,,
l‘w
K;

‘ E

l
5*
5
J
i:
B

‘ i ' ;
- 131213-53
" figi‘fiasggz *

t n

 

25.3.?“

 

2f.-

100?

This is to certify that the
dissertation entitled

Local Structure of lntercalants and Host Nanoporous Materials

presented by
Mouath Ghazi Shatnawi

has been accepted towards fulfillment
of the requirements for the

Doctoral degree in Physics

._ f /
Majo rofessor’s Signature /

5/2/07-

Date

 

 

MSU is an afflnnative-action, equal-opportunity employer

.- “.._-_——-r

 

 

LIBRARY
lt/lachiqan State
Uflltftil'Slty

 

 

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/CIRC/Date0ue.indd-p.1

 

Local Structure of lntercalants and Host N anoporous Materials
By

Mouath Ghazi Shatnawi

A DISSERTATION

' Submitted to
Michigan State University
in partial fulfillment of the requirements
for the Degree of

DOCTOR OF PHILOSOPHY
Department of Physics and Astronomy

2007

ABSTRACT

Local Structure of lntercalants and Host Nanoporous Materials
By

Mouath Ghazi Shatnawi

The Atomic pair distribution function (PDF) method was used in this Ph.D. thesis
to study the structure of intercalant and host Nanoporous materials. Because they
are not crystalline and involve aperiodic intercalated species considering only Bragg
reflections, as done in traditional crystallographic approaches, is not sufficient and
the diffuse scattering will play a major rule in studying their structure. PDF has
the advantage of considering both the Bragg and diffuse scattering, which enables
the study of the average as well as the local structure of materials. All of the data
for this research were collected using the recently developed rapid acquisition PDF
(RA-PDF) technique, which facilitated the collection of the data in greatly reduced
time.

Nanoporous materials have high surface to volume ratio with high surface area
and large porosity which enable them to discriminate and interact with molecules and
clusters. They have wide applications such as environmental separation, clean energy
production, catalysis and biological applications.

In this research we have studied three different examples of important nanoporous
materials that includes silica gel loaded with alkali metals, which are promising ma-
terials for hydrogen production and chemical reduction, mesostructured silica loaded
with mercury, which adsorbs mercury as a “sponge” and boehmite to gamma alumina

transformation, where the gamma alumina is a well known and important catalyst.

To my parents, wife, daughters, brother, cousinry, friends, and all

those who once helped me To my family and friends

iii

Acknowledgments

Reaching the end of my thesis research, I would like to thank all of the people
who participated in making this thesis possible.

First of all, I would like to express my sincere appreciation and gratitude to my
supervisor Professor Simon Billinge for his patient guidance, encouragement and ex-
cellent advice throughout this thesis work. It was a great pleasure to me to conduct
this thesis under his supervision. Through out of my research he inspired and mo-
tivated me with his great knowledge and excellent skills for analyzing and reaching
conclusions from data. I am grateful for the ideal working environment he provided
us in his group with a lot of encouragements and continuous follow ups at all different
stages of our research.

I would like to thank our collaborators: Professor Thomas. J. Pinnavaia and Pro-
fessor J. L. Dye from the chemistry department at Michigan state university. Whom,
with their great knowledge and experience, not only provided us with the very valu-
able samples that have been studied in this research, but also with useful discussions
and results from different structure studies methods that helped in supporting and
completing the results obtained from the PDF method.

My great appreciation to all of my thesis committee members, Norman Birge,
Hendrik Schatz, S. D. Mahnti, Thomas J. Pinnavaia and J. L. Dye. Who gener-
ously, gave me from their valuable time and accepted being members for my thesis
committee.

I would like to thank Emil Bozin, Pavol Juhas, Gianluca Paglia, Ahmad Masadeh,
Hyunjeong Kim,Richard James Worhatch, Moneeb Taiseer Shatnawi, He Lin and
Xiangyun Qiu, current former members of Billinge group. Who were available each
time I needed help. At the start of my research work they were here to provide me
with everything needed to start learning the PDF method and after that a great help

in useful discussions and sharing knowledge.

iv

I would like to thank Debbie Simmons and Cathy Cords for providing everything
possible to make our study here proceed easily. I also acknowledge help from Di-
dier Werrneille, Doug Robinson while performing the RA-PDF measurements at the
MUCAT station in the APS.

My final, and most heartfelt, acknowledgment goes to all of my family. Particu-
larly, like to thank my parents for their patience with me and their daily encourage-
ments and for always being there when I needed them . My sister and my brothers
Malik, Juber, who gave me a great support and encouragements and took care of all
my different issues at my home country Jordan, to keep all of my time free for my
study here. My great appreciation for my wife who worked diligently and successfully
to help me to overcome all kinds of difficulty I faced. For more than five years here
patience, understanding and encouragement during my study time gave me a great
support to proceed with my graduate study. My dearest daughters, Maise, Ramah
and Yara, who were my pleasure and happiness that helped me in forgetting all of
the work stress I faced during my study.

The work at MSU benefited from support from National Science Foundation
through NIRT grant DMR-0304391. X—ray data were collected at the 6IDD beam-line
Advanced Photon Source (APS). Use of the APS is supported by the US. DOE under
Contract No. W-31-109-Eng-38. The MUCAT sector at the APS is supported by the
US. DOE under Contract No. W-7405-Eng—82.

Contents

1 Introduction 1

1.1

2.5

2.6

Nanoporous Materials .......................... 2

1.1.1 Alkali metals and silica gel: Candidate source for hydrogen

production ............................. 3

1.1.2 Functionalized mesostructued silica: Heavy metals adsorbs . . 4

1.1.3 Gamma-Alumina: Most widely used catalyst .......... 4

1.2 Thesis layout ............................... 5

2 Atomic Pair Distribution Function Technique 6
2.1 Introduction ................................ 6

2.2 Local and average structure ....................... 7
2.3 Definition of the pair distribution function ............... 8
2.4 PDF from experiment .......................... 10

2.4.1 Rapid-acquisition pair distribution function (RA-PDF) technique 11

2.4.2 Getting the PDF ......................... 12
Experimental errors in the PDF ..................... 13
2.5.1 Termination errors ........................ 14
2.5.2 Statistical errors ......................... 15
2.5.3 Instrument resolution ....................... 16
Extracting information from the PDF .................. 17
2.6.1 Peak position ........................... 17

vi

2.7

2.6.2 Peak width ............................ 18
2.6.3 Peak intensity ........................... 18
2.6.4 Real Space refinement ...................... 19

Other methods to probe the local structure: Nuclear magnetic resonance 22

2.7.1 Solid state NMR ......................... 23

Structure study of novel alkali metal intercalated silica gel with po-

tential as hydrogen source 25
3.1 Introduction ................................ 25
3.2 Alkali metals and silica gel: New novel materials ............ 25
3.2.1 Silica gel .............................. 26
3.2.2 Sample preparation ........................ 27
3.3 Structure of alkali metals doped into silica gel ............. 28
3.3.1 The PDF experiment ....................... 28
3.3.2 The PDF results ......................... 29
3.4 Other measurements ........................... 34
3.4.1 The NMR results ......................... 34
3.4.2 Differential scanning calorimetry (DSC) ............ 41
3.4.3 Magnetic susceptibilities and electrical conductivity ...... 41
3.5 Discussion ...... A ........................... 42

Mercury Binding Sites in Thiol—Functionalized Mesostructured Sil-

ica
4.1
4.2

4.3

45
Introduction ................................ 45
Sample preparation ............................ 48
4.2.1 Mercury Adsorption ....................... 49
4.2.2 PDF experiments ......................... 49
Results and discussion .......................... 50

vii

5 In-situ study of the boehmite to gamma-alumina phase transition

using the atomic pair distribution function technique
5.1 Introduction ................................
5.2 Experimental methods ..........................

5.3 Results ...................................

5.4 Conclusion .................................

6 Concluding remarks
6.1 Future work ................................
6.2 Nanoporous carbon ............................

6.3 Trapping lead by mesostructured silica .................

A Parametric studies using the RA-PDF technique
A.1 Low temperature RA—PDF experiment .................
A.1.1 Closed cycle refrigerator (CCR) cooling .............
A.1.2 Simplified instruction for using the displex ...........
A.2 High temperature RA-PDF experiment .................

A.2.1 Simplified instruction For using the MUCAT station furnace .

Bibliography

viii

61
61
64
66
78

79
81
81
82

85
85
86
91
94
98

101

List of Tables

3.1
3.2

4.1

4.2

4.3

5.1

5.2

Refined dual-phase structure of the N a—SC-II material. ........
23Na MAS N MR data for Stage I Metal-Silica Gel samples ......

Mercury Content of (SiOg)1_z(LSiOl,5)I Mesostructures for PDF Anal-
ysis, L = Mercaptopropyl ........................
Lattice parameters and fractional coordinates for selected atomic po-
sitions reported in the literature and the refined values obtained from
the experimental PDF profiles for HgO and HgS ............
Atom Pair Distances for the HgS [1] and HgO [2] from the Models

Reported in the Literature ........................

The pr values and the fraction of ’Y-A1203, f7, from the refinement
of the average structure both from Rietveld and PDFfit2 refinements.
The models in each case are single-phase boehmite, single-phase c-
symmetry and a 2—phase mixture of these two models. ........

Refinement result of the local structure .................

ix

37

55

58

75

List of Figures

2.1

2.2

2.3
2.4

2.5

2.6

(a) random distribution of vacancies and (b) preferred ordering in a:-
and y- direction. Panel (c) shows the corresponding powder diffraction
patterns. The inset shows the background enlarged by a factor of 100.
Panel (d) shows the corresponding PDFs for the random distribution
(filled circles) and short range ordered distribution (solid line). The
difference is Shown offset below. Adapted from [3] ...........
Fhllerene (C60 single molecule structure) that forms fcc lattice in the
solid C60. Alse shown the neutron powder diffraction data and the
corresponding PDF. The sharp peaks in the low-7‘ yields from the C-C
correlations that extends up to 7.1 A , (dashed line) the diameter of
the of the Bucky ball. The broad oscillation that exists above 7.1 A
represent the inter-ball correlations ...................
RA-PDF experiment setup. Adapted from [4] .............
The calculated PDF of silicon. Solid line corresponds to termination
at Qmaz = 30 A”1 and the dotted line for the same data terminated
at (2mmc =10 A"1. ............................
PDF of ZnSeo,5Te0,5. Solid line corresponds to termination at Qmax =
40 A-1 and the dotted line for the same data terminated at 62mm =
17 A '1. Adapted from [3] ........................

Calculating the pair distribution function. ...............

10
12

15

16

3.1

3.2

3.3

3.4

3.5

3.6

3.7

3.8

Room-temperature absorption of NagK alloy into. (A) The SC coated
with alkali metal alloy just after mixing. (B) The final loose black
powder (stage 0) formed by shaking the sample shown in A for a few
minutes. Adapted from [5] ........................
Experimental powder diffraction patterns for (a) pure silica, (b) NagK-
SG—I, (c) Na—SG-I, and (d) Na-SG-II. The pure silica data were scaled
to facilitate easier visual comparison. ..................
Experimental reduced structure functions for (a) pure silica, (b) NagK-
SG-I, (c) Na—SG-I and (d) Na-SG-II ...................
Experimental PDFs, scaled with respect to the Si—O peak amplitude for
(a) pure silica, (b) NagK-SG-I, (c) NarSG-I, (d)Na—SG-II, (e) Na—SG-II
on different length scale .........................
Experimental reduced structure functions for NagK-SG-I (blue sym-
bols) and the calculated PDF from metallic bcc sodium ........
Experimental reduced structure functions for Na—SG-I (blue symbols,
and the calculated PDF from metallic bcc sodium ...........
Comparison between experimental (circles) and calculated (solid lines)
PDFs for the Na—SG-II sample, using (a) a single phase Na28i03 model
and (b) (offset) a dual-phase model containing NagSiOg and NaSi. The
residual plots are shown displaced below the fits overlaid on each other.
Magenta for the single phase residual and gray for the dual phase. . .
23Na MAS NMR spectra of N agK-SG-I in silica gel with an average pore
size of 150 A at various metal loadings. The weight percent loadings
are from the amounts of metal and SC used. The spectra are scaled

and displaced for clarity. .........................

xi

27

29

30

32

33

35

38

3.9

3.10

4.1
4.2

4.3

4.4
4.5

23Na MAS NMR spectra of Na2K-SG-I in silica gel as a function of
average pore size. Loadings are 40, 35, and 25 wt % metal for 150, 60
and 30 A SG, respectively. The spectra are scaled and displaced for
clarity. ...................................
Ambient temperature 23Na MAS N MR spectra of Na—SG-II with av-
erage pore size of 150 A. The two peaks on the left are attributed to
Na4Si4, while that on the right is from the product sodium silicate and
initial Na+ formed by reaction of Na with defects. This sample has no

peak at the chemical shift of Nao. ....................

Cinnabar (HgS) structure (yellow = S, gray = Hg). ..........
(a) Experimental F(Q) for the mercury-loaded compositions of Ta-
ble. 4.1. Included for comparison are analogous data for bulk silica
glass, labeled SiOg, which does not contain mercury. The curves are
offset for clarity ...............................
(a) Experimental C(r) for the mercury-loaded compositions of Ta-
ble. 4.1. Included for comparison are analogous data for bulk silica
glass, labeled 8102, which does not contain mercury. The curves are
offset for clarity ...............................
Rescaled PDF profiles for the Hg—loaded samples ............
Experimental PDF (blue circles) and the PDF profiles calculated from
the refined crystal structure (red line) for (a) HgO and (b) HgS refer—
ence compounds. The difference curves are represented by the offset

green lines. ................................

xii

39

40

47

51

52
54

4.6

4.7
4.8

5.1

5.2

5.3

5.4

5.5

Calculated PDF profiles (red lines) assuming that (a) 0%, (b) 10%, and
(c) 50% of the mercury centers in MP—HMS sample 3 (a: = 0.50; Hg/ S =
1.30) participated in Hg-O bond formation and the remaining mercury
atoms bind to sulfur. The experimentally observed PDF profile for this
mesostructure is shown by the blue circles. The difference curves are
shown underneath in green. The Hg-O bond, if present, is expected to
give rise to a peak at 2.02. The peaks at 1.6 and 2.36 are Si-O and
Hg—S pairs, respectively. .........................
Suggested structure for at low mercury loadings .............

Suggested structure for at high mercury loading ............

The powder diffraction intensities for boehmite heated to different tem-
peratures .................................
The powder diffraction intensities for boehmite heated for different
temperatures ...............................
The result of the Rietveld single phase refinements with boehmite as
initial structure (a) and c-symmetry as initial structure (b) for the
460°C. Experimental data (symbols), the calculated data (line) and
the difference curves shown as offset for clarity .............
The result of the Rietveld for the dual phase refinement with boehmite
and c—symmetry as initial structures for the temperatures: (a) 440°C,
(b) 450°C, and (c) 460°C. Experimental data (symbols), the calculated
data (line) and the difference curves shown as offset for clarity . . . .
The result of the PDFfit2 refinement for the dual phase refinement with
the boehmite and c—symmetry as initial structures for the: (a) 440°C
(b) 450°C and (c) 460°C. Experimental data (symbols), the calculated

data (line) and the difference curves shown as offset for clarity . . . .

xiii

59
60

69

70

71

5.6

5.7

5.8

5.9

5.10
5.11

6.1
6.2

A.1

A2

A3

The change of the ’7-A1203 phase fraction resulted from the PDFfit2
average structure (square), Rietveld refinement (circle) and PDFfit2
local structure (triangle) refinements with temperature ........
The reduced structure functions (F (Q)) for boehmite measured at dif-
ferent heating temperatures: (a) 440°C (b) 450°C (c) 460°C and ((1)
470°C ...................................
The PDFS for boehmite heated for different temperatures ......
The boehmite structure. Aluminum (green), oxygen (red) and hydro-
gen (white) ................................
The nano model structure. Aluminum (green) and oxygen (red)

The result of the PDFfit2 refinement for the local structure with the
dual phase refinement; boehmite and c—symmetry as initial structures:
(a) 420°C (b) 430°C , (c) 440°C , ((1) 450°C , (e) 460°C and (f) 470°C.
Experimental data (symbols), the calculated data (line) and the dif-

ference curves shown as offset for clarity ................

Experimental PDFs for N PC at different pyrolization temperatures. .
Experimental PDFs for (a) loaded CTAB (b) loaded TTeA, (c) pure
CTAB and ((1) pure TTeA. Data were scaled with respect to the am-
plitude of the Si—O peak (1.6A). .....................

The copper finger that cools the sample holder in the CCR refrigerator
that is available in the MUCAT station at the APS ..........
Sample holders for low T measurements that can be used in the MU-
CAT station at the APS .........................
The beryllium window that is used for shielding the sample holder in
the CCR refrigerator that is available at the MUCAT station at the
APS .....................................

xiv

72

83

89

A4

A5

A6

A7

A8
A9

The copper shielding used in the CCR refrigerator that is available at
the MUCAT station at the APS; Kapton is covering the windows. . .
The Kapton displex used to obtain extra shielding of the copper used
in the CCR refrigerator that is available at the MUCAT station at the
APS; Kapton is covering the windows. .................
The XRD data from ZnSeo_5Teo,5 compared to that of the beryllium
window background,copper shielding used in the CCR refrigerator that
is available in the MUCAT station at the APS. ............
The tube that is used to mount the capillary inside the furnace that is
available at the MUCAT station at the APS ..............
The furnace that is available at the MUCAT station at the APS . . .
(a) the PDF of the boehmite sample heated up to 460 °C with 15 min-
utes waiting time (red) and the PDF at the same temperature but
heated for a longer waiting time (blue). (b)the PDFs of the same sam-
ple heated to 480 °C with 15 minute waiting time (red) and the one
at 460 °C heated to a longer time (blue),the PDFs are identical due to

the longer waiting time while measuring the 460 °C PDF .......

Images in this dissertation are presented in color

XV

90

91

92

95
96

97

Chapter 1

Introduction

Atoms are the basic units for all materials. By existing in different arrangements, the
approximately one hundred elements in the periodic table build up all of the known
materials in this world from the air we breath to our own bodies. Arranging the
same atoms in different structures constitutes new materials that differ not only in
their shapes but also in their properties. These differences in many cases are very
dramatic, as is the case when comparing the cheap pencil to the expensive diamond;
both are made from the same element carbon in a different structural form. Over
time, great developments have been achieved in the methods and equipment used
for investigating the structure of the materials, allowing us to study the relationship
between atomic structure and properties.

For the last decade the trend was toward miniaturization in different aspects of
technology and science, where recently the focus of the research was on producing
nanostructures. A nanostructure is an object possessing at least one critical dimension
less than 100 nm in extent. This science has been around for many years where
chemists have been using chemical reactions to combine several atoms to create new
types of molecules. However, working with and controlling atoms and molecules was
limited to large numbers of these nanometer-sized objects. In the last twenty years

there has been a great development of new tools that has enabled scientists to see

single atoms on surfaces (STM) [6] and isolate in a single location a collection of
several hundred molecules (optical tweezers) [7].

the fact that some physical interference, that matter at this scale. Also new
reduction in the number small value, 1000, where dramatically.

In this work we will be dealing with a class of nano-materials known as nanoporous
materials. The following sections give a brief description of these materials and their

importance.

1.1 Nanoporous Materials

Porous materials are materials with pores (holes). The term nanoporous refers to the
class of porous materials having pore diameters between 1 and 100 nm [8]. These
materials are often also referred to as mesoporous. What distinguishes nanoporous
materials, and gives them their great importance in applications, is the high surface
to volume ratio, with high surface area and large porosity, which enables them to
discriminate and interact with molecules and clusters. Nanoporous materials are
sometimes called molecular sieves. The term ”molecular sieve” refers to a particular
property of these materials, i.e. the ability to selectively absorb molecules based
primarily on a size exclusion process. This is due to a very regular pore structure
of molecular dimensions. The pores of these materials are classified into two types:
Open pores in which the pores are connected to the surface of the material. Such
materials are important in separation, catalysis and filtration. Closed pores in which
the pores are isolated from the outside, which are useful in thermal isolation and
lightweight structural applications [9].

