.
While established water models with fixed molecular structures are capable of
improved representations of H-bonding in liquid water (Table 4.1), they are unable to
provide vibrational information available from the COMPASS force field through
calculation of model IR spectra. The model IR spectrum of the bulk water system
features O-H stretching vibrations centered at 3490 and 3370 cm'1, an O-H bending
vibration at 1650 cm'1, and a broad region of librations (restricted rotations) centered
around 650 cm'1(Figure 4.3), reflecting the general structure of experimental water
spectra (Robinson et al., 1996). The COMPASS O-H stretching modes appear at higher
frequencies than typically observed for liquid water or ice (Robinson et al., 1996), and
possess a narrow and distinctly bimodal structure not found in experimental spectra. The
position and width of the COMPASS O-H stretching signals may indicate reduced
248
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
influence of H-bonds on water molecule vibrations relative to real liquid water. In fact,
the minimized bulk water simulation cell contains only 10 H-bonds (defined in Section
4.2.2), though before minimization 47 H-bonds were present. After EM, the more
ordered, tetrahedral arrangement of COMPASS water molecules with relatively small
molecular angles results in the disruption of these H-bonds. The bimodal peak shape of
the O-H stretching mode may be a result of the narrow peak widths (10 cm'1) used in the
calculation. The application of broader peaks would result in less distinct signal
structure, and could produce a broader and less bimodal peak shape. The position of the
O-H bending vibration is consistent with experimental spectra, though its relative
absorbance is much higher than expected (Robinson et al., 1996). The estimation of
absorbance intensity performed by the IR-Raman module is based on assumptions of
fixed partial charge and a small vibrational amplitude (Diem, 1993), and it is possible that
these assumptions are not valid for this motion under the COMPASS force field. The
librations produce signals similar to those found in experimental spectra. The Cerius2 IR-
Raman module does not determine combination or overtone bands seen in experimental
spectra.
Overall, the model IR spectrum compares well to experimental spectra of
condensed-phase water (Robinson et al., 1996). Differences listed above are consistent
with the reduced influence of H-bonds on molecule motions relative to experimental
systems, and suggest that absorbance intensity estimates may not be accurate for specific
vibrations. As a whole, data from the COMPASS bulk water simulation indicate that the
force field is capable of a satisfactory representation of the potential energy of liquid
water, and that structural behavior, while largely consistent with empirical measurements,
249
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
reflects the reduced ability of COMPASS water to form H-bonded networks as a result of
the smaller average bond angle of the water molecule. Predictions of dynamic properties
appear unreliable, perhaps also due to the lack of H-bonding interactions.
0.6
0.0
500 0 1000 2500 2000 1500 3500 3000 4000
wav enumb er (cm1)
Figure 4.3. Model IR spectrum of COMPASS bulk water system.
4.3.2 Energetic and Structural Properties o f the DOM Molecule
The potential energy distribution of the Schulten DOM molecule under dry,
densely packed conditions appears in Table 4.3. With the DOM system we see the
introduction of COMPASS torsion (bond twisting) and inversion (out-of-plane
interaction) valence energy terms, absent from calculations based on the simple, three
atom water molecule. Negative electrostatic and torsion energies are the most significant
factors resulting in the low total potential energy of the DOM molecule. It is not possible
to compare directly this potential energy distribution with that published by Schulten
(1999a) for a similar DOM molecule, due to the use of a different force field, MM+
(Allinger, 1977), different methods to sum interaction energy over long distances, and
250
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
different simulation conditions, including the use of a nonperiodic simulation cell, and
the presence of 35 water molecules associated with the organic matter. However, it is
interesting to note that electrostatic interactions play a relatively small role in the MM+
DOM energy distribution, resulting in a positive value for total potential energy
(Schulten, 1999a).
The final configuration of the DOM molecule features a bulk density of
1.36 g cm'3 (cell dimensions a = 24.233 A, b = 22.570 A, c = 23.649 A, a = 99.216,
P = 92.964, y = 86.004, cell volume 12726.2 A3). This value lies within the
1.2 - 1.4 g cm'3 range of experimental estimates for the bulk density of humic substances
cited by Diallo et al. (2001). The DOM model has a molecular surface area of 8672.4 A2,
and a solvent accessible surface area of 4397.0 A2. Both of these values are ~1000 A2
smaller than those provided by Schulten (1999a), which makes sense given that his DOM
configuration includes additional water molecules and has not been subjected to a
Table 4.3. Energy Distribution of the Minimized, Periodic COMPASS DOM Structure
Energy Term Energy (kJ mol'1)
Total Potential -18090.8
Bond 668.649
Angle 2603.79
Torsion -6826.20
Inversion 54.6539
Cross-term -982.165
Van der Waals -324.828
Electrostatic -13284.7
Gas Phase Strain -15265.0
251
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
packing procedure. The solvent accessible surface area can be divided by the mass of the
molecule to provide the specific surface area of 2541.3 m2 g'1, similar to the value
provided by Schulten (1999a), indicating that for this property, the additional surface area
provided by the water molecules is balanced by their added mass. The DOM molecules
volume is 8190.4 A3, marginally smaller than that documented by Schulten (1999a), due
again to an absence of water molecules.
The Hildebrand solubility parameter calculated from the potential (strain)
1/9 T/9
energies of the DOM molecule in condensed and gaseous phases is 19.2 J cm' , just
below the range of experimental estimates of 20.5 - 27.6 J1/2cm3/2 (Diallo et al., 2001).
Kopinke et al. (1995) consider the solubility parameter to be a rough indication of
polarizability. The solubility parameter calculated here provides evidence that the DOM
molecule, as represented by the COMPASS force field, may be somewhat less polar, or
more hydrophobic, than natural humic substances (Diallo et al., 2001). Despite a
potentially reduced polarity, the minimized DOM molecule exhibits many H-bonds, 47 in
the isolated or nonperiodic state (intramolecular H-bonds), and 41 more when examined
as a periodic system, indicating substantial intermolecular interaction with periodic
reflections of the organic compound. The isolated molecule described by Schulten
(1999a) possessed only 24 H-bonds after minimization, likely due to the less dense
configuration. The dipole moment of the minimized molecule is 44.4 D, indicating the
presence of significant organization of dipoles within the model DOM.
A simple minimization of the original DOM structure produces a molecule with a
noisy and featureless model IR spectrum (not shown). After the lengthy packing
procedure designed to bring the organic molecule into a well-equilibrated configuration,
252
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
it is possible to obtain a model IR spectrum (Figure 4.4) with clear signal structure, in the
form of clusters of peaks, which strongly resemble IR spectra of complex organic
molecules like humic substances (Stevenson, 1994). Though it is not possible to link
specific IR peaks to the vibrations of specific functional groups for the extremely large
DOM molecule due to limitations of the Cerius2 software, it is possible to do this for
smaller molecules. Examination of model IR spectra of several small fragments of the
DOM molecule provides information concerning the peak assignments described below.
1
0.8
0.6
o
0.2
o 4
4000 3500 3000 2500 2000 1500 1000 500 0
wavenumber (cm )
Figure 4.4. Model IR spectrum of dry DOM molecule.
The peaks around 3500 cm'1represent O-H stretching motions from alcohol,
phenol, and carboxyl groups, with much smaller contributions from N-H stretching of
primary and secondary amines and amides, especially below 3500 cm'1. These peaks are
shifted to higher wavenumbers relative to experimental values, and the O-H peaks are
quite narrow, as described in Section 4.3.1. The peaks at or above 3000 cm'1represent
C-H stretching movements of cyclic, aromatic, and alkene functional groups, while the
253
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
peaks just below 3000 cm'1represent C-H stretching motions from alkyl moieties. The
tiny peak near 2200 cm'1is caused by the stretching of a single nitrile group, and
indicates that the presence of a few unusual groups within the DOM structure will not
significantly impact the IR spectrum. Peaks in the 2000-1500 cm'1range are primarily
due to carbonyl 0 = 0 stretching motions of carboxyl, ester, aldehyde, and ketone
functional groups, along with aromatic ring vibrations, alkyl C-H bending movements,
and small contributions of N-H bending from amine and amide groups. Peak frequencies
calculated by the Cerius2IR-Raman algorithm occur over a much broader range than seen
experimentally. For example, model IR spectra of numerous humic fragments indicate
that carbonyl C=0 stretching vibrations contribute to peaks from 1860 to 1570 cm'1,
while the experimental range is 1750-1660 cm'1(Bellamy, 1975). In addition, calculated
carbonyl peaks have low absorbance intensities when compared with experimental
spectra, indicating the assumptions involved in the calculation of model IR spectra may
not hold for these vibrations, as described in Section 4.3.1.
The region from 1500-1000 cm'1contains contributions from a variety of different
functional groups, including C=C aromatic vibrations, C-H bending motions, C-O
stretching movements of carboxyl and ether groups, in-plane O-H bending vibrations of
carboxyls, alcohols, and phenols, and in-plane bending motions of C-N and N-H bonds.
The IR region of 1000-400 cm'1is equally tumultuous, containing signals caused by a
variety of C-H and O-H bending vibrations, carbonyl C=0 movements, carboxyl and
ether C-0 stretching and bending, and aromatic ring modes, with minor contributions
from N and S functional groups.
254
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
The DOM molecule, modeled with the COMPASS force field and a lengthy
packing procedure, is capable of density predictions consistent with experimental data for
dehydrated humic substances (Diallo et al., 2001). The Hildebrand solubility parameter,
calculated from potential energy data, is slightly lower than experimental estimates
(Diallo et al., 2001), perhaps an indication of increased hydrophobicity relative to natural
humic materials. However, a large dipole moment and the presence of numerous
H-bonds indicate the molecule does possess significant polarity. The model IR spectrum
calculated from the dry DOM structure is structurally quite similar to experimental
spectra (Stevenson, 1994), indicating that the Cerius2IR-Raman module is capable of
producing reasonable spectra for such large and complex organic molecules. The DOM
molecule reproduces some experimental properties of humic substances, and must be
considered a valuable model suitable for further exploration of molecular-scale
interactions via simulation.
4.3.3 Energetic and Structural Properties o f the Hydrated DOM Molecule
The potential energy distributions for the DOM hydrate with and without its water
shell, along with the energy distribution of the water shell in isolation, are presented in
Table 4.4. The hydrated DOM system has a low total potential energy due primarily to
electrostatic interactions, and offset partially by angle, bond, and van der Waals
interactions. When the hydrated system is compared to individual organic and water
components, it becomes evident that a substantial amount of electrostatic energy, and a
smaller amount of van der Waals energy, can be found in interactions between the DOM
and associated water molecules. The DOM molecule without its water shell has a total
255
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
potential energy higher than that of the dry DOM molecule in isolation (Table 4.3),
suggesting that structural reorganization to optimize interactions with water results in a
configuration less suited to dehydrated conditions. When the potential energy of the
water shell is divided by the number of water molecules present (733), a value of
-231.9 kJ mol'1water is obtained. This quantity is far lower than the -49.20 kJ mol'1
value obtained from minimization of bulk water (Section 4.3.1), an indication of the
strength with which the COMPASS force field favors the hydration process. However, a
direct comparison of these two molar potential energy values is unwise given differences
in the simulation conditions used, including periodic vs. nonperiodic simulation cells, and
alternate methods for summation of nonbond interactions.
Table 4.4. Energy Distributions of the DOM Hydrate, the Isolated DOM Molecule, and
its Water Shell
Energy Term Hydrated DOM
(kJ mol'1)*
Isolated DOM
(kJ mol"1)*
Water Shell
(kJ mol'1)*
Total Potential
-256001 -11584.7 -170001
Bond 8101.60 1783.18 6318.43
Angle 9385.00 5684.72 3700.32
Torsion -4795.07 -4795.07
Inversion 401.757 401.757
Cross-term -1783.51 -1095.99 -687.523
Van der Waals 5437.74 1783.43 4194.04
Electrostatic -272748 -15346.7 -183526
* mole refers to the entire system being simulated, not to its components; i.e. kJ per mole system.
The dipole moment of the hydrated DOM molecule after minimization is 63.9 D
with its water shell, and 42.5 D without it. The latter value is similar to that provided by
the dry DOM simulation (Section 4.3.2). In order to accommodate its aqueous
256
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
environment, the hydrated DOM has adopted subtle structural changes, resulting in the
reduction of intramolecular H-bonds and the appearance of DOM-water H-bonds. The
hydrated DOM features 15 intramolecular H-bonds and 13 H-bonds with associated
water molecules, for a total of 28 H-bonds, far fewer than the isolated, dry DOM
molecule. More DOM-water H-bonds would be expected if the added water molecules
were able to penetrate the organic matrix, rather than simply coating its exterior surfaces.
The same number of H-bonds was present prior to EM, indicating no substantial change
in this type of interaction due to the minimization run, in contrast to the behavior noted
for the bulk water structure (Section 4. 3. 1) . The porosity of the hydrated DOM system,
calculated via the volume ratio [ ( V t o t a i - V d o m ) / V t o t a i ] , is 0. 637, while the solvent
accessible surface area of the DOM molecule in its hydrated conformation is 4415.4 A2.
As discussed in Section 4.1.1, experimental studies indicate that in the presence of
water, humic substances rearrange themselves to form hydrophobic interior domains and
hydrophilic exterior surfaces (Chien and Bleam, 1997; von Wandruszka et al., 1999;
Ferreira et al., 2001; Simpson et al., 2001b; Nanny and Kontas, 2002). The distributions
of distances of C, O, and H atoms relative to the geometric center of the DOM molecule,
in both dry (Section 4.3.2) and hydrated states, were evaluated to determine whether this
type of structural reorganization had occurred (Figure 4.5). Trimodal distance
distributions are evident in the histograms calculated for the DOM hydrate system, a
product of the nonspherical shape of the organic molecule (Figure 4.6), and an indication
of greater organization relative to the dry, densely packed DOM structure. The
C histogram of the DOM hydrate features peaks within distance bins of 9-10, 11-12, and
14-15 A. The distribution of O distances within the DOM hydrate mimics that of C,
257
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Carbon
70
60 -
DOM hydrate dry DOM
50 -
30 -
20 -
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Oxygen
30 n
DOM hydrate
dry DOM
10 -
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hydrogen
50
DOM hydrate
40 -
dry DOM
30 -
20 -
10 -
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Figure 4.5. Histograms of organic C, O, and H distances relative to the geometric center
of the DOM molecule in dry (Section 4.3.2) and hydrated configurations. Values along
the x axes represent the upper boundaries of each 1 A distance bin.
258
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
although the two peaks at larger distances are 1 A further away, and there are
proportionally more O atoms at distances greater than 18 A. These differences are
consistent with small structural alterations that concentrate more polar O functional
groups towards outer regions of the DOM molecule. The H histogram exhibits peaks in
the same positions as the O histogram, due to the large number of protons associated with
carboxyl and alcohol or phenol groups within the organic matter. Visual comparisons of
the dry and hydrated DOM molecule (Figure 4.6) reveal numerous small changes to
molecular structure that allow the concentration of hydrophilic moieties on the exterior of
the organic matter with hydration.
Given the dramatic effect that the presence of water molecules has on the
configuration of the DOM molecule, it seems appropriate to expect the complex organic
substance to exert in turn some influence over the configurations of solvent molecules.
Inspection of water molecules reveals the average O-H bond distance within the hydrated
DOM system to be 1.000 A ( 0.016 A), and the average H-O-H bond angle to be 98.80
( 7.03). In the presence of the DOM molecule, COMPASS water molecules have
adopted smaller molecular angles on average than found under bulk solution conditions,
as well as longer O-H bond lengths (Table 4.1). Perhaps the reduced ability of water
molecules to form the tetrahedral structure typical of bulk water, given their distribution
in a layer only 5 A thick around a large organic molecule, allows them to relax into
conformations with relatively small H-O-H angles, evidently favored by the COMPASS
force field. The altered configuration of water molecules is unlikely to affect their ability
to form H-bonds with the DOM molecule.
259
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 4.6. Dry (left) and hydrated (right) DOM molecules (waters removed from views)
in the XY (top), YZ (middle), and ZX (bottom) planes. Grey represents C, red represents
O, green represents H (for visibility), and N and S are blue and yellow, respectively. In
the lower views, light blue highlights hydrophobic regions occupying more protected
positions with hydration.
260
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Information gathered from simulation of the hydrated DOM molecule provides no
evidence suggesting the organic substance is unsuitable as a model humic substance. In
fact, the segregation of DOM into interior, more hydrophobic portions and exterior, more
hydrophilic portions, as predicted by experiment, is further indication of the utility of this
model molecule in simulation work. The model IR spectrum of the DOM hydrate will be
discussed in Section 4.3.5.
4.3.4 Energetic and Structural Properties o f the Deprotonated, Na- and Ca-saturated
DOM Complexes
A comparison of deprotonated DOM polyanions saturated by different hydrated
cations must begin with a comparison of the cation hydrates themselves. Energy
minimized, octahedrally coordinated COMPASS Na+ features a Na-0 distance of
2.395 A, while Ca2+features a Ca-O distance of 2.357 A. The Ca2+hydrate possesses
over twice as much total potential energy as the Na+hydrate (Table 4.5), primarily
because of a larger electrostatic contribution given the divalent charge of the former, and
the slightly smaller Ca-O distance. The effect of the smaller cation-0 distance also can
be seen in the higher van der Waals energy contribution within the Ca2+hydrate. Valence
(bond, angle, and cross-term) energy contributions from the water molecules in the Na+
hydrate are one-tenth that of the water molecules in the Ca2+hydrate, indicating that the
structure of water is different in each system. Measurements of the minimized molecules
indicate that the H-O-H bond angle of water near the monovalent cation is 102.2, and the
O-H bond length is 0.963 A, while the bond angle of water near the divalent cation is
97.3, and the bond length is 0.977 A. The more intense electrostatic field surrounding
261
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ca2+appears to pull the water molecules into conformations with smaller molecular
angles and more elongated bond lengths than seen in COMPASS bulk water simulation
(Table 4.1). The weaker electrostatic field of Na+ helps water molecules attain a
conformation more similar to real water than that seen within the bulk water system,
perhaps indicating that the COMPASS force field parameters associated with water
molecules were designed for use in solutions of monovalent ions.
Table 4.5. Energy Distributions of Cation Hydrates
Energy Term Na+Hydrate
(kJ mol'1)
Ca2+Hydrate
(kJ mol'1)
Total Potential -449.893 -1036.19
Bond 1.05515 10.5097
Angle 1.88577 18.9406
Cross-term -0.288535 -2.91241
Van der Waals 48.5135 166.704
Electrostatic -501.059 -1229.43
After inspecting the differences between Na+ and Ca2+ hydrates in isolation, we
can now examine differences between deprotonated DOM molecules saturated by these
hydrates. Energy distributions collected after each step of the Na- and Ca-DOM titration
procedures (not shown) reveal a relatively monotonic decrease in electrostatic energy
proportional to the number of cation hydrates added to the system. Van der Waals energy
increases in a more exponential manner with titration, reflecting the increasingly strained
conditions created as hydrated cation begin to occupy sites in the interior of the organic
matrix. Among valence terms, angle and bond energies show steady increases with
titration, while torsion, inversion, and cross-term energies display more varied responses.
262
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Energy distribution information collected upon completion of the simulation
procedures indicates that the low total potential energies of the Na- and Ca-DOM
complexes are due primarily to electrostatic contributions, which are partially
compensated by the bond, angle, and van der Waals energy terms (Table 4.6). Although
the systems cannot be compared directly, as they contain different quantities of both
cations and water molecules, contrasting energy contributions from the organic-salt and
water shell components in isolation provide some interesting trends. The dehydrated,
Na-saturated organic polyanion possesses a total energy roughly three times that of the
fully protonated DOM structure after its equilibration in a water system, while the
Ca-saturated version possesses a total energy about six times that of the protonated DOM
(Table 4.4). The presence of carboxylate sites and compensating cations in the
Na-DOM and Ca-DOM systems creates favorable electrostatic conditions resulting in
substantially lower total potential energies overall. Although the Ca-saturated system
contains just under half the number of cations as the Na-saturated system, the divalent
Table 4.6. Energy Distributions of Na-DOM and Ca-DOM with and without Associated
Water Molecules, and of Water Molecules without Associated Organic Complexes
Energy Term Hydrate
(kJ mol'1)*
Na-DOM
Isolated
(kJ mol'1)*
Water Shell
(kJ mol'1)*
Hydrate
(kJ mol'1)*
Ca-DOM
Isolated
(kJ mol'1)*
Water Shell
(kJ mol1)*
Total Potential -726238 -32332.6 -501783 -558468 -67753.6 -290802
Bond 28037.5 3210.21 24827.3 18768.9 2682.88 16086.1
Angle 20233.7 7514.92 12718.8 15257.0 6355.96 8901.04
Torsion -4779.97 -4779.97 -4857.12 -4857.12
Inversion 459.265 459.265 469.838 469.838
Cross-term -3939.07 -1456.07 -2483.00 -2780.69 -982.679 -1798.01
Van der Waals 25848.9 4458.39 13911.8 16955.8 4160.18 8427.29
Electrostatic -792098 -41739.3 -550757 -602278 -75582.7 -322418
* mole refers to the entire system being simulated, not to its components; i.e. kJ per mole system.
263
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
charges on these cations, and their close interactions with an average of two monovalent
negative organic charge sites per cation, results in a near doubling of the electrostatic
energy found in isolated Ca-DOM as compared to Na-DOM.
The molar potential energies of the water molecules in each system, calculated as
described in Section 4.3.3, provide values that can be compared directly for all three
DOM hydrates. Water from the protonated DOM hydrate has the highest molar potential
energy, -231.9 kJ mol'1, followed by water from the Ca-DOM hydrate, with
-263.8 kJ mol'1, and finally water from the Na-DOM hydrate, with -317.6 kJ mol'1. The
variation in these values indicates that both the number and the arrangement of water
molecules affects the potential energy of the system. A final observation of interest may
be drawn from calculation of the non-additive energy present in the complete DOM
hydrate systems, or the difference between the total potential energy of the complete
system and its organic-salt and water components. Non-additive energy values are
similar for both Na- and Ca-DOM, at -192,122.4 and -199,912.4 kJ mol'1, but over twice
that present in the protonated DOM hydrate, -74,415.3 kJ mol'1. Although a direct
comparison is impossible given the variation in abundance of water and cations present in
the systems, the values suggest that more extensive interactions occur between the
deprotonated, cation-saturated organic molecules and their water shells.
Porosities of 0.815 and 0.753 calculated for the Na- and Ca-DOM complexes,
respectively, reveal a more porous organic structure than seen in simulations of
protonated DOM molecules (Section 4.3.3). A more porous organic matrix may provide
more opportunities for organic interactions with water molecules, as indicated by the
non-additive energy values calculated above. A visual inspection of the DOM salts
264
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
without associated water molecules (Figure 4.7) reflects this increased porosity relative to
the protonated DOM hydrate (Figure 4.6). Reorganization into less compact structures
containing many voids after deprotonation and cation saturation is consistent with AFM
images of dilute humic substances adsorbed to muscovites under aqueous conditions,
which indicate that these natural organic molecules form dense clusters when influenced
by acidic solutions, and more elongated networks when subjected to solutions close to
neutral (Maurice and Namjesnik-Dejanovic, 1999). Increased porosity is reflected in the
Na- and Ca-DOM systems, respectively. The Ca-DOM complex may possess lower
porosity than the Na-DOM complex as a result of tighter packing of the organic structure
organized by the divalent cations, or because fewer cations were introduced into the
interior of the organic molecule. Distributions of C, O, and H atoms from the geometric
center of the molecules (not shown) do not feature the distinct peak structures associated
with segregation of organic matter into more hydrophobic interior regions and more
hydrophilic exterior regions, as seen for the protonated DOM hydrate (Section 4.3.3).
This is to be expected, as the water molecules in Na- and Ca-DOM systems are located
throughout the organic matrices, not just at the outer surfaces, and therefore do not limit
the concentration of polar groups to exterior zones.
The Na-DOM system contains a total of 109 organic H-bonds, or 20
intramolecular H-bonds and 89 H-bonds between the organic matter and associated water
molecules. This dramatic increase in DOM-water H-bonds, relative to the protonated
DOM hydrate system, is linked to both the intrusion of water molecules into pores, larger
solvent accessible surface areas of 5731.4 and 5638.6 A2 for the organic portion of
resulting in increased porosity and a larger solvent accessible surface area, and the
265
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
'Y.
n ' S-5Bi
., v--.'<r Y ' Y V
' Av~'~
T L
A ':
> V, ,
> ' v Y Y / ; ' '/,-
f- WC*S- <i
V11
1<<f \ ? ''#'
j a y y a N;
y> ' - v
- s>^ -
m-v- ,
- A ' '
! ,V' iO' V ' * 1
. C_ - 4
: . * s 4, is -a
-' \' (
' Y a ", ."
w v - m ' A - . A /v
va . ^ r< ; .
i-.. V'C''?"'V'' \ -fi1 */,* / ,, s
r ' -Vv' ./ , Xr\ . A . : " '
' - '' - - ,<< ? '
V- '*' ' 2 ' A *
- ^ . V Y ' ' * v<.v
- ' A / * ; V
7 ^ \
> J J
4 .
Y J
> A ' /
-J '
S
i j
lY V
- . ) 7 l V r o V j , . - * * *. *
. -, - ' ( - A ,/ V-AA ' T,j- /<
" >v: **sY'*'/JY Y ,'*1
,. '* ' Y Y >>?r^U *I' V v i Y
-:M ? '*~J ('! J- . .. -
11'( 1/
'" A
Y >
'' * w /. , ''^'yy.
Y . Y r 7 Y Y r>'-
r - / , Y /; -v"Y ' .*
V:Y ' , '' ' f
''* / '-v v' U " vi
J T
r-
-<g f , \ *<-
' il'1
,vti i
1 - Y 1:'
. . ,vyY
'" '''" h' ' \ ' A-'
>J 4
A-
.
' ' ~if v V)/,o' > *7* I
2 ,K' \ i '
j-t- , /
'; J
*-'\\\
S t-'- ~s>
, Y . 7 . ^
. v
,<, T e\J> lA A, /s
-rntv-ite: . > " * * * -f.l Y r T A
) *^ % " ^ i ' it r ' | ^ \
/ * \ Vv--c -4 . i -
'. J . ,
'' *-,/ AT" ? 2' / ^
r T'' ;-! | 1 \ ' ,. ; .
Y - : 1 '
i 7 7 i ' v
Figure 4.7. Views of Na- (left) and Ca-DOM (right) molecules (water removed from
images) as seen in Figure 4.6. Brown symbols represent cation positions.
266
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
increased polarity of the Na-DOM molecule. The dipole moment of Na-DOM is 243.6
and 229.3 D under hydrated and dehydrated conditions, respectively. The Ca-DOM
system contains a total of 61 organic H-bonds, or 17 intramolecular H-bonds and 44
H-bonds between the organic matter and associated water molecules. The Ca-DOM
complex features fewer H-bonds than the Na-DOM complex because it has fewer total
water molecules, a lower porosity, and perhaps because it possesses a less polar
conformation, with a dipole moment of 133.5 D for the hydrate and 143.2 D upon
removal of associated water molecules. It is possible that the stronger electrostatic field
produced by the divalent cation forces water molecules and humic functional groups into
rigid positions near the Ca2+, and discourages any subtle changes to the structure of the
complex that would lead to the formation of H-bonds. Recent experimental evidence
indicates that the identity of the cation saturating humic substances can affect sorption of
nonionic organic compounds via both H-bonding and hydrophobic mechanisms (Yuan
and Xing, 2001).
Cation identity is expected to affect the shape of surrounding water molecules in
these hydrated DOM systems, just as it did in the isolated cation hydrates. Average
H-O-H bond angles and O-H bond lengths of water molecules in each of the DOM
hydrates are listed in Table 4.7. The bond angles noted for all three systems are
statistically indistinguishable. Though the average molecular angle of water molecules
surrounding Ca-DOM is similar to that found in the isolated Ca2+ hydrate, the same
cannot be said for Na-DOM relative to the Na+hydrate. Rather than featuring the larger
angle present in the isolated Na-6H20 system, Na-DOM water tends toward angles
similar to that of the protonated and Ca-saturated organic systems. The water O-H bond
267
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
lengths of all hydrated organic systems are longer than those found in the isolated cation
hydrates or in the bulk water simulations (Section 4.3.1). The altered conformation of
water molecules surrounding all three organic substances, regardless of cation identity,
provides further evidence to indicate that without enough space to form an approximation
of the tetrahedral structure of bulk water, COMPASS water molecules tend to adopt less
realistic configurations.
Table 4.7. Water Molecule Characteristics in Hydrated DOM Systems
Hydrates DOM Na-DOM Ca-DOM
O-H Distance (A) 1.000 0.016 1.013 0.026 1.010 0.024
H-O-H Angle () 98.80 7.03 98.02 8.74 97.66 8.39
Plots of N a - 0 and C a - O RDFs provide information concerning the average
coordination state of these cations (Figure 4 . 8 ) . The N a - O totai profile indicates substantial
coordination at N a - 0 distances of 2 A or less, and the N a - O carboxyiate profile specifies that
most of the O involved in these complexes is part of carboxylate functional groups.
Peaks present in the N a - O carboxyiate and N a - O water RDFs at slightly longer distances extend
the maximum r-value for inner sphere coordination to 2 . 6 A . The C N s calculated over
this range are presented in Table 4 . 8 , and may be compared to the experimental C N of ~ 6
for N a + in the aqueous state (Ohtaki and Radnai, 1993). Evidently, inner sphere
coordination with DOM, or complexation, results in substantially reduced coordination
for N a + , perhaps due to the restricted stereochemical environment of the cation. Because
many N a + are coordinated directly to carboxylate O, they are located close to other
organic moieties as well. Further coordination of these cations by water molecules is
2 6 8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.00025
total O
0.00020
0.00015
o
0.00010
carboxylate
0.00005
water
0.00000
5 6 3 4 1 2 0
r(A)
0.0005
total O
0.0004
0.0003
0.0002
0.0001
carboxylate
water
0
3 4 5 6 1 2 0
r (A)
Figure 4.8. Radial distribution functions between the cation and total O (grey),
carboxylate O (black), and water O (blue) for Na- (top) and Ca-DOM (bottom)
complexes.
