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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ma campus-p.14

 

ENCAPSULATION OF METAL SULFIDE AND OXIDE
CLUSTERS IN Mos2 AND ws2 AND CHARACTERIZATION OF
RESTACKED Mos2 AND ws2
By

Joy Marie Heising

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry

1999

ABSTRACT

ENCAPSULATION OF METAL SULFIDE AND OXIDE
CLUSTERS IN Mos2 AND ws2 AND CHARACTERIZATION OF
RESTACKED Mos2 AND WS2
By

Joy Marie Heising

The synthesis Of [A1,3O4(OH)24(HZO),2],‘MS2 (M = MO,W; x = 0.02 -
0.05; d spacing = 16.0A) was accomplished by a precipitative encapsulation
technique in which a suspension Of single MS2 layers in H20 was added to
an aqueous solution of the [A1,3O4(OH)24(HZO),2]7* cluster. One-
dimensional electron density mapping and Rietveld refinement revealed
that the cluster is oriented with its ”C3 axis perpendicular to the layers.
Thermal analysis of the samples indicate that the layers remain expanded
up to 100°C and partially expanded to 300°C, but surface area
measurements reveal that void space (if any) in the material is inaccessible.

The synthesis of [Fe688(PEt3)y],‘MS2 (x = 0.0502; y = 2-3; d-spacing
= 10.5 -1 1.4 A ) and [Ni989(PEt3)6]xMSZ (x = 0.04-0.077; d-spacing = 14.0
- 14.2 °) is similar but a solution of the cluster in a non-aqueous, water
miscible solvent is used. [C0688(PPh3)6]0mMoS2 (d-spacing = 15.3 A) was

synthesized by the addition of a solution of the CoéSs(PPh3)6 cluster in

CHzCl2 to an aqueous suspension Of single MOS2 layers. Thermal analysis
reveals that the volatile product in all samples is phosphine sulfide. The
hydrodesulfurization (HDS) activity of [C0688(PPh3),.«,]o.02MOS2 is comparable
tO that Of commercially available catalysts; the activity Of
[N iS,SS,(PEt3)6]xMS2 is inferior.

The structure of restacked MS2 (M = MO,W) was solved by electron

crystallographic studies. Using two-dimensional hk0 data indexed to a 3a

x a orthorhombic unit cell, Patterson projections were calculated which
revealed a severe distortion in the Mo/W plane, forming infinite zig-zag
chains. The projection of the structure suggests M-M distances of 2.92 A
and 2.74 A for MOS2 and W82, respectively. Least squares refinement
from the single crystal data gives R,=13.3% for WS2 and R1=15.3% for
M052, and reveals a WTe2 type structure.

The relationship between charge and structure in restacked MS2 has
been probed by encapsulation of alkali cations, forming A,,(HZO)YMS2 (x =
0.15-0.25; y = 0.3045) and chemical oxidation with I2 and Brz. Chemical

oxidation results in a change in the structure of restacked M082, giving rise

to a V51 x 1/51 superlattice. Differential Scanning Calorimetry studies

show an irreversible exothermic transition to 2H-MS2 which shifts in
temperature with oxidation. Thermopower measurements indicate that

restacked MOS2 and WS2 are p-type metallic conductors.

ACKNOWLEDGMENTS

Nothing presented here would have been possible without the
support, enthusiasm, patience, and guidance Of my advisor, Professor
Mercouri Kanatzidis, for which I am deeply grateful. I would also like to
thank Professor Pinnavaia for his helpful advice, for allowing me to use his
equipment, and his comments as a second reader. Much thanks to Professor
McCracken and Professor Baker for serving as members Of my graduate
committee as well. I thank Dr. Chris Marshall and Dr. Jim Brenner for
HRTEM and HDS studies, Dr. Eric Prouzet for EXAFS measurements, and
Dr. Duck-Young Chung for thermopower measurements.

I owe a great deal to Dr. John Heckman, who taught me to use a
TEM and Offered many useful suggestions throughout my studies. I would
also like tO thank the members Of the Kanatzidis group for their kindness
and all that I have learned from them, especially Rabin Bissessur, Wakgari
Hirpo, Hui-Lien Tsai, Lei Wang, Jason Hanko, and Rhonda Patschke.
Much thanks also to the chemistry department staff. Thanks to my friends,
especially Dean Lantero and Carl Iverson for help with the NMR and Gary

Lavine for just listening. Finally, I thank my family for their support.

iv

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES
ABBREVIATIONS

CHAPTER 1
Introduction
1.1 Insertion Reactions
1.2 Open Framework Materials
1.3 MOS, and WS, as hosts
1.4 Pillaring agents
References

CHAPTER 2

viii
ix

xiv

Encapsulation and Rietveld Structural Characterization of the

[AI,,O4(OH),.,(H,O),,]7+ cluster into MOS, and WS,.
2.0 Abstract
2.1 Introduction
2.2 Experimental
2.2.1 Synthesis
2.2.2 Characterization
2.3 Results and Discussion
2.3.1 Evidence Of Cluster Encapsulation
2.3.2 One-Dimensional Electron Density Mapping
2.3.3 One-Dimensional Rietveld Refinement
2.3.4 Thermal Stability and Absorption Measurements
2.4 Conclusions
References

CHAPTER 3
Encapsulation of Neutral and Cationic Metal Chalcogenide
Clusters in MOS, and WS,
3.0 Abstract
3.1 Introduction
3.2 Experimental
3.2.1 Synthesis
3.2.2 Characterization

3 5
36
37
38
38
40
41
41
49
57
60
63
65

68
69
71
72
72
7S

3.3 Results and Discussion 77

3.3.1 Evidence Of Cluster Encapsulation 77
3.3.2 Thermal Analysis 90
3.3.3 HDS Activity 101
3.4 Conclusions 109
References 1 13
CHAPTER 4
Structure of Restacked MOS, and WS, Elucidated by Electron
Crystallography 1 1 7
4.0 Abstract 118
4.1 Introduction 119
4.2 Experimental 124
4.3 Results and Discussion 125
4.3.1 Data Collection 125
4.3.2 Structure solution from the “triplet” crystal of WS, 128
4.3.3 Dynamic Scattering: Two-Beam Approximation 136
4.3.4 Structure solution from “single” crystals of WS,. 146
4.3.5 Dynamic Range Correction 153
4.3.6 Structure Solution from “single” crystals Of MOS,. 155
4.3.7 Dynamic scattering: n-beam approximation. 160
4.3.8 Electron Crystallographic Studies of an alternate superlattice
in MOS, 168
4.4 Conclusions 174
References 175
CHAPTER 5

Exfoliated and Restacked MOS, and WS,: Ionic or Neutral
Species? Encapsulation and Ordering Of Hard Electropositive

Cations l 7 9
5.0 Abstract 180
5.1 Introduction 181
5.2 Experimental 184

5.2.1 Synthesis 184
5.2.2. Characterization. 191
5.3 Results and Discussion 192
5.3.1 Cation encapsulated MS,. 192
5.3.2 Electron diffraction studies of (cation),MS,. 194
5.3.3 Modeling the cation ordering. 202
5.3.4. Oxidation state of M (M = MO,W) and the charge of the
layers. 209
5.3.5 Oxidation reactions of LiMS,. 212

5.3.6 X-ray diffraction studies of oxidized MS,. 218

vi

5.3.7 DSC studies Of oxidized M8,. 225

5.3.8 Acid restacked MS,. 231
5.3.9 Thermopower measurements. 237
5.4 Conclusions 241
References 246

vii

LIST OF TABLES

Table 2.1. IF(l)obsl and F(1)calc for [Al,,O4(OH),4(H,O)1,]0_05MOS,. 52
Table 2-2 IF(l)obsl and F(1)calc for [N1304(OH)24(H20)1210.055W82- 53
Table 2.3. Rietveld refinement summary Of

[M13O4(OH)24(H20)12]o.ossws2 59

Table 3.1. HDS catalytic activity and surface area (SA) of selected samples
and references. 103

Table 4.1. Crystal data and structure refinement for restacked WS,. 134
Table 4.2. Preliminary atomic coordinates ( x 104) and equivalent
isotropic displacement parameters (A2 x 103) for restacked WS,. 135
Table 4.3. Refinement for WS, as a function of data correction
methods. 141
Table 4.4. Crystal data and structure refinement for crystals of WS,. 150
Table 4.5. Atomic coordinates ( x 104) and equivalent isotropic
displacement parameters (A2 x 103) Of restacked “single” and “triple”
WS,. 15 1
Table 4.6. Crystal data and structure refinement Of restacked MOS,. 158
Table 4.7. Atomic coordinates ( x 10‘) and equivalent isotropic
displacement parameters (A2 x 103) in restacked MOS,. 159
Table 4.8. A comparison Of simulated dynamic electron diffraction data to
the experimental data for MOS, and WS, as a function Of layer
thickness. 167

viii

LIST OF FIGURES

Figure 1.1. The chemical insertion of Li atoms into TaS,. 2
Figure 1.2. Projection of 2H-MOS2 (A) in the ab plane (one layer) and (B)
along the c axis. 2H-WS, is isostructural. 11
Figure 1.3. Structure Of lT-TiS, viewed (A) in the ab plane and (B) along
the c axis. 13
Figure 1.4. Illustration Of the exfoliation and flocculation process used to
encapsulate neutral and cationic species. 15

Figure 1.5. (A) the [A111,O,,(OH),,,(H,O),,]7+ cluster. (B) A tn'mer Of edge-
sharing octahedral Al” ions with a common vertex
(Open circle). 19
Figure 1.6. (A) MGQ, core of Co,S,(PPh,)6. (B) M6Q8 core with phosphine
ligands. 21
Figure 1.7. (A) Ni3S3 units which are stacked in a staggered fashion to
form (B), the [Ni,S,(PEt3)6] 2* cluster. 24

Figure 2.1. X-ray diffraction patterns of (A) [A1,3O,,(OH),4(H,O)1,],05MOS2
and (B) [A11304(OH),4(H,O),,]0.055WS, used for electron density maps

and Rietveld refinements. 42
Figure 2.2. Three possible orientations of the [Al,,O4(OH),4(H,O),,]7*
cluster between the layers. 43

Figure 2.3. Solid State MAS-NMR 27Al spectra of (A)
[N1304(0H)24(H20)1210.03M052 and (B) Na[A11304(0H)24(H20)121600246
Figure 2.4. HRTEM Micrographs Of (A) [A1,,O,(OH),,(H,O),,]0_03MOS, and
(B) [A]1304(OH)24(H2O)12]O.04WS2- 47
Figure 2.5. One-dimensional electron density maps of (A)
[Al1304(OH),4(H,O)1,]005MOS, and (B) [A]1304(OH)24(H20)1210.055W52- 54
Figure 2.6. (A) Electron density maps calculated from
[A1,,O,,(OH),,,(H,O)1,]0055WS2 F(1)obs data using 12-19 001 reflections
(a-h). (B) Maps calculated from theoretical data for a) a W atom and
b) a W atom and a S atom, illustrating the Fourier truncation error and
the dynamic range problem. 56
Figure 2.7. (A) TGA of [Al,,O,(OH),4(H,O)1,],03MOS, under nitrogen,
showing gradual weight loss with an inflection point near 300°C. (B)
XRD patterns of the material (a) at room temperature (b) after heating
to 120°C and (c) after heating tO 300°C. 62

Figure 3.1. X-ray diffraction patterns of (A) [C068, PPh3)6]O,4MOS, and
restacked MOS,, showing an expansion of 15.4 upon encapsulation
of the cluster (B) [C0688(PPh,)6]0'96MoS,, showing only 4 001
reflections. 78

Figure 3.2. The C068,,(PPh3)6 cluster oriented with its (A) C3 axis
perpendicular to the layers and its (B) C4 axis perpendicular to the
layers. 79

Figure 3.3. HRTEM micrographs of [C0688(PPh3)x]0.02MOS,. (A) Region
with larnellar features showing expanded and unexpanded layers.
Expansion suggests that PPh, ligands have been removed. Small (B)

Region showing in-plane structure. 80
Figure 3.4. X-ray diffraction patterns of (A) [Fe6S,(PEt3),]O_096MOS, (A =
10.5 A) and (B) [FeéS,(PEt3),]0,0WS, (A = 11.4 A). 84
Figure 3.5. HRTEM micrograph Of [Fe6Ss(PEt3)3]o,0WS, showing highly
ordered in-plane structure. 85
Figure 3.6. X-ray diffraction patterns Of (A) [N1939(PEI3)6]0.07MOS, (A =
14.2 A) and (B) [Ni,s,(PEt,),]o,,,ws, (A = 14.0 A). 88
Figure 3.7. A view of the [Ni,S,(PEt3)6]2* cluster with its Ni,S3 planes (A)
parallel to and (B) perpendicular to the MS, layers 89
Figure 3.8. TGA Of (A) CO6S,(PPh3)6. (B) [Co,,S,(PPh,)6]0,,MoS,. 91

Figure 3.9. An illustration of the original plan to create a microporous
sulfide: encapsulation Of the cluster with phosphine ligands acting as
“spacers”, followed by thermal treatment to remove the phosphine
ligands. 92

Figure 3.10. An illustration of the observed data: encapsulation Of the
cluster with phosphine ligands acting as “spacers”, followed by thermal
treatment. The phosphine ligands attack the layers as they depart,

forming SPR3 species and possible 82' anion vacancies. 93
Figure 3.11. X-ray diffraction patterns upon thermal treatment of

[C0638(PPh3)6]0,4MOS,. 94
Figure 3.12. TGA of (A) [Fe6S,(PEt3)6](BPh4),, (B) [F6633(PEI3)2]0.096MOS,

(solid line) and [Fe6S,(PEt,)3]0_,OWS, (dashed line). 97
Figure 3.13. X-ray diffraction patterns upon thermal treatment of

[F66$8(PEt3),]o'096MOS,. 98
Figure 3.14. TGA of (A) [Ni,S,(PEt,)6](ClO4),. (B) TGA of

[Ni,S,(PEt,)6]OMWS,. 100
Figure 3.15. HRTEM micrograph of [C068,(PPh3)6]0'0,MOS,

after HDS. 105

Figure 3.16. Illustration of the “rim-edge” model for (A) unpillared and
(B) pillared MOS,. 107

Figure 4.1. Schematic illustrating the proposed superstructures Of
restacked/ 1T-MOS,. 121
Figure 4.2. BFI and SAED Of “triple” and “single” WS,. 126

Figure 4.3. (A) Relationship between the 2a x a and 1/371 x a lattice. (B)
Illustration Of how three flu x a patterns can be overlapped to form

a 2a x 2a pattern. 127
Figure 4.4. Illustration of the “Gel Plotting Macro”
in NIH Image 1.60. 130

Figure 4.5. Two-dimensional Pattersons along the c-axis: (A) lT-MS,
(theoretical), (B) restacked WS, with background subtraction and
Lorentz correction, (C) restacked WS, with only Lorentz correction,

(D) restacked WS, with no corrections. 131
Figure 4.6. Schematic illustrating (A) kinematic scattering and (B)

dynamic scattering. 137
Figure 4.7. Graph Of the function K(A,Q). 139
Figure 4.8. The intersection Of the Ewald sphere with the reciprocal lattice

at an accelerating voltage Of (A) ~50kV and (B) ~100kV. 143

Figure 4.9. (A) The limit Of detectable reflections (in a perfectly oriented
crystal) as a function Of the thickness (t) Of the sample (q = 1/t), the
radius Of the Ewald sphere (r), and angle 7. (B) An illustration Of

secondary scattering. 144
Figure 4.10. Selected Area Electron Diffraction of “single” WS, patterns
used to generate Patterson contour plots. 148
Figure 4.11. Two-dimensional Patterson projections along c-axis: (A)
ideal MS, and (B-D) three “single” WS, data sets. 149
Figure 4.12. (A) Two-dimensional projection Of a restacked MS, layer and
one WTe, layer. (B) View Of WTe, parallel to c-axis. 152
Figure 4.13. Surface plots Of the (3-10) and the (-2-20) reflections from
the patterns in Figures 4.2d and 4.11a. 154
Figure 4.14. Bright Field images and Selected Area Diffraction Patterns of
“single” crystal restacked MOS,. 156
Figure 4.15. Two-dimensional Patterson projections Of
restacked MOS,. 157
Figure 4.16. Simulated dynamic Electron Diffraction patterns of restacked
WS,. 162

Figure 4.17. Patterson projections calculated from the data in 4.16. 165
Figure 4.18. Simulated Patterson projections for WS, calculated as a

function Of thickness. 166
Figure 4.19. Selected Area Electron Diffraction pattern Of alternate MOS,

superlattice. 171
Figure 4.20. Two-dimensional Patterson projections calculated for

(A) an ideal MOS, lattice and (B) from experimental data 172

Figure 4.21. Simulated Patterson projections of alternate MOS,
superlattice. 173

Figure 5.1. (A) X-ray diffraction pattern Of Nao‘,8(H,O)0.45WS,. (B) TGA Of
Nao.18(H20)0.45W82 ’ Ko.21(H20)o.4oW52, and Rbo.24(H20)o.. 193
Figure 5.2. Selected Area Electron Diffraction patterns of (A) CSO_,3MOS,
and (B) CstS, 196
Figure 5.3. Selected Area Electron Diffraction patterns Of (A) K0_,,MOS,;
(B) K0,,MOS,; (C) Rb0.,5MOS,; and (D) K0,,MOS,. 199
Figure 5.4. Selected Area Electron Diffraction patterns from (A) a thick
crystal Of LiXWS, and (B) a thin crystal Of waMOS, 200
Figure 5.5. (A) Structural model and (B) simulated electron diffraction
pattern to explain the Observed patterns in Figures 5.2-5.3. 203
Figure 5.6. Selected Area Electron Diffraction patterns Of (A) BamMOS,
and (B) NaOMMOS, 207
Figure 5.7. (A) Structural model and (B) simulated electron diffraction
pattern to explain the Observed Selected Area Electron Diffraction

patterns in Figure 5.5. 208
Figure 5.8. Selected Area Electron Diffraction pattern Of LiMoS, oxidized
with Br, in MeCN. 214

Figure 5.9. (A) Patterns which show only the alternate superlattice. (B)
Selected Area Electron Diffraction pattern of LiMoS, oxidized with
Br, in MeCN over a longer period Of time with sonication. 215

Figure 5.10. Selected Area Electron Diffraction pattern Of a minority
phase formed upon oxidation of LiWS, with Br, in MeCN. 216

Figure 5.11. Transmission X-ray diffraction (XRD) pattern Of LiMoS,
oxidized in MeCN with increasing time. 219

Figure 5.12. Transmission XRD pattern of LiWS, oxidized with Br, in
MeCN. The black circle marks a peak at 2.55 A which does not appear
to be an hk0 reflection Of any Of the three ex ected lattices. The

asterisk marks a contribution from the Saran wrap substrate. 220
Figure 5.13. Transmission XRD pattern Of LiWS, oxidized with Br, (A)
before and (B) after heating. 224
Figure 5.14. Differential Scanning Calorimetry (DSC) plots of restacked
MOS, and LiMoS, oxidized with Br,. 226
Figure 5.15. DSC plots showing the irreversible exothermic phase
transition to 2H-WS,. 227
Figure 5.16. Transmission XRD pattern Of (A) LiWS, and (B) HXMOS,
Obtained from LiMoS, exfoliated in concentrated HCl. 232
Figure 5.17. DSC plots Of H,WS, Obtained from the reaction Of HCl with
(A) LiWS, (155°C) and Li,,n(H,O),WS, (191°C). 233

Figure 5.18. Schematic illustrating (A) an ideal lT-MS, lattice, (B) a flu
x a lattice with infinite zig-zag metal chains, (C) a 1/3a x I/Ba lattice
with trimers, and (D) a 21/361 x 21/3a lattice with possible tetramers

formed due to M-M distortions. 236
Figure 5.19. Schematic band diagram illustrating the distortion of an ideal

octahedral system to form zig-zag chains. 239
Figure 5.20. Thermopower measurements Of restacked

MOS, and WS,. 240
Figure 5.21. Reaction scheme illustrating the different phases Obtained

with various treatment Of (A) MOS, and (B) WS,. 243

ABBREVIATIONS

BET:Brunauer, Emmett, and Teller (Surface Area)

BFI: Bright Field Image

DSCzDifferential Scanning Calorimetry

EDS:Energy Dispersive Spectroscopy

EXAFS:Extended X-ray Absorption Fine Structure
HDS:HydrOdesulfurization

HRTEMzHigh Resolution Transmission Electron Microscopy
MAS-NMR: Magic Angle Spinning Nuclear Magnetic Resonance
SAED:Selected Area Electron Diffraction

SEM:Scanning Electron Microscope

TEM:Transmission Electron Microscope
TGA:Thermogravimetic Analysis

XRD: X-ray Diffraction

PEt3: triethyl phosphine

PPh,: triphenyl phosphine

A1130“: [M1304(0H) 24(H20) 12]”

xiv

CHAPTER 1

Introduction

1.1 Insertion Reactions

The demand for materials with new and unique combinations Of
properties has required scientists to adopt many clever synthetic strategies.
One approach is tO alter the properties Of two existing materials by
combining them. For example, the combination of nylon—6 with clays
results in nanocomposites with enhanced thermal stability and mechanical
properties relative tO those Of the individual components.’ Materials which
are mixed on a molecular level are of particular interest to chemists, and
can be synthesized through processes such as encapsulation and insertion.

In insertion or encapsulation reactions, a guest species is
incorporated into a host species, resulting in a net gain in the energy Of the
system. The “host” is usually a one-dimensional, two-dimensional, or
three-dimensional materialz'3 and the “guest” includes molecular species
which range from a proton4 to large clusters92 or polymers.‘5'16 An
example is the chemical or electrochemical insertion of Li atoms into the

layered compound TaS,:5

 

 

 

 

n-butyl Lithium

 

 

 

Li Li Li Li Li Li
ti Li Li Li Li ti
l I

l' l' l' !' I' l'

 

 

 

Figure 1.1. The chemical insertion Of Li atoms into TaS,.

An insertion is said to be “topotactic” or “topochernical” when the structure
Of the host framework is not significantly altered by the insertion process.
Layered materials have been explored perhaps most extensively as hosts
because their structure is sufficiently robust to allow topotactic insertion,
but flexible enough to allow species of many different sizes tO be inserted.
Intercalation, a special kind of insertion process, is a very popular example
of host/guest chemistry. In the most general chemical sense, “intercalation”
refers to a reversible insertion. The word “intercalation” is sometimes
loosely employed in the literature to describe irreversible insertion
reactions as well.

The driving force for the insertion process is the net gain in energy
Of the system, which can be realized through a variety Of mechanisms. One
Of the most common ways is through electron transfer from the guest
species to the host species. For example, in the insertion Of the Li atoms
into TaS, the layers are reduced and the lithium atoms are oxidized.5
Another frequently encountered insertion mechanism is ion-exchange,
which has been used to replace gallery N a” ions in smectite clays with other
cations such as tetramethyl or tetraethylammonium.6 This is accomplished
by swelling the clay with water, then exposing it to a solution with an
excess Of the cation which is to be intercalated. A third manner is through
coordination of the guest species to the host, which is Observed when

alcohols replace coordinated water in vanadyl phosphonates.’ Other

methods exist, and some layered hosts, such as V,Os, can undergo insertion
through more than one method.8 The variety in mechanisms through
which guest species can be inserted, coupled with the large number of
known layered structures, has resulted in an enormous body of literature
on the subject Of the insertion Of guest species in lamellar materials.2'3"9'10
The earliest reports of intercalation involve graphite as a host material.‘l
Layered hosts which can undergo insertion reactions also include smectite
clays”; double hydroxides”; metal oxidesg'”, oxyhalides‘s, and halides9;
transition metal phosphates, phosphonates, and sulfophosphonates”; and
transition metal dichalcogenides.3'5'lo Even the high temperature
superconducting cuprates can undergo insertion reactions.‘4

The diversity of available layered hosts and guest species has led tO
the study of these systems for many applications. One of the most
important applications which has been explored for those systems which
undergo insertion via redox reactions is their use as solid state battery
cathodes. In particular, the insertion of alkali metals in the transition metal
dichalcogenides both chemically and electrochemically has been studied
extensively due to the interesting electronic properties of the chalcogenides,
which can be affected by the intercalation reaction.”lo Related to this is
the study of polymers as guest species, desirable because they lead to an

increase in ion mobility in the lattice. Studies have not been limited to the

layered dichalcogenides, but include hosts such as clays, V,O,, FeOCl, and

M003. Strategies to incorporate polymers have included insertion Of the
monomer followed by polymerizationls or direct incorporation of the
polymer.16 Another important application for layers which undergo
insertion processes is as ion-exchange materials17 or open framework
catalysts.18 The layers in smectite clays can be separated by solvation,
called swelling, which allows ion exchange.19 When larger cations are

used, void spaces may be created.6

1.2 Open Framework Materials

The demand for open framework structures, which originates from
their application as size selective molecular sieves and catalysts, has
inspired the discovery and characterization of a vast array of materials in
the last fifty years. Those with accessible void spaces in the range 3 - 20 A
are called microporous; compounds with larger void spaces (up tO 500 A)
are mesoporous. The most widely used Open framework materials are the
aluminosilicate zeolites, which possess three dimensional structures with
pore sizes in the range 3-9 A”, and zeolite-like aluminophosphates, which
contain micropores up to 14 A.21 Applications for zeolites include ion
exchange17 and a variety of size selective catalytic processes.22 Zeolites
with larger void spaces are desirable but have not been discovered. A

breakthrough in the synthesis of mesoporous materials was the discovery Of

MCM-41 in 1992. The synthesis involves the polymerization Of silicate
around micelles Of long-chain alkylammonium cations, followed by
calcination, and results in a material which has uniform channels with
accessible void space.23 The dimensions Of these channels can be controlled
synthetically by adjusting the micelles, allowing the pore size tO range from
16 - 100 A.

Prior to MCM-41, one strategy for attaining pore sizes larger than
14 A was the “pillaring” of smectite clays with large cations, creating
materials with accessible void space. Typically void space is quantified
through surface area values measured by N, absorption. The exchange of
small gallery cations for larger cations requires an expansion of the layers,
and is possible in smectites due tO their ability to swell and even delaminate
in water. The first example of pillaring was reported by Barrer and
MacLeOd in 1955, and involved the exchange Of Na+ for tetramethyl and
tetraethyl ammonium in smectite clays.6 These cations are small pillars,
however, and interest in pillared clays only began to develop in the 1970’s,
particularly with the discovery that pillaring with polynuclear metal
hydroxyl cations such as A113O4(OH),4(H,O),,7‘° and Zr,,(OH),,“+ results in
materials with thermal stability above 500°C and moderately high surface
area values.77 These pillars convert to metal oxide clusters upon
calcination, resulting in a material which retains the layer expansion.

Other pillaring agents have included bicyclic amine cations“, metal and

silicon chelate complexes”, metal halide clusters such as Nb6C1,,2*,'8" and
imogolite, a large tubular structure.26

Unfortunately Open framework silicates are insulators. The
chalcogenides, however, usually have semiconducting or metallic
properties, which, when combined with porosity, could result in a unique
new class Of materials with interesting practical applications which include
use as chemical sensors or catalysts. For this reason many researchers are
using a variety Of techniques in an effort to synthesize Open framework
chalcogenides.27

A logical approach is hydrothermal synthesis, as it is the method
used to make zeolites. An important breakthrough came in 1989, when
Bedard and coworkers discovered A,Sn3S7 (A = alkyl ammonium cation)
and A,MGe,,Sw.2“'34 Perhaps the most exhaustively studied open framework
chalcogenide is A,Sn3S,, which has a layered structure with large pores in
the basal plane. The solution precursors have been isolated and a
mechanism Of formation proposed.29 The effects of progressive selenium
substitution and included water upon the band gap have been reported.30
Unfortunately, adjustment of the layer stacking to minimize the void space
in the basal plane occurs when organic cations Of various sizes are used in
the synthesis.31 The tremendous flexibility Of the framework seems tO
confound efforts to maximize void space in the structure. The materials

are only stable to 250-350°C (depending on the identity Of the organic

cation), and microporous behavior is only Observed when CO, is used in
the absorption studies.32 Cesium sulfide and selenide analogs have been
isolated, but the selenide analog is fairly air sensitive and the sulfide analog
contains an S8 molecule included in the structure which cannot be
removed.33

Most other Open framework chalcogenides are not so thoroughly
investigated. Members Of the family A,MGe,,Qlo (M = transition metal; Q
= S, Se) are perhaps the second most reported and characterized
compounds.34 Some interesting antimony sulfides and selenides have also
been published.” Molten salt synthesis has resulted in KBi3S5, a three
dimensional material with channels, and (Ph4P)[M(Se6),] (M = Ga, In, T1),
a two-dimensional framework with void spaces in the basal plane occupied
by tetraphenylphosphonium cations.36 Perhaps the most promising recent
development in the search for chalcogenide analogs of zeolites are the
hydrothermally synthesized indium sulfide frameworks ASU-31 and ASU-
32, but no absorption studies are reported.37

Concurrent with the hydrothermal self-assembly approach to
synthesize microporous chalcogenides is an effort to create mesoporous
chalcogenides through supramolecular assembly, as in the synthesis Of
MCM-41. The polymerization reaction about the micelles is accomplished
by the linkage of Ge4Qm" clusters with transition metals. Preliminary

results indicate that structures with worrnholes and channels are formed;

however, the thermal stability Of the materials is poor and no absorption
studies are reported.38

As hydrothermal synthesis and supramolecular assembly are both
synthetic techniques developed for the silicates, a logical extension is the
application Of pillaring to the chalcogenides as well. Pillaring has been
extended to Other non-silicate layered materials with some success.
Pillared phosphonates have been reported, but these are not synthesized by
an insertion process.39 The layered materials which are most Obvious
candidates for pillaring by insertion are ones which can undergo ion
exchange, as this is the mechanism for pillaring in clays. Hydrotalcite has
been pillared with polyoxometallates in this manner.12 The cations
[Bi,5(OH),,]6+ [A1,3O,,(OI-I),4(H,O)1,]7+ have been inserted in M003, and the

latter cation into V,Os as well, by ion exchange.”82

Insertion of large
molecules via a redox reaction has been Observed, however, in the insertion
Of [Fe4(n-C5H4Me)4(u3-S)4] and [MO4(n-C5H4Pr‘)4(|.t3-Se)4] into M003 and
FeOCl.40

As mentioned previously, the layered dichalcogenides undergo
intercalation chemistry and have been explored extensively as battery
cathode materials due to their ability to accept alkali metal cations through
redox intercalation.2'5'10 Redox intercalation has also been used to insert

larger molecules such as pyridine" and phthalocyanine dye derivatives42 in

TaS,. Also by redox chemistry, metallocenes can be inserted into group IV

and V layered dichalcogenides43 and into SnS, and SnSe,.44 Redox insertion
is limited, however, in that the pillaring agent must be capable Of reducing
the host. An important development in the pillaring of the layered
dichalcogenides was the insertion of [Al,,O,,(OH),,,(H,O),,]7° and
[Fe,,S,(PEt_,)6]2+ into Nax(H,O)yTaS, by ion exchange.“92 The layers were
dispersed (similar to delamination) in a mixture Of n-methylformarnide and
water and the clusters introduced, resulting in layer flocculation. The use i
of ion exchange to pillar TaS, expands the identity of guest species which
can be incorporated to include species which cannot be inserted by redox

chemistry, which encompasses most Of the pillaring agents used in clays.

1.3 MOS, and WS, as hosts

Unlike the group IV and V chalcogenides, MOS, was notorious for its
resistance to intercalation until 1983, when it was discovered that redox
insertion is possible but induces a transformation in the coordination

environment Of the metal.47

MOS, is found in nature as the mineral
molybdenite. The structure, first solved in 1923 by Dickinson and
Pauling," consists of a layer of MO atoms sandwiched by layers of sulfur

atoms such that the coordination environment of the MO atom is trigonal

prismatic. The stacking of the MOS, layers is staggered (Figure 1.2).

10

 

 

 

Figure 1.2. Projection of 2H-MOS, (A) in the ab plane (one layer) and
(B) along the c axis. 2H-WS, is isostructural.

ll

Each MOS, layer has only van der Waals interactions with the
neighboring MOS, layers. This allows the layers to slide with respect to
one another with relative ease; as such, other polytypes of MOS, can be
found with trigonal prismatic coordination Of MO, but a different stacking
arrangement. Using a simple system Of prefixes for naming the polytypes,
molybdenite is called 2H—MOS, because there are two layers per unit cell
and the crystal has hexagonal symmetry. Another polytype, 3R-MoS,, has
three layers per unit cell and rhombohedral symmetry. The structure and
polytypes Of WS, are analogous to MOS,.46

As mentioned previously, it was discovered that upon intercalation of
MOS, and WS, with lithium a structural transformation takes place in which
the metal coordination environment changes from trigonal prismatic to
octahedral." The new structure, denoted lT-MOS, (and WS,), has been
called analogous to the structure type observed for TiS2 (Figure 1.3).“

The practical applications for MOS, (and to a lesser extent WS,) are
diverse, and include use as a high temperature lubricant.49 Cobalt and
nickel supported MOS, and WS, are among the commercially available
catalysts for hydrodesulfurization.’0 Exfoliated MOS, has also been
explored as a catalyst for coal liquefaction5l and methanation of CO.52 2H-
MOS2 is a semiconductor with a band gap of 1.1 eV. Because it is found in

nature, molybdenite is a relatively inexpensive material.

12

*tA‘t-‘Tw.
var/aw Va:
warm

iii;

 

 

 

Figure 1.3. Structure of 1T-TiS, viewed (A) in the ab plane and (B)
along the c axis. Restacked MOS, and WS, are resemble this structure type
if distortions due to metal-metal associations are neglected.

13

Electrochemical studies Of Li insertion in MOS, for solid state battery
applications have been conducted,53 but a more important development in
the behavior Of LiMoS, was the discovery in 1986 that it exfoliates in
water, much like the delamination of clays. The following redox reaction

with water was proposed: 5"
LiMoS, + H,O --> MOS, (single layers) + LiOH + 1/2 H,(g) Eq. (1.1)

The suspension of MOS, layers can be recovered tO a lamellar form by
centrifugation, evaporation, or filtration. Furthermore, the layers can
encapsulate guest molecules in the restacking process. This is accomplished
by first washing away the LiOH generated in the exfoliation reaction, then
introducing a water-immiscible solvent such as benzene. An emulsion is
formed, and eventually the MOS, begins to collect at the water/solvent
interface. If a guest species is dissolved in the organic solvent, it can be
incorporated in this manner (Figure 1.4).55 Subsequent studies have shown
that some species can also be incorporated without the presence of the
organic solvent?6 It is remarkable that it is possible to incorporate neutral
species in this manner. Neutral species Of all sizes have been encapsulated
and include organic molecules,” polymers,55"56 ferrocene,55m fullerenes,58

and metal chalcogenide clusters.873 The behavior Of WS, is similar.”60

14

 

Li Li Li Li Li Li Li

Li Li Li Li Li Li Li

Li Li Li Li Li Li Li

Li Li Li Li Li Li Li
Li MOS,

‘Hzo, stir

W centrifuge, w
/\\ “/0” decant H 20 \

\\ \|>_:./\

‘ cationic/aqueous

 

 

 

 

 

 

 

 

 

 

 

i .).

$44: 1.6%

 

 

 

 

 

 

Ml‘éfik

Figure 1.4. Illustration of the exfoliation and flocculation process used
to encapsulate neutral and cationic species.