The following sections introduce important nanoporous materials, whose struc-

tures have been the subject of this thesis research:

1.1.1 Alkali metals and silica gel: Candidate source for hy-

drogen production

Hydrogen has the potential to solve our dependence on the limited petroleum re-
sources. More important than that, hydrogen is considered a clean source of energy
that does not cause damage to the environment, as opposed to the petroleum resources
which are a major contributer to poor air quality and greenhouse gas emissions. Wa-
ter which makes up 70% of the Earth’s surface, is a rich source of hydrogen. Hydrogen
can be produced by breaking its bonds with oxygen, such as in the electrolysis process.
However, the efficiency of hydrogen production from water is still a challenge.

A possibility to produce hydrogen is by interacting alkali metals with water. Alkali
metals are highly reactive and react with water violently producing hydrogen gas
with enough excess energy to ignite the hydrogen. However, because they react
spontaneously with air, their use on a large scale is a continuing concern for synthetic
and industrial chemists.

New materials with essentially the same reducing power as the parent alkali metals
were developed by intercalation of up to 40 wt% of liquid alkali metals into the
nanopores of amorphous silica gel [5]. These materials are nonpyrophoric, have good
stability in dry air, and slowly adsorb moisture from air without ignition, making
them easy to handle, easing storage concerns.

Despite their promise, the precise nature of the intercalated metals is not known.
Knowing the structure of these important materials, such as the binding of the alkali
metal to the pores inside the silica, is essential to understand the nature of these
materials which will help in improving their properties. Traditional crystallographic
methods that consider only the Bragg reflections (they assume periodicity), are un-

suitable for studying structures of such disordered materials.

1.1.2 Functionalized mesostructued silica: Heavy metals ad-

sorbs

The interest in removing or separating carcinogenic elements has increased lately, as
the regulation limits on environmental emissions become more and more strict. There
has been very active research on making and developing materials that can trap heavy
metals, such as mercury and lead [10].

Functionalizing mesostructured silica with various forms of thiol is one of the
promising materials to trap low levels of mercuric ions from groundwater [10]. These
trapping agents offer very high surface areas, well—defined pore sizes, and high thiol
group loadings of up to 50% [11, 12]. Linking the organic moiety to the mesostructure
can be accomplished either by grafting a thiol—functionalized silane reagent to the
surface silanol groups of the mesostructure [13, 14, 15], or by the direct incorporation
of the organosilyl group into the framework walls during synthesis [11, 12, 16].

Developing these materials requires precise knowledge of their structure. Since

these materials are amorphous, conventional crystallographic methods do not apply.

1.1.3 Gamma-Alumina: Most widely used catalyst

Among the various transition aluminas, gamma-alumina is the most widely used as
catalyst and adsorbent in many chemical and industrial applications. Ranging from
pharmaceuticals and support structures to paint pigments [8]. Despite its importance
and wide range of applications, the structure of gamma-alumina has been a subject
of long debate. This is mainly because gamma-alumina is a porous material. Making
single crystals from such a material is not possible, so that it can not be studied
using single-crystal diffraction methods. Gamma-alumina may be obtained by heating
boehmite. The temperature at which this transformation occurs varies depending on

the starting material and the heating process [17].

Understanding the local and average structure evolution of boehmite and the steps
it passes through before transforming to 'y-AlgOg may lead to better insight into the
structure of this important material and how it is formed.

Traditionally, structure is studied from the Bragg scattering only, where the struc-
ture of the materials is assumed to be periodic. From the the discussion in the previous
sections, disorder is a common feature of the structure of the different nanoporous
materials. As is true for any material, studying the structure of nanoporous materials
is important to understand and improve their properties. Investigating the structure
of these materials will be a challenge and cannot be trivially achieved by studying
only their Bragg scattering. This is due to the pores and aperiodic intercalants in-
side the pores, which cause another kind of scattering called diffuse scattering, that
lies below and between the Bragg peaks. For that reason the atomic pair distribu-
tion function (PDF) method, which is a total scattering method and considers both
components of scattering, will be used to study the structure of these important

nanoporous materials.

1 .2 Thesis layout

This thesis is organized in five chapters, as follows: The next chapter provides a brief
description of the atomic pair distribution function (PDF) method and the rapid
acquisition PDF experimental technique, which was used in collecting all of the data
that are presented in this thesis. Chapter 3 presents a study for the structure of
alkali metals loaded into silica gel. In chapter 4 we study the structure of mercury
binding sites in mesostructured silica. The in situ study of the structure evolution
of boehmite to gamma is presented in chapter 5. Chapter 6 contains concluding
remarks on the structure studies presented in the previous chapters with suggested
future studies for other nanoporous materials. Finally, a detailed description of low

and high temperature RA-PDF experiments is presented in the appendix.

Chapter 2

Atomic Pair Distribution Function

Technique

2. 1 Introduction

The structure of a material plays a major role in Specifying its properties. For example
carbonaceous materials (graphite, diamond and nanoporous carbon) are all composed
of the same element carbon but with different structures. This causes them to have
different properties and as a result different applications. Developments in the tech-
niques that are used to study the atomic structure of materials have led to great
advances regarding the knowledge of materials. Single crystal diffraction is the best
way to determine the structure of materials, but there are limitations for the appli-
cability of this method. High quality single crystals are required to perform such
measurements, which are not obtainable for many of the materials of technological
interest. In such cases, powder diffraction becomes the method of choice. The sam-
ples are easily prepared by grinding coarse crystalline materials, or as fine grained
precipitates, or in the case of metals by making filings [18]. Powder samples con-
sist of an enormous number of very small crystallites (much less than 10‘3 cm) with

completely random orientation. The basic assumption in crystallographic methods is

that the material consists of a periodic lattice which gives rise to discrete diffraction
peaks called Bragg reflections [18]. As advances in science and technology proceed,
the complexity of the structure of the materials used increases. Traditional crystal-
lographic methods do not reveal all of the information needed to study the structure
of complex materials. This is because they yield information only about the long
range order of the materials, while in many cases the most important properties of
the material are due to its short range order. For example, the optoelectronic prop—
erties of many materials and many electrical properties of semiconductors and high
temperature superconductivity depend upon the presence of various types of disor-
der [19, 20]. Another approach to follow in studying such materials is the atomic pair
distribution function (PDF) technique since it considers both the Bragg and diffuse
scattering [21]. The PDF analysis of powder diffraction data has been limited for
many years to the study of short range order in materials such as glasses and liquids
and for many it was considered as the method of last choice [22]. This is because the
PDFs used to have large spurious oscillations which appear because the beam sources
could not provide short enough wavelengths, a condition that is needed to measure ac-
curate PDFs. Another reason is that the computational methods were not advanced
enough to handle the intensive data analysis and modeling required for the PDF
method. With the great developments achieved in computational methods and the
availability of the synchrotron based radiation sources which provide high-intensity
short-wavelength particles, the PDF technique has been used to study a variety of

materials; with crystalline, nanocrystalline and amorphous structures [4, 21].

2.2 Local and average structure

Xray, neutron or electron diffraction experiments, rely upon the fact that when the
beam hits the powder sample it gets diffracted due to its interaction with the elec-

trons (x-ray and electrons) or with nuclei (neutrons) [18]. If the crystals consist of a

perfect 3-D stacking of atoms at rest the scattered intensity would exclusively con-
sist of the discrete set of Bragg reflections. However, the perfect crystal does not
exist because at least thermal motion brings some disorder. Also in many materials
the atoms are not fully ordered with point defects and displacements present. The
deviations from perfect order give rise to diffuse scattering which is a weak contin-
uous background located outside the Bragg reflections. Considering only the Bragg
diffraction yields information about the average structure, while the diffuse scattering
contains information about the short range order. Figure 2.1 shows two simulated
structures with the same lattice and the same concentration of vacancies [3]. It can be
seen that the Bragg peaks for the structure with (a) random distribution and (b) the
one with short range order are identical. The difference between the two structures
can bee seen through its diffuse scattering, shown in the inset to Figure 2.1(c) where
the powder diffraction has been magnified by 100x. To study the specific properties
of materials that are linked to the disorder we need to consider the total scattering

including both the Bragg and diffuse scattering.

2.3 Definition of the pair distribution function

The PDF gives the probability of finding two atoms separated by the distance, 7' [21].
It is like a distance map of the inside of the material. To clarify the meaning of the
PDF we will study the PDF of the Bucky ball (C60). C60 is a molecule that consists
of 60 carbon atoms, arranged as 12 pentagons and 20 hexagons. Figure 2.2 shows a
picture of the C60 molecule and its PDF C(r). Sitting on an atom in the molecule,
the nearest neighbor atom will be at 1.4 A which corresponds to the first peak in the
PDF. Going to the next neighbor C atom, which is at distance 2.2 A, corresponds
to the second peak in the PDF and so on. As can be seen from the PDF there are
no sharp peaks after 7.1 A, which is the diameter of the Bucky ball. Instead broad

oscillations extend up to 20 A, which corresponds to the ball-ball distances. The

      
   

 

   

 
  

I. 0..

OI

.

O

:z....
0... I. .0.

  

.C.... I
I. 0.. .0.

 

 

 

 

 

 

 

 

 

 

2
l

0 02 0 04
t

 

 

Intensity

1
l
I

 

 

 

 

 

 

 

20 30 40 50
© 28 (degrees)

 

C(r) (3'2)

 

 

 

I . l A l . l I l . l n l . l

 

4 5 6
@ r (A)

Fig. 2.1 (a) Random distribution of vacancies and (b) preferred ordering in x- and y-
direction. Panel (c) shows the corresponding powder diffraction patterns. The inset shows
the background enlarged by a factor of 100. Panel (d) shows the corresponding PDFs for
the random distribution (filled circles) and the short-range ordered distribution (line) [1].

 

10

A
F.
I“)
A
0‘ I
E .
0°.

 

1.1 A
—5

 

 

 

all")

4-80246

    

 

A

. l . l a l . l L
10 12 14 16 18 20
r“)

”-
p

*-
s

.—

 

Figure 2.2: Fullerene (C60 single molecule structure) that forms fcc lattice in the solid
060- Alse shown the neutron powder diffraction data and the corresponding PDF.
The sharp peaks in the low-r yields from the C—C correlations that extends up to
7.1 A , (dashed line) the diameter of the of the Bucky ball. The broad oscillation
that exists above 7.1 A represent the inter—ball correlations

oscillations are broad due to the spinning of the balls. This shows the capability of
the PDF method in the study of both the intra and inter domain structure of the

materials.

2.4 PDF from experiment

As is the case in any experimental method, obtaining the PDF requires two steps:
collecting and processing the data. The details and accuracy of the information that
can be obtained from the PDF depend essentially on the quality of the collected
data. Obtaining a high real space resolution requires one to measure the scattered

intensity up to a high value of scattering vector Q with good statistics [21], where Q

10

= 47r sin O/A, 26 is the angle between the incoming and outgoing radiation and A is
the wavelength of the x-ray radiation. Another important factor is the time needed
to collect such high quality data, especially in parametric studies which require the

measurement of many data sets.

2.4.1 Rapid-acquisition pair distribution function (RA-PDF)
technique

Conventionally x-ray diffraction experiments were performed using energy-resolving
point detectors. Since obtaining good statistics over a wide range of angles is a pre-
requisite for obtaining high resolution PDFs, these measurements are time consuming
even when using synchrotron radiation. For example, collecting the data for back-
ground may need more than eight hours. To overcome this issue Chupas et al [23]
developed a rapid-acquisition pair distribution function (RA-PDF) method. Using
this method they were able to obtain a high-quality PDF (Qmwg 28.5 A”) in a
few seconds of data collection time. The technique uses the Mar345 image plate
camera (IP) for collecting the scattered intensity instead of point detectors. Area
detectors offer a large solid angle coverage and rapid readout. By measuring the
entire scattering range simultaneously and averaging the measured intensity over a
large solid angle, the total structure function, S (Q), can be determined accurately.
The IP camera is a round disk with usable diameter of 345 mm. This is mounted
orthogonally to the beam path with the beam centered on the IP, Figure 2.3. The
diffracted x-rays from isotropic powder samples result in a set of concentric rings
(Debye—Scherrer rings [23]. To avoid saturation of the detector, measurements are
performed by using multiple exposure durations depending on the sample. Using the
program Fit2D [24] the raw data sets are averaged to improve the statistical accuracy
and reduce some systematic errors in the experimental setup such as powder averag-

ing. The data are then normalized with respect to the monitor counts and integrated

11

area detector

        

sample monochromator

 

 

 

 

 

Q(S(Q)-1)

 

 

 

Q (A‘)
Figure 2.3: RA-PDF experiment setup. Adapted from [4]

to produce 1-dimensional x—ray diffraction data. The sample to detector distance,
which is an essential parameter for doing the integration, is determined by using a
standard sample, usually silicon. The samples may be measured either in flat plates
or capillaries. By obtaining high quality PDFs in a short time, the RA-PDF methods

opens the way to study structure changes under in situ conditions.

2.4.2 Getting the PDF

Collecting the diffracted beam from the samples and integrating the 2D images result
in the intensity versus 20. The measured intensity (IT) will be composed of several

parts:

IT 2 Icoh + IinC+Ims + Ibg (2.1)

where I°°h is the coherent scattering intensity, If”C is the incoherent scattering which
arises from Compton scattering, Ims is the multiple scattering intensity and I” the

background intensity which includes scattering without sample due to sample holder,

12

air, optical system, etc..

It is important to ensure that the processed PDF data contain signal stemming
solely from the sample so data from an empty sample holder and the scattering
environment is collected for background subtractions. The information related to the
structure is contained in the coherent scattering. For the corrections related to the
other forms of scattering we use the program PDFgetX2, [25] such that the total

scattering function S (Q) is given by the following relation:

: 16"" (Q) — mils-(0)12

S (Q) I: cu:- (on?

 

+ 1 (2.2)

where I C""(Q) is the measured scattering intensity and c,- and f,(Q) are the atomic
concentration and x-ray scattering factor, respectively, for atomic species 2' [18, 21].
The corresponding atomic PDF, C(r), can be calculated by Fourier transformation

of the reduced structure function, F (Q) = Q[S (Q) — 1], as follows:

QMGZ

C(r) = (7%) / QlS(Q) — 11sin<Qr>do (2.3)

min

Here, Q is the magnitude of the wave vector and Qmam is where the reduced structure

function F(Q) is truncated before the PDF is calculated.

2.5 Experimental errors in the PDF

The experimental PDF will always have some errors due to the limitation of the
experiment setup and the processing of the data, but fortunately the effect of these

errors can be reduced such that the results obtained from the PDF are worthy trust.

13

2.5.1 Termination errors

The ideal experimental PDF is the one obtained by performing the Fourier transfor—
mation in Eq. 2.3 from zero to infinity. Unfortunately this is not achievable practically.
There will be always a limitation on the available minimum (QM-n) and maximum
(Qmax) values of the momentum transfer Q due to the experiment setup. Usually
the effect of Qmin on the quality of the PDF is negligible, since small enough values
of Qmin can be obtained easily from most experimental setups, which is not the case
for Qmax. The cut—off in F (Q) for Q > Qmax has two different effects: Termination
ripples and the decrease in the 1' resolution. The termination ripples are false os-
cillations that are not related to the structure that appear in the PDF beside the
real peaks [26]. Figure 2.4 shows the PDF for silicon calculated for different values
of Qmax. As can be seen, by increasing Qm up to 30 A"1 the termination ripple
around 3.2A vanishes. The other effect of terminating the F (Q) is on the r-resolution
of the PDF, which may be approximated by (27r/Qmax). This effect is explained in
Figure 2.5 [3]. Here the PDF of (ZnSeo,5Teo_5) was terminated at two different val-
ues of Qmax. For QM = 40 A"1 two distinct peaks can be seen at 2.4 A (Zn-Se
bond) and 2.6 A (Zn-Te bond). While for Qm = 17 A only one sharp peak is
observed. Figure 2.4 and Figure 2.5 show the importance of obtaining a high Qm:
first to eliminate the termination ripples and secondly to increase the r-resolution.
Also from these examples it can be seen that the effects of the Qmaa, termination can
be reduced greatly within experimentally accessible ranges of Qm. Since Qm<54f,
obtaining a high Qmaz requires short wavelength. Laboratory x-ray sources such as
CuKa give Qmamrv 8 A‘l, and for MoKa Qmax~ 16 A‘l. Depending on the nature
of the materials we want to study Qm> 20 A‘1 is required. Such a high Qm is
obtainable by using synchrotron based x-ray sources, which provide high intensity

and short wavelength x-rays.

14

 

 

 

' I ' I ' I ' I ' I ' I ' I
O- . .-
*- a

tf‘ui- , - -

5 I .0 O a

‘90
N I T
' . o o I
* ..
I

. I . I I . I . I I . I '

 

 

1.6 2 2.4 2.8 3.2 3.6 4
r (1)

Figure 2.4: The calculated PDF of silicon. Solid line corresponds to termination at
Qmam = 30 A ‘1 and the dotted line for the same data terminated at QM = 10 A ‘1.

2.5.2 Statistical errors

Getting high Qmax is the cure for obtaining a PDF with a high r—resolution and to
reduce the termination ripples. The intensity of the diffracted beam falls off with
increasing Q. As a result the signal at high Q becomes progressively dominated
by noise, which limits the accessible range to which the experimental F(Q) can be
extended. The effect of the noisy data can be decreased by increasing the collection
time of the data. In general a balance between the noise and the termination errors

is needed to obtain the best possible PDF from the data.

15

 

 

 

 

 

I l I 1 I 1 I l
2.2 2.4 2.6 2.8
r (K)

 

Figure 2.5: PDF of ZnSeo_5Teo_5. Solid line corresponds to termination at Qmax =
40 A“1 and the dotted line for the same data terminated at Qmaa, = 17 A '1. Adapted
from [3]

2.5.3 Instrument resolution

Ideally the PDF experiment would involve a detector that can resolve all of the
details of the I (Q) Real equipment has a limited resolution which, depending on
the sample, may have a major effect on the PDF. The effect of the Q resolution is
equivalent to multiplying the PDF by a function such as a Gaussian, which causes
the PDF to decay at high-r values [26, 21]. For amorphous materials the effect of this
damping is negligible, because these materials have short range order such that their
PDFs already diminish at low 7' (around 10 A) before the effect of the Q—resolution
starts to take effect. For crystalline materials, where theoretically the PDF extends

to infinity, this issue becomes important, especially when investigating the high-r

16

region. The same applies for the materials that are composed of limited domains of
limited structure coherence, for which the PDF will decay at the limit of the domains.
In these cases it becomes crucial to determine if the decay in the PDF is due to the
structure (domain size) or due to the instrument used (Q-resolution). This requires
the performance of a high resolution PDF experiment from which we can specify the

coherent size of the domains for the material under study.

2.6 Extracting information from the PDF

The PDF for a perfectly ordered crystalline structure is the sum of well-defined delta
functions, the positions of which give the separations of pairs of atoms in the structure.
In real materials atoms are displaced from their perfect position due to the thermal
motion and/ or static displacements of the atoms. This gives rise to a distribution
of atom-atom distances, which causes the PDF peaks to be broadened. Three basic
properties of the PDF peaks may reveal great information about the structure: The
position of the peak yields information about the atom-pair distances, the width of
the peak reveals information about the disorder of the atoms involved in the pair
and the integrated intensity under the peak gives the coordination number of the
origin atom. These three important properties of the PDF peaks can be extracted
easily by fitting Gaussian peaks convoluted with a sinc function that accounts for the

termination effects.