269
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
restricted to positions not impinged upon by fragments of the DOM molecule. Alcohols,
phenols, and the carbonyls of ester and ketone functional groups contribute the portion
inner sphere coordinated O not attributed to carboxylate groups or water molecules.
Table 4.8. Cation Coordination Numbers for Na- and Ca-DOM Complexes
Ca-O Pairs Na+ Ca2+
Total 0 2.66 4.15
Carboxylate 0 0.65 1.29
Water 0 1.84 2.70
Other 0 0.17 0.16
The extent of complexation of Na+by carboxylate groups is inconsistent with
currently accepted models of cation complexation by humic substances, which suggest
that Na+ should associate with natural organic matter primarily as fully hydrated cations
in a layer of diffuse ions removed from specific organic charge sites (Tipping, 2002).
Acidimetric titration experiments show that the amount of cation associated with a
sample HA is dependent only on pH, and not on the concentration of Na+ in solution,
supporting the prevalence of a non-specific charge neutralization interaction, rather than a
direct organic complexation interaction (Bonn and Fish, 1993). The model titration
procedure used for these simulations may produce results favoring organic complexation
interactions over more distant, charge neutralization activity, as cations are placed near
newly deprotonated carboxylate groups at the beginning of each step in the sequence.
The first peak of the Ca-Ototai RDF occurs at a slightly higher r-value than seen in
the Na-DOM system, and is due primarily to the presence of carboxylate O (Figure 4.8).
Although Ca2+and Na+have similar radii, the higher charge of Ca2+ attracts more
270
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
O atoms, and this crowding discourages occupation of sites closer to the cation. The
Ca-O CNs, calculated through an r-value of 2.8 A (Table 4.8), reflect the high O
coordination of these cations relative to Na+. The Ca-Ototai CN is still lower than the
experimentally determined aqueous CN of ~6 (Ohtaki and Radnai, 1993), indicating that
the presence of the DOM molecule creates stereochemical barriers to coordination even
for a divalent cation with a relatively high energy of hydration. An average Ca2+is
coordinated to at least one carboxylate O, demonstrating the presence of a significant
degree of inner sphere complexation with organic functional groups. Conductimetric
titration of a HA sample with Ca(OH)2revealed a large region in the titration curve in
which additions of Ca2+produced almost no increase in conductivity, evidence for strong
association between the divalent cation and the organic molecules (van den Hoop et al.,
1990), as seen in these simulations. The same titration performed with NaOH indicated
minimal binding of Na+to the HA (van den Hoop et al., 1990). Analysis of the small
competition effect that Ca2+has on Cu2+binding to humic substances suggests that the
former is complexed primarily to carboxylate sites, while the latter is complexed
primarily to phenolic sites (Lu and Allen, 2002). Analysis of Ca2+ and Cu2+adsorption
isotherms with the bimodal non-ideal competitive adsorption (NIC A) model supports this
conclusion (Benedetti et al., 1995), providing more evidence for the strong complexation
interaction of Ca2+ with carboxylate functional groups. Infrared spectroscopic studies,
discussed in Section 4.3.5 below, provide additional support for complexation of Ca2+by
carboxylate groups within humic substances (Griffith and Schnitzer, 1975; Griffith et al.,
1984; Alberts and Filip, 1998). Further coordination of cations in the Ca-DOM molecule
is made up of a relatively large water contribution relative to Na-DOM, due to the larger
271
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
energy of hydration of the divalent cation, and a small contribution from organic, non-
carboxylate O functional groups, similar to that seen in the Na-DOM molecule.
Examination of the local environment of each cation individually reveals a more
complex view of cation-carboxylate interactions. An ion may be defined as singly inner
sphere coordinated when it is within the maximum r-value for inner sphere coordination
for only one of the carboxylate O, and may be defined as doubly inner sphere coordinated
when it is within this r-value for both carboxylate O (Figure 4.9). An ion may be inner
sphere coordinated to more carboxylate groups than are needed to balance its charge. A
carboxylate group may participate in two single inner sphere coordination interactions,
one for each O. An ion is considered outer sphere coordinated when the Os of the nearest
carboxylate group are outside the inner sphere r-values. The Na-DOM complex contains
9 Na+that are doubly inner sphere coordinated, and 34 that are singly inner sphere
coordinated. Two of the former and 8 of the latter are sufficiently close to an O of
another carboxylate group as to be considered inner sphere coordinated to it as well.
Neighboring carboxylate groups attached to benzene rings are capable of capturing Na+
via separate inner sphere coordination interactions, perhaps contributing to the preference
seen in this model system for benzene carboxylates to form inner sphere associations
with cations.
The 34 remaining Na+ are in outer sphere positions relative to carboxylate groups.
Two of these cations are over 6 A away from the nearest carboxylate, likely indicating
transition from a direct, outer sphere association with the organic functional group to a
more independent existence within the shell of water surrounding the deprotonated DOM
molecule. This movement is evidence of the weak interaction of Na+ with carboxylate
272
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
Figure 4.9. A portion of the Na-DOM structure featuring an outer sphere Na-carboxylate
complex (left), a doubly inner sphere Na-carboxylate complex (top), and a singly inner
sphere Na-carboxylate complex, which also happens to possess inner sphere coordination
with an alcohol group (right).
groups, as well as its high mobility in the hydrated organic system. The movement of
42 Na+, over half of the cations present, away from the carboxylate groups to which they
were originally paired, is further evidence for a highly mobile, weakly bound cation.
Although the titration procedure used to model the Na-DOM system may favor the
formation of inner sphere coordinated complexes, it is evident that Na+ is not strongly
bound to the organic polyanion, consistent with experimental evidence (van den Hoop et
al., 1990; Bonn and Fish, 1993).
Evaluation of the local cation environments of the Ca-DOM complex reveals a
different story. Because each Ca2+is considered to be associated with at least two
carboxylate groups given its divalent charge, a total of at least 76 Ca-carboxylate
273
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
interactions must be evaluated, two for each cation present. In fact, the system contains
79 Ca-carboxylate interactions, as there are 3 cations that are doubly inner sphere
coordinated to one carboxylate, and singly inner sphere coordinated to two others. The
Ca-DOM complex features 15 doubly inner sphere complexes, 22 singly inner sphere
complexes, and 42 outer sphere associations. Aromatic carboxylate groups show a
preference for forming doubly inner sphere complexes, as seen in the Na-DOM complex.
A total of just 10 of the 38 Ca2+are involved in outer sphere coordination with both
nearby carboxylate groups, and none of these display drift into the water shell as seen in
the Na-DOM complex. Exchange of the original carboxylate group coordinated to a
specific Ca2+for another has occurred for only 22 of the 76 initial associations created
during titration. In comparison to Na+, Ca2+is more likely to be doubly inner sphere
coordinated with carboxylate groups, and to display fewer exchange reactions and no
movement into the water shell, characteristics that support the idea of a tightly bound
cation. At the same time, the divalent cations occupy more outer sphere coordination
positions near carboxylate groups than the monovalent cations, due to the stronger energy
of hydration of the former.
Deprotonation and subsequent saturation of the DOM molecule with hydrated Na+
and Ca2+results in formation of porous organic structures, consistent with AFM images
of humic substances experiencing near-neutral pH conditions (Maurice and Namjesnik-
Dejanovic, 1999). The expanded porosities, solvent accessible surface areas, and
polarities of the molecules provided conditions amenable to the formation of numerous
H-bonds with associated water molecules. Differences in the number of H-bonds present
in the Na-DOM and Ca-DOM complexes indicate that the identity of the counterions
274
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
associated with humic substances may affect their ability to sorb and retain xenobiotic
organic molecules via these bonds, as suggested by recent experimental evidence (Yuan
and Xing, 2001). The two cations display contrasting coordination behavior and
mobility. The low charge of Na+, and the associated low energy of hydration, provides
little incentive for formation of inner sphere interactions with carboxylate or water O, and
results in a relatively mobile cation. Initial placement of the cations near carboxylate
groups during the titration procedure may lead to a bias favoring organic complexation,
but the high mobility of Na+suggests it is weakly bound at best. The higher charge of
Ca2+ allows the cation to take part in strong complexation interactions with carboxylate,
which reduces its mobility, as seen in experimental work (van den Hoop et al., 1990;
Benedetti et al., 1995; Lu and Allen, 2002). The divalent cation also has a higher
hydration energy, and therefore maintains a more substantial hydration shell around
itself, despite steric hindrance provided by the presence of the organic molecule. Model
IR spectra of the two systems will be discussed in Section 4.3.5 below.
4.3.5 Model Infrared Spectra o f Protonated and Na- and Ca-saturated DOM Hydrates
Figure 4.10 features model IR spectra for the DOM molecule in dry and hydrated
states, and in deprotonated and Na- and Ca-saturated states. Water molecules were
removed lfom the last three systems prior to calculation of the spectra, because the
Cerius2 IR-Raman algorithm cannot operate on systems containing as many atoms and
molecules as were present when the organic molecules were hydrated. As a result, the
DOM hydrate and Na- and Ca-DOM complexes used to create model spectra cannot be
considered to occupy highly minimized states, as the final EM computations were made
275
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
on the fully hydrated systems. This may lead to inaccuracies in the model IR spectra
produced from these configurations (Accelrys Inc., 2001). Such accuracies were deemed
preferable to the alternative, performing EM on the dehydrated systems and then
calculating model IR spectra, as the resulting organic systems would not reflect the
hydrated conditions of interest here.
DOM alone
DOM hydrate (dry)
Na-DOM (dry)
0.9
0.7
0.0
0 2000 1500 1000 500 3500 3000 2500 4000
wavenumber (cm )
DOM alone
DOM hydrate (dry)
Ca-DOM (dry)
0.7
3 0.6
CD
i 05
-e o.4
o
0.3
CQ
2000 1500 1000 500 0 3500 3000 2500 4000
wavenumber (cm1)
Figure 4.10. Model IR spectra for the dry and hydrated DOM molecules, as well as the
Na-DOM (top) and Ca-DOM (bottom) systems, normalized using the intensity of the
nitrile peak.
276
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
The spectra in Figure 4.10 are normalized by the peak intensity of the nitrile
functional group near 2200 cm'1. Experimental studies of simple nitrile compounds
indicate that the frequency of the nitrile vibration does not vary substantially with the
addition of a polar solvent like water (Figeys and Mahieu, 1968; Bernstein et al., 1997),
but the intensity can, perhaps due to dipole interactions (Caldow et al., 1960; Figeys and
Mahieu, 1968) or to the formation of H-bonds (Ritchie et al., 1962; Bernstein et al.,
1997). An examination of the local environment of the nitrile group in the three hydrated
systems indicates that only in the Ca-DOM system are water molecules close enough to
exert influence over this stretching frequency, possibly causing changes to relative
intensity that would invalidate normalization. However, the normalized Ca-DOM IR
spectrum is so similar to that of the Na-DOM complex that it appears unlikely that the
presence of water molecules has affected the intensity of the nitrile vibration.
The frequency of the nitrile functional group shows variation among the four
systems. Dry and hydrated DOM molecules possess peaks at 2193 and 2185 cm'1,
respectively, while Na- and Ca-DOM complexes feature peaks at 2160 and 2120 cm'1,
respectively. Experimental IR studies indicate that the presence of cations can produce a
shift in the location of the nitrile peak to higher frequencies for simple nitrile compounds
(Fawcett and Liu, 1992). That the shift observed here is to lower frequencies may
indicate a shortcoming of the algorithm used to create the model spectra, or may be an
example of the inaccuracies possible when using a poorly minimized structure for
calculation of model IR spectra.
The spectrum of the DOM hydrate is quite different from that of its dry, densely
packed predecessor. The O-H stretching modes, originally located above 3000 cm'1,
277
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
have taken on a diffuse distribution extending to just below 2700 cm'1. Greater variation
in the length and strength of organic O-H bonds, caused by H-bonding interactions with
surrounding water molecules, could explain this change. Experimental spectra of
purified humic substances often contain a broad region of IR absorbance extending to
2500 cm'1, and attributed to H-bonded carboxylic acid groups (Griffith and Schnitzer,
1975). The clusters of peaks located in the ranges 1500 - 1000 and 800 - 500 cm'1in the
dry DOM molecule have shifted to higher positions in the DOM hydrate. Changes in
molecular vibrations due to H-bonding may account for this shift, as may the poorly
equilibrated state of the organic structure.
The model spectra of the Na- and Ca-DOM complexes feature substantial
reduction in peaks representing organic O-H stretching near 3500 cm'1, C=0 stretching
from 1800 to 1600 cm'1, C-0 stretching near 1400 cm'1, and O-H in-plane bending in the
vicinity of 1300 cm'1, due to the absence or near absence of carboxyl groups. Instead,
carboxylate symmetric stretch peaks occur near 1600 cm'1, with associated asymmetric
stretch peaks near 1800 cm'1. The disappearance of carboxyl peaks and the appearance
of carboxylate peaks seen in these model spectra effectively mimics the changes noted in
experimental spectra when humic substances are examined under near neutral pH
conditions (Davis et al., 1999). The Na- and Ca-DOM complexes also exhibit the wide
and moderately intense cluster of O-H stretching peaks seen in the DOM hydrate. This
absorbance pattern cannot be associated with H-bonded carboxyl groups for these
deprotonated molecules, as suggested by experiment (Schnitzer and Griffith, 1975) and
instead may be caused by hydration-induced bond variation, or an organic structure
featuring poor minimization. Transformation of carboxyl groups into carboxylate, and
278
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
subsequent hydration, has resulted in spectra with few strong features and broad peak
distributions.
It was possible to calculate a model IR spectrum for a partially hydrated Ca-DOM
system, containing the 228 water molecules originally associated with each Ca2+ inserted
via titration. This spectrum is displayed in Figure 4.11, along with the spectrum of the
dehydrated Ca-DOM complex. Normalization could not be performed for the partially
hydrated Ca-DOM spectrum, because a distinct nitrile peak could not be identified.
However, the presence of water molecules may make this system and its associated
spectrum more appropriate for comparison with experimental work.
1.0
0.9
0.8
c/5
0.7

0 6
<D
s
0.5
e
o
0.4
s
0.3
0.2 -
0.1
0.0
_Ca-DOM (dry)
Ca-DOM
(some water present)
4000 3500 3000 2500 2000 1500
wavenumber (cm )
1000 500
Figure 4.11. Model IR spectra for the Ca-DOM complex in dry and partially hydrated
states (no normalization).
Several experimental IR spectra of humic substances complexed with Ca2+and
then dehydrated consistently feature a reduction in the carboxyl peak, usually located
near 1700 cm'1, and a simultaneous increase in the carboxylate peak, often observed near
1650 c m1, regardless of pH (Griffith and Schnitzer, 1975; Griffith et al., 1984; Alberts
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
and Filip, 1998). This pattern indicates that the cation interacts with carboxylate groups.
A unique set of IR subtraction spectra of a HA in the aqueous state, not in the dry form
2"b
used in the studies above, indicates little complexation when low concentrations of Ca
are available in solution (Bonn, 1992). With higher concentrations of the cation, a
physical change occurs to the humic substance, as evidenced by decreased intensity of the
IR spectra and increased viscosity of the humic solution. A conformational change
attributed to cross-linking of humic fragments through the complexation of Ca2+by
multiple carboxylate groups may be responsible for the altered state. The assumption that
Ca2+ in solution will be attracted to the carboxylate groups present in humic substances
was used in creating the cation-saturated complexes modeled here. The model IR
spectrum of the partially hydrated Ca-DOM complex does not display the behavior
predicted from the experimental IR spectra of Ca-saturated, dehydrated humic materials
with respect to carboxyl and carboxylate signals, though it is observed in the dehydrated
Ca-DOM complex (Figure 4.10). Instead, the cluster of peaks that include carboxyl and
carboxylate signals appears to have shifted as a group to higher frequencies.
Further experimental evidence favoring Ca2+ complexation by carboxylate groups
may be taken from observation of the absence of IR absorbance from 2700 to
2500 cm'1, indicating reduced presence of carboxyl groups capable of forming
intermolecular H-bonds (Schnitzer and Griffith, 1975). The model IR spectrum of the
partially hydrated Ca-DOM complex features an even broader distribution of O-H
stretching signals in this region, contrary to predictions based on experiment. Evidently,
the partially hydrated cation-organic complex does not produce a realistic model IR
280
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
* * 2
spectrum, perhaps a result of poor minimization or reduced effectiveness of the Cerius
IR algorithm on multi-body systems.
4.4 Conclusions
The DOM molecule designed by Schulten (1999a) displays properties consistent
with experimental data when modeled using the COMPASS force field. This model
organic compound is much larger than the average humic molecule. However, it does
feature realistic elemental ratios, as well as distributions of both nonpolar and reactive
functional groups largely consistent with experimental data. Protein, carbohydrate, lipid,
and lignin-like fragments incorporated into the structure help to capture the complexity of
humic substances at a scale larger than that of component functional groups. Despite the
considerable effort employed to devise this potentially powerful humic molecule,
previous simulations employing it have relied on modeling methods entirely inadequate
to the task. The work presented here is the first rigorous computational exploration of
this class of model humic molecules.
The DOM molecule reproduces a value for dry bulk density within the range of
those determined for humic substances, and a value for the Hildebrand solubility
parameter quite close to those estimated from experimental measurements. A model IR
spectrum created from the fully minimized dry DOM configuration is structurally
comparable to experimental spectra. After calculations to assure that the COMPASS
bulk water system reproduces adequately the structure of liquid water, despite smaller H-
O-H bond angles within individual water molecules, explorations of a hydrated DOM
system could begin. These indicated the organic molecule possesses the flexibility
281
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
required to reorient itself in the presence of a polar solvent, concentrating polar functional
groups toward outer regions in contact with water.
In order to examine the structure of humic substances under pH conditions closer
to neutral, a novel method of stepwise titration was developed, in which hydrated cations
were placed near deprotonated carboxyl functional groups. Both Na- and Ca-DOM
structures created in this manner were highly porous and featured many intermolecular
H-bonds. Ca2+displayed stronger organic complexation and reduced mobility relative to
Na+, as well as a greater level of hydration by water molecules, all a result of its greater
charge. The model IR spectra created from the DOM hydrate and the deprotonated
complexes display trends that include broadened distributions of O-H stretching
vibrations and the appearance of carboxylate signals. However, it is difficult to
determine whether these effects were a result of relevant changes to bond vibrations, or
merely indications of a poorly minimized structure due to complete or partial removal of
water molecules. The spectrum of the partially hydrated Ca-DOM complex was so
poorly structured as to indicate that the Cerius2IR-Raman algorithm may not produce
realistic results for complex multi-body systems.
The DOM molecule, as simulated by the COMPASS force field, must be
considered a valuable tool for investigating the structure and function of aqueous humic
molecules under acidic and near neutral pH conditions. Given the ability of COMPASS
to model both organic and inorganic species, the next logical step in this series of
modeling exercises is an investigation of interactions between DOM and a soil mineral
using similar techniques.
282
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.5 References
Accelrys Inc. (2001) Cerius2 Release 4.6. Accelrys Inc.
Akim L. G., Bailey G. W., and Shevchenko S. M. (1998) A computation chemistry
approach to study the interactions of humic substances with mineral surfaces. In Humic
Substances: Structures, Properties, Uses (ed. G. Davies andE. A. Ghabbour), pp. 133-
145. Royal Society of Chemistry.
Alberts J. J. and Filip Z. (1998) Metal binding in estuarine humic and fulvic acids: FTIR
analysis of humic acid-metal complexes. Environ. Technol. 19, 923-931.
Allen M. P. and Tildesley D. J. (1987) Computer Simulation o f Liquids. Oxford
University Press.
Allinger N. L. (1977) Conformational analysis. 130. MM2. A hydrocarbon force field
utilizing V) and V2 torsional terms. J. Am. Chem. Soc. 99, 8127-8134.
Almendros G., Guadalix M. E., Gonzalez-Vila F. J., and Martin F. (1996) Preservation of
aliphatic macromolecules in soil humins. Org. Geochem. 24, 651-659.
Almendros G., Guadalix M. E., Gonzalez-Vila F. J., and Martin F. (1998) Distribution of
structural units in humic substances as revealed by multi-step selective degradations and
13C-NMRof successive residues. Soil Biol. Biochem. 30, 755-765.
American Chemical Society. (2001) SciFinder Scholar, database accessed April 30, 2003.
Bailey G. W., Akim L. G., and Shevchenko S. M. (2001) Predicting chemical reactivity
of humic substances for minerals and xenobiotics. Use of computational chemistry,
scanning probe microscopy and virtual reality. In Humic Substances and Chemical
Contaminants (ed. C. E. Clapp, M. H. B. Hayes, N. Senesi, P. R. Bloom, and P. M.
Jardine), pp. 41-72. Soil Science Society of America.
Bellamy L. J. (1975) The Infra-Red Spectra o f Complex Molecules. John Wiley & Sons.
Benedetti M. F., Milne C. J., Kinniburgh D. G., van Riemsdijk W. H., and Koopal L. K.
(1995) Metal ion binding to humic substances: Application of the non-ideal competitive
adsorption model. Environ. Sci. Technol. 29, 446-457.
Berendsen H. J. C., Grigera J. R., and Straatsma T. P. (1987) The missing term in
effective pair potentials. J. Phys. Chem. 91, 6269-6271.
Berendsen H. J. C., Postma J. P. M., van Gunsteren W. F., DiNola A., and Haak J. R.
(1984) Molecular dynamics with coupling to an external bath. J. Chem. Phys. 81, 3684-
3690.
Bernstein M. P., Sandford S. A., and Allamandola L. J. (1997) The infrared spectra of
nitriles and related compounds frozen in Ar and H2O. Astrophys. J. 476, 932-942.
283
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Beveridge D. L., Mezei M., Mehrotra P. K., Marchese F. T., Ravi-Shanker G., Vasu T.,
and Swaminathan S. (1983) Monte Carlo computer simulation studies of the equilibrium
properties and structure of liquid water. In Molecular-Based Study o f Fluids (ed. J. M.
Haile and G. A. Mansoori), pp. 297-351. American Chemical Society.
Bleam W. F. (1991) Soil science applications of nuclear magnetic resonance
spectroscopy. Adv. Agron. 46, 91-155.
Bonn B. A. (1992) Interactions of a soil humic acid with alkali metal cations and alkaline
earth metal cations. Ph.D. thesis, Lewis and Clark College.
Bonn B. A. and Fish W. (1993) Measurement of electrostatic and site-specific
associations of alkali metal cations with humic acid. J. Soil Sci. 44, 335-345.
Brown T. L., Novotny F. J., and Rice J. A. (1998) Comparison of desorption mass
spectrometry techniques for the characterization of fulvic acid. In Humic Substances:
Structures, Properties and Uses (ed. G. Davies and E. A. Ghabbour), pp. 91-107. Royal
Society of Chemistry.
Brown T. L. and Rice J. A. (2000a) Effects of experimental parameters on the ESI FT-
ICR mass spectrum of fulvic acid. Anal. Chem. 72, 384-390.
Brown T. L. and Rice J. A. (2000b) The effect of laser wavelength and power density on
the laser desorption mass spectrum of fulvic acid. Org. Geochem. 31, 627-634.
Bruccoleri A. G., Sorenson B. T., and Langford C. H. (2001) Molecular modeling of
humic structures. In Humic Substances: Structures, Models and Functions (ed. E. A.
Ghabbour and G. Davies), pp. 193-208. Royal Society of Chemistry.
Bubert H., Lambert J., and Burba P. (2000) Structural and elemental investigations of
isolated aquatic humic substances using X-ray photoelectron spectroscopy. Fresenius J.
Anal. Chem. 368, 274-280.
Bunte S. W. and Sun H. (2000) Molecular modeling of energetic materials: The
parameterization and validation of nitrate esters in the COMPASS force field. J. Phys.
Chem. B 104, 2477-2489.
Burdon J. (2001) Are the traditional concepts of the structures of humic substances
realistic? Soil Sci. 166, 752-769.
Caldow G. L., Cunliffe-Jones D., and Thompson H. W. (1960) Intermolecular forces and
solvent effects. II. Band intensities. Proc. Royal Soc. London. Ser. A, Math. Phys. Sci.
254, 17-29.
Cameron R. S., Thornton B. K., Swift R. S., and Posner A. M. (1972) Molecular weight
and shape of humic acid from sedimentation and diffusion measurements on fractionated
extracts. J. Soil Sci. 23, 394-408.
284
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cech N. B. and Enke C. G. (2001) Practical implications of some recent studies in
electrospray ionization fundamentals. Mass Spec. Rev. 20, 362-387.
Celi L., Schnitzer M., and Negre M. (1997) Analysis of carboxyl groups in soil humic
acids by a wet chemical method, Fourier-transform infrared spectrophotometry, and
solution-state carbon-13 nuclear magnetic resonance. A comparative study. Soil Sci. 162,
189-197.
Chefetz B., Tarchitzky J., Deshmukh A. P., Hatcher P. G., and Chen Y. (2002) Structural
characterization of soil organic matter and humic acids in particle-size fractions of an
agricultural soil. Soil Sci. Soc. Am. J. 66, 129-141.
Chiavari G. and Galletti G. C. (1992) Pyrolysis-gas chromatography/mass spectrometry
of amino acids. J. Anal. Appl. Pyrolysis 24,123-137.
Chien Y.-Y. and Bleam W. F. (1997) Fluorine-19 nuclear magnetic resonance study of
atrazine in humic and sodium dodecyl sulfate micelles swollen by polar and nonpolar
solvents. Langmuir 13, 5283-5288.
Christl I., Knicker H., Kogel-Knabner I., and Kretzschmar R. (2000) Chemical
heterogeneity of humic substances: Characterization of size fractions obtained by hollow-
fibre ultrafiltration. Euro. J. Soil Sci. 51, 617-625.
Clark M., Cramer R. D., and van Opdenbosch N. (1989) Validation of the general-
purpose Tripos 5.2 force field. J. Comput. Chem. 10, 982-1012.
Connolly M. L. (1983) Solvent-accessible surfaces of proteins and nucleic acids. Science
221,709-713.
Conte P., Piccolo A., van Lagen B., Buurman P., and de Jager P. A. (1997) Quantitative
differences in evaluating soil humic substances by liquid- and solid-state 13C-NMR
spectroscopy. Geoderma 80, 339-352.
Cook R. L. and Langford C. H. (1998) Structural characterization of a fulvic acid and a
humic acid using solid-state Ramp-CP-MAS 13C nuclear magnetic resonance. Environ.
Sci. Technol. 32, 719-725.
Cornell W. D., Cieplak P., Bayly C. I., Gould I. R., Merz K. M., Ferguson D. M.,
Spellmeyer D. C., Fox T., Caldwell J. W., and Kollman P. A. (1995) A second generation
force field for the simulation of proteins, nucleic acids, and organic molecules. J. Am.
Chem. Soc. 117, 5179-5197.
Dai X. Y., Ping C. L., Candler R., Haumaier L., and Zech W. (2001) Characterization of
soil organic matter fractions of tundra soils in arctic Alaska by carbon-13 nuclear
magnetic resonance spectroscopy. Soil Sci. Soc. Am. J. 65, 87-93.
Davies G., Fataftah A., Cherkasskiy A., Ghabbour E. A., Radwan A., Jansen S. A., Kolia
S., Paciolla M. D., Sein L. T., Buermann W., Balasubramanian M., Budnick J., and Xing
285
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
B. (1997) Tight metal binding by humic acids and its role in biomineralization. J. Chem.
Soc., Dalton Trans. 1997, 4047-4060.
Davis W. M., Erickson C. L., Johnston C. T., Delfino J. J., and Porter J. E. (1999)
Quantitative Fourier transform infrared spectroscopic investigation of humic substance
functional group composition. Chemosphere 38, 2913-2928.
Del Rio J. C., McKinney D. E., Knicker H., Nanny M. A., Minard R. D., and Hatcher P.
G. (1998) Structural characterization of bio- and geo-macromolecules by off-line
thermochemolysis with tetramethylammonium hydroxide. J. Chrom. A 823, 433-448.
Diallo M. S., Faulon J.-L., Goddard W. A., and Johnson J. H. (2001) Binding of
hydrophobic organic compounds to dissolved humic substances: A predictive approach
based on computer assisted structure elucidation, atomistic simulations and Flory-
Huggins solution theory. In Humic Substances: Structures, Models and Functions (ed. E.
A. Ghabbour and G. Davies), pp. 221-237. Royal Society of Chemistry.
Diallo M. S., Simpson A., Gassman P., Faulon J. L., Johnson J. H., Goddard W. A., and
Hatcher P. G. (2003) 3-D structural modeling of humic acids through experimental
characterization, computer assisted structure elucidation and atomistic simulations. 1.
Chelsea soil humic acid. Environ. Sci. Technol. 37,1783-1793.
Diem M. (1993) Introduction to Modern Vibrational Spectroscopy. John Wiley & Sons.
Engebretson R. R. and von Wandruszka R. (1997) The effects of molecular size on humic
acid associations. Org. Geochem. 26, 759-767.
Engebretson R. R. and von Wandruszka R. (1998) Kinetic aspects of cation-enhanced
aggregation in aqueous humic acids. Environ. Sci. Technol. 32, 488-493.
Fabbri D., Chiavari G., and Galletti G. C. (1996) Characterization of soil humin by
pyrolysis(/methylation)-gas chromatography/mass spectrometry: Structural relationships
with humic acids. J. Anal. Appl. Pyrolysis 37, 161-172.
Fabbri D., Mongardi M., Montanari L., Galletti G. C., Chiavari G., and Scotti R. (1998)
Comparison between CP/MAS 13C-NMR and pyrolysis-GC/MS in the structural
characterization of humins and humic acids of soil and sediments. Fresenius J. Anal.
Chem. 362, 299-306.