 

15

Due tO the ability Of MOS, to encapsulate neutral species, it was
believed by many scientists that the oxidation of MOS, in the exfoliation
process was complete, resulting in neutral layers in suspension. An
increasing number Of studies have revealed that it is possible to incorporate
cationic species as well, without the detectable presence Of co-encapsulated
anions. Transition metals61 and small alkyl ammonium cations,62 along with
larger cations such as protonated phenanthroline,63 iron porphyrins,64
dihexadecyldimethyl ammonium,“ and poly(allylamine hydrochloride)“
have been incorporated by the exfoliation/encapsulative precipitation
method. These results suggest that MOS, (and WS,) may in fact retain some
negative charge. The results of the studies reported in Chapters 2 and 3
support this assertion. Experiments designed specifically to address the
issue Of charge on the MS, layers are reported in Chapter 5.

The octahedral coordination environment, induced by the lithiation
reaction, appears to be retained in restacked MOS, and WS,.“ ’9 However,
there are several superlattices which have been Observed in lithiated,
restacked, and oxidized MOS, (and WS,) due to metal-metal associations
which cause a deviation from the ideal 1T structure type depicted in Figure
2. Initially a Za x 2a superlattice was reported based on transmission X-ray
diffraction pattems.‘7°59 Other researchers found evidence for a 2a x a

lattice by Scanning Tunneling Microscopy (STM).68 Still others have

16

Observed a 3a x [/30 lattice by X-ray diffraction and STM.69 The

layered metal dichalcogenides are prone to these kinds of structural
distortions due to the formation of charge density waves (CDWs).7O The
structural distortion in restacked MOS, and WS, will be discussed in detail

in Chapter 4.

1.4 Pillaring agents

Due to the unique ability of exfoliated MOS, and WS, to accept both
neutral and cationic species, there are an enormous variety of metal
chalcogenide, metal oxide, metal halide, and metal pnictide clusters which

could be used as pillaring agents?”73

Obviously, metal chalcogenide
clusters are requisite for the synthesis Of an open framework chalcogenide.
The four clusters selected for this work were chosen for their history as a

successful pillaring agent in other layered materials and/or the desirability

of their chemical composition for potential catalytic materials.

17

[Al,,O4(OH),,(H,O),,]Cl.,. The aqueous Al'” ion exhibits diverse
coordination chemistry at various pH.74 It was known for years that
soluble polynuclear species are formed at pH just below 775 before the
[Al,,O4(OH),4(H,O),,]7* cluster, sometimes abbreviated as [A1,,O4o]7*, was
crystallized from solution as a sulfate salt.76 The structure of the cation is a
distorted version Of a Keggin ion structure Observed in
polyoxometallates,73W in which twelve octahedrally coordinated A13+ ions
surround one tetrahedrally coordinated Al3+ ion (Figure 1.5a). The
octahedral Al3+ ions may be broken into four groups of edge-sharing
trimers, each with a common vertex (Figure 1.5b). These trimers are
linked to one another through edge sharing, forming a cluster with
tetrahedral symmetry. The tetrahedral Al3+ ion at the center is coordinated
by the common vertex Of each Of the four trimers. The cluster is
approximately 10 A in its smallest projection (visible in Figure 1.5a) and
nearly 13 A in its largest dimension.

Polynuclear hydroxy aluminum species, particularly the [A1,,O4o]7*
cluster, have been explored by many as pillaring agents for clays.”78
Solution and solid state 27A1 NMR have proved useful tOOls in the

identification of the cluster in solution and in the intercalated samples.79

The thermal stability of clays pillared with this species permits activity as a

18

 

Figure 1.5. (A) the [A1,,O,,(OH),,,(H,O),,]7+ cluster. The tetrahedral Al3+
polyhedron is shown in white. (B) A trimer of edge-sharing octahedral
A13+ ions with a common vertex (open circle). Four of these trimers are
linked together though edge sharing (black lines) to form the cluster.

molecular sieve and acid catalyst.78 The cluster has also been used as a
pillaring agent for M00380, TaS,“, and V,Os.82 The encapsulation of this

cluster will be described in Chapter 2.

Co,S,(PPh,)6. This cluster is a member Of a large family Of
compounds of the general formula CofiQ,(PR,)6n+ (Q = S, Se, Te; R = ethyl,
butyl, phenyl; n = 0,1,2).83"84 The structure is comprised Of a cube Of eight
chalcogen atoms containing a cobalt atom in the center Of each face (Figure
1.6a). TO each cobalt atom is ligated one phosphine group (Figure 1.6b).

This MGQ, core appears frequently in metal chalcogenide cluster
chemistry.71 The size of the C0688 core is about 5 A along the faces Of the
cube and 7.5 A along the body diagonal. The overall dimensions Of the
cluster, including the phosphine ligands, range from 14 - 15.5 A.

Research involving molecular clusters such as C06S,(PPh,)6 is
motivated (at least in part) by the hope that they might be used as
molecular precursors to extended solids.84a One particularly interesting
feature Of the CO,,Q,(PR3 6‘” clusters is their structural and compositional
similarity to catalytic materials. As mentioned previously, cobalt
supported MOS, is used as a catalyst in hydrodesulfurization (HDS), but the
mechanism Of the activity is not well understood?0 One theory to explain

the activity, called the “contact synergy” model, involves separate C098,

20

 

Figure 1.6. (A) M6Q, core (M, black circles, Q, gray circles) Of
C068,(PPh3)6. (B) MGQ8 core with phosphine ligands (P, dark gray circles).

21

clusters and MOS, domains.85 Another group of HDS catalysts, called the
Chevrel phases, are ternary compounds containing a three dimensional
array of linked MOGQB (Q = S, Se, Te) units.86

The encapsulation Of neutral members Of the C06Q,(PR3)6 family into
MOS, was initiated in the Kanatzidis laboratory by Bissessur.87 Surface
area measurements and magnetic properties were reported, but I-IDS
activity was not explored. The CO,,S,(PPh,),5 cluster was selected for
further studies due to the relative ease of the cluster synthesis, the large
expansion afforded by the bulky ligands, and the relatively high thermal
stability of the unencapsulated cluster. The encapsulation of this cluster

will be discussed in Chapter 3.

[Fe6S,(PEt,)6]2‘°. In addition to the possible use for metal
chalcogenide clusters as molecular ”precursors to extended solids, the
synthesis of many such clusters, particularly iron chalcogenide clusters, is
motivated by a need for models Of compositionally and structurally similar
redox centers in biological systems.88 Because they are of interest tO
materials scientists as well as bioinorganic chemists, the family Of
[Fe6Q,(PR3)6]'” (Q = S, Se, Te; R = methyl, ethyl; n = 0 - 4 ) compounds is
also fairly large. 8990'” ‘[Fe6S,(PEt,)6]2* is isostructural to C06S8(PPh3)6 with
similar core dimensions (4.5 A across faces and 7.5 A down the body

diagonal), but smaller overall dimensions due to the smaller phosphine

22

ligands (10 - 12 A). The [F e6Sg(PEt3)6] 2+ cluster was successfully used as
a pillaring agent for TaS, but little was reported beyond the structural

81.92

characterization Of the material The encapsulation of this cluster will

be discussed in Chapter 3.

[Ni,S,(PEt,)6] 2*. Polynuclear nickel chalcogenide species with the
M6Q, structure have not been Observed; rather, a prismane structure is
found in Ni,,Se5(PPh,)6.72'93 In general, nickel chalcogenide cluster
chemistry seems to be dominated by the linkage of nickel trimers to form
larger clusters. The [Ni,S,(PEt,)6] 2* cluster is assembled from three Ni,S3
units (Figure 1.7a) stacked in alternating directions.94 The coordination
environment of the nickel atoms in the top and bottom trimers is completed
by triethyl phosphine ligands (Figure 1.7b). The dimensions Of the core
are 6 A perpendicular to and 6.8 A parallel to the N1383 planes. The
overall dimensions of [Ni9S9(PEt,)6] 2* range from 11.5 A to 13 A.

This cluster has not been used as a pillaring agent for any lamellar
material. Ni supported MOS, and WS,, however, are active HDS catalysts;
hence, a nickel chalcogenide pillaring agent is desirable. The encapsulation

of this cluster will be discussed in Chapter 3.

23

 

Figure 1.7. (A) Ni3S3 units (Ni = black circles, S = gray circles) which
are stacked in a staggered fashion to form (B), the [Ni,S,(PEt3)6] 2* cluster.
The top and bottom trimers are ligated by triethyl phosphine (P = dark
gray circles, C = Open circles).

24

The insertion Of metal sulfide and metal oxide clusters into the
layered dichalcogenides is a step toward open framework sulfides. The
hosts MOS, and WS, are particularly well suited for this task due to their
remarkable ability tO exfoliate in water, allowing the encapsulation of large
guest molecules between the layers. The clusters [Al,;.,O,,(OH),,,(H,O),,]7+
(Chapter 2); CO6S,(PPh,)6, [Fe688(PEt3)6]2*, and [Ni,S,(PEt3)6]2+ (Chapter 3)
have been incorporated and characterized. The structure Of the layered
hosts MOS, and WS, has been solved by electron crystallography (Chapter
4), and studies regarding the oxidation state of the layers have been

conducted (Chapter 5).

25

References

l. a) Fukushima, Y.; Okada, A.; Kawasumi, M.; Kurachi, T.; Kamigaito,
0. Clay Minerals, 1988, 23, 27. b) Okada, A.; Kawakado, M. Japanese
Patent Application 1985, NO. SHO 60-217396.

2. SchOllhom, R. Comments Inorg. Chem, 1983, 2, 271.
3. Whittingham, M.S. Prog. Solid St. Chem, 1978, 12, 41.

4. Murphy, D.W.; DiSalvo, F.J.; Hull, G.W., Jr.; Waszczak, J.V.; Mayer,
S.F.; Steward, G.R.; Early, S.; Acrivos, J.V.; Geballe, T.H. J. Chem.
Phys., 1975, 62, 967.

5. a) Dines, M.B. Mater. Res. Bull, 1975, 10, 287. b) Murphy, D.W.;
DiSalvo, F.J.; Hull, G.W., Jr., Waszczak, J.V. Inorg. Chem, 1976,
15, 17. c) Hung, P. L.-K.; Sundararajan, G.; San Filippo, J., Jr.
Organometallics, 1982, l, 957.

6. Barrer, R.M.; MacLeOd, D.M. Trans. Faraday Soc., 1955, 51, 1290.

7. a) Johnson, J.W.; Jacobson, A.J.; Butler, W.M.; Rosenthal, S.E.; Brody,
J.F.; Lewandowski, J.T. J. Am. Chem. Soc., 1989, 111, 381. b)
Gendraud, P.; de Roy, M.E.; Besse, J.P. J. Solid State Chem, 1993,
106, 577.

8. a) Bouhaouss, A.; Aldebert, P.; Mater. Res. Bull, 1983, 18, 1247. b)
Aldebert, P.; Baffier, N.; Gharbi, N.; Livage, J. Mater. Res. Bull,
1981, 16, 949.

9. O’Hare, D. New J. Chem, 1994, 18, 989.

10. Rouxel, J. Comprehensive Supramolecular Chemistry, Vol. 7.,
Alberti, G.; Bein, T., Eds. Pergamon Press, New York, 1996, 77-105.

11. a) Fredenhagen, K.; Cadenbach, G. Z. Anorg. Allg. Chem,1926, 158,
249. b) Schaflrautl, C. J. Prakt. Chem, 1840, 3, 129.

12. a) Li, L.; Ma, S.; Liu, X.; Yue, Y.; Hui, J.; Xu, R.; Bao, Y.; Rocha, J.

Chem. Mater., 1996, 8, 204. b) Drezdzon, M.A.; Inorg. Chem, 1988,
27, 4628.

26

13.
14.

15.

16.

17.

18.

19.
20.

Clearfield, A. Prog. Inorg. Chem, 1998, 47,371.

a) Choy, J.—H.; Park, N.-G.; Kin, Y.-I.; Hwang, S.-H.; Lee, J .-S.; YOO,
H.-I. J. Phys. Chem, 1995, 99, 7845. b) Choy, J.-H.; Park, N-G.;
Hwang, S.-J.; Kim, D.-H. J. Am. Chem. Soc., 1994, 116, 11564.

a) Kanatzidis, M.G.; Marcy, H.O.; McCarthy, W.J.; Kannewurf, C.R.;
Marks, T.J. Solid State Ionics, 1989, 32/33, 594. b) Bissessur, R.;
DeGroot, D.C.; Schindler, J .L.; Kannewurf, C.R.; Kanatidis, M.G. J.
Chem. Soc., Chem. Commun., 1993, 687.

a) Wang, L.; Schindler, J.; Kannewurf, C.R.; Kanatzidis, M.G. J.
Mater. Chem, 1997, 7, 1277. b) Liu, Y.-J.; Schindler, J.L.; DeGroot,
D.C.; Kannewurf, C.R.; Hirpo, W.; Kanatzidis, M.G. Chem. Mater.,
1996, 8, 525.

Sherry, H.S., Ion Exchange, 2, J. A. Marinsky, Ed., Marcel Dekker,
New York, 1969.

a) Pinnavaia, T.J. Science, 1983, 220, 365. b) Figueras, F. Catal.
Rev. -Sci. Eng., 1988, 30, 457.

Walker, G.F.; Garret, W.G. Science, 1967, 156, 385.
a) Barrer, R.M. Hydrothermal Chemistry of Zeolites, Academic Press,

New York, 1972. b) Breck, D.W. Zeolite Molecular Sieves, Wiley,
New York, 1984.

21. a) Davis, ME. Acc. Chem. Res., 1993, 26, 111. b) Bu, X.; Feng, P.; Stucky, 6D.
Science, 1997, 278, 2080.

22.

23.

24.

25.

a) Vaughan, D.E.W., Chem. Eng. Prog., 1988, 25. b) Chen, N.Y.;
Garwood, W.E.; Dwyer, F.G. Shape Selective Catalysis in Industrial
Applications, Marcel Dekker, New York, 1989.

Kresge, C.T.; Leonowicz, M.E.; Roth, W.J.; Vartuli, J.C.; Beck, J.S.
Nature, 1992, 359, 710.

Mortland, M.M.; Berkheiser, V.E. Clays Clay Miner., 1976, 24, 60.

a) Knudson, M.I.; McAtee, J.L. Clays Clay Miner., 1973, 21, 19. b)
Pinnavaia, T.J. Clays Clay Miner., 1978, 26, 319.

27

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

Johnson, I.D.; Werpy, T.A.; Pinnavaia, T]. J. Am. Chem. Soc., 1988,
110, 8545.

Bowes, C.L.; Ozin, G.A. Adv. Mater., 1996, 8, 13.

Bedard, B.L.; Wilson, s.T.; Vail, L.D.; Bennet, J.M.; Flanigen, E.M.
Zeolites, Facts, Figures, Future. Jacobs, P.A.; van Santen, R.A.; Eds.
Elsevier Science Publishers B., Amsterdam, the Netherlands, 1989,
375.

Jiang, T.; Lough, A.; Ozin, G.A.; Bedard, R.L. J. Mater. Chem,
1998, 8, 733.

a) Ahari, H.; Ozin, G.A.; Bedard, R.L.; Petrov, 8.; Young, D. Adv.
Mater., 1995, 7, 370. b) Ahari, H.; Bowes, C.L.; Jiang, T.; Lough, A.;
G.A. Ozin, Bedard, R.L.; Petrov, S.; Young, D. Adv. Mater., 1995, 7,
375.

Jiang, T.; Lough, A.; Ozin, G.A.; Bedard, R.L.; Broach, R. J. Mater.
Chem, 1998, 8, 721.

Bowes, C.L.; Petrov, s.; Vovk, G.; Young, D.; Ozin, G.A.; Bedard,
R.L. J. Mater. Chem, 1998, 8, 711.

a) Sheldrick, W.S.; Braunbeck, H.-G. Z. Naturforsch, 1990, 45b,
1643. b) Marking, G.A.; Kanatzidis, M.G. Chem. Mater., 1995, 7,
1915.

a) Yaghi, O.M.; Sun, 2.; Richardson, D.A.; Groy, T.L. J. Am. Chem.
Soc., 1994, 116, 807. b) Tan, K.; Darovsky, A.; Parise, J.B. J. Am.
Chem. Soc., 1995, 117, 7039. c) Ahari, H.; Garcia, A.; Kirkby, S.;
Ozin, G.A.; Young, D.; Lough, A.J. J. Chem. Soc., Dalton Trans,
1998, 2023.

a) Parise, J.B. Science, 1991, 251, 293. b) Hanko, J.A.; Kanatzidis,
M.G. Angew. Chem. Int. Ed., 1998, 37, 342. c) Stephan, H.-O.;
Kanatzidis, M.G. J. Am. Chem. Soc., 1996, 118, 12226.

a) McCarthy, T.J.; Tanzer, T.A.; Kanatzidis, M.G. J. Am. Chem. Soc.,

1995, 117 , 1294. b) Dhingra, S.; Kanatzidis, M.G. Science, 1992,
258, 1769.

28

37.

38.

39.

40.

41.

42.

43.

45.

46.

47

Li, H.; Laine, A.; O’Keefe, M.; Yaghi, O.M. Science, 1999, 283, 1145.

a) Wachhold, M.; Rangan, K.K.; Billinge, S.J.L.; Petkov, V.; Heising,
J.; Kanatzidis, M.G. submitted for publication. b) MacLachlan, M.J.;
Coombs, N .; Ozin, G.A. Nature, 1999, 397, 681.

Dines, M.D.; DiGiacomo, P.M.; Callahan, K.P.; Griffith, P.C.; Lane,
R.H.; COOksey, R.E. Chemically Modified Surfaces in Catalysis and
Electrocatalysis, Miller, J .S., Ed., Amer. Chem. Soc. Symp. Ser. ,

1982, 192, Washington, DC, Chapter 12.

Chatakondu, K.; Green, M.L.H.; Qin, J.; Thompson, M.E.; Wiseman,
P.J. J. Chem. Soc., Chem. Commun., 1988, 223.

Gamble, F.R.; DiSalvo, F.J.; Klemm, R.A.; Geballe, T.H. Science,
1970, 168, 568.

Preobazhensky, V.B.; Zhemova, A.Ja.; Babichev, A.P.; KObrin, I.K.
Solid State Commun., 1978, 27, 127.

a) Dines, M. B. Science, 1975, 188, 1210. b) Clement, R.P.; Davies,
W.B.; Ford, K.A.; Green, M.L.H.; Jacobson, A.J. Inorg. Chem, 1978,
17, 2754.

. a) O’Hare, D.; Jaegermann, W.; Williamson, D.L.; Ohuchi, F.S.;

Parkinson, B.A. Inorg. Chem, 1988, 27, 1537. b) Forrnstone, C.A.;
FitzGerald, E.T.; O’Hare, D.; Cox, P.A.; Kurmoo, M.; HOdby, J.W.;
Lillicrap, D.; Goss-Custard, M. J. Chem. Soc., Chem. Commun.,
1990, 501.

a) Dickinson, R.G.; Pauling, L. J. Am. Chem. Soc., 1923, 45, 1466.
b) Bronsema, K.D.; de Boer, J .L.; Jellinek, F. Z. Anorg. Allg. Chem,
1986, 540/541, 15.

a) Wildervanck, J.C.; Jellinek, F. Z. Anorg. Allg. Chem, 1964, 328,
309. b) Glemser, O.; Sauer, H.; KOnig, P. Z Anorg. Chem, 1948,
257, 241. c) van Arkel, A.E., Rec. Trav., Chim, Pays-Bas, 1925, 45,
437. d) Ehrlich, R; Z. Anorg. Chem, 1948, 257 , 247.

Py, M.A.; Haering, R. R. Can. J. Phys, 1983, 61, 76.

29

48.

49.
50.

51.

52.

53.

54.

55.

56.

57.

Chianelli, R.R.; Scanlon, J.C.; Thompson, A.H. Mater. Res. Bull,
1975, 10, 1379.

Fleischauer, P.D. Thin Solid Films, 1987, 154, 309.

a) Chianelli, R.R.; Ruppert, A.P.; Jose-Yacaman, M.; Vézquez-Zavala,
A. Catal. Today, 1995, 23, 269. b) Topsoe, H.; Clausen, B.S. Catal.
Rev. -Sci. Eng., 1984, 26, 395. c) Weisser, 0.; Landa, S.
Sulfided Catalysts, their Properties and Applications. Pergamon Press,
New York, 1973.

Bockrath, B.C.; Parfitt, D.S. Catal. Lett., 1995, 33, 201.
Miremadi, B.K.; Morrison, S.R. J. Catal., 1987, 103, 334.

a) Mulhem, P.J. Can. J. Phys, 1989, 67, 1049. b) Julien, C.
Microionics: Solid State Integrable Batteries. Balkanski, M., Ed.
Elsevier Science, 1991, New York, 309.

Joensen, P.; Frindt, R.F..; Morrison, S.R. Mater. Res. Bull, 1986, 21,
547.

a) Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. Science, 1989,
246, 369. b) Zhou, X.; Yang, D.; Frindt, R.F. J. Phys. Chem. Solids,
1996, 57, 1137. c) Kosidowski, L.; Powell, A.V. Chem. Commun.,
1998, 2201.

a) Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. J. Mater. Res.,
1991, 6, 1103. b) Ruiz-I-Iitzky, E.; Jimenes, R.; Casal, B.; Manriquez,
v.; Santa Ana, A.; Gonzalez, G. Adv. Mater., 1993, 5, 738. c)
Kanatzidis, M.G.; Bissessur, R.; DeGroot, D.C.; Schindler, J.L.;
Kannewurf, C.R. Chem. Mater., 1993, 5, 595. d) Bissessur, R.;
Kanatzidis, M.G.; Schindler, J.L.; Kannewurf, C.R. J. Chem. Soc.,
Chem. Commun., 1993, 1582. e) Lemmon, J.P.; Lerner, M.M. Chem.
Mater., 1994, 6, 207. 1) Wang, L.; Schindler, J.; Thomas, J.A.;
Kannewurf, C.R.; Kanatzidis, M.G. Chem. Mater., 1995, 7, 1753. g)
Oriakhi, C.O.; Nafshun, R.L.; Lerner, M.M. Mater. Res. Bull, 1996,
31, 1513.

Tagaya, H.; Hashimoto, T.; Karasu, M.; Izumi, T.; Chiba, K. Chem.
Lett., 1991, 2113.

30

58.

59.

61.

62.

63.

65.

67.

68.

Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. Mater. Res.
Symp. Proc., 1992, Boston.

a) Yang, D.; Frindt, R.F. J. Phys. Chem. Solids, 1996, 57, 1113. b)
Tsai, H.-L.; Heising, J.; Schindler, J.L.; Kannewurf, C.R.; Kanatzidis,
M.G. Chem. Mater., 1997, 9, 879.

. a) Wang, L. Ph.D. Dissertation, Michigan State University, 1999. b)

Tsai, H.-L.; Kanatzidis, M.G., tO be published.

a) Danot, M.; Manson, J.L.; Golub, A.S.; Protzenko, G.A.;
Fabritchnyi, P.B.; Novikov, Yu.N.; Rouxel, J. Mater. Res. Bull.,
1994, 29, 833. b) Golub, A.S.; Payen, c.; Protzenko, G.A.; Novikov,
Yu.N.; Danot, M. Solid State Commun., 1997, 102, 419. c) Gash,
AE.; Spain, A.L.; Dysleski, L.M.; Flaschenreim, C.J.; Kalaveshi, A.;
Dorhout, P.K.; Strauss, S.H. Environ. Sci. Tech, 198, 32, 1007. d)
Golub, A.S.; Shumilova, I.B.; Zubavichus, Y.V.; Jahncke, M.; 81155-
Fink, G.; Danot, M.; Novikov, Y.N. J. Mater. Chem, 1997, 7, 163.
e) Gee, M.A.; Frindt, R.F.; Joensen, R; Morrison, S.R. Mater. Res.
Bull., 1986, 21, 543.

Golub, A.S.; Protsenko, G.A.; Burnileva, L.V.; Buyanovskaya, A.G.;
Novikov, Yu.N. Russ. Chem. bull, 1993, 42, 632.

Golub, A.S.; Shumilova, I.B.; Novikov, Yu.N.; Mansot, J.L.; Danot,
M. Solid State Ionics, 1996, 91, 307.

Nakagaki, S.; Mangrich, A.S.; Wypych, F. Inorg. Chim. Acta, 1997,
254, 213.

Taguchi, Y.; Kimura, R.; Azumi, R.; Tachibana, H.; Koshizaki, N.;
Shimomura, M.; Momozawa, N.; Sakai, H.; Abe, M.; Matsumoto, M.
Langmuir, 1998, 14, 6550.

. Ollivier, P.J.; Kovtyukhova, N.I.; Keller, S.W.; Mallouk, T.E. Chem.

Commun., 1998, 1563.

Yang, D.; Sandoval, S.J.; Divigalpitiya, W.M.R.; Irwin, J.C.; Frindt,
R.F. Phys. Rev. B., 1991, 43, 12053.

Qin, X.R.; Yang, R.; Frindt, R.F.; Irwin, J .C. Ultramicroscopy, 1992,
42-44, 630.

31

69.

70.

71.
72.

73.

74.

75.

76.

77.

a) Wypych, F.; SchOllhorn, R. J. Chem. Soc., Chem. Commun., 1992,
1386. b) Wypych, F.; Weber, Th.; Prins, R. Chem. Mater., 1998, 10,
723.

a) Wilson, J.A.; DiSalvo, F.J.; Mahajan, 5. Adv. Phys, 1975, 10,
1379. b) Wilson, J.A.; DiSalvo, F.J.; Mahajan, S. Phys. Rev. Lett.,
1974, 32, 882.

Dance, 1.; Fisher, K. Prog. Inorg. Chem, 1994, 41, 637.

Fenske, D.; Ohmer, J.; Hachgenei, J.; Merzweiler, K. Angew. Chem.
Int. Ed. Engl., 1988, 27, 1277.

a) Pope, M.T. Heteropoly and Isopoly Oxometallates. Springer-
Verlag, New York, New York, 1983. b) Fleming, P.B.; Mueller,
L.A.; McCarley, B.E. Inorg. Chem, 1967, 6, l. c) Koknat, F.W.;
Adaway, T.J.; Erzerum, S.I.; Syet, S. Inorg. Nucl. Chem. Lett., 1980,
16, 307.

Cotton, F.A.; Wilkinson, G. Advanced Inorganic Chemistry, 5th Ed.
John Wiley and Sons, New York, 1988, 215-217; 813-819.

a) Kohlschiitter, H.W.; Hantelrnann, P. Z. Anorg. All. Chem, 1941,
248, 319. b) Broset, C.; Biederrnann, G.; Sillén, L.G. Acta. Chem.
Scand., 1954, 8, 1917.

a) Johansson, G.; Lundgren, G.; Sillen, L.G.; Soderquist, R. Acta.
Chem. Scand., 1960, 14, 769. b) Johansson, G. Acta. Chem. Scand.,
1960, 14, 771. c) Tsai, P.P.; Hsu, P.H. Soil Sci. Soc. Am. J., 1985,
49, 1060.

a) Brindley, G.W.; Sempels, R.E. Clay Miner., 1977, 12, 229. b)
Lahav, N.; Shani, U.; Shabtai, J. Clays Clay Miner., 1978, 26, 107. c)
Singh, S.S.; Kodama, H. Clays Clay Miner., 1988, 36, 397. d)
Schoonheydt, R.A.; van den Eynde, J.; Tubbax, H.; Leeman, H.;
Stuyckens, M.; Lenotte, 1.; Stone, W.B.E. Clays Clay Miner., 1993,
41, 598. e) Coelho, A.V.; Poncelet, G. Appl. Catal., 1991, 77, 303.
f) Bukka, K.; Miller, I.D.; Shabtai, J. Clays Clay Miner., 1992, 40,
92. g) Malla, P.B.; Komarneni, S. Clays Clay Miner., 1993, 41, 472.

32

78.

79.

80.

81.
82.
83.

84.

85.
86.

87.

a) Vaughan, D.W. E.; Lussier, R.J.; Magee, J .S. US. Patent 4,176,090
(1979). b) Vaughan, D.W. E.; Lussier, R.J.; Magee, J.S. US. Patent
4,271,043 (1981). c) Shabtai, J.; US. Patent 4,238,364 (1980).

a) Bottero, J.Y.; Cases, J.M.; Flessinger,F.; Poirier, J.B. J. Phys.
Chem, 1980, 84, 2933. b) Akitt, J.W.; Farthing, A.; J. Chem. Soc.,
Dalton Trans, 1981, 1617. c) Plee, D.; Borg, F.; Gatineau, L.;
Fripiat, L]. J. Am. Chem. Soc., 1985, 107, 2362.

a) Lerf, A.; Lalik, E.; KOlOdziejski, W.; Klinowski, J. J. Phys. Chem,
1992, 96, 7389. b) Nazar, L.F.; Liblong, W.S.; Yin, X.T. J. Am.
Chem. Soc., 1991, 113, 5889. c) Nazar, L.F.; Yin, X.T.; Zinkweg,
D.; Zhang, Z.; Liblong, S. Mat. Res. Soc. Symp. Proc., 1991, 210,
417.

Nazar, L.F.; Jacobson, A.J. J. Mater. Chem, 1994, 4, 149.
Mori, M.; Isobe, T.; Senna, M. Solid State Ionics, 1995, 81, 157.

a) Hong, M.; Huang, 2.; Lei, X.; Wei, G.; Kang, B.; Liu, H. Inorg.
Chim. Acta, 1989, 159, l. b) Jiang, F.; Huang, X.; Cao, R.; Hong,
M.; Liu, H. Acta Cryst. C., 1995, 1275. c) Hong, M.; Huang, 2.; Lei,
X.; Wei, G.; Kang, B.; Liu, H. Polyhedron, 1991, 10, 927.

a) Stuczynski, S.M.; Kwon, Y.-U.; Steigerwald, M.L. J. Organomet.
Chem, 1993, 449, 167. b) Steigerwald, M.L.; Siegrist, T.;
Stuczynski, S.M. Inorg. Chem, 1991, 30, 2256. c) Cecconi, F.;
Ghilardi, C.A.; Midollini, S.; Orlandini, A.; Bencini, A. J. Chem.
Soc., Dalton Trans, 1996, 3991. d) Cecconi, F.; Ghilardi, C.A.;
Midollini, S.; Orlandini, A. Inorg. Chim. Acta, 1983, 76, L183. e)
Cecconi, F.; Ghilardi, C.A.; Midollini, 8.; Orlandini, A.; Zanello, P.
Polyhedron, 1986, 5, 2021.

Delmon, B. Prepr. Am. Chem. Soc. Div. Pet. Chem, 1977, 22, 503.
a) Chevrel, R.; Sergent, M.; Prigent, J. J. Solid State Chem, 1971, 3,
515. b) Chevrel, R.; Hirrien, M.; Sergent, M. Polyhedron, 1986, 5,
87. c) McCarty, K.F.; Anderegg, J.W.; Schrader, G.L. J. Catal.,
1985, 93, 375.

a) Bissessur, R.; Heising, J.; Hirpo, W.; Kanatzidis, M.G. Chem.
Mater., 1996, 8, 318. b) Bissessur, R. “Synthesis and Characterization

33

88.

89.

90.

91.

92.

93.
94.

of Novel Intercalation Compounds of Molybdenum Trioxide and
Molybdenum Disulfide.” Ph.D. Dissertation, Michigan State
University, 1994.

a) Holm, R.H. Chem. Soc. Rev., 1981, 10, 455. b) Coucouvanis, D.
Acct. Chem. Res., 1981, 14, 201.

Steigerwald, M.L.; Siegrist, T.; Gyorgy, E.M.; Hessen, B.; Kwon, Y.-
U., Tanzler, S.M. Inorg. Chem, 1994, 33, 3389.

a) Cecconi, F.; Ghilardi, C.A.; Midollini, S.; Orlandini, A. Inorg.
Chim. Acta, 1997, 254, 387. b) Agresti, A.; Bacci, M.; Cecconi, F.;
Ghilardi, C.A.; Midollini, S. Inorg. Chem, 1985, 24, 689. c)
Cecooni, F.; Ghilardi, C.A.; Midollini, S.; Orlandini, A.; Zanello, P.
J. Chem. Soc., Dalton Trans, 1987, 831. d) Cecooni, F.; Ghilardi,
C.A.; Midollini, S. J. Chem. Soc., Chem. Commun., 1981, 640.

Goddard, C.A.; Long, J.R.; Holm, R.H. Inorg. Chem, 1996, 35,
4347.

Nazar, L.F.; Jacobson, A.J. J.Chem. Soc., Chem. Commun., 1986,
570.

Fenske, D.; Ohmer, J. Angew. Chem. Int. Ed. Engl., 1987, 26, 148.
a) Cecconi, F.; Ghilardi, C.A.; Midollini, s. Inorg. Chem, 1983,22,

3802. b) Ghilardi, C.A.; Midollini, S.; Sacconi, L. J. Chem. Soc.,
Chem. Commun., 1981, 47.

34

CHAPTER 2

Encapsulation and Rietveld Structural Characterization Of the
[AI,,O,.(OH),,(H,O),,]7+ cluster into MOS, and WS,.

35

2.0 Abstract

The synthesis Of [Al,,O4(OH),4(H,O).2]xMS, (x=0.02-0.05, M=MO,
x=0.02-0.055, M=W) was accomplished by a precipitative encapsulation
technique using single layers of MS,. The products were characterized by
powder X-ray diffraction, energy dispersive spectroscopy (EDS),
thermogravimetric analysis (TGA), 27A1 Magic Angle Spinning Nuclear
Magnetic Resonance (MAS-NMR), High Resolution Transmission Electron
Microscopy (HRTEM), room temperature electrical conductivity
measurements, and surface area measurements. Powder X-ray diffraction
patterns show an expansion of approximately 9.9A. 27Al MAS-NMR
indicates that the cluster is intact between the MS, layers. One-dimensional
electron density mapping and Rietveld refinement performed on the
powder diffraction data shows that the cluster is oriented with its C3
symmetry axis perpendicular to the layers. The samples exhibit
conductivity values from 3-14 S/cm. TGA shows that the layers remain
expanded to 100°C, and partially expanded to 300°C. Surface area

measurements suggest that the space between the clusters is not accessible.

36

2.1 Introduction

Since Barrer and MacLeOd reported the first pillared species in
1955‘, much effort has gone into the development of microporous
materials for use as size selective molecular sieves and as catalysts.
Zeolites and pillared clays have been studied extensively for these
applications”; however, these materials are insulators. Because WS, and
MOS, have interesting electronic and optical properties, pillared WS, and
MOS, may be useful for unique applications similar to but distinct from the
applications for zeolites and pillared clays. The availability and relatively
low cost of MOS, make it an especially desirable chalcogenide host.

The practical applications of MOS, itself are quite diverse. These
include use as a solid lubricant“, a catalyst’, and as a host material for solid
state batteries.6 MOS, has been intercalated with a variety of compounds
which include polymers”, small organic molecules9, and inorganic
complexes such as ferrocene10 and the cobalt clusters CO,,Q,(PR,)6 (Q=S, Se,
Te; R=alkyl)”. WS, is isostructural to MOS,, with similar intercalation
chemistry, although less well explored.‘2

Despite the considerable intercalation history of layered
dichalcogenides, microporous pillared chalcogenides have not been reported
to date. The intercalation of TaS, with [A1,,O,,(OH),,,(H,O),,]7+ and the iron

cluster [Fe68,(PEt,)6]2‘° has been achieved, but no information about the

37

porosity of these materials is reported.13 Intercalation of the C068,,(PPh3)6
clusters into MOS, increase the surface area three or fourfold, but TEM
studies show that the lamellar expansion due to the clusters is localized.14 We
have chosen to explore [A1,3O,,(OH),,,(H,O),,]7+ as a pillaring agent for MOS,
because of its successful history as a pillaring agent for clays” and because

alumina-supported MOS, is used as a catalyst for hydrodesulfurization.5

2.2 Experimental

2.2.1 Synthesis

MOS, was purchased from Cerac and WS, from Alfa Aesar. LiBH4, n-Butyl
lithium, hexane, and AlCl3 were purchased from Aldrich. BaCl,'2H,O,
Na,SO,,010H,O, and NaOH were purchased from J .T. Baker. All compounds
were used as received except hexane, which was dried over CaH, (Aldrich).
LiMS,”. LiMoS, was prepared by a) stirring 2H-MOS, in threefold excess
of n-butyl lithium in dry hexane for 2 days or b) heating with LiBH4 at
300°C for 2 days. LiWS2 was prepared by the latter method at 350°C.