2.6.1 Peak position

The peak position yields bond-lengths directly, which can be very useful in under-
standing the local atomic structure. For example while investigating the structure
of the semiconductor alloy In1_xGaxAs Petkov et al [27] studied the evolution of the

bond lengths with alloy composition. Based on the dependence of the peak position

17

on a: they concluded that local atomic-level strain exists.

2.6.2 Peak width

The peak width may be obtained by fitting a Gaussian function. Since the number
of neighbors is constant the integrated area under the peak is invariant and the peak
height, extracted directly from the data, gives the inverse peak width. This often gives
a more accurate determination of the evolution of peak width with some experimental
parameter such as temperature or composition. Three kinds of information may be

obtained by studying the peak width:

1. The width as a function of temperature yields information about the Debye

temperature of a bond [28].

2. The width as a function of atomic separation yields information about correlated

atomic dynamics [29].

3. The width as a function of doping gives information about doping induced

disorder [30]

2.6.3 Peak intensity

The integrated intensity under the PDF peak yields information about the number of
atoms at a specific distance i.e. coordination number). This type of analysis is widely
used in studies of glasses [31] and in partially crystalline samples [32]. For example;
while studying the structure of nanoporous carbon; Petkov et al [33] used the fall of
the first peak peak intensity with temperature as evidence of disorder in the sample

heated to 800°C compared to the one heated to 1000°C.

18

2.6.4 Real Space refinement

Once the experimental PDF is obtained we want to extract all of the possible infor-
mation that we can get from it. As discussed in Section 2.6.3, we can obtain a lot of
information from the peak position, width and intensity. However determining the
structure is still an essential part to understand and possibly improve the properties
of the material under study. Conventional structure determination depends on the
intensity and position of Bragg peaks. The most common method for such analy-
sis is the Rietveld method [34]. A least-squares refinement between the calculated
and observed intensities is performed until the best match with the measured profile
is obtained. The calculated intensities are obtained based on the crystal structure,
thermal factors, diffraction optics, instrumental factors, lattice parameters and other
specimen characteristics. The success of the method can be gauged by the publication
of more than a thousand scientific papers yearly using it [35]. Considering only the
Bragg peaks assumes perfect long range periodicity of the crystal. Such a presump—
tion prevents the studying of non-periodic structures, or aperiodic modifications to
otherwise crystallographic materials, to be studied. However many important mate-
rials are disordered and many of these materials owe their important properties to
these deviations from the average structure. These deviations result in the occur-
rence of diffuse scattering, which contains information about two-body interactions
and which is disregarded as a background in the Rietveld method. This shows the
need for performing a similar refinement to the PDF since it considers both kinds of
scattering. Full profile refinement of the PDF can be carried out using the program
PDFfit2 [36]. In this method the model is defined in a small unit cell with atom
positions specified in terms of fractional coordinates. The PDF is calculated from
the model structure as follows (Figure 2.6), (1) assume that we have a sample that
consists of N atoms at position 1",, with respect to some origin. Place the origin of

our space randomly at any atom. (2) Systematically find every other atom in the

19

 

 

 

 

 

 

 

12 3 4 5.6
NA)

Figure 2.6: Calculating the pair distribution function.

sample and measure its distance from the origin. (3) Each time we find an atom we
place a unit of intensity at that position, rm, on the axis of the radial distribution
R(r). When we have cycled over all of the atoms in the sample we move the origin
to another atom and repeat the process, adding the intensity to the R(r) function.

We multiply the unit of intensity for each atom-pair by bmbn/ < b >2 where b, is the

20

scattering length of the ith atom. Dividing by the number of atoms to keep R(r) an

intrinsic function, we obtain the relation

=N lzbm lblbgna — _ Tm» (24)

which is related to the PDF,

Gcazc(r) = R(r) — 47rrpo (2.5)

the sum goes over all the pairs of atoms i and j within the crystal. (b), is the
average scattering power of the sample. The goodness of the fit is determined by the

parameter Ru, which is computed according to the relation:

Zi=1w(ri)lGobS(ri) " Gcalc(ri)l2
12...: l/z w<r.-)G:. .(n-l (2'6)

 

 

Here w is the weighting factor, w(r,-) = 1/0,2 where a is the estimated standard
deviation on the ith data-point at position 73-. The sum goes over all measured
data points 1‘,- in the experimental PDF. Performing the sum for all the atoms in the
sample (1023) is impractical, but the sample is made of many equivalent cells which are
periodically repeated in the space so that we can limit the calculation to one unit cell.
The refined parameters are the same as those used in Rietveld. The main difference
from Rietveld is that it allows for different 1" scaling refinement, which enables one
to study the local structure for different r—ranges [21]. The ability to refine the local
structure yields information about disordered and short-range atomic correlations.
Because of the similarity between the Rietveld methods and the PDF modeling a
quantitative comparison between the resulting structures of both refinements may be
made. This is an important first step in revealing the existence of local distortions

beyond the average structure. It should be noted that the pr for the PDFfit2 is

21

similar to that used in Rietveld analysis but the functions being fit are significantly
different. Hence, direct comparison of pr from PDF and Rietveld analysis should
not be made. pr values are useful measures of the goodness-of-fit when comparing
how different models fit to the same PDF data. For well-crystallized samples, PDF
pr values greater than 10%, are not uncommon. Obtaining a pr value of less than

20%, for nanocrystalline structures is excellent.

2.7 Other methods to probe the local structure:
Nuclear magnetic resonance

The Nuclear magnetic resonance (NMR) is a spectroscopic technique that uses the
absorption of radio frequency radiation (RF) and magnetic fields to determine the
structure of compounds. It is limited to nuclei that posses magnetic moment and spin
angular momentum (13C not 12C). The physical principle behind these techniques may
be simplified as following: Consider a single proton which has spin angular momentum
$45., placed in a magnetic field. The proton will have two possible states to occupy
with potential energy iuh. Applying a radio frequency field with resonant frequency
1! = 2pH0 will cause the nuclear spin to realign or flip in the higher—energy direction.
For a large samples of protons, the protons will align themselves in the two states
with slightly more protons (~ 10’6) in the lower state. In that case, applying the
resonant RF will lead to transitions between the two states with small absorption of
energy. After absorbing energy the nuclei will re-emit RF radiation and return to the
lower—energI state. The energy of an NMR transition depends on the magnetic-field
strength and a proportionality factor for each nucleus called the magnetogyric ratio.
The local environment around a given nucleus in a molecule will slightly perturb the
local magnetic field exerted on that nucleus and affect its exact transition energy.

This dependence of the transition energy on the position of a particular atom in a

22

molecule makes NMR spectroscopy extremely useful for determining the structure of

molecules.

2.7 .1 Solid state NMR

The N MR method can be applied to liquids and solids. The solid state N MR absorp-
tion lines are much broader than those of liquids. Still, in both cases they provide
information about the structure of the material under study. There are many reasons

that may cause the broadening in the NMR spectra. We mention some of them:
1. Magnetic dipolar interaction:

This broadening arises due to the interaction between the spins with each other,
which produces a small magnetic field (local field) that affects the other nuclei in
the sample. The strength of the local field changes from nuclear site to another,

and as a result they will have different resonance frequencies.

2. Chemical Shift:

The chemical shift is a shift of the nucleus resonance frequency due to the orbital
motion of the electrons. The chemical shift of a nucleus is calculated from the
difference between the resonance frequency of the nucleus and a standard. This

quantity is reported in ppm and given by the relation:

(5: (V—VTCf)

1 6 2.7
11..., x 0 ( )

The standard for protons or silicon is often tetramethylsilane, Si(CH3)4, abbre-
viated TMS. The chemical shift arises due to the magnetic field produced by
the electrons around the nucleus. This local field opposes the applied magnetic
causing a shield around the nucleus, such that the field around the nucleus is
H = Happ(1 — 0). Here, 0 depends on the electron density around the nu—

cleus. The more the electrons surround the nucleus (heavy nuclei) the more the

23

chemical shielding affects the NMR spectra. The shift is very sensitive to the
environment around a nucleus. If a is anisotropic this will yield broadening in

the NMR spectrum.

. Knight Shift:

This shift exists for metals and arises due to the polarization of the conduction
electrons in a metal, which causes an increase in the magnetic field the nucleus
experiences. It was first observed by Walter Knight [37]. Applying the same
static field, he found that the resonance frequency of 63Cu in metallic copper
was 0.23% higher than that of the diamagnetic CuCl. For non-cubic metals the

Knight shift is anisotropic which leads to a broadening in the NMR spectrum

24

Chapter 3

Structure study of novel alkali
metal intercalated silica gel with

potential as hydrogen source

3.1 Introduction

3.2 Alkali metals and silica gel: New novel mate-
rials

Alkali metals are extremely reactive reducing agents that produce hydrogen when
they react directly with water, making them a hydrogen source candidate for the
production of clean fuel. Because their reactivity in air renders them difficult to
handle, they are typically present in other forms such as oxides, hydroxides and
halides. Taking full advantage of the reducing the power of the alkali metals has
therefore been a big concern for synthetic and industrial chemists for a long time [38,
39,40]

Recently, candidate materials with good air stability for producing hydrogen were

25

developed by doping liquid alkali metals into silica gels [5]. These materials have been
found to absorb as much as 60 mole percent alkali metal from the liquid phase into
the silica. By comparison, alumino—silicates and silica zeolites absorb up to 12 mole
percent of alkali metals from the vapor phase [41, 42, 43, 44]. These materials are
relatively easy to handle and react quantitatively with water to produce hydrogen.
These novel materials are highly promising in hydrogen production and reducing

applications because of their reactivity but ease of handling.

3.2.1 Silica gel

Silica gel is a form of silicon dioxide, SiOg, the material that occurs in nature as
sand. The difference between silica gel and sand is that sand is a crystalline, non-
porous form, whereas silica gel is non-crystalline and highly porous. It has an internal
network of interconnecting microscopic pores, yielding a typical surface area of 700-
800 square meters per gram; or, stated another way, the internal surface area of a
teaspoon full of silica gel is equivalent to that of a football field [45]. As opposed to
zeolites, silica gels have larger pores with a wide range of diameters between 5 A and
3000 A and do not allow for the separation of molecules solely dependent on their size.
Chemically, silica gel is an inert, non-toxic, material synthetically manufactured from
the chemical reaction between sodium silicate and sulfuric acid. It was known from as
early as 1640, but had remained a curiosity until its adsorbent properties were found
useful during World War I by the chemistry professor Walter A. Patrick, who had
found it useful for the adsorption of vapors and gases in gas mask canisters [46]. Water
molecules are adsorbed or absorbed by these micro-capillaries until vapor pressure
equilibrium is achieved with the relative humidity of the surrounding air. During
World War II, it was commonly used as a dehydrating agent to protect military and
pharmaceutical supplies, among a number of other applications. Its use as a buffering

agent to control relative humidity within the mid-range rather than as a desiccant is

26

 

 

 

 

 

Figure 3.1: Room-temperature absorption of Na2K alloy into. (A) The SG coated
with alkali metal alloy just after mixing. (B) The final loose black powder (stage 0)
formed by shaking the sample shown in A for a few minutes. Adapted from [5]

unique to museum applications.

3.2.2 Sample preparation

The samples are straightforwardly prepared by exposing silica gel to molten alkali
metal, M and shaking in a beaker at room temperature [5]. This results in a loose
black powder (denoted M-SG—O, stage 0) as it is shown in Fig. 3.1. This powder is
reactive and must be handled in an inert atmosphere to avoid ignition. The material
undergoes a reaction on heating in inert atmosphere to 150 °C, or by reacting Na
with SG at N 165 C with continual agitation (M—SG, stage I). The product is also
a black powder. It does not ignite in dry air and exposure to dry oxygen for up to
one day does not reduce its reducing capacity. However, it quickly degrades in moist
air. Slowly heating the M—SG—O to 400 °C results in a second reaction. This material
(denoted M—SG, stage II) was found to be stable in open air and its reducing capacity
does not diminish with time. It is the material of choice for hydrogen production and
solvent drying.

Three powder samples were prepared for room temperature X—ray scattering ex-

periments:

1. NagK-SG-I: In this case the alkali metal, M, was the Na2K alloy which is liquid

27

at room temperature. The alloy was mixed with silica gel at room temperature

resulting in a product with an alloy content of 40 wt%.

2. N a—SG-I: Pure sodium was used as the metal and heated with the silica-gel to

150 °C resulting in a product with a Na to SI02 mole ratio of about 1.521.

3. Na—SG-II: Pure sodium was heated with silica gel, while agitating, to 400 C

resulting in a product with a Na to SiOg mole ratio of about 1.521.

3.3 Structure of alkali metals doped into silica gel

As is true for any material, knowing its structure is a prerequisite for understand-
ing and developing its properties. Determining the structure of the products is not
straightforward because they are not crystalline and involve species intercalated in-
side an amorphous silica host. Traditional crystallographic methods cannot be used
in this case because the intercalants are not periodically arranged [47]. For that we
used the atomic pair pair distribution function (PDF) to study the structure of these
materials. The method has been shown to be useful in studying the structure in sim-
ilar situations [47, 48]. Combined with the other measurements (NMR, conductivity,
differential scanning calorimetry (DSC) and magnetic susceptibility) we were able to

obtain a comprehensive idea about the structure of these important materials.

3.3.1 The PDF experiment

These samples were packed and sealed with kapton tape in flat plate sample holders
of 1.0 mm thickness. X-ray diffraction data were collected using the rapid acquisi-
tion pair distribution function (RA-PDF) technique [23] using an incident energy of
87.005 keV (A = 0.142479 A) at the 6—IDD beam line at the Advanced Photon Source
(APS), Argonne National Laboratory. A 2—dimensional circular image plate camera

(Mar345®) 345 mm in diameter was mounted orthogonal to the beam with sample to

28

 

 

ij (<3).
L (by
(of

l (orb.)
O 30 . 60 ' 90 120 150 180 210

 

 

 

mi
‘é‘ é ‘110‘1.2‘1I4A1l6.1.8.
Q (3“)

Figure 3.2: Experimental powder diffraction patterns for (a) pure silica, (b) Na2K-
SG-I, (c) Na—SG-I, and (d) Na-SG-II. The pure silica data were scaled to facilitate
easier visual comparison.

2 4

O

detector beam path calibrated to be 203.87 mm [23]. In order to avoid saturation of
the detector, each measurement was carried out by using multiple exposure durations
to the X-rays, ranging from 100 to 1000 3, depending on the sample. Using Fit2D [24]
the raw data sets obtained were averaged to improve the statistical accuracy and re-
duce any systematic error in the experimental setup, normalized with respect to the

monitor counts, and then integrated to produce 1-dimensional X-ray diffraction data.

3.3.2 The PDF results

The resulting 1D x-ray powder diffraction patterns obtained for the samples are shown
in Figure 3.2. It can be seen from Figure 3.2(b) and (c) that the X-ray diffraction
patterns of NagK-SG-I and Na—SG-I, respectively, are diffuse in nature and similar
in appearance to the diffraction pattern of pure silica gel, which has been shown for

comparison (Figure 3.2(a)). Na—SG-II exhibits a higher but limited number of broad

29

 

I

(a).

 

'10'12

 

(n ..

9 - (b)
8°" '
L Q. I. (C)

N
o (a)

2 4 6 810112141618
C(r)

Figure 3.3: Experimental reduced structure functions for (a) pure silica, (b) Na2K-
SG-I, (c) Na—SG-I and (d) Na-SG-II

Bragg peaks yet still has a large diffuse component. This suggests that the reaction
products in this material are nanostructured. The amorphous materials cannot be
studied by crystallographic techniques and require a PDF-type of analysis. Recent
experience also suggests that the nanocrystalline Na-SG-II sample will also benefit
from a similar kind of analysis over traditional crystallographic methods [49]. The
reduced structure functions F (Q) = Q[S (Q) — 1] were truncated at QM of 20 A"1
before the PDF was calculated. The F(Q) was multiplied by the Lorch function
before Fourier transformation to damp the data in the high-Q region and reduce
noise in the PDF due to a sharp termination [50] The structure functions and PDFs
obtained are shown in Figs. 3.3 and 3.4, respectively. Conversion of the XRD data to

the scattering functions and PDFs was performed using the program PDFgetX2 [25].

Comparison between the data in Figs. 3.2, 3.3 and 3.4 illustrates why this ap-

30

 

 

 

a VA
‘71. b)
or. o
VP" A; (0]
o A, v _
(d)
O

l A l n I A I n l ‘44 l . lLl

4 6 81012141618
r(fi)

 

 

30 35

5 1o 15 2o 25
r01)

 

Figure 3.4: Experimental PDFs, scaled with respect to the Si—O peak amplitude for
(a) pure silica, (b) NagK-SG-I, (c) Na-SG-I, (d)Na—SG-II, (e) Na—SG-II on different
length scale

proach yields quantitative local structural information in nanostructured materials.
First, the data-reduction step to obtain F(Q) results in an amplification of the im-
portant diffuse signal at high-Q. In the reduced structure functions (Figure 3.3) all

diffraction features, including those at higher wave vectors, appear equally strong and

are therefore equally important in the structure determination.

Stage 0 and Stage I

The features of the reduced structure functions (Figure 3.3) for NagK-SG-I and N a-
SG-I are broad, synonymous with the disordered nature of the materials discussed
above. This disorder is also reflected in the PDF’s (Figure 3.4) which exhibit a rapid
fall-off in the structural features with increasing 1'. When compared to pure silica gel,

the general features of the reduced structure functions and PDFs of the NagK-SG-I

31

 

 

10‘15 20 25 30‘35
r(fi)

Figure 3.5: Experimental reduced structure functions for NagK-SG-I (blue symbols)
and the calculated PDF from metallic bcc sodium
(red line)

 

and Na—SG-I are similar but still show some differences. In particular there are broad
oscillations present in the PDFs of NagK-SG—I and Na—SG-I that extend up to 15 A
that are absent for the pure silica gel. The broad oscillations in the PDFs are due to
the presence of the nano clusters. This is characteristic of presence of small particles
in a system [51]. To investigate the origin of these clusters in Na2K-SG-I we tried
to refine it by using the Na2K structure model [52]. The result of this refinement
is shown in Figure 3.5(a), where it can be seen the refinement was not successful.
Finally we tried to fit the ripples by using a simple metallic bcc assuming that the
cluster was a disordered alloy with the same bcc structure as the end members. We
calculated the PDF from the structure of metallic sodium and compared it to the
PDF as shown in Figure 3.5. As it can bee seen from this Figure a good agreement
is obtained between the NagK-SG-I measured PDF and the calculated PDF from
the metallic bcc sodium. Large thermal factors (0.15 A2) were used, indicating a

significant amount of structural disorder. This indicates that significant displacive

32

 

 

 

 

       
         

 

-§' 1 l ' 'iTT ' U I ' l I I U r7.-
N. I
.3 O .1
—m -
Q)
,Q; 0 .
%
6‘ ’:° ? I . I . LLI . I LI . ‘
O
l "'00 10 15 20 25 3O 35 "
S .0 rm .
00
00 °
00

 

 

 

IIIIIIIIJI

I I
24681012141618
r01)

Figure 3.6: Experimental reduced structure functions for Na—SG-I (blue symbols, and
the calculated PDF from metallic bcc sodium

 

disorder exists in the atomic positions in the cluster in the silica gel. From Figure 3.5
it is evident that the diameter of at least some of the clusters reaches at least 37 A.
This is rather smaller than the largest pore-sizes of the gel 150 A which also suggests
that the clusters have limited structural coherence.

For Na-SG-I, we calculated the PDF from the structure of metallic sodium and
compared it to the PDF of Na—SG-I. As evident in the inset of Figure 3.6 the broad
oscillations in the PDF of Na-SG-I that appear in the range of 9-40 A matched
perfectly with peaks from the Na PDF. This indicates the existence of Na nano

clusters inside the Na—SG-I sample.

