A Multi-Technique Comparative Evaluation of Ag Dispersion on
Polycrystalline TiO2
by
Zenda Deena Ann Davis
A thesis submitted to the Graduate Faculty of
Auburn University
in partial ful llment of the
requirements for the Degree of
Master of Science
Auburn, Alabama
December 8, 2012
Keywords: silver, titanium oxide, desulfurization, x-ray photoelectron spectroscopy,
adsorbents, crystallite size, dispersion
Copyright 2012 by Zenda Deena Ann Davis
Approved by
Bruce J. Tatarchuk, Chair, Professor of Chemical Engineering
W. Robert Ashurst, Associate Professor of Chemical Engineering
Minseo Park, Associate Professor of Physics
Christopher B. Roberts, Professor of Chemical Engineering
Abstract
Energy independence, coupled with the increasing demands on energy supply, and ris-
ing environmental, health, and e ciency concerns have fueled the need for more appropriate
re ning processes in the petroleum industry. The North American pursuit of energy inde-
pendence can be advanced through the re ning of regionally available petroleum reserves
{ deposits of shale oil, tar sands, and bitumen. These unconventional deposits typically
contain extra heavy oils that are extremely viscous and high in concentrations of sulfur. The
processing of these sulfur rich feedstock presents many new challenges. One of these is the
removal of heterocyclic sulfur compounds and their alkylated derivatives. Sulfur removal is
necessary for the evolution of fuel cell technology and for cleaner fuel oil.
Catalytic and adsorption processes that target undesirable refractory sulfur compounds
are being investigated for use in the desulfurization of hydrocarbon streams. A silver-titania
adsorbent developed at Auburn University is one such compound that can be used in the
desulfurization processes. The readily available raw materials, facile synthesis, and multi-
cycle regenerability at ambient conditions make this adsorbent appealing. Ag/TiO2 adsor-
bents are capable of removing various refractory species in the presence of 160 fold excess of
competing aromatics in logistic fuels, such as JP5. Ag/TiO2 adsorbents showed a consistent
capacity of 8.5 mg S/g of competing aromatics for JP5 fuels ( 1200 ppmw sulfur) for 10
cycles.
There is a continual e ort to improve the sulfur sorption capacity of Ag/TiO2 adsor-
bents. The  rst step in improving the quality of an adsorbent is to have a fundamental
understanding of its properties. Literature has shown that there is a strong correlation
between the e cacy of supported catalyst and particle dimension and dispersion. It is im-
portant to note that dispersion is important to gain the greatest access to the active atoms
ii
of the adsorbent. Thus, the size of metal particles is crucial to the activity and selectivity
of such systems. The purpose of this study was to perform fundamental studies to acquire
a molecular level understanding of Ag/TiO2 adsorbents used in the liquid phase desulfur-
ization of logistics fuels. The ultimate goal of this study is the enhancement of performance
capabilities and development of new sorbent materials.
In the literature, various techniques for the estimation of crystallite size are described.
The intrinsic strengths and weaknesses of the four such techniques, Electron Microscopy
(EM), X-ray Di raction (XRD), Gaseous Chemisorption, and X-ray Photoelectron Spec-
troscopy (XPS) have been analyzed in their application to the Ag=TiO2 adsorbent system.
Oxygen chemisorption and XPS have proven to be useful in de ning the catalytic system.
Both techniques have demonstrated respective nanometer scale and sub-nanometer scale oc-
currence of Ag supported on titania. The results from these techniques complemented with
information from Electron Paramagnetic Resonance (EPR) support the hypothesis that well
dispersed silver oxide crystallite decorate defects on the titania support at low loading (<4
wt%). These crystallites are responsible for the preferential selectivity of Ag=TiO2 for sulfur
removal at low loadings.
The XPS particle size estimation was approximately a factor of approximately 5 less
than the particle size found estimated by oxygen chemisorption for the same range. Particle
size estimation using XRD was not achieved. Silver atoms are mono{dispersed between 0
{ 4 wt % loading, after this loading there is a critical point beyond which particle growth
occurs.
iii
Acknowledgments
I would like to express my sincere gratitude to the many people who have supported me
throughout this journey. My research adviser, Dr. Tatarchuk for his invaluable and enthusi-
astic, support and guidance. My family for their unconditional love and encouragement and
my friends for keeping me grounded. I would also like to thank my colleagues { the past and
present members of the Center for Micro brous Materials Manufacturing (CM3).
To everyone who has challenged, encouraged and empowered me. Thank you.