MS, aqueous suspension. LiMS, was exfoliatedl7 in deionized,
deoxygenated H,O via a redox reaction, generating single layers, lithium

hydroxide, and hydrogen gas.

38

LIMSZ(S) ‘1' H200) ---> M82 (single layers) '1' LlOH(aq) + 0.5H2 (g) Eq. (2.1)

Eq. 2.1 has been proposed for the exfoliation of MS, by Divigalpitiya et
al.9 The exfoliated suspension (pH>12) was centrifuged for 0.5 hr and
rinsed with deionized, deoxygenated H,O three times in order to reduce the
pH of the solution to about 7 (pH ~12, pH ~9, and pH ~7 after each rinse).
The product was then re-suspended in H,O and stirred 0.5 hr prior to use.
[A1,,O4(OH),,,(H,O),,]CI7 solution.18 The [A1,3O4(OH),4(H,O),,]7* was
prepared by the slow addition Of 500 ml of NaOH (0.24M) to 500 ml of an
aqueous AlCl3 solution (0.1M) at 80°C with stirring. It was then
crystallized as the sulfate salt Na[Al,3O4(OH),4(H,O),,](SO4), in order to
isolate it from any other soluble Al-containing species. It was then
redissolved by stirring in excess BaCl, solution, and the BaSO4 precipitate
removed by filtration. This method results in excess Ba2+ ions also present
in the solution.

[Al,,O4(OH),4(H,O),,],MS,. The aqueous MS, suspension was added to
various amounts of a 0.03M solution (approximately) of
[A1,_,,O,,(OH),4(I-I,O),,]Cl7 in H,O.l9 After stirring for several hours the
product was isolated via centrifuge, rinsed, and dried on a glass slide. The
product was a shiny film which could be scraped off the glass and ground

to a fine black powder.

39

2.2.2 Characterization

Powder X-ray diffraction patterns were recorded using a Rigaku-
Denki/RW400F2 (Rotaflex) rotating anode X-ray diffractometer using Cu-

Ka radiation. HRTEM micrographs were acquired using a J EOL 4000EX

II at Argonne National Laboratory, operating at 400kV. Energy
Dispersive Spectroscopy (EDS) was used to determine the ratios of Al to
M0 (or W) on a JEOL-JSM-6400V at an accelerating voltage of 20kV with
samples mounted on non-aluminum stubs using carbon paint or carbon
tape, equipped with a Tracor Northern 5500 X-ray microanalysis
attachment. Solid state 27Al MAS-NMR measurements were performed
using a Varian 400MHz instrument tuned to 130.4 MHz pulse frequency,
and at MAS frequencies of 4 and 6.2 MHz. Therrnogravimetric

measurements were Obtained with a Shimadzu TG 50 instrument using

oxygen or nitrogen flow and a heating rate of 50C or 20C/min. BET
surface area measurements were performed on a Quantasorb Jr. Sorption
System using ultra-pure nitrogen gas as the adsorbate and ultra-pure
helium gas as the carrier. The surface areas of the samples were calculated
using the BET equation.20 Conductivity measurements of the materials at
room temperature were obtained for a pressed pellet using the four prong

probe and a Keithley 236 source measure unit.

40

2.3 Results and Discussion

2.3.1 Evidence Of Cluster Encapsulation

The predominance of the 001 reflections in the X-ray powder
diffraction pattern indicate that the layers are well oriented, with a d-
spacing of about 16 A (see Figure 2.1). Restacked MOS, has a d-spacing of
6.2 A, so the expansion of the layers is about 9.9 A. This value is
consistent with the dimensions of the cluster, which are about 10.2 A along
the C3 symmetry axis, 12.8 A along one C, symmetry axis and 11.5 A
along the other C, axis (average diameter 11.5 A) (see orientations (A),
(B), and (C), respectively, in Figure 2.2). Our observed expansion
suggests that the cluster is oriented with its C3 axis perpendicular to the
layers.

Loadings of the cluster were calculated using Energy Dispersive
Spectroscopy (EDS) to quantitate an Ale0 (or Al:W) ratio. Conversion to
oxides by heating to 650°C under a stream of O, prior to quantitation was
necessary for MOS,-containing samples because the X-ray emission lines
(for EDS) of Mo and S overlap completely, hindering quantitative analysis.
The Observed values ranged from [A1,,O4(OH),4(H,O)1,]O,O,MOS, to

[A1,304(OH),4(H,O)1,]0_05MOS,. The theoretical maximum loading of the

41

(A

II1 TII III III IIr I71 II
,_ F I l l T I

O
O

 

002

fiTII'IIIjTII'III

 

 

 

lllllllllllllllllllllllllll

Intensity (arbitrary units)

bIIII'III'II
E0

 

 

 

lllllJllllJlLllllllll

 

4o 60 80 100 120 140
29 (degrees)

C
N
O

(B)

IIIFTI—ITIIIIIIIIIIIIIIIII

 

I
F
o
o

 

 

 

 

Intensity (arbitrary units)

 

 

illllllllllllLlllll

CIIIIlITIIITIIIITIII
h— .1

lllllllllllelllJJlll

o 20 4o 60 80 100 120 140
26 (degrees)

 

Figure 2 . l . X-ray diffraction patterns of (A)
[M1304(OH)24(H20)12]o.osM082 and (B) [M13O4(OI'I),4(H,O)1,]0055WS, used for
electron density maps and Rietveld refinements. Strong preferred
orientation of the samples leads to the predominance of the 001 reflections.

42

 

The gray circles represent Al atoms, open

Figure 2.2. Three possible orientations of the [Al1,0,,(OH),,,(I-I,O),,]7+

cluster between the layers.
circles are 0 atoms. H atoms are omitted for clarity.

43

cluster (if one assumes hcp packing of the spheres) is
[A1,304(OH),4(H,O),,]0_06MOS,. Conversion to oxide was not necessary for
samples containing WS,, and comparable stoichiometries (up to
[A1,3O4(OH),4(H,O)1,]0_055WS,) were observed by EDS. The stoichiometry
of selected samples was confirmed by elemental analysis.

Lithiated and exfoliated MOS, was presumed to be neutral because it
is possible to intercalate neutral molecules.9 One surprising discovery was
that no chloride is detected in the samples by EDS. It appears that this
cation intercalates without its chloride anion. The strong intensity of
reflections, the relatively high degree of order, and the ease of
encapsulation of the cation seems to indicate that the MOS, and WS, have
some negative charge. This would explain the driving force for the
intercalation. We have observed intercalation of other cations without
anions, consistent with a negative charge on the layers.21 The observed
loadings of the cluster, presuming no change in its charge, suggest that the
negative charge on the layers is between 0.1 and 0.4. This wide range Of
possible negative charge suggests that the charge on the aluminum cluster
may vary, depending on the loading. Another surprising result is that the
excess Ba2+ ions found in solution with the cluster do not appear to be
encapsulated. The layers appear to favor the soft [A1,,O4(OH),4(H,O),,]7*
ion over the hard, electropositive Ba2+ ion. Studies to probe the ionicity of

the layers in detail are described in Chapter 5.

44

MAS-NMR 27Al spectra of several samples with various loadings of
cluster exhibited a peak at about 64 ppm corresponding to the tetrahedral
aluminum at the center of the cluster (Figure 2.3). This peak has been
observed by others who have intercalated the cation into various hosts and
is found in the sulfate salt of the unintercalated cluster.'0'ls The spectra also
exhibit a broad peak centered around 0 ppm which can be attributed to
octahedral aluminum. Intermediate coordination environments may
contribute to the broad signal found between the two peaks in
[A113O4(OH),4(H,O)1,]0.0,MOS,. This suggests that, although some of the
cluster is clearly intact, there may be some decomposition products
encapsulated as well.

An effort was made to image the encapsulated clusters, or at least the
lamellar expansion, by HRTEM. HRTEM studies of thin sectioned
[Al,,O4(OH),4(I-I,O),,]7*-pillared clays: revealed a lamellar expansion (the
cluster, as it is comprised of light atoms, apparently does not scatter
sufficiently to be directly visible); however, it was smaller than the
expansion indicated by the powder X-ray diffraction patterns.22 The
authors attributed this to dehydration of the cluster in the high vacuum of
the electron microscope. Rather than thin sectioning,
[A1,,O4(OH),,,(H,O)1,],03MOS, and [Al,,O4(OH),4(I-I,O),,]0_03WS, were dusted

on a carbon supported copper grid and images were acquired at the edges

45

W

I IIII IIII III IIIIIIIIIIIII IIFI FIIII- I71
200 150 100 so 0 -50 -10'0 .150 -260 ppm

 

 

 

IWWTIVI r rII1TT Ti

1' .r.lwx rl'T #71: . 1.11.11. .ir
200 150 100 50 o -50 -160 -15'0 ado ppm
Fi re 2.3. Solid State MAS-NMR 27A1 s ctra of (A)

[A 1304(0H)24(H20)12]o.03M0$2 and (B) a[A11304(OH)24(H20)12](SO4)2.
The peak at 64 ppm (asterisk) rs attributed to the tetrahedral Al3+ atom at

the center of the cluster; the broad peak at 0 ppm (circle) is attributed to
octahedral Al3+.

46

 

Figure 2.4. HRTEM Micrographs of (A) [A113O4(OH)2,4(H20)12]o,03M082
and (B) [A11304(0H)74(H20)12]o_04WSZ. Lamellar regions are visible in
both; however, d-spacings are the same as the collapsed material,
suggesting a "calzone"—like structure.

47

of the particles. A view parallel to the layers is visible in Figure 2.4. In
contrast to the powder X-ray diffraction pattern, a more reliable method
which probes the bulk of the material, the layers do not appear to be
expanded. Figure 2.4b contains regions in which the layers appear to be
locally expanded, but scattering centers due to the cluster are not visible,
and regions of unexpanded WS, are also found in the micrograph. This
suggests that the interior of the particle is pillared, but at the edges the
layers are mostly collapsed, resembling a “calzone”.

The electrical conductivity of thin films of the materials, dried on
glass microscope slides, were measured. If measured within twenty four
hours of the exfoliation, the conductivity of the cluster intercalated MOS,
ranges from 3-14 S/cm. In contrast, the room temperature conductivity of
2H-MOS, is only ~10'5 S/cm. The metallic properties of these samples can

be attributed to a phase transition in MOS, upon lithiation.23

48

2.3.2 One-Dimensional Electron Density Mapping

In order to probe the structure and orientation of the encapsulated
cluster, one-dimensional electron density mapping calculations were
carried out on samples Of the cluster in MOS, and in WS, prepared using
excess cluster to generate particularly well ordered materials (see Figure
2.1).24 The integrated intensities (I) of the 001 reflections in the X-ray
powder diffraction patterns were extracted and converted to structure
factors by the following relation:

|F(l).b.| = (I/Lp)“2 Eq- (2.2)
in which F(l)obs is the structure factor for each reflection and Lp, a
Lorentz-polarization correction, is
Lp = (1 + c05220)/(sin20c080) Eq. (2.3)
Sixteen 001 reflections were used in the calculations for the sample
containing MOS, and eighteen 001 reflections for the sample containing
WS,.

In order to calculate an electron density map it is necessary to know
the signs [(+) or (-) for a centrosymmetric structure] as well as the
magnitudes of the F(l)obs values. The recovery of these signs is the essence
of (centrosymmetric) single crystal structure determination. In order to
recover the signs of the F(1)obs values, an assumption is made that the

contribution of the [Al,3O.,(OH),,,(I-I,O),,]7+ cluster to each structure factor

49

is small relative to the contribution from the MOS, and WS, layers.
Therefore, we can calculate the structure factors for the expanded, empty
layers and apply the signs of these structure factors (F(l)c,,c) to the |F(l)obs|
values. This is a reasonable assumption because the cluster is comprised of
light atoms which will not scatter nearly as well as the heavy atoms found
in the MOS, and WS, layers. Subsequent calculations including the cluster
(computed for modeling purposes) resulted in no changes in the signs of
the structure factors.

Calculation of F(l)calc for the expanded, empty layers is more tedious
than calculation of |F(l)obs| because one must first calculate the atomic
scattering factors (f) for MO (or W) and S by the following equation:

4
f(9) = [2a,e<-b.-/12 sinz 6') + Cid-198m2 9/ 12) Eq. (2.4)
i=1
where a,, bi, and c are a series of 3 nine coefficients which describe an
exponential decay that is specific to each element (in this case Mo, W, or
S), 0 is the Bragg angle, A is the wavelength of the radiation (1.54184 A
for Cu K,), and B is the temperature factor (B = 2 in these calculations).
The value of (f) for each element varies as a function Of 0, and as such will
be different for each F(1)c,,c.

Once f(0) for each atom is known, each F(l)calc can be calculated:

50

N
F(l)calc = Zij COS(27Tle) Eq. (2.5)
i=1
where, if there are N atoms in the structure, fj is the scattering factor Of

atom j (at the particular 0 value associated with each F(l) ), l is the 001

calc
index, and zj is the fractional coordinate of atom j on the c-axis. The signs
Of these F(l)c,,c are then applied to the corresponding |F(1)obsl values, and

electron density (p) is calculated:

p(z) = (1 / c)[22 F(l)0,,_, cos(27tlz)] Eq.(2.6)
l

where c is the c-axis, l is the 001 index, and z is a fractional coordinate
along the c-axis (varied from -0.2 to 1.2 in increments of 0.01 in the
calculation to create a “map” of the electron density).

In order to interpret a one-dimensional electron density map, it is
necessary to compare it to at least one structural model. The
[A1,3O4(OH),4(H,O),,]7+ cluster is a polyhedral species with well-defined

66 ‘99

faces; as such, it may prefer to Slt in certain orientations over a
completely random arrangement in the gallery. Three plausible
orientations for the cluster are depicted in Figure 2.2. Orientation (A), in
which the C3 axis of the cluster is perpendicular to the layers, seems the

most plausible on the basis of the Observed d-spacing in the X—ray powder

51

Table 2.1. 113(1),) and Fa)calc for [A1,,O,(OH),,(H,O),,],,,Mos,.

 

 

(th1 d... (A) d... (A) F... F....(A) F....<B) F....(C)
(001) 16.699 16.065 26.646 274.22 270.41 271.51
(002) 8.1199 8.032 52.606 247.67 266.53 257.94
(003) 5.4380 5.355 58.736 171.17 146.79 160.11
(004) 4.0530 4.016 19.715 74.776 78.750 72.375
(005) 3.2148 3.213 8.7890 44.046 50.528 50.806
(006) 2.6945 2.678 13.220 63.536 62.353 75.002
(007) 2.3010 2.295 24.328 131.01 147.03 108.80
(008) 2.0115 2.008 35.250 152.74 117.28 161.35
(009) 1.7940 1.785 27.271 188.86 212.96 186.74
(0010) 1.6069 1.606 30.300 186.01 186.74 207.76
(0011) 1.4650 1.460 36.311 166.52 157.87 129.79
(0012) 1.3438 1.339 22.789 121.93 121.44 141.82
(0013) 1.2338 1.236 14.951 56.119 69.016 65.835
(0014) 1.1507 1.148 18.734 59.011 46.407 38.087
(0015) 1.0717 1.071 10.520 18.270 13.598 28.652
(0016) 1.0026 1.004 10.942 26.389 45.069 30.572

 

52

Table 2.2 112(1),) and Pa)“, for [A1,,O,(OH),,(H,O),,],,,,ws,.

 

 

(hkt) d... (A) d... (A) F... F....(A) F....(B) F....(C)
(001) 16.127 15.957 41.993 233.64 229.81 231.01
(002) 8.0412 7.978 78.622 235.87 254.44 245.44
(003) 5.3510 5.319 87.163 192.42 169.80 183.48
(004) 4.0043 3.989 49.378 130.21 131.42 124.92
(005) 3.2001 3.191 36.795 108.94 117.22 117.33
(006) 2.6653 2.660 37.353 127.78 124.82 138.89
(007) 2.2829 2.280 59.930 176.82 196.28 155.88
(008) 1.9977 1.995 57.041 178.46 139.94 185.13
(009) 1.7754 1.773 60.342 189.72 214.66 187.26
(0010) 1.5973 1.596 68.732 172.88 173.03 197.09
(0011) 1.4524 1.451 54.920 152.13 145.55 114.76
(0012) 1.3310 1.330 44.387 116.22 112.17 133.66
(0013) 1.2279 1.227 33.418 64.414 80.720 76.615
(0014) 1.1406 1.140 42.592 83.899 70.454 63.151
(0015) 1.0645 1.064 25.648 56.567 50.907 65.094
(0016) 1.0004 0.997 26.249 64.253 82.880 69.740
(0017) 0.93866 0.939 30.428 83.784 71.011 74.567
(0018) 0.89063 0.887 23.845 80.485 82.466 84.517

 

53

 

1.2;-

0.8
0.6

0.4
0.2

c axis

IIj'jII‘WIIlIIIIIII II

C
I

 

-o.2 '

 

lJllll llllllll+LlllLLl

electron density

A
‘ W
v

 

I'IIIII'IIrIlII I'IIII'III

0:
mi
0

I
.\

axrs

O

O)
rTITjIII'III'IIT'IIIIIII'III—IT—I

 

.llLlllllllllJJllllllllllllAll

electron density

Figure 2.5. One-dimensional electron density maps of (A)
[A1,3O4(OH),4(H,O)1,]0.05MOS, and (B) [N13O4(OH),4(H,O)1,]0.055WS,. The
patterns labeled (0) were generated from the X-ray diffraction data
(Figure 2.1). The remaining patterns were calculated from three possible
orientations of the cluster depicted in Figure 2.2. Orientation (A) matches
best with the experimental data for both samples.

54

diffraction pattern. Theoretical electron density maps were generated for

these three orientations by computing F(l) values for each of these

calc
models in the same way that F(l)c,,c values were generated for the
expanded, empty layers (Eq. 2.4 and Eq. 2.5), and then calculating the
electron density map for each (Eq. 2.6).

Comparison Of the observed data (O) to theoretical data for the
intercalated cluster in three possible orientations for both samples supports
the initial supposition that the cluster is oriented with its C3 axis oriented
perpendicular to the layers in both hosts (see Figure 2.5). In this
orientation, which has been observed in clays with an expansion of 9.6 A,25
qualitatively one would expect to see four peaks for the four planes
containing oxygen atoms interspersed with three, somewhat weaker peaks
corresponding to the less densely occupied aluminum containing planes.
The electron density 'map calculated “from the diffraction data for MOS,
(Figure 2.5a, pattern (0)) matches the calculated pattern for orientation
(A) (Figure 2.5a, pattern (A)) in shape and location of peaks (although
two peaks from the aluminum planes seem enhanced, and two of the
oxygen planes seem weak) in the experimental pattern. Both the locations
and intensities of the peaks for orientations (B) and (C) do not match the

pattern calculated from the diffraction data.

The data from the cluster in WS, also matches orientation (A), but in

55

12:

electron density
_1 I I I I I I l I T I I I
G. (D - (D 3'
L I l I l J J J l l 1 +

 

 

.1

ITfIl’IqA

 

 

 

-mO'O

llLl lllll JIJJILLJI

-o.2 -o.1 o, 0.1 0.2 0.3
c-axrs

 

(B)

rIII IIIIIIIfTrTIjIIIII’TI
p-
i-
i-
h-
I-
D

 

‘
d
d

 

 

electron density

 

 

llllllllllllllllLillll

 

 

D
llllllllLll_l_LlllllllIJLJJLIJI

 

0 0.5 1
c-axis
Figure 2.6. (A) Electron density maps calculated from

[Al,,O4(OH),,(I-I,O),,]0.055WS, F(l)obs data using 12—19 001 reflections (a-h).
The S atoms are split and appear to move. (B) Maps calculated from
theoretical data for a) a W atom and b) a W atom and a S atom, illustrating
the Fourier truncation error and the dynamic range problem.

56

this case both the patterns from the diffraction data (0) and for the
theoretical pattern (A) have somewhat asymmetric peaks for two of the
oxygen planes (Figure 2.5b). In addition, the peaks for the sulfur atoms
are split. In fact, if the number of reflections used to calculate the map is
altered, the sulfur peaks seem to move (Figure 2.6a). This strange feature
is due to a combination of a dynamic range problem between the heavy
tungsten and lighter sulfur (i.e., the very strong atomic scattering factor Of
W), and a Fourier truncation caused by the use of a finite number of

reflections (Figure 2.6b).

2.3.3 One-Dimensional Rietveld Refinement

In order to confirm the cluster orientation found from the analysis
of the electron density maps, a Rietveld refinement of the three structural
models was carried out using the program FULLPROF”, on the powder
X-ray diffraction data from intercalated WS,. The cautious application of
the Rietveld method, which uses the raw experimental diffractogram and
does not require extensive and error prone data treatment, seems to be a
straightforward and efficient way to discriminate between various models
of cluster orientation. It is not subject to some of the problems of electron
density mapping, such as Fourier truncation errors and lack of resolution

due to an insufficient number diffraction peaks. Furthermore, it is possible

57

to exclude the first diffraction peaks, which are most affected by systematic
errors in the geometrical factor because of the high degree Of preferred
orientation. In electron density calculations, the absence of those peaks
could preclude the interpretation of the resulting map.

For this refinement, the first two reflections were excluded because
their calculated intensities were 2.7 and 1.7 times higher than the observed
reflections. These huge discrepancies are mostly due to the deviation of the
Lorentz factor from the ideal random powder case.27 The experimental
data were fit with the Pearson VII peak profile function, and a model for
the asymmetry of the peaks was included.28 Sixteen 001 reflections were
used, falling within the range 13.5° S 20 S 125°, with a step size of 016°.
Owing to the small number Of Bragg peaks in the diffraction pattern, the
number of free structural parameters was kept to a minimum. Reasonable
and fixed isotropic displacement parameters were assigned to the various
atoms types: BW = 1.5 A2, BS = 2.0 A2, BAl = 2.5 A2, B0 = 3.0 A2, and only
an overall temperature factor was refined. The [A1,3O,,0]7+ cluster was
treated as a rigid body, the z positional parameter of all the atoms in the
cluster being constrained to follow the same shift. The population
parameter was fixed at the value determined from EDS measurements.
The position of the sulfur layer was also refined, giving a total of four free

intensity dependent parameters: Bovemn, chuster. 25, and the scale factor.

58

Table 2.3. Rietveld refinement summary of [A1,3O4(OH),4(H,O),,]0.055WS,“

 

 

WS, orientation A orientation B orientation C
pr 12.0% 7.8% 13.6% 9.4%
x2 29.1 12.9 37.6 17.7
Rl 13.6 5.8 16.1 7.3
dDW 0.170 0.391 0.135 0.271
zS 0.095(3) 0.096(1) 0.094(3) 0.096(1)
Azdm, --- 0.010(6) 0.005(5) 0.031(7)

 

“ pr is the weighted agreement factor based on the Observations, x2 is the
goodness of fit, R1 is the agreement indices based on “Observed” integrated
intensities and dDW is the value of the Durbin-Watson statistic”, an indicator
of the degree of correlation among the residuals. Azcluster is the displacement
of the center of the cluster from its original position at z=1/2. The
standard deviations given by the refinement program are typically
underestimated and they were multiplied by a scale factor in order to be
more realistic}0

59

The results of the refinements are summarized in Table 2.3. The
WS, column corresponds to a model without cluster. Orientation (B) is
clearly to be rejected, giving worse results than the WS, host alone.
Orientation (A), which was favored by the electron density maps, gives
better results than orientation (C). There is almost certainly water between
the layers as well; however, while including ordered water in the models
improves the correlation between the theoretical and experimental patterns,
it does not change the relative ranking of the models. A larger d spacing
(11A) and electron density mapping calculations indicate that the cluster

has a different orientation ((B) in Figure 2.2) in MOO3.3|

2.3.4 Thermal Stability and Absorption Measurements

Thermogravimetric analysis under nitrogen shows continuous
gradual weight loss with an inflection point around 300°C (Figure 2.7a).
Powder X-ray diffraction indicates that the layers have not collapsed after
heating to 120°C (Figure 2.7b, pattern B). After heating to 300°C the
layers are still expanded, but the material does not diffract as well and a
“restacked” peak at 6.1 A begins to appear (Figure 2.7b, pattern C). Upon
heating the sample to 650°C the cluster degrades and the layers collapse,

resulting in a material with a d-spacing only slightly larger than the

60

restacked, unintercalated material. For [A1,3O4(OH),4(H,O)1,]0.04MOS,, there
are about 0.72 water molecules per molybdenum (about 20 per aluminum
cluster). The amount of water in the sample can be estimated if one
assumes that when heated under oxygen, the aluminum cluster will
decompose to form A1,O3 and MOS, will convert to M003. Additional
weight loss can be attributed to the water.

Surface area measurements of the samples were conducted; all
exhibited low surface areas (ex. 9.4 mzlg) relative to restacked MOS, (10
mzlg). Drying under vacuum overnight, over P,Os, soaking in ethanol and
critical point drying in C0,, and heating to 100°C, 120°C, or 300°C under
flowing nitrogen or vacuum did not improve the surface areas of the
samples. For one sample heated to 120°C, the vacuum in the outgassing
apparatus even reached ~10'5 torr, and powder X-ray diffraction after the
measurements showed little change in the d-spacing, but the surface area
was no better than restacked MOS,. Given that there is sufficient space
between the clusters to accommodate intercalated water molecules, even in
the most fully loaded samples which have been observed, it is surprising
that the surface area values of the materials are not better than restacked
MOS,. It indicates that the volume vacated by the departing water

molecules becomes inaccessible to the nitrogen molecules. In addition, the

61

 

structural rearrangement

/ by 120°C

layer collapse
begins at 300°C

 
  
  

CO (O (O
N b O)
'1 [IIIII

IIlTrleII

  

°/o weight loss

coco
(DO

 

 

I‘Llllllllllllllllllllllllllllllll

0 100 200 300 400 500 600
Temperature (°C)

 

(D
O)

 

 

 

 

B
8 §
.35 8
C CD
3
2‘
g A
lé
5!.
.E‘ B
U)
C
0
E c
11111 111.111.111.1111111111111111.1
5 10 15 20
2theta(9)

Figure 2.7. (A) TGA Of [A1,,O,(OH),4(H,O)1,]003MOS, under nitrogen,
showing gradual weight loss with an inflection point near 300°C. (B) XRD
patterns of the material (a) at room temperature (b) after heating to 120°C
and (c) after heating to 300°C.

62

collapsed layers at the edges of the particles may inhibit nitrogen
absorption.

After conducting surface area measurements on a heated sample
(120°C for 18 hours), 27Al solid state MAS-NMR was performed on the
product. Although X-ray powder diffraction indicated little change in the
d—spacing of the material, the resonance at 64 ppm due to the tetrahedral
Al3+ atoms disappeared, indicating that the structure of the intercalated
species was altered, possibly through oligomerization. This also could
result in blockage of void space and may account for the observed low

surface areas.

2.4 Conclusions

We have seen through X-ray diffraction, MAS-NMR, and EDS that
[A1,,O4(OH),4(H,O),,]7" can be encapsulated in MOS, and WS,. The cation
appears to intercalate without its anion, suggesting a negative charge on the
layers. Electron density mapping and Rietveld powder refinement show
that the cluster maintains its structural integrity while encapsulated, and it
is oriented with its C, symmetry axis perpendicular to the layers. The
materials are conductive and the layers remain expanded up to 300°C;

however, 27Al solid state MAS-NMR of the heated material indicates that

63

the cluster has decomposed. Surface area measurements indicate that void

space in the material (if any exists) is inaccessible.

References
1. Barter, R.M.; MacLeOd,D.L. Trans. Faraday Soc., 1955, 51, 1290.

2. Barrer, R.M. “Zeolites and Clay Minerals as Sorbents and Molecular
Sieves,” Academic Press, New York, 1978, 407.

3. (a) Pinnavaia, T.J. “Characterization of Catalytic Materials,” Wachs,
I.E. (Ed.), Manning Publications Co., Greenwich CT, 1992, 149. (b)
Johnson, I.D.; Werpy, T.A.; Pinnavaia, T]. J. Am. Chem. Soc., 1988,
110, 8545.

4. Fleischauer, P.D. Thin Solid Films 1987, 154, 309.

5. (a) Harris, 8.; Chianelli, R.R. J. Catalysis 1984, 86, 400. (b) Weisser,
O.; Landa, S. “Sulfided Catalysts, Their Properties and Applications,”
Pergamon, New York, 1973. (c) Prins, R.; De Beer, H.J.; Samorjai,
G.A. Cat. Rev. -Sci. Eng., 1989, 32, 1.

6. (a) Julien, C.; Saikh, S.I.; Nazri, G.A. Mater. Sci. Eng., 1992, B15,
73. (b) Tributsch, H. Faraday Discuss. Chem. Soc., 1980, 70, 190.

7. (a) Bissessur, R.; Kanatzidis, M.G.; Schindler, J .L.; Kannewurf, C.R. J.
Chem. Soc. Chem. Commun., 1993, 1582. (b) Kanatzidis, M.G.;
Bissessur, R.; De Groot, D.C.; Schindler, J.L.; Kannewurf, C.R. Chem.
Mater., 1993, 5, 595. (c) Wang, L.; Schindler, J.L.; Thomas, J.A.;
Kannewurf, C.R.; Kanatzidis, M.G. Chem. Mater., 1995, 7, 1753.

8. Lemon, J .P.; Lerner, M.M. Chem. Mater., 1994, 6, 207.

9. (a) Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. Science, 1989,
246, 371. (b) Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. J.
Mater. Res., 1991, 6, 1103.

10. Tigaya, H.; Hashmoto, T.; Karasu, M.; Izumi, T.; Chiba, K. Chem.
Lett., 1991, 2113.

11. Bissessur, R.; Heising, J.; Hirpo, W.; Kanatzidis, M.G. Chem. Mater.,
1996, 8, 318.

12. a) Rudorff, W. Chimia, 1965, 19, 489. b) Bayer, E.; Rudorff, W. Z.
Naturforsch, 1972, 27b, 1336. c) SchOllhorn, R.; Bethel, J.; Paulus,

65

13.

14.

15.

16.

17.

18.
19.
20.

21.
22.

23.

24.

W. Rev. Chim. Miner., 1984, 21, 545. d) Miremadi, B.K.; Morrison,
S.R. J. Appl. Phys, 1988, 63, 4970.

(a) Lerf, A.; Lalik, E.; KOlOdziejski, W.; Klinowski, J. J. Phys. Chem,
1992, 96, 7389. (b) Nazar L.F.; Jacobson, A.J. J. Chem. Soc. Chem.
Commun., 1986, 570. (c) Nazar, L.F.; Jacobson, A.J. J. Mater. Chem,
1994, 4, 149.

Brenner, J.; Marshall, C.; Ellis, L.; Tomczyk, N.; Heising, J.;
Kanatzidis, M.G. Chem. Mater., 1998, 10, 1244.

Plee, D.; Borg, F.; Gatinau, L.; Fripiat, 1.]. J. Am. Chem. Soc., 1985,
107, 2362.

a) Murphy, D.W.; DiSalvo, F.J.; Hull G.W., Jr.; Wasczak, J.V. Inorg.
Chem, 1976, 15, 17. b) Tsai, H.L.; Heising, J.; Schindler, J.L.;
Kannewurf, C.R.; Kanatzidis, M.G. Chem. Mater., 1997, 9, 879.

Joensen, P.; Frindt, R.F.; Morrison, S.R. Mater. Res. Bull., 1986, 21,
457.

Pu, G.; Nazar, L.F.; Bain, A.D. Chem. Mater., 1991, 3, 602.
Tsai, P.P.; Hsu, P.H. Soil. Sci. Soc. Am. J., 1985, 49, 1060.

Brunauer, S.; Emmett, P.H.; Teller, Z. J. Am. Chem. Soc., 1938, 60,
309.

Heising, J.; Tsai, H.L.; Kanatzidis, M.G. unpublished results.

Gil, A.; Diaz, A.; Montes, M.; Acosta, D. R. J. Mater. Sci., 1994, 29,
4927.

a) Yang, D.; Sandoval, J. J.; Divigalpitiya, W.M.R.; Irwin, J.C.; Frindt,
R.F. Phys. Rev. B, 1991, 43, 12053. b) Py, M.A.; Haering, R. R. Can.
J. Phys, 1983, 61, 76.

a) Liu, Y.J.; Schindler, J.L.; DeGroot, D.C.; Kannewurf, C.R.; Hirpo,

W.; Kanatzidis, M.G. Chem. Mater., 1996, 8, 525. b) Liu, Y.J. Ph.D.
dissertation, Michigan State University, East Lansing, MI, 1994.

66

25.

26.

Pinnavaia, T.J.; Tzou, M.S.; Landau, S.D.; Raythatha, R.J. J. Mol.
Catal., 1984, 27, 195.

(a) Rietveld, J. Appl. Cryst., 1969, 2, 65. (b) Rodriguez-Carvajal, J.
“Abstracts of the Satellite Meeting on Powder Diffraction of the XV
Congress of the IUCr,” Toulouse, France, 1990, 127.

27. (a) Reynolds, R.C. Clays and Clay Minerals, 1986, 34, 359. (b) Bish,

28.

29.
30.
31.

D.L.; Post, J.E. (Eds), “Modern Powder Diffraction, Reviews in
Mineralogy,” 20, The Mineralogical Society of America, Washington
DC, 1989, Chapter 6. (d) Sheldrick, G.M.; Kruger, C;. Goddard, R.
(eds), “Crystallographic Computing 3: Data Collection, Structure
Determination, Proteins, and Databases,” Clarendon Press, Oxford,

1985, 247.
(a) Young, R.A.; Wiles, D.B. J. Appl. Cryst., 1982, 15, 430. (b)
Cagliotti, G.; Pelotti, A.; Ricci, R.P. Nucl. Instrum. Methods, 1958,

35, 223. (c) Berar, J.P.; Baldinozzi, G. J. Appl. Cryst., 1993, 26,
128.

Durbin, J .; Watson, G.S. Biometrika , 1971, 58, 1.
Berar, J .F.; Lelann, P. J. Appl. Cryst., 1991, 24, l.

Nazar, L.F.; Liblong, S.W.; Yin, X.T. J. Am. Chem. Soc., 1991, 113,
5889.