Fan T. W.-M., Higashi R. M., and Lane A. N. (2000) Chemical characterization of a
chelator-treated soil humate by solution-state multinuclear two-dimensional NMR with
FTIR and pyrolysis-GCMS. Environ. Sci. Technol. 34, 1636-1646.
Fawcett W. R. and Liu G. (1992) A study of ion pairing in acetonitrile solutions
containing magnesium perchlorate using attenuated total reflection FTIR spectroscopy. J.
Phys. Chem. 96, 4231-4236.
286
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ferrario M. and Tani A. (1985) A molecular dynamics study of the TIP4P model of
water. Chem. Phys. Letters 121, 182-186.
Ferreira J., Nascimiento O. R., and Martin-Neto L. (2001) Hydrophobic interactions
between spin-label 5-SASL and humic acid as revealed by ESR spectroscopy. Environ.
Sci. Technol. 35, 761-765.
Figeys H. P. and Mahieu V. (1968) Influence of molecular structure on the frequency and
integrated intensity of the vC=-N vibration of a series of polycyclic aromatic nitriles - II
Solvent effects on band intensities. Spectrochim. Acta 24A, 1553-1560.
Fried J. R. and Li B. (2001) Atomistic simulation of the glass transition of di-substituted
polysilanes. Comp. Theor. Polym. Sci. 11, 273-281.
Gasteiger J. and Marsili M. (1980) Iterative partial equalization of orbital
electronegativity - a rapid access to atomic charges. Tetrahedron 36, 3219-3228.
Gonzalez-Vila F. J., Almendros G., Tinoco P., and Rodriguez J. (2001b) Nitrogen
speciation and pyrolytic patterns of 15N-labelled soil and compost fractions. J. Anal.
Appl. Pyrolysis 58-59, 329-339.
Gonzalez-Vila F. J., Lankes U., and Liidemann H.-D. (2001a) Comparison of the
information gained by pyrolytic techniques and NMR spectroscopy on the structural
features of aquatic humic substances. J. Anal. Appl. Pyrolysis 58-59, 349-359.
Grasset L. and Ambles A. (1998) Structure of humin and humic acid from an acid soil as
revealed by phase transfer catalyzed hydrolysis. Org. Geochem. 29, 881-891.
Grasset L., Guignard C., and Ambles A. (2002) Free and esterified aliphatic carboxylic
acids in humin and humic acids from a peat sample as revealed by pyrolysis with
tetramethylammonium hydroxide or tetraethylammonium acetate. Org. Geochem. 33,
181-188.
Griffith S. M., Holder M. B., and Munro S. (1984) Chemistry of organic matter colloids
in Andepts and Vertisols of the Caribbean. Trop. Agric. 61, 213-220.
Griffith S. M. and Schnitzer M. (1975) The isolation and characterization of stable metal-
organic complexes from tropical volcanic soils. Soil Sci. 120, 126-131.
Grote M., Klinnert S., and Bechmann W. (2000) Comparison of degradation state and
stability of different humic acids by means of chemolysis with tetramethylammonium
hydroxide. J. Environ. Monit. 2, 165-169.
Gunasekara A. S., Dickinson L. C., and Xing B. (2001) Solid state NMR evidence for
multiple domains in humic substances. In Humic Substances: Structures, Models and
Functions (ed. E. A. Ghabbour and G. Davies), pp. 63-71. Royal Society of Chemistry.
287
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Gunasekara A. S., Simpson M. J., and Xing B. (2003) Identification and characterization
of sorption domains in soil organic matter using structurally modified humic acids.
Environ. Sci. Technol. 37, 852-858.
Haiber S., Herzog H., Burba P., Gosciniak B., and Lambert J. (2001) Two-dimensional
NMR studies of size fractionated Suwannee River fulvic and humic acid reference.
Environ. Sci. Technol. 35, 4289-4294.
Halgren T. A. (1992) Representation of van der Waals (vdW) interactions in molecular
mechanics force fields: Potential form, combination rules, and vdW parameters. J. Am.
Chem. Soc. 114, 7827-7843.
Hatcher P. G., Dria K. J., Kim S., and Frazier S. F. (2001) Modem analytical studies of
humic substances. Soil Sci. 166, 770-794.
Hatcher P. G., Maciel G. E., and Dennis L. W. (1981) Aliphatic structure of humic acid; a
clue to their origin. Org. Geochem. 3, 43-48.
Hertkorn N., Permin A., Perminova I., Kovalevskii D., Yudov M., Petrosyan V., and
Kettrup A. (2002) Comparative analysis of partial structures of a peat humic and fulvic
acid using one- and two-dimensional nuclear magnetic resonance spectroscopy. J.
Environ. Qual. 31, 375-387.
Heyes D. M. (1994) Physical properties of liquid water by molecular dynamics
simulations. J. Chem. Soc., Faraday Trans. 90, 3039-3049.
Hu W.-G., Mao J., Xing B., and Schmidt-Rohr K. (2000) Poly(methylene) crystallites in
humic substances detected by nuclear magnetic resonance. Environ. Sci. Technol. 34,
530-534.
Hundal L. S., Carmo A. M., Bleam W. F., and Thompson M. L. (2000) Sulfur in
biosolids-derived fulvic acid: Characterization by XANES spectroscopy and selective
dissolution approaches. Environ. Sci. Technol. 34, 5184-5188.
Hutchison K. J., Hesterberg D., and Chou J. W. (2001) Stability of reduced organic sulfur
in humic acid as affected by aeration and pH. Soil Sci. Soc. Am. J. 65, 704-709.
Jandl G., Schulten H.-R., and Leinweber P. (2002) Quantification of long-chain fatty
acids in dissolved organic matter and soils. J. Plant Nutr. Soil Sci. 165, 133-139.
Jansen S. A., Malaty M., Nwabara S., Johnson E., Ghabbour E., Davies G., and Vamum
J. M. (1996) Structural modeling in humic acids. Materials Sci. Eng. C 4, 175-179.
Jorgensen W. L., Chandrasekhar J., Madura J. D., Impey R. W., and Klein M. L. (1983)
Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 70,
926-935.
288
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Kacker T., Haupt E. T. K., Garms C., Francke W., and Steinhart H. (2002) Structural
13
characterisation of humic acid-bound PAH residues in soil by C-CPMAS-NMR-
spectroscopy: Evidence of covalent bonds. Chemosphere 48,117-131.
Kerner M., Hohenberg H., Ertl S., Reckermann M., and Spitzy A. (2003) Self
organization of dissolved organic matter to micelle-like microparticles in river water.
Nature 422, 150-154.
Khalaf M., Kohl S. D., Klumpp E., Rice J. A., and Tombacz E. (2003) Comparison of
sorption domains in molecular weight fractions of a soil humic acid using solid-state 19F
NMR. Environ. Sci. Technol. 37, 2855-2860.
Kingery W. L., Simpson A. J., Hayes M. H. B., Locke M. A., and Hicks R. P. (2000) The
application of multidimensional NMR to the study of soil humic substances. Soil Sci.
165,483-494.
Klapper L., McKnight D. M., Fulton J. R., Blunt-Harris E. L., Nevin K. P., Lovley D. R.,
and Hatcher P. G. (2002) Fulvic acid oxidation state detection using fluorescence
spectroscopy. Environ. Sci. Technol. 36, 3170-3175.
Knicker H. (2000) Biogenic nitrogen in soils as revealed by solid-state carbon-13 and
nitrogen-15 nuclear magnetic resonance spectroscopy. J. Environ. Qual. 29, 715-723.
Knicker H. (2002) The feasibility of using DCPMAS 15N 13C NMR spectroscopy for a
better characterization of immobilized 15N during incubation of 13C and 15N-enriched
plant material. Org. Geochem. 33, 237-246.
Knicker H., Almendros G., Gonzalez-Vila F. J., Ludemann H.-D., and Martin F. (1995)
13C and 15N NMR analysis of some fungal melanins in comparison with soil organic
matter. Org. Geochem. 23, 1023-1028.
Knicker H., del Rio J. C., Hatcher P. G., and Minard R. D. (2001) Identification of
protein remnants in insoluble geopolymers using TMAH thermochemolysis/GC-MS.
Org. Geochem. 32, 397-409.
Knicker H. and Ludemann H.-D. (1995) N-15 and C-13 CPMAS and solution NMR
studies of N-15 enriched plant material during 600 days of microbial degradation. Org.
Geochem. 23, 329-341.
Knicker H., Scaroni A. W., and Hatcher P. G. (1996) 13C and 15N NMR spectroscopic
investigation on the formation of fossil algal residues. Org. Geochem. 24, 661-669.
Knicker H., Schmidt M. W. I., and Kogel-Knabner I. (1999) The structure of organic
nitrogen in particle size fractions determined by 15N CPMAS NMR. In Effect o f Mineral-
Organic-Microorganism Interactions on Soil and Freshwater Environments (ed. J.
Berthelin, P. M. Huang, F. Bollag, and F. Andreux), pp. 143-149. Kluwer
Academic/Plenum.
289
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Kopinke F.-D., Porschmann J., and Stottmeister U. (1995) Sorption of organic pollutants
on anthropogenic humic matter. Environ. Sci. Technol. 29, 941-950.
Kramer R. W., Kujawinski E. B., Zang X., Green-Church K. B., Jones R. B., Freitas M.
A., and Hatcher P. G. (2001) Studies of the structure of humic substances by electrospray
ionization coupled to a quadrupole-time of flight (QQ-TOF) mass spectrometer. In Humic
Substances: Structures, Models and Functions (ed. E. A. Ghabbour and G. Davies), pp.
95-107. Royal Society of Chemistry.
Kubicki J. D. (2000) Molecular mechanics and quantum mechanical modeling of hexane
soot structure and interactions with pyrene. Geochem. Trans. 7, 41-46.
Kubicki J. D. and Apitz S. E. (1999) Models of natural organic matter and interactions
with organic contaminants. Org. Geochem. 30, 911-927.
Kubicki J. D. and Trout C. C. (2003) Molecular modeling of fulvic and humic acids:
Charging effects and interactions with Al3+, benzene, and pyridine. In Geochemical and
Hydrological Reactivity o f Heavy Metals in Soils (ed. H. M. Selim and W. L. Kingery),
pp. 113-143. Lewis Publishers.
Kujawinski E. B. (2002) Electrospray ionization Fourier transform ion cyclotron
resonance mass spectrometry (ESI FT-ICR MS): Characterization of complex
environmental mixtures. Environ. Forens. 3, 207-216.
Launne T., Neelov I., and Sundholm F. (2001) Molecular dynamics simulations of
polymers of unsubstituted and substituted poly(p-phenylene terephthalate)s in the bulk
state. Macromolec. Theory Simul., 10,137-143.
Leenheer J. A. (1994) Chemistry of dissolved organic matter in rivers, lakes, and
reservoirs. In Environmental Chemistry o f Lakes and Reservoirs (ed. L. A. Baker),
pp. 195-221. American Chemical Society.
Leenheer J. A., McKnight D. M., Thurman E. M., and MacCarthy P. (1989) Structural
components and proposed structural models of fulvic acid from the Suwannee River. In
Humic Substances in the Suwannee River, Georgia: Interactions, Properties, and
Proposed Structures, Vol. Open-File Report 87-557 (ed. R. C. Averett, J. A. Leenheer, D.
M. McKnight, and K. A. Thom), pp. 335-359. United States Geological Survey.
Leenheer J. A., Rostad C. E., Gates P. M., Furlong E. T., and Ferrer I. (2001) Molecular
resolution and fragmentation of fulvic acid by electrospray ionization/multistage tandem
mass spectrometry. Anal. Chem. 73, 1461-1471.
Leenheer J. A., Wershaw R. L., Brown G. K., and Reddy M. M. (2003) Characterization
and diagenesis of strong-acid carboxyl groups in humic substances. Appl. Geochem. 18,
471-482.
290
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Leenheer J. A., Wershaw R. L., and Reddy M. M. (1995a) Strong-acid, carboxyl-group
structures in fulvic acid from the Suwannee River, Georgia. 1. Minor structures. Environ.
Sci. Technol. 29, 393-398.
Leenheer J. A., Wershaw R. L., and Reddy M. M. (1995b) Strong-acid, carboxyl-group
structures in fulvic acid from the Suwannee River, Georgia. 2. Major structures. Environ.
Sci. Technol. 29, 399-405.
Lehtonen T., Peuravuori J., and Pihlaja K. (2000) Characterisation of lake-aquatic humic
matter isolated with two different sorbing solid techniques: Tetramethylammonium
hydroxide treatment and pyrolysis-gas chromatography/mass spectrometry. Anal. Chim.
Acta 424, 91-103.
Leinweber P. and Schulten H.-R. (1998) Advances in analytical pyrolysis of soil organic
matter. J. Anal. Appl. Pyrolysis 47, 165-189.
Leinweber P. and Schulten H.-R. (2000) Nonhydrolyzable forms of soil organic nitrogen:
Extractability and composition. J. Plant Nutr. Soil Sci. 163,433-439.
Lu X. Q., Hanna J. V., and Johnson W. D. (2000) Source indicators of humic substances:
An elemental composition, solid state 13C CP/MAS NMR and Py-GC/MS study. Appl.
Geochem. 15, 1019-1033.
Lu Y. and Allen H. E. (2002) Characterization of copper complexation with natural
dissolved organic matter (DOM) - link to acidic moieties of DOM and competition by Ca
and Mg. Water Res. 36, 5083-5101.
Mahieu N., Oik D. C., and Randall E. W. (2000) Accumulation of heterocyclic nitrogen
in humified organic matter: A 15N-NMR study of lowland rice soils. Euro. J. Soil Sci. 51,
379-389.
Mahieu N., Oik D. C., and Randall E. W. (2002) Multinuclear magnetic resonance
analysis of two humic acid fractions from lowland rice soils. J. Environ. Qual. 31, 421-
430.
Mahieu N., Powlson D. S., and Randall E. W. (1999) Statistical analysis of published
carbon-13 CPMAS NMR spectra of soil organic matter. Soil Sci. Soc. Amer. J. 63, 307-
319.
Mao J., Ding G., and Xing B. (2002b) Domain mobility of humic acids investigated with
one- and two-dimensional nuclear magnetic resonance: Support for dual-mode sorption
model. Commun. Soil Sci. Plant Anal. 33, 1679-1688.
Mao J., Hu W., Schmidt-Rohr K., Davies G., Ghabbour E. A., and Xing B. (1998)
Structure and elemental composition of humic acids: Comparison of solid-state 13C
NMR calculations and chemical analyses. In Humic Substances: Structures, Properties
and Uses (ed. G. Davies and E. A. Ghabbour), pp. 79-90. Royal Society of Chemistry.
291
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Mao J. D., Hu W. G., Schmidt-Rohr K., Davies G., Ghabbour E. A., and Xing B. (2000)
13
Quantitative characterization of humic substances by solid-state C nuclear magnetic
resonance. Soil Sci. Soc. Am. J. 64, 873-884.
Mao J.-D., Hundal L. S., Thompson M. L., and Schmidt-Rohr K. (2002a) Correlation of
poly(methylene)-rich amorphous aliphatic domains in humic substances with sorption of
a nonpolar organic contaminant, phenanthrene. Environ. Sci. Technol. 36, 929-936.
Mao J.-D., Xing B., and Schmidt-Rohr K. (2001) New structural information on a humic
acid from two-dimensional correlation solid-state nuclear magnetic resonance.
Environ. Sci. Technol. 35, 1928-1934.
Martin F., Almendros G., Gonzalez-Vila F. J., and Verdejo T. (2001) Experimental
reappraisal of flash pyrolysis and low-temperature thermally assisted hydrolysis and
methylation using tetramethylammonium hydroxide for the molecular characterization of
humic acids. J. Anal. Appl. Pyrolysis 61,133-145.
Martinez C. E., McBride M. B., Kandianis M. T., Duxbury J. M., Yoon S.-J., and Bleam
W. F. (2002) Zinc-sulfur and cadmium-sulfur associations in metalliferous peats:
Evidence from spectrscopy, distribution coefficients, and phytoavailability. Environ. Sci.
Technol. 36, 3683-3689.
Mathers N. J., Mao X. A., Xu Z. H., Saffigna P. G., Bemers-Price S. J., and Perera M. C.
S. (2000) Recent advances in the application of 13C and 15N NMR spectroscopy to soil
organic matter studies. Aust. J. Soil Res. 38, 769-787.
Matsuoka O., Clementi E., and Yoshimine M. (1976) Cl study of the water dimer
potential surface. J. Chem. Phys. 64,1351-1361.
Maurice P. A. and Namjesnik-Dejanovic K. (1999) Aggregate structures of sorbed humic
substances observed in aqueous solutions. Environ. Sci. Technol. 33, 1538-1541.
Mayo S. L., Olafson B. D., and Goddard W. A. (1990) DREIDING: A generic force field
for molecular simulations. J. Phys. Chem. 94, 8897-8909.
Mehlman J. (2003) atomicDistance. v0.2a.
Monteil-Rivera F., Brouwer E. B., Masset S., Deslandes Y., and Dumonceau J. (2000)
Combination of x-ray photoelectron and solid-state 13C nuclear magnetic resonance
spectroscopy in the structural characterisation of humic acids. Anal. Chim. Acta 424, 243-
255.
Morra M. J., Fendorf S. E., and Brown P. D. (1997) Speciation of sulfur in humic and
fulvic acids using X-ray absorption near edge structure (XANES) spectroscopy.
Geochim. Cosmochim. Acta 61, 683-688.
Myneni S. C. B. (2002) Soft x-ray spectroscopy and spectromicroscopy studies of
organic molecules in the environment. In Applications o f Synchrotron Radiation in Low-
292
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Temperature Geochemistry and Environmental Science (ed. P. A. Fenter, M. L. Rivers,
N. C. Sturchio, and S. R. Sutton), pp. 485-579. Mineralogical Society of America.
Myneni S. C. B., Warwick T. A., Martinez G. A., and Meigs G. (1998) C-functional
group chemistry of humic substances and their spatial variation in soils, Advanced Light
Source Compendium o f User Abstracts, accessed June 21, 2003 from:
http://www-als.lbl.gov/als/compendium.
Namjesnik-Dejanovic K. and Maurice P. A. (2000) Conformations and aggregate
structures of sorbed natural organic matter on muscovite and hematite. Geochim.
Cosmochim. Acta 65, 1047-1057.
Nanny M. A. and Kontas C. (2002) Fluroescent spectroscopic analysis of humic and
fulvic acid micelle-like structures with 6-propionyl-2-dimethylaminonaphthalene
(Prodan). 223rd American Chemical Society National Meeting. Orlando, FL.
Nantsis E. A. and Carper W. R. (1998a) Molecular structure of divalent metal ion-fulvic
acid complexes. J. Molec. Struc. 423, 203-212.
Nantsis E. A. and Carper W. R. (1998b) Effects of hydration on the molecular structure
of divalent metal ion-fulvic acid complexes: A MOP AC (PM3) study. J. Molec. Struc.
431, 267-275.
Nantsis E. A. and Carper W. R. (1998c) Effects of hydration on the molecular structure
of magnesium-fulvic acid complexes: A MOPAC (PM3) study. J. Molec. Struc. 468, 51-
58.
Negre M., Schulten H.-R., Gennari M., and Vindrola D. (2001) Interactions of
imidazolinone herbicides with soil humic acids. Experimental results and molecular
modeling. J. Environ. Sci. Health B 36, 107-125.
Nierop K. G. J., Pulleman M. M., and Marinissen J. C. Y. (2001) Management induced
organic matter differentiation in grassland and arable soil: A study using pyrolysis
techniques. Soil Biol. Biochem. 33, 755-764.
Nurmi J. T. and Tratnyek P. G. (2002) Electrochemical properties of natural organic
matter (NOM), fractions of NOM, and model biogeochemical electron shuttles. Environ.
Sci. Technol. 36, 617-624.
Ohtaki H. and Radnai T. (1993) Structure and dynamics of hydrated ions. Chem. Rev. 93,
1157-1204.
Oik D. C., Brunetti G., and Senesi N. (2000) Decrease in humification of organic matter
with intensified lowland rice cropping: A wet chemical and spectroscopic investigation.
Soil Sci. Soc. Am. J. 64, 1337-1347.
Orlov D. S. (1995) Humic Substances o f Soils and General Theory o f Humification. A A
Balkema Publishers.
293
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Page D. W., van Leeuwen J. A., Spark K. M., and Mulcahy D. E. (2002) Pyrolysis
characterization of plant, humus and soil extracts from Australian catchments. J. Anal.
Appl. Pyrolysis 65, 269-285.
Park S.-H. and Sposito G. (?) Molecular modeling of clay mineral structure and surface
chemistry. In ...
Perminova I. V. (1999) Size exclusion chromatography of humic substances:
Complexities of data interpretation attributable to non-size exclusion effects. Soil Sci.
164, 834-840.
Peters K. E. (2000) Petroleum tricyclic terpanes: Predicted physicochemical behavior
from molecular mechanics calculations. Org. Geochem. 31, 497-507.
Piccolo A. (2001) The supramolecular structure of humic substances. Soil Sci. 166, 810-
832.
Piccolo A., Conte P., Cozzolino A., and Spaccini R. (2001) Molecular sizes and
association forces of humic substances in solution. In Humic Substances and Chemical
Contaminants (ed. C. E. Clapp, M. H. B. Hayes, N. Senesi, P. R. Bloom, and P. M.
Jardine), pp. 89-118. Soil Science Society of America.
Pignatello J. J. (1999) A revised physical concept of natural organic matter as a sorbent
of organic compounds. In Mineral-Water Interfacial Reactions: Kinetics and
Mechanisms (ed. D. L. Sparks and T. J. Grundl), pp. 204-221. American Chemical
Society.
Plancque G., Amekraz B., Moulin V., Toulhoat P., and Moulin C. (2001) Molecular
structure of fulvic acids by electrospray wtih quadrupole time-of-flight mass
spectrometry. Rapid Comm. Mass Spec. 15, 827-835.
Plaschke M., Romer J., Klenze R., and Kim J. I. (1999) In situ AFM study of sorbed
humic acid colloids at different pH. Coll. Surf. A 160, 269-279.
Porquet A., Bianchi L., and Stoll S. (2003) Molecular dynamic simulations of fulvic acid
clusters in water. Coll. Surf. A 217, 49-54.
Preston C. M. and Ripmeester J. A. (1982) Application of solution and solid-state 13C
NMR to four organic soils, their humic acids, fulvic acids, humins and hydrolysis
residues. Can. J. Spectrosc. 27, 99-105.
Purcell W. P. and Singer J. A. (1967) A brief review and table of semiempirical
parameters used in the Hueckel molecular orbital method. J. Chem. Eng. Data 12, 235-
246.
Qian J., Skyllberg U., Freeh W., Bleam W. F Bloom P. R and Petit P. E. (2002)
Bonding of methyl mercury to reduced sulfur groups in soil and stream organic matter as
294
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
determined by X-ray absorption spectroscopy and binding affinity studies. Geochim.
Cosmochim. Acta 66, 3873-3885.
Rappe A. K., Casewit C. J., Colwell K. S., Goddard W. A , and Skiff W. M. (1992) UFF,
a full periodic table force field for molecular mechanics and molecular dynamics
simulations. J. Am. Chem. Soc. 114, 10024-10035.
Ritchie C. D., Bierl B. A., and Honour R. J. (1962) The solvation of polar groups. I. A
study of solvent effects on infrared band intensities. J. Am. Chem. Soc. 84, 4687-4692.
Ritchie J. D. and Perdue E. M. (2003) Proton-binding study of standard and reference
fulvic acids, humic acids, and natural organic matter. Geochim. Cosmochim. Acta 67, 85-
96.
Robinson G. W., Zhu S.-B., Singh S., and Evans M. W. (1996) Water in Biology,
Chemistry and Physics: Experimental Overviews and Computational Methodologies.
World Scientific.
Saiz-Jimenez C. (1992) Applications of pyrolysis-gas chromatography/mass spectrometry
to the study of humic substances: Evidence of aliphatic biopolymers in sedimentary and
terrestrial humic acids. Sci.TotalEnviron. 117/118, 13-25.
Saiz-Jimenez C. (1994) Pyrolysis/methylation of soil fulvic acids: Benzenecarboxylic
acids revisited. Environ. Sci. Technol. 28, 197-200.
Saiz-Jimenez C. (1995) The origin of alkylbenzenes and thiophenes in pyrolysates of
geochemical samples. Org. Geochem. 23, 81-85.
Saiz-Jimenez C. (1996) The chemical structure of humic substances: Recent advances. In
Humic Substances in Terrestrial Ecosystems (ed. A. Piccolo), pp. 1-44. Elsevier.
Scheinost A. C., Kretzschmar R., Christl I., and Jacobsen C. (2001) Carbon group
chemistry of humic and fulvic acid: A comparison of C-ls NEXAFS and 13C-NMR
spectroscopies. In Humic Substances: Structures, Models and Functions (ed. E. A.
Ghabbour and G. Davies), pp. 39-47. Royal Society of Chemistry.
Schmidt M. W. I., Knicker H., Hatcher P. G., and Kogel-Knabner I. (1997) Improvement
of 13-C and 15-N CPMAS NMR spectra of bulk soils, particle size fractions and organic
material by treatment with 10% hydrofluoric acid. Euro. J. Soil Sci. 48, 319-328.
Schnitzer M. (1990) Selected methods for the characterization of soil humic substances.
In Humic Substances in Soil and Crop Sciences: Selected Readings (ed. P. MacCarthy, C.
E. Clapp, R. L. Malcolm, and P. R. Bloom), pp. 65-89. American Society of Agronomy
and Soil Science Society of America.
Schnitzer M. and Griffith S. M. (1975) Novel method for estimating hydrogen-bonded
CO2H groups in humic substances. Can. J. Soil Sci. 55, 491-493.
295
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Schnitzer M. and Preston C. M. (1986) Analysis of humic acids by solution and solid-
state carbon-13 nuclear magnetic resonance. Soil Sci. Soc. Amer. J. 50, 326-331.
Schnitzer M. and Schulten H.-R. (1998) New ideas on the chemical make-up of soil
humic and fulvic acids. In Future Prospects for Soil Chemistry, Vol. 55 (ed. P. M.
Huang, D. L. Sparks, and S. A. Boyds), pp. 153-177. Soil Science Society of America.
Schulten H.-R. (1999a) Analytical pyrolysis and computational chemistry of aquatic
humic substances and dissolved organic matter. J. Anal. Appl. Pyrolysis 49, 385-415.
Schulten H.-R. (1999b) Interactions of dissolved organic matter with xenobiotic
compounds: Molecular modeling in water. Environ. Toxicol. Chem. 18, 1643-1655.
Schulten H. R. and Schnitzer M. (1997) Chemical model structures for soil organic
matter and soils. Soil Sci. 162, 115-130.
Schulten H.-R. and Leinweber P. (1996) Characterization of humic and soil particles by
analytical pyrolysis and computer modeling. J. Anal. Appl. Pyrolysis 38, 1-53.
Schulten H.-R. and Leinweber P. (2000) New insights into organic-mineral particles:
Composition, properties and models of molecular structure. Biol. Fertil. Soils 30, 399-
432.
Schulten H.-R., Leinweber P., and Schnitzer M. (1998) Analytical pyrolysis and
computer modelling of humic and soil particles. In Structure and Surface Reactions o f
Soil Particles (ed. P. M. Huang, N. Senesi, and J. Buffle), pp. 281-324. John Wiley &
Sons Ltd.
Schulten H.-R. and Schnitzer M. (1993) A state of the art structural concept for humic
substances. Naturwissenschaften 80, 29-30.
Schulten H.-R. and Schnitzer M. (1998) The chemistry of soil organic nitrogen: A
review. Biol. Fertil. Soils 26, 1-15.
Schulten H.-R., Sorge-Lewin C., and Schnitzer M. (1997) Structure of "unknown" soil
nitrogen investigated by analytical pyrolysis. Biol. Fertil. Soils 24, 249-254.
Schulten H.-R., Thomsen M., and Carlsen L. (2001) Humic complexes of diethyl
phthalate: Molecular modelling of the sorption process. Chemosphere 45, 357-369.
Sein L. T., Varnum J. M., and Jansen S. A. (1999) Conformational modeling of a new
building block of humic acid: Approaches to the lowest energy conformer. Environ. Sci.
Technol. 33, 546-552.
Senesi N. (1992) Application of electron spin resonance and fluorescence spectroscopies
to the study of soil humic substances. In Humus, Its Structure and Role in Agriculture
and Environment (ed. J. Kubat), pp. 11-26. Elsevier Science Publishing Company.
296
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Shevchenko S. M. and Bailey G. W. (1996) The mystery of the lignin-carbohydrate
complex: A computational approach. J. Molec. Struc. 364, 197-208.
Shevchenko S. M. and Bailey G. W. (1998a) Modeling sorption of soil organic matter on
mineral surfaces: Wood-derived polymers on mica. Supramolecular Sci. 5, 143-157.
Shevchenko S. M. and Bailey G. W. (1998b) Non-bonded organo-mineral interactions
and sorption of organic compounds on soil surfaces: A model approach. J. Molec. Struc.
422,259-270.
Shevchenko S. M., Bailey G. W., and Akim L. G. (1999) The conformational dynamics
of humic polyanions in model organic and organo-mineral aggregates. J. Molec. Struc.
460,179-190.
Shin H.-S., Monsallier J. M., and Choppin G. R. (1999) Spectroscopic and chemical
characterizations of molecular size fractionated humic acid. Talanta 50, 641-647.
Simpson A. (2001) Multidimensional solution state NMR of humic substances: A
practical guide and review. Soil Sci. 166, 795-809.
Simpson A. J., Burdon J., Graham C. L., Hayes M. H. B., Spencer N., and Kingery W. L.
(2001a) Interpretation of heteronuclear and multidimensional NMR spectroscopy of
humic substances. Euro. J. Soil Sci. 52, 495-509.
Simpson A. J., Kingery W. L., and Hatcher P. G. (2003) The identification of plant
derived structures in humic materials using three-dimensional NMR spectroscopy.
Environ. Sci. Technol. 37, 337-342.