iv
Table of Contents
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1 Surface Chemistry of Silver-Titania Adsorbents . . . . . . . . . . . . . . . . 5
2.1.1 Adsorbent Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.2 Surface Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Silver Crystal Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Determination of Particle Size . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.1 Chemisorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.2 X-ray Photoelectron Spectroscopy . . . . . . . . . . . . . . . . . . . . 13
2.3.3 X-ray Di raction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.4 Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Comparison of the Various Methods of Particle Size Determination . . . . . 18
2.4.1 Impact of Sampling Perspective on Particle Size . . . . . . . . . . . . 18
3 Experimental Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2 X-ray Photoelectron Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . 26
v
3.2.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.3 Oxygen Chemisorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.4 Electron Paramagnetic Resonance . . . . . . . . . . . . . . . . . . . . . . . 31
3.4.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.1 XPS Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.1.1 Atomic Ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.1.2 Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1.3 Particle Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2 Evaluation of Particle Size using XPS Experimental Data . . . . . . . . . . 54
4.3 Particle Growth Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.4 EPR Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
A Physical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
B Sample Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
B.1 Calculating the Ag Mole Fraction in Silver-Titania Adsorbent . . . . . . . . 92
B.2 Calculating Mean Free Path of Titanium Dioxide - Penn Method . . . . . . 94
C Crystal Size Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
C.1 Quantitative Estimation of Surface Atoms . . . . . . . . . . . . . . . . . . . 96
C.2 Monolayer Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
C.3 Chemisorption Particle Geometry Analysis . . . . . . . . . . . . . . . . . . 108
C.4 Theoretical XPS Intensity Calculation . . . . . . . . . . . . . . . . . . . . . 109
vi
List of Figures
2.1 4 wt% Ag/TiO2 adsorbent - 2.5 magni cation . . . . . . . . . . . . . . . . . . 6
2.2 4 wt% Ag/TiO2 adsorbent - 15 magni cation . . . . . . . . . . . . . . . . . . 6
2.3 4 wt% Ag/TiO2 adsorbent - 25 magni cation . . . . . . . . . . . . . . . . . . 7
2.4 Birds eye view of an fcc structure of (111) plane . . . . . . . . . . . . . . . . . . 8
2.5 FCC structure of (111) plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.6 FCC (111) plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.7 Diagram showing the perspective of number, volume, and surface weighted dis-
tributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.8 Bar chart showing number weighted sampling distribution perspective . . . . . . 20
2.9 Bar chart showing volume weighted sampling distribution perspective . . . . . . 20
2.10 Bar chart showing surface area weighted sampling distribution perspective . . . 21
2.11 Frequency of the comparative techniques used for crystallite size determination
as documented by ScienceDirect, June 2012 . . . . . . . . . . . . . . . . . . . . 22
2.12 Layers of a representative silver particle . . . . . . . . . . . . . . . . . . . . . . 25
4.1 XPS spectra of the Ag=TiO2 Ag(3d3=2;3d5=2) peaks . . . . . . . . . . . . . . . 34
4.2 Deconvolution of XPS spectra of the Ag=TiO2 Ag(3d3=2;3d5=2) peaks . . . . . . 36
vii
4.3 Oxygen 1s spectra of various loading of Ag=TiO2 adsorbents . . . . . . . . . . 39
4.4 Ag/Ti ratio as a function of weight loading of Ag . . . . . . . . . . . . . . . . . 42
4.5 Isometric View - Particles of di erent geometries - Not dawn to scale . . . . . . 44
4.6 Top View - Particles of di erent geometries - Not dawn to scale . . . . . . . . . 45
4.7 Trend in Ag average crystallite size in reported in Nair [1] . . . . . . . . . . . . 48
4.8 Trend in Ag average crystallite size (dimensions calculated based on reported
ASA value Nair[1]) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.9 Case 1 - Estimation of IAg/ITiO2 using known particle size values from chemisorp-
tion using (i) spherical, (ii) hemispherical, and (iii) cubic models. . . . . . . . . 50
4.10 Case 2 - Estimation of IAg/ITiO2 using thickness calculated from chemisorption
dimensions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.11 Case 3- Particle size estimation using XPS experimental data . . . . . . . . . . 53
4.12 Schematic of the TiO2 support . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.13 Ag/Ti ratio as a function of weight loading of Ag . . . . . . . . . . . . . . . . . 57
4.14 Ag/Ti ratio as a function of weight loading of Ag . . . . . . . . . . . . . . . . . 58
4.15 Ag/Ti ratio as a function of weight loading of Ag . . . . . . . . . . . . . . . . . 59
4.16 Ag/Ti ratio as a function of weight loading of Ag . . . . . . . . . . . . . . . . . 60
4.17 Schematic representation of the most probable morphological con guration of
Ag=TiO2. 0 { 4 wt% Ag loading. . . . . . . . . . . . . . . . . . . . . . . . . . . 62
viii
4.18 Schematic representation of the most probable morphological con guration of
Ag=TiO28{ 20 wt% Ag loading. . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.19 Monolayer detected versus moles deposited of Ag . . . . . . . . . . . . . . . . . 65
4.20 EPR spectra of the Ag=TiO2 at 77 K . . . . . . . . . . . . . . . . . . . . . . . . 67