67

CHAPTER 3

Encapsulation of Neutral and Cationic Metal Chalcogenide
Clusters in M08, and WS,

68

3.0 Abstract

The synthesis of the compounds [C06S8(PPh,)6],MoS, (x = 0.02-
0.094), [FeéS,(PEt,)y],MS, (x = 0.05-0.2, y = 2-3, M = Mo, W), and
[Ni,S,(PEt,)y],MS, (x = 0.040077, y = 6, M = M0, W) was accomplished
by a precipitative encapsulation technique using single layers of MS,. The
products were characterized by X-ray diffraction, energy dispersive
spectroscopy (EDS), High Resolution Transmission Electron Microscopy
(HRTEM), Thermogravimetric analysis (TGA), direct probe mass
spectrometry, room temperature electrical conductivity, and
hydrodesulfurization catalytic testing (HDS). Powder X-ray diffraction
patterns show an expansion of ~15.3 A for [C068,(PPh,)6]o_o,MoS,, 10.5 A
for [Fe,s,(PEr,),],,,6Mos,, 11.4 A for [Fe6S,(PEt,),]O,WS,, 14.2 A for
[Ni,S,(PEt,)6]omMOS,, and 14.0 A for [Ni,s,(PE1,),],,,,ws,. Variation in
cluster loading can result in different expansions. HRTEM micrographs
show evidence of localized pillaring in [C0688(Pph3)6]0_O,MOS,. The
products exhibit metallic conductivity originating from the metallic nature
of the MS, host. TGA of the samples show weight loss at just below 300°C
for [C068,(PPh3)6]o.onoS,, 140°C and 150°C for [Fe688(PEt,),],o,6MOS, and
[Fe6S,(PEt,),]O,WS,, respectively, and 160°C for [Ni,S,(PEt,)6]0_O,MOS,.
The X-ray powder patterns become amorphous close to the temperature at

which weight loss takes place; eventually the layers collapse in all

69

materials. The unencapsulated C06S,(PPh,)6 and [FeGS,(PEt,)6](BPh4),
clusters exhibit weight loss due to departure of the phosphine ligands,
whereas [Ni,S,(PEt,)6] (C104), loses weight due to loss of triethyl phosphine
sulfide. In all encapsulated samples the volatile product was determined by
mass spectrometry to be phosphine sulfide. [C06S,(PPh,)6]O.O,MoS, was
found to be as active as the commercially available catalyst Crosfield 465
(CO-Mo) at normal reaction temperatures, and more active at lower
temperatures; however, the product was less selective toward desired
byproducts. Catalytic activity Of [Ni.,S,(PEt,)6],04MOS, and
[Ni,S,(PEt,)6]o.04WS, was found to be inferior to Crosfield 504 (Ni-MO)
under all conditions tested. The HDS behavior of these samples has

interesting implications for the “rim-edge” model for HDS selectivity.

70

3.1 Introduction

Zeolites', pillared claysz, and more recently materials such as MCM-
413 have dominated modern developments in micro and mesoporous
materials. These open framework oxides, however, are insulators. The
pursuit Of non-oxidic porous materials, particularly open framework
chalcogenides, is driven by a vision of materials which combine porosity
with semiconducting or metallic electronic and optical properties.
Strategies to create such a material have mainly involved self-assembly
through hydrotherrna14'5'6, molten salt7, and high temperature8 synthesis and
have led to a rich diversity of new compounds.9 Recently very promising
zeolite-like indium sulfides10 have been synthesized through this approach,
as well as mesostructured metal germanium sulfides through
supramolecular self-assembly.”

An alternate strategy to create a microporous chalcogenide is
through pillaring. Many of the binary sulfides adopt a layered structure
and can be intercalated with guest species.12 The availability and relatively
low cost of MOS, make it an especially desirable chalcogenide host. MOS,
has been intercalated with a variety of compounds which include

13,14

polymers , small organic molecules”, ferrocene",

[A1,,O4(OH),4(H,O),,]7* 17and the cobalt clusters Co,,Q,(PR,)6 (Q=S, Se, Te;

71

R=alkyl)'8‘ WS, is isostructural to MOS,, with similar intercalation
chemistry, although less well explored.19

The encapsulation of oxidic clusters such as [A1,,O,,(OH),,,(H,O),,]7+
into hosts MOS, and WS, was discussed in the previous chapter and
elsewhere.17 The only precedents for the combination of lamellar
dichalcogenides with sulfide pillaring agents are the intercalation of the
iron cluster [FeGS,(PEt,)6]2+ in TaS,20 and the pillaring of MOS, with the
cobalt clusters Co,,Q,(PR,),S (Q=S, Se, Te; R=alkyl) which was initiated in
this laboratory.18 The following data are a continuation and an expansion

of this work.

3.2 Experimental

3.2.1 Synthesis

Literature procedures were used in the preparation of Co,,S,(PPh,),,,21
[Fe,,S,(PEt,),,](BPh,,),,22 and [Ni,,S,(PEt,)6](ClO.,),.23 Chemicals used were
Na,S (Columbus Chemical Industries), PPh, (Baker), CoBr, (Aldrich),
dimethylformamide (dmf, Aldrich), H,S (Aldrich), [Fe(I-I,O)6](BF4), (Alfa
Aesar), PEt, (Aldrich), dichloromethane (CH,C1,, Aldrich), ethanol (EtOH,
Aldrich), NaBPh4 (Lancaster), and Ni(H,O)6](ClO4), (Alfa Aesar). LiMS,

72

(M = M0, W) was prepared as described in Chapter 2, or by a literature
method.24 Chemicals used were 2H-MS, (Cerac, MOS,; Alfa Aesar, WS,),
n-butyl lithium (Aldrich), hexane Aldrich, and CaH, (Fisher).

MS, aqueous suspension. Method 1. LiMoS, (0.1 g, 0.6 mmol)
was placed in a 125 ml flask equipped with a stir bar under a nitrogen
atmosphere and capped with a rubber septa. The flask was removed from
the glove box and 10 ml deionized H,O was introduced. This results in the
exfoliation of MOS, by a redox reaction described previously (Chapter 2,
equation 1), generating single layers in suspension, H, gas, and LiOH. The
reaction was then sonicated for 1.5 hours. Method 2. The preparation
was identical to method 1 (WS, = 0.1 g, 0.2 mmol) until H,O was
introduced; then, instead of sonicating, the mixture was stirred for a half
hour. The contents of the flask were then transferred to a centrifuge tube
and centrifuged for a half hour. The supernatant was decanted, 10 more
ml H,O was added to the product to rinse away the excess LiOH, and the
mixture centrifuged again. This rinsing was repeated two more times. To
the product 10 more ml H,O was added, the contents of the centrifuge tube
were returned to the 125 ml flask, and stirred for a half hour prior to use.

[C068,(PPh,)6],MOS,. To an aqueous suspension of MOS, (prepared
by method 1) was added a solution of various amounts (ranging from 1/50
to 1/16 cluster:MoS, molar ratio) of Co,,S,(PPh,)6 in CH,C1,. The flask was

shaken to help promote the formation of an emulsion; then stirred for two

73

days. The product was isolated by filtration and rinsed with H,O to
remove excess LiOH and CH,Cl, to remove unencapsulated cluster.
[Fe6S,(PEt,)y],MS,. An aqueous suspension of MS, (prepared by
method 2) was added to a solution of various amounts of
[Fe6S,(PEt,)6](BPh4), in acetone (Aldrich) (example: 0.12 g
[Fe6S,(PEt,)6](BPh4),, 0.06 mmol in 50 ml for MOS,; 0.46 g
[Fe6S8(PEt,)6](BPh4),, 0.24 mmol in 80 ml acetone for WS,). Flocculation
occurred immediately. The reaction mixture centrifuged and the
supernatant decanted. The product was rinsed with acetone until colorless,
then rinsed with a little H,O and deposited on a glass slide to dry.
[Ni,S,(PEt,)6],MOS,. One fifth of an aqueous suspension of MOS,
(prepared by method 2: ~ 0.02 g LiMoS,, 0.12 mmol) was added to a
solution of [Ni.,S,(PEt,)6](ClO,), (0.0516 g, 0.03 mmol) in 5 ml of dimethyl
sulfoxide (dmso, Aldrich), acetonitrile (MeCN, Aldrich), or dmf.
Flocculation occurred immediately. The reaction mixture was centrifuged
and the supernatant decanted. The product was rinsed with the solvent used
to dissolve the cluster (dmso, MeCN, or dmf) until colorless, then rinsed
with a little H,O and deposited on a glass slide to dry. Best results were
achieved with dmso. Larger scale preparations used to prepare samples for
HDS measurements (in dmso) employed 1.2 mmol MOS, and 0.2 mmol of

[Ni989(PEt3)6](C104)2'

74

[Ni,S,(PEt,)6],WS,. The reaction procedure was similar except that
dmso and acetone were used (0.2 mmol WS,, 0.02-0.09 mmol
[N198,(PEt,)6](ClO4),, 25-50 ml acetone). Best results were achieved with
acetone. Larger scale preparations used to prepare samples for HDS
measurements (in dmso) employed 2 mmol WS, and 0.2 mmol of

[Ni989(PEt3)6](ClO4)2'

3.2.2 Characterization

Powder X-ray diffraction patterns, HRTEM micrographs, Energy
Dispersive Spectroscopy (EDS) measurements, Thermogravimetric
Analyses (TGAs), Surface Area measurements, and conductivity values
were collected as described previously (Chapter 2, section 2.2.2).
Hydrodesulfurization (HDS) measurements were conducted by Dr. James
Brenner or Dr. Di Wei (post-doctoral associates of Dr. Chris Marshall,
Chemical Technology Division) at Argonne National Laboratory using a
pilot plant scale HDS unit designed with the help of Dr. John Young of EC
Technologies. Conditions were chosen so as to mimic those of an industrial
middle distillate desulfurization unit. Weight loadings of samples (0.1 g)
and the commercial catalysts (1.0 g) have been adjusted so as to make the

total metal (CO or Ni + Mo or W) loadings equivalent. Liquid velocity was

75

0.25 g/minute (or 0.33 g/minute for [Ni,S,(PEt,)6]O.04MS, samples); gas
velocity was 700 cm3/min; ratio of H,/H,S/N, was 5/0/2; pressure was 400
psig; furnace temperature was BOO-400°C; tabular alumina diluent was 2.5-
3.0 g; and liquid feed was 1.0 weight % sulfur as dibenzothiophene in
hexadecane (or 0.75 weight % sulfur feed for [Ni,S,(PEt,)6]o_O,MS,
samples). Reported values were collected at a Liquid Hourly Space
Velocity (LHSV) of 20/hr. Prior to catalytic testing the samples were
purged in N, at 20°C for 30 minutes, dried in N, at 150°C for 60 minutes
and at 200°C for 60 minutes, then heated in H, for 2 hours at 200°C. The
commercially available materials were pre-treated in an analogous fashion
except the second drying step was at 400°C and the final heating step was in

an 5% H,S/H, mixture at 400°C.

76

3.3 Results and Discussion

3.3.1 Evidence of Cluster Encapsulation

[CO6S,(PPh,)6],MOS,. X-ray diffraction patterns exhibiting an
expansion of the MOS, layers due to encapsulation of the CO,S,(PPh,)6
cluster can be found in Figure 3.1. As noted during the discussion of the
diffraction patterns in Chapter 2, the 001 reflections predominate due to the
tendency of lamellar materials to adopt a preferred orientation. These
patterns exhibit d-spacings around 21.6 A (Figure 3.1a, pattern a), which is
a 15.4 A expansion over the d-spacing of restacked MOS,, 6.2 A (Figure
3.1a, pattern b). This expansion is consistent with the dimensions of the
cluster. The patterns from contain very broad and weak peaks, in contrast
to the diffraction patterns found for [Al,,O4(OH),4(H,O)1,]0.05MOS,. This
seems to be particularly true for samples with lower loadings of the cluster
(x = 0.02-0.03). EDS measurements indicate a loading of
[COGS,(PPh,)6],O,,MoS, for the sample found in Figure 3.1a, pattern b, and
[Co,S,(PPh,)6]0.o,4MoS, for the sample in Figure 3.1b.18c Samples with
higher loadings of the cluster (x z 0.10) give sharper powder patterns, but
are still not as ordered as those found for the encapsulation Of cationic

species. Only 4 001 reflections are observed in Figure 3.1b, and at higher

77

(A)

 

 

 

 

 

 

 

 

 

 

 

 

5
o
E N ('0
s e e
b
‘3
e: A
E
5!
.3
(I)
C
d)
.15
5 10 15 20
29 (degrees)
(B)
N
A O
a 5 °
'15 o
3
E;
1!
3§
5!
.é‘
(D
C
o
E
llllllljglLLIIII

 

 

 

0 1 0 2 0 3 0 40 5 0 6 0
20 (degrees)

Figure 3.1. X-ray diffraction patterns of (A) [C0638(PPh3)6]0_,4MOS, and
restacked MoS,, showing an expansion of 15.4 upon encapsulation of the
cluster (B) [CO6S,(PPh,)do.%MoS,, showing only 4 001 reflections (in
contrast to 18 001 in [A1,,O4(OH),,,(H,O),,]0_05MOS,) for even high loadings
of the cluster.

78

 

Figure 3.2. The Co,S,,(PPh,)6 cluster oriented with its (A) C, axis
perpendicular to the layers and its (B) C4 axis perpendicular to the layers.
Expansion in (B) is slightly larger than in (A). Black circles represent Mo
or W atoms; large light gray circles, S atoms; dark gray circles, Co atoms;
small light gray circles, P atoms; open circles, C atoms.

79

 

Figure 3.3. HRTEM micrographs of [C0688(PPh3)5]o_02M082. (A Region
with lamellar features showing expanded and unexpanded (white arrow)
layers. Small scattering centers may be attributed to CO6S3 cores (black
arrows). (B) Region showing in-plane structure which contains dark grey
circles (white arrows) which may be attributed to CO6S8 cores.

80

20 values the hk0 reflections predominate. Although some of the
broadening of the peaks might be attributed to small particle size, disorder
probably contributes to the broadening as well.

A rather broad range of expansions (from 14 - 15.4 A) has been
observed for these samples'8 which might be explained by a variety of
factors. If the cluster is oriented with one of its C, axes perpendicular to
the layers the expansion will be at least 14 A; if one of its C4 axes is
perpendicular to the layers the expansion should be at least one angstrom
greater (Figure 3.2). Also, the phosphine ligands in the Co,,S,(PPh,)6
cluster contain phenyl rings which, although not very flexible compared to
other phosphine ligands, may move slightly to alter the dimensions of the
cluster. Finally, unlike the 12 H,O ligands in the [Al,,O,(OI-I),4(l-I,O),,]7”
cluster, some of the bulky triphenyl phosphine ligands may detach from the
cluster. The partially deligated species would have noticeably smaller
dimensions.

Room temperature conductivity measurements on pressed pellets of
the materials revealed metallic behavior, with conductivity values around
2-3 S/cm. Lithiation results in a structural transformation of the MS, from
trigonal prismatic coordination about the metal atoms to octahedral
coordination” which is retained in restacked MS, (vide infra). The
structural transformation results in a change in the electrical properties

from semiconducting to metallic behavior.

81

HRTEM micrographs acquired at the edges of [C0688(PPh,)6]OmMOS,
particles dusted on a carbon-supported copper grid are found in Figure
3.3. In Figure 3.3a an intimate mixture of expanded and un-expanded
lamellar regions is visible. The observed d-spacing in the expanded region
(13.9 A, A = 7.7 A) is smaller than that observed by transmission X-ray
diffraction (21.5 A), which may be attributed to loss of the phosphine
ligands and is consistent with the dimensions of the C068, core (~8 A). In
the expanded region it is possible to see weak scattering centers which may
be interpreted as C068, cores (black arrows). Phosphine ligands, even if
present, would be too light to scatter sufficiently to be visible. The fact
that expansion of the layers is not continuous over the entire sample
suggests that pillaring is for the most part localized. In Figure 3.3b a view
Of the in-plane structure is visible. Superimposed on the scattering due to
MO and S centers are large, spherical scattering centers which may be
attributed to in-plane Co,,S,(PPh,)x species (white. arrows). This implies
that the clusters are dispersed throughout the material, as indicated by X-

ray diffraction measurements.

[Feés,(PEt,)6],MS,. X-ray diffraction patterns of Fe,,S,,(PEt,)62+
encapsulated in MOS, and WS, are shown in Figure 3.4. The patterns are
more well-ordered than those for the neutral Co,,S,(PPh,),5 cluster,

exhibiting 10 001 reflections in the MoS,—containing sample and 15 001

82

reflections in the WS,-containing sample. The d-spacing of [Fe6S,(PEt,)6]2‘°
in MOS, is 16.7 A, corresponding to an expansion of 10.5 A (Figure 3.4a),
whereas the d-spacing of [Fe6S,(PEt,)6] 2* in WS, is 17.6 A, corresponding
to an expansion of 11.4 A (Figure 3.4b). These expansions are consistent
with the dimensions of the cluster (~10.3 A along the C, axis; ~11.7 A
along the C4 axis). The orientation of the cluster between the layers is not
known; however, the cluster was Observed to be oriented with its C, axis
perpendicular to the layers upon intercalation into TaS, and gave rise to an
11.5 A expansion.20

EDS measurements indicated that the chemical formulas of the
compounds shown in Figure 3.4 are [FeGS,(PEt,),]o_096MoS, and
[Fe6S,(PEt,),]0_,WS,. There is no evidence of co-intercalation of the BPh,‘
anion. These represent the upper limit of observed loadings. The values
are slightly higher than those observed for the maximum loadings of the
Co,,S,(PPh,)6 cluster, which is consistent with the smaller dimensions of the
Fe,,S,(PEt,),,2+ cluster. The maximum observed loading of the
[Al,,O,,(OH),,,(H,O),,]7+ cluster suggested a negative charge as high as
0.35/0.38 per Mo/W atom on the layers; the maximum loadings observed
here give rise to comparable negative charge on the layers (0.2/0.4 per
Mo/W atom). However, this is not a reliable way to determine the charge
per Mo/W atom as the value may be artificially inflated due to partial

decomposition of either cluster.

83

(A)

 

001
002

 

 

 

 

0010

Intensity (arbitrary units)

0)
mo
00
O

llllllLllll JLII 111111111

1 0 2 0 30 4 0 5 0 6 0
29 (degrees)

 

 

 

(B)

 

 

 

 

 

 

  

P
8
A
(I)
a:
C
3 tn
2‘ o
w o '-
L. O'-
a: ,.<3
9 8 0°
to 0 °
v
a: V
I— o
0) O
8
u:
E 8§Sgg
All llLllllllJ

 

 

 

0 2 0 4 0 6 0 8 0 1 0 0
20 (degrees)

Figure 3.4. X-ray diffraction patterns of (A) [Fe,,S8 PEt,),]o.o,6MoS, (A =
10.5 A) and (B) (Fe,s,(1>131,),],,,ws, (A = 11.4 ). Both exhibit a
predominance of 001 reflections and greater order than the patterns for
[Co,S,(PPh,)6],,6MOS,,but less order than for [All,O4(OH),4(H,O)1,]0.O,MOS,.

84

 

er ._,

Figure 3.5. HRTEM micrograph Of [Fe583(PEt3)x]y]o,2WS2, showing
highly ordered in-plane structure. Large dark grey areas could be attributed
to aggregated [Fe683121’ cores.

85

The triethyl phosphine ligands are more flexible than the triphenyl
phosphine ligands because they are less hindered sterically; hence, a wider
range of d—spacings are possible for any given orientation of the cluster.
However, given the dimensions of the cluster, it is impossible to Observe
these loadings in WS, (5 W atom per cluster) without loss of some of the
triethyl phosphine ligands. It is barely possible to retain them all in the
loading observed in MOS, (10.4 MO per cluster). The number of
phosphines ligated to the clusters will be discussed in conjunction with the
thermal analysis of the materials.

Room temperature conductivity measurements on thin films of the
materials indicated metallic behavior, with conductivity values around 0.4
S/cm.

A HRTEM micrograph acquired at the edge of [Fe68,(PEt,),]o_,WS,
particles dusted on a carbon-supported copper grid shows the in-plane
structure, Figure 3.5. The in-plane structure in this sample is significantly
more ordered than the structure visible in Figure 3.3b, and can be indexed
as 110 planes. NO discrete scattering centers due to [FeGS,,]2+ cores are
observed; however, larger dark gray regions are visible which may be due
to aggregation of [Fe,,S,,]2+ cores. Aggregation would be consistent with the
higher loading of [Fe,,S,,(PEt,),,]2+ clusters found in this sample (0.2 per W

atom) than Co,,S,(PPh,)x cores in MOS, (0.02 per Mo atom).

86

[Ni,S,(PEt,)6],MS,. X-ray diffraction patterns of Ni,S,,(PEt,),,2+
encapsulated in MOS, and WS, are found in Figure 3.6. As with the other
cationic clusters, the patterns are more well ordered than those from
samples containing the neutral Co,,S,,(PPh,)6 cluster. Fourteen 001
reflections are Observed for are [Ni,S,(PEt,)6]2+ in MoS,, which exhibits an
expansion of 14.2 A; and nine 001 for [Ni,S,(PEt,)6]2+ in WS,, which is
expanded by 14.0 A. The dimensions of the cluster are approximately 13
A perpendicular to and 11.5 A parallel to the Ni,S, planes (Figure 3.7).

EDS measurements indicated that the chemical formulas of the
compounds shown in Figure 3.6 are [Ni,S,(PEt,)6]0mMoS, and
[Ni,S,(PEt,)6]0m7WS,. There is no evidence of co-intercalation of the C10,;
ion. These loadings represent the upper limit; lower loadings (down to
~0.04) have been observed. The maximum charge per Mo/W atom is
0.14/0.15, which is lower than that observed for samples containing the
[Fe,,S,(PEt,)6]2+ and the [Al],O,,(OH),,,(H,O),,]7+ clusters. The dimensions,
coupled with the observed loadings (13-14 Mo/W atoms per cluster),
suggest that the clusters may be packed in the gallery space with the Ni,S,
planes parallel to the layers. It is possible to accommodate the cluster in
this loading without loss of phosphine ligands. At lower loadings smaller

d-spacings have been observed (10.5 A - 13 A).

87

(A)

 

 

 

 

 

 

 

 

 

 

 

1- N
O O
O O
A
U)
1‘:
C
3
S
:3 CD
._ o [x
e o 3
CU Vt!)
" 88 2.2 - a
,3: Coco'- Pro
. CO C v- V
8 ° 8 ° 3
a) o 0
fl
5
1111111111111141111111194411’111
10 20 30 40 50 60 70
29 (degrees)
'- N
0
1° 3
A
(D
.2
C
3
Z‘
.2
E
N
3 8 8 no
0
E‘ ,9 ° 8
a o °
m 0
fl
5
llllLJlLALLLJllllll lLJJlllllli

 

 

 

51015 20 25 30 35 40
29 (degrees)

Figure 3.6. X-ray diffraction patterns of (A) [Ni,S,(PEt,)6]omMOS, (A =
14.2 A) and (B) [Ni,s,(PE1,),],,,,ws, (A = 14.0 A). Again the
predominance and number of observed 001 reflections indicates a greater
degree of order than in [C068,(PPh,)6]o.,6MOS,.

88

(B)

 

Figure 3.7 . A view of the [Ni,,S,,(PEt,),,]2+ cluster with its Ni,S, planes
(A) parallel to and (B) perpendicular to the MS, layers. Orientation (A)
gives rise to a larger expansion which is consistent with the d—spacing
Observed in the X-ray diffraction pattern.

89

3.3.2 Thermal Analysis

[Co,S,(PPh,)6]o.o,,MOS,. The TGA of unintercalated Co,,S,(PPh,)6
(Figure 3.8a) shows a clean weight loss, beginning at about 250°C,
corresponding to the departure of the triphenyl phosphine ligands
(expected weight loss, 72%; observed, 70.6%). The identity of the volatile
product was confirmed by direct probe mass spectrometry, and the
nonvolatile decomposition product is CoS.'8 In contrast, the TGA of
[C0,,S,(PPh,)6]0.0,,MOS2 shows a broad, continuous weight loss, beginning at
about 220°C, to 650°C. Our hope was that the triphenyl phosphine ligands
could be removed from the cluster by thermal treatment, resulting in a
microporous sulfide (Figure 3.9). If all six phosphine ligands are still
attached to the cluster core, the expected weight loss due to the departure
of triphenyl phosphine ligands is 16.6%; the Observed weight loss is 18.9%.
Although additional weight loss might be explained by co-encapsulated
solvent, direct probe mass-spectrometry indicates that the volatile product
is actually triphenyl phosphine sulfide. Predicted weight loss for triphenyl
phosphine sulfide is 18.7%. In fact, when the selenide analog
Co,,Se,,(PPh,)6 is encapsulated the volatile product is still triphenyl
phosphine sulfide. This surprising result suggests that the phosphine

ligands are extracting sulfur from MOS, as they depart (Figure 3.10).

90

(A)

 

120

100 -

80-

60*-

°/e weight loss

40L

 

 

lllll[Ill—LlJllllllllllell—LJ

 

20 “4‘
0 100 200 300 400 500 600
Temperature (°C)

(3)

 

105

100 -

90+

% weight loss

 

 

 

80 l I l l l l l l l l l l l l l l l l I l l l l L l l l l l l l l
0 100 200 300 400 500 600
Temperature (°C)

Figure 3.8. TGA of (A) CO6S,(PPh,)6. The PPh, ligands depart in a clean
step just below 300°C (expected weight loss, 72%; observed, 70.6%). (B)
[Co,S,(PPh,)6]0,,MoS,. The PPh, ligands begin to depart at a comparable
temperature, but the weight loss is gradual to 650°C (predicted weight loss
due to PPh, departure, 16.6%; observed, 18.9%).

91

 

 

HEAT TREATMENT

 

 

 

Figure 3.9. An illustration of the original plan to create a microporous
sulfide: encapsulation of the cluster with phosphine ligands acting as
“spacers”, followed by thermal treatment to remove the phosphine ligands.

92

 

HEAT TREATMENT

 

 

 

Figure 3.10. An illustration of the observed data: encapsulation of the
cluster with phosphine ligands acting as “spacers”, followed by thermal
treatment. The phosphine ligands attack the layers as they depart, forming
SPR3 species and possible 82' anion vacancies.

93

 

Intensity (arbitrary units)

 

 

20 (degrees)

Figure 3.11. X-ray diffraction patterns upon thermal treatment of
[CO6S,(PPh,)6]0,4MOS,. [C068,(PPh,)6]0,4MOS, becomes amorphous upon
heating to 250°C, then collapses to a d-spacing only slightly larger than
restacked MOS, by 650°C (b).

94

X-ray diffraction studies upon thermal treatment show that the
material becomes amorphous after onset of the removal of the phosphine
ligands; if heated to 650°C, the material collapses to a product with a d-
spacing only slightly larger than restacked MOS, (Figure 3.11). It is likely
that, as vacancies are created in the MOS, layers by the departing phosphine

ligands, sulfur atoms are transferred from the [C0688] cores:

[CO6S,(PPh,)6],MoS, --> [C06S,]XMOS,,Y --> [C0,,s,,_,],Mos2 (Eq. 3.1)

[FeGS,(PEt,),],MS,. The TGA of [Fe68,(PEt,)6](BPh4), (Figure
3.12a) shows weight loss, beginning shortly after 100°C, due to the
departure of the BPh,‘ anions as well as the triethyl phosphine ligands
(expected weight loss, 69.5%; observed, 69.2%). The majority of the
triethyl phosphine appears by mass spectrometry to depart as a non-sulfide
product. [Fe6S,(PEt,),]o.o,6MoS, and [Fe6S,(PEt,),]o,WS,, however, show a
gradual weight loss starting at about 140°C for the MoS,-containing sample
and 150°C for the WS,-containing sample (Figure 3.12b). No BPh,‘ anions
were detected by mass spectrometry. If all six phosphine ligands were
present in each sample, and the volatile product was only triethyl
phosphine, the predicted weight loss for each compound would be 23.9%
and 28.0%, respectively; the observed values are 12.7% for

[FeGS,(PEt,),]O.096MOS, and 18.7% for [Fe6S,(PEt,)x]o_,WS,. It seems that, as

95

predicted based on the loadings and dimensions of the cluster, the [Fe,,S,]2+
core is not fully ligated. Furthermore, direct probe mass spectrometry
revealed that the majority of the volatile product is triethyl phosphine
sulfide, analogous to the behavior of the triphenyl phosphine ligand in
[Co,S,(PPh,)6]O.O,,MoS,. For [Fe6S,(PEt,)x]o_096MOS,, if there are two triethyl
phosphine ligands still attached to the cluster the predicted weight loss (due
to SPEt, loss) is 12.0%. For [FeéS,(PEt,),]O_,WS,, if there are three
triphenyl phosphine ligands still attached to the cluster the predicted weight
loss (again due to SPEt, loss) is 20.7%. Loss of three phosphine ligands
was Observed upon encapsulation Of the [Fe,,S,(PEt,)6]2+ cluster in TaS,.20
X-ray diffraction patterns probing the thermal stability of
[FeGS,(PEt,),]O.o,6MOS, are found in Figure 3.13. The material remains
pillared up to 100°C (13b), the temperature at which MOS, converts from
an octahedral geometry about the ”Mo atom to a trigonal prismatic
coordination environment. At 150°C, after which weight loss has begun,
the powder pattern is amorphous (13c). By 200°C the pillars have
collapsed, resulting in Fe,MoS,, a material with a d-spacing only slightly
larger than restacked MOS2 (13d). [Fe6S,(PEt,),]0,WS, exhibits slightly
better thermal stability. At 150°C the material is still expanded; however,

by 210°C partial decomposition to Fe,WS, has taken place.

96

110
100
90
80
70
60
50
40
30

% weight loss

105

100

95

90

% weight loss

85

80

Figure 3.12. TGA of (A) [FeGS,(PEt,)6](BPh4),, which loses weight due
to the departure of the PEt, ligands and the BPh4 anion (expected, 69.5%;

 

 

 

 

 

 

 

 

 

observed, 69.15%)

1111111111111114 111_.J '
0 100 200 300 400 500 600
Temperature (°C)
(3)
\
r \
\ __
.1....1141.I....1....1.LA11~.:
0 100 200 300 400 500 600

Temperature (°C)

and (B) [Fe6S,(PEt,),]O_O,6MOS, (solid line) and
[FeGS,(PEt,),]0,0WS, (dashed line). The volatile species, SPEt,, is lost more

gradually than the PEt, ligands in the unintercalated cluster.

97

 

 

Intensity (arbitrary units)
1
CD >

 

 

 

 

5 1 0 1 5 2 0
29 (degrees)

Figure 3.13. X-ray diffraction patterns upon thermal treatment of
[Fe6S,(PEt,),]0.0,6MoS,. The room temperature pattern (a) resembles the
pattern upon heating to 100°C (b), but becomes amorphous by 150°C (c)
and collapses to FexMOS, (6.3 A) by 200°C ((1).

98

[Ni,S,(PEt,)6]o.o,.,MOS,. The TGA of unintercalated
[Ni,S,(PEt,)5](ClO4), is found in Figure 3.14a. The material loses most of
its weight in a sharp step beginning at approximately 240°C, but continues
to lose weight gradually until 650°C. Theoretical weight loss due to
departure of triethyl phosphine ligands is 41%; the observed weight loss is
56.6%. Direct probe mass spectrometry indicates that the volatile product
from the thermolysis of the unintercalated cluster is predominantly triethyl
phosphine sulfide; predicted weight loss due to loss of triethyl phosphine
sulfide is 52.3%, much closer to the experimentally observed value.

[Ni9S9(PEt,)6]O.047MoS, also loses weight gradually all the way to
650°C; weight loss begins sooner but the material exhibits a significant
drop around 160°C, a slightly lower temperature than the sharp step
observed for the unintercalated cluster. Direct probe mass spectrometry
indicates that the volatile product from the thermolysis of the
unintercalated cluster is also triethyl phosphine sulfide (predicted weight
loss due to SPEt,, 13.4%; observed, 16.5%). The sulfur source in this case
is probably the [Ni,S,]2+ core, as even the unintercalated cluster loses its
phosphine ligands as SPEt,, leading to layer collapse. Additional weight

loss may be due to co-encapsulated solvent.

99

(A)

100r

 

110

% weight loss

 

 

llllllllll

O 100 200 300 400 500 600
Temperature (°C)

 

(B)

 

105

% weight loss .1
(D co o

o 01 o

I l I

06
(II
I

 

 

lllll

0 100 200 300 400 500 600
Temperature (°C)

 

on

O
i"
L
l—
i-
l—
b

Figure 3.14. TGA of (A) [Ni,S,(PEt,)6](ClO,),. Volatile product, lost in
a clean step before 250°C, is SPEt, (expected weight loss due to SPEt,,
52.3%; observed, 56.6%). (B) TGA of [Ni,S,(PEt,)6]0_0,,WS,. The volatile
product is also SPEt, (expected loss, 13.4%; observed, 16.5%).

3.3.3 HDS Activity

HDS, or hydrodesulfurization, is a catalytic process used to remove
sulfur-containing impurities from crude oil. The removal of these
impurities prevents the formation of sulfur oxides upon combustion, which
are precursors to acid rain. For example, an efficient HDS catalyst

converts dibenzothiophene to biphenyl:

The effectiveness of an HDS catalyst depends not only on its activity, but
also on its selectivity toward certain products. In the desulfurization of
dibenzothiophene, the desired product is biphenyl. An undesirable product

is cyclohexylbenzene:

xs H, + ———> 0—0 + H,S
5 Eq. (3.3)

because it consumes additional H, gas.

The transition metal sulfides are a class of compounds which exhibit
good HDS activity, with rhenium, osmium, and iridium sulfides being
among the most effective.26 Unfortunately, those particular metal sulfides
are rather expensive. It has been known for many years that mixed-metal

sulfide catalysts such as Co or Ni promoted MOS, or WS, (over an alumina

101

support) can provide comparable or better activity” Although the more
reasonable cost of these mixed-metal sulfides has resulted in widespread
commercial use, the mechanism for their activity is not well understood,
and a variety of models have been developed to try to explain their
effectiveness” 29' 30' 3‘

As mentioned previously, the Co,,S,,(PPh,),5 and [Ni,S,,(PEt,),,]2+
clusters were chosen as pillaring agents due to the similarity in chemical
composition of the pillared products to the mixed-metal HDS catalysts.
Most Of the models place the active sites on the surfaces or edges of the
particles; however, a material such as ours, in which the MS, layers have
been propped apart with these clusters, may allow catalysis to take place in
the interior of the particles, thereby enhancing the efficiency of the
catalyst. Accordingly, the products were tested for HDS activity by
reaction with dibenzothiophene (Eq. 3.2). The activities for a number Of
reference materials were collected as well. The results are summarized in
Table 3.1. The selectivity of each catalyst toward biphenyl is reported in
parentheses (as a percentage) next to the overall activity.

The results indicate that [Co,S,(PPh,)6]o.o,MOS, exhibits comparable
catalytic activity to Crosfield 465 (Co-Mo) at 400°C. At 300°C its activity

(38%) is actually better than the commercially available catalyst (20%).

102

Table 3.1. HDS catalytic activity and surface area (SA) of selected
samples and references. Reported (as a function Of temperature) are the
percent conversion of dibenzothiophene to a non-sulfur containing product,
and the percentage of biphenyl in the desulfurized products (in
parentheses)“

 

 

Catalyst 400°C 350°C 300°C 200°C SA
(mi/g)
[C068,(PPh,),,]O.O,MoS2 85 38 15 (7) 26.8
(50) (61)
Crosfield 465 (Co-Mo) 82 56 20 <5
(80) (91) (85)
0.2[COGS,(PPh,)6] + 25 10 <5 <5 9.3
restacked MOS, (73) (76)
[Ni989(PEt3)6]0.04MOS2 58 35 32
(24) (17) (16)
1Ni.S.(PEt.>.1...WS. 44 41 39 (8)
(21) (16)
Crosfield 504 (Ni-MO) 98 58 27 10

(44) (46) (76) (55)
[A1,,O,(OH),4(I-I,O),,]O_O,MOS, 6 (89) <5 <5 <5 10.0

restacked MOS, 22 9 (78) <5 <5 10.0
(75)
2H-MoS, 8 (88) <5 <5 <5 5.8

 

aRun-to-run conversion reproducibility = :1: 15% (e.g., 20 :l: 3%). Amount
of “active” metals (Co + Mo) in each run = 0.1 g in all cases.

103

Unpromoted 2H-MOS, is clearly an inferior catalyst, with an activity of
only 8% at 400°C. Restacked MOS,, although better than 2H—MOS, (22%),
is still not as effective as [C06S,(PPh,)6]0,O,MOS,. In addition to these
reference samples, a material consisting of a physical mixture of restacked
MOS, and the Co,,S,(PPh,),5 cluster in a ratio comparable to the
stoichiometry of [CO6S,(PPh,)6]O_O,MoS, was prepared and tested
(0.2[Co,S,(PPh,)6] + restacked MoS,). The results are only slightly better
than restacked MOS, (25%), verifying that the increased catalytic activity in
[CO6S,(PPh,)6]0.O,MoS, requires the insertion of the cluster between the
layers. Unfortunately [Co,S,(PPh,)6]o_0,MOS, is not as selective toward
biphenyl as Crosfield 465.