Simpson A. J., Kingery W. L., Hayes M. H. B., Spraul M., Humpfer E., Dvortsak P.,
Kerssebaum R., Godejohann M., and Hofmann M. (2002) Molecular structures and
associations of humic substances in the terrestrial environment. Naturwissenschaften 89,
84-88.
Simpson A. J., Kingery W. L., Shaw D. R., Spraul M., Humpfer E., and Dvortsak P.
(2001c) The application of HR-MAS NMR spectroscopy for the study of structures
and associations of organic components at the solid-aqueous interface of a whole soil.
Environ. Sci. Technol. 35, 3321-3325.
Simpson A. J., Kingery W. L., Spraul M., Humpfer E., Dvortsak P., and Kerssebaum R.
(2001b) Separation of structural components in soil organic matter by diffusion ordered
spectroscopy. Environ. Sci. Technol. 35, 4421-4425.
Skouras E. D., Burganos V. N., and Payatakes A. C. (2001) Improved atomistic
simulation of diffusion and sorption in metal oxides. J. Chem. Phys. 114, 545-552.
Smemik R. J. and Oades J. M. (2000a) The use of spin counting for determining
quantitation in solid state 13C NMR spectra of natural organic matter. 1. Model systems
and the effects of paramagnetic impurities. Geoderma 96, 101-129.
297
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Smemik R. J. and Oades J. M. (2000b) The use of spin counting for determining
quantitation in solid state 13C NMR spectra of natural organic matter. 2. HF-treated soil
fractions. Geoderma 96, 159-171.
Smemik R. J. and Oades J. M. (2003) Spin accounting and RESTORE - two new
methods to improve quantification in solid-state 13C NMR analysis of soil organic matter.
Euro. J. Soil Sci. 54, 103-116.
Soper A. K., Bmni F., and Ricci M. A. (1997) Site-site pair correlation functions of water
from 25 to 400 degrees C: Revised analysis of new and old diffraction data. J. Chem.
Phys. 106, 247-254.
Soper A. K. and Phillips M. G. (1986) A new determination of the structure of water at
25 C. Chem Phys. 107, 47-60.
Sposito G., Park S. H., and Sutton R. (1999) Monte Carlo simulation of the total radial
distribution function for interlayer water in sodium and potassium montmorillonites.
Clays Clay Min. 47, 192-200.
Steelink C. (1985) Implications of elemental characteristics of humic substances. In
Humic Substances in Soil, Sediment, and Water (ed. G. R. Aiken, D. M. McKnight, and
R. L. Wershaw), pp. 457-476. John Wiley & Sons.
Stenson A. C., Landing W. M., Marshall A. G., and Cooper W. T. (2002) Ionization and
fragmentation of humic substances in electrospray ionization Fourier transform-ion
cyclotron resonance mass spectrometry. Anal. Chem. 74, 4397-4409.
Stevenson F. J. (1994) Humus chemistry: Genesis, Composition, Reactions, 2nd Ed. John
Wiley & Sons, Ltd.
Stevenson F. J. (1982) Humus chemistry: Genesis, Composition, Reactions. John Wiley
& Sons, Ltd.
Stewart J. J. P. (1989a) Optimization of parameters for semiempirical methods. 1.
Method. J. Comput. Chem. 10, 209-220.
Stewart J. J. P. (1989b) Optimization of parameters for semiempirical methods. 2.
Applications. J. Comput. Chem. 10, 221-264.
Struyk Z. and Sposito G. (2001) Redox properties of standard humic acids. Geoderma
102, 329-346.
Sun H. (1998) COMPASS: An ab initio force-field optimized for condensed-phase
applications - Overview with details on alkane and benzene compounds. J. Phys. Chem.
B 102, 7338-7364.
Sun H., Ren P., and Fried J. R. (1998) The COMPASS force field: Parameterization and
validation for phosphazenes. Comp. Theor. Polym. Sci. 8 , 229-246.
298
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Swift R. S. (1999) Macromolecular properties of soil humic substances: Fact, fiction, and
opinion. Soil Sci. 164, 790-802.
Swift R. S., Leonard R. L., Newman R. H., and Theng B. K. G. (1992) Changes in humic
acid composition with molecular weight as detected by 13C-nuclear magnetic resonance
spectroscopy. Sci. Total Environ. 117/118, 53-61.
Szulczewski M. D., Helmke P. A., and Bleam W. F. (2001) XANES spectrscopy studies
of Cr(VI) reduction by thiols in organosulfur compounds and humic substances. Environ.
Sci. Technol. 35, 1134-1141.
Tipping E. (2002) Cation Binding by Humic Substances. Cambridge University Press.
Ussiri D. A. N. and Johnson C. E. (2003) Characterization of organic matter in a northern
hardwood forest soil by 13C NMR spectroscopy and chemical methods. Geoderma 111,
123-149.
Vairavamurthy A. and Wang S. (2002) Organic nitrogen in geomacromolecules: Insights
on speciation and transformation with K-edge XANES spectroscopy. Environ. Sci.
Technol. 36, 3050-3056.
Vairavamurthy M. A., Malefic D., Wang S., Manowitz B., Eglinton T., and Lyons T.
(1997) Characterization of sulfur-containing functional groups in sedimentary humic
substances by X-ray absorption near-edge structure spectroscopy. Energy Fuels 11, 546-
553.
Van den Hoop M. A. G. T., van Leeuwen H. P., and Cleven R. F. M. J. (1990) Study of
the polyelectrolyte properties of humic acids by conductimetric titration. Anal. Chim.
Acta 232, 141-148.
Van Heemst J. D. H., van Bergen P. F., Stankiewicz B. A., and de Leeuw J. W. (1999)
Multiple sources of alkylphenols produced upon pyrolysis of DOM, POM and recent
sediments. J. Anal. Appl. Pyrolysis 52, 239-256.
Van Loon W. M. G. M., Boon J. J., and de Groot B. (1993) Quantitative analysis of
sulfonic acid groups in macromolecular lignosulfonic acids and aquatic humic substances
by temperature-resolved pyrolysis-mass spectrometry. Environ. Sci. Technol. 27, 2387-
2396.
Varga B., Kiss G., Galambos I., Gelencser A., Hlavay J., and Krivacsy Z. (2000)
Secondary structure o f humic acids. Can micelle-like conformation be proved by aqueous
size exclusion chromatography? Environ. Sci. Technol. 34, 3303-3306.
Vempati R. K., Hess T. R., and Cocke D. L. (1996) X-ray photoelectron spectroscopy. In
Methods o f Soil Analysis. Part 3 - Chemical Methods (ed. D. L. Sparks), pp. 357-375.
Soil Science Society of America.
299
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Von Wandruszka R. and Engebretson R. (2001) Kinetics of humic acid associations. In
Humic Substances and Chemical Contaminants (ed. C. E. Clapp, M. H. B. Hayes, N.
Senesi, P. R. Bloom, and P. M. Jardine), pp. 119-126. Soil Science Society of America.
Von Wandruszka R., Engebretson R. R., and Yates L. M. (1999) Humic acid
pseudomicelles in dilute aqueous solution: Fluorescence and surface tension
measurements. In Understanding Humic Substances: Advanced Methods, Properties and
Applications (ed. E. A. Ghabbour and G. Davies), pp. 79-85. Royal Society of Chemistry.
Wang K., Dickinson L. C., Ghabbour E. A., Davies G., and Xing B. (2001) Investigation
of molecular motion of humic acids with 1-D and 2-D solution NMR. In Humic
Substances: Structures, Models and Functions (ed. E. A. Ghabbour and G. Davies), pp.
73-82. Royal Society of Chemistry.
Winter M. (2003) WebElements. The University of Sheffield and WebElements Ltd.,
UK. Accessed 14 November 2003 from: http://www.webelements.com.
Xia K., Weesner F., Bleam W. F., Bloom P. R., Skyllberg U. L., and Helmke P. A.
(1998) XANES studies of oxidation states of sulfur in aquatic and soil humic substances.
Soil Sci. Soc. Am. J. 62, 1240-1246.
Yang J., Ren Y., Tian A.-M., and Sun H. (2000) COMPASS force field for 14 inorganic
molecules, He, Ne, Ar, Kr, Xe, H2 , O2 , N2 , NO, CO, CO2 , NO2 , CS2 , and SO2 , in liquid
phases. J. Phys. Chem. B 104, 4951-4957.
Yuan G. and Xing B. (2001) Effects of metal cations on sorption and desorption of
organic compounds in humic acids. Soil Sci. 166, 107-115.
Zang X., Brown J. C., van Heemst J. D. H., Palumbo A., and Hatcher P. G. (2001)
Characterization of amino acids and proteinaceous materials using online
tetramethylammonium hydroxide (TMAH) thermochemolysis and gas chromatography-
mass spectrometry technique. J. Anal. Appl. Pyrolysis 61, 181-193.
Zang X., van Heemst J. D. H., Dria K. J., and Hatcher P. G. (2000) Encapsulation of
protein in humic acid from a histosol as an explanation for the occurrence of organic
nitrogen in soil and sediment. Org. Geochem. 31, 679-695.
300
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Chapter Five
Molecular modeling of a DOM-montmorillonite system
5.1 Introduction
An understanding of organo-mineral interactions that lends itself to prediction of
soil C storage capacity is essential to the design of land management methods promoting
soil C sequestration as a means of reducing atmospheric CO2 concentrations and resulting
climate change. Particularly important in temperate ecosystems are the interactions
between organic molecules and common aluminosilicate minerals, such as
montmorillonite, that lead to protection of the organic material from microbial decay over
long time periods. Humic substances, the poorly defined mix of partially degraded
organic molecules described extensively in Section 4.1, are stabilized by such organo-
mineral interactions (Oades, 1988; Schulten and Leinweber, 1996; Zech et al., 1997;
Baldock and Skjemstad, 2000), so much so that the C atoms making up these molecules
have mean residence times of hundreds to thousands of years (Stevenson, 1994;
Eusterhues et al., 2003).
Organic molecules can adsorb to mineral surfaces through a wide variety of
mechanisms, including hydrophobic interactions, H-bonding, protonation, ligand or ion
exchange, and cation or water bridging (Sposito, 1989) (Table 1.1). The prevalence of
each of these types of interactions is determined by the organic functional groups
available and the mineral surface under scrutiny, as influenced by soil conditions.
Because humic substances are derived from biological molecules, they contain functional
groups typical of most natural organic substances, including alkyl and aromatic
301
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
backbones, carboxyl groups, which under typical soil pH regimes are deprotonated to
form carboxylates, and numerous other moieties featuring O, N, and S atoms, as
described in Sections 4.1.3-4.1.5. The diverse functionality contained within the humic
fraction allows these molecules to participate in a variety of adsorption interactions
simultaneously. The heterogeneity of the siloxane surface of smectite minerals such as
montmorillonite also results in the availability of a number of different interaction sites.
Negative charge sites within the clay mineral attract cations, which allow negatively
charged organic molecules to form cation bridges (Sposito, 1989). Water bridges form
when a water molecule intercedes between the cation and the organic moiety (Sposito,
1989). Such bridging interactions involving polar organic molecules are also known as
ion-dipole interactions. In addition, organic cations can replace inorganic ones near these
sites, becoming adsorbed through an exchange reaction. Locations distant from these
charge sites are in fact hydrophobic, providing surfaces suitable for hydrophobic
interactions (Jaynes and Boyd, 1991a, 1991b; Boyd and Jaynes, 1992; van Oss and Giese,
1995; Giese et al., 1996; Laird, 1996; Chiou and Rutherford, 1997). Finally, the O atoms
covering the siloxane surface are capable of forming H-bonds with suitable organic
functional groups.
An organo-mineral complex results from the formation of multiple organo-
mineral interactions. The structure of such a complex determines the level of protection
from microbial degradation that the mineral matrix provides to the adsorbed organic
molecule. An organic molecule adsorbed to the external surfaces of several clay particles
receives partial physical protection from surrounding minerals, because microbes and
their enzymes can approach the substrate from only a few different directions (Baldock
302
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
and Skjemstad, 2000). Such microaggregates undoubtedly contain most of the organic
material associated with smectite minerals. It may be possible for humic molecules to
slip between smectite layers and into the interlayer region, creating organic interlayer
complexes that receive more complete shielding from microbial attack given the narrow
dimensions of the interlayer space. Organic molecules tucked within a clay interlayer,
and thus protected from microbial degradation, could form a very stable and long-lived
pool of soil carbon. Organic materials have been found within the interlayers of smectite
minerals taken from a few different soil ecosystems (Kodama and Schnitzer, 1971; Satoh
and Yamane, 1971; Theng et al., 1986; Righi et al., 1995). Radiocarbon dating
performed on one of these samples indicated the interlayer organic substance was
thousands of years old (Theng et al., 1992).
Natural organo-mineral complexes contain an incredible level of heterogeneity,
such that identification of specific organo-mineral interactions is quite impossible using
typical spatially averaged spectroscopic means. Instead, it is only possible to infer which
mechanisms are active through simple analysis of basic properties of the complexes, and
their reactions to changes in conditions. For example, the organic materials found within
smectite interlayers in soils are predominantly alkyl (Theng et al., 1986; Schnitzer et al.,
1988; Schulten et al., 1996) or alkylaromatic (Righi et al., 1995), suggesting that
hydrophobic interactions may play a large role in the formation of these organo-mineral
complexes. On the other hand, the addition of Ca2+ often leads to reduced degradation
and release of C in soils, suggesting that cation bridges are essential to the formation of
fresh microaggregates that protect organic matter from microbial attack (Sokoloff, 1938;
Muneer and Oades, 1989a-c). Simplified organo-mineral systems, synthesized from pure
303
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
clay minerals and well defined chemicals such as pesticides, can be used to investigate
specific binding mechanisms, but cannot capture the complexity of a humic molecule
adsorbing to a smectite surface. Molecular modeling, however, can be used to study such
i
complex systems. The analysis of interactions that form between a model humic
molecule and a model montmorillonite surface may help to further our understanding of
the mechanisms of adsorption available to organic molecules in soil systems.
Molecular modeling techniques are applied most frequently to systems that are
either completely inorganic, like those cited in Section 2.1, or completely organic, some
of which are described in Sections 3.1 and 4.1.7. Organo-mineral systems have received
less attention because force fields that adequately describe the interactions of atoms in
both organic and inorganic molecules are rare. Some researchers have performed
organo-mineral simulations using less accurate, generalized force fields, often originally
designed to capture only approximately the traits of organic molecules (Keldsen et al.,
1994; Schulten and Leinweber, 1996, 2000; Park et al., 1997; Schulten and Schnitzer,
1997; Akim et al., 1998; Janeba et al., 1998; Leinweber and Schulten, 1998; Schulten et
al., 1998; Shevchenko and Bailey, 1998a, b; Shevchenko et al., 1999; Capkova et al.,
2000,2003; Titiloye and Skipper, 2000; Bailey et al., 2001; Pospisil et al., 2001, 2002,
2003; Newman et al., 2002; Boulet et al., 2003; Gaudel-Siri et al., 2003; Park and
Sposito, 2003), while others have constructed force fields tailored to systems of interest
by combining aspects of several relevant force fields (Sato et al., 1992a, b, 1996; Breu
and Catlow, 1995; Hackett et al., 1998, 2000; Teppen et al., 1998; Breu et al., 1999;
Bujdak et al., 2000; Yu et al., 2000a, b, 2003; Pintore et al., 2001; Kuppa and Manias,
2002, 2003; Heinz et al., 2003; Kuppa et al., 2003). Quantum chemical calculations are
304
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
capable of simulating organo-mineral systems, as long as they are quite small (Kubicki et
al., 1997; Gorb et al., 2000; Haberhauer et al., 2001; van Duin and Larter, 2001; Yu et al.,
2001; Teppen et al., 2002; Tunega et al., 2002).
A review of simulations of organo-mineral systems featuring smectites or other
siloxane surfaces provides examples of numerous organo-mineral adsorption
mechanisms, including hydrophobic interactions (Keldsen et al., 1994; Akim et al., 1998;
Shevchenko and Bailey, 1998a, b; Teppen et al., 1998, 2002; Yu et al., 2000a, b, 2001,
2003; Titiloye and Skipper, 2000; Haberhauer et al., 2001; van Duin and Larter, 2001;
Park and Sposito, 2003), H-bonding (Gorb et al., 2000; Yu et al., 2000b; Haberhauer et
al., 2001; Newman et al., 2002; Tunega et al., 2002), cation exchange (Sato et al., 1992a,
b, 1996; Breu and Catlow, 1995; Park et al., 1997; Janeba et al., 1998; Breu et al., 1999;
Bujdak et al., 2000; Capkova et al., 2000, 2003; Yu et al., 2000a; Pospisil et al., 2001,
2002,2003; Heinz et al., 2003), ion-dipole interactions involving polar organic
molecules, which feature cation bridges within dehydrated systems and water bridges
within hydrated systems (Hackett et al., 1998, 2000; Bujdak et al., 2000; Pintore et al.,
2001; Kuppa and Manias, 2002, 2003; Pospisil et al., 2002; Boulet et al., 2003; Capkova
et al., 2003; Gaudel-Siri et al., 2003; Kuppa et al., 2003), and cation bridging interactions
involving charged functional groups (Akim et al., 1998; Shevchenko and Bailey, 1998a,
b; Bailey et al., 2001; Pintore et al., 2001; Yu et al., 2003). Studies that describe cation
bridges have examined monovalent cations only, and have never accounted for the fully
hydrated state such cations would possess prior to the formation of the adsorption
complex, conditions that are not particularly relevant to natural soil systems. Simulations
between model humic substances and siloxane surfaces indicate that when such
305
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
molecules are fully protonated, H-bonding (Schulten and Leinweber, 1996, 2000;
Schulten and Schnitzer, 1997; Leinweber and Schulten, 1998; Schulten et al., 1998) and
hydrophobic (Akim et al., 1998; Shevchenko and Bailey, 1998a, b) interactions dominate,
and when partial deprotonation results in the formation of carboxylate groups, cation
bridging interactions prevail (Akim et al., 1998; Shevchenko and Bailey, 1998a;
Shevchenko et al., 1999).
The development of the COMPASS force field (Accelrys Inc., 2001), designed
specifically for modeling condensed phase organo-mineral systems, provides a new
opportunity to perform simulations that may elucidate the mechanisms by which C is
retained in soils. This Chapter describes the use of the COMPASS force field to model
the interactions of protonated and Ca-saturated DOM molecules within a hydrated Ca-
montmorillonite interlayer. The variation in protonation state of the two forms of the
DOM molecule corresponds to that expected under highly acidic and near neutral soil pH
regimes, allowing conclusions to be made concerning the effect of pH on organo-mineral
adsorption. The use of hydrated Ca2+within these simulations is expected to provide
information concerning the formation of cation bridges more relevant to soil systems than
that seen in previous work, as cation bridges are thought to form around polyvalent ions
like Ca2+ (Theng, 1979; Oades, 1988,1995; Baldock and Skjemstad, 2000), which are
undoubtedly fully hydrated in solution prior to participation in these bridges. The
hydrated interlayers provide a conveniently periodic system for these simulations.
Although most soil C is not found within smectite interlayers, the organic molecules that
do find their way into this region may be protected from degradation for thousands of
years, forming an extremely stable C pool of interest to those attempting to encourage C
306
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
sequestration in soils. Any conclusions regarding organo-mineral interactions drawn
from these simulations can be applied readily to the adsorption of organic molecules to
external siloxane surfaces of smectite particles as well.
Presented below are the results of simulations of Ca-montmorillonite in dry and
hydrated states using the COMPASS force field (Sun, 1998), performed in order to assess
the suitability of this force field for modeling smectite minerals. Subsequent simulations
of both the protonated and the deprotonated, Ca-saturated DOM molecules (as described
in Sections 4.3.3 and 4.3.4) within a hydrated Ca-montmorillonite interlayer were used to
investigate the effect of protonation on the types of organo-mineral interactions that
form.
5.2 Methods
5.2.1 Dry Ca-montmorillonite System
The montmorillonite clay layer used for these simulations featured atomic
coordinates and partial charges identical to those described in Section 2.2.1, and
originally developed by Skipper et al. (1995a, b). The COMPASS force field (Sun,
1998), described in Section 4.2.1, provided the format for the potential energy equation
employed in these calculations, as well as all valence and van der Waals interaction
parameters. The COMPASS force field does not possess interaction parameters for
octahedrally coordinated Al, necessitating the use of a rigid clay structure.
The simulation cell encompassed eight unit cells of the mineral and six
substitution sites (two tetrahedral charge sites and four octahedral charge sites). Three-
dimensional periodic boundary conditions were applied as described in Section 2.2.1.
307
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Long-range electrostatic interactions were summed using the three-dimensional Ewald
summation method and a reciprocal space cutoff k of 3 A'1, while short-range van der
Waals interactions were summed up to a real-space cutoff of 9 A, as stated in Section
2.2. 1.
To simulate dry Ca-montmorillonite, three Ca2+ were inserted into an interlayer
region between two halves of a clay mineral with an initial layer spacing of 12 A. The
constrained geometry of a dehydrated interlayer region may encourage entrapment of the
dry Ca-montmorillonite system in conformations corresponding to local potential energy
minima, rather than those representing a fully equilibrated state (Section 1.4.1). To
ensure adequate sampling of conformational space, several different Ca-montmorillonite
systems were created, featuring Ca2+ in a variety of starting positions. Ca2+ were placed
near tetrahedral charge sites, octahedral charge sites, or in locations roughly equidistant
from multiple charge sites. Further variation was added through positioning of Ca2+at
the midplane or within six-oxygen cavities.
Each of these initial configurations was minimized using the Cerius2 Smart
Minimizer (Accelrys Inc., 2001) and high convergence criteria, as described in Section
4.2.2. During minimization, all cell parameters were allowed to vary, as the registration
of clay layers can affect interactions that develop in a collapsed clay system (Smith,
1998). Addition of 100 kPa pressure along the z axis of the Ca-montmorillonite
simulation cell was explored as a means of mimicking natural conditions involving
application of pressure normal to the clay layer (Section 2.2.2). However, changes to cell
parameters during minimization inevitably shifted the vector normal to the clay layer
away from an orientation parallel to the z axis, so pressure applied along the z axis was
308
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
not distributed evenly over the mineral surfaces. Consequently, EM calculations were
performed without the addition of pressure.
Minimizations never ran until completion, because the elevated potential energy
of the rigid mineral structure prevented the systems from meeting the high convergence
criteria used. Instead, the calculations continued until no further progress could be made
using the EM algorithms employed. The conformation with a final layer spacing closest
to experimental measurements was designated the best equilibrated Ca-montmorillonite.
Analysis of the dry Ca-montmorillonite systems consisted of collection of two potential
energy distributions, that which included and that which excluded energy contributions
from interactions among atoms within the rigid clay layer, as well as calculation of Ca-0
RDFs and CNs (Section 2.2.2), and calculation of the model IR spectrum for the best
equilibrated conformation (Section 4.2.1).
5.2.2 Hydrated Ca-montmorillonite System
Several initial configurations of the hydrated Ca-montmorillonite system were
created to assess the effects of starting conditions on hydration and adsorption of
interlayer cations. First, two halves of the rigid montmorillonite mineral described in
Section 5.2.1 were positioned with a layer spacing of 15.2 - 15.6 A, a range taken from
XRD measurements of Ca-montmorillonite at -25 C and relative humidities in the range
of 30 - 90% (Hendricks et al., 1940; Mooney et al., 1952; Cases et al., 1997; Chiou and
Rutherford, 1997; Dios Cancela et al., 1997). Then three Ca2+ were added to the
midplane of the inter layer region. Four different sets of Ca2+starting positions were
created, through variations in location near tetrahedral or octahedral charge sites, as well
309
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
as between six-oxygen cavities or oxygen triads. Duplication of each of these partially
constructed Ca-montmorillonites followed, in order to create systems with different
initial cation hydration states. The cations of one set of four systems were surrounded by
six water molecules, arranged in two planes parallel to the clay layers, with a roughly
octahedral geometry. The cations of the other set of four systems were surrounded by
eight water molecules, also arranged in two planes, with a square antiprism geometry.
Finally, interlayer water was added to the system in two parallel layers, bringing the total
number of water molecules for each simulation cell to 64, an amount that has produced
structural and dynamic simulation output consistent with experimental data for smectites
with layer spacings of ~15 A (Chang et al., 1995, 1997,1998, 1999; Sposito et al., 1999;
Park and Sposito, 2000; Greathouse et al., 2000; Greathouse and Storm, 2002). Water
molecules were described by the COMPASS force field, as seen in Section 4.2.2.
Initial simulations of the Ca-montmorillonite hydrate, which allowed cell
parameters to vary, resulted in systems with layer spacings near 14.7 A, lower than seen
in experimental work. Further exploration, via minimization calculations of hydrates
with fixed cell parameters and layer spacings of 15.2, 15.4, and 15.6 A, indicated that the
COMPASS water molecules maintained lower potential energy configurations for
systems with the narrowest layer spacing. Therefore, all Ca-montmorillonite hydrates
were simulated with a fixed 15.2 A layer spacing. An initial minimization was followed
by an annealing MD (NVT) run, in which the temperature was raised in steps of 100 K
from 300 to 600 K and back, and each temperature was held for the equivalent of 5 ps
(Section 4.2.2). After a second minimization, the two systems with the highest potential
energy values were thrown out, while the six systems with lower potential energy values
310
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
were subjected to MD (NVT) calculations at 300 K for 20 ps, then to a final EM
calculation. Potential energy data, as well as Ca-0 RDF and CN information, were
collected for each of these systems, and the final positions of all cations with respect to
clay mineral surface features and charge sites were recorded. Potential energy values
provided for the hydrated system with the lowest energy do not include energy derived
from interactions between the fixed atoms of montmorillonite, but do include energy
derived from interactions between mobile atoms in the interlayer and fixed atoms in the
mineral. Average interlayer water H-O-H bond angles and O-H bond lengths were
computed for the Ca-montmorillonite hydrate structure with the lowest potential energy
after minimization. Hydrogen bonds were calculated as described in Section 4.2.2. In
addition, a model IR spectrum was calculated for the system with the lowest potential
energy.
5.2.3 The DOM-montmorillomte Systems
Two hydrated organo-smectite interlayer systems were examined, one in which
the DOM molecule was fully protonated (Section 4.3.3), and one in which the DOM
2+
molecules carboxyl groups were deprotonated, and Ca hydrates were added to
compensate for the resulting negative charge (Section 4.3.4). To prepare these systems,
first a 32 unit cell Ca-montmorillonite clay layer was constructed from a 2 x 2 array of
the final configuration of the 8 unit Ca-montmorillonite hydrate system with the lowest
potential energy (Section 5.3.1). The lateral (ab) dimensions of the 32 unit simulation
cell were 42.24 A x 36.56 A. This Ca-montmorillonite layer contained 24 isomorphic
substitution sites ( 8 tetrahedral and 16 octahedral), creating a negative charge of 24e that
311
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
was balanced by the presence of 12 Ca2+in the interlayer region. These cations were
surrounded by water molecules within the hydration spheres observed in the original 8
unit Ca-montmorillonite hydrate (Section 5.3.1); all other interlayer water molecules
were eliminated from the system. To allow the organic molecules to fit easily within the
inter layer space, the initial layer spacing of the large Ca-montmorillonite cell was set to
50 A for the protonated DOM molecule, and 55 A for the Ca-DOM complex. Interlayer
cations and associated waters of hydration were distributed such that an equal number of
cations resided near each clay surface.
Next, the DOM systems were prepared for insertion into the clay interlayer
region. The layers of water molecules surrounding the final conformations of the
hydrated, protonated DOM molecule (Section 4.3.3) and the hydrated Ca-DOM complex
2d"
(Section 4.3.4) were removed. Those water molecules added as part of a hydrated Ca
unit in the latter system were not removed. The organic systems were pasted into the
centers of the clay interlayer regions using an orientation adjusted so that none of the
inserted molecules were positioned unrealistically close to other interlayer or mineral
species. The resulting initial protonated DOM-Ca-montmorillonite system consisted of a
2+
rigid, periodic 32 unit Ca-montmorillonite clay layer, 12 Ca ,92 mterlayer water
molecules, and a humic molecule with the formula C447H421O272N15S2 . The initial Ca-
saturated DOM-Ca-montmorillonite system featured a rigid, periodic 32 unit Ca-
montmorillonite clay layer, 50 Ca2+, 320 interlayer water molecules, and a humic
polyanion with the formula C 4 4 7 H 3 4 5 O 2 7 2 N 1 5 S 2 ' 7 6 .
In order to identify a layer spacing more appropriate to each organo-mineral
system, a series of EM calculations was performed that included the application of
312
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
pressure normal to the clay layers. The use of pressure in this sequence of minimizations
encouraged the mineral surfaces in each system to move closer to each other, reducing
the size of the interlayer region. A typical run in this series applied 0.0001 GPa of
pressure along the z axis of the simulation cell, and proceeded for 40,000 EM steps.
Because clay layer registration is unlikely to affect systems with basal spacings greater
than those observed with adsorption of a layer or two of water molecules, only the
c dimensions of the organo-smectite systems were allowed to vary. The final state of the
previous EM run in the sequence was used as the initial state of the next EM run. The
pressure applied to the organo-montmorillonite system containing the protonated DOM
molecule was doubled to 0.0002 GPa during a few of the earlier EM calculations in the
cycle, because little change in layer spacing was observed when using the lower pressure.
The layer spacings and potential energies of the systems were recorded after each
minimization. Although, due to their large size and complexity, the systems never
reached a fully minimized state according to the Cerius2 Smart Minimizer (Accelrys Inc.,
2001), when an EM calculation produced a layer spacing reduction of less than 0.1 A
(Ca-saturated DOM), or produced a slightly increased layer spacing (protonated DOM),
this was considered an indication of minimization sufficient for identification of a
suitable layer spacing for the organo-mineral systems. The layer spacing selected in this
manner was 30.53 A for the protonated system, and 33.32 A for the Ca-saturated system.