4.21 EPR spectra of the Ag=TiO2 at 298 K . . . . . . . . . . . . . . . . . . . . . . . 68
C.1 Square matrix with evenly spaced spherical particles . . . . . . . . . . . . . . . 96
C.2 Square matrix with evenly spaced hemispherical particles . . . . . . . . . . . . 99
C.3 Square matrix with evenly spaced cubic particles . . . . . . . . . . . . . . . . . 101
C.4 Schematic representation of 20 wt% Ag particles (particles size derived from
chemisorption) on TiO2 surface . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
ix
List of Tables
2.1 Surface Properties of Ag=TiO2 Sorbent. . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Characterization Techniques Discussed in this Work . . . . . . . . . . . . . . . . 23
4.1 BE of Ag(3d5=2) Peak in Ag, Ag2O, and AgO . . . . . . . . . . . . . . . . . . . 35
4.2 FWHM and BE of Ag-3d Peaks . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.3 Atomic Ratio of Surface Species on Ag=TiO2 Adsorbent Pellets . . . . . . . . . 40
4.4 Silver-Titania Intensity Ratios - Experimental vs Calculated . . . . . . . . . . . 42
4.5 Trend in average crystal size for Ag particles on TiO2 determined by O2 chemisorp-
tion -(i) spherical, (ii) hemispherical and, (iii) cubic models . . . . . . . . . . . 47
4.6 Crystal Size due to Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.7 IAg/ITiO2 Experimental versus Theoretical -Kuipers - Case 1 . . . . . . . . . . . 50
4.8 IAg/ITiO2 Experimental versus Theoretical - Case 2 . . . . . . . . . . . . . . . . 52
4.9 Particle Size Dimensions - Oxygen Chemisorption versus XPS . . . . . . . . . . 53
4.10 IAg/ITiO2 -Experimental versus Theoretical . . . . . . . . . . . . . . . . . . . . 57
4.11 Particle Growth - after 8 wt% Loading . . . . . . . . . . . . . . . . . . . . . . . 59
4.12 Particle Growth- after 4 wt% Loading . . . . . . . . . . . . . . . . . . . . . . . 61
4.13 XPS Ag Crystallite Growth Estimation . . . . . . . . . . . . . . . . . . . . . . . 61
B.1 Atomic Ratio Ag=TiO2 Sorbent . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
B.2 Atomic Ratio of Surface Species on Ag=TiO2 Adsorbent Pellets . . . . . . . . . 93
C.1 Surface Coverage Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
C.1 Surface Coverage Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
C.1 Surface Coverage Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
x
C.2 Crystallite Geometry Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . 108
C.3 Ag(111) Intensity Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
C.3 Ag(111) Intensity Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
C.4 Ti(001) Intensity Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
C.4 Ti(001) Intensity Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
C.5 IAg/ITiO2 Theoretical Estimation-Kuipers Case-1 . . . . . . . . . . . . . . . . . . 113
C.6 IAg/ITiO2 Theoretical Estimation-Kuipers Case 2 . . . . . . . . . . . . . . . . . . 114
C.7 IAg/ITiO2 Theoretical Estimation-Kuipers Case 3 . . . . . . . . . . . . . . . . . . 115
xi
List of Abbreviations
Acronyms
Ac Cross-sectional Area
ADS Adsorptive Desulfurization
Am Cross-sectional area of the active metal surface atom
ASA Active Surface Area
BE Binding Energy
BET Brunaur-Emmett-Teller
CEN European Standards Organization
EM Electron Microscopy
EPA Environmental Protection Agency
EPR Electron Paramagnetic Resonance
ESCA Electron Spectroscopy Chemical Analysis
FCC face-centered cubic
FPM First Principles Model
FTIR Fourier Transform Infrared Spectroscopy
FWHM Full Width Half Maximum
HDS Hydrodesulfurization
xii
HSA Hemispherical Analyzer
IMFP Inelastic Mean Free Path
IR Infrared Spectroscopy
KM Kerkhof-Moulijn
MM Molecular Weight
NA Avogadro?s number
Nm monolayer uptake
NN Nearest Neighbor
PEMFC Proton Exchange Membrane Fuel Cell or Polymer Electrolyte Membrane Fuel Cell
S/V Surface-to-Volume
SA Surface Area
SF Sensitivity Factor
SMSI Strong Metal Support Interaction
SOFC Solid Oxide Fuel Cell
SPM Scanning Probe Microscope
TEM Transmission Electron Microscopy
TPD Temperature Programmed Desorption
TPRS Temperature Programmed Reaction Spectroscopy
UHV Ultra-High Vacuum
ULS Ultra Low Sulfur
xiii
XPS X-ray Photoelectron Spectroscopy
XRD X-ray Di raction
Xi Mole fraction of i
English letter symbols
d particle dimension
f particle shape factor
I Intensity
L percentage metal loading
m metal
S adsorption stoichiometry
s support
Greek letter symbols
 inelastic mean free path; wavelength
 work function; angular frequency
 density
 cross-section; speci c surface area
Units
G Gauss
ppm parts per million
ppmw parts per million weight
xiv
eV electron volts eV
xv
De nitions
Agglomeration: The gathering of smaller crystallites into larger particles.
Experimental: When used in tables mean experimental results.
Crystallite: A small single crystal. Crystallites are constituents of particles.
Interatomic spacing: The distance between the nuclei of atoms in a material
Mono-dispersed: The distribution of a quantity for example, atoms in such a way that
they are separated from each other i.e., not touching.
Monolayer: A singly packed layer of atoms or crystallites.
Nucleate: To aggregate as a result of the combination smaller crystallites into larger
particles.
Particles: A general term for small object of any size. Particles can be formed from
one or more crystallites. In this work, particles correspond to the active
phases rather than the support.
Sinter: To form larger particles or masses by heating, by pressure or both.
xvi
Chapter 1
Introduction
Worldwide concerns about increasing anthropogenic air pollution have resulted in the
establishment of stricter standards on transportation fuels. The aim of these standards is
to reduce exhaust pollutants thus improving air quality and subsequently public health.