As discussed previously, the structure of [Co,S,(PPh,)6]0.O,MoS, under
HDS conditions is ambiguous. HRTEM micrographs acquired after the
catalysis suggest that some regions remain pillared whereas other regions
have collapsed (Figure 3.15), but this method cannot probe the structure of
the bulk of the sample. If in fact the sample is still pillared, as the
micrograph suggests, the increase in activity may be due to the larger
surface area of the sample (26.8 mZ/g) as compared to the surface areas of
pristine MOS, (5.8 mzlg), restacked MOS, (1.0.0 mz/g),
[Al,,O,,(OI-l),,,(I-I,O),,]0_(,,MOS2 (10.0 m2/g), and (0.2[COGS,(PPh,)6] +

restacked MoS,) (9.3 mzlg). Also, if one compares the surface areas of the

104

 

. . ) , , . , a, . A,

Figure 3.15. HRTEM micrograph of [C06S3(PPh3)6]o_02M082 after HDS.
Although some regions do not appear to be expanded (black arrow), other
regions are still separated (white arrow). The observed d-spacing, 14.7 A,
corresponds to an expansion of 8.5 A, consistent with the dimensions of
the C0683 core.

105

samples to their selectivities toward biphenyl, it appears that higher surface
area coincides with lower selectivity.

This possible relationship between surface area and selectivity
supports the rim-edge model31a of the active sites. In this model all
catalysis takes place on the exterior of the particle, not on the basal planes
but at the perimeter of each layer. There are two different kinds of active
sites: the “edge” sites and the “rim” sites. To understand the difference
between the two, consider a stack of MOS, layers (Figure 3.16a). The
perimeters of the layers at the top and bottom of the stack are the “rim”
sites (much like the rim of a soup can). The perimeters of the interior
layers, which comprise the bulk of the active sites in Figure 3.16a, are
called “edge” sites. Unlike the “edge” sites, which are sterically hindered
on both top and bottom by the presence of other layers, the “rim” sites are
unrestricted on one side. According to the authors of this model, selective
HDS takes place at “edge” sites, whereas HDS combined with
hydrogenation takes place at the “rim” sites. The exfoliation/flocculation
process used to synthesize restacked MOS, (and [C0688(PPh3)6]0'0,MOS,)
decreases the particle size (and increases the surface area) relative to 2H-
MOS,, resulting in more “rim” sites, which may explain why restacked
MOS, shows higher activity (but less selectivity) than 2H-MOS,.

Furthermore, by propping the layers apart with the Co,,S,,(PPh,)6 cluster,

106

(A)

“rim” site “rim” site

1 K

 

_
— ‘— M82 layer
“edge” —
sites —
—
_
—

“rim" site‘ “rim” site

(B)

“rim” site “rim” site

\A

.. - .._ . . Q—MSzlayer
“rim” sites[ _, 2 _

1 ° ‘-——clusters
—

“rim” site‘ ~‘rim” site

Figure 3.16. Illustration of the “rim-edge” model for (A) unpillared and
(B) pillared MOS,. In unpillared MoS,, the “rim” sites are less selective
toward biphenyl. In pillared MOS, the propping apart of the layers results
in a higher number of “rim” sites, reducing the selectivity of the catalyst.

107

the surface area increases and even more “rim” sites are generated, which
also results in a decrease in the selectivity (Figure 3.16b). The rim-edge
model does not address the effects of CO- or Ni- containing species on the
activity, however.

It is interesting to note that, although we considered the attack of the
layers by the departing PPh3 ligand to be a serious liability, in fact it may
promote the activity of the material. In the “anion vacancy” model, the
HDS active sites are sulfur vacancies in MOS, which are stabilized by the
promoter species. Although sulfur vacancies induced by phosphine ligand
evolution could explain increased activity in [CO6S,(PPh,)6]o_o,MOS,, it does
not explain the increase in the activity of restacked MOS, relative to 2H-
MOS,.

The activities of [Ni,S9(PEt,)6]0.0,MOS, and [Ni9S9(PEt,)6]0_0,WS, were
disappointing in comparison to the activity of [CO6S,(PPh,)6]0.0,MoS,.
Although they are somewhat increased relative to restacked MOS, (58% and
44% over 22%), they are far lower than the activity of the commercially
available catalyst Crosfield 504 (Ni-Mo) (98%) and their selectivity is the
worst of all the catalysts. This behavior can be understood in terms of
several factors. First, at HDS temperatures the encapsulated cluster has
degraded (as discussed previously), most likely resulting in a material

which is no longer pillared. Second, because the sulfide source of the

108

volatile triethyl phosphine sulfide is the [Ni,S,]2+ cluster core, no
additional sulfur vacancies are created by the departing ligand upon
thermal treatment. The fact that there is any increase in the activity
relative to restacked MOS, is more of a reflection on the effects of the Ni-
containing species on the activity than it is on the rim-edge model or the

anion vacancy model.

3.4 Conclusions

In conclusion, CO6S,(PPh,)6, [Fe6S,(PEt,)6]2‘° , and [Ni,S9(PEt,)6]2°
can be encapsulated in MOS, and WS, and exhibit expansions of 14.0 - 15.4
A, 10.5 - 11.4 A, and ~14 A, respectively, which are consistent with the
dimensions of the clusters. X-ray diffraction patterns from samples
containing the neutral Co,,S,(PPh,)6 species are weaker than the patterns
from samples containing the cationic clusters. Although X-ray diffraction
and TGA studies suggest little loss of phosphine prior to thermal treatment
in [CO6S,(PPh,)6]O_0,MoS,, HRTEM micrographs show loss of PPh, ligands,
perhaps due to the ultrahigh vacuum or the limited applicability of the
technique, which allows only the examination of the edges Of thin particles.
Furthermore, expansion of the layers appears to be localized. TGA studies
indicate the loss of PEt3 ligands prior to thermal treatment in

[FeGS,(PEt,),]O.096MoS, and [FeGS,(PEt,),]0,WS,, and HRTEM micrographs

109

suggest some aggregation of the clusters. The orientation of the
Co,,S,,(PPh,)6 or the [Fe,,S,(PEt,),]2‘° clusters in the layers is not known; it
may change with loading and/or degree of phosphine coordination
remaining in the encapsulated cluster. The orientation of the cluster in
[Fe,,S,(PEt,),],,,,,TaS,20 is with its C, axis perpendicular to the MS, layers;
this kind of arrangement makes sense, particularly at lower loadings,
because it minimizes the Observed expansion. The rather large expansion
Observed for [Ni,S,(PEt,)6]omMOS, and [Ni,S,(PEt,)6]0_o,,WS,, coupled with
the high loadings of the cluster and TGA studies revealing no loss of PEt3
ligands prior to thermal treatment, imply that the clusters are packed with
their Ni,S, planes parallel to the MS, layers. Metallic conductivity of the
samples verifies that the layers have not structurally transformed from the
octahedral coordination found in LiMoS, and restacked MS, to the trigonal
prismatic coordination found in 2H-MS,.

The ability of MOS, and WS, to encapsulate all three of these clusters
is both remarkable and perplexing. What is the driving force for the
reaction, particularly the encapsulation of neutral species such as the
Co,S,,(PPh,)6 cluster? This ability of MOS, (and WS,) to accept neutral
molecules initially convinced some scientists that the layers were in fact
completely oxidized in the exfoliation process.9 However, increasing
evidence of facile encapsulation of cationic species, Often with a high

degree of order, suggests that the layers retain some negative charge. The

110

driving force for the encapsulation of cationic species would simply be an
electrostatic interaction. The degree of charge remaining on the layers is
difficult to determine from these samples, as a range of stoichiometries for
all clusters can be Observed. Furthermore, residual negative charge on the
layers may explain the encapsulation of species such as the [Fe,,S,(PEt,),,]2+
and the [Ni,S,(PEt,),,]2+ clusters, but it still does not explain the ability of
these layers to also incorporate neutral species. Admittedly, the
encapsulation of the neutral cluster is more difficult, requiring sonication
and two days of stirring, and is sometimes difficult to reproduce.
Encapsulation can be expedited by bubbling oxygen or air through the
reaction mixture, or by (cautiously) introducing a neutral oxidizing agent
such as 1,. This indicates that the layers may require further oxidation,
either with time or with additional oxidizing agents in the reaction, before
neutral species can be encapsulated. In addition, it is probable that small
amounts of Li“ cations, which are extremely difficult tO detect, are co-
encapsulated with the Co,,S,(PPh,)6 cluster. However, the driving force for
neutral cluster encapsulation remains a mystery.

Thermal treatment combined with mass spectrometry reveals that all
phosphine ligands depart from the samples as phosphine sulfide. In
[CO6S,(PPh,)6]o_o,MoS, and [Fe6S,(PEt,)y],MS, the phosphines appear to

attack the layers as they depart; even the unencapsulated [Ni,S,(PEt,)6]2°

111

cluster decomposes to SPEt,, indicating that the sulfur source in all samples
is the [Ni,S,]2+ core.

The HDS activity of [C0658(PPh3)6]0.O,MOS, is comparable to the
commercially available catalyst Crosfield 465 (Co-Mo); however, it is less
selective toward desirable products such as biphenyl. Although the poor
selectivity prohibits it from becoming commercially viable, the correlation
between surface area and activity seems to support the “rim-edge” model to
describe a mechanism for selectivity. In addition, the attack of the layers
by the departing phosphine ligand does not seem to inhibit catalytic

activity; in fact, it may help promote it.

112

References

1. Barrer, R.M. “Zeolites and Clay Minerals as Sorbents and Molecular
Sieves,” Academic Press, New York, 1978, p407.

2. a) Pinnavaia, T.J. “Characterization of Catalytic Materials,” Wachs, I.E.
(Ed.), Manning Publications Co, Greenwich CT, 1992, p149. b)
Johnson, I.D.; Werpy, T.A.; Pinnavaia, T]. J. Am. Chem. Soc., 1988,
110, 8545. c) Pinnavaia, T.J. Science, 1983, 220, 365.

3. Kresge, C.T.; Leonowizc, M.E.; Roth, W.J.; Vartuli, J.C.; and Beck, J .5.
Nature, 1992, 359, 710.

4. a) Hanko, J.A.; Kanatzidis, M.G. Angew. Chem. Int. Ed., 1998, 37 ,
342. b), Stephan, H.-O.; Kanatzidis, M.G. J. Am. Chem. Soc., 1996,
118, 12226. c) Chou, J.-H., Kanatzidis, M.G. Chem. Mater., 1995, 7,
5. d) Chou, J.-H., Kanatzidis, M.G. Inorg. Chem, 1995, 33, 1001. e)
Parise, J.B. Science, 1991, 251, 293.

5. a) Yaghi, O.M.; Sun, 2.; Richardson, D.A.; Groy, T.L. J. Am. Chem.
Soc., 1994, 116, 807. b) Tan, K.; Darovsky, A.; Parise, J.B. J. Am.
Chem. Soc., 1995, 117 , 7039. c) Ahari, H.; Garcia, A.; Kirkby, S.;
Ozin, G.A.; Young, D.; and Lough, A.J. J. Chem. Soc., Dalton Trans,
1998, 2023.

6. a) Bedard, B.L.; Wilson, S.T.; Vail, L.D.; Bennet, J.M.; Flanigen, E.M.
Zeolites, Facts, Figures, Future. Jacobs, P.A.; van Santen, R.A. (Eds)
Elsevier Science Publishers 3.; Amsterdam, the Netherlands, 1989,
375-387. b) Jiang, T.; Lough, A.; Ozin, G.A.; Bedard, R.L.; Broach,
R. J. Mater. Chem, 1998, 8, 721. c) Jiang, T.; Lough, A.; Ozin,
G.A.; Bedard, R.L. J. Mater. Chem, 1998, 8, 733. d) Ahari, H.;
Bowes, C.L.; Jiang, T.; Lough, A.; Ozin, G.A.; Bedard, R.L.; Petrov,
8.; Young, D. Adv. Mater., 1995, 7, 375. e) Ahari, H.; Ozin, g.A.;
Bedard, R.L.; Petrov, S.; Young, D. Adv. Mater., 1995, 7, 370. f)
Bowes, C.L.; Petrov, S.; Vovk, G.; Young, D.; Ozin, G.A.; Bedard,
R.L. J. Mater. Chem, 1998, 8, 711.

7. a) Dhingra, S.; Kanatzidis, M.G. Science, 1992, 258, 1769. b)
Chondroudis, K.; Kanatzidis, M.G. J. Am. Chem. Soc., 1997, 119,
2574. c) Sheldrick, W.S.; Braunbeck, H.-G. Z. Naturforsch, 1990,
45b, 1643. d) Marking, G.A.; Kanatzidis, M.G. Chem. Mater., 1995,

113

7, 1915. e) McCarthy, T.J.; Tanzer, T.A.; Kanatzidis, M.G. J. Am.
Chem. Soc., 1995, 117, 1294.

8. Eisenmann, B.; Jackowski, J.; Schafer, H. Rev. Chim. Min., 1983, 20,

295.

9. Bowes, C.L.; Ozin, G.A.; Adv. Mater., 1996, 8, 13.

10. Li, H.; Laine, A.; O’Keefe, M.; Yaghi, O.M. Science, 1999, 283,

11.

12.

13.

14.

15.

16.

17.

18.

1145.

a) Wachhold, M.; Rangan, K.K.; Billinge, S.J.L.; Petkov, V.; Heising,
J.; Kanatzidis, M.G. submitted. b) MacLachlan, M.J.; Coombs, N.;
Ozin, G.A. Nature, 1999, 397, 681.

a) Whittingham, M.S.Prog. Solid St. Chem, 1978, 12, 41-99. h)
Rouxel, J. “Layered Metal Chalcogenides and their Intercalation
Chemistry.” Comprehensive Supramolecular Chemistry, Alberti, G.;
Bein, T., Eds. Vol. 7, Pergamon Press, New York, 1996, p77-105.

(a) Bissessur, R.; Kanatzidis, M.G.; Schindler, J.L.; Kannewurf, C.R.

J. Chem. Soc. Chem. Commun. 1993, 1582. (b) Kanatzidis, M.G.;
Bissessur, R.; De Groot, D.C.; Schindler, J .L.; Kannewurf, C.R. Chem.
Mater. 1993, 5, 595. (c) Wang, L.; Schindler, J.L.; Thomas, J.A.;
Kannewurf, C.R.; Kanatzidis, M.G. Chem. Mater., 1995, 7, 1753.

Lemmon , J .P.; Lerner, M.M. Chem. Mater., 1994, 6, 207.

(a) Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. Science 1989,
246, 371. (b) Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. J.
Mater. Res. 1991, 6, 1103.

Tigaya, H.; Hashmoto, T.; Karasu, M.; Izumi, T.; Chiba, K. Chem.
Lett. 1991, 2113.

Heising, J.; Bonhomme, F.; Kanatzidis, M.G. J. Solid State Chem,
1998, 139, 22.

a) Bissessur, R.; Heising, J.; Hirpo, W.; Kanatzidis, M.G. Chem.
Mater., 1996, 8, 318. b) Brenner, J.; Marshall, C.; Ellis,L.;
Tomczyk,N.; Heising, J.; Kanatzidis, M.G.Chem. Mater., 1998, 10,
124-4. c) Bissessur, R. “Synthesis and Characterization Of Novel

114

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

Intercalation Compounds of Molybdenum Trioxide and Molybdenum
Disulfide.” Ph.D. Dissertation, Michigan State University, 1994.

a) Rudorff, W. Chimia 1965 19, 489. b) Bayer, E.; Rudorff, W. Z.
Naturforsch 1972, 27b, 1336. c) SchOllhom, R.; Bethel, J.; Paulus,
W. Rev. Chim. Miner., 1984, 21, 545. d) Miremadi, B.K.;. Morrison,
S.R. J. Appl. Phys. 1988, 63, 4970.

(a) Nazar, L.F.; Jacobson, A.J. J. Chem. Soc. Chem. Commun. 1986,
570. (b) Nazar, L.F.; Jacobson, A.J. J. Mater. Chem, 1994, 4, 149.

Hong, M.; Huang, 2.; Lei, Z.; Wei, G.; Kang, B.; Liu, H. Polyhedron
1991, 10, 927.

Agresti, A.; Bacci, M.; Cecconi, F.; Ghilardi, C.A.; Midollini, S.
Inorg. Chem. 1985, 24, 689.

Cecconi, F.; Ghilardi, C.A.; Midollini, S. Inorg. Chem. 1983, 22,
3802.

Murphy, D.W.; DiSalvo, F.J.; Hull, G.W. Jr.; Wasczak, J.V. Inorg.
Chem. 1976, 15, 17.

Py, M.A.; Haering, R.R.. Can. J. Phys. 1983, 61, 76.

a) Pecoraro, T.A.; Chianelli, R.R. J. Catal., 1981, 67, 430. b) Harris,
8.; Chianelli, R.R. J. Catal., 1984, 86, 400. c) Ledoux, M.J.;
Michaux, O.; Agostini, G.; Panissod, P. J. Catal., 1986, 102, 275.
Massoth, F.E. Adv. Catal., 1978, 27, 265.

a) Lipsch, J.M.J.G.; Schuit, g.C.A. J. Catal., 1969, 15, 179. b)
Massoth, F.E. J. Catal., 1977, 50, 190.

a) Farragher, A.L.; Cossee, P. “Proceedings, 5th International
Congress on Catalysis, Palm Beach, 1972”, Hightower, J.W., Ed.,
North-Holland, Amsterdam, 1973, 1301. b) Dungey, K.E.; Curtis,
M.D.; Penner—Hahn, J.E. J. Catal., 1998, 175, 129.

Topsoe, H.; Clausen, B.S. Catal. Rev. -Sci. Eng., 1984, 26, 395.

115

31. a) Daage, M.; Chianelli, R.R. J. Catal., 1994, 149, 414. b) Del Valle,
M.; Cruz-Reyes, J .; Avalos, Borja, M.; Fuentes, S. Catal. Lett., 1998,
54, 59.

116

CHAPTER 4

Structure Of Restacked MOS, and WS, Elucidated by Electron
Crystallography

117

4.0 Abstract

There has been a lot of confusion about the nature of restacked MOS,
and WS,. The structure has been proposed to be trigonal TiS, type with
octahedral M4+ and called 1T-MoS,. The presence of a distortion in the
metal plane which gives rise to a superstructure has been suggested. We
have performed electron crystallographic studies on small (sub micron)
single crystal domains of restacked WS, and MOS, in order to solve their
superstructure. We find that what initially seems to be a trigonal crystal is
actually a “triplet” of three individual orthorhombic crystals. Using two-
dimensional hk0 data from films for both “triple” and “single” crystals we
calculated corresponding Patterson projections, which reveal a severe
distortion in the Mo/W plane, forming infinite zigzag chains. The
projection of the structure suggests M-M distances of 2.92 A and 2.74 A
for MOS, and WS,, respectively. Least squares refinement from the single
crystal data gives R,=13.3% for WS, and R,=15.3% for MOS,. Therefore,

we submit that restacked MOS, and WS, are not lT-form but rather WTe,

type.

118

4.1 Introduction

Due to a unique combination of valuable structural, electronic, and
optical properties, the layered dichalcogenides have been studied and used
for a litany of practical applications.”4 One Of the most versatile members
of this class of compounds is MOS,. Found in nature in its 2H form as the
mineral molybdenitez, its inexpensiveness and availability have permitted
its use as a solid lubricant, a catalyst for hydrodesulfurization3, a host for
intercalation chemistry and an electrode material for solid state batteries“.

Many layered transition metal chalcogenides can be treated with n-
Butyl lithium to form a reduced species in which lithium occupies the space
between the layerss. (This ability to undergo reduction and accept lithium
is one important reason why these materials have been investigated for
solid state batteries.) The redox properties of these reduced
dichalcogenides vary from one compound to the next, but some have the
remarkable ability to form suspensions in water, much like clays. The
chemistry of LiMoS, (and LiWS,) in water is particularly fascinating
because, after undergoing a redox reaction with water to form H, (g) and
LiOH, a suspension of colloidally dispersed single layers is formed, and
the layers can remain separated from one another in water for days. The

material can be “restacked” by filtration, precipitation, centrifugation, or

119

evaporation. Due to this remarkable ability, many guest species have been
encapsulated between the layers.“8

There has been a lot of confusion about the nature of restacked MOS,
and WS,. The presence of a distortion in the metal plane which gives rise
to a superstructure has been suggested. The reduction of the 2H form
appears to induce a structural transformation from trigonal prismatic
coordination about the metal to octahedral coordination? The structure has
been proposed to be trigonal TiS, type10 with octahedral M‘” and called 1T-
MoS,. Originally, lT-MoS, was synthesized by the oxidation of
K,(H,O),MOS, instead of the exfoliation/restacking method, but the
products of both synthetic methods exhibit metallic conductivity and an
exothermic transition around 90-100°C which suggests that they are
identical.” The layered dichalcogenides are prone to a wide variety of
structural distortions caused by Charge Density Waves (CDWs), resulting
in significant changes in the properties of the materials.'2 The source of
confusion is that several superlattices have been reported for lT-MOS,

(Figure 4.1).

120

 

Figure 4.1. Schematic illustrating the proposed superstructures of
restacked/lT-MOS,. The superstructures derive from bonding associations
of metal atoms. (A) An ideal undistorted lattice, (B) a tetramerization, (C)
a trimerization, and (D) zig-zag chain formation. Black circles represent
Mo, gray circles, S.

121

EXAFS (Extended X—ray Absorption Fine Structure) analyses have
indicated M-M associations, but are not able to provide a structural model.
X-ray and electron diffraction studies of the restacked MOS, (and WS,)
have indicated a 2a x 2a superstructure in the ab plane. A tetramerization
model was proposed based on the X-ray diffraction studies.”"“ The
electron diffraction studies on restacked WS,, however, found that the
hexagonal symmetry was preserved, but the proposed tetramerization
model is incompatible with this finding.15 To add to the confusion, another
electron diffraction study on LiMoS, found the same 2a x 2a
superstructure, but the authors attributed it to lithium ordering between the
layers.16 Later, an STM study of the surface layer of restacked MOS,

suggested a 2a x a superstructure caused by the formation of zig-zag

chains.17 The original lT-MOS,, however, is reported to exhibit a fla x

[[321 superstructure, which would be best described by a trimerization

model?” In an effort to clarify the structure, we have performed 2—
dimensional electron crystallographic studies on restacked WS, and MOS,.
This method has enabled us to probe the bulk material, not just the local
environments of the atoms or the structure at the surface. Furthermore, it
has allowed us not only to directly determine the structure, but to refine it

as well.

122

X-rays are scattered by the electron shells of atoms; electrons, on the
other hand, are scattered by the electrostatic potential due to the atomic
nuclei and their electron clouds. Despite the different origins of scattering,
much of the theory developed for X-ray crystallography can be applied to
electron crystallography and thus, in principle, the two techniques provide

similar structural infonnation.‘9’2°

Most experimentalists have avoided
electron diffraction as a technique for structure solution, however, because
electrons interact more strongly with the sample than X-rays; hence,
multiple scattering events (secondary and dynamic scattering) may take
place in electron diffraction.”20 Because a TEM (Transmission Electron
Microscope) is capable of high magnification, it is possible to acquire data
for many samples which cannot be investigated by single crystal X-ray
methods?“20 Furthermore, the wavelength of an electron (at an
accelerating voltage of l20kV) is 0.0335 A; as compared to 0.71 A for Ka
Mo radiation, which (theoretically) can lead to. higher resolution data.
Because the Ewald sphere is much flatter, many reflections can be observed
simultaneously for one orientation of the crystalm"22

We have discovered that, despite the presence of Mo/W in our

samples, quasi-kinematical data are acquired, from which a plausible

projection of the structure can be obtained and refined.

123

4.2 Experimental

LiMS, (M=Mo, W) was synthesized by reacting 2H-MS2 with excess
LiBH4 at 300-350°C for 3 days. Restacked MS,“'13 was synthesized by
reacting LiMS, with H,O, rinsing several times to remove the LiOH
generated in the exfoliation process, and depositing the solution on a
copper carbon coated grid. The grids were examined at 120kV on a JEOL
120CX TEM. Suitable crystals were located and their diffraction patterns
captured on film. The negatives of the diffraction patterns were scanned

into the computer at 600 dpi. The patterns were indexed to an

orthorhombic cell with a =5.56A and b=3.21A, related to the subcell by the

relation 3a,“, x am. The “Gel Plotting Macro” in NIH Image 1.60 was

used to extract the integrated intensities (Im) from the patterns.23 Accurate
cell parameters a and b were determined from powder diffraction data
using a Rigaku-Denki/RW400F2 (Rotaflex) rotating anode powder
diffractometer.

The data were converted into .hkl file format for use in SHELXTL
programs.24 The two-dimensional Patterson maps were calculated from the

data by Equation 4.1:

P(u,v) = %2:2|<I>(hk)|2 cos 27t(hu + kv) Eq (4 l)
h k . .

124

Where P corresponds to electron density overlap in the structure (i.e. the

Patterson function), A is the unit cell area, and l<I>(hk)l = 1“,.19 Least

squares refinement of the structure in SHELXTL (version 5)24 were
carried out after the coefficients for the electron scattering factors were

obtained by fitting the sine/2t curves using the program Curve Expert.”

4.3 Results and Discussion

4.3.1 Data Collection

Because restacked MOS, and WS, are layered compounds with plate-
like morphology, they exhibit preferred orientation which causes the ab
plane to be perpendicular to the electron beam, and the diffraction patterns
consistently contained only hk0 data. Our initial efforts to locate a suitable
crystal resulted in pictures similar to Figure 4.2a-4.2b. The patterns
suggested a 2a x 2a superstructure. Upon consideration, however, we
recognized that the exfoliation/flocculation process, involved in the
synthesis of restacked MOS, and WS,, would result in turbostratic
materials, or at least materials susceptible to stacking faults, and that this

could result in pronounced twinning phenomena. We then targeted

125

 

Figure 4.2. (A) Bright field image and (B) selected area electron diffraction
pattern from a "triple" WS2 crystal, giving rise to an apparent 2a x 2a
hexagonal superstructure. (C) Bright field image and (D) selected area
electron diffraction pattern from a "single" WS2 crystal, giving rise to a

Zn x a or ‘13a x a superstructure.

126

 

 

 

 

 

 

a
o 0
2a ° °
‘3' o o
. . . .
a o o . o
O . o
. . g . o
B O . o - O
.........
...........
.............
...... . . ..
...... ...
...... ...
...... ...
...... . . ....
.............
...........
o..........
.........

Figure 4.3. (A) Relationship between the 2a x a and flu x a lattice. (B)

Illustration of how three flu x a patterns can be overlapped to form a 2a x
2a pattern. Note that only the sublattice reflections overlap.

127

extremely thin crystals in our investigations and found that, in fact,
restacked WS, and MOS, have a 2a x a superstructure (Figure 4.2c-4.2d),
which is consistent with the results of the STM studies of restacked MOS,.l7

This 2a x a superstructure can actually be more simply described by an

orthorhombic [Xi—a x a cell (Figure 4.3a). Note that a = a but a* ¢ a* in

the two unit cells because of the change from a hexagonal to an

orthorhombic crystal system. The 2a x 2a superstructure is in fact a

“triplet” of three 1/3—a x a crystals rotated by 120 degrees relative to each

other. This causes their diffraction spots to be aligned with respect to the

sublattice reflections but not the superlattice reflections (Figure 4.3b).

4.3.2 Structure solution from the “triplet” crystal Of WS,

Because a “single” crystal pattern of sufficient quality was not
initially available, we proceeded with the structure determination using the
twinned crystal depicted in Figure 4.2a-4.2b. Intensities of the reflections
were extracted from the diffraction pattern using the “Gel Plotting Macro”
of NIH Image 1.60.” This Macro was developed to calculate

concentrations in gel electrophoresis experiments by a comparison of the

128

bands in a sample to the same bands in reference samples of known
concentrations. In order to make this comparison it is necessary to
consider not only how dark (or light) the band, but also the area occupied
by the band. Similarly, the intensity of a reflection in an electron
diffraction pattern is a function of the area of the spot as well as its optical
density. NO external standard is required because only the “relative”
intensities of the reflections are necessary for structure determination.
Accordingly, the diffraction pattern was indexed to the three orthorhombic
cells and the relative intensities of the reflections were extracted from the
pattern (Figure 4.4). First the macro creates a one-dimensional set of
peaks from the data; then the width and baseline of the peaks are specified
and a numerical value for the integrated intensity is extracted. A 2-D
rolling ball background subtraction suggested in the “Gel Plotting Macros”
manual was applied, as well as a Lorentz correction recommended by early
electron crystallographers.‘9 Intensities of reflections from all three of the
“triplet” cells were extracted, but only the cell with the strongest
reflections was used. The intensity of the subcell reflections were

corrected for the twinning by dividing by three.

129

 

 

 

 

 

 

 

 

Figure 4.4. Illustration of the "Gel Plotting Macro" in NIH Image 1.60.
(A) Lanes of reflections are specified, then (B) converted to a set of one-
dimensional peaks from which the integrated intensities of the reflections
can be extracted.

130

b axis

a axis

 

Figure c-axis: (A)
ideal 1T-M82 (from simulated data) (B) restacked W82 with background
subtraction and Lorentz correction (C) restacked WSz with only the
Lorentz correction (D) restacked WS2 with no corrections. All
experimental patterns indicate a deviation in the W—W atom vector.

131

A two-dimensional Patterson contour plot generated from the
extracted data clearly indicates a substantial deviation in the metal atom
position from the ideal position in the lT-TiS, structure type, which gives
rise to a short M-M distance (Figure 4.5b). In an ideal 1T structure the
electron density overlap at the center of the Patterson map, which is

indicative of the M-M vector in the structure, is circular (Figure 4.5a).

This translation of the metal atom along the /3_a axis can only be explained

by the formation of zig-zag chains. This kind of distortion has been

observed in WTe,,26

another layered dichalcogenide with (distorted)
octahedral coordination about the metal atom, and represents a significant
departure from the ideal structure, TiS,.lo Distortion of layered octahedral
ML, (12 systems to form zig-zag chains due to a charge density wave has
been predicted by Rovira and Whangbo.27 This kind of distortion gives
rise to changes in the band structure of the material which are consistent
with the properties we have observed.”

The Patterson map in Figure 4.5b contains some strange features
which had to be addressed before refinement of the structure was possible,
however. The M-S vector peaks are very weak, and there are other
puzzling features which seem to indicate short atom-atom vectors in the

structure. These vectors became more pronounced with the elimination of

higher order reflections. They disappeared completely, however, when the

132

2-D rolling ball background subtraction was eliminated from the data
extraction procedure. Electron diffraction patterns have a rather high,
sloping background due to diffuse scattering which occurs in all samples.
The peaks closest to the main beam are most severely affected. A
background subtraction seemed an ideal manner to address this problem;
however, use of the background subtraction method in the “Gel Plotting
Macro” Obviously results in perturbation of the data, particularly the data
closest to the main beam, because the contribution of the background is less
significant in gel electrophoresis samples than in electron diffraction
patterns. The background can be removed manually (with much better
results) during the step in which the baseline is specified for each peak.
Also, data from two different camera lengths were combined. Use of a
larger camera length allows reflections to be collected further away from
the main beam on the negative, whereas the smaller camera length allows
higher order reflections (which provide better resolution) to be collected.
The modified data set gave rise to the Patterson map found in Figure 4.5c.
Least squares refinement of this data set in SHELXTL (version 5)24
against a WTe, type model gives an R, value of 36.8% (Tables 4.1 and
4.2). Structures with comparable R, values solved by electron diffraction
have actually been published by early electron crystallographers.19 Such a

high R, value is considered acceptable for many systems due to the

133

Table 4.1. Crystal data and structure refinement for restacked WS,.

 

Empirical formula
Formula weight
Temperature
Wavelength

Crystal system
Plane group
Unit cell dimensions

Area (A2)
z
6 (000)

20 range for data collection
Data resolution

Index ranges

Reflections collected
Independent reflections
Refinement method

Data / restraints / parameters
Goodness-of-fit on (1)2
R indices (all data)“

we,
248.00
293 K
0.0335 A

Rectangular
pg (#4)
a=556A
b=321A
y = 90 deg.
17.295

2

46

0.36 to 4.20 deg.

0.09 to 1.1 sin0/A
-12<=h<=10, -5<=k<=5
173

99

Full-matrix Is. on (I) 2

9910/7
0.874
R, = 0.3683

 

“R, = 201601 - K1¢,11)/ 21601

134

Table 4.2. Preliminary atomic coordinates ( x 10°) and equivalent
isotropic displacement parameters (A2 x 103) for restacked WS,.

 

 

x y U(eq) Occ.
W( 1) -2022 0 l 1
S( 1) 423 1 0 724 1
8(2) 997 0 1 100 1

 

U(eq) is defined as one third of the trace of the orthogonalized Uij tensor.

susceptibility of the electron diffraction data to perturbation by multiple
scattering events, as mentioned previously. In order to improve the

refinement, we attempted a two-beam dynamic scattering correction.

135

4.3.3 Dynamic Scattering: Two-Beam Approximation

Dynamic scattering has inhibited the widespread use of electron
diffraction data for structure determination, particularly in samples
containing heavy elements. A schematic depicting the perturbation of the
electron diffraction data by dynamic scattering is found in Figure 4.6. In
the ideal situation (found in X-ray diffraction), called “kinematic”
scattering, the radiation source is diffracted only one time as it passes
through the sample. Due to the strong interaction of the electron beam
with the sample, however, electrons can be diffracted many times as they
pass through the sample. The thicker the sample, the greater its
susceptibility to multiple scattering events. One can then consider the
intensity of each diffracted beam to be “contaminated” with intensity from
the other diffracted beams, and also “contaminated” due to the loss of some
of its original intensity to other reflections:

I = (Ikin - Ilost ) + I,gained Eq’(4'2)

dyn

Historically, correction for this phenomenon has been attempted in more
than one way. The two-beam approximation tries to simplify the problem
by treating the data as if the pattern comes from a “mosaic” of small

crystals, each with a unique and perfect orientation such that only that one

136

A main beam

 

sample

 

 

 

diffracted diffracted
beams beams
B main beam

 

sample

   

diffracted diffracted
beams beams

Figure 4.6. Schematic illustrating (A) kinematic scattering, in which the
electron beam is diffracted only once as it passes through the sample and (B)
dynamic scattering, in which the beam is diffracted multiple times, resulting
in “contamination” of the intensity of all the reflections.

137

reflection is produced per crystal. Each reflection therefore functions
completely independently of all the other reflections in the pattern.'9'”'35 In
this case, the intensity of each diffracted beam can be conceptualized as
1d,“: Ikin - 1,0,, Eq.(4.3)
If one imagines I,ost tO be a constant fraction (say, 30%) Of the original
intensity of each reflection, it becomes clear that the strongest reflections
will be most noticeably affected. The mathematics are actually somewhat
more complicated, but the net effect is a homogenization of the diffraction
pattern. In order to adjust for this ‘homogenization’, a method has been
developed to correct the experimental data.” It operates under the
assumption that most reflections scatter kinematically, but a few of the
most intense reflections exhibit “deviations toward dynamic scattering.”19
The following equation applies in a kinematically scattering crystal:
I = Q24 Eq.(4.4)

in which Q is a normalized structure factor”0 and .4 is a Lorentz correction

(in this case, to divide each reflection by its d-spacing). In a dynamically

scattering crystal the equation is slightly modified:
I=Q°A3K(A3Q)-4 Eq.(4.5)
such that it now includes a term for the thickness of the crystal (A,) and a

function K(A,Q), shown in Figure 4.7, which describes dynamic scattering

138

as a function of the product of the thickness and the normalized structure

factor.