At this point, the layer spacing for each system was recorded and fixed. Water
molecules were added to fully hydrate the interlayer using the Assembly/Soak (periodic
boundary conditions) command within Insightll v2000.1 (Accelrys Inc.). Each organo-
mineral system was saved within Cerius2using the Biosym (.car) format, loaded into
313
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Insightll, and soaked with water molecules. Periodic systems featuring only these added
water molecules were then exported from Insightll using the Brookhaven file format,
imported into Cerius2 and loaded with the COMPASS force field (Sun, 1998), then
pasted into the relevant organo-mineral system. 451 water molecules were added to the
interlayer containing the protonated DOM molecule, bringing the total number of
interlayer water molecules to 543, while 400 water molecules were added to the
interlayer region containing the Ca-saturated DOM polyanion, for a total of 720
interlayer water molecules.
The hydrated organo-montmorillonite interlayer systems were minimized, then
subjected to annealing MD (NVT) runs, in which the temperature was raised in steps of
100 K from 300 to 600 K and back, with each temperature condition held for calculations
representing 5 ps (Section 4.2.2). Further minimization of each annealed structure was
followed by a 10 ps MD (NVT) computation at 300 K. Afterward, the systems were
soaked a second time using Insightll (Accelrys Inc.), as described above. The protonated
DOM-montmorillonite system was unable to accommodate any more water molecules,
while the Ca-saturated DOM-montmorillonite system absorbed 132 additional water
molecules, for a total of 852. The densities of the organo-mineral systems as well as the
hydrated interlayer regions were calculated at this point.
Minimization of the Ca-saturated DOM-montmorillonite system was followed by
an annealing MD (NVT) run for both organo-mineral structures. This time each
temperature in the annealing cycle was held for 10 ps. The annealed structures were
minimized, then subjected to 20 ps of MD (NVT) calculations at 300 K. A final
minimization of each system resulted in hydrated organo-mineral complexes assumed to
314
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
be free of physically unrealistic configurations and interactions, and likely associated
with the global energy minimum for each DOM-montmorillonite cell. Potential energy
data were collected after each of the minimizations described above, to monitor the
approach toward these stable organo-mineral structures. Potential energy distributions
were calculated for the clay, water, and DOM molecule components of the two systems
in their final conformations, as well as for different combinations of these components.
Analysis of the organo-montmorillonite systems began with a visual examination
of the protonated and Ca-saturated DOM molecules apart from their hydrated interlayer
environments, and calculation of their surface areas and molecular volumes, as described
in Section 4.2.3. Subsequent inspection of interlayer cations included calculation of
Ca-O RDFs, CNs, and examination of the local structure around each individual Ca2+, as
described in Section 4.2.6. Then, calculation of H-O-H bond angles and O-H bond
lengths of interlayer water molecules proceeded, as described in Section 4.2.2. The
H-bonds occurring within the DOM molecules, and between organic and water atoms,
organic and clay atoms, and clay and water atoms, were defined and counted as described
in Sections 4.2.2 and 4.2.4. This was followed by individual analysis of each organo-
mineral interaction occurring within the DOM-montmorillonite systems. Model IR
spectra were calculated as described in Section 4.2.2 for three periodic systems, the
protonated DOM molecule without surrounding water and clay molecules, and the Ca-
saturated DOM complex, both with and without the water molecules originally hydrating
each Ca2+, but without additional water and clay molecules. All other systems examined
contained too many atoms for the Cerius2IR-Raman module. When possible, spectra of
the DOM molecules were normalized via the peak absorbance of the nitrile band located
315
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
near 2200 cm'1, using the signal from the dry, densely packed DOM molecule IR
spectrum as a reference (Section 4.3.2). Normalization is needed to allow quantitative
comparison of model IR spectra produced from systems containing different molecules,
as the signal absorbances reported in each spectrum are provided in relative units that are
not consistent across spectra. The nitrile peak is the only peak in these spectra that can be
assigned uniquely to a single functional group, and one that is assumed to be relatively
unresponsive to environmental change (Section 4.3.5). Normalization via the nitrile
signal was performed for IR spectra of the DOM molecule in hydrated states (Section
4.3.5), and no obvious problems were observed.
5.3 Results
5.3.1 Simulations o f Ca-montmorillonite
The dehydration of a natural sample of Ca-montmorillonite does not entail the
removal of all water molecules, as could be inferred from the dry Ca-montmorillonite
system described in Section 5.2.1. Although most water does leave the interlayer, a few
molecules become so tightly bound to the divalent cations present that they deprotonate,
forming CaOH+and H3 0 + ionic complexes (Sposito, 1989; Yariv, 1992). Because the
counterion complexes now on hand to balance the charge of the dry montmorillonite are
monovalent, an interlayer ion is available to balance the charge of each individual
substitution site within the mineral. The counterions can nestle within six-oxygen
cavities near each charge site, allowing collapse of the clay layers until the basal spacing
reaches values of 9.6 - 9.8 A (Cases et al., 1997; Chiou and Rutherford, 1997).
316
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
The dry Ca-montmorillonite systems described in Section 5.2.1 contain Ca2+in
the interlayer region, not the more complex CaOH+ and H30 +species, and thus cannot be
considered accurate representations of a dehydrated Ca-montmorillonite mineral.
Although the model mineral systems feature unrealistic interlayer species, potential
energy values and cation-mineral interactions obtained from these calculations can
provide a useful frame of reference for simulations of hydrated Ca-montmorillonite
systems. Because natural mineral systems are hydrated, results from these models of dry
Ca-montmorillonite are meant to provide comparative data for subsequent simulations of
hydrated Ca-montmorillonite, rather than predictive values relevant to a natural
dehydrated smectite system.
Layer spacings of the minimized dry Ca-montmorillonite systems ranged from
10.26 to 11.15 A. One might expect larger simulated than experimental layer spacings,
given the attraction of each of the Ca2+for two separate substitution sites. A divalent
cation drawn in two different directions might not be able to sink within a six-oxygen
cavity, preventing full collapse of the interlayer region. However, only the dry
montmorillonite system with the largest layer spacing (11.15 A) features a single Ca2+in
a location intermediate between two (octahedral) charge sites. All other systems feature a
cation adjacent to each of the two tetrahedral charge sites, and a cation near one of the
four octahedral charge sites.
Closer examination of the dry Ca-montmorillonite systems reveals that cations
attracted to tetrahedral charge sites typically interact with the oxygen triad bonded to the
tetrahedral Al, while those attracted to octahedral charge sites are more likely to reside in
the midplane between six-oxygen cavities present in the two mineral layers. Radial
317
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
distribution functions computed from the final dry clay structures (not shown) provide
distance cutoffs for Ca-0 inner sphere complexation in the range 2.4 - 2.7 A, with each
Ca2+directly coordinated to between two and five surface O atoms. A cation adsorbed to
an oxygen triad typically exhibits inner sphere coordination with two to three O of the
triad, as well as with one to two O of the opposite clay layer. A cation adsorbed between
six-oxygen cavities often possesses inner sphere coordination to one or two O from each
clay layer.
The best equilibrated dry clay system (Figure 5.1) is considered to be that with the
lowest layer spacing, 10.26 A (cell dimensions = 21.05 A x 18.13 A x 10.35 A, cell
angles = 90.46 x 97.69 x 90.27). Potential energy information for this system indicates
that electrostatic interactions dominate both energy distributions provided (Table 5.1).
The two Ca2+adsorbed to tetrahedral charge sites are coordinated to all three of the
associated triad oxygens, as well as one of the Os from the opposite clay layer. The Ca2+
attracted to an octahedral charge site is located between a cavity within each of the clay
layers, and is coordinated to one surface O from each surface. As a result, the average
Ca-0 CN for this system is 3.33.
The Ca-montmorillonite present in soil systems is always hydrated. The ~15 A
Ca-montmorillonite layer spacings observed via XRD under relative humidities in the
range of 30 - 90% (Hendricks et al., 1940; Mooney et al., 1952; Cases et al., 1997; Chiou
and Rutherford, 1997; Dios Cancela et al., 1997) suggest that an average of two layers of
water molecules are adsorbed within each interlayer region. X-ray diffraction analysis
(Slade et al., 1985; Cases et al., 1997) and molecular simulation data (Chavez-Paez et al.,
2001; Greathouse and Storm, 2002) support a bilayer water structure within 15 A Ca-
318
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Figure 5.1. The dry Ca-montmorillonite system with layer spacing 10.26 A, considered
to occupy the global potential energy minimum. Ca2+are represented by large brown
spheres. Red spheres represent O, white spheres represent H, grey spheres represent Si,
blue-green spheres represent Al, and blue spheres represent Mg. The Ca2+to the right
and left of the cell are adsorbed to oxygen triads near tetrahedral charge sites, while the
Ca2+in the center is adsorbed between six-oxygen cavities near an octahedral charge site.
Table 5.1. Energy Distributions for Dry Ca-montmorillonite System
Energy Term Energy (kJ mol'1)
with rigid mineral interactions
Total Potential -96396.0
Bond 664.829
Angle 1143.93
Torsion -24.7045
Inversion 0
Cross-term 15.4236
Van der Waals 2484.97
Electrostatic -100680
without rigid mineral interactions
Total Potential -7313.00
Van der Waals 297.081
Electrostatic -7610.07
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
montmorillonite. Insight regarding the arrangement of interlayer water molecules, as
well as information concerning possible hydration geometries of interlayer cations, can
be used to construct simulation cells with interlayer structures similar to that expected
after equilibration. An examination of the distribution of water O along the z axis of the
hydrated Ca-montmorillonite interlayer, averaged for the six clay systems subjected to
the complete simulation cycle described in Section 5.2.2, indicates that a bilayer of water
has formed, as expected (Figure 5.2). However, a small peak in distribution near the
midplane of the interlayer suggests a third layer is beginning to develop under these
conditions.
12
10
6
4
2
0
3 4 0 1 2 2 1 4 3
z(A)
Figure 5.2. Average distribution of water O along the z axis, normal to the clay layer, for
the six hydrated Ca-montmorillonite systems (layer spacing 15.2 A) subjected to an EM -
annealing MD - EM - MD (T = 300 K) - EM cycle. The origin represents the midplane
of the interlayer.
Although the initial Ca-montmorillonite systems featured two levels of cation
hydration, six-coordinated and eight-coordinated, the levels of hydration present in the
320
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
fully minimized systems are similar. Radial distribution functions calculated for each
system (not shown) indicate that inner sphere coordination of water O occurs within
2.8 A of each Ca2+. Four of the six systems simulated contain two cations hydrated by
seven water molecules, and one cation hydrated by eight water molecules (CN = 7.33).
The system with the lowest potential energy contains two cations hydrated by eight water
molecules, and one cation hydrated by seven (CN = 7.67). A sixth system contains a
Ca2+that has formed an inner sphere complex with an oxygen triad attached to a
tetrahedral substitution site, such that three clay O and five water O are within its sphere
of hydration.
All other cations occupy positions near the midplane of the interlayer. These
counterions may be associated with individual charge sites, or may be located in
intermediate positions near a few different charge sites. Most cations are located between
oxygen triads found in each clay layer, or between an oxygen triad of one layer and a six-
oxygen cavity of the other. Positioning with respect to these surface sites is constrained
by the initial arrangement of the mineral surfaces, because fixed cell parameters
prevented adjustments to clay registration.
The hydrated Ca-montmorillonite configuration with the lowest potential energy
is displayed in Figure 5.3. The higher than average cation solvation within the system
with lowest potential energy, mentioned above, is the primary distinguishing feature of
this version of hydrated Ca-montmorillonite. One of the eight-coordinated cations is
located between oxygen triads located on opposing sides of the interlayer, while the other
two cations are located between an oxygen triad on one side and a mineral six-oxygen
321
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
cavity on the other. All cations are located midway between a tetrahedral and an
octahedral charge site.
The total potential energy of the dry Ca-montmorillonite system without rigid
atom energy contributions (Table 5.1) can be subtracted from the total potential energy of
the hydrated system (Table 5.2), then divided by the number of water molecules present
in the simulation cell (64), to provide the potential energy per mole of interlayer water
molecules (Table 2.1), -58.7 kJ mol'1. This value is lower than that provided in Table 4.2
for the COMPASS bulk water system, an indication of the appeal of the interlayer region
for water molecules as captured by the COMPASS force field.
Figure 5.3. Ca-montmorillonite hydrate system with lowest potential energy (layer
spacing 15.2 A). Elements are color-coded as described in Figure 5.1.
322
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Table 5.2. Energy Distribution of Ca-montmorillonite Hydrate
Energy Term Energy (kJ mol'1)
Total Potential -11072.5
Bond 102.914
Angle 139.454
Cross-term -23.7933
Van der Waals 507.214
Electrostatic -11798.3
The model IR spectrum of dry Ca-montmorillonite is nearly featureless
(Figure 5.4), and bears little resemblance to experimental spectra of smectite minerals
(Farmer, 1974). This is not surprising, as the COMPASS force field is not capable of
modeling a realistic montmorillonite structure unless the clay atoms are fixed in position
relative to one another, resulting in a system that cannot be considered highly minimized.
Because the dry Ca-montmorillonite spectrum is transparent over a broad frequency
range, it appears likely that model IR spectra of montmorillonite systems containing
water or organic molecules within the interlayer will consist primarily of signals
produced by interlayer species.
The similarity of the model IR spectrum of hydrated Ca-montmorillonite to that of
bulk water (Section 4.3.1) indicates that vibrations of the interlayer water molecules
provide essentially all of the signals displayed (Figure 5.4). The O-H stretching region of
interlayer water (3700 - 3200 cm'1) is broader than that of bulk water, suggesting the loss
of a more ordered tetrahedral structure upon adsorption to a mineral surface, as seen
experimentally (Farmer, 1974). Although the average minimized O-H bond length of
interlayer water molecules is 0.976 A ( 0.006 A), similar to both experimental and bulk
323
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
COMPASS water MD values (Table 4.1), the average minimized water H-O-H angle of
98.84 ( 2.25) is smaller than that found in the bulk water simulation (Section 4.3.1),
and much smaller than the experimentally observed liquid water angle of 105.4 (Table
4.1). This reduced molecular angle would inhibit the formation of tetrahedral structure,
as would the narrow space available in the interlayer region. The reduction in 3D
tetrahedral organization of water molecules may also explain the shift from a bimodal
distribution of O-H stretching signals under bulk water conditions to an asymmetric
unimodal distribution with adsorption.
e
o
1.0
0.8
0.6
0.4
0.2
0.0
- dry Ca-montmorillonite
- hydrated Ca-montmorillonite
4000 3500 3000 2500 2000 1500
wavenumber (cm'1)
1000 500
bulk water

u
o

0.4
0.2
0.0
1500 1000 500 0 3000 2500 2000 4000 3500
wavenumber (cm" )
Figure 5.4. Model IR spectra of dry and hydrated Ca-montmorillonite (top, no
normalization), as well as bulk water (bottom, Section 4.3.1).
324
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
In contrast, the distribution of O-H bending vibrations produced by mineral-
adsorbed water is distinctly bimodal, with peaks at 1685 and 1658 cm'1, while the bulk
water IR spectrum contains only a shoulder near the higher frequency and a peak near the
lower. Experimental IR spectra of Ca-montmorillonite hydrates feature a reduction in the
frequency of the O-H bending band upon adsorption to a mineral surface (Xu et al.,
2000), evidence of relatively few, strained H-bonds (Pimentel and McClellan, 1960),
often with mineral surface O as the acceptor (Yariv, 1992). Calculation of H-bonds
present in the fully equilibrated Ca-montmorillonite hydrate structure reveals the
existence of 24 water-mineral surface H-bonds, and just one water-water H-bond. When
normalized by the number of water molecules under examination, the clay hydrate
system contains far more H-bonds than the bulk water system (Section 4.3.1), and
therefore should feature a higher frequency O-H bending mode (Pimentel and McClellan,
1960), in this case represented by the growth of a shoulder into a dominant peak at
1685 cm'1.
5.3.2 The DOM-montmorillonite Systems
The potential energy distributions of the final conformations of the protonated and
Ca-saturated DOM-montmorillonite interlayer complexes are listed in Tables 5.3 and 5.4,
respectively. Both systems owe their low potential energies primarily to electrostatic
energy, with additional contributions from torsion energy found in the organic structures.
The increased importance of torsion to the distribution of potential energy in these
interlayer organic molecules is highlighted when comparing the nonperiodic DOM
energies in Tables 5.3 and 5.4 to those in Tables 4.4 and 4.6, respectively. The
325
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
conformations of the DOM molecules taken from within the interlayer region feature
lower potential energies than the conformations of the same molecules taken from dilute,
hydrated states, largely due to decreased torsion and angle energies. Though the
interlayer region presents physical constraints to movement along the z axis, it appears
that the lateral volume available to the flexible DOM molecules in this environment
allows them to adopt conformations with lower torsional and angular energies than is
possible under the fixed volume conditions of the hydrated simulations described in
Sections 4.3.3 and 4.3.4.
While a direct comparison of the distributions for the two complete organo-
mineral systems is impossible given significant differences in layer spacing and water
content, it is possible to compare the energy contained in direct interactions between the
organic molecule and the Ca-montmorillonite. By subtracting the individual potential
energies of the periodic DOM and Ca-montmorillonite molecules from the potential
energy of the combined DOM-mineral potential energy, a value o f - 1 119.5 kJ mol'1is
obtained for the protonated system, while a value of -5780.2 kJ mol'1is obtained for the
Ca-saturated system. Evidently, the Ca-saturated DOM interacts more strongly with the
mineral surface than does the protonated DOM, likely because the Ca-saturated system
features greater substantial electrostatic interactions. As for interlayer water, a lower
limit for the potential energy of these molecules can be calculated by subtracting the
periodic DOM energy and four times the dry Ca-montmorillonite energy calculated in
Section 5.3.1 from the total system energy for each organo-mineral complex, then
dividing the value obtained by the number of interlayer water molecules. Water
molecules of the protonated DOM-montmorillonite system have a minimum potential
326
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
energy o f -56.5 kJ mol'1, while water molecules of the Ca-saturated DOM-
montmorillonite system possess a minimum potential energy o f -74.1 kJ mol'1. Though
without data from dry DOM-montmorillonite simulations it is impossible to obtain an
exact potential energy for water molecules in these systems, the minimum values listed
above do indicate that the interlayer of the protonated DOM-montmorillonite is less
hydrophilic than the interlayer of hydrated Ca-montmorillonite.
Table 5.3. The Potential Energy Distributions of the Protonated DOM-montmorillonite
Interlayer Complex, its Individual Components, and their Combinations
Energy Term Total System
(kJ mol'1)*
DOM - Water
(kJ mol'1)*
DOM - Mineral
(kJ mol'1)*
Mineral - Water
(kJ mol'1)*
Total -7537E8 -33670.3 -34292.4 -52259.0
Bonds 1473.07 1473.07 702.732 770.337
Angles 3389.60 3389.60 2545.37 844.227
Torsions -7178.53 -7178.53 -7178.53
Inversions 50.2126 50.2126 50.2126
Cross-terms -1115.36 -1115.36 -952.621 -162.742
Van der Waals 3673.86 3152.27 375.506 4348.26
Electrostatic -75664.7 -33441.5 -29835.0 -58059.3
Energy Term Periodic DOM Nonperiodic DOM Water Ca-montmorillonite
(kJ mol'1)* (kJ mol"1)* (kJ mol'1)* (kJ mol'1)*
Total -15426.0 -15327.9 -11861.6 -17746.9
Bonds 702.732 702.732 770.337
Angles 2545.37 2545.37 844.227
Torsions -7178.53 -7178.53
Inversions 50.2126 50.2126
Cross-terms -952.621 -952.621 -162.742
Van der Waals 514.230 432.354 3377.83 310.109
Electrostatic -11107.4 -10927.4 -16691.3 -18057.0
* mole refers to the entire system being simulated, not to its components; i.e. kJ per mole system.
327
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
An inspection of the structure of the interlayer region of each of these organo-
mineral systems begins with an examination of the structure of the DOM molecules. The
protonated and deprotonated DOM molecules have molecular volumes of 8253.7 A3and
7918.4 A3, respectively. These volumes are similar to those measured for the same
molecules in hydrated states (Sections 4.3.3 and 4.3.4), and illustrate the loss in volume
caused by removal of protons. The solvent accessible surface area of the deprotonated
DOM molecule is 5558.4 A2, similar to that recorded for the hydrated form examined in
Section 4.3.4. However, the solvent accessible surface area for the protonated DOM
Table 5.4. The Potential Energy Distributions of the Ca-saturated DOM-montmorillonite
Interlayer Complex, its Individual Components, and their Combinations
Energy Term Total System
(kJ mol'1)*
DOM - Water
(kJ mol1)*
DOM - Mineral
(kJ mol'1)*
Mineral - Water
(kJ mol1)*
Total -169611 -124634 -97484.3 -55970.6
Bonds 2041.93 2041.93 647.102 1394.82
Angles 4186.93 4186.93 2647.00 1539.93
Torsions -7563.29 -7563.29 -7563.29
Inversions 99.9733 99.9733 99.9733
Cross-terms -1120.55 -1120.55 -825.675 -294.873
Van der Waals 13254.8 12463.09 5639.24 5282.43
Electrostatic -180511 -134742.0 -98128.6 -63892.6
Energy Term Periodic DOM Nonperiodic DOM Water Ca-montmorillonite
(kJ mol1)* (kJ mol1)* (kJ mol1)* (kJ mol"1)*
Total -77206.1 -73068.1 -16974.3 -14498.0
Bonds 647.102 647.102 1394.82
Angles 2647.00 2647.00 1539.93
Torsions -7563.29 -7563.29
Inversions 99.9733 99.9733
Cross-terms -825.675 -825.675 -294.873
Van der Waals 5392.76 5308.95 4544.41 192.786
Electrostatic -77604.0 -73381.9 -24158.6 -14690.8
* mole refers to the entire system being simulated, not to its components; i.e. kJ per mole system.
328
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
r-TvT-v /tvTv /TvT\ ^T^X-. /^vTs tvT\ /TvTS /-Ts^Tv xXv
>!r * j
' : /
'> i i- =
t
LA
\ i %i
\ 'Vv'.-V
" / I s' ",
~r~'
,
V/ A
IV. ,<<)
! o) t J ; Y : A " ^ sC:Y ~ <Y
I t v ' . ' \ , y - > - a ' ' 1 r \ y - v C ~ i - * r v \ *\ ? \ .
,,t ! ~y- ' * -
L.
k / - > ; ) > Va V A Y '> "r ^ '
~ t* S.-N . 4../ \A- V^ V *7 *
V ' V ,
' 1 Y Y Y ^ *<f iAX.% '
' V, ' A
*'r* -<
I YY* Y,
AH
: ? J . . ' A > - , / . ' V - . . J )
^ 4a y >*<J * Y- / - 1' * '
* a ) * a U - A ^ * -1 * ' x / ,
r C r ^ ) a a a v a '
Y ; H V# *Y Y y ^ Y v
" - ^ Y a " V f e ' " / ' Y
<A M J" v '<v ' .
* v ' J '
Figure 5.5. Views of protonated (left) and Ca-saturated (right) DOM molecules within
interlayers of hydrated Ca-montmorillonite (waters removed from images) in the XZ
(top), YZ (middle), and XY (bottom) planes. Grey denotes C, red denotes O, green
denotes H (for visibility), N and S are blue and yellow, respectively, and Ca2+ is brown.
The clay layer is color-coded as described in Figure 5.1. The clay layer and associated
Ca2+were removed from the XY plane view to improve visibility of organic structures.
329
Re pr oduc e d with per mi ssi on of t he copyright owner. Fur t her reproduct i on prohibited without per mi ssi on.
molecule is 4951.9 A2, 12 % greater than the value for the hydrated version (Section
4.3.3). Apparently, the angular and torsional changes that led to an interlayer DOM
molecule with a lower potential energy also resulted in the formation of a more diffuse
organic structure with more external surface available for interaction with water and
mineral components. Figure 5.5 contains views of the organic structures along each of
three perspectives. Although the layer spacing of the protonated system is smaller, the
protonated DOM molecule is larger than the corresponding Ca-saturated molecule in the
z dimension, to maximize contact with the mineral surface. The Ca-saturated DOM
configuration, on the other hand, shows extensive spreading in the x and y directions and
diminished contact with the mineral surface. Its structure is quite porous, in part because
carboxylate groups attached to different parts of the DOM molecule can become attracted
to the same Ca2+, creating large loops of organic material that define available pores. The
Ca-saturated DOM hydrate described in Section 4.3.4 also featured extensive porosity.
Differences in layer spacing and hydration of the two systems illustrate the effect
of the protonation state of interlayer organic species on the surrounding interlayer
environment. The density of the fully hydrated interlayer region of the protonated DOM-
montmorillonite complex, with a layer spacing of 30.53 A and a total of 543 water
molecules, is 1.06 g cm3. On the other hand, the deprotonated, cation-saturated DOM-
montmorillonite complex, with a layer spacing of 33.32 A and an interlayer featuring
852 water molecules, has an interlayer density of 1.26 g cm'3. The deprotonated, Ca-
saturated humic polyanion is also extremely polar, and facilitates the introduction of
more water molecules through numerous DOM-water and cation-water interactions. The
negative charge of the organic molecule must be screened from the negative charge of the
330
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
clay layer using an intermediate layer of hydrated Ca2+. These Ca2+cushions contribute
to the larger layer spacing of the Ca-saturated system, which in turn provides a larger
volume available for hydration. The protonated DOM molecule, with its neutral charge
and reduced polarity, does not require extensive charge screening by cations, which
allows the clay layers to move closer together. However, the hydrophobic nature of the
protonated humic molecule reduces the amount of water that can fit into the interlayer
region. It is not surprising, given both the lower water content and the less polar
environment, that the protonated DOM-montmorillonite features fewer H-bonding
interactions for all types of H-bond pairs, with one notable exception (Table 5.5).
Table 5.5. Hydrogen Bonding within the DOM-montmorillonite Interlayer Complexes
H-bonds Protonated DOM Ca-saturated DOM
intramolecular
DOM 37 43
intermolecular
DOM-DOM 0 0
DOM-water 60 152
DOM-mineral 6 0
Water-mineral 84 97
Water-water 17 33
The final configuration of the protonated DOM-montmorillonite simulation
system features multiple direct organo-mineral H-bonds, while the Ca-saturated system
contains none. Without a dense layer of charge screening hydrated counterions, flexible
portions of the protonated DOM molecule move to positions near the mineral surface.
There, organic hydroxyl protons can form H-bonds with mineral surface O atoms (Figure
331
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
5.6). Hydrogen bonding occurs preferentially near octahedral charge sites abandoned by
interlayer cations. The Ca-saturated DOM molecule, excluded from positions near the
2"b
mineral surface due to its negative charge and associated charge screening Ca , takes
part in a less direct organo-mineral interaction mediated by H-bonds. Several water
molecules form H-bonds with organic and mineral constituents simultaneously, creating a
series of indirect H-bonded structures in this system (Figure 5.6). The protonated system
features only two such water mediated H-bonding interactions.
Figure 5.6. The illustration to the left is an example of a direct organo-mineral H-bond
taken from the protonated system ( 0 -0 distance 2.58 A, O-H-O angle 149.3), while the
illustration to the right is an example of an indirect organo-mineral H-bonded structure
taken from the Ca-saturated system ( 0-0 distances 2.53 A and 2.68 A, O-H-O angles
161.4 and 157.3 for upper and lower H-bonds, respectively). Atoms are color-coded as
described in Figure 5.5, except that H is now white instead of green.
The close proximity of organic and mineral components in the protonated system
creates conditions conducive to the formation of two other types of organo-mineral
interactions. Hydrophobic interactions take place between organic moieties, typically
substituted alkyl chains, located close to mineral surfaces (Figure 5.7). The C backbones
332
per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited without per mi ssi on.
of the organic components are usually around 4 A from the mineral surface, and water
molecules are excluded between the DOM molecule and the Ca-montmorillonite,
reducing the number of water-mineral H-bonds that can form in the protonated system.
Few regions of the Ca-saturated DOM molecule reside close enough to the clay surface to
be considered part of hydrophobic interactions.
Also evident in the protonated system is an organo-mineral interaction based on
electrostatic attraction of the negative portions of polar molecules for the six-oxygen
surface cavities covering the montmorillonite surface. Mineral hydroxyl groups lie near
the center of these cavities, and a few water molecules in the organo-mineral systems
orient such that their oxygen atoms point directly into these surface features, drawn
toward the inner mineral proton. Some organic functional groups can be seen mimicking
this effect, positioning organic 0 atoms just 2 or 3 A away from six-oxygen cavities,
while organic H atoms are placed such that the O-H bond is nearly parallel with the
mineral surface (Figure 5.7). This type of interaction is absent from the Ca-saturated
system.
The hydrophobic environment created by the protonated DOM molecule closely
surrounded by mineral surfaces may influence the formation of Ca2+-mineral inner sphere
complexes. Six of these complexes are observed in the protonated system, while four can
be identified in the Ca-saturated system (Figure 5.5). The greater number of these
complexes is a factor in the lower potential energy of the clay component of the
protonated system, as compared to the Ca-saturated one (Tables 5.3 and 5.4). The
contribution of this inner sphere mineral O to the Ca-0 CN can be seen in Table 5.6;
greater mineral contact and fewer Ca2+overall explain the larger value for the protonated
333
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Figure 5.7. A portion of the protonated DOM-montmorillonite system displaying a
typical hydrophobic organo-mineral interaction as well as the attraction of an organic O
to a six-oxygen cavity. Atoms are color-coded as described in Figure 5.6.