Stage II

The reduced structure function and PDF for Na—SG—II (Figs. 3.3d and 3.4d) contain
more well defined peaks than the lower stage materials, indicating more structural

order. Figure 3.4(e) shows the PDF for Na-SG-II plotted out to 48 A. The PDF

33

signal is observed to be taper out at about 40 A, indicating the size of the coherent
domain. When attempting to determine the structure of the stage II material its

proposed formation reaction was considered:
5Na+BSiOg ——> 2NagSiOg + NaSi

Na28103 has an orthorhombic structure with lattice parameters a = 10.43 A, b =
6.02 A and c = 4.81 A [53]. Refinement of this structure to the experimental PDF
that is shown in Figure 3.7(a), yielded an pr of 37%. Overall this fit appears
satisfactory except for the peaks around 3 A where there is an obvious difference
between the experimental and the calculated PDF. To improve the quality of the fit,
sodium silicide (NaSi) was incorporated as the second phase in the refinement. N aSi
is known to have the base-centered monoclinic structure, with lattice parameters:
a = 12.19 A, b = 6.55 A, c = 11.18 A, and a = 119° [54]. This resulted in a
considerable improvement in the refinement fit (Figure 3.7b) with an Ru, of 17%.

The structural parameters for the dual-phase model are presented in Table 3.1.

3.4 Other measurements

The PDF results discussed so far yielded a clear insight about the structure of these
new materials, but still other methods were used to support and to give a complete
understanding of the structure of the materials. NMR, Magnetic susceptibilities and
conductivity measurements were performed by the Professor Dye group. The follow-

ing description of these experiments is taken from a joint publication [55]

3.4.1 The NMR results

23N a MAS N MR. Solid-state 23Na MAS N MR spectra provide a good diagnostic for
metallic sodium. The paramagnetic Knight shift of metallic sodium is 1040 ppm from

that of Na+(aq). Thus peaks in the region of 1100 ppm provide direct evidence for

34

 

 

 

 

I A I A I A 1 J I I

24881012141818
r01)

Figure 3.7: Comparison between experimental (circles) and calculated (solid lines)
PDFs for the Na-SG-II sample, using (a) a single phase Na28i03 model and (b)
(offset) a dual—phase model containing NagSiO3 and N aSi. The residual plots are
shown displaced below the fits overlaid on each other. Magenta for the single phase
residual and gray for the dual phase.

the presence of Na", while those closer to 0 ppm are due to Na+. Although integrated
intensities are not quantitative for such solid-state spectra, they provide qualitative
estimates of the relative amounts of metal and cation present. An early indication of
the metallic nature of our samples was their effect on sample spinning. Conductive
samples are hard to spin in a strong magnetic field because of opposing Eddy currents.
We found it necessary to dilute many samples with solid boron nitride in order to
spin them at 3—4 kHz. This indicates the presence of inter-grain conductivity that
was confirmed by measurements of the dc conductivity of both stage 0 and stage
I samples (vide infra). The 23Na MAS NMR spectra show that the sodium in all
stage 0 Na—K-SG compositions is present as N a0, with very low concentrations of
N a+. This confirms that very little reaction occurs between the alkali metal alloys

and silica gel at ambient temperatures. When converted to stage I by heating (or

35

Table 3.1: Refined dual-phase structure of the Na-SG-II material.

 

 

Na23i03

a(A)= 10.43, b(A) = 6.11, c(A)= 4.93

 

 

 

Atom x y z U(A2)
Na(8b) 0.165 0.347 0.00 0.015
Si(4a) 0.00 0.118 0.481 0.068
01(4a) 0.000 0.113 0.811 0.041
02(80) 0.123 0.281 0.5 0.024
NaSi

a(A)= 12.72, b(A)= 7.25, c(A)= 10.56
Atom x y z U(A2)
Na1(8f) 0.292 0.732 0.353 0.009
Na2(8f) 0.678 0.0458 0.513 0.011
Si1(8f) 0.447 0.252 0.323 0.011
Si2(8f) 0.624 0.553 0.359 0.031

 

 

when Na—SG-I is formed by heating sodium metal with silica gel) some Na+ is formed,
but, in large-pore silica gel, a substantial fraction of Na0 remains. There are several
possible sources of N a+. Reaction of Na with defects, such as SiO‘ groups or residual
SiOH, would both decrease the amount of reducing material and produce sodium
cations. Collection of hydrogen from the reaction with water and comparison with
the masses of metal and SG used, show that this is a minor effect, amounting to not
more than 4 wt % metal. Ionization of Na0 near or in the walls of the cavities would
also produce Na+, accompanied either by electron donation to the SI02 lattice, or
formation of an “electron gas” in the void spaces, as occurred with Cs in an all-silica
zeolite. The driving force for such ionization is the strong interaction of the resulting
cations with the oxygens of the silica gel lattice. With large-pore SG, however, a
fully loaded sample could have only a fraction of the encapsulated alkali metal atoms

close enough to the walls for effective charge—balance. Ionization to form M+ and an

36

Table 3.2: 23N a MAS NMR data for Stage I Metal-Silica Gel samples

 

 

 

average N a0 N 8+ N a0
pore size wt% ratio line width line width chemshift

(A) metal Nao / N a+ (Hz) (Hz) (ppm)
150 10 0.084 7500 4500 1049
150 20 0.34 5700 6600 1092
1 50 30 1.02 4400 4200 1 1 19
150 40 1.40 2600 3200 1 154
60 35 0.58 4500 2200 1123
30 25 0.25 3400 3900 1120
150 40 1.41 9000 8800 1056

 

 

electron gas would be unfavorable if the cations could not be stabilized by the silica
lattice. Thus, except for ionized alkali metals in small pores or near the walls or
intercalated into the SG lattice, we anticipate that additional alkali metal would be
present as the neutral metal. Support for this picture and for ionization near the walls
is provided by the 23Na MAS NMR spectra of NagK-SG in silica gel with different
pore diameters and different metal loadings. The ratio of N a0 to N a+ is strikingly
dependent on both the loading and the average pore size of the silica gel used, as
shown in Figure 3.8 and Figure 3.9. This indicates the existence of Na nano clusters
inside the Na—SG-I sample.

Figures 3.8 and 3.9 show this dependence, and Table 3.2 gives the approximate
metal-to—cation ratios for various preparations, together with chemical shifts and line
widths.

In addition to the pronounced effect of loading and pore size on the NaO/Na+
ratio, the chemical shift of Na0 and the line widths of both Na0 and Na+ are strongly
affected. The chemical shift of Na+ is constant at 4:l:1 ppm, but that of Na0 varies
systematically with the Nao/Na+ ratio from 1049 to 1154 ppm (Table 3.2). The
line widths of both Na0 and Na+ are much broader than is common for either the
pure metal or for sodium salts. This could result from quadrupole broadening or a

distribution of environments, but the broadening of both lines and the systematic

37

 

."
o
I

Intensity (Arb.)

 

 

l
1 5 _ Metal Loading
' — 40%
— 30%
— 20%
_ 10%

 

0.0 Na”
l I
ma 1200 1100 1000 11in o .1011 .200
Chemical Shift (ppm)

Figure 3.8: 23Na MAS NMR spectra of NagK-SG-I in silica gel with an average pore
size of 150 A at various metal loadings. The weight percent loadings are from the
amounts of metal and SG used. The spectra are scaled and displaced for clarity.

 

   

1.4 - Average Pore Size
_150:

— m

‘42 ‘1 m A

0.2-

 

 

0.0 o

 
     

Na'

IIIIIIFTW
I

 

WWIIII‘IIIII III Jill-III

Figure 3.9: 23Na MAS NMR spectra of Na2K-SG—I in silica gel as a function of

average pore size. Loadings are 40, 35, and 25 wt % metal for 150, 60 and 30 A SG,
respectively. The spectra are scaled and displaced for clarity.

39

 

 

 

 

 

 

 

111 4 f‘
.... 0.8 —
13'
S.
Q as -
10
5
IE.

0.4 -

02 -

l I l l l
150 100 so u 511 .1.
Chemical Shift (ppm)

Figure 3.10: Ambient temperature 23N a MAS NMR spectra of Na—SG-II with average
pore size of 150 A. The two peaks on the left are attributed to Na4Si4, while that on
the right is from the product sodium silicate and initial Na+ formed by reaction of
Na with defects. This sample has no peak at the chemical shift of N a”.

chemical shift changes of the signal of N a0 suggest that distributed paramagnetism,
a common source of line-broadening, may be responsible. Exchange broadening by
interconversion of Na0 and N 8+ is unlikely, since the line widths are unchanged from
~60 °C to +60 °C. Attempts to obtain more information about the nature of stage I
M-SG materials from 2QSi MAS NMR studies were unsuccessful because of extremely
long T1 relaxation times, as has been observed for metal-zeolite systems [44]. Data
collection for 19 h yielded a signal-to—noise ratio of only 4 : 1. The chemical shift of
the single peak observed was at -83 ppm compared with -118 ppm for calcined silica
gel. As shown in Figure 3.10, conversion of stage I Na-SG to stage II by heating
to 400 ° yields a complex pattern that includes two peaks of Na+ in the region

found previously [56] for Na4Si4. The sodium silicate to sodium silicide ratio (5.5),

40

obtained by deconvolution, is close to that expected (4.5) on the basis of the reaction
stoichiometry and defect N a+. There is no peak in the chemical shift range of Nao.
Thus, the 23Na MAS N MR results are consistent with the conclusion from the PDF
studies that N 80 reacts completely with Si02 to form N84814 and Na2SiOg. Attempts
were made to confirm this with 2QSi MAS N MR studies, but the expected peak [56]
at -364 ppm from Si44“ was not observed. Only silicate (or silica) peaks around -100
ppm were observed above the noise. As was the case with Na-SG-I, the extremely

long 2QSi relaxation time made it difficult to achieve good spectra.

3.4.2 Differential scanning calorimetry (DSC)

DSC measurements [5], show that the conversion of stage 0 to stage I is slow and
exothermic, merging with the exothermic formation of stage II material above about
300 °C. The disappearance of the endotherm of melting upon formation of stage I
materials has been verified by many experiments with NarSG, Na2K-SG and K2Na—
SC. Although these samples contain alkali metals in the pores, they do not undergo
conventional melting transitions. The observation of melting for stage 0 samples
shows that mere confinement of metals in the pores does not eliminate the enthalpy of
melting. We have no current explanation for the elimination of the melting endotherm

when stage 0 M-SG is converted to stage I.

3.4.3 Magnetic susceptibilities and electrical conductivity

The magnetic susceptibilities of calcined silica gel, Na—SG-I, and K2Na-SG-I (150 A
average pore size) were measured from 5 K to 300 K and fit by the Curie-Weiss
equation, including a temperature-independent term. The silica gel sample had the
most pronounced “Curie Tail” with a paramagnetic content of 0.33 mole %, based
on SiOZ. Na-SG yielded 0.027 mole % and KgNa-SG had 0.066 mole %, based on

the moles of metal present. The temperature independent contribution to the molar

41

susceptibilities, after correction for the diamagnetism of the SG (obtained by an
exponential fit of the data above 100 K) was +14x10‘6 mol'1 for Na—SG-I and
+24><10‘6 for KgNa-SG-I. The molar susceptibility of sodium metal is 15.6x 10'6
mol‘l, while that of potassium is 21.5x10‘6 mol‘l. Thus, the susceptibility data are
compatible with the presence of metal in the pores. The small paramagnetism shows
that most of the electrons formed by ionization of the metal near or in the SG walls
are spin—paired, as was the case with cesium in an all-silica zeolite. Semi-quantitative
measurements of the DC electrical resistance of various M-SG samples showed that
samples with high NaO/Na+ ratios are very conducting, while those that show little
or no Na0 NMR peak are insulating. For example, for samples in SC with 150 A
pore size, the resistance of stage 0 Na2K-SG (40 wt%) was <3 ohms, while those of
stage I at this concentration as well as at the 25 wt% level were each < 200 ohms. In
contrast, the resistances of stage I samples of NagK-SG-I at the 20 wt% level and of
Na—SG-I (25 wt%) in SC with average pore size 30 A were each > 2x107 ohms. The
blank resistance of the lead wires and the empty cell was 1.2 ohms. These results,
while not quantitative, indicate that inter-particle conductivity occurs by formation of
conducting paths between the metal-containing pores, and that samples are insulating

when complete or nearly complete ionization of the metal occurs.

3.5 Discussion

A plausible picture that emerges from all of the data for stages 0 and I alkali metal-
silica gel materials is that the liquid alloys go into the void spaces at room temperature
as neutral metals, without significant ionization. DSC scans of stage 0 powders Show
the expected low temperature melting endotherms, the 23Na MAS NMR spectra show
only Na“, and the electrical conductivity is very high. Heating the stage 0 materials
(or Na with SG) results in ionization of some of the alkali metal to produce M+ at or in

the silica gel walls. The cations are stabilized by interaction with the electronegative

42

oxygens of the silica gel, but the nature of the released electrons is open to question.
Since the reducing ability (amount of hydrogen formed by reaction with water) is
unchanged from that of stage 0, the electrons released by ionization of the included
metal are still available for chemical reduction. Indeed, the reduction potential is
essentially the same as that of alkali metals. Samples whose Na N MR spectra show
only Na+ and the absence of N80 are still able to reduce biphenyl to the radical
anion. Most of the released electrons are spin-paired, as is the case with zeolite-
based ”inorganic electrides”. An intriguing possibility is the formation of a “skin”
of electrons on the surface of the pores, similar to the trapping of electrons in the
cavities of organic electrides. Organic electrides that have large channels between the
cavities show extensive electron spin-pairing. An alternative fate of the electrons is
attachment to the 8102 lattice, perhaps to form localized “lone pairs” by expansion
of the coordination number of Si from four to five, or by occupation of the conduction
band of the silica gel. When all of the available sites for M+ are occupied, additional
metal remains in the pores as the neutral metal. The broad NMR lines (temperature
independent) and the variation of the line width and chemical shift of the N80 NMR
peak with loading and pore size indicate that the neutral metal species is influenced
by its location relative to the ionic shell. Silica gels loaded with alkali metals produce
hydrogen when water is added. Stage I samples retain the reducing ability, but not
the pyrophoric character of the parent metals, and are thus useful in synthesis.

The structures of these materials and the chemical nature of the intercalated
species were previously not well defined and could not be determined by traditional
crystallographic techniques due to the diffuse nature of their X—ray diffraction pat-
terns. Here we have determined the structures of these materials by using the atomic
PDF local structure probe, which considers equally all components of the scattering
signal.

Comparison of the experimental PDFs with those of pure silica gel indicates that

43

amorphous silica is present in N agK-SG-I and Na—SG-I, with additional features com-
ing from the alkali metals. Broad oscillations in the PDF show that alkali metal
nanocrystalline aggregates are present. This result confirms the presence of metallic
sodium in the pores of this material despite the lack of a melting isotherm from dif-
ferential scanning calorimetry (DSC) measurements. The PDF result was confirmed
by the results of the NMR measurements, which, as we discussed earlier, confirmed
the presence of the alkali metals inside the pores of the silica gel. One possibility for
this absence of the melting endotherm is that the silica pores suppress the metals
from melting. This possibility will be a subject of a future high temperature PDF
study.

The structure of the Na-SG-II was determined and indicated that Na-SG—II is a
dual-phase sample consisting of N a4Si4 and Na28103, in accordance with the proposed
chemical reaction between Na and 8102. This result motivated the sample makers
to test the use of Na4Si4 instead of the Na or Na2K alloy in silica gel, where it was
found that it can provide at least 90 grams of hydrogen per kilogram of fuel [57].
Also determining the other reaction product in the stage II materials (Nagle3),
may help in developing methods to remove or convert them to other forms that
can produce hydrogen that may increase the efficiency of the stage II materials in

producing hydrogen.

44

Chapter 4

Mercury Binding Sites in
Thiol—Functionalized

Mesostructured Silica

4. 1 Introduction

Mercury, a toxin known to cause neurological impairment in humans, is of great envi-
ronmental concern. One promising technique for achieving the removal of low levels
of mercuric ions from groundwater is to trap them using complexing ligands (e.g.,
thiols) that are covalently linked to a high surface area support [10]. Accordingly,
various forms of thiol-functionalized mesostructured silica have been examined as can-
didates for mercury remediation [13, 14, 15, 58]. These trapping agents offer very high
surface areas, well-defined pore sizes, and high thiol group loadings with up to 50%
of the framework silicon centers being functionalized [11, 12]. Linking the organic
moiety to the mesostructure can be accomplished either by grafting a thiol-functional
silane reagent to the surface silanol groups of the mesostructure [13, 14, 15] or by the
direct incorporation of the organosilyl group into the framework walls during synthe—

sis [11, 12, 16]. The direct assembly pathway is generally preferred over grafting, in

45

part, because the trapping agent can be prepared in a one-pot process, as opposed to
the multiple processing steps involved in grafting. Also, the direct assembly process
provides a higher concentration and a more uniform distribution of organo groups on
the surface of the support [16].

Recently, an XAS study [59] was reported for mercury bound to the thiol groups
of a mercaptopropyl—functionalized silica mesostructure with a wormhole framework
structure, denoted MP-HMS. The EXAFS data indicated the presence of Hg—S bonds,
as expected, as well as an equivalent population of Hg—O bonds. These findings
favored a predominance of S-Hg-O over S-Hg—S linkages for Hg/ S ratios S 1.0A. A
shortcoming of EXAFS is that structural information falls off quickly with increasing
distance from the probe element, and it is difficult to learn about the positions of
atoms beyond the nearest neighbor coordination shell. This limitation is especially
acute for Hg L3 spectra [60, 61]. In the earlier XAS work [59] it was not possible
to differentiate between localized and extended networks or to establish the precise
nature of the Hg binding. For thiol-functionalized mesostructures in which 30% to
50%, of the silicon centers are functionalized [14, 62] and available for binding to
mercury, it is plausible that extended -Hg—S- chains could form within the pores
through the bridging of thiolate ligands to two mercury centers. Such chain structures
are found for the HgS structure, as illustrated in Fig. 4.1. The presence of bridging
thiolate centers would be verified by observing a mercury-mercury contact similar
to the intrachain Hg—Hg distance in HgS. If extended chain structures formed, it
would be possible to observe interchain Hg-Hg distances. However, the limitations
of the XAS technique, particularly when applied to mercury, preclude the detection
of chain structures. For that we used the atomic pair distribution function (PDF)
which provides information complementary to EXAFS. Since the PDF technique does
not presume periodicity it can be applied to disordered materials. It yields high

quality information about structure on short and intermediate ranges and is a useful

46

 

Figure 4.1: Cinnabar (HgS) structure (yellow = S, gray = Hg).

complement to the EXAFS study. Here we applied the PDF method to study a
series of thiol-functionalized mesoporous silicas with different mercaptopropyl content
and Hg loading. Two types of wormhole framework structures with the anhydrous
compositions (Si02)1_$(LSi01_5)I, where L is the mercaptopropyl group, are prepared
by direct assembly methods. The MP—HMS series of trapping agents made use of
tetraethyl orthosilicate (TEOS) as the source of SiO4 units [13, 11]. The MPMSU— SA
compositions are made from sodium silicate as the $0.; source [12]. For both families
of materials, mercaptopropyltrimethoxysilane was the precursor for the LSiO3 units
in the framework and dodecylamine is the structure-directing porogen. We find clear
evidence for the presence of bridging sulfur centers on the pore surfaces, analogous to

the bridging sulfur centers in Cinnabar and mercuric alkylthiolates. Also, in contrast

47

to the XAS result for these trapping agents, [59] we see no evidence for Hg-O bonds
and can certainly rule out oxygen coordination of the mercury at greater than the
10% level. A revised model is proposed for Hg binding in which S binds to two Hg
neighbors and each mercury is bound to at least two sulfur centers, regardless of the

mercury loading on the surface.