 

0:1 . . . . . . A .
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 A30

Figure 4.7. Graph of the function K(A,Q), which describes dynamic
scattering as a function of the thickness (A,) of a crystal and the normalized
structure factor (Q) according to the two-beam approximation. In a
kinematically scattering crystal K = 1.

After a preliminary refinement has been conducted, the calculated

structure factors are used to determine an experimental K value for each
reflection:
(I.../ 49692.... = K... Eq.(4.6)
If Kexp < 1 the reflection is considered to be decreased in intensity due to
dynamic scattering. The purpose of calculating Kexp is purely diagnostic,
and Kexp < 1 values are only significant for reflections of large intensity.
Once the reflections in need of correction have been determined,
(I)calc and the thickness (A,) of the crystal can be used to calculate a

theoretical A,Q value for each reflection, and the graph in Figure 4.7 is

139

used to find a Km value. The corrected intensity can then be calculated by

the following relation:
Icorr = (Iobs/ A/Kcon’ Eqfl(47)

Two major drawbacks of this method (aside from the fact that the 1’,ained
term is neglected) are that some knowledge of the structure is required to
correct the data, and the thickness of the crystal must be determined.
Unfortunately the thickness of the “triplet” crystal used in this structure
determination was not known. Several different X-ray diffraction patterns

containing 001 reflections were acquired, and Debye-Scherrer line

broadening31 was used to estimate an average thickness for the particles:

th, = 0.9 x A, x 57.3
[3,,, cos0 Eq.(4.8)

where B,,, is the full width at half maximum of the 001 reflection. The
results indicated a thickness range of 150 A to 220 A. Km values were
calculated for 12 reflections at 150 A and up to 34 reflections at 220 A,
new Patterson maps were generated, and refinements of the corrected data

were conducted.

140

Table 4.3. Refinement for WS, as a function of data correction methods.

 

 

R,a Goodness of Fit (GOF)

Original data 0.3683 0.874

(no background subtraction)

Dynamic scatterin correction 0.3580 0.871

(2 beam) t= 150

Dynamic scatterin correction 0.3161 0.756

(2 beam) t = 220

Lorentz correction eliminated 0.2454 1.301

 

1R1 = 201601 - 106,10 / 216,1

The results (Table 4.3) indicated little improvement in the
refinement at either thickness (from 36.8% to 35.8% and 31.6%). This led
us to re-assess the quality of our integrated data prior to the dynamic
scattering correction. As mentioned previously, only data from the
strongest supercell was used in the refinement, and subcell reflections were
divided by three to correct for the twinning. In reality, this correction
results in an underestimation of intensity values of the subcell reflections.
The relative intensities of the three different supercells indicated that 50%
of the intensities of the subcell reflections are due to that particular
supercell. Accordingly the data was adjusted, but the refinement was not

much improved (33.66%).

141

Next the Lorentz correction suggested by Vainshtein was eliminated.
This correction, which is simply to divide the intensities of the reflections
by their d-spacings, is a crude method to compensate for the curvature of
the Ewald sphere (Figure 4.8). Because most crystals examined by TEM
are very thin, their reciprocal lattice points are more “cigar”-shaped than
spherical along the axis parallel to the electron beam. At the center of the
negative the Ewald sphere intersects the lattice points through the center of
the cigars, capturing their maximum intensity. Towards the edge of the
negative, when the Ewald sphere has begun to curve away from the plane,
the sphere no longer intersects the lattice points through the center of the
cigars. The end result is that the intensities of the reflections at the limit
of the Ewald sphere, which have the smallest d-spacings, are
underestimated. In the early days of electron diffraction, smaller
accelerating voltages (which give rise to smaller Ewald spheres) were used;
hence the effect of the curvature was more significant and a d-spacing
correction was more important. Most modern electron microscopists no
longer apply this correction, however.” With the elimination of the
Lorentz correction the intensities of the vectors in the Patterson contour
plots generated from the uncorrected data increased, (Figure 4.5d) and the

refinement improved to 24.54% (Table 4.3).

142

 

-_‘_ _—“‘

Figure 4.8. The intersection of the Ewald sphere with the reciprocal
lattice at an accelerating voltage of (A) ~50kV and (B) ~100kV. at 100kV
the Ewald sphere is much flatter, and the effects Of its curvature become
significant at smaller d-spacings than at 50kV.

143

 

1

\

ii
011‘ -<

 

B
O
o O
0
° 0
O
. O
O
0 ° 0
O
O . O
O O
o ° 0
O
O . O
. O
0 ° C
O O O
O o O . O . O .
O O O
0 ° 0 ° 0 ° C
O O
0 ° 0 ° C
0
° 0 ° 0 9 O
0 ° 0 ° 0

Figure 4.9. (A) The limit of detectable reflections (in a perfectly
oriented crystal) as a function of the thickness (t) of the sample (q = 1/t),
the radius of the Ewald sphere (r), and angle 7. (B) An illustration of
secondary scattering, in which the diffracted beams from one crystal act as
“main beams” as they pass through a second crystal, resulting in the
appearance of reflections with artificially small d-spacings.

144

Based on the accelerating voltage and the d-spacing of the highest
order reflections Observed it is possible to estimate the thickness (t) of the
crystal (Figure 4.9a) through the following relationships:

d*/2r = sin'y

tan'yd* = q Eq.(4.9)
where d* is the reciprocal of the d-spacing of the outermost observed
reflection, r is the radius of the Ewald sphere (29.8 A"), and q is 1/2 the
length Of the cigar and is equal to the reciprocal of the thickness (t). Based
on this calculation, the “triplet” crystal would have to be 4 layers thick or
less. This immediately seemed suspicious because in order to observe these
higher order reflections it was necessary to examine twinned crystals of
“intermediate” thickness, indicating that the crystal is probably more than 4
layers thick. Therefore it seems that, in addition to dynamic scattering, the
data is perturbed by secondary scattering.

Secondary scattering, like dynamic scattering, involves multiple
scattering of the electrons as they pass through the sample. Unlike dynamic
scattering, however, which is a multiple scattering event within a single
crystal domain (or, in the approximation, a “mosaic” of independently
scattering crystals forming a nearly single crystal domain), secondary
scattering involves multiple scattering events encompassing two or more
crystals which are significantly dislocated relative to one another. It is

possible that the two crystals could even be two different compounds.

145

Consider two stacked layers rotated by 120 degrees relative to one another,
as in the “triplet” crystal. The electron beam is scattered at least once as it
passes through the first crystal; but before the pattern reaches the negative
the undiffracted and diffracted beams must pass through the second crystal,
where they can act as “main beams” and be scattered again. These new
“main beams” each have their own Ewald sphere that is slightly shifted
relative to the original main beam, which results in a pattern which
appears to have higher resolution data than is possible (Figure 4%) given
the thickness of the sample and the curvature Of the Ewald sphere
associated with the original main beam. Secondary scattering can be treated
mathematically; however, when one combines correction of the data for
secondary scattering as well as dynamic scattering the reliability of the
experimental data becomes questionable. For this reason, we decided to

concentrate our efforts on patterns which were relatively “single” crystal.

4.3.4 Structure solution from “single” crystals of WS,.

Data was extracted from the diffraction pattern of restacked WS, in
Figure 4.2d, and also the patterns from restacked WS, in Figure 4.10,
using the Gel Plotting Macro as described previously. All Patterson maps

generated from these data sets indicate a deviation in the M-M vector from

146

the ideal position (Figure 4.11). The other peaks in the Patterson maps,
which contain information about the M-S vectors, are elongated because
there are two non-equivalent metal atom positions.

Least squares refinement Of the data from Figure 4.2d in SHELXTL
(version 5)” gave rise to an R, value of 13.3% (Tables 4.4,4.5). The
structure is non-centrosymmetric, belonging to the plane group pg (#4).

Due to a convention regarding the cell parameters in this plane group,

which contains a glide plane along one axis, l/Ta = am, and a = b,,,,,. The

structure, shown in Figure 4.12, contains the W atoms in distorted
octahedral coordination, forming zigzag chains. The structure of WTe,,
which is very similar, is also shown.” It is interesting to note that, in
WTe,, the W atoms are also slightly distorted along the c-axis of the
structure. Because the electron diffraction data contains only hk0
reflections, the refinement of the structure of ,WS, produces only a
projection of the layer structure. The projection of the short W-W
distance in the structure is 2.74 A, significantly shorter than the 3.16 A

distance found in the ideal 2H-WS, structure.

147

 

Figure 4.10. Slected area eelctron diffracotin pattern from "single
crystal" restacked WS2 used to generate Patterson contour plots.

148

 

 

Figure c-axis: (A)
ideal lT-M82 (from simulated data) (B) restacked WS2 from the data in
Figure 4.2d (C) restacked WS; from the data in Figure 4.10a (D) restacked
W82 from the data in Figure 4.10b. All patterns indicate a deviation in the
W-W atom vector.

149

Table 4.4. Crystal data and structure refinement for crystals of WS,.

 

Title

Empirical formula
Formula weight

Temperature
Wavelength

Crystal system
Plane group

Unit cell dimensions

Area (A2)

Z
(I) (000)

20 range
Data resolution
Index ranges

Reflections collected

Independent reflections

Refinement method

Data / restraints /
parameters

Goodness-of-fit on (D2
R indices (all data)“

“single” WS,

248.00

293 K
0.0335 A

Rectangular
Pg (#4)

a = 5.56 A
b = 3.21 A
y: 90 deg.
17.295

2
46

0.36 to 2.52 deg.
0.09 to 0.66 sine/A
--7<=h<=7,
3<=k<:3

78
43 [R(int) = 0.1556]
Full-matrix 1.8. on (1)2

43/0/5

1.305
R, = 0.1328

“triple” ws,

ws2
248.00

293 K
0.0335 A

Rectangular
Pg (#4)

a = 5.56 A
b = 3.21 A
Y = 90 deg.
17.295

2
46

0.36 to 4.20 deg.
0.09 to 1.1 sine/A
—12<=h<=10,
-5<=k<=5

173
99 [R(int) = 0.0870]
Full-matrix 1.8. on (1)2

99/0/7

1.301
R, = 0.2456

 

“R, = 2(ll<l>,| - 106,11) / 216,1

150

Table 4.5 . Atomic coordinates ( x 10°) and equivalent isotropic
displacement parameters (A2 x 103) of restacked “single” and “triple” WS,.

 

x y U(eq) Occ.

 

“single” WS,

W(1) -2021 0 2 1

S(1) 4290 0 8 1

8(2) 950 0 8 1
“triple” WS,

W( 1) -2066 0 4 1

8(1) 4182 0 105 1

8(2) 883 0 127 1

 

U(eq) is defined as one third of the trace of the orthogonalized Uij tensor.

151

      

   
  
   
       
 
 
 

 

    

   
  

A Z‘9 ' zag . H“? O
chain Twfifi?
tat-st

b axis 7%“;gfg,‘
may.

      
 

.Ilk.’ 1‘.
O . .

B

zig - zag ‘
chain .‘V’A‘Y’AV’. ii
b axis .’/3%‘ 741% We$. E

 

   

M82 layer

Figure 4.12. (A) Two-dimensional projection of a restacked MS, layer
(M=MO,W) and one WTe, layer. (B) View of WTe, parallel to c-axis.
Dark Circles are M; gray circles, S; open circles, Te.

152

4.3.5 Dynamic Range Correction

One of the challenges in the data collection for this system is
balancing the dynamic range in the intensities of the reflections with the
limited capability of the film. In the electron diffraction pattern depicted
in Figure 4.2d, a number of the strongest peaks (usually sublattice peaks)
are overexposed, creating a “flat top” on the negative (Figure 4.13a), but
the exposure is a good length of time to capture most of the weaker
reflections (Figure 4.13b). Conversely, if a shorter exposure time is used
as in the electron diffraction patterns in Figure 4.10, the strongest peaks
will not have a “flat top” (Figure 4.13c), but the weaker peaks, which
encompass most of the superlattice reflections, are so weak that it is
extremely difficult to evaluate their relative intensities (Figure 4.13d). For

this reason, the data collected at longer exposures were corrected for this

“flat top” problem. The intensities of the (1020), (13110), and (1300)
reflections (8 total, 5 unique) were increased by 30%. Least squares
refinement against corrected and uncorrected data sets revealed no
differences in positional parameters and only minor changes in thermal

parameters; the net result was a virtually identical solution with slightly

lower Rm, (Am=0.5%); R, (Am=1.6%); and wR, (AM=8%) values.

153

 

Figure 4.13. Surface plot of the (3-10) (A) and the (-2-20) (C) reflections
in the electron diffraction pattern in Figure 4.2d. The (3-10) has a "flat top"
due to overexposure. Surface plot of the (3-10) (B) and the (-2-20) (D)
reflections in the electron diffraction pattern in Figure 4.1 la. The (3-10) is
not overexposed, but the (-2-20) is very weak.

p—i

54

4.3.6 Structure Solution from “single” crystals Of MOS,.

Data was also collected for restacked MOS, (Figure 4.14). The
lighter MO atom does not scatter as well as W; hence, it was more difficult
to find “single crystal” patterns which diffracted well enough for a

refinement. Patterson maps calculated from both diffraction patterns in

Figure 4.14 indicate a distortion in the M-M vector along the WT: axis,

forming an oval (Figure 4.15). In restacked WS, the distortion is more
pronounced, as the M-M vector has split into two resolvable peaks which
correspond to two different M-M vectors. Least squares refinement in
SHELXTL (version 5)24 was conducted on the data shown in Figure 4.15b,
and gave rise to an R, value of 15.3% (Tables 4.6, 4.7). The short MO—Mo
distance is 2.92 A, longer than the value found for restacked WS, (2.74 A)
but still significantly shorter than the ideal MO-MO distance in 2H-MOS,
(3.16 A).

155

      

Figure 4.14. Bright field images (A,C) andselected area electron
diffraction patterns (B,D) of restacked M082 , giving rise to 21 2a x a or a
13a x a superstructure.

156

a axis

 

Figure 4.15. Two-dimensional Patterson projections along the c-axis: (A)
restacked M082 calculated from the pattern in Figure 4.14b. (B) restacked
MOS2 calculated from the pattern in Figure 4.14d.

157

Table 4.6. Crystal data and structure refinement Of restacked MOS,.

 

Empirical formula
Formula weight
Temperature
Wavelength

Crystal system
Plane group
Unit cell dimensions

Area (A2)
Z
(I) (000)

20 range for data collection
Data resolution

Index ranges

Reflections collected
Independent reflections
Refinement method

Data / restraints / parameters
Goodness-of-fit on (D2
R indices (all data)“

MOS,
160.00
293 K
0.0335 A

Rectangular
pg (#4)

a = 5.47 A
b = 3.16 A
y = 90 deg.
17 .295

2

40

0.36 to 2.78 deg.
0.09 to 0.73 sine/2t
-7<=h<=5, -3<=k<=3
73

46 [R(int) = 0.2120]
Full-matrix 1.8. on (I)2

46/0/7
1.267
R, = 0.1525

 

“R, = 201601 - 106,11) / 2|<I>°|

158

Table 4.7. Atomic coordinates ( x 10°) and equivalent isotropic
displacement parameters (A2 x 103) in restacked MOS,.

 

 

x y U(eq) Occ.
Mo( 1) -225 1 0 9 1
8(1) 4240 0 38 1
8(2) 925 0 54 1

 

U(eq) is defined as one third of the trace of the orthogonalized Uij tensor.

The M-M distances have been measured by EXAFS to be 2.74 A and
2.8 A for WS2 and MoS,, respectively.” Shorter M-M distances than those
observed by EXAFS might be explained by displacement of the metal atom
along the c-axis. As mentioned previously, this displacement is actually
Observed in WTe,. Longer M-M distances than those observed by EXAFS
may be an artifact of dynamic scattering: we have observed by simulation
that the M-M distances appear to increase with increasing thickness (vide
infra).33 Despite these difficulties, electron crystallography is a more
powerful tool for the structure elucidation of these compounds than
EXAFS. The latter probes only the local structure, while electron

diffraction provides a direct structure determination.

159

4.3.7 Dynamic scattering: n-beam approximation.

Although including the effects of dynamic scattering in the
refinement should improve the results, it is non-essential for determination
of the structure of these materials. This suggests that the use of electron
diffraction for structure determination of inorganic materials may not be
as unrealistic as previously supposed. Nonetheless, an effort was made to
address the effects of dynamic scattering in these data sets.

The two-beam approximation is severely limited by the assumption
that each diffracted beam functions independently of all the other diffracted
beams in the diffraction pattern. The assumption is particularly false when
large accelerating voltages are used, resulting in very large Ewald spheres
which can excite many reflections at one time. Therefore, alternate
approximations, most notably the n-beam approximation, have been
developed.”'3°'” The n-beam approximation permits the calculation of the
dynamic contribution to each reflection from multiple beams in a series of
slices through the crystal.”5 Although more accurate, it is significantly
more complicated than the two-beam approximation, and solving a
structure using a dynamical treatment of the electron diffraction data is a
mathematically challenging procedure. Some researchers have combined
HRTEM (High Resolution TEM) images with the electron diffraction data

in order to obtain phase information and then used direct methods.37

160

Others have developed software which enables one to combine least-squares
refinement with a multislice calculation.”

As part of an effort to include a correction for dynamic scattering,
the models which resulted from kinematic refinement in SHELX programs
were constructed using the molecular modeling program CERIUSZ.33 This
program contains a HRTEM module which employs the multislice n-beam
calculation to simulate high resolution images. One step in this process
involves the generation of electron diffraction patterns. Simulated
diffraction data for varying thickness of the WS, crystal are shown in
Figure 4.16. Figure 4.16a contains essentially a kinematic pattern, as it
comes from a crystal which is only one layer thick. Note that the ( 200) is
less intense than the (300). At three layers thick (Figure 4.16b) the
sublattice reflections have increased in intensity, and at six layers thick
(Figure 4.16c) the intensity of the (200) is equal to the (300). At eight
layers thick (Figure 4.16d), the hexagonal sublattice reflections are the
most intense reflections in the diffraction pattern. It is interesting to note
that, using the two-beam approximation, one might have predicted that the
sublattice reflections, which are generally more intense than the
superlattice reflections, would have been decreased in intensity with
increasing thickness rather than increased. One can qualitatively use the

values of the (200) and the (300) to estimate the degree of dynamic

161

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A i I B
1'07 3 . 1.01 i
053 O O O O . O O .0.5_ o O . T . O .
O O O . O O O O O .
. . O O O o . O O O . o
0.0 O O O 0.0 O O O
0 o O O O O . 0 O Q a O
. . . 3 . . . . i . . .
'0.57 O O . é . O O r05" O O . ‘ . O O
-1.0— . ° , , ‘ . -1.0— 3
-110 05 0T0 05 {To -1i0 -015 050 015 110
1.0-C 3 1.0-D , 3 ,
o 0 § 0 o O O O .
05- ' ' 9 ' 05- ° ' 9 ' °
' O O O O O ' O O ¢ O ‘O
O o . O . O O O . O . o O
0.0 O O O 0.0 O O O
O o . O . o O O o . O . o O
as: O O O O O O O ‘ O O
-0.5‘ . . O . e '05‘ . . O . .
. o O O O o O O O
-1.0- ' -1.0-
b*
-110 0.5 010 015 110 -110 -015 010 015 110

Figure 4.16. Electron Diffraction patterns of restacked W82 simulated
using the multi-slice dynamic scattering calculation employed in the
CERIUS2 HRTEM module: (A) one layer thick (kinematic scattering)
(B) three layers thick (C) six layers thick and (D) eight layers thick.

As thickness increases, the relative intensities of the hexagonal sublattice
reflections increases.

162

scattering present in the sample. In this case, the data in Figure 4.2d is the
most free of dynamic scattering. Simulations of the diffraction patterns
from the models of MOS, were comparable, with the exception that the
(200) surpassed the (300) before 5 layers thick rather than after 6 layers
thick, as in the simulations involving WS,. One can also see that in both
data sets for MOS, the (200) is more intense than the (300), suggesting
contamination due to dynamic scattering.

The intensities were extracted from these simulated diffraction
patterns (an option in CERIUS”), and Patterson contour plots were
generated (Figure 4.17). In the kinematic pattern (Figure 4.17a), the
distortion in the W—W vector can be resolved as two distinct peaks, and the
W-S vectors are elongated and fairly intense. At three layers thick (Figure
4.17b) the W-W vectors are slightly more blurred and the W-S vectors are
more rounded. At six layers thick (Figure 4.17c) the W-W vectors are no
longer resolvable as two separate peaks, and the W-S vectors are
significantly reduced in intensity. At eight layers thick (Figure 4.17d)
there is almost no intensity in the W-S vectors and the distortion in the W-
W vector is only evident as a large, oval peak. If the data from this last
Patterson map is refined against the model that was used to create it the W-
W distance increases and the temperature factors of all the atoms become

very large. Based on the fact that the simulated electron diffraction

163

patterns for MOS, are comparable, would expect it to follow a similar
trend, but the deterioration of the data with thickness would be more rapid.

This loss of resolution of the M-M vector with increasing dynamic
scattering led us to conduct a kinematic simulation in which the W-W
distance was varied from 2.65 A to 2.90 A (Figure 4.18). The W-W
vector starts to blur between 2.70 and 2.80 A (Figures 4.18b and 4.18C),
and blurs completely into one oval peak between 2.80 and 2.90 A (Figure
4.18d). The W—S vectors become more rounded due to the decrease in the
distortion of the W atoms. The oval which describes the W-W vector in
these Patterson contour plots is smaller than the oval in the plots generated
as a function of thickness, but the most significant difference is that the W-
S vectors are not decreased in intensity with increasing W-W distance. One
might expect this feature to manifest itself as elevated temperature factors
for the S atoms, which is observed to a small degree in the MOS, data.

The CERIUS2 program is not designed to conduct least squares
refinements in conjunction with multislice n-beam calculations. In order to
try to evaluate the degree to which our data is contaminated by dynamic
scattering, R, values were calculated as a method to compare the simulated
dynamic data” with the experimental data:

211 6.1—K16. 11
—Ak
1” ' 20>.

hk

 

 

Eq.(4.10)

164

 

>b axis

 

a axis

 

Figure . c-axis
calculated from the data in Figure 4.16: (A) one layer thick (kinematic
scattering) (B) three layers thick (C) six layers thick and (D) eight layers
thick. As thickness increases, the W-W vectors become blurred and the
W-S vectors decrease in intensity.

165

b axis

a axis

 

Figure c-axis
calculated from simulated electron diffraction data (kinematic) as a
function of WW distance: (A) 2.65 A (B) 2.70 A (C) 2.80 A and

(D) 2.90 A. As the W-W distance increases, the W-W vectors become
blurred, but the W-S vectors retain their initial intensity.

166

Table 4.8. A comparison of simulated dynamic electron diffraction data
to the experimental data for M08, and WS, as a function of layer thickness.

 

 

Layer thickness R, for MOS, (%) R, for WS, (%)
1 15.38 13.27
2 14.75 13.21
3 15.44 14.77
4 19.03 19.58
5 23.15 24.98
6 26.19 29.92
7 28.14 34.76
8 30.13 39.81

 

Unfortunately the simulations did not result in significant
improvements over kinematic least squares refinement: the best R, value
calculated for the “single” W8, data set is only ~0.1% lower than the value
from the kinematic refinement, and the best value for M08, is only ~0.9%
lower. Part of the problem is that the atomic coordinates and the thermal
parameters have already been minimized in a kinematic refinement in
SHELX; if we had access to a program which would allow those

parameters to refine against a dynamic model the R, value might

167

improve.38

The lack of improvement also suggests that there are other
sources of perturbation of the data besides dynamic scattering. For
example, the correction which was applied to compensate for the limited
dynamic range of the electron microscope film may not have been
adequate. Another possible source of perturbation Of the data is the
flatbed scanner used to digitize the negative, as it is also somewhat limited

in dynamic range. Despite these difficulties, meaningful structural

information has been extracted from the electron diffraction patterns.

4.3.8 Electron Crystallographic Studies Of an alternate
superlattice in M08,

Recently, STM studies on the original lT-MOS,, prepared by the

oxidation of K,(H,O),Mos2 with 1,, indicated that it has a 3a x fl.

superstructure”, as proposed in the original publication reporting the
material.11 It is becoming clear that restacked M08, is not the same as IT-
MOS,. Other studies have indicated that restacked MoS,, previously
believed to be neutral because of its ability to encapsulate neutral species,
has some residual negative charge."0 The charge balancing species could be

Li°, H“, or H,O". This residual negative charge apparently stabilizes the

168

structure of restacked MOS,. STM studies of K,(H,O) yMOS, suggest that it
has the same superstructure as restacked MOS,.°l

As will be discussed in the next chapter, it is possible to produce this

alternate fl; x 1/3_a superlattice in restacked MOS, upon oxidation of the

sample with Br,. It is very unusual to observe this lattice without the
presence of the orthorhombic lattice solved in this chapter, but a few
“single crystal” patterns were obtained (Figure 4.19). We attempted to
elucidate the structure of this alternate lattice from the electron diffraction
patterns.

Figure 4.20a contains the Patterson map of an idealized octahedral

species with a 1/3_a x l/3_a lattice. The M-M vectors can be found at (1/3,

2/3) and (2/3, 1/3). Figures 4.20b and 4.20c are the Patterson maps
calculated from the experimental data in Figure 4.19. Although the M-M
vector is not spherical, there does not appear to be a splitting or even an
elongation of the M-M vectors as seen in the orthorhombic Patterson maps
Actually it seems that, for this system, even if there is a distortion in the
M-M distance it will not be visible in the Patterson map. In Figure 4.21a a
Patterson map has been simulated from a model (created using the
distortion model shown in Figure 4.1c) with a short M-M distance of 2.8
A. Although the M-M vectors are no longer spherical, there is no evidence

of a pair of M-M vectors representing the short and long distances in the

169

structure, even though the map was generated from a model which
contained such distances. In fact, if one distorts the M-M distance to a
ridiculously short value, such as 2.2 A, the M-M vector disappears almost
completely and the overall intensity of the peaks in the pattern is reduced
(Figure 4.21b). The problem is that the Patterson map is not a plot of the
electron density in the structure, but a plot Of the atom-atom vectors in the
structure. The M-M distortion generates distortion in the M-8 vectors as
well, which in this case complicates the Patterson map significantly more
than in the orthorhombic cell and results in an overall increase in the
background noise in the map. When the 8 atoms are eliminated from the
model, the distorted M-M vectors are easily observed (Figure 4.21c).
Although the Patterson did not provide much information about the
structure, we attempted unsuccessfully to refine the structure using the
proposed trimerization model shown in Figure 4.10, which belongs to the
plane group p31m (#15). This unit cell is larger than the orthorhombic
cell, and consequently contains more atoms. Even with maximum
constraint Of the parameters the data:parameter ratio was 4:1, which is
insufficient for a refinement. Data collection at higher accelerating
voltages, or employing tilt studies to collect higher order reflections, are
required for the determination of the structure of this compound from

electron diffraction data.

170

A A frctoals of A

.u

.91. Seede electron draction pattern
restacked M082 which have been oxidized with Br2, giving rise to a
43a x 43a superstructure.

 

 

 

 

 

 

(A) ideal 1T-MS2 (computed from simulated data) (B) from the electron
diffraction data shown in Figure 19a (C) from the data shown in

Figure 19b. Although the M-M vector is not spherical, the map does not
reveal a short M-M vector.

172

 

 

 

 

 

 

 

Figure 4. 21. Two- dimensional Patterson projections along the c- axis
calculated from simulated data: (A) a trimerization with a short M- M
distance of 2.8 A (B) a trimerization with a short M-M distance of 2.2 A
and (C) the same model without 8 atoms.

173

4.4 Conclusions

In summary, the structure of restacked WS, and M08, has been
determined from electron diffraction data. The 2-D Patterson projection
indicates unequivocally that the metal atoms are distorting to form zigzag
chains with a short W-W distance of 2.74 A and a short Mo-Mo distance of
2.92 A. These distortions are in agreement with earlier theoretical
predictions for layered octahedral ML, (12 systems.27 The results are
consistent with those from EXAFS studies, and they are more informative
because they provide direct structural information. Structure refinement
has been conducted to R,=l3.3% for W8, and R,=15.3% for MOS,. We
conclude that restacked M08, and WS, are not lT-TiS, type, but rather
WTe, type, and that quasi-kinematical electron diffraction data are

sufficient for meaningful structure elucidation.

174

References
l. Fleischauer, P.D. Thin Solid Films, 1987, 154, 309.
2. Dickinson, R.G.; Pauling, L. J. Am. Chem. Soc., 1923,45, 1466.

3. a) Harris, 8.; Chianelli, R.R. J. Catalysis, 1984, 86, 400. b) Brenner,
J.; Marshall, C.L.; Ellis, L.; Tomczyk, N.; Heising, J.; Kanatzidis,
M.G. Chem. Mater, 1998, 5, 1244.

4. Julien, C.; Saikh, S.I.; Nazri, G.A. Mater. Sci. Eng., 1992, B15, 73.

5. Murphy, D.W.; DiSalvo, F.J.; Hull, G.W.,Jr.; Waszczak, J.V. Inorg.
Chem, 1976, 15, 17.

6. a) Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, 8.R. Science, 1989,
246, 369. b) Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. J.
Mater. Res., 1991, 6, 1103. c) Gee, M.A.; Frindt, R.F.; Morrison,
8.R. Mater. Res. Bull., 1986, 21, 543.

7. a) Bissessur, R.; Heising, J.; Hirpo, W.; Kanatzidis, M.G.; Chem.
Mater., 1996, 8, 318. b) Wang, L.; Schindler, J.L.; Thomas, J.A.;
Kannewurf, C.R.; Kanatzidis, M.G. Chem. Mater., 1995, 7, 1753. c)
Kanatzidis, M.G.; Bissessur, R.; DeGroot, D.C.; Schindler, J.L.;
Kannewurf, C.R. Chem. Mater., 1993, 5, 595.

8. a) Lemmon, J.P.; Lerner, M.M. Chem. Mater.,. 1994, 6, 207. b)
Dungey, K.E.; Curtis, M.D.; Penner-Hahn, J.B. J. Catalysis, 1998,
175, 129.

9. Py, M.A.; Haering, R.R. Can. J. Phys, 1983, 61, 76.

10. Chianelli, R.R.; Scanlon, J.C.; Thompson, A.H. Mater. Res. Bull.,
1975, 10, 1379.

11. Wypych, F.; SchOllhorn, R. J. Chem. Soc., Chem. Commun., 1992,
1386.

12. a) Wilson, J.A.; DiSalvo, F.J.; Mahajan, 8. Adv. Phys, 1975, 24,
117. b) Wilson, J.A.; DiSalvo, F.J.; Mahajan, 8.; Phys. Rev. Lett.,
1974, 32, 882.

175

13. Yang, D.; Frindt, R.F. J. Phys. Chem. Solids, 1996, 57, 1113.

14. Yang, D.; Sandovals, S.J.; Divigalpitiya, W.M.R.; Irwin, J.C.; Frindt,
R.F. Phys. Rev. B. , 1991, 43, 12053.

15. Tsai, H.L.; Heising, J.; Schindler, J.L.; Kannewurf, C.R.; Kanatzidis,
M.G. Chem. Mater., 1997, 9, 879.

16. Chrissafis, K.; Zamani, M.; Kambas, K.; Stoemenos, J.; Economou,
N.A.; Samaras, I.; Julien, C. Mater. Sci. Eng., 1989, B3, 145.

17. Qin, X.R.; Yang, D.; Frindt, R.F.; Irwin, J.C. Ultramicroscopy,
1992, 42-44, 630.

18. Wypych, F. Weber, Th. Prins, R. Chem. Mater., 1998, 10, 723.

19. Vainshtein, B.K. Structure Analysis by Electron Difiraction;
Pergamon Press, MacMillan Co., New York, 1964.

20. Dorset, D.L. Structural Electron Crystallography ; Plenum Press,
New York, 1995.

21. Stout, G.H.; Jensen, L.H. X-ray Structure Determination: a Practical
Guide (2nd ed.); John Wiley & Sons, New York, 1989.

22. Ladd, M.F.C.; Palmer, R.A. Structure Determination by X-ray
Crystallography (3rd ed.); Plenum Press, New York, 1994.

23. Analysis of integrated intensities was performed on a Macintosh
Performa 6214CD computer using the public domain NIH Image 1.60
program (developed at the US. National Institutes of Health and
available on the Internet at http://7/2/98/rsb.info.nih.gov/nih~image/)
using the “Gel Plotting Macros” plug-in.

24. Sheldrick, GM. “SHEL'I'XTM Version 5”, Bruker Analytical X-Ray
Instruments, Inc., Madison, WI.

25. Hyarns, Daniel. “CurveExpert Version 1.34”, 1995-1997. Portions
copyright (1993) Microsoft Corporation, Seattle, WA.

26. a) Brown, B.E. Acta Crystallogn, 1966, 20, 268. b) Mar, A.; Jobic,
8.; Ibers, J. A. J. Am. Chem. Soc., 1992, 114, 9587.

176

27.

28.

29.

30.

31.

32.

33.

34.
35.
36.
37.

38.

39.

a) Rovira, C.; Whangbo, M.-H. Inorganic Chem, 1993, 32, 4094. b)
Whangbo, M.-H.; Canadell, E. J. Am. Chem. Soc., 1992, 114, 9587.

Bissessur, R.; Kanatzidis, M.G.; Schindler, J.L.; Kannewurf, C.R. J.
Chem. Soc., Chem. Commun., 1993, 1582.

a) Pinsker, Z.G.; Dvoryankina, G.G. Kristallografiya, 1958, 3, 438.
b) Nagakura, 8. Acta Crystallogr., 1957, 10, 601.

Q = A |<I>/ Q I, where (I) is the structure factor, Q is the volume of the
unit cell, and A is the wavelength of the electrons.

Klug, P.; Alexander, L.E. X-Ray Diffraction Procedures for
Polycrystalline and Amorphous Materials, John Wiley and Sons, New
York, 1954.

a) Joensen, P.; Crozier, E.D.; Alberding, N.; Frindt, R.F. J. Phys. C.:
Solid State Phys, 1987, 20, 4043. b) Prouzet, E.; Heising, J.;
Kanatzidis, M.G. Unpublished results.

“CERIUS2 Version 3.5 HRTEM module”, Molecular Simulations, Inc.,
San Diego, CA.

Cowley, J.M.; Moodie, A.F. Acta Cryst., 1957, 10, 609.
Cowley, J .M. Difi’raction Physics; Elsevier, New York, 1975.
Spence, J.C. H. Acta Cryst., 1998, A54, 7.

Sinkler, W.; Marks, L.D.; Edwards, D.D.; Mason, T.O.;
Poeppelmeier, K.R.; Hu, 2.; Jorgensen, J .D. J. Solid State Chem,
1998, 136, 145.

a) Zandbergen, H.W.; Andersen, S.J.; Jansen, J. Science, 1997, 277,
1221. b) Jansen, J.; Tang, D.; Zandbergen, H.W.; Schenk, H. Acta
Cryst., 1998, A54, 91.

“CERIUS2 Version 3.8 HRTEM module”, Molecular Simulations, Inc.,

San Diego, CA. (According to the company, this newer software
version contains an “improved” dynamic scattering calculation,

177

particularly with regard to the incorporation of the temperature factors
of the atoms in the calculation.)