Currently, there are no international standards mandating sulfur emission content. The
European Union, the United States of America, Japan and China have been leading the
world with the most severe environmental restrictions as pertaining to sulfur levels in fuels
[2, 3]. The European Union, Japan, and China have forged ahead of the United States
by restricting sulfur levels to 10 ppmw before or by 2008 [3]. While, in 2006, the United
States limited ultra low sulfur diesel (ULSD) to 15 ppmw, a regulation established by the
EPA under the Tier 2 Vehicle and Gasoline Sulfur Program Final Rule published in 2000
[4, 5]. Currently, there is no singular international body mandating strict worldwide sulfur
emission conformity. The European Standards Organization (CEN) regulates Europe and
the Environmental Protection Agency (EPA) regulates the United States. The environmental
directives in other regions are regulated by local government and are driven by their economic
needs. As a result, there is often little or no regulation in third world countries. Several
protocols have been implemented to reduce sulfur emissions in a stage-wise process but none
has had widespread global participation.
The emergence of fuel cells, which is a technology highly dependent on sulfur free fuels,
has increased the need for ULS fuels. Fuel cells are desirable for several reasons: their
ability to generate electric power directly from fuel, high e ciency, high energy, high power
density, portability, simplicity, small size, silent operation, and extremely low emissions.
Fuel cells require a high purity hydrogen input that is free of trace catalytic poisons. One
1
such identi ed poison is sulfur. Sulfur has deleterious e ects on the performance of fuel cells
and its removal is necessary for e cient fuel reforming. Polymer electrolyte membrane fuel
cells (PEMFC) and solid oxide fuel cells (SOFC) require sulfur levels below 1 ppmw and 10
ppmw respectively for e cient operation [6{9]. Consequently, researchers need to develop
more practical and e cient means to process ULS fuels to facilitate this technology and to
meet rising environmental regulations. As a result, several processes are being investigated
to achieve low sulfur standards; among them: bio-desulfurization, extractive desulfurization,
oxidative desulfurization and the use of adsorbents.
Adsorptive desulfurization (ADS) is the most economical process currently available,
with regards to industrial scale up and commercialization. The process of ADS relies on
using a sorbent as a separation tool for the adsorption of sulfur compounds from re nery
streams. In addition, the other alternative technologies have signi cant limitations in wide
spread utilization. Bio-desulfurization has to overcome issues with catalysts inferior activity
rates, thermal instability and slow reaction processes. Extractive desulfurization solvents
extract large amounts of energy intensive hydrocarbons and typically cannot decrease re nery
streams to ULS levels of 10 ppmw. Oxidative desulfurization is a comparably competitive
process, but only after the chemical oxidation of the organic sulfur compounds, and it relies
on ADS or extractive desulfurization to complete this process. Consequently, ADS has
emerged a highly promising step in fuel re ning [10, 11].
Future advancement in ADS technology requires the development of adsorbents with
high sulfur capacity, high selectivity and multi-cycle regenerability. High sorption capacity
(g of S per g of sorbent) is needed to remove as much sulfur as possible from a large volume of
fuel in a single hydrocarbon stream pass. Selectivity is required to target speci c compounds,
particularly sulfur containing heterocyclic compounds such as thiophenes derivates and their
alkylated derivatives that are di cult to chemically reform [12, 13]. It is also integral that
the absorbents be able to resist chemical changes that causes deactivation and has multi-
cycle retention properties to achieve competitive and economical sulfur removal. Ambient
2
temperature and pressure ADS of liquid fuels o ers great advantages when compared to
higher temperature and higher pressure HDS. The primary advantages being reduction in
energy usage, capital and operating costs.
Many adsorbents have been studied for liquid phase desulfurization of logistic fuels in
an e ort to further develop e ective adsorbents [6, 8{11, 14{19]. Adsorbents that strongly
bind and isolate sulfur-containing compounds are needed to enable the transition into more
environmentally benign sulfur products.
The novel Ag/Titania adsorbents developed at the Center for Micro brous Materials
Manufacturing (CM3) at Auburn University has been found to be a more e ective adsorbent
for the liquid phase desulfurization of high sulfur logistic fuels [20]. Ag=TiO2 adsorbents
overcome two obstacles usually encountered in ADS namely, low capacity and selectivity.
Ag=TiO2 based adsorbents have been found to selectively remove sulfur containing aromatic
compounds. The sulfur content in fuels such as JP5 and JP8 have been reduced to levels
below 1 ppmw at ambient operating conditions.
A molecular level understanding of the surface chemical processes involved in the ADS
process is needed to more fully explore the advantages. Information about the active sites,
fundamental mechanisms, and chemical species at the adsorption interface is needed to
truly understand the adsorbent system. Characterization of Ag=TiO2 is a reverse problem
formulation because the adsorbent was initially, found to possess desirable attributes instead
of being tailored with the desirable attributes. Subsequently, investigations were carried out
in a systematic manner to elucidate the surface chemistry interactions with the ultimate goal
of improving performance and designing a superior adsorbent.
1.1 Objectives
The  rst step in improving the quality of an adsorbent is to have a fundamental appre-
ciation of its properties. The characteristics of adsorbent are determined by the chemistry
occurring at the topmost atomic layers. Fundamental studies were performed to extract
3
information at the surface interface. Research has shown that there is a strong correlation
between the e cacy of supported adsorbent and crystallite dimension and dispersion. The
purpose of this study was to acquire a molecular level understanding of Ag/TiO2 adsorbents
with the ultimate goal of improving performance capabilities and developing new sorbent
materials. Thus, in an e ort to achieve stepwise characterization of the adsorbents a multi-
technique comparative evaluation of Ag dispersion on polycrystalline TiO2 was performed.