4.2 Sample preparation

Using tetraethyl orthosilicate (TEOS) and sodium silicate as the silica source were
used to prepare mesostructured compositions with wormhole framework structures
and anhydrous compositions of (SiOg)1_x(LSiOl,5)m, where L is mercaptopropyl and a:
is the fraction of framework silicon centers that have been fimctionalized. Mesostruc—
tures assembled from tetraethyl orthosilicate (TEOS) were denoted MP-HMS, and
those prepared from sodium silicate were denoted MP-MSU—SA. For both classes of
materials, dodecylamine (DDA) was used as the structure-directing porogen and mer-
captopropyltrimethoxysilane (MPTMS) was the source of the organosilicon centers.
The MP-HMS compositions were assembled at 65 °C in accord with a previously
reported method [11], except that the order of addition of reagents differed, with
water being the last reagent added to the reaction mixture at 65 °C. The mixture
then was aged in a reciprocal water bath shaker at 65 °C for 36 h. Finally, the so-
lution was filtered and the precipitate collected, air-dried, and subjected to Soxhlet
extraction with ethanol to remove the surfactant. The overall molar stoichiometry
employed was (1 —— 2:) TEOS::r MPTMS:0.22 DDA:6.7 EtOH:160 H20. The synthesis
of MP-MSU-SA materials from sodium silicate was carried out at 45 °C by adding
the sodium silicate reagent to a mixture containing MPTMS, surfactant, ethanol, and
an amount of glacial acetic acid equivalent to the formal sodium hydroxide content
of the sodium silicate [12]. After a reaction time of 20 h, the surfactant was ex-

tracted from the mesostructure with hot ethanol. The overall reaction stoichiometry

48

was (1 — as) SiOg:0.80(1 — 2:) NaOH:0.252 DDAza: MPTMS:0.80(1 —- 2:) acetic acid:3.4
EtOH:134 + 7.9(1 —— as) H20. Nitrogen adsorption measurements indicated the pore
size of the micelle-templated mesostructures to be in the small mesopore to large

micropore range (2.2 -1.5 nm), in accord with previously reported findings.

4.2.1 Mercury Adsorption

A 500 mg quantity of (SiOg)1_$(LSiOl_5)x mesostructure with a: = 0.30 or 0.50 was
added to various volumes of a 1000 ppm Hg(NOg)2 to achieve specific Hg2+ / SH ratios.
The suspensions were agitated for 48(:l:3) h, and the filtrate was analyzed by colori-
metric assay using diphenylthiocarbazone as an indicator. The amount of mercury
adsorbed by the mesostructure was determined by difference. Within experimental

uncertainty, the uptake of mercury was quantitative up to a total Hg2+ / SH ratio of

1.00.

4.2.2 PDF experiments

Powder samples were packed in flat plates with 1 mm thickness that were sealed
with Kapton tape. The X—ray scattering experiments were conducted on the powder
samples using X-rays of energy 76 keV (A = 0.16248A) at the 6-IDD station of
MUCAT at the Advanced Photon Source (APS) at Argonne National Laboratory.
Diffraction data were collected using the recently developed rapid acquisition pair
distribution function (RA-PDF) technique that are described in Section 2.4.1 The
data were collected using a circular image plate camera (mar345) with diameter of
345 mm, which was mounted orthogonal to the beam path. The sample holder to
detector distance was 161.94 mm. Different collection times were used to obtain each
data set. For the HgO and HgS reference compounds, five scans with a total collection
time of about 4 min were needed. For the MP-HMS and MP-MSU-SA samples five to

six scans were necessary with a longer collection time than the previous two standards.

49

The collection time lasted from half an hour to an hour to get high-quality results
for these mesostructured samples. The transformation of F(Q) to G (r) was carried
out with a Qmw 30.0 A for both HgO and the HgS samples. For the MP—HMS and
MP-MSU-SA samples the counting statistics were not as good as for the HgS and
HgO, which forced us to truncate our F(Q) at QM 16.0 A for all the mesoporous
silicas. Data from a MP-HMS sample with :1: = 0.50 with no adsorbed mercury
were also collected, this time at the BESSRC—CAT 11-ID-C beam line at the APS.
The sample was sealed in a capillary and measured with X—rays of energy 114.67 keV
(A = 0.108 A). Scattered radiation was collected with an intrinsic germanium detector
coupled to a multichannel analyzer. Several runs were conducted and the resulting
XRD patterns were averaged to improve the statistical accuracy and to reduce any
systematic effect due to instabilities in the experimental setup. In this case the data

processing was done with the help of the program RAD [63].

4.3 Results and discussion

Representative mercaptopropyl—functionalized mesostructures with 3D wormhole frame-
work structures and anhydrous compositions of (SiOg)1_$(LSiOl,5)x, where L is mer-
captopropyl, were prepared and equilibrated at ambient temperature with known
amounts of aqueous mercuric nitrate. At the initial ambient pH of the mercuric ion
solution (2.3) the predominant solution species is Hg2+(aq) and the minor species
is Hg(OH)+ [64, 21]. At equilibrium, the final pH of the solution was near neutral,
so that any mercury remaining in the solution was mainly in the form of Hg(OH)2.
For the MP-HMS mesostructures assembled from tetraethyl orthosilicate, the frac-
tion of framework silicon centers containing a covalently bound thiol group was held
constant at :1: = 0.50 and the bound Hg2+ to total S ratio was varied from 0.50 to
1.30. The compositions assembled from sodium silicate, denoted MPMSU- SA, had

:1: = 0.30 or 0.50 and a bound Hg2+ to total S ratio of 0.91. The compositions are

50

Table 4.1: Mercury Content of (Si02)1_$(LSiOl,5)x Mesostructures for PDF Analysis,
L = Mercaptopropyl

sample mesostructure :1: mmol Hg2+ / g mmol (Hg2+/ SH)

 

 

1 MP-HMS 0.5 2.52 0.5
2 MP-HMS 0.5 4.85 0.96
3 MP-HMS 0.5 6.33 1.30
4 MP-MSU—SA 0.3 2.99 0.92
5 MP—MSU-SA 0.5 4.16 0.91
g I ' r ' I ' I I I l I I I l I ' I l
1 MP—Hus
’
N
92 2 MP-l-IMS
N9. 3 MP-HM
'1' N
d '-
Va, 4 MP-MSU-SA
L A? Aw
A, 5 MP-MSU-SA
n 5102
O

I A I A I A I A I A I A
0 8 10 12 14 16
o (1")

Figure 4.2: (8) Experimental F(Q) for the mercury-loaded compositions of Table. 4.1.
Included for comparison are analogous data for bulk silica glass, labeled 8102, which
does not contain mercury. The curves are offset for clarity.

I l l

I I
2 4

summarized in table 4.1. The structure functions for the Hg-loaded mesostructures
are shown in Fig. 4.2, and the resulting PDFs, C(r), are given in Fig. 4.3. The curves
are offset from each other for clarity. Features in F(Q) are broad, indicating the
disordered nature of these materials, and this is reflected in G (7‘) as a rapid falloff
in structural features with increasing r. Also shown for comparison is the F(Q) and

PDF of mercury-free bulk silica glass. The curves for the Hg-loaded mesostructures

51

 

‘I'I'I'I'I'I'I'I'u

1 tip-Hus;

‘

 

__2 MP—I-ms-

 

 

 

 

3 6 91215182124

6‘ 3 MID-Hus:
1‘ 4,. A...
I; A 4 334150-511

5 uP-usu-stj

‘ $102 '.
o A. A. v

 

 

 

Figure 4.3: (a) Experimental C(r) for the mercury-loaded compositions of Table. 4.1.
Included for comparison are analogous data for bulk silica glass, labeled $102, which
does not contain mercury. The curves are offset for clarity.

are similar in shape, but quite distinct from the bulk silica glass. The presence of
the strongly scattering Hg center is clearly evident. In particular, strong peaks in
G (r) at 2.4 and 3.7 are evident in the Hg-loaded samples that are not present in the
silica. The peak at 1.6, which is present in all of the samples, originates from the
Si—O distance in the silica network, but the 2.4 and 3.7 peaks arise from the pairing
of mercury with atomic neighbors. As indicated by the definition of C(r) (see 2.3),
the intensity of a peak in the PDF coming from the pair of atoms 1' and j is scaled by
n,Z,-nj ZJ- / (Z)2, where Z,- is the atomic number and n,- the coordination number of the
ith ion and (Z) is the compositionally averaged atomic number of the entire sample.
Thus, as the Hg loading increases, increases and the Si-O peaks at r = 1.6 change in
magnitude, even though they have the same origin in all the samples. The PDFs were
rescaled in such a way that the Si-O peak in each sample has the same integrated

intensity. This was accomplished not by average scattering intensity, but rather by

52

rescaling of the silica volume using the assumption that the mesostructures have sim-
ilar morphologies and approximately the same sample volume. These are the same
G(r)’s shown in Figure 4.3, except they have been rescaled so the Si—O silica peaks
at r = 1.6 have the same integrated area. Included for comparison is the PDF of
the MP-HMS trapping agent (a: = 0.50) without bound mercury. The rescaled PDF
profiles and the assignment of peaks in the profile are shown in Fig. 4.4. Referring to
Table. 4.1, we see that samples 1 and 4 have similar Hg loadings and have the least
adsorbed Hg, samples 2 and 5 have 50% more adsorbed Hg and are similar to each
other in total Hg loading, and sample 3 has the highest Hg loading. It is evident from
Fig. 4.4 that the peaks at 2.4 and 3.7 follow this trend and arise principally from
pairs of atoms involving Hg, more specifically, Hg-S and Hg—Hg pairs, respectively, as
discussed more fully below. In the low-loaded samples 1 and 4 the 2.4 and 3.7 peaks
are weak. They grow in the medium loadings of 2 and 5, and they dominate the silica
peaks in the heavily mercury loaded 3 composition. The broad bumps centered near
5.75 and 6.25 most likely originate from Hg pairs beyond the second nearest neighbor,
but these Hg pairs are quite disordered. To verify the peak assignments in the PDF
profiles of the mercury-loaded mesostructures, we consider the isostructural reference
compounds HgS [2] and HgO [1]. These show representative bonding environments
of Hg“ with S and O, respectively. The Hg prefers to have two neighbors arranged
linearly with Hg-S bonds of 2.36 A. The Hg—S-Hg bond angle is 104.73°, resulting
in 1D zigzag -Hg-S-Hg-S- chains [2], as represented in Fig. 4.1. The situation is the
same in HgO except that ng_0 = 2.02 A and the Hg-O—Hg bond angle is 109.8° [1].
For comparison, we have measured the PDFs of these structures. They are shown
in Fig. 4.5 with the average crystal structure refined to the data using the real-space
profile-fitting program PDFFIT [65] plotted on top. The experimental and calcu-
lated PDF profiles agree well, indicating that the RAPDF method clearly results in

an accurate PDF. The refined values of the structural parameters from the fits of the

53

 

 

 

93 Hg-l-OHg—S l-lg-Hg _
I : -
. | .
S'”-°| 3 MP—HMS
9 .
2'. ,
5 ' f 2 MP-HMS -
”‘1‘ 3’1 l £2: 5 MP-MSU-SA
do 57.77 : 2V]; 4»- a. ‘ ‘
(D

' 5 5 3 1 MP-HMS ~—___27
, ' 5 j 4 MP-MSU-SA

   
 
     

 

 

(A)

Figure 4.4: Rescaled PDF profiles for the Hg—loaded samples

 

I
II

'l'l'l IJII
234 6789

d—

I
5
r

54

 

   
 
  

 

 

 

 

 

 

 

' I I I l I.' I ' I r I I I I l I
- (a)‘
”indigo-853111] '
or“ .1:
«It 2
V I
(9° » :
Tim MVM;.V vv _ -Av
'I L LA I A'I A I A I A I A I A I A
l I l I I I:': I ' I I I I '(bi
”h I I -
- (1mm); mm.) -
17"“ : :
so- . 5
o 'j‘Mw : :
*_ I I _
I .--v.v. “WA; -_A' _.
'I A I A I A'I' A I A I A I A I A I A‘

 

Figure 4.5: Experimental PDF (blue circles) and the PDF profiles calculated from
the refined crystal structure (red line) for (a) HgO and (b) HgS reference compounds.
The difference curves are represented by the offset green lines.

Table 4.2: Lattice parameters and fractional coordinates for selected atomic positions
reported in the literature and the refined values obtained from the experimental PDF
profiles for HgO and HgS

HgO HgS
literature refined literature refined
a 3.311 3.3013 a = b 4.149 4.1289
b 5.526 5.5146 c 9.495 9.4591
c 3.526 3.5163 Hg(r/a) 0.72 0.719
0(z/c) 0.17 0.157 S(r/a) 0.485 0.4913

experimental data are summarized in Table. 4.2, along with the literature values. An
earlier XAS study [59] indicated the presence of a similar number of Hg-O as Hg-S

bonds for Hg—loaded mesostructures equivalent to those studied in the present work.

55

We thus seek a peak in the PDF of our samples at 2.02 that might originate from
Hg-O. In the PDF of crystalline HgO, shown in Figure 4, the Hg—O peak is somewhat
weak due to the weak scattering power of oxygen, but it is clearly apparent. However,
there is no evidence for this peak in the data from the Hg—loaded mesoporous silicas
(cf., Fig. 4.3 and Fig. 4.4). Oxygen has half the number of electrons and, therefore,
half the scattering power of sulfur. Therefore, if there were an equal number of Hg-S
and Hg—O bonds, as suggested from the earlier XAS study, we would expect a Hg—O
peak at 2.02 A with an intensity roughly half as large as the Hg-S peak at 2.36 A. In
fact, rather than a peak, a deep valley is evident at the expected Hg—O bond distance
of 2.0 A for all of the mesostructured samples (of, Fig. 4.3 and Fig. 4.6).

It is worth noting that the previously reported XAS studies of Hg binding to MP-
HMS [59] mesostructures revealed the presence of only one significant RDF peak due
to near-neighbor contacts. Well-crystallized Cinnabar and meta-Cinnabar also exhib—
ited a single RDF peak corresponding to the inner coordination sphere of mercury
only. The lack of outer sphere coordination information generally is unexpected for an
XAS structure analysis. But in the case of mercury this result is not unusual, because
several other reports of Hg L3 spectra for HgO, HgS, and other mercury-containing
compounds [60, 61] failed to provide next nearest neighbor distances for reasons that
are yet unclear. The earlier XAS study [59] indicated that 50% of mercury neighbors
were oxygen. However, there is little or no evidence for a Hg-O peak at the expected
position of r 2.02 A in Fig. 4.4. We would like to quantify the maximum number
of Hg—O neighbors consistent with our PDF data. The PDF that would be obtained
if mercury were bonded to oxygen and sulfur with 50 : 50 probability as indicated
in the XAS result [59] was simulated. The results are shown in Fig. 4.6c. Mercury
coordinated to oxygen at the 50% level is clearly inconsistent with the PDF data. To
investigate this further, we simulated PDFs with 10% (Fig. 4.6b) and 0% (Fig. 4.6a)

Hg—O neighbors. The best agreement is the model with no Hg—O bonds present, al-

56

 

"jiai '

 

0 (A?)

n
I

 

 

AIAIA AIAIAI Illllli

1.5 2 2.5 1.5 2 2.5 1.5 2 2.5
r (l) r (I) r (1)

Figure 4.6: Calculated PDF profiles (red lines) assuming that (a) 0%, (b) 10%, and
(c) 50% of the mercury centers in MP—HMS sample 3 (:1: = 0.50; Hg/S = 1.30)
participated in Hg—O bond formation and the remaining mercury atoms bind to sulfur.
The experimentally observed PDF profile for this mesostructure is shown by the blue
circles. The difference curves are shown underneath in green. The Hg-O bond, if
present, is expected to give rise to a peak at 2.02. The peaks at 1.6 and 2.36 are Si-O
and Hg—S pairs, respectively.

though we cannot rule out the presence of Hg-O bonds below the ~ 10% level. We
next consider the Hg—Hg second-neighbor peak at 3.7. In composition 1 with a low
Hg2+ loading, this peak is double- valued, as it overlaps peaks in the same region due
to the Si—Si and O-O pairs from the silica framework (see the PDF profiles for the
mercury-free bulk silica glass in Fig. 4.3 and the MP-HMS mesostructure in Fig. 4.4).
In the more heavily loaded compositions 2 and 3, the Hg-Hg peak dominates and
it is clear that it is peaked at 3.7. The high-r shoulder evident in the low-loaded
samples is coming from the silica. The fact that this peak is strong, rather sharp,
and well defined indicates that there are a significant number of Hg-Hg contacts in
the pores and that they have a rather well-defined separation. This is expected if the
thiolate sulfur atom bridges two mercury centers to form Hg—S-Hg linkages. Binding

of thiolate sulfur to a single mercury center would lead to an irregular distribution

57

Table 4.3: Atom Pair Distances for the HgS [1] and HgO [2] from the Models Reported
in the Literature

 

 

atom pair distance (A)
Hg—S (intrachain) 2.36
Hg—O (intrachain) 2.02
Hg—Hg (Hg-S intrachain) 3.75
Hg—Hg (Hg-S interchain) 4.14
Hg—Hg (Hg-O intrachain) 3.31
Hg—Hg (Hg-O interchain) 3.67

 

 

of Hg and result in a rather broad distribution of Hg-Hg distances, in contrast to
the well-defined peak that is observed. Bridging thiolate ligands, however, provide
a well—defined Hg— Hg separation, because of the well-defined bond lengths and an-
gles associated with covalent bonding. The relevant distances are available from the
reference structures (Table. 4.3) and Fig. 4.5). The intrachain Hg-Hg contacts are
3.75 A in HgS and 4.14 A between the chains (of, Fig. 4.1). The 3.7 peak in the
PDF corresponds closely to the expected Hg-Hg distance for a Hg-S—Hg linkage.

In agreement with the absence of Hg-O bonds, we also do not observe a strong
Hg—Hg peak at 3.31 that would correspond to Hg—O-Hg linkages. Nor is there evidence
of a strong interchain Hg—Hg peak at 4.14 that would correspond to zigzag chains of
-Hg—S-Hg- packed alongside each other, as in the Cinnabar structure. Because the
thiolate sulfur centers are tethered to flexible hydrocarbon chains and encased inside
nanoscale pores, it is not surprising that the surface -Hg—S—Hg- units do not order on
the pore surfaces to give an interchain Hg-Hg contact.

The results of the PDF measurements indicate that no more than 10% of the bonds
to mercury are in the form of Hg—O bonds. This means that the binding of mercury
is almost exclusively to the sulfur atoms of the mecaptopropyl group (MP). Raman
spectra provided by the Pinnavaia group [48], allow for the assignment of at least
two binding modes for mercury(II) cations to (SiOg)1_z(LSiOl,5)x mesostructures,

where L is a mercaptopropyl group. Under all mercury loading conditions (ie., Hg / S

58

Figure 4.7: Suggested structure for at low mercury loadings.

= 0.50 — 1.30) the thiolate groups bridge at least two mercury centers, leading to
the a Hg-Hg contact in the PDF of 3.7A. At low mercury loadings, ie. Hg/SS
0.5, the predominant surface species is electrically neutral and not associated with
a counter anion, as evidenced by the presence of little or no nitrate ion by Raman
spectroscopy. For mercury binding under these conditions, the presence of polymeric
HgL2 compositions in which the mercury adopts tetrahedral coordination to bridging
thiolates as illustrated in Figure 4.7. This structure is analogous to the bonding mode
of mercury in Hg(SBut)2.