DOM-montmorillonite system. The Ca2+within the protonated complex collect more
water molecules in their solvation shell, and have a total Ca-0 CN higher than found in
the Ca-saturated complex. The lack of organic atoms within the coordination spheres of
Ca2+near the protonated DOM molecule provides a likely explanation for this
observation, as organic structures attached to coordinating organic O can block water
Table 5.6. Cation Coordination Numbers for Protonated and Ca-saturated Systems,
Using a Maximum Radius of 2.75 A
Ca-0 Pairs Protonated DOM Ca-saturated DOM
Total 0 7.58 6.91
Water 6.08 4.02
Mineral 1.50 0.22
Organic 0 2.67
Carboxylate 2.45
Other Organic 0.22
334
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
molecules from fully hydrating the cation. In the Ca-saturated system, numerous inner
sphere Ca-carboxylate complexes form, along with a few inner sphere complexes
between Ca2+ and other organic functional groups. The Ca-carboxylate complexes
typically feature smaller Ca-0 distances than found in hydrated Ca2+, as can be seen in
the Ca-0 RDFs for each system (Figures 5.8 and 5.9). The Ca-total O RDFs presented
for each of these systems feature peak heights smaller than those for Ca-water O, and in
Figure 5.9 for Ca-carboxylate O. Because this function is normalized by the number of O
atoms relevant to each plot (Equation 2.2), the greater number and broader distribution of
Ca-total O radial distances results in a less extreme RDF structure than seen from some
specific O subgroups.
50
45
water
40
35
30
25
total 0
20
15
10
organic
mineral
5
0
5.5 6 4 4.5 5 2.5 3 3.5 1.5 2 1
r(A)
Figure 5.8. Ca-O RDFs for the protonated DOM-montmorillonite system.
An examination of the local environment around each Ca2+ reveals some
interesting trends. Within the protonated system, Ca2+prefer positions near tetrahedral
335
per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
charge sites and above surface O triads. The cations remain in locations fairly close to
the mineral surface (Figure 5.5), some forming water bridges involving outer sphere
coordination to organic O functional groups and inner or outer sphere coordination to the
25
20 i-
water
15
oc
10
total Q
5
rmrieral
0
5.5 6 3 3.5 4 4.5 5 1 1.5 2 2.5
r(A)
BO
carboxyl ate
70
60
total organic
50
40
30
20
Other organic O
10
0
5.5 6 4.5 5 1.5 2 2.5 3 3.5 4 1
r(A)
Figure 5.9. Ca-0 RDFs for the Ca-saturated DOM-montmorillonite system.
336
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
mineral surface (Figure 5.10). Within the Ca-saturated system, a more complex structure
has evolved. Cations in this interlayer region also prefer positions near surface O triads.
The Ca2+ complexed to carboxylate groups show a preference for octahedral charge sites,
while the Ca2+ not involved in organic inner sphere complexes favor placement near
tetrahedral charge sites. Some cations originally located near mineral charge sites have
moved away from this surface and into the DOM molecule, providing evidence of cation
exchange reactions. Typical cation bridges in this system involve a cation directly
coordinated to carboxylate groups and outer sphere coordinated to the mineral surface
(Figure 5.10). Two water bridges involving Ca2+ outer sphere coordinated to organic
functional groups and inner sphere coordinated to surface O triads are present as well.
Figure 5.10. A water bridge within the protonated DOM-montmorillonite system (left),
featuring outer sphere complexation between Ca2+ and a carboxyl O atom marked by a
black line, and inner sphere coordination with the mineral surface, and a cation bridge
within the Ca-saturated DOM-montmorillonite system (right), featuring inner sphere
complexation between Ca2+ and carboxylate groups marked by black lines, and outer
sphere coordination to the mineral surface. Atoms are color-coded as described in
Figure 5.6.
337
with per mi ssi on of t he copyri ght owne r . Fur t her reproduct i on prohibited without per mi ssi on
Model IR spectra provide a means of assessing the overall effects of protonation
state on a DOM molecule within a montmorillonite interlayer. Though constraints of the
Cerius2IR-Raman algorithm prevented inclusion of the mineral layer in the structures
used to calculate these spectra, the featureless spectrum established for dry Ca-
montmorillonite (Figure 5.4) indicates its spectral contribution may be limited. The
protonated DOM-clay (dry)
c
3
<U
o

o.o
1000 500 0 2000 1500 3000 2500 4000 3500
wavenumber (cm'1)

x>
Lh
o
1.0
0.8
0.6
0.4
0.2
0.0
' iiWi IM
\ I .
A # A i H a - V ' i Aa. a
DOM alone
DOM hydrate (dry)
4000 3500 3000 2500 2000 1500
wavenumber (cm-1)
1000 500
Figure 5.11. Model IR spectra of the protonated, periodic DOM molecule extracted from
the interlayer of a Ca-montmorillonite hydrate (top), as well as the dry, densely packed
DOM molecule (Section 4.3.2), and the DOM hydrate after water molecules were
removed from the structure (Section 4.3.5) (bottom). The absorbance values for all
spectra were normalized via the absorbance of the nitrile peak of the dry, densely packed
DOM molecule.
338
per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
model IR spectrum calculated for the periodic, protonated DOM molecule (Figure 5.11),
without associated water and mineral molecules, possesses features similar to those
calculated for the dry, densely packed DOM molecule (Section 4.3.2). The only
significant difference seen is a shift in the O-H stretching bands to lower frequencies,
perhaps an indication of fewer Fl-bonds or the reduced potential energy of the interlayer
e
3
o
o
G
-P
1.0 -1
0.8
0.6
0.4
0. 2 -
0.0
Ca-DOM-clay (dry)
4000 3500 3000 2500 2000 1500
wavenumber (cm"1)
1000 500
8
1.0
0.8 -
0.6
0.4
0.2
0.0
DOM alone
Ca-DOM (dry)
i' W'1:
b'
/ '\r \
4000 3500 3000 2500 2000 1500
wavenumber (cm"1)
1000 500
Figure 5.12. Model IR spectra of the periodic Ca-saturated DOM molecule extracted
from the interlayer of a Ca-montmorillonite hydrate (top), as well as the dry, densely
packed DOM molecule (Section 4.3.2), and the Ca-DOM hydrate after water molecules
were removed from the structure (Section 4.3.5) (bottom). The absorbance values for all
spectra were normalized via the absorbance of the nitrile peak of the dry, densely packed
DOM molecule.
339
per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
organic structure as compared to the dry structure (Section 4.3.2). The model IR
spectrum of the Ca-saturated DOM molecule without associated mineral and water
molecules (Figure 5.12) also shows many similarities to the dry, densely packed DOM
molecule. The dramatic reduction in O-H stretching and bending vibrations is a direct
result of deprotonation of carboxyl groups. Peaks near 1600 and 1400 cm'1may be
linked to vibrations of carboxylate functional groups. The spectra of the interlayer
molecules show little resemblance to the spectra of the DOM hydrates (Section 4.3.5), an
indication that dehydration of the nonperiodic hydrated DOM structures prior to
computation of the latter spectra may have created poorly minimized systems that led to
flawed IR calculations.
It was possible to produce a model IR spectrum for a periodic interlayer Ca-DOM
molecule accompanied by the water molecules originally associated with each Ca2+
(Figure 5.13). This spectrum shows essentially no similarity to that of the nonperiodic,
partially hydrated Ca-DOM complex described in Section 4.3.5. However, the IR
spectrum of the interlayer complex does resemble that of bulk water, indicating that the
signal of the water molecules can hide that of the organic material almost completely.
The O-H stretching and bending modes of the interlayer complex have shifted to higher
frequencies, an indication of greater H-bonding within the interlayer region. Because the
average structures of water molecules are statistically identical in both Ca-saturated and
deprotonated systems, with O-H bond lengths of 0.974 0.005 and 0.976 0.005 A and
H-O-H bond angles of 99.27 1.53 and 99.61 1.40, respectively, it is possible that a
model IR spectrum of the protonated interlayer DOM molecule in a hydrated state would
be quite similar to that of the Ca-saturated interlayer DOM molecule.
340
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
1.0
0.8
.5

0.6 -
i
x>
t- 0.4
u
C/5

0.2
0.0
4000
Ca-DOM-clay (with water)
3500 3000 2500 2000 1500
wavenumber (cm-1)
1000 500
Ca-DOM (with water)
water
e
S 06
o

e
0.0
500 0 1500 1000 3000 2500 2000 4000 3500
wavenumber ( c m 1)
Figure 5.13. Model IR spectra of the periodic Ca-saturated DOM molecule extracted
from the interlayer of a Ca-montmorillonite hydrate and including water molecules
associated with Ca2+ hydrates (top), as well as the Ca-DOM complex including water
molecules associated with Ca2+hydrates (Section 4.3.5), and bulk water (Section 4.3.1)
(bottom). These spectra are not normalized.
5.4 Discussion
5.4.1 Ca-montmorillonite and the COMPASS Force Field
The work presented in this chapter may be considered an exploration of the
interactions of an interlayer-adsorbed model humic substance with surrounding water
molecules and mineral surfaces. Before this investigation can begin, it is essential to
ascertain whether the COMPASS force field can be used to model an interlayer structure
341
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
for hydrated Ca-montmorillonite that is consistent with experimental data. If simulations
using COMPASS result in an agreeable interlayer structure, then the force field can be
considered acceptable for further simulation of hydrated organo-smectite complexes.
The COMPASS force field was not designed to model the structure of
montmorillonite. Because COMPASS lacks parameters to describe octahedrally
coordinated Al, the positions of all clay atoms must be fixed relative to each other during
all simulations. This deficiency additionally impacts the distribution of electrostatic
charge throughout the mineral layer. For this reason, the electrostatic charges established
by Skipper et al. (1995a, b) for a model montmorillonite were used here. Though this
particular charge distribution has been used successfully for simulations involving
multiple water models (Skipper et al., 1995a, b; Chang et al., 1995, 1997,1998,1999;
Greathouse and Sposito, 1999; Chavez-Paez et al., 2001), it has never been combined
with the COMPASS force field. Caution must be exercised when altering parameters of
a complex force field like COMPASS, as such changes may affect the accuracy with
which the force field reproduces molecular interactions.
The dry Ca-montmorillonite system described in Section 5.3.1 features an
equilibrium layer spacing of 10.26 A, larger than observed experimentally (Hendricks et
al., 1940; Mooney et al., 1952; Cases et al., 1997; Chiou and Rutherford, 1997; Dios
Cancela et al., 1997). The adsorption of many Ca2+to oxygen triads, and the midplanar
location of other cations hovering near six-oxygen cavities, likely prevents full collapse
of the interlayer. Perhaps a modeling method utilizing extreme conditions, for example
the annealing MD regime used by Smith (1998) to produce a collapsed Cs-
montmorillonite that included simulation at 2000 K (Section 2.1), would force
342
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
counterions to sink within six-oxygen cavities, and facilitate a reduction in layer spacing.
In fact, many simulations of dry smectite systems experiencing less extreme conditions
result in layer spacings of 10 - 11 A (Chang et al., 1998; Greathouse and Storm, 2001;
Sutton and Sposito, 2001; Hensen and Smit, 2002), so the deviation between model and
experimental layer spacing seen for the dry Ca-montmorillonite system is not unusual.
With hydration, we can observe the interaction of COMPASS water molecules
with the clay mineral surface. The attraction of water molecules to these surfaces,
evidenced by the numerous water-surface O H-bonds that form, creates a layered water
structure consistent with modeling (Chavez-Paez et al., 2001; Greathouse and Storm,
2002) and experimental work (Slade et al., 1985; Cases et al., 1997). A comparison of
the total number of H-bonds observed within the Ca-montmorillonite hydrate and the
bulk water structure prior to final minimization (Section 4.3.1) reveals that the clay
system contains far more of these interactions relative to the number of water molecules
present in each system. Experimental IR analysis of Ca-smectite samples usually
indicates a decrease in H-bonds (Xu et al., 2000), while the model spectrum generated by
Cerius2provides further evidence for an increase in these bonds. The proportional
increase in H-bonds found within the hydrated mineral is a direct result of reduced H-
bonding present in the COMPASS bulk water system, due to a reduction in tetrahedral
structure caused by the small average H-O-H angle for COMPASS water molecules. The
number of water-clay H-bonds that can form in a Ca-montmorillonite hydrate modeled
with COMPASS are unaffected by the small molecular angle of water, as these H-bonds
do not require the formation of a 3D tetrahedral structure. Water-water H-bonds are
further inhibited within the clay hydrate due to the constrained geometry of the interlayer
343
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her reproduct i on prohibited without per mi ssi on.
region. Although the H-bonding behavior of COMPASS water molecules is not
2
consistent with experimentally observed trends, it is interesting to note that the Cerius
IR-Raman module is sufficiently sophisticated to provide a spectrum that reflects the
increase in H-bonds observed within the model Ca-montmorillonite hydrate.
Initial simulations of the Ca-montmorillonite hydrate that allowed fluctuations in
cell dimensions resulted in systems with layer spacings of 14.7 A. Although smaller than
observed experimentally in typical two layer Ca-montmorillonites (Hendricks et al.,
1940; Mooney et al., 1952; Cases et al., 1997; Chiou and Rutherford, 1997; Dios Cancela
et al., 1997), similar layer spacings were observed in model Ca-montmorillonite systems
hydrated with TIP4P water (Chavez-Paez et al., 2001; Greathouse and Storm, 2002).
COMPASS systems with fixed 15.2 A layer spacings often featured a small concentration
of water molecules at the midplane, easily differentiated from the water layers directly
adsorbed to the mineral surface (Figure 5.2). Though a relatively minor feature, the
presence of this emerging midplanar water layer, along with the preference of the
hydrated Ca-montmorillonite systems for layer spacings < 15 A (Section 5.2.2), indicates
that the majority of COMPASS water molecules occupy positions close to the mineral
surface, creating just enough room in the 15.2 A systems to allow the remaining water
molecules to adopt positions near the midplane. The electrostatic charge of-0.8e
associated with mineral surface O, nearly identical to the electrostatic charge of-0.84e
associated with water O, may draw COMPASS water molecules especially close to the
clay layers.
The Ca2+in each of the fully minimized montmorillonite hydrates are coordinated
by 7 or 8 O atoms. Experimental studies of Ca2+ hydration within mineral interlayers
344
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
suggest that the cation is surrounded by six or eight water molecules. Hall (1982)
provides neutron scattering data suggesting that Ca2+within montmorillonite interlayers
is six-coordinated. Skipper et al. (1994) detect only six-coordinated Ca2+within the
interlayer of the more highly charged vermiculite via neutron diffraction, while Slade et
al. (1985) present XRD evidence for the presence of both six- and eight-coordinated Ca2+
within Ca-vermiculite, and calculate an average Ca-0 CN of 7.79. The Ca-0 CNs
calculated from the model hydrates described here cannot be considered to support the
existence of more highly coordinated Ca2+ within natural smectites, as a careful review
by Greathouse and Storm (2002) of simulations of this cation in bulk solution reveals that
hydration structure is highly dependent on the Ca-0 potential function employed. For
their simulations of Ca-montmorillonite, Greathouse and Storm (2002) used a Ca-O
potential designed to produce an 8-coordinated structure in bulk solution, and
consequently observed the same CN within the interlayer. Thus, the tendency of
COMPASS Ca2+to draw 7 to 8 O atoms into its solvation shell merely reflects the
preference for greater hydration of Ca2+built into COMPASS Ca-O parameters. The Ca-
montmorillonite hydrate system with the lowest potential energy features more cations
coordinated by 8 water molecules, further evidence that COMPASS Ca2+ tends toward an
8-coordinated state.
The unlikely inner sphere Ca2+complex that forms in one of the model Ca-
montmorillonite hydrates is also 8-coordinated, as it is adsorbed to 3 surface O and
hydrated by 5 water molecules. The appearance of this inner sphere complex in a single
simulation cell indicates that it is possible for physically unrealistic interactions to occur
when using the COMPASS force field and the simulation regime described in Section
345
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
5.2.2. This inner sphere adsorption may be facilitated by the electrostatic charge
associated with surface O atoms, a charge nearly identical to that of COMPASS water O
atoms.
A detailed inspection of the simulation results for COMPASS Ca-montmorillonite
hydrates indicates that the force field adequately captures all major structural features
observed in experimental Ca-smectite samples. The majority of water molecules are
found in two layers, one adsorbed to each mineral surface. Water-mineral O H-bonds are
abundant, while water-water H-bonds have been reduced due to the inhibition of
tetrahedral structure within the constrained geometry of the interlayer region. Ca2+prefer
midplanar positions, and are hydrated by up to 8 water molecules. Both the formation of
an inner sphere Ca2+complex, and the adsorption o f interlayer water to positions
sufficiently close to the mineral layer to allow the emergence of a third, midplanar water
layer, suggest that the Skipper et al. (1995a, b) electrostatic charges associated with
mineral surface O may be a little too attractive to positively charged interlayer atoms.
The COMPASS force field produces acceptable Ca-montmorillonite hydrate structures
when given appropriate initial conditions, and therefore can be considered a useful tool
for the study of organo-clay complexes.
5.4.2 Hydrophobic, Hydrogen Bonding, and Water Bridging Organo-mineral
Interactions within Protonated DOM-montmorillonite
The protonated DOM molecule adsorbs weakly to Ca-montmorillonite through
hydrophobic, H-bonding, and water bridging interactions. Hydrophobic interactions can
be observed through a visual inspection of the organic molecule, which reveals that
346
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
several primarily alkyl functional groups have aligned themselves close to and parallel
with the clay mineral surface, such that water molecules are excluded from the vicinity.
This rearrangement of the organic molecule results in an expanded DOM surface area,
while the elimination of water molecules from some surface microsites reduces the
number of water-mineral H-bonds possible in this system. The interlayer of the
protonated organo-mineral system can be considered more hydrophobic than the
interlayer of hydrated Ca-montmorillonite, because the minimum potential energy of
water molecules in the former is higher than the calculated potential energy of water
molecules in the latter, indicating that the presence of neutral organic material has made
the region less appealing to water molecules. Few organic moieties within the Ca-
saturated DOM-montmorillonite system are positioned close enough to the clay layer to
be considered part of a hydrophobic interaction.
Simulations of a variety of organic materials near smectite and other siloxane
surfaces under dry conditions frequently suggest that organic molecules prefer to
maximize contact with the mineral, promoting hydrophobic interaction (Keldsen et al.,
1994; Shevchenko and Bailey, 1998; Teppen et al., 1998,2002; Haberhauer et al., 2001;
Yu et al., 2000a, b, 2001, 2003). With hydration, many small, nonpolar organic
molecules partially or fully desorb from the mineral surface, in response to both
competition from water molecules for mineral surface sites, as well as the availability of
a hydrophilic environment attractive to these organic materials (Teppen et al., 1998,
2002; Yu et al., 2003). However, hydration does not hinder the formation of hydrophobic
organo-mineral interactions involving molecules with great reduced polarity, such as
methane (Titiloye and Skipper, 2000; Park and Sposito, 2003) and carbazole (van Duin
347
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
and Larter, 2001). Evidently the hydrophobicity of alkyl chains near the mineral surfaces
within the protonated DOM-montmorillonite system, as described by the COMPASS
force field, is sufficient to allow the formation and maintenance of hydrophobic organo-
mineral interactions despite hydration.
Adsorption and desorption studies on smectites indicate that hydrophobic
microsites far from clay charge sites are capable of strongly adsorbing small organic
molecules (Laird et al., 1992; Barriuso et al., 1994; Ghosh and Keinath, 1994; Laird,
1996; Celis et al., 1997; Laird and Fleming, 1999) and larger organic polymers (Theng,
1979,1982) through hydrophobic interactions with interlayer and external surfaces.
When exposed to solutions of natural organic matter, smectite minerals selectively adsorb
large amounts of alkyl C functional groups (Chorover and Amistadi, 2001), perhaps in
part through hydrophobic mechanisms. Studies of natural soils indicate that alkyl C is
preferentially preserved, especially in clay size fractions, which are often enriched in
smectite minerals (Oades et al., 1987; Oades, 1988; Baldock et al., 1992; Ristori et al.,
1992; Leinweber et al., 1999; Mahieu et al., 1999; Baldock and Skjemstad, 2000; Chen
and Chiu, 2003). Many 13C CP/MAS NMR spectra of smectitic soils are dominated by
alkyl signals (Wattel-Koekkoek et al., 2001), and radiocarbon dating of such soils reveals
that the C is more recalcitrant than any found in kaolinitic soils (Wattel-Koekkoek et al.,
2003). An examination of the chemical composition of the few natural organo-smectite
interlayer complexes recovered from soils indicates that the organic material retained in
these interlayers is primarily alkyl (Theng et al., 1986; Schnitzer et al., 1988; Schulten et
al., 1996) or alkylaromatic (Righi et al., 1995). This collection of laboratory and field
348
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
evidence suggests that hydrophobic interactions may play an important role in
sequestration of organic matter in soils.
Another interaction observed within the protonated DOM-montmorillonite system
is that of a H-bond between the organic proton of an alcohol or carboxyl functional group
and an O atom from the mineral surface. Numerous organo-mineral simulations, ranging
from atomistic EM and MD calculations of montmorillonite and model silicate minerals
with amino acids (Newman et al., 2002), proteins (Yu et al., 2000b), and model humic
substances similar to the DOM molecule (Schulten and Leinweber, 1996, 2000; Schulten
and Schnitzer, 1997), to quantum chemical analyses of the interactions of tiny fragments
of aluminosilicates with simple organic molecules such as acetic acid (Kubicki et al.,
1997; Haberhauer et al., 2001; Tunega et al., 2002), also feature such H-bonds.
Experimental evidence for the existence of weak organo-mineral H-bonds frequently
takes the form of IR spectra that indicate shifts in key vibrational frequencies of organic
molecules upon introduction into the mineral matrix (Theng, 1979; Mortland, 1986;
Ristori et al., 1992). The IR spectrum of the protonated DOM molecule extracted from
the montmorillonite interlayer (Figure 5.11) illustrates that O-H stretching vibrations
have shifted to lower frequencies relative to the dry DOM molecule, an indication of the
presence of longer and more strained hydroxyl groups involved in H-bonds (Pimentel and
McClellan, 1960). However, as with experimental spectra, it is difficult to determine the
extent to which this modification is due to interactions with the mineral rather than with
surrounding water molecules. The IR spectrum of the hydrated DOM molecular system
described in Section 4.2.4 is so different from those calculated for the dry and interlayer
DOM molecules as to be useless in distinguishing the effects of organic H-bonds with
349
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
mineral and water components. It appears likely that the IR spectra calculated for the
hydrated systems described in Sections 4.2.4 and 4.2.5 are flawed.
Another organo-mineral interaction observed within this system, involving the
attraction of organic O atoms for positions near six-oxygen surface cavities, might have
resulted in the formation of more H-bonds, if the mineral hydroxyl groups located within
these cavities were allowed to move. Hydroxyl groups within a flexible montmorillonite
structure could respond to the presence of a nearby organic O atom by adopting a
configuration more perpendicular to the mineral surface, possibly allowing the interior
proton to form a H-bond with the organic functional group. While most water molecules
near the mineral surface are oriented such that a single proton is able to form a H-bond
with mineral surface O atoms, as is typical of montmorillonite interlayers (Section 2.4.2),
a few in each simulation can be seen in orientations in which the water O is pointed
directly toward a six-oxygen cavity (Figure 5.14), more typical of the interlayer of a
hectorite clay mineral, featuring hydroxyls with orientations perpendicular to the mineral
surface (Section 2.4.2). Though direct interaction of organic and water O atoms with
hydroxyl groups in the clay interior is an extremely minor component of the protonated
DOM-montmorillonite system described here, the use of a flexible mineral might expand
the role of this type of surface contact.
Inner sphere complexes between cations and polar organic functional groups, or
cation bridges, were not observed in the protonated system. Only outer sphere
complexes, or water bridges, formed between the Ca2+hovering near the mineral surface
and organic moieties including carboxyl and alcohol groups. Direct cation bridging ion-
dipole interactions are observed in many model organo-smectite interlayers under
350
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
5? i f ' V # /-.
>Y h. ^ 4
u L & ^ i a ;
Figure 5.14. Views of the protonated (top) and Ca-saturated (bottom) DOM-
montmorillonite systems. Water molecules are represented using cylinders or thin sticks
to reduce visual clutter, while DOM and clay molecules are represented in the ball and
stick style. The O atoms of a few water molecules near mineral surfaces are directed
toward surface cavities. Atoms are color-coded by element as described in Figure 5.5,
except that H is now white instead of green.
351
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
dehydrated conditions (Hackett et al., 2000; Pintore et al., 2001; Pospisil et al., 2002;
Boulet et al., 2003; Gaudel-Siri et al., 2003; Kuppa and Manias, 2003), but these
interactions are rarely preserved upon hydration because water molecules replace the
polar organic groups occupying positions near cations, creating water bridges (Bujdak et
al., 2000; Hackett et al., 2000; Boulet et al., 2003). While some experimental work
supports the importance of direct ion-dipole interactions as a mechanism for adsorption
of organic materials to minerals (Dios Cancela et al., 1996, 2000), much of this work is
conducted under artificially dehydrated conditions. In an environment that includes
water, ions with large hydration energies like Ca2+ are typically fully hydrated, and
therefore must form water bridges with polar organic functional groups (Theng, 1979).
Early definitions of the water bridging interaction specified that the water
molecule hydrating a cation must take part simultaneously in a H-bond with an organic
functional group (Theng, 1979; Mortland, 1986). Although several outer sphere Ca-
DOM complexes can be identified in the protonated DOM-montmorillonite system, the
lack of H-bonds between hydrating water molecules and organic functional groups means
that this type of water bridge is absent. Only two water bridging interactions including
such H-bonds are observed in the Ca-saturated system. Apparently, Ca2+exerts such a
strong influence over the position and orientation of hydrating water molecules within
these simulations that their ability to form secondary H-bonds with organic molecules is
hampered. Experimental data including the removal of significant levels of clay-
adsorbed organic molecules through washes with water (Theng, 1974,1979; Theng and
Scharpenseel, 1975; Mirabella et al., 1996), the trend observed in montmorillonite
suspensions of increasing organic adsorption with increasing cation charge to radius ratio
352
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
(Theng and Scharpenseel, 1975), and distinctive shifts, or lack thereof, in IR signals
(Theng, 1974,1979; Bosetto et al., 1994), are interpreted to be evidence of adsorption
through water bridges. None of these experimental observations provides compelling
evidence that the water molecule between cation and organic is H-bonded to the organic
functional group. Based on the modeling data presented here, it appears that a water
bridge structure that includes a H-bond is not common.
Some of the Ca2+in the protonated DOM-montmorillonite system are not fully
hydrated because they have formed inner sphere complexes not with organic groups, but
with the mineral surface. Inner sphere Ca-montmorillonite complexes are seen in the Ca-
saturated DOM-montmorillonite system as well, although to a lesser extent. A single
such complex was produced during simulations of hydrated Ca-montmorillonite,
consistent with the suggestion that the partial charge on the montmorillonite O atoms in
these simulations is excessive. Such inner sphere Ca-montmorillonite complexes would
not be predicted based on the large hydration energy of Ca2+, and have not been seen in
previous studies on Ca-montmorillonite hydrates conducted using both modeling
(Chavez-Paez et al., 2001; Greathouse and Storm, 2002) and experimental (Hall, 1982;
Slade et al., 1985; Cases et al., 1997) methods. These unusual complexes form during the
annealing stage of these simulations, an indication that the use of high temperature to
hasten the location of low potential energy configurations resulted in less stable cation
hydrate structures inappropriate to clay hydrates at 300 K.
I f the annealing procedure employed here encourages the formation of unrealistic
inner sphere complexes at the mineral surface, it may produce an excessive number of
inner sphere complexes with organic functional groups as well. While the exact number
353
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
of inner sphere complexes observed in the DOM systems described in Sections 4.2.4 and
5.2.2 may be elevated due to this simulation artifact, the chemical trends observed remain
valid. Similarly, the proportional increase in prevalence of inner sphere Ca-
montmorillonite complexes in the organo-montmorillonite systems relative to the
montmorillonite hydrate, even after identical annealing time periods, implies that the
presence of organic matter in the interlayer may disrupt sufficiently the structure of
interlayer water molecules to weaken the level of hydration seen around counterions.
That the protonated system contains more of these complexes than the Ca-saturated
system may reflect the reduced water or increased hydrophobicity of the former.
5.4.3 Indirect Hydrogen Bonding and Cation Bridging Organo-mineral Interactions
within Ca-saturated DOM-montmorillonite
The presence of a charge screening layer of hydrated cations between organic and
mineral components of the Ca-saturated DOM-montmorillonite system, similar to that
observed experimentally between the natural organic matter adsorbed to a mica surface
using surface x-ray reflectivity measurements (Nagy et al., 2001), prevents the formation
of many direct organo-mineral interactions seen in the protonated systems. Additionally,
the deprotonation of carboxyl groups in the Ca-saturated system reduces the number of
organic protons capable of forming H-bonds directly with the montmorillonite surface.
As a result, no DOM-montmorillonite H-bonds are observed in the Ca-saturated system
(Table 5.5). However, the presence of numerous carboxylate groups facilitates an
indirect form of H-bonding with the mineral surface, as now water molecules can
participate in H-bonds with both a carboxylate O atom and a surface mineral O atom
354
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
simultaneously. Perhaps this increase in H-bonding opportunities accounts for the shifts
in O-H stretching and bending frequencies in the partially hydrated, Ca-saturated DOM-
montmorillonite IR spectrum, when compared to that of bulk water (Figure 5.13). Few
such indirect H-bonding structures are present in the protonated system, as the reduced
polarity and neutral charge of this organic molecule lead to reduced formation of H-
bonds with water molecules (Figure 5.5), and even the exclusion of water molecules from
sites near the mineral surface.
Because water-mediated organo-mineral H-bonding is an indirect interaction, it
may be missed during typical analysis of model organo-mineral systems. A few studies
do provide clear evidence supporting the existence of this type of interaction. Molecular
dynamics simulations of amino acids between montmorillonite layers feature such
indirect H-bonds (Newman et al., 2002), as does a quantum chemical simulation of
nitrobenzene near a montmorillonite surface (Gorb et ah, 2000). The addition of water
molecules between organic and mineral components of simulation systems can weaken
organo-mineral adsorption, leading some to conclude that these interactions are not
significant (Shevchenko and Bailey, 1998a; Bailey et ah, 2001). However, it is important
to remember that though adsorption may be stronger within similar, dehydrated systems,
such systems are not relevant to the study of natural soil processes, as soil environments
are always hydrated. It seems unlikely that current spectroscopic techniques would be
able to identify indirect organo-mineral H-bonding structures mediated by ubiquitous
water molecules. This weak interaction can be explored only through molecular
modeling methods at this point.