40. Heising, J.; Bonhomme, F.; Kanatzidis, M.G. J. Solid State. Chem,
1998, 139, 22.

41. Wypych, F.; Weber, Th.; Prins, R. Surf. Sci., 1997, 380, L474.

178

CHAPTER 5
Exfoliated and Restacked M08, and WS,: Ionic or Neutral

Species? Encapsulation and Ordering of Hard Electropositive
Cations

179

5.0 Abstract

The relationship between charge and structure in restacked M8, (M =
MO,W) has been probed by cation encapsulation and chemical oxidation and
characterized by elemental analysis, electron diffraction, X-ray diffraction,
and Differential Scanning Calorimetry. Alkali cations have been
encapsulated in M08, and WS, without the presence of a co-intercalated
counter-ion, suggesting a negative charge in the range 0.15-0.25 electrons
per M atom. Electron diffraction studies show ordering of these cations

between the layers. Chemical oxidation with I, or Br, results in a change

in the structure of restacked MoS,, giving rise to a l/3_a x /3—a

superlattice, whereas no change is observed in the structure of restacked
WS,. Differential Scanning Calorimetry studies show an irreversible
exothermic transition to 2H-MS, upon heating, which shifts in temperature
with oxidation. Thermopower measurements indicate that restacked M08,
and WS, are p-type metallic conductors, consistent with an observed

structural distortion and residual negative charge.

180

5.1 Introduction

2H-MOS, and 2H—W8, are the most stable members of the family of
layered dichalcogenides, and find practical application in such processes as
hydrodesulfurization (HDS) due to their availability and relatively low
cost. ' Both materials strongly resist attempts to incorporate guest species
between their layers.2 Li atoms can be inserted into M08, (and WS,) only
upon treatment with a strong reducing agent such as n-butyllithium (n-
BuLi) or LiBH,.3'° LiMoS, and LiWS,, however, exhibit a remarkable

ability to exfoliate in water by the following redox reaction:

LiMS, + H,O --—> (MS,),,,,,, m + LiOH + H, (g) (Eq. 5.1)

resulting in a colloidally dispersed suspension of single layers.”5 The
layers can be recovered in a restacked form by filtration, centrifugation, or
precipitation. The ability of these materials to exfoliate and to be restacked
with relative ease has permitted the encapsulation of a wide variety of guest
species which include neutral organic molecules,6 polymers,7 metal
chalcogenide8 and metal oxide9 clusters, metallocenes”, porphyrins”, and
metal cations?”13

The authors who first reported the exfoliation Of MOS, believed that

the redox reaction between H,O and M08, resulted in complete oxidation of

the layers, returning the molybdenum to a 4+ oxidation state, because many

181

of the molecules which have been encapsulated are neutral.‘5°8'14 A net drift
of the suspended layers toward the anode in an electrophoresis apparatus
was observed by these authors, but was attributed to OH' ions associated
with the layers which could be displaced by the organic molecules.15

Subsequent pH studies seemed to support this hypothesis.“5

The increasing
number of examples of cationic species which have been encapsulated have
suggested to other authors that the layers may retain some negative
charge.“"°“3'l7 It is rather curious that restacked M082 (and W82) can
incorporate both cationic and neutral species. The ambiguity of the
situation stems from an abundance of both Li* and OH‘ generated in the
exfoliation process: Li‘, which could counterbalance a negative charge on
the layers when neutral species are intercalated, is a notoriously difficult
element to detect; OH’, which could co—intercalate with the cationic species,
can be confused with residual co-intercalated H20.

In addition to the unresolved issues about the oxidation state of the
layers, there has also been confusion about the atomic structure of the
layers. Reduction of 2H-MoS2 with n-BuLi or LiBH4 results in a structural
transformation in the layers, causing the geometry around the metal atoms
to shift from trigonal prismatic to octahedral.18 This results in a change in
the conductive properties of the material from semiconducting to

18.19

metallic. This structural change appears to be retained in exfoliated

182

MoSz, with a structural distortion which results in a a x l/3_a

orthorhombic superlattice?“21 Restacked MoS2 has also been called 1T—
MoS2 because of the octahedral coordination of the Mo atom. Restacked
WS2 appears to be analogous to restacked MoSz.“'21 Another material which
has been called 1T--MoS2 is prepared by high temperature synthesis and
subsequent oxidation of K(,_33(H20)yMoSZ.22 Restacked M082 and lT-MoS2
were thought to be the same material because both appear to have

octahedral metal coordination and both undergo an irreversible exothermic

transition to 2H-MoS2 at around 100°C.”23 lT-MoSZ, however, exhibits a

1/51 x 1/371 superlattice.”24 The a x 1&1 superlattice found in restacked

M082 is due to M-M associations resulting in the formation of zigzag

chains, whereas a 1/321 x [[321 superlattice would involve M-M

trimerization.

Are these layers neutral or cationic? Why do 1T-MoS2 and restacked
MoS2 have different superlattices? What is the impact of negative charge
on the structure of these materials? If the M082 layers are slightly reduced,
how does it affect the conductivity? In this chapter we describe our
experimental efforts toward answering these questions. The encapsulation
of hard, electropositive alkali metal cations is described, chosen due to

their relatively poor affinity for OH,’ which make them perhaps the best

183

species for chemical analyses to determine negative charge. The materials
are characterized by X-ray diffraction, electron diffraction, TGA, and
elemental analyses. The treatment of LiMS2 and exfoliated MS2 with the
oxidizing agent Br2 and concentrated HCl has also been conducted and the
products characterized by X-ray diffraction, electron diffraction, and DSC
measurements. Thermopower measurements of restacked M082 and W82,
which address the effects of structure and charge on the nature of their

conductivity, are also presented.

5.2 Experimental

5.2.1 Synthesis

LiMoS2 and LiWS2 were synthesized by the LiBH4 method.4 Acetonitrile,
RbCl, CsF, and CsI were purchased from Aldrich. KCl and were
purchased from J .T. Baker. Br2 and NaCl were purchased from EM
Science. I2 was purchased from Mallinckrodt. HCl was purchased from
Columbus Chemical Industries. All compounds were used as received.

MoS2 aqueous suspension. In a glove box under nitrogen atmosphere
LiMoS2 (0.1 g, 0.6 mmol) was placed in a 125 ml Erlenmeyer flask

equipped with a stir bar and rubber stopper. The flask was removed from

184

the glove box and 10 ml deoxygenated deionized H20 was added. The
mixture was allowed to stir for 0.5 hr, then the contents of the flask were
transferred to a centrifuge tube and centrifuged for 0.5 hr. The
supernatant was decanted (pH >12) and to the black gooey solid at the
bottom of the tube 10 ml fresh deoxygenated deionized H20 was added to
rinse away the LiOH. The tube was agitated to re-suspend the product, and
the mixture again centrifuged for 0.5 hr. This rinsing was conducted
three times, the supernatant having pH ~12, pH ~9, and pH ~7 after each
rinse. Then 10 ml more deoxygenated deionized H20 was added, the tube
was agitated to re-suspend the product, and the contents of the centrifuge
tube were returned to a 125 ml Erlenmeyer flask with stir bar. The
suspension was stirred for at least 0.5 hr before further use.

WS2 aqueous suspension. In a glove box under nitrogen atmosphere
LiWS2 (0.1 g, 0.39 mmol) was placed in a 125 ml Erlenmeyer flask
equipped with stir bar and rubber stopper. The flask was removed from
the glove box and 10 ml deoxygenated deionized H20 was added. The
material was then centrifuged, rinsed, and re-suspended in the same
manner as MoSz.

Na0.,4(H,O)mMoSZ. NaCl (0.11 g, 1.8 mmol) was dissolved in about 3.5
ml H20. To this solution was added an aqueous suspension of MoS2 (3:1
excess Na”). Flocculation occurred immediately. The reaction mixture

was stirred for one half hour, centrifuged and the supernatant decanted.

185

The product was rinsed and centrifuged three times as described in the
preparation of aqueous MoS,, then deposited on a glass slide to dry.
K013(H20)MMoSZ. KCl (0.13 g, 1.8 mmol) was dissolved in about 3.5 ml
H20. To this solution was added an aqueous suspension of MoS2 (3:1 excess
K*). Flocculation occurred immediately. The product isolation was
analogous to that used for Na0.,4(H20)0.45MoSZ.

Rbo.,5(H20)oJMoSr RbCl (0.22 g, 1.8 mmol) was dissolved in about 3.5
ml H20. To this solution was added an aqueous suspension of MoS2 (3:1
excess Rb*). Flocculation occurred immediately. The product isolation
was analogous to that used for Nam 4(I-120)0.45M082.

Cso.,3(HzO)yMoS,. CsF (0.27g, 1.8 mmol) was dissolved in about 2 ml
H20. To this solution was added an aqueous suspension of MoS2 (3:1 excess
Rb“). Flocculation occurred immediately. The product isolation was
analogous to that used for Na0.,4(H20)0_45MoSZ.

Ba,(H,0),Mos,. BaC12'2H20(0.88g, 3.6 mmol) was dissolved in about 5
ml H20. To this solution was added an aqueous suspension of MoS2 (6:1
excess Ba2+ ). Flocculation occurred immediately. The product isolation
was analogous to that used for Na0.,4(H20)0_45MoSZ.

Na0.,,(H20)M,WS,. NaCl (0.14 g, 2.4 mmol) was dissolved in about 3.5
ml H20. To this solution was added an aqueous suspension of WS,, (6:1
excess Na*). Flocculation occurred immediately. The product isolation

was analogous to that used for Na0.,4(H20)0.45M082.

186

K03,(H20)M,W82. KCl (0.18 g, 2.4 mmol) was dissolved in about 5 ml
H20. To this solution was added an aqueous suspension of WS2 (6:1 excess
K”). Flocculation occurred immediately. The product isolation was
analogous to that used for Nao.,4(H20)0.45MoS2.
Rbo.24(H,O)oMWSZ. RbCl (0.15 g, 1.2 mmol) was dissolved in about 3.5
ml H20. To this solution was added an aqueous suspension of WS2 (3:1
excess Rb“). Flocculation occurred immediately. The product isolation
was analogous to that used for Na0.,4(H20)0.45MoSZ.
Cso.,3(H20)yWS,. Method 1. An aqueous suspension of WS2 was added
to 5 ml of a concentrated CsF solution in H20. To this solution was added
an aqueous suspension of WS2 (7:1 excess Cs*). Flocculation occurred
immediately. The product isolation was analogous to that used for
Nao,14(HzO)o,45MoSZ.

Method 2. C51 (0.73 g, 2.8 mmol) was dissolved in about 5 m1
H20. To this solution was added an aqueous suspension of WS2 (7:1 excess
Cs“). Flocculation occurred immediately. The product isolation was
analogous to that used for Na0_,,(H20)0_45MoSZ.
Baom(H20),WS,. BaC12°2H20 (0.29 g, 1.2 mmol) was dissolved in about 5
ml H20. To this solution was added an aqueous suspension of WS2 (3:1
excess Ba2+ ). Flocculation occurred immediately. The product isolation

was analogous to that used for Na0.,4(H20)0.45MoS2.

187

HxMoSZ. In a glove box under nitrogen atmosphere LiMoS2 (0.3 g, 1.8
mmol) was placed in a 125 ml Erlenmeyer flask equipped with a stir bar
and rubber stopper. The flask was removed from the glove box and 30 ml
cold concentrated HCl was added. The reaction mixture was stirred for a
few minutes, then was transferred to a centrifuge tube and centrifuged for
a few minutes. The supernatant was decanted, then 30 ml cold H20 was
added to the solid to rinse, and the slurry was centrifuged again. This
rinsing was repeated two more times, the supernatant after each rinse
having pH~1, pH~1, and pH~7, respectively. The solid was then slurried in
a few ml of H20 and deposited on a glass slide to dry overnight.

H,WS,. In a glove box under nitrogen atmosphere LiWS2 (0.3 g, 1.2
mmol) was placed in a 125 ml Erlenmeyer flask equipped with a stir bar
and rubber stopper. The reaction procedure and rinsing was the same as
the one used to prepare HxMoSz, with the supernatant after each rinse
having pH~1, pH~4, and pH~7, respectively. The solid was then slurried in
a few ml of H20 and deposited on a glass slide to dry overnight.

Oxidized LiMoS,. Method 1. In a glove box under nitrogen
atmosphere LiMoS2 (0.3 g, 1.8 mmol) was placed in a 125 ml Erlenmeyer
flask equipped with a stir bar and rubber stopper. The flask was removed
from the box, and 50 ml of a 0.18 M solution of I2 in acetonitrile or 50 ml
of a 45 mM solution of Br2 in acetonitrile was added to the solid (5:1

excess Ilerz). The mixture was stirred for different amounts of time

188

ranging from 15 minutes to 4 days, then was centrifuged and the
supernatant decanted. The product was rinsed by re-suspending it in 15 m1
acetonitrile and re-centrifuging 2 times and then deposited on a glass slide
and a saran wrapped transmission XRD slide to dry.

Method 2. In a glove box under nitrogen atmosphere LiMoS, (0.15
g, 0.9 mmol) was placed in a 125 ml Erlenmeyer flask equipped with a stir
bar and rubber stopper. The flask was removed from the box and an
exfoliated suspension of MoS, was prepared as described previously, then
centrifuged and rinsed once with acetonitrile. To the solid 50 ml of a 0.18
M MeCN solution of I, or 50 ml of a corresponding 45 mM MeCN solution
of Br2 was added (5:1 excess I,/Br,). The mixture stirred for different
amounts of time ranging from 15 minutes to a week, was centrifuged and
the supernatant decanted. The product was rinsed by re-suspending in
acetonitrile, re-centrifuging and decanting the supernatant. The product
was then deposited on a glass slide and a SaranTM wrapped transmission X-
ray Diffraction (XRD) slide to dry.
Oxidized LiWS,. Method 1. In a glove box under nitrogen
atmosphere LiWS, (0.3 g, 1.2 mmol) was placed in a 125 ml Erlenmeyer
flask equipped with a stir bar and rubber stopper. The flask was removed
from the box, and to the solid 50 ml of a 0.12 M MeCN solution of I, or
corresponding 75 ml 80 mM MeCN solution of Br, was added (10:1 excess

I,/Br,). The mixture stirred for varying amounts of time from ~15

189

minutes to a week, was centrifuged and the supernatant decanted. The
product was isolated as described for oxidized LiMoS,.

Method 2. In a glove box under nitrogen atmosphere LiWS, (0.15
g, 0.6 mmol) was placed in a 125 ml Erlenmeyer flask equipped with a stir
bar and rubber stopper. The flask was removed from the box and an
exfoliated suspension of WS, was prepared as described previously, then
centrifuged and rinsed with acetonitrile. To the solid 50 ml of a 0.12 M
MeCN solution of I, or corresponding 75 ml 80 mM MeCN solution of Br,
was added (10:1 excess I,/Br,). The reaction time and product isolation
was the same as described above.

Method 3. LiWS, (0.15g, 0.6mmol) was used to prepare an
exfoliated suspension as described previously. Next 100ml of a saturated
aqueous solution of Br, was added. The product was centrifuged
immediately for a few minutes, then rinsed with water and centrifuged
three times. It was deposited on a glass slide and a SaranTM wrapped

transmission XRD slide to dry.

190

5.2.2.Characterization.

Selected Area Electron Diffraction patterns were collected using a
JEOL 120 CX Transmission Electron Microscope (TEM). Transmission
and reflection powder X-ray diffraction patterns were collected using a
Rigaku-Denki/RW400F2 (Rotaflex) rotating anode X-ray diffractometer
using Cu-Ka radiation. Differential Scanning Calorimetry (DSC)
measurements were conducted using a Shimadzu DSC-50 instrument under
nitrogen flow in sealed aluminum containers at a heating rate of
5°C/minute. Thermogravimetric Analyses (TGA) employed a Shimadzu
TGA-50 instrument under nitrogen flow in quartz containers at a heating
rate of 2°C/minute. Energy Dispersive Spectroscopy (EDS) was used to
determine the alkali metal to Mo/W ratio, and employed a JEOL-ISM-
6400V Scanning Electron Microscope was operated at an accelerating
voltage of 30kV using a Tracor Northern 5500 X-ray microanalysis
attachment. Thermopower measurements utilized a MMR Technologies

Seebeck System.

191

5.3 Results and Discussion

5.3.1 Cation encapsulated MS,.

Addition of exfoliated MOS,./W S, to the alkali halide solution results
in immediate flocculation of the suspension. This rapid precipitation,
which is not usually observed upon addition of neutral species, strongly
suggests that the interaction of the layers with the alkali cations is ionic in
nature. Similar behavior is observed upon addition of exfoliated MoS,/W S,
to solutions containing Al,,0,,(OH),,,(H,O)1,7+ or tetraphenylphosphonium
cations.”25

The samples were characterized by reflectance X-ray diffraction
(Figure 5.1a) and were found to be well ordered, particularly compared to
the X-ray diffraction patterns of samples containing neutral species such as
the cobalt chalcogenide clusters.8 The predominance of the 001 reflections
in the pattern is due to the tendency of lamellar materials to adopt a
preferred orientation. All samples exhibited interlamellar d-spacings in the

range 9.3-9.7 A (an expansion of 3.1 - 3.5 A), indicating co-encapsulation

of approximately one monolayer of water with the cations.

192

 

 

Intensity (arbitrary units)

 

 

 

 

 

 

weight loss (%)

 

 

 

 

96.5E‘J—‘l‘411111111111r111
50 100 150 200 250

Temperature (°C)

 

Figure 5.1. (A) X-ray diffraction pattern of Nao.,,(H,O)o.45WS,.,
exhibiting a d-spacing of 9.7 A (A: 3.5 A). (B) TGA of Na0_,,(I-I,O)o.45WS,
(dotted line), Ko.21(1'120)o.4owsz (daShed line)’ and Rbo.24(HzO)o.34WS2 (SOIid
line). Weight loss corresponds to H,O loss.

193

The flocculated products were found to have the following formulas:
Nao,|4(1'120)xMOS,, K0,,(H,O),MoS,, Rbo.,5(H,O),MoS,, Cso_,3(H,O),MoS,,
Nao.13(HzO),WS,, Ko_,,(H,O),WS,, Rb0_,4(H,O),WS,, Cso_,3(H,O),WS,, and
Ba0.08(H,O),WS,. . TGA (Thermal Gravimetric Analysis), performed in
order to determine the amount of co-encapsulated water, indicates a range
of 03-045 for x (Figure 5.1b). Higher x values were found for samples
containing Na+ and K*; lower values for those containing Cs” and Rb“. No
halide counter-ions were detected. Although the presence OH- ions cannot
be completely discounted, the pH of the suspension (~7) and the relatively
poor affinity of the alkali cations for OH- ions make their co-encapsulation

unlikely.

5.3.2 Electron diffraction studies of (cation),MS,.

The cation intercalated M08, and WS, samples were also examined
by electron diffraction. Both restacked M08, and WS, give rise to electron
diffraction patterns which appear to be more simple than they actually
are.21 In Chapter 4, we solved and refined the structure of restacked M08,
and WS, from electron diffraction data.“ The materials exhibit a
superlattice caused by metal-metal associations to form zig-zag chains,

similar to those observed in W'I'e,.26 This zig-zag distortion results in the

194

doubling of one axis of the hexagonal sublattice. As the hexagonal

symmetry is destroyed, the new lattice is best described by an I/3_a x a

orthorhombic cell (Figure 4.2d, Figure 4.3). Due to their turbostratic
nature and the disorder introduced by the exfoliation/flocculation process,
the diffraction patterns of exfoliated and restacked MoS, and WS, are
often a “triplet” of these orthorhombic cells, forming an apparent 2a x 2a
superlattice (Figure 4.2b). Failure to recognize this twinning can lead to
incorrect interpretations of the M-M interactions.

The cation intercalated samples are represented by an array of
diffraction patterns (Figure 5.2) which contain an additional superlattice
due to ordering of the electropositive cations in the gallery. The high
vacuum found in a TEM (~10'7 torr) probably results in the removal of
most of the co-encapsulated water from the samples. The majority of the
patterns appear at first glance to contain a weak 6a x 6a superlattice
relative to the hexagonal sublattice, with distinct variations in the
reflections present and the intensity distribution from one pattern to the
next. Because of the propensity of these materials for twinning, however,

the diffraction patterns are actually quite complicated.

195

 

u

‘n

#14, I

. ‘ mi ' -’ ' '4.
Figure 5.2. SAED patterns of (A) Cso,23M082 and (B) Cso,13W82. Super-
lattice reflections caused by Cs+ ordering indicate a tripling of the a axis
and the 43a axis, indicating a 3430 x 30 supercell. Twinning of the lattice
projects a 6a x 6a unit cell.

196

Considering only the orthorhombic lattice caused by the formation

of zig-zag M-M chains in the host material,21 Figures 5.2a and 5.2b contain

one dominant flh x a orthorhombic cell with only weak twinning. In the

interest of clarity, this /3_a x a orthorhombic lattice will henceforth be

called the “host-lattice”. A closer examination of Figure 5.2a reveals that

the superlattice caused by the intercalated cations form strong columns of

reflections along the 1/3_a* axis of the host-lattice. The superlattice

reflections are spaced in thirds between the host-lattice reflections. This
suggests that at least one cell parameter should be 3a. The reflections at
1/2 the spacing between the columns of host-lattice reflections can be
attributed to the weak twinning of the host-lattice; thus, the same
phenomena which give rise to a false 2a x 2a lattice also give rise to a false
6a superlattice parameter. Interestingly, there do not appear to be any
superlattice reflections from the cations contained in the columns or rows
of host-lattice reflections in Figure 5.2a. Considering one of the columns
of superlattice reflections to find the second cell parameter, the smallest
distance between two reflections is 1/3 the length of the (010) reflection of

the host-lattice; therefore, the superlattice may be minimally described by

the orthorhombic cell 3a’ x 3b relative to the host-lattice (a’ = 3a; b =

197

3a), and 31/32: x 30 (orthorhombic) relative to the hexagonal sublattice of

the host material (Figure 5.2b).
It is logical to conclude that, if the host-lattice is somewhat twinned,
the superlattice will also be somewhat twinned. In this manner the

variations in the other diffraction patterns (Figures 5.3 and 5.4) can be

explained by the degree of twinning of the 1/51 x a orthorhombic host-

lattice and the 3a' x 3b cation superlattice. Figure 5.3a is very similar to
Figures 5.2a and 5.2b, but comes from K0,,MoS, rather than Cs0,3MoS,
(Figure 5.2a) or CstS, (Figure 5.2b). The host-lattice is more twinned
than the pattern in Figure 5.2a, but the pattern of cation ordering is
essentially the same. Figures 5.3b and 5.3c are from crystals of K0,,3MoS,
and Rb0_,4MoS,, respectively, that are twinned but are still relatively thin.
The distribution of orthorhombic host-lattice cells is uneven, appearing to
be stronger in one direction. Consequently, the cation superlattice in these
pictures is still stronger in one direction, but does show evidence of

twinning. Unlike the previous patterns, the cation superlattice reflections

are now found in the columns of reflections along the l/3_a * axis of the

dominant orthorhombic host-lattice (arrows, Figure

198

 

3'.’
v
.-

Figure 5.3. SAED patterns of (A) K023MoSZ, which resembles the patterns
in Figure 5.2 but the degree of twinning is slightly greater; (B) K033MoSZ
and (C) Rbo. tsMoSZ, in which the host-lattice is more twinned but the
crystals are still relatively thin; and (D) Ko_23Mosz, from a thick crystal
with a very twinned host-lattice and a "halo" of cation superlattice
reflections. As twinning of the host-lattice increases, twinning of the cation
superlattice increases, leading to a variety of diffraction patterns.

199

 

. a: ‘

Figure 5.4. SAED patterns from (A) a thick crystal of LixWS2 and (B) a
thin crystal of RbonMoSz. Streaking is observed due to disorder of the
cations. The cations are more susceptible to disorder along the ‘13a axis.

200

5.3b). This seems to be due to cation superlattices associated with the other
(weaker) host-lattice cells. Figure 5.3d, on the other hand, came from a
thicker crystal of K0_,3MoS,. In this case, the twinning of the host-lattice is
so pronounced that that the host material exhibits the pseudo 2a x 2a
hexagonal lattice which is common in thicker crystals. The crystal is thin
enough that the superlattice from the intercalated cations is still visible, but
extremely weak. It is twinned in the same manner as the host-lattice,
creating a hexagonal “halo” around the pseudo 2a x 2a lattice. The
systematic absences of the superlattice (for example, the 020, 130, and the
310 ) are completely destroyed due to the twinning, generating the pseudo
60 x 60 cation superlattice.

Not all crystals gave rise to well ordered cation superlattices. Figure
5.4a comes from a rather thick sample of exfoliated WS, that contained a
large excess of LiOH in solution when it was flocculated; there is evidence
of Li” cations ordering in the sample, but the pattern is very weak and,
instead of spots, one sees faint streaking. The streaking is likely due to the
fact that Li, in addition to being a very light atom, is probably coordinated
by four water molecules, which would prevent it from resting easily in the
sites created by the S atoms. The pattern best resembles Figure 5.3d.

Crystals which contain discrete superlattice spots are somewhat unusual;

201

more common are patterns which contain disorder or streaking along the

l/3_a* axis of the host lattice, as in the waMoS, crystal in Figure 5.4b.

5.3.3 Modeling the cation ordering.

It is virtually impossible to obtain a diffraction pattern which is
completely “single crystal” with respect to the host-lattice and the cation
superlattice so that this manner of indexing the patterns might be
confirmed; however, it is the best available indexing which can account for
the variations in intensities and reflections from one diffraction pattern to
the next, all of which can be found within one sample. As such, we have
developed a cation ordering model from which an electron diffraction
pattern was simulated. It has been designed to resemble the diffraction
patterns in Figures 5.2a, 5.2b, and 5.3a, since they are the closest to a
“single crystal” pattern (Figure 5). As noted previously, no superlattice
reflection is found within a column or row of host-lattice reflections
(Figure 5.2a). The model of cation ordering, depicted in Figure 5.5a,

gives rise to such an electron diffraction pattern (Figure 5.5b).27

202

 

 

 

A 1 . . . é . 0 O

E o o o o O i... O o

c _ . g .

I: '5

m 0 o . o . o 5.0 . o . o 0

an, . . 5 .

g o o o O o 0.0; Q o .0 O o

_ O : o

8 m--. ------ Q ------ -o ----- .--no-o-a-u-«En---o-o-o----.----o--o «- .— o
O : 0

e o O o. o Q g... o O 0.0 o

a“ o E O

.5 o o . o o O 0.; o O .0 . o O

o 3.

I-t — , ,

V 0 0 DE 0 O o o

9(- o E

'6 d 0 Q o . o
o E o

l l l I i l l l I

 

 

 

d* (reciprocal angstroms)

Figure 5.5. (A) Structural model and (B) simulated electron diffraction
pattern to explain the observed SAED patterns in Figures 5.2-5.4. Filled
circles represent cations in one layer; open circles represent cations in a
second layer. The projection of the two results in a 3V3a x 30 supercell.

203

The cell contains three chains. Alkali cations are placed along the chains in
pockets defined by sulfur atoms, where a cation would be expected to
reside in the gallery. One third of the sites are empty in two of the chains;
two thirds of the sites are empty in the third chain. So many alkali cations
would not normally be in such close proximity to one another within one
layer, however, particularly when one considers the elemental analyses of
these samples. More likely this picture represents the superposition of two
alternating layers of cations, depicted as white and black. Hence, four of
the cations have an occupancy of 0.5 and one cation is fully occupied. This
results in a stoichiometry of A0.,67MS,, which is consistent with the
experimentally determined formula. Note that it is a necessary condition
of the systematic absences that all the alkali cations be in the same kind of
site, either “down” or “up” relative to the zig-zag chain. The alkali cations
occupy only sites which are pointing “up” in the model presented in Figure
5.5a; however, the sites which are pointing “down“ are equally valid sites,

so it becomes apparent that this is a possible source of disorder in the

samples. The streaking which is observed along the 1/3-a * axis in Figure

5.4b indicates that the cations are more disordered along the 1/33 axis of

the host-lattice. This is reasonable when one considers that the M atom
distorts from its ideal octahedral position along this axis, causing an

accompanying shift of the S atoms and distorting the site on which the

204

alkali cation sits. Also, the fact that there seems to be no disorder along the
0 axis indicates that the cations are better ordered along the zig-zag chains
than between zig-zag chains.

A few of the electron diffraction patterns did not have this 3a x 3b

superlattice, but had a rather different pattern (Figure 5.6) corresponding

to a 1/3-a x 1/3—a lattice combined with the pseudo 2a x 2a lattice of the
host material, forming a “honeycomb“ pattern. The projection of this 1/33

x 1/3_a unit cell occupies a smaller area than the previous superlattice (31

A2 vs. 161 A2), which is puzzling when one considers that it is most readily
observed for samples in which Ba“ is encapsulated. To reconcile such a
lattice with the observed loading of cation one must again employ
alternating layers of cations and consider their projection onto the ab plane.
For example, in Figure 5.7a a model has been constructed with four layers

of cations, depicted as white, black, gray, and striped circles. The unit cell

per MS, layer is 2 3a x 2 3a (solid line), but the projection is the

smaller 3a x [[321 lattice (dashed line). This gives a stoichiometry of

A0.09MS,. The simulated diffraction pattern generated from this cation
model (Figure 5.7b) resembles the experimental data.27 It is now possible
to understand why cations of different charge can have the same diffraction

pattern - they can occupy half as many sites and require twice as many

205

layers to project the appropriate unit cell. This is consistent with the fact
that the superlattice is weaker in the samples containing Ba2+ than it is in

the samples containing N a+ or Rb“.

206

 

Figure 5.6. SAED patterns of (A) BaoggMOSz and (B) Nao_14MoSZ. The
combination of a V3a x ‘13a cation superlattice with a pseudo 2a x 2a host-
lattice gives rise to a "honeycomb" pattern. This alternate cation super-
lattice, although most common for samples with encapsulated Ba2+, was
also observed within samples which exhibited the 3V3a x 3a superlattice.

207

 

 

 

 

A l
C
B o E o

o . o 3 o . o
A '7 5
U2 0 ; o o
a o o o : o o O
O _ o + o
I: o 0 Q o g Q . Q o o
{I} i
3.. 0 O 9 O o
a o 0 g 0 o
a -—----o ------ .. ------ —O ------ .- ------ ...-----.§ ------- O ------ . ------ -o ------ .— ------ on-
8 O O i O o
a— o . o . o $ 0 . o . o
o_ i
Q I o . o . E . o . o 0
cu :
b — o O ? O

o o o . o o 0
cu. o o E o o
c d O 0 Q ‘ O 0 O
.1 O . O : O . O O
l
I o
l l I I i l l l l

 

 

 

d* (reciprocal angstroms)
Figure 5.7. (A) Structural model and (B) simulated ED pattern to explain
the SAED patterns in Figure 5.6. Black circles represent one cation layer;
open circles a second; grey a third; and striped a fourth. Each layer has a
2V3a x 243a lattice, but the ‘13a x V3a lattice is formed by the projection.

208

5.3.4. Oxidation state of M (M = Mo,W) and the charge Of
the layers.

The fact that one can see ordering of alkali cations in the MOS, and
WS, layers without co-intercalation of the halide counter-ions supports the
conclusion that the exfoliated and restacked layers retain residual negative
charge. The layers in suspension behave as if they are solvated
macroanions, precipitating readily in the presence of most cationic species.
The hypothesis that the charge is due to the association of OH' ions with the
basal planes seems unlikely because the driving force for the association of
the OH' ions with the layers is not obvious. Furthermore, the pH Of the
exfoliated suspension has been reduced to ~7 by repeated rinsing. It seems
surprising that any remaining OH' ions would not be at least partially
displaced in a solution with an excess of halide ions, resulting in materials
with co-encapsulated halide ions. Unfortunately the. high degree of disorder
and the propensity of the systems for twinning make it impossible to
determine the degree of negative charge on the layers based on the cell
parameters of the samples with encapsulated cations.

The amount of negative charge on exfoliated and restacked MOS, has
been studied rather extensively by a group of Russian scientists.”l7 They
assert that MOS, which has been freshly exfoliated under an inert

atmosphere has a negative charge of approximately 015-029 per Mo atom

209

(Li+ counter-ion), but if the dispersion is allowed to sediment over time the
value decreases to 0.07-0.12 per Mo atom.17m This slow, continuous
oxidation of MOS, suspensions in solution over time may explain how it is
possible to intercalate the neutral cobalt chalcogenide clusters, which
require stirring for at least 2 days before encapsulation is observed.8 This
same group has also observed that in some transition metal systems it is
possible to co—intercalate OH’ anions, thereby increasing the number of
cations which may be encapsulated between the layers beyond the range of
negative charge attributed to MOS,.'3° The ability of MOS, and WS, to
accept co-intercalated OH' ions hinders the quantitation of the negative
charge through chemical analyses, as the exfoliation process creates an
abundance of OH' ions. These difficulties necessitate the use of “innocent”
cations such as alkali metals, as they have poor affinity for OH‘. This is
particularly true for the heavier alkali cations Rb+ and Cs,+ and they are
easier to quantify than Li“, Na+ or K”. Although the external OH' ions can
largely be removed by rinsing, it is difficult to be certain of their total
removal without acidification of the solution. This complicates the
situation by increasing the ionic strength of the solution and providing
protons, a competing cation which is virtually impossible to unambiguously
detect. In addition, if the MOS, is in fact slowly oxidized further in
solution over time, more OH’ ions will be generated. Even restacked MOS,

or MOS, with encapsulated hydrated alkali cations may continue to undergo

210

redox chemistry with intercalated H,O (see Eq. (5.1)). Presumably the
cations/OH' pairs would remain trapped between the layers, indicating at
least the oxidation state at the time of flocculation, but would not reflect the
continuum of oxidation states possible for the material over time.

The results of our chemical analyses, which indicate a negative
charge in the range 0.10—0.25 per Mo/W atom, are consistent with the

‘3'” and are more

results of the analysis of fresh Li+ encapsulated samples
reliable due to the use of the heavier alkali metal cations. Other studies
involving the selective encapsulation of heavy metal ions in such as Ag+ and
Pb2+ into MOS, and WS, have been conducted, and the results indicate a
higher negative charge (0.5-0.8 per Mo/W atom).28 These chalcophilic
cations, like the alkali cations, have a poor affinity for OH’; however, the
samples have been prepared under significantly different experimental
conditions, in which LiMS, was exfOliated in the presence of the cations
under anaerobic conditions in solutions of moderate ionic strength,
resulting in extremely rapid precipitation of the layers. These results

suggest that a continuum of oxidation states are possible for MOS, and WS,,

depending on the experimental conditions.

211

5.3.5 Oxidation reactions of LiMS,.

In light of increasing evidence that these materials have a negative
charge, LiMS, and exfoliated MS, were treated with oxidizing agents to see
if truly neutral MS, could be obtained by oxidizing with I, or Br,. The
oxidized products were characterized by X-ray and electron diffraction
studies and their thermal behavior.

Figure 5.8a contains the Selected Area Electron Diffraction pattern
of LiMoS, oxidized with Br, in acetonitrile (MeCN). If LiMoS, is first
exfoliated in water, then rinsed with acetonitrile, forming Li,_
n(MeCN),MoS, (n=0.75-0.85), and subsequently oxidized with Br, in
acetonitrile, similar patterns can be observed. The pattern resembles the
one found for the sample Ba0.08MoS,, in which a hexagonal array of

diffraction spots resembling a “honeycomb” is formed by the superposition

of the twinned 3a x a lattice caused by Mo-Mo distortions and a [[371 x

1/51 lattice formed by the ordering of the Ba2+ cations. The SAED pattern

from oxidation of MOS, with Br, is caused by a superposition of two types

of lattices as well; however, in this case the fit x fizz lattice is not due to

Li+ cations ordering between the layers, but different Mo-Mo associations.