As the mercury loading is increased to a maximum value of 1.3, the predominant
binding mode becomes cationic. The shift from an electrically neutral Hg(SR)2 com-
plex to a cationic complex with increasing Hg loading is supported by the increase in
the free nitrate ion in the Raman spectrum. The cationic species also is associated
with the presence of bridging thiolate ligands, as supported by the continued presence
of a Hg-Hg contact in the PDF. As shown in Figure 4.8, the predominant formation
of a polymeric Hg(SR)+species at high mercury loadings in which the mercury cen-
ters adopt a linear two—fold coordination to bridging thiolate ligands. This type of
structure accounts for a Hg/ S stoichiometry of 1.0. Because the observed Hg/ S ra-
tio can be as high as 1.3, it is possible that some of the thiolate ligands bridge to

three mercury centers. Under no conditions do we find PDF evidence for binding of

59

 

 

A Hg T
RS/‘L "H3 SR

Figure 4.8: Suggested structure for at high mercury loading

mercury to oxygen centers.

60

Chapter 5

In—situ study of the boehmite to
gamma-alumina phase transition
using the atomic pair distribution

function technique

5.1 Introduction

The dehydration of aluminum hydroxides produces a series of metastable transition
aluminas, depending on the properties of the starting material, before transforming
to corundum (Oz-alumina), the stable form of aluminum oxide. For example, the

dehydration of boehmite yields the following sequence [17]:
boehmite—r 7 —> 5 —> 6’ —+ a

The temperature at which these phase transitions occur depends on the crystallinity
and previous history of the boehmite and on the heating time [17]. Due to their
important applications the transition aluminas have been the focus of many stud-

ies [66, 67, 68, 69, 70].

61

The precursor boehmite has a well known orthorhombic layered structure with
lattice parameters a = 2.8681, b = 12.2336 and c = 3.6923. The layers are held
together by hydrogen bonds between hydroxyl groups. Within each layer the oxygen
forms an fcc lattice [71].

Among the different transition aluminas; gamma-alumina is the most widely used
as a catalyst and adsorbent in many chemical processes, such as purification of gas oil
fractions and control of automotive emissions [8, 72]. ’7-A1203 cannot be obtained as a
single crystal. Diffraction from 7 -A1203 samples has broad peaks suggesting disorder
and nanocrystallinity, and its detailed structure has long been debated. Historically
'7 -A1203 was thought to have a cubic structure with the Fd3m space group [73].
Li et al. [74] instead suggested the tetragonal I4/amd space group with the Al atoms
restricted to spinel positions and vacancies on octahedral sites. Based on extensive
computational study Paglia et al. [75] proposed a modified tetragonal model where
they incorporated the A1 into the Wyckoff 80 position (the so-called c—symmetry
model). They performed Rietveld refinement on the diffraction data to compare the
c—symmetry model with other known models used to describe the structure of 'y-
A1203. The results of these refinements indicate the asymmetry model as the perfect
model to describe the average structure of ’7-A1203. The c-symmetry model has the
I 4/ amd space group, withe lattice parameters a = 55.652(1) and c = 57.871(5).

Most of the studies concerned with investigating the structure of ’Y-A1203 applied
traditional crystallographic structure methods [76, 70]. These methods consider only
the Bragg scattering and neglect any diffuse scattering. As a result they yield infor-
mation only about the long range order. To be able to study both of the short range
order and the average structure, a method that considers both kinds of scattering is
needed. This is possible using the atomic pair distribution function (PDF) [21] anal-
ysis of x-ray or neutron powder diffraction. The PDF gives the probability of finding

pairs of atoms separated by a distance r. The advantage of using this technique is that

62

it considers the total scattering of the material; the Bragg and diffuse components.
The PDF method has been widely used to study liquid and glasses [22] and more
recently it has been applied successfully to study nanostructured materials [4, 21].

A recent study [77] applied the PDF to study the structure of ’7-A1203. Two
models were refined against the experimental data: the spinel-based model and the
c—symmetry model. Compared to the spinel model the c—symmetry model gave the
best agreement, but only for the high-r region. None of the models were able to give
a good fit to the low-r region below 8 A. This indicates the existence of a fine-scale
nanostructure with a domain size ~ 1 nm. A new model was constructed to describe
the local structure of ’Y-A1203 (it will be abbreviated here as the nano model). The
skeleton of this model consists of six Al-O layers that remain after the hydrogen
leaves the structure during formation of ’7-A1203 from boehmite. Then the model
was constructed by considering the migration of Al into site positions within the
inter-skeletal layers, with the oxygen moving to accommodate the cation migration.
The refinement of this model for r < 8 A yielded a very good agreement with the
experimental data. [7 7] The most novel feature of this model is that locally the oxygen
sublattice is modified from the average structure, with stacking faults in the 7-A1203
matrix.

As mentioned before, 7-A1203 may be obtained by heating boehmite. Under-
standing the local and average structure evolution of boehmite and the steps it passes
through before transforming to ’Y-A1203 may lead to more insight into the structure
of this important material and how it is formed. Here we have carried out an in situ
high temperature structural study of the formation of "Y-A1203 from boehmite. As
discussed, diffuse scattering plays a major role in understanding the structure of ry—
A1203 and so we have collected data that can be analyzed using conventional Rietveld
refinement and PDF analysis. Applying the PDF to study such a transformation has

the advantage over other methods of providing both local and average structural in-

63

formation during the phase transition. Such an in situ study is now possible because

of the rapid acquisition data collection mode [23].

5.2 Experimental methods

Boehmite samples were obtained from the Alumina and Ceramics Laboratory, Malakoff
Industries, Arkansas, USA. The Powder sample was loaded in quartz capillaries with
2 mm diameter. The data were collected using the rapid acquisition pair distribution
function (RA-PDF) approach. [23] The samples were heated in a compact Debye-
scherrer geometry furnace built by Ames Lab [78] and available to users at the 6ID
beamline of the APS. The furnace has many properties that make it usable for syn-
chrotron experiments: It has compact size allowing for easy and quick mounting and
alignment on a standard four circle goniometer. The sample environment can be ad—
justed easily. The furnace can be heated up to 1200°C . The samples were inserted
into the core of the furnace positioned perpendicular to the incoming X-ray beam.
Samples are held at the end of a tube and aligned with an opening in the side of the
furnace. The X—ray beam enters through the opening and the diffracted rays emerge
through 3 mm wide slot allowing diffraction over a range to 20m = 90°.

The experiments were conducted using x—rays of energy 100.00 keV (A = 0.1235 A).
Data were collected using a circular image plate camera (Mar345) 345 mm in diame-
ter. The camera was mounted orthogonal to the beam path, with a sample to detector
distance of 203 mm, which was determined using the silicon standard.

In order to avoid saturation of the detector, each measurement was carried out
by multiple exposure to the x-rays. Each sample was exposed for 300 seconds at
a time, and to improve the statistics we took two scans for the measurements at
temperatures from 300°C to 380°C and four scans for the ones from 400—880°C. Steps
of 20°C were used for the temperatures from 300°C to 380°C and 540°C to 880°C

while 10°C steps were used for temperatures from 430°C to 530°C, where the main

64

features of the transformation were expected to happen, an expectation that was
confirmed during the experiment by the changes in the peaks of the X—ray diffraction
intensities (XRD). To ensure that all parts of the powder sample are in equilibrium
with the thermocouple temperature we waited two minutes at each temperature before
collecting the data. This resulted in thirty seven data sets. Data for each temperature
were combined and integrated using the program FIT2D [24].

Data from an empty capillary were also collected and subtracted from the raw
data. Other standard corrections were made, as described in detail by Egami and
Billinge [21], to obtain the total scattering structure function, 5' (Q), using the pro-
gram PDFgetX2 [25]. Finally, the PDF, which gives the probability of finding an atom
at a distance r away from another atom, was obtained by a Fourier transformation of
S (Q) according to G (r) = % f0°° Q[.S' (Q) - 1] sin(Qr) dQ, where Q is the magnitude of
the scattering vector. S (Q) was truncated at Qmw = 20 A, since beyond that value
of Q the signal-to—noise ratio was unfavorable.

To investigate the nature of the boehmite to 7 -A1203 transition we performed two
types of refinements, Rietveld and PDF analogous:

(1) Rietveld analysis which gives information about the average structure was per-
formed using the GSAS code [34]. A least-squares refinement between the calculated
and observed intensities is performed until the best match with the measured profile is
obtained. Rietveld refinement considers only the Bragg peaks, such that every other
component in the diffraction data including the diffuse scattering will be subtracted
as background. The background in Rietveld is obtained from the refinement and
there is no need to worry about performing background measurement, as opposed to
the PDF measurements.

(2) The PDF real space refinement was performed using the PDFfit2 code [36]. In this

method the model is defined in a small unit cell with atom positions specified in terms

of fractional coordinates. Again as it is in the Rietveld, a least-squares refinement

65

 

 

Figure 5.1: The powder diffraction intensities for boehmite heated to different tem-
peratures

between the calculated PDF from the structure and the measured PDF is performed
until the best match is achieved. Depending on the refinement range, this refinement
can yield information about both the local and average structure. For investigating
the average structure we restricted the range of the refinement to 8 < r < 20. For
the local structure we restricted it to 1.3 < r < 8, which are the same ranges used in
the previous PDF study of the 7-A1203 structure [77].

PDFfit2 and Rietveld refinements were performed. For the PDFfit2 refinements
thermal parameters, lattice parameters and phase fractions were refined. In addition
to these parameters, we refined the background and the peak profiles parameter for
the Rietveld refinements. Atoms were kept fixed at their ideal positions in both
Rietveld and PDFfit2 refinements.

5.3 Results

The complete set of x—ray powder difiraction patterns obtained at different temper-

atures are shown in Fig.5.1 with a selection of patterns are shown in Fig.5.2. The

66

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A I A l A I
0 4 8 12
20°

 

Figure 5.2: The powder diffraction intensities for boehmite heated for different tem-
peratures

evolution of the boehmite structure toward 7-Al2O3 is evident from the change in
the diffraction peaks. For example the amplitude of the peak around 20 = 1° (020)
starts to decrease with increasing temperature until it disappears around 470°C.

To investigate the boehmite to 7-A1203 transformation process we performed Ri-
etveld refinement with boehmite and c—symmetry as initial structures. Since the XRD
pattern measured at room temperature and up to 430°C did not show any signifi—
cant changes, the boehmite single phase refinement gave a very good agreement as
expected, as evident from the resulting pr’s that are summarized in Tab. 5.1. The
XRD pattern begins to change significantly above 430°C. Fig. 5.3 shows a compar-

ison between the single phase refinement results for the XRD measured at 460°C

67

Table 5.1: The Rm? values and the fraction of 7-A1203, f7, from the refinement of
the average structure both from Rietveld and PDFfit2 refinements. The models in
each case are single-phase boehmite, single-phase c-symmetry and a 2-phase mixture
of these two models.

 

 

 

T (°C) boehmite c-symm 2-phase f,
Rietveld
25 0.059 - - -
430 0.059 - - -
440 0.064 0.225 0.055 0.14
450 0.081 0.186 0.055 0.27
460 0.130 0.130 0.064 0.62
470 0.146 0.074 - -
PDFfit2
25 0.168 - - -
430 0.194 - 0.199 -
440 0.253 0.954 0.210 0.29
450 0.425 0.890 0.212 0.40
460 0.822 0.375 0.240 0.68
470 - 0.144 0.167 -

 

 

 

with boehmite and the c-symmetry 7-A1203 model as initial structures. Both refine-
ments yielded poor fits suggesting that the sample is neither pure 7-A1203 or pure
boehmite. It is evident in the Figure that the peak around 29 = 1° comes solely
from the boehmite structure, while the peak around 20 = 5° comes solely from the
7-A1203 structure. Since both peaks are present in the data, this suggests that both
the boehmite and 7-A1203 structures coexist at this temperature. To confirm this
possibility we performed a dual phase (boehmite and c-symmetry) refinement for the
temperatures from 430°C up to 480°C. The prs from these refinements are summa-
rized in Tab. 5.1. Although the dual phase refinement at 470°C yielded lower RM,
than that of the single phase refinements, the results of this refinement can not be
accepted, because they yielded negative thermal parameters for the boehmite. This
is opposed to the single phase c—symmetry refinement, which yielded thermal pa-
rameters within the accepted range for high temperature measurements. The same

argument applies for the 430°C where the dual phase refinement yielded negative

68

  
  

- «mug-1.1”": ..
. 30.60""

I (a. u.)
-2 51.5 -1 -0.5 0 0.5

 

01234567891011
20°

Figure 5.3: The result of the Rietveld single phase refinements with boehmite as initial
structure (a) and c—symmetry as initial structure (b) for the 460°C. Experimental
data (symbols), the calculated data (line) and the difference curves shown as offset
for clarity

thermal parameters for the c-symmetry model as opposed the boehmite single phase
refinement. For temperatures from 440°C up to 460°C the dual phase refinements
yielded lower Rum and thermal parameters value.

This shows that for temperatures from 440°C up to 460°C the dual phase refine-
ments yield a better fit than either the boehmite or the c-symmetry single phase
refinements. The dual phase refinement fits are shown in Fig. 5.4. For the 480°C
sample the c—symmetry single phase refinement yielded a better result than the dual

phase, suggesting the boehmite has completely transformed to 7-A1203 . The phase

69

 

I (o. u.)

-6.4-5.6-4.8 -4 -3.2-2.4-1 .6-O.8

 

 

 

0 2 4 8 81012
20°

Figure 5.4: The result of the Rietveld for the dual phase refinement with boehmite
and c—symmetry as initial structures for the temperatures: (8) 440°C, (b) 450°C, and
(c) 460°C. Experimental data (symbols), the calculated data (line) and the difference
curves shown as offset for clarity

fraction of 7-A1203 vs. temperature is shown in Fig. 5.6.

For comparison with the Rietveld refinement results we performed a real space
PDF refinement for the average structure by restricting the refinement range to the
high-r region from 8 < r < 20. representative fits are shown in Fig. 5.5 and the pr
values resulted from the refinements are listed in Table. 5.1. For phase fractions; the
PDFfit2 code provide values for what is called the scale factor for each phase, which
depends on the phase fraction as well as the scattering power of the phases that the

material is composed of. The phase fractions were calculated from the scale factors

70

8 (r2)

-18-15-12-9 -6 -3 O 3 6

 

9 121518 21 24 27
r (1)

Figure 5.5: The result of the PDFfit2 refinement for the dual phase refinement with
the boehmite and c—symmetry as initial structures for the: (a) 440°C (b) 450°C and
(0) 460°C. Experimental data (symbols), the calculated data (line) and the difference
curves shown as offset for clarity

by the method described by Lin et al [79]. The GSAS code gave directly what is
called the weighted fraction [34].

The real-space refinements are in qualitative agreement with the Rietveld refine-
ments, indicating a better fit from the dual phase refinement in the region 440°C-
460°C. In all cases a slightly large value is refined for the phase fraction of 7-A1203
from the PDF data.

Representative F(Q)s of the data are shown in Fig. 5.7. The good quality of

the data, up to the maximum Q we used of 20 A”, can be seen in the Figure.

71

 

400 420 440 480
T(°C)

Figure 5.6: The change of the ’7-A1203 phase fraction resulted from the PDFfit2
average structure (square), Rietveld refinement (circle) and PDFfit2 local structure
(triangle) refinements with temperature

72

21 28

14

r (1“)
7

-_14 —7 0

 

2 4 6 81012141618
0(1)

Figure 5.7: The reduced structure functions (F (Q)) for boehmite measured at differ-
ent heating temperatures: (8) 440°C (b) 450°C (0) 460°C and (d) 470°C

73

 

 

‘ '2.

3
‘ KP“

Figure 5.8: The PDFs for boehmite heated for different temperatures

Fig. 5.8 shows the PDFs of all the data obtained by heating the boehmite from room
temperature up to 880°C. The evolution of boehmite toward the 7-A1203 is evident
from the change in the PDF peaks especially the peaks around 1.9 A, 2.8 A and 3.5 A
which correspond to the Al—O, Al—Al and Al—O(H) bonds in the boehmite structure.

These changes in the PDF peaks can be understood from the differences between
the boehmite and 7-A1203 nano structure which are shown in Figure 5.9 and Fig-
ure 5.10 respectively. In boehmite all of the Al cations are octahedrally coordinated
as opposed to the fine scale model model in which Al cations have migrated to tetrahe-
dral and bridging positions. These migrations of the cations cause significant changes
in the local geometry are evident form the large changes in the PDF. the 7—A1203
to have a distribution of different Al—O bond lengths (1.4 A to 1.9 A), and the Al—Al
and Al-O(H) bond lengths number to increase and decrease respectively. It can be
seen that these changes do not begin until the temperature reaches 440°C where the
transformation process starts to develop. The changes in the PDFs increase with
increasing temperature until the temperature reaches 470°C, where above this tem-
perature no noticeable change can be seen in the PDFs, which indicates that the

boehmite to 7—Al203 transformation was completed at 480°C.

74

 

 

 

 

 

Figure 5.9: The boehmite structure. Aluminum (green), oxygen (red) and hydrogen
(white)

Table 5.2: Refinement result of the local structure
T (°C ) boehmite Nano 2-phase f,

25 0.173 - —
420 0.183 — 0.175 0.12
430 0.192 — 0.173 0.13

440 0.291 0.664 0.18 0.21
450 0.389 0.602 0.211 0.41
460 0.658 0.350 0.230 0.74
470 0.666 0.256 0.252 0.96
480 - 0.242 0.252 -

 

 

So far we have been concerned with the average structure (the Rietveld and PDF-
fit2 refinement in the range 8 < 7‘ < 28A). We are also interested in investigating the
local structure. Two models were used, the same boehmite model that we used in the
average structure refinement and the nano model that we have described before [77].
The result of these refinements are shown in Fig. 5.11 and Table 5.2. The results of

the local structure refinement results differers from that of the average refinements.

75

 

 

 

 

 

 

 

Figure 5.10: The nano model structure. Aluminum (green) and oxygen (red)

76

 

 

 

 

 

Figure 5.11: The result of the PDFfit2 refinement for the local structure with the
dual phase refinement; boehmite and c—symmetry as initial structures: (a) 420°C (b)
430°C , (0) 440°C , (d) 450°C , (e) 460°C and (f) 470°C. Experimental data (symbols),
the calculated data (line) and the difference curves shown as offset for clarity

Where the transformation toward 7-A1203 starts to take effect at 520°C as oppose
to the average structure where it seems to start at 430°C. This can bee seen from the
results of the refinements where the dual phase refinements yielded better results for
temperatures from 420°C up to 470°C , while the single phase refinements with ei-

ther of the models yielded bad results compared to that of the dual phase refinement,
Which it yielded a better results for the 400°C and 480°C PDF refinements.

77

5.4 Conclusion

The boehmite to 7 -A1203 transformation was studied by insitu experiment using the
RA-PDF technique. The XRD and the PDFs show that the transformation begins
at 440 °C and completes at 470 °C. Refinements were made for both the average and
local structure, using for that the XRD and the PDFs. The results of both refinement
the real space and the Rietveld showed that the transformation of boehmite proceeds
in inhomogeneous way where part of the sample transforms to 7 -A1203 while the rest
stays as boehmite. As the temperature increases the more of the boehmite transforms

and the ratio of the 7 -AlgO3 in the sample increases until the whole sample transforms

to ’7 -A1203.

78

Chapter 6

Concluding remarks

We used the atomic pair distribution function method (PDF) to study the structures
of different nanoporous and intercalant nanoporous materials. Traditional crystallo-
graphic methods -which assume periodicity— cannot be used for studying such mate-
rials because the intercalants are not periodically arranged [47]. The advantage of
using the PDF over other traditional crystallographic methods is that it considers
both kinds of scattering; the diffuse (short range order) and the Bragg (long range
order).

Data collection for all of the experiments in this thesis was performed using the
rapid acquisition PDF (RA-PDF) method [23]. This recently developed technique
uses area detectors which offer a large solid angle coverage and rapid data collection.
Collecting the data in a short time makes performing parametric studies accessible.
This was the other part of the thesis where we used the RA-PDF technique to perform
a high temperature experiment.