1.2 Overview
In this chapter, Chapter 1, the motivation of this work is presented along with objective
and a brief outline of the subsequent chapters. Chapter 2 presents the literature review {
an overview of the surface properties of the Ag/TiO2 adsorbents previously determined is
presented, four techniques of crystallite size determination are discussed along the funda-
mental physics and their sampling perspectives in their application the Ag/TiO2 adsorbents
system. Chapters 3, Experimental Methodology gives a brief overview of the techniques and
the experimental conditions used. In Chapter 4, the experimental results are presented and
discussed. Also, atheoretical method of crystallite size determination by XPS based on geo-
metric probability developed by Kuipers et al. is applied.The  nal chapters, 5 and 6 detail
the conclusions of this work and the potential for future work in clarifying outstanding issues
and questions.
4
Chapter 2
Literature Review
The use of heterogeneous catalysts is widespread in the petroleum and chemical indus-
tries. Billions of dollars are spent each year on tons of heterogeneous catalysts. Thus, there
is a continual e ort to improve the quality of the catalysts for e ciency and economy. The
 rst step in improving the quality of a catalyst is to have a fundamental understanding of
its properties. In this chapter, the fundamental properties of the Ag was used along with
simple theoretical concepts and experimental facts in the particle size and surface atom dis-
persion determination of Ag promoted on a TiO2 support. Various methods can be used
for the estimation of particle dimension and surface dispersion. Here, EM, XRD, gaseous
chemisorption, and XPS have been explored; the fundamental physics along with their sam-
pling perspective and biases were analyzed in the determination of Ag particles size in the
Ag/TiO2 adsorbents system.
2.1 Surface Chemistry of Silver-Titania Adsorbents
2.1.1 Adsorbent Preparation
Silver-titania adsorbents were made from 3.2 mm TiO2 pellets [St. Gobain Norpro: Type
ST61120]. The pellets were crushed, sieved and dried in a convection oven for at least 6 hours
at 110  C. Ag was dispersed on the support in the form of AgNO3 solution [Alfa Aesar Co.
99.9 % purity] via incipient wetness impregnation to maintain 90 % of the pore volume of
the TiO2. The concentration of the AgNO3 solution was varied according to the required
load. The sample was subsequently dried at 110 C for 6 hours followed by calcination in air
at 400 C for 2 hours.
5
Figure 2.1: 4 wt% Ag/TiO2 adsorbent - 2.5 magni cation
Figure 2.2: 4 wt% Ag/TiO2 adsorbent - 15 magni cation
6
Figure 2.3: 4 wt% Ag/TiO2 adsorbent - 25 magni cation
2.1.2 Surface Properties
Table 2.1: Surface Properties of Ag=TiO2 Sorbent.
Ag BET Pore Active Ag Average Dispersion
Loading surface area Volume surface area crystal size
(wt% ) (m2/g) (ml/g) (m2/g) (nm) (% )
0.00 153 0.46 { { {
4.00 114 0.27 6.7 3.4 34.4
8.00 89 0.23 10.7 4.1 28.7
12.00 79 0.21 12.1 5.3 22.4
20.00 58 0.13 14.3 6.9 17.0
Table 2.1 states the surface properties of Ag=TiO2 sorbent used in this study. This data
was previously published by Nair [1]. At 0 { 20 wt % loading Ag is present in the oxide
phase. EPR and TPR analyses indicated that Ag is present in its oxide state as Ag1+ with
minor concentrations present as Ag2+ [21]. TPR showed that the majority of Ag was present
in the oxide phase at higher loading (greater than 4 wt%). Oxygen chemisorption was used
to determine the average crystallite size of the silver particles on the surface.
7
2.2 Silver Crystal Structure
Ag has a face centered cubic (FCC) crystallographic structure with 4 atoms per unit
cell { the Ag unit cell contributes 1=2 of an atom on each of its 6 faces to each unit, and 1=8 th
of an atom at each of its 8 corners totaling 4 atoms. In the bulk, each atom has 12 nearest
neighbors thus a coordination number of 12. The Ag (111) structure exposes a surface
atomic arrangement of 3 fold symmetry which appears to be hexagonal. Fig 2.4 depicts the
hexagonal packing of the surface atoms, which is the most e cient (closely packed) manner
of atomic arrangement. The coordination number of the surface layer on the fcc (111) surface
is 9; each atom has 6 nearest neighbors in the 1st layer and 3 in the layer below, totaling 9
atoms.
Figure 2.4: Birds eye view of an fcc structure of (111) plane
The density of atoms in Ag FCC (111) crystal plane is calculated from the lattice
parameter, a. Since two corner atoms and and a central atom make up the length of the side
diagonal of the FCC (see Figure 2.5), the lattice parameter is given by:
a2 +a2 = (4R)2
a = 4p2 = 2Rp2
a = 4:08  A
The atomic density per cubic cm3 of Ag(111) surface is:
8
a
p 2
a
(a) 3d view of a unit cell (b) cross-sectional view
Figure 2.5: FCC structure of (111) plane
Figure 2.6: FCC (111) plane
9
Ag atomic density = # of atoms per unit cellVolume of unit cell = 4a3
= 4(4:08 10 10 m)3
= 5:88 1028 atoms/m3
The number of surface atoms is dependent on the surface structure. Considering the
crystallographic surface heterogeneity of polycrystalline surfaces for dispersed metals, the
usual approximation is to assume that such surfaces are formed from equal proportions of
the main lox index planes [22]. Using this assumption, equal portions of the (100), (110)
and (111) planes was used to estimate the polycrystalline planar concentration.