355
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Cation bridges are the dominant organo-mineral adsorption complex present in
the Ca-saturated DOM-montmorillonite system by far, though they are completely absent
in the protonated system. While other models indicate the importance of cation bridging
in the adsorption of polyanions to negatively charged aluminosilicates (Akim et al., 1998;
Shevchenko and Bailey, 1998a, b; Shevchenko et al., 1999; Bailey et al., 2001; Yu et al.,
2003), no previous study has begun with fully hydrated cations as found in the solution
state. By not accounting for cation hydration in this way, earlier models may contain a
strong, inherent bias favoring the formation of cation bridges. All previous work also
involved the use of monovalent cations, although in natural soil environments it is
expected that multivalent cations such as Ca2+are necessary to the formation of strong
organo-mineral cation bridges (Theng, 1979; Oades, 1988,1995; Baldock and Skjemstad,
2000). The simulations presented in this Chapter provide molecular modeling evidence
that Ca2+ can form cation bridges between organic functional groups and the mineral
surface, even when introduced to the system in a fully hydrated form. The strong
attraction of carboxylate groups for nearby cations is evident in the simulations of cation-
saturated DOM molecules (Section 4.3.4). When a polyanion is placed within the
interlayer of negatively charged montmorillonite, a dense layer of hydrated cations is
required to screen the charge; carboxylate groups attracted to these charge screening
counterions are linked to the mineral surface as well. Cations participating in these
interactions frequently occupy positions near octahedral charge sites, while cations
involved in the occasional water bridge, or far from organic moieties, prefer locations
near tetrahedral charge sites. Perhaps the recessed position of the octahedral charge
356
Re pr oduc ed with per mi ssi on of t he copyright owner. Fur t her reproduct i on prohibited without per mi ssi on.
allows carboxylate functional groups to exert a stronger competing pull on nearby
cations, such that cation-organic inner sphere complexes become favorable.
Considerable experimental evidence indicates Ca2+bridging interactions play an
important role in adsorption and retention of humic substances and other organic
molecules. Adsorption experiments show that the addition of Ca2+ to montmorillonite
suspensions increases the adsorption of natural polyanions (Theng, 1979; Franchi et al.,
2003), while desorption experiments indicate that soils release less soluble organic matter
after addition of Ca2+(Sokoloff, 1938; Muneer and Oades, 1989a). Incubation
experiments reveal that additions of Ca2+ increase soil retention of fresh organic matter,
as well as aggregate stability (Sokoloff, 1938; Muneer and Oades, 1989a-c). Extraction
experiments using chemicals that selectively remove Ca2+ and other multivalent cations
indicate that natural organo-mineral complexes are held together with such cations
(Gaiffe et al., 1984; Muneer and Oades, 1989a), especially in smectitic soils (Wattel-
Koekkoek et al., 2001), which typically contain C that is thousands of years old
(Anderson and Paul, 1984; Arai et al., 1996; Wattel-Koekkoek et al., 2003). When
contrasting calcareous and acidic versions of otherwise similar smectite-rich Vertisols,
Holder and Griffith (1983) note that organic matter from the former is far more difficult
to extract. Oades (1988) observes that calcareous soils in general tend to have much
more organic matter than similar, noncalcareous soils, while noting that high base status
and high clay content are strongly correlated in soil systems because both are traits
derived primarily from parent material, such that the influence of each on C retention
cannot be separated. In addition, Ca2+can influence organo-mineral interaction in less
direct ways, through interactions with clay minerals that modify surface area and
357
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
structure (Oades, 1988). Nevertheless, the body of evidence presented above supports
the presence of strong cation bridging organo-mineral interactions in soil systems, as
observed in the DOM-montmorillonite simulations.
While we can assume that such DOM-montmorillonite cation bridging
interactions are able to form on external mineral surfaces, is it possible that they can
occur within the smectite interlayer region, the specific system modeled in this study?
Acidic solutions of humic fractions have been shown to penetrate montmorillonite
interlayers when in the presence of Ca2+ (Mirabella et al., 1996), but such interlayer
complexation is virtually absent above pH 5 (Schnitzer and Kodama, 1966; Martin
Martinez and Perez Rodriguez, 1969). The few natural organo-smectite interlayer
complexes described in the literature were all recovered from acidic soils (Kodama and
Schnitzer, 1971; Satoh and Yamane, 1971; Theng et al., 1986; Righi et al., 1995). The
protonated DOM molecule might represent better the organic material present under such
acidic soil conditions. Although the Ca-saturated DOM-montmorillonite system
indicates that once within the interlayer, a partially deprotonated humic substance in the
presence of sufficient Ca2+takes part in cation bridges with the mineral surface,
producing potential energies suggesting greater stability than evident in the protonated
DOM-montmorillonite system, the models examined here do not address the mechanism
of entry into the clay interlayer region. Simulation of the interactions of a model humic
molecule with appropriately protonated broken edge sites of a flexible Ca-
montmorillonite is necessary to ascertain the feasibility of interlayer complexation of
organic materials under a range of typical soil pH conditions.
358
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
5.5 Conclusions
Simulations of a model humic molecule in protonated and Ca-saturated forms
within a hydrated Ca-montmorillonite interlayer were conducted as a means of refining
concepts of organo-mineral adsorption to improve understanding of the mechanisms of C
stabilization within soils. The COMPASS force field, designed for condensed phase,
organo-mineral systems, was not equipped to simulate a flexible montmorillonite, so a
rigid mineral layer with partial charges taken from Skipper et al. (1995a, b) was used.
Initial simulations of dry and hydrated Ca-montmorillonite produced acceptable results,
with some deviations from experimental interlayer water structure resulting from the
reduced H-O-H bond angle of COMPASS water molecules, and perhaps the use of
excessive negative charge on the O atoms of the montmorillonite.
The protonated DOM-montmorillonite system featured significant direct
hydrophobic and H-bonding interactions between organic and mineral components. No
direct ion-dipole interactions were observed; instead, polar organic functional groups
formed water bridging complexes with nearby Ca2+. Some cations formed inner sphere
complexes with the mineral surface, their hydration shells stripped after annealing
calculations, perhaps partially in response to the disruption of water structure created by
large organic molecules. The Ca-saturated DOM-montmorillonite system exhibited
organo-mineral cation bridges, as well as a few water bridges and an indirect form of H-
bonding mediated by water molecules. Water bridges consisting of a water molecule
simultaneously hydrating a cation and H-bonding to an organic moiety were rare,
suggesting that a definition of water bridging that includes H-bonding is inappropriate.
359
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Although calculations based on the potential energy of the organic and mineral
components of these systems indicate that the Ca-saturated DOM is more strongly bound
to the montmorillonite than is the protonated DOM, experimental evidence suggests that
protonated humic molecules can penetrate the interlayer region, and deprotonated ones
cannot. This means a Ca-saturated DOM-montmorillonite interlayer complex, such as
that described above, may not exist in nature. Nevertheless, ideas generated by this study
concerning the nature of the organo-mineral interactions may be applied to any
adsorption system featuring organic molecules with any of the functional groups present
in the DOM molecule, and any negatively charged siloxane surface, external or internal.
It is hoped that a better understanding of the molecular mechanisms leading to adsorption
of organic matter by soil minerals will inform land management decisions designed to
encourage C sequestration.
5.6 References
Accelrys Inc. (2001) Cerius Release 4.6. Accelrys Inc.
Akim L. G., Bailey G. W., and Shevchenko S. M. (1998) A computation chemistry
approach to study the interactions of humic substances with mineral surfaces. In Humic
Substances: Structures, Properties, Uses (ed. G. Davies and E. A. Ghabbour), pp. 133-
145. Royal Society of Chemistry.
Anderson D. W. and Paul E. A. (1984) Organo-mineral complexes and their study by
radiocarbon dating. Soil Sci. Soc. Am. J. 48, 298-301.
Arai S., Hatta T., Tanaka U., Hayamizu K., Kogoshi K., and Ito O. (1996)
Characterization of the organic components of an Alfisol and a Vertisol in adjacent
1^
locations in Indian semi-arid tropics using optical spectroscopy, C NMR spectroscopy,
and 14C dating. Geoderma 69, 59-70.
Bailey G. W., Akim L. G., and Shevchenko S. M. (2001) Predicting chemical reactivity
of humic substances for minerals and xenobiotics. Use of computational chemistry,
scanning probe microscopy and virtual reality. In Humic Substances and Chemical
360
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Contaminants (ed. C. E. Clapp, M. H. B. Hayes, N. Senesi, P. R. Bloom, and P. M.
Jardine), pp. 41-72. Soil Science Society of America.
Baldock J. A., Oades J. M., Waters A. G., Peng X., Vassallo A. M., and Wilson M. A.
(1992) Aspects of the chemical structure of soil organic materials as revealed by solid-
state 13C NMR spectroscopy. Biogeochem. 16, 1-42.
Baldock J. A. and Skjemstad J. O. (2000) Role of the soil matrix and minerals in
protecting natural organic materials against biological attack. Org. Geochem. 31, 697-
710.
Barriuso E., Laird D. A., Koskinen W. C., and Dowdy R. H. (1994) Atrazine desorption
from smectites. Soil Sci. Soc. Am. J. 58, 1632-1638.
Bosetto M., Arfaioli P., and Fusi P. (1994) Adsorption of metolachlor on homoionic
montmorillonites. Agrochim. 38, 14-24.
Boulet P., Bowden A. A., Coveney P. V., and Whiting A. (2003) Combined experimental
and theoretical investigations of clay-polymer nanocomposites: Intercalation of single
bifunctional organic compounds in Na+-montmorillonite and Na+-hectorite clays for the
design of new materials. J. Mater. Chem. 13, 2540-2550.
Boyd S. A. and Jaynes W. F. (1992) Role of layer charge in organic contaminant sorption
by organo-clays. Layer charge characteristics o f clays. Pre-Meeting Workshop CMS and
SSSA, 48-77.
Breu J. and Catlow C. R. A. (1995) Chiral recognition among tris(diimine)-Metal
Complexes. 4. Atomistic computer modeling of a monolayer of [Ru(bpy)3]2+ intercalated
into a smectite clay. Inorg. Chem. 34, 4504-4510.
Breu J., Raj N., and Catlow C. R. A. (1999) Atomistic computer modeling of
[Ru(bpy)3]2+and [Ru(phen)3]2+intercalated into low charged smectites. J. Chem. Soc.,
Dalton Trans., 835-845.
Bujdak J., Hackett E., and Giannelis E. P. (2000) Effect of layer charge on the
intercalation of poly(ethylene oxide) in layered silicates: Implications on nanocomposite
polymer electrolytes. Chem. Mater. 12, 2168-2174.
Capkova P., Pospisil M., Miehe-Brendle J., Trchova M., Weiss Z., and Dred R. L. (2000)
Montmorillonite and beidellite intercalated with tetramethylammonium cations. J. Mol.
Model. 6 , 600-607.
Capkova P., Pospisil M., and Weiss Z. (2003) Combination of modeling and experiment
in structure analysis of intercalated layer silicates. J. Mol. Model. 9, 195-205.
Cases J. M., Berend I., Franfois M., Uriot J. P., Michot L. J., and Thomas F. (1997)
Mechanism of adsorption and desorption of water vapor by homoionic montmorillonite:
3. The Mg2+, Ca2+, Sr2+and Ba2+ exchanged forms. Clays Clay Miner. 45, 8-22.
361
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Celis R., Comejo J., Hermosin M. C., and Koskinen W. C. (1997) Sorption-desorption of
atrazine and simazine by model soil colloidal components. Soil Sci. Soc. Am. J. 61, 436-
443.
Chang F.-R. C., Skipper N. T., Refson K., Greathouse J. A., and Sposito G. (1999)
Interlayer molecular structure and dynamics in Li-, Na-, and K-montmorillonite-water
systems. In Mineral-Water Interfacial Reactions: Kinetics and Mechanisms (ed. D. L.
Sparks and T. Grundl), pp. 88-106. American Chemical Society.
Chang F.-R. C., Skipper N. T., and Sposito G. (1995) Computer simulation of interlayer
molecular structure in sodium montmorillonite hydrates. Langmuir 11, 2734-2741.
Chang F.-R. C., Skipper N. T., and Sposito G. (1997) Monte Carlo and molecular
dynamics simulations of interfacial structure in lithium-montmorillonite hydrates.
Langmuir 13, 2074-2082.
Chang F.-R. C., Skipper N. T., and Sposito G. (1998) Monte Carlo and molecular
dynamics simulations of electrical double-layer structure in potassium-montmorillonite
hydrates. Langmuir 14, 1201-1207.
Chavez Paez M., dePablo L., and dePablo J. J. (2001) Monte Carlo simulations of Ca-
montmorillonite hydrates. J. Chem. Phys. 114, 10948-10953.
Chen J.-S. and Chiu C.-Y. (2003) Characterization of soil organic matter in different
particle-size fractions in humid subalpine soils by CP/MAS 13C NMR. Geoderma 117,
129-141.
Chiou C. T. and Rutherford D. W. (1997) Effects of exchanged cation and layer charge
on the sorption of water and EGME vapors on montmorillonite clays. Clays Clay Miner.
45, 867-880.
Chorover J. and Amistadi M. K. (2001) Reaction of forest floor organic matter at
goethite, bimessite, and smectite surfaces. Geochim. Cosmochim. Acta 65, 95-109.
Dios Cancela G., Alfonso Mendez L., Huertas F. J., Romero Taboada E., Sainz Diaz C.
I., and Hernandez Laguna A. (2000) Adsorption mechanism and structure of the
montmorillonite complexes with (CH3)2XO (X=C, and S), (CFLOhPO, and CH3-CN
molecules./. Coll. Interface Sci. 222, 125-136.
Dios Cancela G., Huertas F. J., Taboada E. R., Sanchez Rasero F., and Hernandez
Laguna A. (1997) Adsorption of water vapor by homoionic montmorillonites. Heats of
adsorption and desorption. J. Coll. Interface Sci. 185, 343-354.
Dios Cancela G., Romero Taboada E., Huertas F. J., Hernandez Laguna A., and Sanchez
Rasero F. (1996) Interaction of trialkyl phosphites with montmorillonites. Clays Clay
Miner. 44,170-180.
362
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
Eusterhues K., Rumpel C., Kleber M., and Kogel-Knabner I. (2003) Stabilisation of soil
organic matter by interactions with minerals as revealed by mineral dissolution and
oxidative degradation. Org. Geochem. 34, 1591-1600.
Farmer V. C. (1974) The layer silicates. In The Infra-Red Spectra o f Minerals (ed. V. C.
Farmer), pp. 331-363. Mineralogical Society.
Franchi M., Ferris J. P., and Gallori E. (2002) Cations as mediators of the adsorption of
nucleic acids on clay surfaces in prebiotic environments. Origins Life Evol. Biosphere 33,
1-16.
Gaiffe M., Duquet B., Tavant H., Tavant Y., and Bruckert S. (1984) Stabilite biologique
et comportement physique d'un complexe argilo-humique place dans differentes
conditions de saturation en calcium ou en potassium. Plant Soil 77, 271-284.
Gaudel-Siri A., Brocorens P., Siri D., Gardebien F., Bredas J.-L., and Lazzaroni R.
(2003) Molecular dynamics of e-Caprolactone intercalated in Wyoming sodium
montmorillonite. Langmuir 19, 8287-8291.
Ghosh D. R. and Keinath T. M. (1994) Effect of clay minerals present in aquifer soils on
the adsorption and desorption of hydrophobic organic compounds. Environ. Progress 13,
51-59.
Giese R. F., Wu W., and van Oss C. J. (1996) Surface and electrokinetic properties of
clays and other mineral particles, untreated and treated wtih organic or inorganic cations.
J. Dispersion Sci. Technol. 17, 527-547.
Gorb L., Gu J., Leszczynska D., and Leszczynski J. (2000) The interaction of
nitrobenzene with the hydrate basal surface of montmorillonite: An ab initio study. Phys.
Chem. Chem. Phys. 2, 5007-5012.
Greathouse J. A., Refson K., and Sposito G. (2000) Molecular dynamics simulation of
water mobility in magnesium-smectite hydrates. J. Am. Chem. Soc. 122, 11459-11464.
Greathouse J. A. and Storm E. W. (2002) Calcium hydration on montmorillonite clay
surfaces studied by Monte Carlo simulation. Mol. Sim. 28, 633-647.
Haberhauer G., Aquino A. J. A., Tunega D., Gerzabek M. H., and Lischka H. (2001)
Modeling of molecular interactions of soil components with organic compounds. In
Humic Substances: Structures, Models and Functions (ed. E. A. Ghabbour and G.
Davies), pp. 209-219. Royal Society of Chemistry.
Hackett E., Manias E., and Giannelis E. P. (1998) Molecular dynamics simulations of
organically modified layered silicates. J. Chem. Phys. 108, 7410-7415.
Hackett E., Manias E., and Giannelis E. P. (2000) Computer simulation studies of
PEO/layer silicate nanocomposites. Chem. Mater. 12, 2161-2167.
363
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Hall P. L. (1982) Neutron scattering techniques for the study of clay minerals. In
Advanced Techniques for Clay Mineral Analysis (ed. J. J. Fripiat), pp. 51-75. Elsevier.
Heinz H., Castelijns H. J., and Suter U. W. (2003) Structure and phase transitions of alkyl
chains on mica. J. Am. Chem. Soc. 125, 9500-9510.
Hendricks S. B., Nelson R. A., and Alexander L. T. (1940) Hydration mechanism of the
clay mineral montmorillonite saturated with various cations. J. Am. Chem. Soc. 62, 1457-
1464.
Hensen E. J. M. and Smit B. (2002) Why clays swell. J. Phys. Chem. B 106, 12664-
12667.
Holder M. B. and Griffith S. M. (1983) Some characteristics of humic materials in
Caribbean Vertisols. Can. J. Soil Sci. 63, 151-159.
Janeba D., Capkova P., and Weiss Z. (1998) Molecular mechanics studies of
montmorillonite intercalated with tetramethylammonium and trimethylphenylammonium.
J. Mol. Model. 4, 176-182.
Jaynes W. F. and Boyd S. A. (1991) Clay mineral type and organic compound sorption
by hexadecyltrimethylammonium-exchanged clays. Soil Sci. Soc. Am. J. 55, 43-48.
Jaynes W. F. and Boyd S. A. (1991) Hydrophobicity of siloxane surfaces in smectites as
revealed by aromatic hydrocarbon adsorption from water. Clays Clay Miner. 39, 428-
436.
Keldsen G. L., Nicholas J. B., Carrado K. A., and Winans R. E. (1994) Molecular
modeling of the enthalpies of adsorption of hydrocarbons on smectite clay. J. Phys.
Chem. 98,279-284.
Kodama H. and Schnitzer M. (1971) Evidence for interlamellar adsorption of organic
matter by clay in a podzol soil. Can. J. Soil Sci. 51, 509-512.
Kubicki J. D., Blake G. A., and Apitz S. E. (1997) Molecular orbital calculations for
modeling acetate-aluminosilicate adsorption and dissolution reactions. Geochim.
Cosmochim. Acta 61, 1031-1046.
Kuppa V. and Manias E. (2002) Computer simulation of PEO/layered silicate
nanocomposites: 2. Lithium dynamics in PEO/Li+ montmorillonite intercalates. Chem.
Mater. 14,2171-2175.
Kuppa V. and Manias E. (2003) Dynamics of poly(ethylene oxide) in nanoscale
confinements: A computer simulations perspective. J. Chem. Phys. 118, 3421-3429.
Kuppa V., Menakanit S., Krishnamoorti R., and Manias E. (2003) Simulation insights on
the structure of nanoscopically confined poly(ethylene oxide). J. Polym. Sci. B: Polym.
Phys. 41, 3285-3298.
364
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Laird D. A. (1996) Interactions between atrazine and smectite surfaces. In Herbicide
Metabolites in Surface Water and Groundwater (ed. M. T. Meyer and E. M. Thurman),
pp. 86-100. American Chemical Society.
Laird D. A., Barriuso E., Dowdy R. H., and Koskinen W. C. (1992) Adsorption of
atrazine on smectites. Soil Sci. Soc. Am. J. 56, 62-67.
Laird D. A. and Fleming P. D. (1999) Mechanisms for adsorption of organic bases on
hydrated smectite surfaces. Environ. Toxicol. Chem. 18, 1668-1672.
Leinweber P. and Schulten H.-R. (1998) Advances in analytical pyrolysis of soil organic
matter. J. Anal. Appl. Pyrolysis 47, 165-189.
Leinweber P., Schulten H.-R., and Jancke H. (1999) New evidence for the molecular
composition of soil organic matter in Vertisols. Soil Sci. 164, 857-870.
Mahieu N., Powlson D. S., and Randall E. W. (1999) Statistical analysis of published
carbon-13 CPMAS NMR spectra of soil organic matter. Soil Sci. Soc. Am. J. 63, 307-319.
Martin Martinez F. and Perez Rodriguez J. L. (1969) Interlamellar adsorption of a
blackearth humic acid on Na-montmorillonite. Z. Pflanzenernahr. Bodenk. 124, 52-57.
Mirabella A., Piccolo A., and Pietramellara G. (1996) Intercalation between a well-
characterized Andisol fulvic acid and montmorillonite. Fresenius Environ. Bull. 5, 430-
435.
Mooney R. W., Keenan A. G., and Wood L. A. (1952) Adsorption of water vapor by
montmorillonite. II. Effect of exchangeable ions and lattice swelling as measured by X-
Ray diffraction. J. Am. Chem. Soc. 74, 1371-1374.
Mortland M. M. (1986) Mechanisms of adsorption of nonhumic organic species by clays.
In Interactions o f Soil Minerals with Natural Organics and Microbes (ed. P. M. Huang
and M. Schnitzer), pp. 59-76. Soil Science Society of America, Inc.
Muneer M. and Oades J. M. (1989a) The role of Ca-organic interactions in soil aggregate
stability. III. Mechanisms and models. Aust. J. Soil Res. 27,411-423.
Muneer M. and Oades J. M. (1989b) The role of Ca-organic interactions in soil aggregate
stability. I. Laboratory studies with 14C-glucose, CaC0 3 and CaS0 4 2 H2 O. Aust. J. Soil
Res. 27, 389-399.
Muneer M. and Oades J. M. (1989c) The role of Ca-organic interactions in soil aggregate
stability. II. Field studies with 14C-labelled straw, CaCOs and CaS0 4 2 H2 O. Aust. J. Soil
Res. 27, 401-409.
Nagy K. L., Schlegel M. L., Fenter P., and Sturchio N. C. (2001) Structure of natural
organic matter sorbed on muscovite as determined by surface x-ray reflectivity. 11th
Annual V. M. Goldschmidt Conference.
365
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Newman S. P., Di Cristina T., Coveney P. V., and Jones W. (2002) Molecular dynamics
simulation of cationic and anionic clays containing amino acids. Langmuir 18, 2933-
2939.
Oades J. M. (1988) The retention of organic matter in soils. Biogeochem. 5, 35-70.
Oades J. M. (1995) Recent advances in organomineral interactions: Implications for
carbon cycling and soil structure. In Environemental Impact o f Soil Component
Interactions: Natural and Anthropogenic Organics, Vol. 1 (ed. P. M. Huang, J. Berthelin,
J.-M. Bollag, W. B. McGill, and A. L. Page). CRC Press.
Oades J. M., Vassallo A. M., Waters A. G., and Wilson M. A. (1987) Characterization of
organic matter in particle size and density fractions from a red-brown earth by solid-state
13C N.M.R. Aust. J. Soil. Res. 25, 71-82.
Park S.-H., Fitch A., and Wang Y. L. (1997) Computational studies compared to
electrochemical measurements of intercalation of cationic compounds in Wyoming
montmorillonite. J. Phys. Chem. B 101, 4889-4896.
Park S.-H. and Sposito G. (2000) Monte Carlo simulation of total radial distribution
functions for interlayer water in Li-, Na- and K-montmorillonite hydrates. J. Phys. Chem.
B 104, 4642-4648.
Park S.-H. and Sposito G. (2003) Do montmorillonite surfaces promote methane hydrate
formation? Monte Carlo and molecular dynamics simulations. J. Phys. Chem. B 107,
2281-2290.
Pimentel G. C. and McClellan A. L. (1960) The Hydrogen Bond. W H Freeman.
Pintore M., Deiana S., Demontis P., Manunza B., Sufffitti G. B., and Gessa C. (2001)
Simulations of interlayer methanol in Ca- and Na-saturated montmorillonites using
molecular dynamics. Clays Clay Miner. 49, 255-262.
Pospisil M., Capkova P., Merinska D., Malac Z., and Simonik J. (2001) Structure
analysis of montmorillonite intercalated with cetylpyridinium and
cetyltrimethylammonium: Molecular simulations and XRD analysis. J. Colloid Int. Sci.
236,127-131.
Pospisil M., Capkova P., Weiss Z., Malac Z., and Simonik J. (2002) Intercalation of
octadecylamine into montmorillonite: Molecular simulations and XRD analysis. J.
Colloid Int. Sci. 245, 126-132.
Pospisil M., Capkova P., Weissmannova H., Klika Z., Trchova M., Chmielova M., and
Weiss Z. (2003) Structure analysis of montmorillonite intercalated with rhodamine B:
Modeling and experiment. J. Mol. Model. 9, 39-46.
366
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Righi D., Dinel H., Schulten H.-R., and Schnitzer M. (1995) Characterization of clay-
organic-matter complexes resistant to oxidation by peroxide. Euro. J. Soil Sci. 46, 423-
429.
Ristori G. G., Sparvoli E., Nobili M. d., and D'Acqui L. P. (1992) Characterization of
organic matter in particle-size fractions of Vertisols. Geoderma 54, 295-305.
Sato H., Yamagishi A., and Kato S. (1992a) Theoretical study on the interactions
between a metal chelate and a clay: Monte Carlo simulations. J. Phys. Chem. 96, 9377-
9382.
Sato H., Yamagishi A., and Kato S. (1992b) Theoretical studies on racemic adsorption of
tris(l,10-phenanthroline)metal(II) by a clay: Monte Carlo simulations. J. Phys. Chem. 96,
9382-9387.
Sato H., Yamagishi A., Naka K., and Kato S. (1996) Monte Carlo simulations on
intercalation oftris(l,10-phenanthroline)Metal(II) by saponite clay. J. Phys. Chem. 100,
1711-1717.
Satoh T. and Yamane I. (1971) On the interlamellar complex between montmorillonite
and organic substance in certain soil. Soil Sci. Plant Nutr. 17, 181-185.
Schnitzer M. and Kodama H. (1966) Montmorillonite: Effect of pH on its adsorption of a
soil humic compound. Science 153, 70-71.
Schnitzer M., Ripmeester J. A., and Kodama H. (1988) Characterization of the organic
matter associated with a soil clay. Soil Sci. 145, 448-454.
Schulten H.-R. and Leinweber P. (1996) Characterization of humic and soil particles by
analytical pyrolysis and computer modeling. J. Anal. Appl. Pyrolysis 38, 1-53.
Schulten H.-R. and Leinweber P. (2000) New insights into organic-mineral particles:
Composition, properties and models of molecular structure. Biol. Fertil. Soils 30, 399-
432.
Schulten H.-R., Leinweber P., and Schnitzer M. (1998) Analytical pyrolysis and
computer modelling of humic and soil particles. In Structure and Surface Reactions o f
Soil Particles (ed. P. M. Huang, N. Senesi, and J. Buffle), pp. 281-324. John Wiley &
Sons Ltd.
Schulten H.-R., Leinweber P., and Theng B. K. G. (1996) Characterization of organic
matter in an interlayer clay-organic complex from soil by pyrolysis methylation-mass
spectrometry. Geoderma 69, 105-118.
Schulten H.-R. and Schnitzer M. (1997) Chemical model structures for soil organic
matter and soils. Soil Sci. 162,115-130.
367
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Shevchenko S. M. and Bailey G. W. (1998a) Modeling sorption of soil organic matter on
mineral surfaces: Wood-derived polymers on mica. Supramolecular Sci. 5, 143-157.
Shevchenko S. M. and Bailey G. W. (1998b) Non-bonded organo-mineral interactions
and sorption of organic compounds on soil surfaces: A model approach. J. Molec. Struc.
422, 259-270.
Shevchenko S. M., Bailey G. W., and Akim L. G. (1999) The conformational dynamics
of humic polyanions in model organic and organo-mineral aggregates. J. Molec. Struc.
460, 179-190.
Skipper N. T., Chang F.-R. C., and Sposito G. (1995a) Monte Carlo simulation of
interlayer molecular structure in swelling clay minerals. 1. Methodology. Clays Clay
Miner. 43, 285-293.
Skipper N. T., Soper A. K., and Smalley M. V. (1994) Neutron diffraction study of
calcium vermiculite: Hydration of calcium ions in a confined environment. J. Phys.
Chem. 98, 942-945.
Skipper N. T., Sposito G., and Chang F.-R. C. (1995b) Monte Carlo simulation of
interlayer molecular structure in swelling clay minerals. 2. Monolayer hydrates. Clays
Clay Miner. 43, 294-303.
Slade P. G., Stone P. A., and Radoslovich E. W. (1985) Interlayer structures of the two-
layer hydrates of Na- and Ca-vermiculites. Clays Clay Miner. 33, 51-61.
Smith D. E. (1998) Molecular computer simulations of the swelling properties and
interlayer structure of cesium montmorillonite. Langmuir 14, 5959-5967.
Sokoloff V. P. (1938) Effect of neutral salts of sodium and calcium on carbon and
nitrogen of soils. J. Agr. Res. 57, 201-216.
Sposito G. (1989) The Chemistry o f Soils. Oxford University Press.