The first intercalant nanoporous material we studied was alkali metals and their
alloys doped into silica gel, which are convenient for chemical reduction of organic
compounds and the production of clean hydrogen [5]. Using the PDF we were able to
address the nature of the reducing species present in these amorphous materials. The

PDF results showed that liquid Na—K alloys added to silica gel at room temperature

79

(stage 0) or heated to 150 C (stage I) as well as stage I Na—SG, retain the overall
pattern of pure silica gel. Broad oscillations in the PDF show that added alkali metals
remain in the pores as nanoscale metal clusters. N a-SG-I when heated to 400°C (stage
II) yields a dual-phase reaction product that consists of Na4Si4 and Na28iOg. This
study was an example where the PDF combined with other methods can provide a
complete picture about the structure of the materials. For stage 0 and stage I the
23N a MAS NMR studies confirm the presence of N a0 and demonstrate that Na+ ions
are formed as well. The relative amounts of Na0 and Na+ depend on both the overall
metal loading and the average pore size. The results suggest that ionization occurs
near or in the 8102 walls, with neutral metal present in the larger cavities. The fate
of the electrons released by ionization is uncertain, but they may add to the silica gel
lattice, or form an “electride—like plasma” near the silica gel walls.

The second intercalant nanoporous material we studied was Mercury in thiol-
functionalized mesostructured silica. Thiol-functionalized mesostructured silica with
anhydrous compositions of (8102) 1.1,, (LSiOl,5),, where L is a mercaptopropyl group
and a: is the fraction of functionalized framework silicon centers, are effective trapping
agents for the removal of mercuric(II) ions from water. We investigated the mercury-
binding mechanism for representative thiol-functionalized mesostructures by atomic
pair distribution function (PDF) analysis of synchrotron X—ray powder diffraction
data. The mesostructures with wormhole framework structures and compositions
corresponding to :r = 0.30 and 0.50 were prepared by direct assembly methods in
the presence of a structure-directing amine porogen. PDF analysis of five mercury-
loaded compositions with Hg/ S ratios of 0.50-1.30 provided evidence for the bridging
of thiolate sulfur atoms to two metal ion centers and the formation of chain structures
on the pore surfaces. We find no evidence for Hg—O bonds and can rule out oxygen
coordination of the mercury at greater than the 10% level. The relative intensities

of the PDF peaks corresponding to Hg—S and Hg- Hg atomic pairs indicate that the

80

fist“.

 

mercury centers cluster on the functionalized surfaces by virtue of thiolate bridging,
regardless of the overall mercury loading.

The boehmite to 7 -A1203 transformation was studied by insitu experiment using
the RA-PDF technique. By measuring data at temperatures from room temperature
up to 880°C, both the XRD and the PDFs showed that the transformation starts
to develop at 440°C and is completed at 470°C. The results of both the real space
and the Rietveld refinement showed that the transformation of boehmite proceeds
in an inhomogeneous way where part of the sample transforms to 7-A1203 while
the rest stays as boehmite. As the temperature increases the more of the boehmite
transforms and the fraction of the ’7-A1203 in the sample increases until the whole
sample transforms to ’7-A1203. Some differences between the results from the local
and average structure were obtained; these will be investigated further and appear in

a publication.

6. 1 Future work

I intend to continue my research with studying structure of different nanoporous

materials using the PDF method. There are a few projects that need more work to

be finished, which include:

6.2 Nanoporous carbon

Due to their controllable narrow porosity distribution nanoporous carbon (NPC) has
many applications such, as fuel cells, super capacitors and in lithium ion batteries.
Nanoporous carbon may be formed by pyrolsis of polyfurfuryl alcohol (PFA) with
existence of some surfactant like Triton or THF. We would like to study the effect
of the different preparation processes on the structure of the NPC. This will include

studying the structure changes with different preparation methods, which includes:

81

lc

 

 

 

 

 

 

 

 

 

 

I ' I ' I ' I t f ' I '
-‘ 1190200 -
‘A A
1190300 .
-4 wv m' 1
1190330
4‘ 1190400 -
I V-‘T' A:_ L“— m
g , ‘4 g - 1190500
-11: A - 5130.090 ‘
1190700
I
l A -
l I
l-O'AII
O “
‘ l L l 1 I l . l . l .
0 3 12 15 18

1'0)

Figure 6.1: Experimental PDFs for NPC at different pyrolization temperatures.

(1) Different pyrolsis temperatures. (2) Also we want to investigate the effect of
using different surfactants (Triton and THF) for preparing the NPC. (3) The effect
of activating the NPC with 002. Figure 6.1, shows sample of the PDFs obtained
from these measurements so far. More data measurement is needed to complete this

study.

6.3 Trapping lead by mesostructured silica

Trapping heavy metals has been a big concern for chemists due to its environmental
applications, such as removing the contamination from drinking water. Similar to
what we did in the study for the mercury (see Chapter 4), we measured four sam-
ples prepared with two different surfactants TTeA and CTAB, two of the samples

loaded with lead and the other used as reference samples. We want to investigate

82

10

 

 

 

a .

o) 1

a -

g .
5““ (O) ‘
S'- I -
o 0
m ..

 

 

 

20 25 30 35
r(l)

Figure 6.2: Experimental PDFs for (a) loaded CTAB (b) loaded TTeA, (c) pure
CTAB and ((1) pure TTeA. Data were scaled with respect to the amplitude of the
Si-O peak (1.6A).

A I A

5 10‘

15

the structure of these materials using the PDF method. Figure 6.2 shows the PDFs
obtained for these samples. Primary results indicate the existence of lead clusters
in the loaded samples, which is indicated by the broad oscillations in their PDFS
that extend to more than 30A as opposed to the unloaded materials where the PDFs
peaks diminished around 15A as it is expected for amorphous materials. We intend

to study the structure of the lead inside the materials and try to find a model that

83

loc

h
dug”

 

 

describes the structure of lead clusters inside it.

84

 

lo

.' .' "0h.

 

1. 5191.197—

Appendix A

Parametric studies using the

RA-PDF technique

The properties of various materials are often related to structural changes such as
those that occur during phase transitions. In these cases in situ experiments are
preferred where a real time data collection can be performed to study the structure
evolution with external parameters such as time, pressure or temperature. The fast
data collection that has been facilitated by the RA-PDF method opened up a pos-
sibility of in situ, time resolved experiments, as complete series of data sets can be
collected rapidly. The results of such experiments are a series of PDFs as a ftmc-
tion of the external parameter. Here we describe the experimental steps of typical

parametric PDF study, using temperature as external parameter.

A.1 Low temperature RA-PDF experiment

Two methods can be used to lower the temperature in the RA-PDF experiment and
these are: ( 1) Gas f low: The flow of cold nitrogen or helium over the sample can be
used to lower its temperature. The advantage of using this method of cooling that

the temperature all over the sample is uniform which allows a shorter temperature

85

lo-

._ Ath

 

 

equilibration for all the data sets. (2) Closed cycle refrigerator (CCR): As the name
implies, closed cycle systems recirculate a fixed volume of gas to provide cooling to
the sample mounting stages, and no gas is added or removed during operation.
Typical low temperature experiment performed in RA-PDF mode was done using
the CCR method at the MUCAT station at the APS. For that the discussion here

will be mainly for describing this method of cooling.

A.1.1 Closed cycle refrigerator (CCR) cooling

A helium compressor provides high pressure gas to the cryocooler through a flexible
gas line. The expansion of the gas at different stages produces the refrigeration. Low
pressure gas is then returned through another gas line where it is recycled through the
compressor. This closed loop cycle does not require daily refilling with liquid nitrogen
and liquid helium - which is an expensive process - and can be continuously repeated
and maintained as needed to produce the desired refrigeration to a displex finger.
The cooled displex finger is in close contact with the sample (see Figure A.1), so that
the sample is cooled by heat conduction. For that it is important to chose suitable
sample holder for the low temperature experiments, which should have a good thermal
conductivity and to have the smallest possible size to increase the efficiency of the
cooling process. Fig A.2 shows a typical sample holders used in one of the recent
low temperature experiments at the APS. Another important issue that should be
considered when choosing the sample holder is that it should be safe to use for various
cooling parts around it. For example its designed should be such that it does not have
sharp edges that may destroy the Kapton cover or damage the beryllium window.
In general powder samples have poor conductivity, even when the material under
study is known to be a good thermal conductor. This is because in spite to whatever
one tries to make a good packing, there will always be gaps between the grains

of the sample acting as thermal insulation. As a result, the temperature of the

86

 

”lib-210' .

Figure A.1: The copper finger that cools the sample holder in the CCR refrigerator
that is available in the MUCAT station at the APS

87

1:811

100

 

 

 

Figure A.2: Sample holders for low T measurements that can be used in the MUCAT
station at the APS

powder sample will not be uniform throughout the sample. This makes determination
of the temperature of the powder sample challenging and proper attention has to
be exercised. The uncertainty in determining the temperature can be reduced by
increasing the time before collecting the data for every given temperature, to ensure
temperature equilibration, that all the parts of the sample have reached the same
temperature. The longer the waiting time the more uniform is the temperature across
the sample, but for some materials the phase change is time dependent; such that
undesired phase change may occur for extended waiting time. In general, before
going to the experiment the waiting time at each temperature should be studied and
decided carefully. The advantage of using the displex finger is that it can achieve
lower temperatures than the one that can be reached by using the gas flow. Because
of that, if the phase transition of the sample under study is not affected by the waiting
time needed to perform the measurement, the closed cycle method is preferred, where
a wider range of temperatures is available.

For increasing the efficiency of the cooling in low temperature measurements, the

sample must be shielded form the surrounding environment and performed under

88

 

lo

 

1J- Jfl ... 0.9V 4 1
A.—

    

.1,. ...» A - “l“ ,1
. _. x ‘ I, . ,., ‘ v .. _'«: “u n. 1:.“ 1 :A=l:.:
a .. .. '. 1 .,,.., -.2.23:.21184820515313.

Figure A3: The beryllium window that is used for shielding the sample holder in the
CCR refrigerator that is available at the MUCAT station at the APS.

     

vacuum. There are two ways to shield the sample from the outer environment at the
MUCAT beamline: either by using the Kapton displex (Figure AA and Figure A5) or
the beryllium window (Figure A3). The beryllium window has the advantage that
the diffracted data can be collected up to a wide range of angles, which is limited in the
case of the Kapton displex. On the other hand, the beryllium window is significantly
more expensive than the Kapton. One major problem of using the beryllium that
it yields a background with sharp peaks in contrast to the Kapton displex which
has a background with a single very broad peak and hence problems with the data
analysis. Having sharp peaks in the background is not favorable since any shift in

the peak positions (for any reason) will yield incorrect background subtraction. The

89

 

Figure A.4: The copper shielding used in the CCR refrigerator that is available at
the MUCAT station at the APS; Kapton is covering the windows.

effects of the incorrect subtraction vary depending on the sample measured. If the
scattering power of the sample is much higher than that of the beryllium window, then
subtracting the background will have very little or no effect on the data. Figure A.6
shows XRD data for ZnSeo_5Teo,5 measured at 15 K compared to that of the beryllium.
It is evident from this figure that the beryllium sharp peaks will have minimal effect
on the data quality. In general, the beryllium window can be used for shielding the
samples in low T measurements, if the scattering power of the sample is expected to

be much higher than that of the beryllium window.

90

 

10c

11 Van.

 

 

 

Figure A5: The Kapton displex used to obtain extra shielding of the copper used in
the CCR refrigerator that is available at the MUCAT station at the APS; Kapton is
covering the windows.

A.1.2 Simplified instruction for using the displex

In order to carry out low temperature RA-PDF perform the following general proce—
dures:

(1) Load the sample in appropriate sized and shaped low temperature sample holder.
You should watch out for the size of the sample, because the beryllium heat shield is
rather small. It’ll be good to measure the dimensions first to make sure. They have

1:1 design diagram you can turn to it.

91

 

loo

 

 

 

 

40L _
3.01
O . .
v

9r -

 

 

 

 

C
I.
'Al

I I

1 O 20 3O 4O
20

Figure A6: The XRD data from ZnSe0,5Te0,5 compared to that of the beryllium
window background,copper shielding used in the CCR refrigerator that is available
in the MUCAT station at the APS.

 

(2) The sample holder can be mounted either horizontally or vertically, and having
some washes can be handy. (3) The thermal couple should be placed into the back
of the sample holder, ensuring a good contact.

(4) Put the heat shield back. Find the threads in a gentle way. Take care never to
touch the Be window with bare hands, wearing gloves all the time is recommended.
Mount in the heat shield as far inside as possible, in order to make sure it is in the
same place every time.

(5) Put the vacuum shroud on. Do it slowly and carefully, wear gloves too. Be aware
that vacuum shroud is only attached by seals to the displex.

(6) Close all the valves: the venting valve, vacuum pump valves, the nitrogen gas
cylinder valve.

(7) Turn on the vacuum pump. The vacuum gauge should show a rapid pressure

decrease.

92

loo

.1]

 

l

 

(8) Open valves one by one starting from the vacuum pump. This should be done in
a slow fashion. do not open the venting valve, as it may damage the pump.

(9) After you opened the valve to the displex, you should see that the pressure gauge
drops fast. One red and one green light are in the front panel of the gauge.

(10) Once the green light is on, the refrigerator can be turned “on”.

(11) At this stage the setup is complete and the experiment is to be controlled from
the data acquisition desk.

For changing the sample, one needs to follow the same steps in reverse order:

1. Raise the temperature of the finger to 300 K. It usually takes one hour to raise

the temperature from 15 K to room temperature.
2. Turn off the refrigerator.

3. Wait until the temperature reading is 300 K. One needs to wait another hour,
for the other parts and sample to reach the room temperature. Attempting to
take the sample before equilibrium is reached, the moisture in the surrounding

air will accumulate on the sample and cause difficulty in taking it out.
4. Close all the valves.
5. Switch off the vacuum pump.

6. Open the venting, or using the nitrogen gas to fill the displex, observe the

pressure going up.
7. Wear the gloves for personnel protection.

8. Take off the vacuum shroud with extreme care, be slow when the shroud is
almost coming off. You may lose control when the shrouds disengaged from the

displex.
9. Take off the heat shield with extreme care, also be slow at the end.

93

10. Take off the thermal couple if there is one.

11. Unmount your sample.
The macro mar _ scan-temp an be used for automated run of the experiment:
mar_scan_temp T0 T1 Steps WaitSecs Repeats ExposureTime

Running this macro will take measurements at temperatures going from initial tem-
perature T0 Kelvin to the final temperature T1 Kelvin with 8 “Steps” divisions.
When “Steps = 0” it will measure only at T0. When a set point is reached, a sleep
for “WaitSecs” will occur to allow sample to equilibrate with the displex finger. At
every temperature take “Repeats” scans for “ExposureTime seconds” to improve the

statistics at a given temperature.

A.2 High temperature RA-PDF experiment

For high temperature experiments a furnace is used for raising the temperature of
the sample. Figure A.8 shows the furnace that is available at the MUCAT station at
the APS built by Ames Lab [78]. The mean feature that makes this furnace usable
for synchrotron experiments is its compact size, where it can be easily and quickly
mounted and aligned on a standard four circle goniometer. The sample environment
can be adjusted easily. The samples should be sealed into capillaries. There are differ-
ent kinds of capillaries depending on the highest temperature intended to use. Quartz
capillaries have a softening temperature around 1700 °C, while the glass capillaries
softening temperatures are around 700 °-800 °C depending on the type of glass used.
The capillary can be inserted into the core of the furnace by using the tube shown
in Figure A.7, through which it can be positioned perpendicularly with the line of
the incoming x-rays. This tube was designed to hold only 2mm diameter capillaries.

If capillary has diameter less than 2mm, the capillary will be loose and off-center in

94

 

lo

 

 

Figure A.7: The tube that is used to mount the capillary inside the furnace that is
available at the MUCAT station at the APS

the tube which produces an error in determining the sample to detector distance, an
important parameter for integrating the data. The samples can be aligned with an
opening in the side of the furnace. The x—ray beam enters through the opening and
the diffracted rays emerge through a slot in the furnace. The furnace can be heated
up to 1200 °C. The heating process can be fast (within minutes), although this is not
preferred as it affects the life time of the furnace. The cooling process is slow: it takes
around 20 minutes to cool down from 800 °C to the room temperature. If the reaction
requires the presence of oxygen (or other gases), or evolve gases, the sample should be
supported in open fused quartz tubes with the gas flowing through the tubes during

the measurements. If no gases are evolved or flowed, then the samples can be sealed

95

1:1 I“

 

 

 

Figure A8: The furnace that is available at the MUCAT station at the APS

in fused capillaries under partial pressure, the value of which depends on the highest
temperature one wants to heat the sample up to such that the capillaries will not ex-
plode. The first law of thermodynamics provides a good approximation for calculating
this pressure value. As is the case in the low temperature measurements, a waiting
time is needed before collecting the data at each temperature to ensure thermal equi-
libration through out the powder sample. For some samples the phase transition is a
time-temperature dependent, in such cases the waiting time should not be long such
that it causes a phase transition earlier than the temperature where it is supposed to
happen. Figure A.9 shows the example PDFs obtained from the boehmite to gamma
alumina transformation as was discussed earlier(5). This experiment ended up with
two different data sets of heated boehmite, one that was heated with regular 15 min-
utes waiting time at each temperature and the other with longer waiting time, due

to an error during the experiment that was found out after the sample temperature

96

 

1.8 _ 2.7 _ 3.8 . 4.5

G (1'2)

-:..7 -1.8 -0.9 0 09‘

 

 

 

 

 

 

 

2.8 '

v

'1.4 . 2.1

G (Fl-2)
0 7

 

 

 

 

 

 

 

 

-2.1 =1.4 —0.7' 0

2 4 6 8 10 12 1 4 1 6 18

r (3)
Figure A9: (a) the PDF of the boehmite sample heated up to 460 °C with 15 minutes
waiting time (red) and the PDF at the same temperature but heated for a longer
waiting time (blue). (b)the PDFs of the same sample heated to 480 °C with 15 minute

waiting time (red) and the one at 460 °C heated to a longer time (blue),the PDFs
are identical due to the longer waiting time while measuring the 460 °C PDF

97

 

lo

 

has reached 400°C which required to wait until the sample cooled down to the initial
temperature (300°C) and then reheating it again. Figure A.9(a) shows the PDF for a
boehmtie that was heated up to 460°C with a waiting time of 15 minutes, compared
to the PDFs with the longer waiting time. The difference between the PDFs is very
clear, and this shows the change in the structure that happened during the waiting
time. As seen from Figure A.9(b) the PDF measured at 480°C with regular waiting
time matches with the PDF measured at 460°C where the waiting time was much
longer. Even though this happened due to an error in performing the experiment,
it shows clearly the effect of the waiting time on the structure evolution for some
materials. This example illustrates that the waiting time is an important parameter

that needs to be decided about carefully before the experiment.

A.2.1 Simplified instruction For using the MUCAT station

furnace

The procedures needed to follow for performing the high T experiment is simple since
the beam scientist will take care of most of the issues regarding mounting the furnace
and calibrating the position of the tube that holds the capillary inside the furnace.
After the beam scientist finishes these time consuming issues (around half a day) we

are ready to run the experiment:

1. Load the sample in 2mm diameter capillaries before the experiment. If sample
loading needed to performed at the beamline the chemical lab at the MUCAT

station can be used for that. Which need to be included in the proposal to the
APS.

2. The beamtime will hit only the lower part of the sample, which makes filling

quarter of the capillary length sufficient.

3. Insert the sample into the tube (Figure A.7), then slowly and carefully insert

98

 

 

both into the upper opening of the furnace (Figure A.8). Capillaries are very
fragile such that any accidental touch may cause to brake it. For that if the
capillary is felt to be not smoothly going inside the furnace, or if it is hitting
something inside the core of the furnace, stop inserting the sample and take
it out to try that again. Breaking the capillary inside the furnace should be
extremely avoided, if this happens the beam scientist will need to move the

furnace and remove the broken capillaries, which will be very a time consuming.