Density of atoms in Ag(111) crystal plane
No. of atoms contained in Ag (111) plane =
 1
6  3 corners
 
+
 1
2  3 sides
 
atoms
= 2 atoms
Area of Ag (111) plane = a
2p3
2
Surface density of atoms in Ag (111) crystal plane = 2a2p3
2
= 4a2p3
= 4(4:08 10 8)2 p3
= 1:38 1019 atoms/m2
10
Density of atoms in Ag(110) crystal plane
No. of atoms contained in Ag (110) plane =
 1
4  4 corners
 
+
 1
2  2 diagonal center
 
atoms
= 2 atoms
Area of Ag (110) plane = a2p2
Surface density of atoms in Ag (110) crystal plane = 2a2p2
= 2(4:08 10 10)2 p2
= 0:85 1019 atoms/m2
Density of atoms in Ag(100) crystal plane
No. of atoms contained in Ag (100) plane =
 1
4  4 corners
 
+ (1 center) atoms
= 2 atoms
Area of Ag (100) plane = a2
Surface density of atoms in Ag (100) crystal plane = 2a2
= 2(4:08 10 10)2
= 1:20 1019 atoms/m2
Generally, in heterogeneous catalysis, only a monolayer (ML) of a chemisorbed species is
adsorbed to the active sites. Assuming equal proportions of the Ag low index planes the
number of Ag atoms in a ML is estimated to be 1.14 1015 atoms=cm2.
11
2.3 Determination of Particle Size
2.3.1 Chemisorption
Gas phase chemisorption is the most extensively used technique for determining particle
size and surface atom dispersion of supported metal catalysts in heterogeneous systems
[24{33]. In this technique, a monolayer of chemisorbing atoms (surface atoms exposed)
per number of total atoms is assumed to be directly proportional to the speci c uptake.
Oxygen has been shown to be the adsorbate that most accurately predicts the crystallite
size dimensions of silver, the use of hydrogen and carbon dioxide as probe gases do not
result in accurate determination of silver particle size [22, 26, 29]. The underlying theoretical
assumptions of chemisorption are:
 Monolayer formation criteria - a monolayer is chemisorbed
 Stoichiometric ratio
 Adsorption selectivity
 Gas is adsorbed on the surface and not absorbed internally by the metal
 Unimodal particle size
 Crystallite formation
 Crystallite geometry
 Isolated atoms do not adsorb oxygen
Perspectives
Chemisorption is a simple, well-established surface weighted means of calculating the
average particle diameter. Of all the techniques considered in this chapter, it is the most
readily available techniques. Chemisorption is a surface area weighted average distribution
12
of the vital size. The accuracy of the techniques relies on background information, namely,
the correct stoichiometric ratio, and the ability of a absorbate gas to selectively chemisorb
on the active metal provided that there is constant gas uptake on the support when the
metal is present or absent. For the Ag/TiO2 adsorbent, there may be problems due to the
variation in the stoichiometry ratio of Ag : O (the number of sites per chemisorbed oxygen
atom) at the adsorption surface [29]. Although Ag(S) : O ? 1 appears to be the best value
as reported in the literature [28, 34, 35] other stiochiometries possibilities have been reported
[28]. Di erences in stoichiometries may be due to the hypothesis that the gas molecule to
metal stoichiometry is dependent on metal particle size. Other discrepancies lie in the fact
that the oxygen adlayer on silver has been reported to be complex [36, 37]. Complexities
arise to carbonaceous contaminants which have an e ect on the sorption behavior of oxygen
due to possibility of subsurface oxidation of silver [28].
In chemisorption, there is size variation depending on the crystallite geometry assumed,
for example, spherical and cubic geometries will produce slightly di erent average dimen-
sions. See results in Table 4.6. Other problems that may arise include, multiple adsorption
of oxygen molecule on metal surface atoms, internal absorption, incomplete adsorption of
the oxygen molecules by the metal particles, and non-metallic silver particles. Of these prob-
lems, explanations for exaggerated crystallite size would be adsorption by titania support {
subsurface adsorption and multiple adsorption of gas molecule on metal surface atoms { un-
der estimation of the stoichiometric ratio [30]. Lower crystallite size would result from over
estimation of the stoichiometric ratio and inability of absorbate gas to selectively chemisorb
a monolayer.
2.3.2 X-ray Photoelectron Spectroscopy
XPS is based on the detection of photoelectrons from a sample irradiated by an X-
ray source. The detected photoelectrons are recorded in a energy spectrum. This energy
spectrum is analyzed to produce a wealth of chemical information about the sample under
13
investigation. XPS is sensitive for trace amounts of all elements with the exception of
hydrogen. Quantitative information is received only from the top 10 nm of the surface
thus, XPS is de ned as a surface analysis technique. Like chemisorption, XPS is a well
established as a technique for studying surface atom dispersion and particle size of metal on
supported systems [38{47], however, in comparison it is less often used to acquired particle
size dimension [48{53].
The basic assumptions underlying XPS are:
 The number of electrons recorded is proportional to the number of atoms in a given
state
 Variation in the chemical environment does not appreciably a ect the overall ejected
photoelectrons
 Mono-dispersion of surface atoms
Perspectives
XPS is a non-destructive, surface analysis technique that gives a near- surface weighted
measurement of the average particle dimension. Although there is general agreement that
XPS o ers a surface weighted characterization of particle dimension, Fung [49] postulated
that XPS also adds a volume aspect such that it gives an account of the particle dimension
based between the surface and volume average size. In XPS, the particle size is determined
without extensive pretreatment with regards to other techniques. This is important as
pretreatment may alter the state of the particle or crystallite in question.