Sposito G., Park S.-H., and Sutton R. (1999) Monte Carlo simulation of the total radial
distribution function for interlayer water in sodium and potassium montmorillonites.
Clays Clay Miner. 47, 192-200.
Stevenson F. J. (1994) Humus chemistry: Genesis, Composition, Reactions. John Wiley
& Sons, Ltd.
Sun H. (1998) COMPASS: An ab initio force-field optimized for condensed-phase
applications - Overview with details on alkane and benzene compounds. J. Phys. Chem.
B 102, 7338-7364.
Sutton R. and Sposito G. (2001) Molecular simulation of interlayer structure and
dynamics of 12.4 A Cs-smectite hydrates. J. Coll. Interface Sci. 237, 174-184.
368
Re pr oduc e d with per mi ssi on of t he copyright owner. Fur t her reproduct i on prohibited without per mi ssi on.
Teppen B. J., Yu C.-H., Miller D. M., and Schafer L. (1998) Molecular dynamics
simulations of sorption of organic compounds at the clay mineral/aqueous solution
interface. J. Comput. Chem. 19,144-153.
Teppen B. J., Yu C.-H., Newton S. Q., Miller D. M., and Schafer L. (2002) Quantum
molecular dynamics simulations regarding the dechlorination of trichloroethene in the
interlayer space of the 2:1 clay mineral nontronite. J. Phys. Chem. A 106, 5498-5503.
Theng B. K. G. (1974) The Chemistry o f Clay-Organic Reactions. Halsted Press.
Theng B. K. G. (1979) Formation and Properties o f Clay-Polymer Complexes. Elsevier
Scientific Publishing Company.
Theng B. K. G. (1982) Clay-polymer interactions: Summary and perspectives. Clays
Clay Miner. 30, 1-10.
Theng B. K. G., Churchman G. J., and Newman R. H. (1986) The occurrence of
interlayer clay-organic complexes in two New Zealand soils. Soil Sci. 142,262-266.
Theng B. K. G. and Scharpenseel H. W. (1975) The adsorption of 14C-labeIled humic
acid by montmorillonite. International Clay Conference, 643-653.
Theng B. K. G., Tate K. R., and Becker-Heidmann P. (1992) Towards establishing the
age, location, and identity of the inert soil organic matter of a spodosol. Z.
Pflanzenernahr. Bodenk. 155, 181-184.
Titiloye J. O. and Skipper N. T. (2000) Computer simulation of the structure and
dynamics of methane in hydrated Na-smectite clay. Chem. Phys. Lett. 329, 23-28.
Tunega D., Haberhauer G., Gerzabek M. H., and Lischka H. (2002) Theoretical study of
adsorption sites on the (001) surfaces of 1:1 clay minerals. Langmuir 18, 139-147.
Van Duin A. C. T. and Larter S. R. (2001) Molecular dynamics investigation into the
adsorption of organic compounds on kaolinite surfaces. Org. Geochem. 32, 143-150.
Van Oss C. J. and Giese R. F. (1995) The hydrophilicity and hydrophobicity of clay
minerals. Clays Clay Miner. 43,474-477.
Wattel-Koekkoek E. J. W., Buurman P., van der Plicht J., Wattel E., and van Breemen N.
(2003) Mean residence time of soil organic matter associated with kaolinite and smectite.
Euro. J. Soil Sci. 54, 269-278.
Wattel-Koekkoek E. J. W., van Genuchten P. P. L., Buurman P., and van Lagen B.
(2001) Amount and composition of clay-associated soil organic matter in a range of
kaolinitic and smectitic soils. Geoderma 99, 27-49.
369
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
Xu W., Johnston C. T., Parker P., and Agnew S. F. (2000) Infrared study of water
sorption on Na-, Li-, Ca-, and Mg-exchanged (SWy-1 and SAz-1) montmorillonite. Clays
Clay Miner. 48, 120-131.
Yariv S. (1992) The effect of tetrahedral substitution of Si by Al on the surface acidity of
the oxygen plane of clay minerals. Intern. Rev. Phys. Chem. 11, 345-375.
Yu C.-H., Newton S. Q., Miller D. M., Teppen B. J., and Schafer L. (2001) Ab initio
study of the nonequivalence of adsorption of D- and L-peptides on clay mineral surfaces.
Struc. Chem. 12, 393-398.
Yu C.-H., Newton S. Q., Norman M. A., Miller D. M., Schafer L., and Teppen B. J.
(2000a) Molecular dynamics simulations of the adsorption of methylene blue at clay
mineral surfaces. Clays Clay Miner. 48, 665-681.
Yu C.-H., Newton S. Q., Norman M. A., Schafer L., and Miller D. M. (2003) Molecular
dynamics simulations of adsorption of organic compounds at the clay mineral/aqueous
solution interface. Struc. Chem. 14, 175-185.
Yu C.-H., Norman M. A., Newton S. Q., Miller D. M., Teppen B. J., and Schafer L.
(2000b) Molecular dynamics simulations of the adsorption of proteins on clay mineral
surfaces. J. Molec. Struc. 556, 95-103.
Zech W., Senesi N., Guggenberger G., Kaiser K., Lehmann J., Miano T. M., Miltner A.,
and Schroth G. (1997) Factors controlling humification and mineralization of soil organic
matter in the tropics. Geoderma 79, 117-161.
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
370
Chapter Six
Conclusion
6.1 Molecular Modeling as a Means of Examining Complex Systems
The heterogeneity characteristic of environmental surfaces at molecular spatial
scales presents a broad range of challenges to the study of interactions relevant to
engineered and natural systems. Smectite minerals are an excellent example of the
complexity common to such surfaces. Even neglecting important properties like the
particle size, edge site protonation state, and stacking of these layered 2 : 1 clay minerals, a
simple study of the siloxane surfaces of smectites must account for the number, type
(tetrahedral versus octahedral), and distribution of isomorphic substitution sites creating
negative charge, the identity of cations hovering near these sites to balance the charge,
and the hydration state. The variation of these characteristics across a clay surface
creates an array of microsites, each inviting a particular set of interactions with
adsorbates of interest. While direct observation of adsorption reactions cannot be
accomplished through microscopy, as the vast majority of the siloxane surface surrounds
interlayer regions within the interior of clay particles, a combination of bulk chemical
methods and spectroscopic studies have provided insights regarding the dominant
interaction mechanisms operating in numerous clay systems.
Of course, these techniques have their limitations. Bulk chemical methods, such
as adsorption/desorption, selectivity, and diffusion experiments, provide no direct
observations of molecular scale interactions, instead revealing trends which may be
considered consistent or inconsistent with hypothesized adsorption mechanisms. While
371
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
spectroscopies can be used to examine the interactions of molecules directly, they
provide only spatially averaged information. Though the individual molecules of a
substance associated with a natural surface encounter a number of different adsorption
microsites, and therefore may adopt a variety of different adsorption mechanisms, there is
no spectroscopic indication of which of these mechanisms is associated with each type of
microsite. Depending on the type of spectroscopy used to examine an adsorption system,
some adsorption interactions may provide weak signals, or no signal at all, compared to
others. Thus, the spectroscopic evidence of a single dominant adsorption mechanism
may hide signals from important corollary mechanisms. Translation of these
spectroscopic signals into molecular interactions requires substantial chemical knowledge
about the system under study, representing the synthesis of a broad range of both bulk
chemical and spectroscopic data. When the combination of available information
provides vague or contradictory views of adsorption interactions, molecular modeling
often proves a useful tool for exploring relevant surface chemistry at a truly molecular
scale.
Take, for example, the seemingly simple, inorganic adsorption system present
within the interlayer of a Cs-smectite hydrate (Section 1.2). The strong selectivity of
smectite minerals for Cs+(Muurinen et al., 1987; Oscarson et al., 1994; Iwasaki and
Onodera, 1995; Staunton and Roubard, 1997), and the reduced diffusion of this cation
upon adsorption within the mineral matrix (Calvet, 1973; Muurinen et al., 1987; Cho et
al., 1993; Wanner et al., 1996; Tsai et al., 2001; Cormenzana et al., 2003), suggests that
Cs+forms inner sphere complexes with interlayer surfaces near isomorphic substitution
sites, due to the low hydration energy of the cation (Ohtaki and Radnai, 1993). Water
372
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
adsorption and desorption experiments in combination with XRD (Mooney et al., 1952;
Calvet, 1972; Prost, 1976; MacEwan and Wilson, 1980; Berend et al., 1995; Chiou and
Rutherford, 1997) and IR (Prost, 1975) measurements supporting a reduced interlayer
water content for Cs-smectites, relative to smectites saturated with cations of high
hydration energy, are consistent with inner sphere adsorption of Cs+, because this cation
cannot draw enough water into the region to become fully hydrated. Additional water
molecules adsorbed by Cs-smectite hydrates likely occupy sites near external surfaces
and in abundant micropores. However, recent NMR (Weiss et al., 1990a, b; Kim et al.,
1995,1996) and EXAFS (Bostick et al., 2002; Nakano et al., 2003) spectra of adsorbed
Cs+have been interpreted to indicate that the interlayer region contains cations in both
inner and outer sphere complexes. In order to understand cation-mineral interactions
relevant to both the use of smectitic barriers at nuclear waste storage facilities (Cho et al.,
1993; Chun et al., 2001), and the movement of radioactive Cs+ contaminants via colloidal
transport through aquifers (Kersting et al., 1999), it is essential that the contradictory
views of adsorption processes presented by these two models of the Cs-smectite
interlayer be reconciled.
Adsorb a polyfunctional organic molecule to a hydrated smectite surface and the
complexity of the system increases drastically. Organic functional groups common to
materials derived from biological molecules are able to interact with a variety of smectite
microsites to form adsorption complexes bound by hydrophobic, H-bonding, ion or
ligand exchange, and cation or water bridging interactions (Table 1.1). A large organic
molecule may be adsorbed simultaneously to multiple mineral microsites via a few
different organo-mineral adsorption mechanisms. These organo-mineral interactions
373
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout per mi ssi on.
offer protection from microbial degradation to the organic molecule so involved, and are
thought to be responsible for the lengthy mean residence times of the complex and poorly
defined organic fraction known as humic substances (Oades, 1988; Schulten and
Leinweber, 1996; Zech et al., 1997; Baldock and Skjemstad, 2000).
Natural organo-mineral complexes have proven to be difficult subjects for both
bulk chemical and spectroscopic analyses, given the complexity of both the humic
substances and the natural mineral assemblages present in soil. Two lines of reasoning
point to the importance of two different adsorption mechanisms in the formation and
stabilization of organo-mineral complexes. Considerable experimental evidence
indicates that the presence of multivalent cations like Ca leads to improved organic
matter retention and aggregate stability within soils (Sokoloff, 1938; Muneer and Oades,
1989a-c), while the removal of these cations induces rapid release of C molecules and
mineral particles (Gaiffe et al., 1984; Muneer and Oades, 1989a), especially in smectitic
soils (Wattel-Koekkoek et al., 2001). This collection of evidence suggests that cation
bridging interactions may result in the creation of organo-mineral microaggregates that
protect soil organic matter from microbial degradation. In fact, Oades (1988) observes
that soils rich in Ca2+, Al3+, and Fe3+tend to have a higher organic matter content than
similar soils lacking substantial reservoirs of these multivalent cations.
An examination of soil organic matter itself reveals another trend relevant to
organo-mineral adsorption mechanisms. Studies of soil particle size fractions indicate
that alkyl C is preserved preferentially, especially in clay size fractions, which are often
enriched in smectite minerals (Oades et al., 1987; Oades, 1988; Baldock et al., 1992;
Ristori et al., 1992; Leinweber et al., 1999; Mahieu et al., 1999; Baldock and Skjemstad,
374
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
2000). When exposed to solutions of natural organic matter, smectite minerals
selectively adsorb large quantities of alkyl C functional groups (Chorover and Amistadi,
2001), and many I3C CP/MAS NMR spectra of smectitic soils are dominated by alkyl
signals (Wattel-Koekkoek et al., 2001). An examination of the chemical makeup of the
few natural organo-smectite interlayer complexes recovered from soils indicates that the
organic material retained in these interlayers is primarily alkyl (Theng et al., 1986;
Schnitzer et al., 1988; Schulten et al., 1996) or alkylaromatic (Righi et al., 1995). This
collection of evidence suggests that hydrophobic interactions may play an important role
in sequestration of organic matter in soils, especially where penetration of smectite
interlayers is possible. In order to better understand the processes involved in soil C
storage, and thus inform land management decisions designed to offset atmospheric CO2
emissions through sequestration of organic matter in soils, we must determine which of
these two organo-mineral interactions is relevant under a range of soil conditions, and
identify any other less obvious adsorption mechanisms.
Atomistic simulations can provide a valuable molecular scale perspective
concerning adsorption processes, one that encompasses the heterogeneous nature of the
smectite surfaces present in the two systems described above. Molecular modeling of Cs-
hectorite, -beidellite, and -montmorillonite interlayers with several different water
contents was performed in order to compare hydration, cation speciation, and water and
cation self-diffusion coefficients with those provided in the experimental literature
(Chapter 2). These simulations indicated that the interlayer of a Cs-smectite hydrate
contains less than a monolayer of water, as suggested by Calvet (1972) and Prost (1975),
but provided no evidence that outer sphere complexes could form in the interlayer, in
375
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
disagreement with recent NMR (Weiss et al., 1990a, b; Kim et al., 1995,1996) and
EXAFS (Bostick et al., 2002; Nakano et al., 2003) spectroscopic analysis. The spatially
averaged spectra of Cs-smectites contains signals from Cs+within interlayers, on external
edge and plane sites, and within extensive micropores. Perhaps a reinterpretation of this
data which accounts for adsorption of Cs+to non-interlayer sites would result in a model
of the interlayer of Cs-smectite hydrates more consistent with the modeling results
presented in Chapter 2.
Molecular dynamics calculations describing these Cs-smectite systems indicate
that the movement of Cs+ is so diminished as a result of the formation of inner sphere
complexes with the smectite surface that most cations exhibit jump diffusion only,
hovering near mineral charge sites for hundreds of picoseconds, then suddenly jumping
to new locations (Chapter 2). The frequency of these jumps increases with increasing
water content, as does the self-diffusion coefficient of interlayer water molecules.
However, an unusual development within the Cs-montmorillonite system with 2/3 water
monolayer, in which two cations exhibited diffusive motion while four cations were
confined to single adsorption sites for 800 ps, required some explanation. Therefore, the
variation in cation location and movement with mineral structure and hydration state was
further explored through animation of MD data (Chapter 3).
While previous experimental work on Cs-smectites suggested a system sensitive
to differences in mineral characteristics and water content (Weiss et al., 1990a, b; Iwasaki
and Onodera, 1995; Kim et al., 1995, 1996; Onodera et al., 1998), animations revealed a
complex array of interactions influenced by charge distribution, mineral hydroxyl
orientation, and hydration state. Particularly responsive to these variables was the
376
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
tendency for water molecule sharing between two Cs+. A network of these water sharing
interactions holds in place the four stationary cations observed in the animation of Cs-
montmorillonite with 2/3 water monolayer. Meanwhile, the two diffusive Cs+can be
seen participating in a cation exchange reaction. The multibody interactions captured in
these animations add a new dimension of complexity to models of adsorption interactions
occurring within Cs-smectite interlayers.
When humic substances enter the picture, the number and variety of potential
adsorption structures increases considerably. Because our understanding of the actual
molecular composition of the diverse set of molecules making up this organic fraction is
so limited (Section 4.1), a molecular model for humic substances must mimic relevant
behaviors of natural humic materials. The DOM molecule created by Schulten (1999)
was exposed to a variety of simulation conditions and exhibited several functional
properties largely consistent with experimental measurements of natural humic
substances, including distribution of functional groups, dry bulk density, Hildebrand
solubility parameter, IR spectrum, pseudomicellar reorientation within polar solvents, and
cation selectivity after deprotonation of carboxyl functional groups (Chapter 4). Though
the model DOM molecule has an unrealistically large molecular size according to our
current understanding of humic molecules (Section 4.1.1), its behavioral properties allow
us to use it as a proxy for the diverse materials making up humic substances.
Simulations of the model humic molecule in both protonated and Ca-saturated
forms within the interlayer of a hydrated Ca-montmorillonite were performed in order to
enhance understanding of organo-mineral adsorption occurring within complex soil
systems (Chapter 5). Organo-mineral adsorption within the protonated DOM-
377
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r eproduct i on prohibited wi t hout permi ssi on.
montmorillonite system was dominated by direct hydrophobic and H-bonding
interactions which tended to exclude water from positions between organic and mineral
components, although a few water bridging complexes involving nearby Ca2+were also
present. The Ca-saturated DOM-montmorillonite system featured cation bridges, as well
as a few water bridges and an indirect type of organo-mineral H-bonding mediated by
water molecules. The layers of Ca2+ required to screen the negative charge of the
deprotonated DOM molecule from the negative charge of the clay mineral, and the waters
of hydration associated with these cations, prevent organic moieties from occupying
positions near the mineral surfaces for the most part. The simulations illustrate that
multiple organo-mineral interactions are involved in the adsorption of a humic molecule
to a mineral surface, and that the protonation state of the organic component has a
dramatic effect on the adsorption mechanisms employed.
Comparison of simulation data to experimental data reported in the literature
reveals an additional level of complexity specific to interlayer adsorption of organic
molecules. Although calculations indicate that the Ca-saturated DOM is more strongly
sorbed to the montmorillonite than the protonated DOM, interlayer adsorption
experiments conducted over a range of pH conditions suggest that protonated humic
molecules can infiltrate the interlayer region, while deprotonated humic molecules cannot
(Schnitzer and Kodama, 1966; Martin Martinez and Perez Rodriguez, 1969). As a result,
the Ca-saturated DOM-montmorillonite interlayer complex described in Chapter 5 may
not form at all. Another series of simulations involving a model humic molecule of
appropriate size near hydrated, flexible montmorillonite edge sites, in which the
protonation state of both organic and mineral components are manipulated to represent a
378
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
range of pH conditions, is necessary to determine when interlayer adsorption is possible,
and when humic adsorption to smectites is limited to externally available surfaces.
In both of these examples, molecular modeling has revealed a hidden world of
complex surface chemistry associated with smectite minerals. The range of multibody
interactions occurring within hydrated Cs-smectite interlayers, as well as the numerous
organo-mineral interactions dependent on protonation state within hydrated DOM-
montmorillonite interlayers, simply cannot be detected through traditional bulk chemical
or spectroscopic methodologies. Continued advances in modeling techniques promise to
bring to light unique insights regarding other complex interactions relevant to soil
systems. Improvements to quantum chemical algorithms, for example, will allow the
simulation of large and fully flexible molecular systems with far more accuracy than can
be supplied by todays atomistic force fields. Quantum chemical simulations describing
the interaction of Cs+ or a model humic molecule with a large, flexible smectite mineral
with realistic edge site structure would advance immensely our understanding of
molecular scale adsorption interactions relevant to these systems.
However, molecular modeling must always be considered an adjunct to
experimental work, rather than a substitution for it. Experimental data provide basic
information essential to setting up simulation systems that resemble natural ones.
Inconclusive or conflicting experimental analyses provide the questions that these models
may answer. And simulation data inform interpretations of experimental data, and
suggest new experiments designed to probe significant molecular interactions. The
synthesis of information taken from experimental and modeling investigations of
interactions taking place on heterogeneous environmental surfaces fosters a more
379
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
complete understanding of these complex molecular environments. We can use this
understanding to make wiser choices about management of the engineered and natural
systems all around us.
6.2 References
Baldock J. A., Oades J. M., Waters A. G., Peng X., Vassallo A. M., and Wilson M. A.
(1992) Aspects of the chemical structure of soil organic materials as revealed by solid-
state 13C NMR spectroscopy. Biogeochem. 16, 1-42.
Baldock J. A. and Skjemstad J. O. (2000) Role of the soil matrix and minerals in
protecting natural organic materials against biological attack. Org. Geochem. 31, 697-
710.
Berend I., Cases J. M., Francis M., Uriot J. P., Michot L., Masion A., and Thomas F.
(1995) Mechanism of adsorption and desorption of water vapor by homoionic
montmorillonites. 2. The Li+, Na+, K+, Rb+and Cs+-exchanged forms. Clays Clay Miner.
43, 324-336.
Bostick B. C., Vairavamurthy M. A., Karthikeyan K. G., and Chorover J. (2002) Cesium
adsorption on clay minerals: An EXAFS spectroscopic investigation. Environ. Sci.
Technol. 36, 2670-2676.
Calvet R. (1972) Water adsorption on clays: Study of the hydration of montmorillonite.
Bull. Soc. Chim. Fr. 8 , 3097-3104.
Calvet R. (1973) Hydration of montmorillonite and diffusion of exchangeable cations.
Part II. Study of exchangeable cation diffusion in montmorillonite. Ann. Agrono. 24,135-
217.
Chiou C. T. and Rutherford D. W. (1997) Effects of exchanged cation and layer charge
on the sorption of water and EGME vapors on montmorillonite clays. Clays Clay Miner.
45, 867-880.
Cho W. J., Oscarson D. W., Gray M. N., and Cheung S. C. H. (1993) Influence of
diffusant concentration on diffusion coefficients in clay. Radiochim. Acta 60,159-163.
Chorover J. and Amistadi M. K. (2001) Reaction of forest floor organic matter at
goethite, bimessite, and smectite surfaces. Geochim. Cosmochim. Acta 65, 95-109.
Chun K. S., Kim S. S., and Kang C. H. (2001) Release of boron and cesium or uranium
from simulated borosilicate waste glasses through a compacted Ca-bentonite layer. J.
Nucl. Mat. 298, 150-154.
380
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Cormenzana J. L., Garcia-Gutierrez M., Missana T., and Junghanns A. (2003)
Simultaneous esimation of effective and apparent diffusion coefficients in compacted
bentonite. J. Contam. Hydrol. 61, 63-72.
Gaiffe M., Duquet B., Tavant H., Tavant Y., and Bruckert S. (1984) Stabilite biologique
et comportement physique d'un complexe argilo-humique place dans differentes
conditions de saturation en calcium ou en potassium. Plant Soil 77, 271-284.
Iwasaki T. and Onodera Y. (1995) Sorption behaviour of caesium ions in smectites. In
Proceedings o f the 10th International Clay Conference, 1993, pp. 67-73. CSIRO
Publishing.
Kersting A. B., Eford D. W., Finnegan D. L., Rokop D. J., Smith D. K., and Thompson J.
L. (1999) Migration of plutonium in ground water at the Nevada Test Site. Nature 397,
56-59.
Kim Y., Cygan R. T., and Kirkpatrick R. J. (1996) 133Cs NMR and XPS investigation of
cesium adsorbed on clay minerals and related phases. Geochim. Cosmochim. Acta 60,
1041-1052.
Kim Y., Kirkpatrick R. J., and Cygan R. T. (1995) 133Cs NMR study of Cs reaction with
clay minerals. In Environmental Issues and Waste Management Techniques in the
Ceramic and Nuclear Industries, Vol. 61 (ed. V. Jain and R. Palmer), pp. 629-636.
American Ceramic Society.
Leinweber P., Schulten H.-R., and Jancke H. (1999) New evidence for the molecular
composition of soil organic matter in Vertisols. Soil Sci. 164, 857-870.
MacEwan D. M. C. and Wilson M. J. (1980) Interlayer and intercalation complexes of
clay minerals. In Crystal Structures o f Clay Minerals and their X-ray Identification (ed.
G. W. Brindley and G. Brown), pp. 197-248. Mineralogical Society.
Mahieu N., Powlson D. S., and Randall E. W. (1999) Statistical analysis of published
carbon-13 CPMAS NMR spectra of soil organic matter. Soil Sci. Soc. Am. J. 63, 307-319.
Martin Martinez F. and Perez Rodriguez J. L. (1969) Interlamellar adsorption of a
blackearth humic acid on Na-montmorillonite. Z. Pflanzenernahr. Bodenk. 124, 52-57.
Mooney R. W., Keenan A. G., and Wood L. A. (1952) Adsorption of water vapor by
montmorillonite. II. Effect of exchangeable ions and lattice swelling as measured by X-
ray diffraction. J. Am. Chem. Soc. 74, 1371-1374.
Muneer M. and Oades J. M. (1989a) The role of Ca-organic interactions in soil aggregate
stability. III. Mechanisms and models. Aust. J. Soil Res. 27, 411-423.
Muneer M. and Oades J. M. (1989b) The role of Ca-organic interactions in soil aggregate
stability. I. Laboratory studies with 14C-glucose, CaC0 3 and CaS0 4 2 H2 O. Aust. J. Soil
Res. 27, 389-399.
381
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Muneer M. and Oades J. M. (1989c) The role of Ca-organic interactions in soil aggregate
stability. II. Field studies with 14C-labelled straw, CaC0 3 and CaS0 4 2 H2 O. Aust. J. Soil
Res. 27, 401-409.
Muurinen A., Penttila-Hiltunen P., and Rantanen J. (1987) Diffusion mechanisms of
strontium and cesium in compacted sodium bentonite. In Scientific Basis for Nuclear
Waste Management X( ed. J. K. Bates and W. B. Seefeldt), pp. 803-812. Materials
Research Society.
Nakano M., Kawamura K., and Ichikawa Y. (2003) Local structural information of Cs in
smectite hydrates by means of an EXAFS study and molecular dynamics simulations.
Appl. Clay Sci. 23, 15-23.
Oades J. M. (1988) The retention of organic matter in soils. Biogeochem. 5, 35-70.
Oades J. M., Vassallo A. M., Waters A. G., and Wilson M. A. (1987) Characterization of
organic matter in particle size and density fractions from a red-brown earth by solid-state
13C N.M.R. Aust. J. Soil. Res. 25, 71-82.
Ohtaki H. and Radnai T. (1993) Structure and dynamics of hydrated ions. Chem. Rev. 93,
1157-1204.
Onodera Y., Iwasaki T., Ebina T., Hayashi H., Torii K., Chatterjee A., and Mimura H.
(1998) Effect of layer charge on fixation of cesium ions in smectites. J. Contam. Hydrol.
35, 131-140.
Oscarson D. W., Hume H. B., and King F. (1994) Sorption of cesium on compacted
bentonite. Clays Clay Miner. 42, 731-736.
Prost R. (1975) Study of the hydration of clays: Water-mineral interactions and
mechanism of water retention. II. Study of a smectite (hectorite). Ann. Agrono. 26, 463-
535.
Prost R. (1976) Interactions between adsorbed water molecules and the structure of clay
minerals: Hydration mechanism of smectites. In Proceedings o f the International Clay
Conference, 1975, Mexico City (ed. S. W. Bailey), pp. 353. Applied Publishing Ltd.
Righi D., Dinel H., Schulten H.-R., and Schnitzer M. (1995) Characterization of clay-
organic-matter complexes resistant to oxidation by peroxide. Euro. J. Soil Sci. 46, 423-
429.
Ristori G. G., Sparvoli E., Nobili M. d., and D'Acqui L. P. (1992) Characterization of
organic matter in particle-size fractions of Vertisols. Geoderma 54, 295-305.
Schnitzer M. and Kodama H. (1966) Montmorillonite: Effect of pH on its adsorption of a
soil humic compound. Science 153, 70-71.
382
Re pr oduc e d with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.
Schnitzer M., Ripmeester J. A., and Kodama H. (1988) Characterization of the organic
matter associated with a soil clay. Soil Sci. 145, 448-454.
Schulten H.-R. (1999) Analytical pyrolysis and computational chemistry of aquatic
humic substances and dissolved organic matter. J. Anal. Appl. Pyrolysis 49, 385-415.
Schulten H.-R. and Leinweber P. (1996) Characterization of humic and soil particles by
analytical pyrolysis and computer modeling. J. Anal. Appl. Pyrolysis 38, 1-53.
Schulten H.-R., Leinweber P., and Theng B. K. G. (1996) Characterization of organic
matter in an interlayer clay-organic complex from soil by pyrolysis methylation-mass
spectrometry. Geoderma 69, 105-118.
Sokoloff V. P. (1938) Effect of neutral salts of sodium and calcium on carbon and
nitrogen of soils. J. Agr. Res. 57, 201-216.
Staunton S. and Roubaud M. (1997) Adsorption of 137Cs on montmorillonite and illite:
Effect of charge compensating cation, ionic strength, concentration of Cs, K and fulvic
acid. Clays Clay Miner. 45, 251-260.
Theng B. K. G., Churchman G. J., and Newman R. H. (1986) The occurrence of
interlayer clay-organic complexes in two New Zealand soils. Soil Sci. 142, 262-266.
Tsai S.-C., Ouyang S., and Hsu C.-N. (2001) Sorption and diffusion behavior of Cs and
Sr on Jih-Hsing bentonite. Appl. Radial Isot. 54, 209-215.
Wanner H., Albinsson Y., and Wieland E. (1996) A thermodynamic surface model for
caesium sorption on bentonite. Fresenius J. Anal. Chem. 354, 763-769.
Wattel-Koekkoek E. J. W., van Genuchten P. P. L., Buurman P., and van Lagen B.
(2001) Amount and composition of clay-associated soil organic matter in a range of
kaolinitic and smectitic soils. Geoderma 99, 27-49.
Weiss C. A., Kirkpatrick R. J., and Altaner S. P. (1990a) The structural environments of
cations adsorbed onto clays - 133Cs variable-temperature MAS NMR spectroscopic study
ofhectorite. Geochim. Cosmochim. Acta 54, 1655-1669.
Weiss C. A., Kirkpatrick R. J., and Altaner S. P. (1990b) Variations in interlayer cation
sites of clay minerals as studied by 133Cs MAS nuclear magnetic resonance spectroscopy.
Am. Miner. 75, 970-982.
Zech W., Senesi N., Guggenberger G., Kaiser K., Lehmann J., Miano T. M., Miltner A.,
and Schroth G. (1997) Factors controlling humification and mineralization of soil organic
matter in the tropics. Geoderma 79,117-161.
383
Re pr oduc ed with per mi ssi on of t he copyri ght owner. Fur t her r epr oduct i on prohibited wi t hout per mi ssi on.

You might also like