4. The Sample position should be fixed through out the entire experiment, and
this mainly for three reasons: (i) To make sure that the x—ray is hitting the
sample. This can be checked also from the Mar image. (ii) Seeing signal from
the sample in the Mar image is not enough yet. The beam size is usually around
(0.5x 0.5 mm) which is less than the capillary diameter, but still there will be
a possibility that the sample may be shifted from its ideal position, in such
a case only part of the beam may be hitting the sample and the rest is not.
Such an error will yield a problematic data that needs to be avoided. (iv) To
obtain correct sample to detector distance from the integration the standard
sample. The position of the sample can be fixed by the eye telescope that will

be mounted by the beam scientist before starting the experiment.

5. Removing the sample should be performed again slowly and carefully to avoid

breaking the capillaries inside the furnace.

6. The eye telescope should be used each time new sample is loaded. The capillary
position should be always exactly at the same position where the standard

sample was mounted.

The MUCAT station has some ready macros to use for performing automated
runs of the experiments. For example the following command can be used to run

a macro available at the MUCAT station, enabling the user to collect data sets for

99

the desired range of temperatures, temperature steps, exposer time for each scan and
waiting time at each temperature. For clarifying the meaning of each parameter of

this command will use the following example:
mar_steptmp T0 T1 Steps WaitSec Repeats ExpouserTime; t8 Tfinsh

This macro is similar to the mor_scon-temp defined earlier for the low temperature
measurements. It differers by the te, which is used for changing the temperature to
“Tfinsh” after finishing collecting (“Repeats” — 1) data sets.

As it can bee seen the macro uses the Kelvin unit for measuring the temperature
as oppose to what is usually used in high-T measurements, for that the values of
the temperatures should be converted from Celsius to Kelvin before it is used in this

macro .

100

 

1cm

.— A. 1"

 

 

Bibliography

[1] W. L. Roth, The structure of mercuric oxide, Acta Crystallogr. 9, 277—280
(1956).

[2] R. W. G. Wyckoff, Crystal Structures, Crystal Structures, volume 1, Wiley,
New York, 2 edition, 1967.

[3] T. Proffen, S. J. L. Billinge, T. Egami, and D. Louca, Structural analysis
of complex materials using the atomic pair distribution function - a
practical guide, Z. Kristallogr. 218, 132—143 (2003).

[4] S. J. L. Billinge and M. G. Kanatzidis, Beyond crystallography: the study
of disorder, nanocrystallinity and crystallographically challenged ma-
terials, Chem. Commun. 2004, 749-760 (2004).

[5] J. L. Dye, K. D. Cram, S. A. Urbin, M. Y. Redko, J. E. Jackson, and M. Lefenfeld,
Alkali metals plus silica gel: powerful reducing agents and convenient
hydrogen sources, J. Am. Chem. Soc. 127, 9338—9339 (2005).

[6] X. Bouju, C. Joachim, and C. Girard, single-atom motion during a lateral
STM manipulation, Phys. Rev. B 59, 7845-7848 (1999).

[7] A. Ashkin, Acceleration and Trapping of Particles by Radiation Pres-
sure, Phys. Rev. Lett. 24, 156—159 (1970).

[8] Y. Kanda, Y. Uemichi, T. Kobayashi, L. Andalaluna, and M. Sugioka, Catalytic
Performance of Noble Metals Supported on A1203-Modified MOM-41
for Thiophene Hedrodesulfurization, in Nanoporous Materials IV, edited
by A. Sayari and M. Jaroniec, pages 747—754, Elsevier, Amstrdam, 2005.

[9] C. Q. Lu and X. S. Zhao, Nanoporous Materials-An overview, in
Nanoporous Material Science and Engineering, edited by C. Q. Lu and X. S.
Zhao, Imperial College Press, 2004.

[10] U. S. EPA, Capsule Report: aqueous mercury treatment, Office of Re-
search and Developmen, 1997.

[11] Y. Mori and T. J. Pinnavaia, Optimizing Organic Functionality in
Mesostructured Silica: Direct Assembly of Mercaptopropyl Groups
in Wormhole Hamework Structures, Chem. Mater. 13, 2173—2178 (2001).

101

 

 

lo

 

[12] S.-S. Shah, J .and Kim and T. J. Pinnavaia, A versatile pathway for the
direct assembly of organo-functional mesostructures from sodium sil-
icate, Chem. Commun. , 572—573 (2004).

[13] L. Mercier and T. J. Pinnavaia, Access in mesoporous materials: advan-
tages of a uniform pore structure in the design of a heavy metal ion
adsorbent for environmental remediation, Adv. Mater. 9, 500—503 (1997).

[14] X. Feng, G. E. Fryxell, L. Q. Wang, A. Y. Kim, J. Liu, and K. Kemner, Func-
tionalized Monolayers on Ordered Mesoporous Supports, Science 276,
923—926 (1997).

[15] A. M. Liu, K. Hidajat, S. Kawi, and D. Y. Zhao, A new class of hybrid meso-
porous materials with functionalized organic monolayers for selective
adsorption of heavy metal ions, Chem. Commun. , 1145—1146 (2000).

[16] A. Walcarius, M. Etienne, and B. Lebeau, Rate of access to the binding
sites in organically modified silicates. 2. ordered mesoporous silicas
grafted with amine or thiol groups, Chem. Mater. 15, 2161—2173 (2003).

[17] B. C. Lippens and J. H. de Boer, Study of phase transformations during
calcination of aluminum hydroxides by selected area electron diffrac-
tion, Acta Crystallogr. 17, 1312—1321 (1964).

[18] B. E. Warren, X-ray diffraction, X-roy diffraction, Dover, New York, 1990.

[19] T. Egami and S. J. L. Billinge, Lattice Effects in high--Tc superconductors,
in Physical properties of high-temperature superconductors V, edited by D. M.
Ginsberg, pages 265—373, Singapore, 1996, World—Scientific.

[20] V. Petkov, I-K. Jeong, J. S. Chung, M. F. Thorpe, S. Kycia, and S. J. L.
Billinge, High real-space resolution measurement of the local structure
of Ga1_mInmAs using X-ray diffraction, Phys. Rev. Lett. 83, 4089—4092
(1999).

[21] T. Egami and S. J. L. Billinge, Underneath the Bragg peaks: structural
analysis of complex materials, Underneath the Bragg peaks: structural anal-
ysis of complex materials, Pergamon Press, Elsevier, Oxford, England, 2003.

[22] A. C. Wright, Diffraction studies of glass structure: the first 70 years,
Glass. Phys. Chem. 24, 148-179 (1998).

[23] P. J. Chupas, X. Qiu, J. C. Hanson, P. L. Lee, C. P. Grey, and S. J. L. Billinge,
Rapid acquisition pair distribution function analysis (RA-PDF), J.
Appl. Crystallogr. 36, 1342—1347 (2003).

[24] A. P. Hammersley, S. O. Svenson, M. Hanfiand, and D. Hauserman, Two-
dimensional detector software: from real detector to idealised image
or two-theta scan, High Pressure Res. 14, 235—248 (1996).

102

 

1c

M—

'A— 3 u: 2'

 

 

[25] X. Qiu, J. W. Thompson, and S. J. L. Billinge, PDFgetX2: a GUI driven
program to obtain the pair distribution function from X-ray powder
diffraction data, J. Appl. Crystallogr. 37, 678 (2004).

[26] B. H. Toby and T. Egami, Accuracy of pair distribution function analysis
applied to crystalline and noncrystalline materials, Acta Crystallogr. A
48, 336—46 (1992).

[27] V. Petkov and S. J. L. Billinge, Local structure of random InxGa1_,,,As al-
loys by full-profile fitting of atomic pair distribution functions, Physica
B 305, 83 (2001).

[28] S. J. L. Billinge, R. G. DiFrancesco, G. H. Kwei, J. J. Neumeier, and J. D.
Thompson, Direct observation of lattice polaron formation in the local
structure of La1_xCa¢MnOg, Phys. Rev. Lett. 77, 715—718 (1996).

[29] I.-K. Jeong, T. Proffen, F. Mohiuddin—Jacobs, and S. J. L. Billinge, Measuring
correlated atomic motion using X-ray diffraction, J. Phys. Chem. A 103,
921—924 (1999).

[30] E. S. Boiin, S. J. L. Billinge, G. H. Kwei, and H. Takagi, Local structural
evidence for inhomogeneous charge distribution in Cu02 planes of su-
perconducting La2-xSrmCuO4, Physica C 341-348, 1793 (2000).

[31] Y. Waseda, The structure of non-crystalline materials, The structure of
non-crystalline materials, McGraw-Hill, New York, 1980.

[32] D. Louca and T. Egami, Local lattice distortions in Lal-xerMnO3 stud-
ied by pulsed neutron scattering, Phys. Rev. B 59, 6193—6204 (1999).

[33] V. Petkov, R. G. DiFrancesco, S. J. L. Billinge, M. Acharya, and H. C. Foley,
Local structure of nanoporous carbons, Philos. Mag. B 79, 1519 (1999).

[34] A. C. Larson and R. B. Von Dreele, General Structure Analysis System,
Report No. LAUR-86-748, Los Alamos National Laboratory, Los Alamos, NM
87545,1987.

[35] http://home.wxs.nl/ rietv025.

[36] C. L. Farrow, P. Juhas, J. W. Liu, D. Bryndin, E. S. BoZin, J. Bloch, T. Proffen,
and S. J. L. Billinge, PDFfit2 and PDFgui: Computer programs for

studying nanostructure in crystals, J. Phys: Condens. Matter (2007), to
be published.

[37] W. D. Knight, Nuclear Magnetic Resonance Shift in Metals, Phys. Rev.
76, 1259—1260 (1949).

[38] J. Levy, D. Tamarkin, H. Selig, and M. Rabinovitz, Potassium Metal Dis-
persed on Silica: A Versatile Reagent in Organic Chemistry, Angew.
Chem. Int. Edit. 20, 1033 (1981).

103

 

[39] D. Savoia, C. Tfombini, and A. Umani-Ronchi, Applications of potassium-
graphite and metals dispersed on graphite in organic synthesis, Pure
and Appl. Chem. 57, 1887—1896 (1985).

[40] A. J. Birch, The Birch reduction in organic synthesis, Pure and Appl.
Chem. 68, 553—556 (1999).

[41] J. A. Rabo, P. H. Angell, P. H. Kasai, and V. Schomaker, tudies of cations
in zeolites: adsorption of carbon monoxide; formation of Ni ions and
Na3+4 centres, Discuss. Faraday Soc. 41, 328-349 (1966).

[42] V. I. Srdanov, G. D. Stucky, E. Lippma, and G. Engelhardt, Evidence for an
Antiferromagnetic Transition in a Zeolite-Supported Cubic Lattice of
F Centers, Phys. Rev. Lett. 80, 2449—2452 (1998).

[43] A. S. Ichimura, J. L. Dye, M. A. Camblor, and L. A. Villaescusa, Toward
Inorganic Electrides, J. Am. Chem. Soc. 124, 1170—1171 (2002).

[44] D. P. Wernette, A. S. Ichimura, S. A. Urbin, and J. L. Dye, Inorganic Elec-
trides Formed by Alkali Metal Addition to Pure Silica Zeolites, Chem.
Mater. 15, 1441—1448 (2003).

[45] http: / / www.dcchem.co.kr / english / product / p_basic / p.basic07.htm.

[46] M. Feldman and P. Derochers, Research Universities and Local Economic
Development: Lessons from the History of the Johns Hopkins Univer-
sity, Industry and Innovation 10, 5—24 (2003).

[47] V. Petkov, S. J. L. Billinge, T. Vogt, A. S. Ichimura, and J. L. Dye, Structure of
intercalated Cs in zeolite ITQ-4: an array of metal ions and electrons
confined in a pseudo-1D nanoporous host, Phys. Rev. Lett. 89, 075502
(2002), (Highlighted in Phys. Rev. Focus: http://focus.aps.org/story/v10/st4).

[48] S. J. L. Billinge, E. J. McKimmey, M. Shatnawi, H. Kim, V. Petkov,
D. Wermeille, and T. J. Pinnavaia, Mercury Binding Sites in Thiol-
Functionalized Mesostructured Silica, J. Am. Chem. Soc. 127, 8492—8498
(2005).

[49] V. Petkov, P. N. ’Ilrikalitis, E. S. BoZin, S. J. L. Billinge, T. Vogt, and M. G.
Kanatzidis, Structure of V205.anO xerogel solved by the atomic pair
distribution function technique, J. Am. Chem. Soc. 124, 10157 (2002).

[50] S. Susman, K. J. Volin, D. G. Montague, and D. L. Price, Temperature de-
pendence of the first sharp diffraction peak in vitreous silica, Phys.
Rev. B 43, 11076—11081 (1991).

[51] K. Unruh and C. Huber, T.E. Huber, Melting and freezing behavior of
indium metal in porous glasses, Phys. Rev. B 48, 9021—9027 (1993).

104

[52]

[53]

[54]

[55]

[56]

[57]

[58]

[59]

[60]

[61]

[62]

P. Villars and L. D. Calvert, Pearson’s Handbook of Crystallographic
Data for Intermetallic Phases, Pearson’s Handbook of Crystallographic Data

for Intermetallic Phases, volume 3, American Society for Metals, Metals Park,

OH, 1985.

P. A. Grund and E. M. M. Pizy, Structure cristalline du metasilicate de
sodium anhydre, Na28i03, Acta Crystallogr. A 5, 837—840 (1952).

M. Tegze and J. Hafner, Electronic strucures of semiconducting alkali-
metal silicides and germanides, Phys. Rev. B 40, 9841 (1989).

M. Shatnawi, G. Paglia, J. L. Dye, K. D. Cram, M. Lefenfeld, and S. J. L.
Billinge, Structures of alkali metals in silica gel nanopores: new mate-
rials for chemical reductions and hydrogen production, J. Am. Chem.
Soc. 129, 1386—1392 (2007).

J. He, D. D. Klug, K. Uehara, K. F. Preston, C. I. Ratcliffe, and J. S. Tse, NNIR
and X-ray Spectroscopy of Sodium-Silicon Clathrates, J. Phys. Chem. B
105, 3475—3485 (2001).

M. Lefenfeld and L. Dye, James, Sodium Silicide and Alkali Metal-
Silica Gel for Convenient Hydrogen Production, Fuel Cell Magazine
August / Septemeber, 18—22 (2005).

M. H. Lim, C. F. Blanford, and A. Stein, Synthesis of ordered microporous
silicates with organosulfur surface groups and their applications as
solid acid catalysts, Chem. Mater. 10, 467—470 (1998).

C.-C. Chen, E. McKimmy, T. Pinnavaia, and K. Hayes, XAS study of mer-
cury(II) ions trapped in mercaptan-functionalized mesostructured sil-

icate with a wormhole framework structure, Environ. Sci. Technol. 38,
4758—4762 (2004).

A. Lennie, J. Charnock, and R. Pattrick, Structure of mercury(II)-sulfur
complexes by EXAFS spectroscopic measurements, Chem. Geol. 199,
199—207 (2003).

D. B. Bishop, G. D. McCool, A. J. Nelson, J. G. Reynolds, T. F. Baumann, G. A.
Fox, J. G. DeWitt, and J. C. Andrews, X—Ray absorption spectroscopy of
thiacrown compounds used in the remediation of mercury contami-
nated water, Microchem. J. 71, 247—254 (2002).

K. Kemner, X. Feng, J. Liu, G. E. Fryxell, L.-Q. Wang, A. Y. Kim, M. Gong, and
S. Mattigod, Investigation of the local chemical interactions between Hg

and self-assembled monolayers on mesoporous supports, J. Synchrotron
Radiat. 6, 633—635 (1999).

105

1c

 

 

[63] Petkov, RAD, a program for analysis of X-ray diffraction data from
amorphous materials for personal computers, J. Appl. Crystallogr. 22,
387—389 (1989).

[64] C. F. J. Baes and R. E. Mesmer, Hydrolysis of Cations, Hydrolysis of Cations,
John Wiley & Sons, New York, 1976.

[65] T. Proffen and S. J. L. Billinge, PDFFIT, a program for full profile struc-
tural refinement of the atomic pair distribution function, J. Appl. Crys-
tallogr. 32, 572—575 (1999).

[66] T. Tsuchida, R. Furuichi, and T. Ishii, Kinetics of the dehydration of
boehmites prepared under different hydrothermal conditions, Ther-
mochim. Acta 39, 103—115 (1980).

[67] S. J. Wilson and J. D. C. McConnell, A kinetic study of the system 7-
AlOOH/ A1203, J. Solid State Chem. 34, 315-322 (1980).

[68] J. A. Wang, X. Bokhimi, A. Morales, O. Novaro, T. Lopez, and R. Gomez,
Aluminum local environment and defects in the crystalline structure
of sol-gel alumina catalyst, J. Phys. Chem. B 103, 299—303 (1999).

[69] G. Paglia, A. L. Rohl, C. E. Buckley, and J. D. Gale, A computational in-
vestigation of the structure of n-alumina using interatomic potentials,
J. Mater. Chem. 11, 3310—3316 (2001).

[70] G. Paglia, C. E. Buckley, A. L. Rohl, K. Winter, R. D. Hart, B. A. Hunter,
A. Studer, and J. V. Hanna, The boehmite derived 7-alumina system,
1: structural evolution with temperature, with the identification and

structural determination of a new transition phase, 7I-alumina, Chem.
Mater. 16, 220—236 (2004).

[71] C. E. Corbato, R. T. Tettenhorst, and G. G. Christoph, Structure refinement
of deuterated boehmite, Clays and Clay Minerals 33, 71—75 (1985).

[72] P. Euzen, P. Raybaud, X. Krokidis, H. Toulhoat, J .-L. Le Loarer, J .-P. Jolivet,
and C. Froidefond, Catalytic Performance of Noble Metals Supported
on AlgOg-MOdified MCM—41 for Thiophene Hedrodesulfurization, in
Nanoporous Materials IV, edited by F. Schuth, K. Sing, and J. Weitkamp, page
1591, Wiley-VCH, Weinheim, 2002.

[73] K. E. Sickafus, J. M. Wills, and N. W. Grimes, Spinel compounds: structure
and property relations, J. Am. Ceram. Soc. 82, 3279—3297 (1999).

[74] D. Y. Li, B. H. O’Connor, G. I. D. Roach, and J. B. Cornell, Structural models
for h and g alumina by x-ray Rietveld Refinement, XVth Congress of the

International Union of Crystallography (Bordeaux) Abstracts volume, C—61
(1990).

106

[75]

[76]

[77]

[78]

[79]

G. Paglia, C. E. Buckley, A. L. Rohl, B. A. Hunter, R. D. Hart, J. V. Hanna,
and L. T. Byrne, Tetragonal structure model for boehmite-derived 7-
Alumina, Phys. Rev. B 68, 144110 (2003).

G. Paglia, Determination of the Structure of 7-Alumina using Empiri-
cal and First Principles Calculations combined with Supporting Experiments,
Phd thesis, Curtin University of Technology, 2004, Published online at
http://adt.caul.edu.au/.

G. Paglia, E. S. BoZin, and S. J. L. Billinge, A novel fine-scale nanostructure
in 7-A1203, Chem. Mater. 18, 3242—3248 (2006).

L. Margulies, M. Kramer, R. W. McCallum, S. Kycia, D. R. Haeffner, J. C.
Lang, and A. I. Goldman, New high temperature furnace for structure
refinement by powder diffraction in controlled atmospheres using syn-
chrotron radiation, Rev. Sci. Instrum. 70, 3554—3561 (1999).

H. Lin, E. S. Boiin, S. J. L. Billinge, E. Quarez, and M. G.
Kanatzidis, Nanoscale clusters in the high performance thermoelectric
AngmeTem+2, Phys. Rev. B 72, 174113 (2005).

107

 

   

111117171711[jg]