XPS is broadly applicable { all elements can be observed in XPS with the exception of
hydrogen. Also, unlike chemisorption, XPS is applicable to metal and metal oxides (reducible
and irreducible phases). XPS o ers good chemical resolution and depth resolution. The
shortcomings of the past regarding poor spatial resolution and detection limit [54] have been
improved upon in recent years [55{57].
14
2.3.3 X-ray Di raction
X-ray di raction XRD, is a non-destructive analytical technique which is used to derive
information about the crystal structure, chemical composition, and physical properties of
materials. In XRD, the scattered intensity of X-ray radiation is used to produce photon
energy spectra as a function of the incident and scattered angle, polarization, and wavelength
on a given sample. The spectra produced are characteristics of the composition of the sample.
The scattered monochromatic X-rays are the result of constructive interference by the atoms
in the crystal plane. Crystalline material produce di raction maxima according to the Bragg
equation:
n = 2dsin (2.1)
where n is an integer,  is the wavelength of incident wave, d is the spacing between the
planes in the atomic lattice, and  is the angle between the incident ray and the scattering
planes. The average crystallite size is determined by X-ray broadening [58, 59] by the Scherrer
expression:
D = k  cos (2.2)
where D is the mean diameter,  is the pure X-ray di raction broadening and k is a constant
 1. k and  are constants dependent on the crystallite shape.
Perspectives
XRD is a non-destructive, volume weighted means of determining crystallite size. The
major weakness of this technique lies in the determination of small crystallites. If the ir-
radiated volume is too small for a sharp di raction maximum to build up, the resulting
di raction pattern is broadened [59]. From the literature, the lower detection limit is re-
ported to be . 5 nm [27, 49, 61]. However, when a synchrotron radiation source is used,
platinum particles as small as 1.5 nm have been estimated [62]. This is a result of a better
signal to noise ratio produced using the high intensity radiation.
15
The agreement between X-ray line broadening (XRD) and gaseous chemisorption is good
when particles are greater than > 50  A and < 1000  A [26] XRD did not provide conclusive
measurements when applied in the crystallite size determination of Ag/TiO2 adsorbents.
The lack of signal suggested that the crystallite was below the detection limit of the XRD
 less than 5 nm. This is a result of X-ray broadening of the di raction beam produced
from very small crystals at angle close to the Bragg angle [59, 60]. Strong di raction beam
pattern form from bigger crystals than smaller crystal. Since, the instrumental peak width
was much larger than the broadening due to crystallite size, the crystallite size could not be
determined. A silver peak was detected at the 20 wt% loading however due to its small size
and broad width a size dimension could not be reliably determined.
2.3.4 Electron Microscopy
Electron microscopy EM, is the near atomic scale microscopic examination of objects,
through the use of highly energetic electrons (produced form a highly focused X-ray beam).
EM can yield information on the topography, morphology, composition, and crystallography
of a given sample. There are several forms of EM available. In this work, only transmis-
sion electron microscope (TEM) will be employed. In TEM, information is projected in
two-dimensional image from the interaction of the electron beam with matter encountered.
Detection is dependent on the existence of su cient contrast between the particle and the
support.
Perspectives
EM is a projected area, perimeter weighted average of size distribution. The main
advantages of TEM are that it is a straightforward technique and the particles can be viewed
from the images created. However, care must be taken during interpretation regarding the
possible sources of contrast. Ideally, contrast should be as a result of the attenuation of
electrons on the surface of the given sample but, there is also contrast due to di raction
16
and interference which lead to misinterpretation [63, 64]. TEM is very prone to contrast
issues. It is di cult to contrast supported catalyst material, thus it is not a good technique
for particle size determination os supported catalysts. Other disadvantages are that it is
destructive and di cult to prepare; extremely thin sections of specimens, typically about
100 nm are needed. Several samples must be examined with a large number of particles to
produce a accurate quantitative representation of the sample.
The application of the TEM using the di raction contrast method proved to be fruitless
in the determination of particle size on the Ag/TiO2. Information on particles size was not
discernible from the bright and dark  eld images produced [65].
17
2.4 Comparison of the Various Methods of Particle Size Determination
Undoubtedly, the most valuable information is derived through a combination of several
characterization techniques. However, when comparing techniques the underlying assump-
tions and perspectives must be considered when evaluating the information received. Since
the information required is average particle size, the basis of the average weighting must be
considered during analysis of the representation of particle size.
2.4.1 Impact of Sampling Perspective on Particle Size
Many techniques make an convenient assumption about the geometry of a particle when
determining particle size. This is done to lessen the complexity, for example, an amorphous
particle is described by many lengths (shortest and longest diameter and perimeter) in various
di erent directions (x-axis, y-axis or a percentage of both). Taking all these length into
consideration to determine the true particle size would be overly complex. Thus, using an
assumption of a regular geometric shape will be valuable. This results in a trade o decreased
complexity at the concession of a small reduction in accuracy, which research has shown has
very little e ect [66].
Each technique derives particle size based on a physical principle. A physical value such
as, volume of a gas, scattered light and electron count is measured and used to determine
average particle dimension. Each technique is based on a particular calculation namely,
number, surface, and volume. For example, the XRD measures light scattered from various
particles and averages the dimensions produced from the distribution of particles.