Wypych and coworkers have reported that, upon treatment with I, or

212

H,Cr,O,, K0_33(H,O)yMoS, is oxidized to a material which contains a 1/3'a x

[[371 lattice attributed to Mo-Mo associations.22 This material has a smaller

a parameter than restacked MOS,. A closer inspection of Figure 5.8a
reveals that the hexagons become increasingly distorted as one moves

further away from the center spot (Figure 5.8b), indicating a mismatch in

the two lattices. The 1/321 x 1/321 lattice has a smaller a parameter than the

1/321 x a lattice. If Li,,n(MeCN)xMoS, is treated with Br, it is possible to

find electron diffraction patterns of a few crystals which contain only the

1/3-a x 1/371 lattice (Figure 5.8c), but the majority of the sample is a

decomposition product. The material with the 1/371 x 1/3_a lattice is

metastable and appears to be more susceptible to conversion to 2H-MOS,

than the material with the 3a x a lattice. If one treats LiMoS, with Br,
for a longer period of time or with sonication, the 1/3—a x l/ib lattice

disappears, leaving a mixture of 2H-MOS, and material with the 1/3-0 x a

lattice (Figure 5.9).

213

 

The pattern, which resembles those in Figure 5.6, is a combination of a
twinned V3a x a orthorhombic lattice and a 43a x ‘13a hexagonal lattice of
M082. The two lattices have slightly different cell parameters, which
results in a distortion of the honeycomb as one moves away from the
center of the pattern (B).

214

 

Figure 5.9. Oxidation Of Li1-n(H20)xM082 with Br; results in some
particles which contain only the V3a x 43a lattice (A). If Li1-n(H20)xMoSZ
or LiMoSz is oxidized over a longer period of time, or with sonication, the
‘13a x \/3a lattice disappears, leaving a mixture of 2H-M082 and M082 with
the twinned 43a x a orthorhombic lattice (B).

215

 

 

 

B a 8 a
9.‘ <2- 8
V 9
(273a x 273a)

 

 

d* (reciprocal angstroms)
Figure 5.10. SAED pattern of a minority phase formed upon oxidation of
LiW82 with Brz in MeCN which exhibits a 2‘13a x 243a lattice.

216

The electron diffraction patterns of WS, samples obtained upon

oxidation with Br, are different from those of MOS, samples. Almost all

patterns contained only the original l/3_a x a lattice, even with repeated

sonication and stirring in excess Br, for a week. After a week a small
amount of 2H-WS, could be found, and the amount of diffuse scattering
was significantly increased due to decomposition to amorphous product(s).
From these results one can conclude that either the product has not been
oxidized or restacked WS, does not exhibit an alternate superlattice upon
oxidation. Although the majority of the product exhibited no change, in
almost every sample examined there was a small minority of crystallites

which gave rise to the diffraction pattern shown in Figure 5.10a. The

pattern contains the reflections attributed to the twinned 1/3—a x a lattice,

but there are additional, extremely weak reflections which could be

indexed as a 21/51 x 2/3_a lattice (Figure 5.10b). This suggests that even

if WS, is able to have an alternate superlattice upon oxidation it is not the

same as the lattice found in oxidized MOS,.

217

5.3.6 X-ray diffraction studies of oxidized MS,.

LiMS, and Li,_n(MeCN)xMS, were also examined by transmission X-
ray diffraction (Figures 5.11, 5.12). Due to the turbostratic nature of these

samples induced by the exfoliation/flocculation process, only hkO

reflections are visible from the 1/3_a x a lattice. Interpretation of the X-ray

diffraction patterns is more difficult than the electron diffraction patterns
in part because the peaks are broad. Furthermore, some hOI reflections of

the 2H form of the dichalcogenide overlap with the unique hkO reflections

from the 1/3_a x a lattice and the Ifla x 1/3—a lattice. For example, the 200
of the 1/3_a x 1/3-0 lattice overlaps with the 103 of 2H-MS,. In fact, the

presence of the l/3_a x z/3—a lattice cannot be reliably determined by

powder X-ray diffraction because the only reflection which is not subject

to an overlap problem is the 100, which is extremely weak. The presence

of the l/fi x a lattice is more easily monitored because the 120, a

moderately strong reflection at d=2.10 A, does not overlap with peaks

from either of the other two phases which may be present in the sample.

218

 

 

 

m C
71? 0 *
a: t
c o
:3 ' o
2‘ N
‘3
9::
'9 c
(U
v z: 'k B
'7)
c o
m '-
o m
E
c
or
F
, A
llLJJJllllllllbllllllLJlIlllllllllllll

 

25 30 35 40 45 50 55 60 65
26 (degrees)

Figure 5.11. Transmission X-ray diffraction (XRD) pattern of LiMoS,
oxidized in MeCN with increasing time. (A) Initially the pattern contains

the a x 1/32 lattice with two new peaks at 2.55 A ( solid black circle) and
1.56 A (asterisk). (B) a peak at 2.3 A appears, indexed as either the 200 of
the I/3—a x 1/3—a lattice or the 103 of 2H-MOS,. (C) The 120 of the a x 1/371
lattice has disappeared. The pattern is mostly 2H-MOS,.

219

 

 

 

 

’0?

9: A

r:

: 0%.,

2‘ co

2 . 9

- o
N

8 E -

V V

a .8—

c

Q

4—0

5

M

L l I 1 L J I l l 1 4L 1 1 l

20 40 60 80 100
29 (degrees)

 

 

 

Figure 5.12. Transmission XRD pattern of LiWS, oxidized with Br, in
MeCN. The black circle marks a peak at 2.55 A which does not appear to
be an hk0 reflection of any of the three expected lattices. The asterisk
marks a contribution from the SaranTM wrap substrate.

220

The time dependent behavior of the oxidation of restacked MOS, (and

LiMoS,) is illustrated in Figure 5.11. Initially the Viz x a lattice

dominates (Figure 5.11a), with the appearance of a shoulder (d = 1.56 A)
on the 310 reflection at 20 = ~59 degrees and a new peak at 20 = 35° ((1 =
2.55 A) which is not visible in the electron diffraction patterns and is not
readily indexed to any of the 3 phases which may be present. At an

intermediate stage a new peak appears at 29 = ~38° (d = 2.35 A) which, as

mentioned previously, can be indexed as either the 200 of the V3—a x 1/3?!

lattice or the 103 of 2H-MOS2 (Figure 5.11b). The 110 and the 310

reflections begin to shift to smaller (1 spacings (d = 2.78 to 2.71 A and 1.61

to 1.59 A). This is not surprising, because the V3_a x a lattice has the

largest a parameter at 3.21 A; the 1561 x V321 lattice is slightly contracted,

as evidenced by the electron diffraction patterns; and in 2H-MOS, a = 3.16

A. The 120 peak of the V371 x a lattice at 29 = ~38° (d = 2.10 A) is still

present, and the shoulder at 20 = ~59° has become more pronounced. In

the final stage the diffraction pattern no longer contains the peak at 29 =

35° (d = 2.55 A) or the 120 peak of the V37: x a lattice at 29 = ~38° (d =

2.10 A), and the 110 and the 310 reflections have shifted to even smaller (1

221

spacings (d = 2.69 A and 1.58 A). The pattern at this stage contains mostly
MOS, which has converted to the 2H form.

The transmission X-ray diffraction patterns of restacked WS, upon
treatment with Br, (Figure 5.12) resemble those of oxidized restacked
MOS,. The peak at 29 = 35° ((1 = 2.55 A) appears but, unlike in MoS,,
does not disappear with time. Under more rigorous conditions or with

time the peak at 29 = ~38° (d = 2.35 A) appears. Given that there is no

evidence of a V371 x 1/321 lattice by electron diffraction, that peak is

probably best indexed as the 103 reflection of 2H-WS,. The disappearance

of the 120 peak of the V51 x a lattice at 29 = ~38° (d = 2.10 A) and the

new peak at 29 = 35° (d = 2.55 A) happens upon heating the material,
which converts it to 2H-WS, (Figure 5.13).

The reflection X-ray diffraction patterns contain the basal 001
reflection at 6.05 A for WS, and 6.21 A for MoS,, indicating the de-
intercalation of most or all of the lithium cations. The basal spacing of
WS, is slightly contracted (0.1 A ) relative to 2H-MS,, but expands upon
heating.

Residual negative charge on MOS, is also substantiated by these
electron diffraction patterns. The structure of restacked MOS, and WS,,
which we recently solved from electron diffraction data and found to be

W'I'e, type, consists of metal-metal associations to form zigzag chains.21

222

As mentioned previously, a material called 'lT-MOS,’ has been prepared by

the oxidation of K0,33(H,O)yMOS2 and found to have a V3_a x V3—a lattice by

X-ray diffraction and also by STM studies.”24 The starting material
K0.3,(H,O)YMOS,, however, has also been studied by STM and found to have
the same structural distortion as restacked MOS, and WS,.29 Furthermore,
it is evident by the electron diffraction studies presented here that restacked
MOS, can be at least partially converted to “IT-MOS,” upon treatment with

Br, or 1,. The exterior of the particle appears to be oxidized, forming the

V3_a x Via lattice, whereas the interior of the particle retains the original
V3_a x a structure. Actually in their STM studies of lT-MOS, the authors

conceded that occasionally they observed the fit: x a lattice found in their

starting material and concluded that it was symptomatic of incomplete

oxidation of those samples.

223

 

 

 

’0?
9:!
C
3
B B
.2
E
m C
v
.E‘ o
m N
c '-
m .
E l
- A

I11111111111111111111111111111111411111

25 30 35 4O 45 50 55 60 65
26 (degrees)

Figure 5.13. Transmission XRD pattern of LiWS, oxidized with Br, (A)
before and (B) after heating. The peak at 2.55 A (black circle) disappears,
leaving 2H-WS,.

224

5.3.7 DSC studies of oxidized M8,.

As mentioned earlier, restacked MOS, and WS, are metastable, and

can be converted to 2H—MS, by an irreversible exothermic transition upon

heating. The lT- form of MOS, with the V3_a x V3_a lattice has also been

reported to be metastable with a comparable conversion.22 Figure 5.14

contains a DSC measurement Of oxidized MoS,, which has been identified

by electron diffraction to contain both the fla x a and fl: x V3_a lattices,

and a DSC measurement of restacked MOS, for comparison. Restacked
MOS, undergoes conversion to 2H-MOS, at 98°C (at a heating rate of
5°C/min). The transition in the MOS, obtained by oxidizing LiMoS, with
Br, in acetonitrile occurs at a lower temperature, the broad peak reaching a
maximum at 90°C. Although by TEM it is evident that the material
contains more than one kind of lattice, it is not possible to resolve two
separate transitions. Li1_n(MeCN),MoS, oxidized with Br, converts rapidly,

and a DSC measurement of the material could not be obtained.

225

-0.5

 

98°C

'IIIIIIIIIIITIITTTT

     

llllllllllllll11111111111

 

 

IIIITIIII

.l

 

40 60 80 100120140160180
Temperature (°C)

Figure 5.14. Differential Scanning Calorimetry (DSC) plots of restacked
and LiMoS, oxidized with Br,. Restacked MOS, exhibits an irreversible
exothermic phase transition at 98°C, corresponding to its conversion to 2H-
MoS,. LiMoS, oxidized with Br, undergoes a similar transition at 90°C.

226

 

 

 

 

‘UTIIYI—ITIIV'I

 

0 50 100 150 200 250 300 350
Temperature (°C)

 

 

 

 

 

h
i- d
-15b1111PJAII11111L111L‘
O

100 1 50 200 250
Temperature (°C)

Figure 5.15. DSC plots showing the irreversible exothermic phase
transition to 2H-WS,. (A) LiWS, (149°C) and Li,,,,(MeCN),WS, (189°C)
oxidized with Br, in MeCN (B) Li,,,,(H,O) xWS, oxidized with Br, in H,O
(183°C) and restacked WS, (195°C).

227

The situation is different and somewhat more complicated for the
oxidation of WS, samples (Figure 5.15). Restacked WS, undergoes
conversion to 2H-WS, at 195°C (at a heating rate of 5°C/min). WS,
obtained by treating LiWS, with Br, in acetonitrile converts at 149°C,
whereas Li,,n(MeCN),WS, treated with Br, in acetonitrile converts at 189°C

(Figure 5.15a). By electron diffraction and transmission X-ray diffraction

the materials are structurally identical, containing the 1/321 x a lattice. The

oxidation was carried out in acetonitrile because aqueous exfoliated WS,
decomposes fairly rapidly upon exposure to Br,, but if the aqueous reaction
is isolated within a few minutes it is possible to see a shift in the transition

to lower temperatures (Figure 5.15b). By transmission X-ray diffraction

this material has the V371 x a lattice with a small shoulder at 2.55 A,

indicating that it has the same structure as the products of the oxidation of
Li,_n(MeCN),WS, and LiWS,.
The structural behavior of restacked WS, and LiWS, is clearly

different from that of restacked MOS, and LiMoS, upon treatment with Br,.

Despite the lack of a M321 x V371 phase in the WS, system, there is a

dramatic shift in conversion temperature, from 195°C to 149°C, which is
reminiscent of the shift in the transition temperature observed in the MOS,

system. This shift suggests that oxidation has in fact taken place.

228

Furthermore, reflection X-ray diffraction indicates no residual cations

between the layers (i.e., the d spacing is 6.1 A). The fact that both

restacked and oxidized WS, have the V3_a x a lattice implies that, at least in

WS,, the shift in conversion temperature is not dependent on a structural
rearrangement, but rather the degree to which the sample is oxidized. The
discrepancy in transition temperature between Br, treated LiWS, and Br,
treated Li,,n(MeCN),WS, is probably due to limited diffusion in Lil,
n(MeCN),,WS,. In LiWS, the acetonitrile can solvate the lithium cations,
swelling the layers apart and allowing the oxidant to access the layers more
intimately, whereas Li,,n(MeCN),WS, does not swell in acetonitrile; hence
the interior of the particles is inaccessible to the oxidizing agent and the
material remains trapped in the incompletely oxidized state that it achieved
upon exfoliation in water.

Studies of the structure dependence upon oxidation state in MOS,
(and WS,) are complicated in part because 2H—MOS, undergoes a phase
transition upon lithiation in which the coordination environment changes
from trigonal prismatic to octahedral. If only a small amount of lithium is
introduced the material can remain in the 2H-MOS, structure.30 It has also
been observed that the material does not convert to the octahedral structure
homogeneously with small amounts of lithium, resulting in a two phase

system.18a Electrochemical studies which start with 2H-MOS, are difficult

229

to interpret due to this problem.“ The V3_a x a octahedral phase can be

used as a starting point, however. Electrochemical studies of Li,MoS,,
K,(H,O)yMoS,, and K,(H,O)yWS, which start from the octahedral structure
indicate the presence of 46 different phases in the range x = 0 to x = l.'8b'
32 Although some authors have attributed these many phases to cation
ordering in the gallery“, others have attributed them to charge density
waves.32 The different observed superlattices in MOS, suggest that charge

density waves are responsible for at least some of these phases. Wypych

and coworkers have proposed a scheme containing three different lattices: a

2a x 2a for KO.,MOS,, V32 x a for K,,,(H,O),MOS,, and a V321 x V3—a for

lT-MOS,.24 A comparison of the electrochemical behavior of
Kx(I-I,O)yMoS, and K,(H,O),WS, reveals that the two materials behave
differently at low x values”, consistent with the different electron

diffraction results observed in the reaction of LiMoS, and LiWS, with Br,.

230

5.3.8 Acid restacked M8,.

When LiMS, is exfoliated in concentrated acid the product should
form a proton bronze, H,MS,. The transmission X—ray diffraction patterns

for the reaction of LiMoS, and LiWS, with concentrated HCl (Figure 5.16)

show that the materials retain the Viz x a lattice and a peak appears at

2.55 A. The products were examined by electron diffraction and exhibited

only the V361 x a lattice with no evidence of an additional reflection at 2.55

A. In this case, the structure of HxMos, is the same as that of PLWS,,

exhibiting no evidence of a Viz x V32 lattice. The reflection X-ray

diffraction patterns exhibit basal 001 spacings of 6.05 A.

If LiWS, is exfoliated directly in concentrated HCl the DSC shows an
exothermic transition at 155°C, whereas if it is first exfoliated in H,O and
then concentrated acid is added it exhibits a transition temperature of
191°C (Figure 5.17). Presumably the x values in the products of each
reaction, H,WS,, are different. The discrepancy in the behavior of LiWS,
and Li,,n(H,O),WS, in acidic conditions is similar to the discrepancy upon

oxidation with Br,, in which the difference in conversion temperatures

231

 

110

A

(D

.".'.'

g 0

5 o 9
E S! m
e

(U

V

.2; ' A
(D

C

Q)

fill

E

 

 

tllllrllerlLUIermrlnllllLtllllrlr

25 30 35 4O 45 50 55 60 65
26(degrees)

 

Figure 5.16. Transmission XRD pattern of (A) LiWS, and (B) I-LMOS,
Obtained from LiMoS, exfoliated in concentrated HCl. Black circle marks

peak at 2.55 A.

232

 

 

I Jillllllllllllll[1111;111111141

 

 

.—
h
-1 bIIILlllIJJIIIIJI11111111]IIII-

0 50 100 150 200 250 300
Temperature (°C)

 

Figure 5.17. DSC plots of HXWS, obtained from the reaction of HCl with
(A) LiWS2 (155°C) and Li,,n(H,O),WS, (191°C).

233

could be attributed to a diffusion problem. Exfoliation in concentrated
acid provides more oxidizing conditions than exfoliation in H,O before
addition of acid. The amount of oxidation during the exfoliation process is
important because there is a competition between oxidation and proton
encapsulation. Once the material has flocculated its oxidation is limited by
diffusion; hence the discrepancy in transition temperatures.

LiMoS, exfoliated in concentrated acid does not exhibit a shift in its
exothermic transition to a lower temperature. The structure of I-LMOS,
samples prepared from the reaction of LiMoS, and Li,_n(H,O),MoS, with

acid resembles that of their W analogues more than in Br, solution. There

is no evidence of the [[321 x V3—a phase in acidic conditions. This suggests

that the Russian scientists’ hypothesis that MOS, is fully oxidized upon
treatment with acid is not correct; only that protons are now the
encapsulated counter-ion.l7

As mentioned previously, the authors who first reported the
exfoliation of MOS, believed that the negative charge of the layers was due
to OH' groups associated with the basal planes of the layers. It was
believed that exfoliated MOS, (and WS,) has a point of zero charge (PZC)
upon acidification. The authors used the different PZC to create a
restacked composite material with alternating M08, and WS, layers,

forming a superstructure along the 001 axis.16c Their evidence for a

234

superstructure of alternating MOS, and WS, layers was poor, however; a
better interpretation of these results is that the introduction of protons to a
homogeneous mixture of exfoliated layers resulted in rapid flocculation,
creating a homogeneous mixture of MOS, and WS, in the product. They
also developed the concept that the PZC occurs at slightly different pH for
the edge OH' sites and basal OH‘ sites, and used this to create a House-Of-
Cards structure for MOS,.16b Their evidence for this kind of structure was
a decrease in the intensity of the 001 reflections in the X-ray diffraction
pattern. However, the intensity of the 001 reflections, which is indicative
of the degree of order in the flocculated layers, can be influenced by a
variety of factors (the ionic strength of the solution, for example), and does
not constitute proof of a point of zero charge on the layers.

The mysterious peak at 2.55 A in the transmission X-ray diffraction
patterns of HXMS2 and LiMS, oxidized with Br, is puzzling, but shifting of
the layers relative to one another could allow hOI reflections to appear. A
102 reflection has been observed in some samples of 2H-MOS, at that d
spacing. The peak at 2.3 A is almost certainly the 103 of 2H-MS,.

An illustration of the different superlattices observed in MOS, and
WS, due to M-M associations can be found in Figure 5.18. An ideal 1T-

MS, lattice (not experimentally observed) can distort to form zig—zag

235

 

Figure 5.18. Schematic illustrating (A) an ideal lT-MS2 lattice, (B) a
V371 x a lattice with infinite zig-zag metal chains, (C) a V3_a x V3_a lattice

with trimers, and (D) a 215a x 2 3a lattice with possible tetramers
formed due to M—M distortions.

236

chains, giving rise to the orthorhombic V51 x a lattice found in restacked

MOS, and WS,. It can trimerize, giving rise to a V3_a x V371 lattice found

in MOS, upon oxidation with Br, or I, (Figure 5.180). Finally, when WS,

is treated with Br,, a minority phase with a 2 V3_a x 2 V3—a lattice is formed.

This distortion could be due to a number of M-M associations, perhaps a

trimerization similar to Figure 5.18c or a tetramerization (Figure 5.18d).

5.3.9 Thermopower measurements.

As mentioned previously, the reduction of 2H-MoS, with lithium
results in a structural transformation from trigonal prismatic to octahedral
metal coordination and a change in properties from semiconducting to
metallic. These changes have been explained using. a depiction of the band
structures for LiMoS, in both coordination environments, which revealed
that there was a net stabilization in energy upon conversion to octahedral
that resulted in a half-filled band with predominantly d orbital character.‘8

Restacked MOS, and WS, retain octahedral coordination about the
metal atom,21 and their conductivity has been reported elsewhere” to be
metallic. If one simply removes electrons from the half filled band which

was published for LiMoS,, an idealized octahedral d3 system, one should

237

expect an n-type conductor for the d2 system (Figure 5.19). Thermopower
measurements on restacked MOS, and WS,, however, indicate that they are
p-type (hole) conductors (Figure 5.20), with room temperature values of
50 and 20 ttV/K, respectively. The structural distortion in restacked MOS,
and WS, to form zig-zag chains obviously should result in a change in the
band structure. In fact, distortion of octahedral ML, (12 systems to form
zig-zag chains has been predicted by Extended Huckel Tight Binding
calculations” and, for a fully oxidized (12 system, results in a half filled
band (Figure 5.19). If, however, the system is actually d”, as is the case
with restacked MOS, and WS,, residual negative charge results in a material
with p-type conductivity, consistent with our experimental observations

(Figure 5.19).

238

 

 

 

dz2

 

 

 

zig-zag

 

_ b
chains

   

Figure 5.19. Schematic band diagram illustrating the distortion of an
ideal octahedral system to form zigzag chains. The ideal system describes
an n-type conductor whether it is d2 (black fill) or d“ (black fill + stripes).
The distorted system is half filled for d2 and more than half filled for d“,

resulting in a p-type metallic conductor.

239

 

60

11111
P

50

IIIIII

40

IIIIIIIIITTI
D
P

—L
o
‘>
>
O
O
O
O

 

 

O
WTI.IIIII
b

lllllllllllllllllllillll11_LIJ
80 120 160 200 240 280 320 360 400
Temperature (°C)

 

Figure 5.20. Thermopower measurements of restacked MOS, (triangles)
and WS, (circles). The relatively small magnitude and slope of the data is
consistent with that of a p-type metallic conductor.

240

5.4 Conclusions

MOS, and WS, retain some negative charge upon exfoliation and
flocculation. Although it is possible to incorporate neutral species, cations
can also be encapsulated without a detectable co—encapsulated anion, usually
giving rise to very ordered reflectance X-ray diffraction patterns.9"3'l7 In
the studies presented here hard, electropositive alkali cations have been
encapsulated in MOS, and WS, under neutral synthetic conditions. The
choice of these cations and these synthetic conditions minimizes the risk of
co-encapsulation of OH’ ions. The cations order in the gallery, forming
additional superlattices. The range of negative charge on the layers
appears to be 015-025 for both MOS, and WS,, which is consistent with
measurements on samples prepared in a similar fashion by other
researchers.17

Structural characterization of restacked MOS, also supports a
negative charge. Restacked MOS, and WS, have been found by electron

diffraction to have the same superlattice as K,(H,O),MOS,.2'2° Treatment

of restacked MOS, with Br, results in a change from a V371 x a superlattice

to a 1/371 x V321 superstructure, which is analogous to the superstructure

observed in “1T-MOS,”, the oxidation product of K(,.,MOS,.22 This change

in structure upon exposure to an oxidizing agent implies that restacked

241

MOS, is incompletely oxidized, and is probably better formulated as
LixMoS, or Lix(H,O)yMoS,.

Restacked WS, does not appear to exhibit a structural change upon
exposure to Br,. However, both restacked MOS, and WS2 exhibit
irreversible exothermic transitions which are shifted to lower transition
temperatures upon treatment with Br,. The shift is quite significant in
WS,, from 195°C to 149°C in the Br,-treated sample, indicating that
oxidation has taken place. It is somewhat surprising, however, that

oxidation of restacked WS, does not result in an alternate superlattice

analogous to the Viz x V3_a superstructure observed in oxidized MOS,.

The products of all thermal conversions are the semiconducting phase 2H-
MS,, which requires an expulsion of the residual negative charge and
explains why the oxidation state affects the transition temperature. The
expulsion of the residual negative charge likely results in the reduction of

residual H,O (or H+ in H,MS,), forming H, gasf"4

242

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A 2H-MoSz (1)]
H20 + * + HCI
+ Br2/MeCN
l Lio.2M032 (2) i HxMOSZ (2)
(130 X01 \ (l/3a Xa)
+ Br2/ _
M eCN LIO.2M082 (2) N3a Xa)
‘A “1T-MOSQ” (3) (1/3a x V3a)
A 98°C A¢9OOC A
V V
l2H-MoSz (1)| |2H-Mosa (1)| I2H-M082 (1)]
B le-ws2 (1 )l
H20 + + HCI
+ Br2/MBCN
I UO.2W32 (2)|~ \ . Br2/ 'ws2“ 13)
(«30 X0) MBCK (‘30 X0) 7
I \ “W82“ (3) (‘30 Xa)
Lio.2W82 (2) minority (2V3ax2\/3a)
+ "we; (3)
A 195°C A *14goc<x<19500 A 149°C Ajlss°c
I2H-W82 (1)] [ 2H-W82 (1 )| I2H-W82 (1)] I2H-WSz (1)|

 

 

 

 

 

Figure 5.2 l . Reaction scheme illustrating the different phases obtained
with various treatment of (A)MOS, and (B) WS,.

243

The different phases of MOS, produced under various experimental
conditions are illustrated in Figure 5.21a. This scheme is intended to
correlate all forms of MOS, and the inter-conversion between forms, and to
help clear up any remaining confusion regarding the nature of this system.
Exfoliation and flocculation of LiMoS, in water results in restacked MoS,,
which can be more accurately formulated as L102M082 (2) based on the
chemical analyses of the alkali cation encapsulated products. Oxidation of
LiMoS, in Br,/MeCN results in a mixture of “IT-MOS,” (3) and Li0.,MoS,.
LiMoS, exfoliated in H,O can be converted to the same mixture upon
treatment with Br,/MeCN. If LiMoS, is exfoliated in concentrated HCl it
forms H,MOS,, a material with the same superstructure as Lio_,MoS, which
does not appear to be oxidized. A similar scheme in Figure 5.21b
illustrates the analogous phases of WS,. A principal difference is that the
product of the oxidation of LiWS, with Br2 in MeCN, called “WS,” (3), is
not structurally different from Li0_,WS,. Depending on the preparation, it
is possible to isolate products exhibiting a continuum of transition
temperatures between 149°C and the transition temperature of Li0.,WS,
(195°C). Another difference is that, unlike in MoS,, when LiWS, reacts
with concentrated HCl, the transition temperature of the product (155°C) is
closer to that of the oxidized material (3) rather than Lio_,WS,. Given the
higher energy of the d orbitals of W, it is not surprising that WS, is more

readily oxidized than MOS, under comparable experimental conditions.

244

Clearly the MOS, and WS, systems are rather complicated, and the
nature of the phase Obtained is very sensitive to the experimental conditions
used to prepare the product. The term “restacked MOS,” has been used

casually in the literature to represent material which has been converted to

the 2H form as well as the metastable compound with the V371 x a lattice,

and is usually written without a formulation of the negative charge present.
In addition, MOS, and WS, resemble one another, but there are distinct
differences between the two systems. To prevent future confusion
researchers should take special care to formulate their products precisely,
and also to recognize that restacked MOS, (and WS,) which has been heated
above its conversion temperature is better described by the formula 2H-

M08, (and 2H-WS,).

245

10.

References

Brenner, J.; Marshall, C.L.; Ellis, L.; Tomczyk, N.; Heising, J.;
Kanatzidis, M.G. Chem. Mater., 1998, 10, 1244.

Whittingham, M.S. Prog. Solid State Chem, 1978, 12, 41.

. a) Dines, M.B. Mater. Res. Bull., 1975, 10, 287. b) Yang, D.; Frindt,

R.F. J. Phys. Chem. Solids, 1996, 57, 1113.

a) Tsai, H.L.; Heising, J .; Schindler, J.L.; Kannewurf, C.R.; Kanatzidis,
M.G. Chem. Mater., 1997, 9, 879. b) Wang, L.; Kanatzidis, M.G.
manuscript in preparation.

Joensen, P.; Frindt, R.F.; Morrison. S.R. Mater. Res. Bull., 1986, 21,
457.

a) Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. Science, 1989,
246, 369. b) Zhou, X.; Yang, D.; Frindt, R.F. J. Phys. Chem. Solids,
1996, 57, 1137. c) Kosidowski, L.; Powell, A.V. Chem. Commun.,
1998, 2201.

a) Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. J. Mater. Res.
1991, 6, 1103. b) Kanatzidis, M.G.; Bissessur, R.; DeGroot, D.C.;
Schindler, J.L.; Kannewurf, C.R. Chem. Mater., 1993, 5, 595. c)
Wang, L.; Schindler, J.L.; Thomas, J.A.; Kannewurf, C.R.; Kanatzidis,
M.G. Chem. Mater., 1995, 7, 1753. d) Ruiz-Hitzky, E.; Jimenez, R.;
Casal, B.; Manriquez, V.; Santa Ana, A.; Gonzalez, G. Adv. Mater.,
1993, 5, 738. e) Lemmon, J.P.; Lerner, M.M. Chem. Mater., 1994, 6,
207.

Bissessur, R.; Heising, J.; Hirpo, W.; Kanatzidis, M.G. Chem. Mater.,
1996, 8, 318.

a) Heising, J.; Bonhomme F.; Kanatzidis, M.G. J. Solid State Chem.,
1998, 139, 22. b) Ollivier, P.J.; Kovtyukhova, N.I.; Keller, S.W.;
Mallouk, T.E. Chem. Commun., 1998, 1563.

Tagaya, H.; Hashimoto, T.; Karasu, M.; Izumi, T.; Chiba, K.
Chemistry Lett., 1991, 2113.

246

ll.

12.

13.

14..

15.

16.

17.

18.

19.

20.

Nakagaki, S.; Mangrich, A.S.; Wypych, R. Inorganica Chimica Acta,
1997, 254, 213.

a) Dungey, K.E.; Curtis, M.D.; Penner-Hahn, J.B. J. Catalysis, 1998,
175, 129 . b) Gee, M.A.; Frindt, R.F.; Morrison, S.R. Mater. Res.
Bull., 1986, 21, 543.

a) Golub, A.S.; Shumilova, I.B.; Zubavichus, Y.V.; Jahncke, M.; Siiss-
Fink, G.; Danot, M.; Novikov, Y.N. J. Mater. Chem, 1997, 7, 163. b)
Danot, M.; Mansot, J.L.; Golub, A.S.; Protzenko, G.A.; Fabritchnyi,
P.B.; Novikov, Y.N.; Rouxel, J. Mater. Res. Bull., 1994, 29, 833. c)
Golub, A.S.; Payen, C.; Protzenko, G.A.; Novikov, Y.N.; Danot, M.
Solid State Commun., 1997, 102, 419.

Divigalpitiya, W.M.R.; Frindt, R.F.; Morrison, S.R. Appl. Surf. Sci.,
1991, 48-49, 572.

Divigalpitiya, W.M.R.; Morrison, S.R.; Frindt, R.F. Thin Solid Films,
1990, 186, 177.

a) Miremadi, B.K.; Morrison, S.R. J. Appl. Phys., 1988, 63, 4970. b)
Miremadi, B.K.; Cowan, T.; Morrison, S.R. J. Appl. Phys., 1991, 69,
6373. c) Miremadi, B.K.; Morrison, S.R. J. Appl. Phys., 1990, 67 ,
1515.

a) Golub, A.S.; Protsenko, G.A.; Gumileva, L.V.; Buyanovskaya,
A.G.; Novikov, Y.N. Russ. Chem. Bull., 1993, 42, 632. b) Golub,
A.S.; Shumilova, I.B.; Novikov, Y.N.; Mansot, J.L.; Danot, M. Solid
State Ionics, 1996, 91, 307.

a) Py, M.A.; Haering, R.R. Can. J. Phys., 1983, 61, 76,. b) Mulhem,
P.J. Can. J. Phys., 1989, 67, 1049.

Bissessur, R.; Kanatzidis, M.G.; Schindler, J.L.; Kannewurf, C.R. J.
Chem. Soc., Chem. Commun., 1993,1582.

a) Yang, D.; Sandoval, S.I.; Divigalpitiya, W.M.R.; Irwin, J.C.; Frindt,
R.F. Phys. Rev. B., 1991, 43, 12053. b) Joensen, P.; Crozier, E.D.;
Alberding, N.; Frindt, R. F. J. Phys. Solid State Phys., 1987, 20, 4043.
c) Qin, X.R.; Yang, D.; Frindt, R.F.; Irwin, J.C. Ultramicroscopy,
1992, 42-44, 630. d) Qin, X.R.; Yang, D.; Frindt, R.F.; Irwin, J.C.

247

21.

22.

23.

24.

25.
26.

27.

28.

29.

30.
31.

32.

Phys. Rev. B., 1991, 44, 3490. e) Sandoval, S.J.; Yang, D.; Frindt,
R.F.; Irwin, J.C. Phys. Rev. B., 1991, 44, 3955-3962.

Heising, J.; Kanatzidis, M.G. J. Am. Chem. Soc., 1999, 121, 638.

a) Wypych, F.; SchOllhom, R. J. Chem. Soc., Chem. Commun., 1992,
1386. b) SchOllhom, R.; Kl'impers, M.; Plorin, D. J. Less-Common.
Metals, 1978, 58, 55.

a) Yang, D.; Frindt, R.F. Mol. Cryst. Liq. Cryst., 1994, 244, 355. b)
Bissessur, R.; Kanatzidis, M.G.; Schindler, J.L.; Kannewurf, C.R. J.
Chem. Soc., Chem. Commun., 1993, 20, 1582.

Wypych, F.; Weber, Th.; Prins, R. Chem. Mater., 1998, 10, 723.
Heising, J .; Kanatzidis, M.G. unpublished results.

a) Brown, B.E. Acta Crystallogr., 1966, 20, 268. b) Mar, A.; Jobic,
S.; Ibers, J.A. J. Am. Chem. Soc., 1992, 114, 8963.

Twinning of the host-lattice has been included in the simulation of the
diffraction patterns in Figures 5.5b and 5.7b to give rise to patterns
which better resemble the experimental data. For clarity, the M-M
(M=Mo, W) zig-zag chains of these other cells, which would be rotated
120 degrees relative to one another, have been omitted from Figures
5.5a and 5.7a.

Gash, A.E.; Spain, A.L.; Dysleski, L.M.; Flashenreim, C.J.; Kalaveshi,
A.; Dorhout, P.K.; Strauss, S.H. Environ. Sci. Techn., 1998, 32,
1007.

Wypych, F.; Weber, Th.; Prins, R. Surf. Sci. Lett., 1997, 380, L474.
Papageorgopoulos, C.A.; Jaegermann, W. Surf. Sci., 1995, 338, 83.
Julien, C. “Lithium Insertion in Chalcogenides: Electrochemical
Properties and Electrode Applications.” Microionics: Solid State
Integrable Batteries. M. Balkanski (ed.) Elsevier Science, New York,
1991, 309-341.

Wypych, F.; Sollmann, K.; Schtillhom, R. Mater. Res. Bull., 1992,
27, 545.

248

33. a) Rovira, C.; Whangbo, M.-H. Inorganic Chem, 1993, 32, 4094. b)
Whangbo, M.-H.; Canadell, E. J. Am. Chem. Soc., 1992, 114, 9587.

34. Proton NMR studies of the gaseous species released upon conversion of

restacked MOS, to 2H-MOS, indicated the presence of hydrogen gas. J.
Heising, M.G. Kanatzidis, unpublished results.

249