In a number distribution, each individual particle has equal weighting. In a volume
distribution, weighting is placed on the number of particles of a particular volume. Likewise,
in surface area distribution, weighting is placed on the number of particles that gives a
particular surface area. In Figure 2.7, 42 particles of three di erent sizes are depicted. Figures
2.8, 2.9, and 2.10 shows the number, volume, and surface area distribution respectively. In
Figure 2.8, each particle size accounts for one-third of the total. The 1 nm, 2 nm, and the
18
3 nm particles all carry equal weight because each group has a total of 14 particles. This
makes it clear that size is not a determinant factor in number distribution. The same result
is translated in a volume distribution in Figure 2.9 where the 1 nm, 2 nm, and the 3 nm
particles account for 2.8 %, 22%, and 75% respectively of the total volume. Likewise, in
 gure 2.10, the surface area distribution the 1 nm, 2 nm, and the 3 nm particles account for
approximately 8 %, 31%, and 61% respectively of the total surface area.
Diameter
1 nm
2 nm
3 nm
Figure 2.7: Diagram showing the perspective of number, volume, and surface weighted
distributions
The number average diameter as determined by TEM is given by:
dn =  nidi n
i
where d is the diameter and, n the the particle number in increment i.
The surface average diameter ds is determined by
ds =  nid
3
i
 nid2i
19
1nm 2nm 3nm
5
10
15
Particle Size
Num
ber
Figure 2.8: Bar chart showing number weighted sampling distribution perspective
1nm 2nm 3nm
1pi
6
8pi
6
27pi
6
Particle Size
Volume
Figure 2.9: Bar chart showing volume weighted sampling distribution perspective
20
1nm 2nm 3nm
pi
4pi
8pi
Particle Size
Surface
Area
Figure 2.10: Bar chart showing surface area weighted sampling distribution perspective
the volume average diameter dv is given by:
dv =  nid
4
i
 nid3i
the volume area average diameter dVA, as determined by chemisorption is given by:
dVA = 6 VMSA
M
where M indicates metal, V is the volume per metal atom, and SA is the surface area.
21
The four main techniques considered for average particle size determination are based
on either a physical or a chemical principle. Electron microscopy and X-ray di raction and
based on direct observation, while X-ray photoelectron spectroscopy can be said to estimate
mean particle size somewhat indirectly. Chemisorption is based on the chemical interaction
of gas molecules with the particles on the surface.
Electron microscopy and X-ray di raction are the most widespread technique for particle
size determination. A simple search on Elsevier?s ScienceDirect website for the technique
and particle size (at the time of the preparation of this document) revealed 224,386 EM
articles, 114, 725 XRD articles, 34,011 XPS articles, and 21,061 chemisorption articles.
EM
XRD
XPS
Chemisorption
2.24 ? 105
1.15 ? 105
34,011
21,061
No. of Articles
Figure 2.11: Frequency of the comparative techniques used for crystallite size determina-
tion as documented by ScienceDirect, June 2012
Comparison techniques that are in agreement with each other that give reproducible and
reliable results are essential in particle size determination. Firstly, when selecting comparison
techniques, it is important that the weighted mean measurement be understood. The notion
of a weighted mean describes how each data point contributes equally to the  nal average.
Table 2.2 lists the four techniques and their corresponding weight basis. The literature
22
has several accounts of particle size comparison using a combination of two or more of the
techniques - XPS, XRD, EM, and/or chemisorption [27{29, 31, 34, 67{71].
Table 2.2: Characterization Techniques Discussed in this Work
Technique In Out Average Weight Information
Chemisorption Gas Gas Surface Surface Area, Dispersion,
Particle Size
XPS X-ray Electron Near-Surface Composition, Dispersion,
Oxidation State, Particle
Size
XRD X-ray X-ray Volume Bulk Structure, Particle
Size
TEM Electron Electron Perimeter, Number Particle Size, Structure
Seyedmonir et al [29] and Mustard & Bartholomew [27] used XRD, TEM, and chemisorp-
tion to characterize Ag/TiO2 and supported nickel particles respectively. In determining
particle size, Seyedmonir et al reported that there were no silver peaks distinguishable from
XRD and that a very small number of crystals (range 3-10 nm) could be easily distinguish-
able from TEM micrographs. Thus, a possibility of error exists due to the small number,
however, there was reasonable agreement between oxygen chemisorption and TEM data for
particle size estimation. Mustard & Bartholomew reported that XRD was not accurate in
the estimating particle size of supported nickel. TEM was accurate but tedious because sam-
ples are di cult to prepare and of the large number of particles that must be counted for
accurate analysis. Chemisorption was also accurate and the most convenient. Chemisorp-
tion and TEM were found to be in very good and good agreement over a wide range of
metal loading Ni/SiO2 and 15 wt% Ni/Al2O3 respectively. There was poor agreement with
Ni/TiO2 suggesting that gaseous adsorption was suppressed. Strong metal support interac-
tions (SMSI) have been reported to cause the causes suppression of thin electron transparent
metal crystallite.
Several researchers [28, 34] have used chemisorption with XRD for crystallite size esti-
mations. In comparing XRD and chemisorption, Strohmayer [28] reported poor quantitative
23
agreement between the two techniques for Ag/TiO2 and Ag/ -Al2O3. The discrepancy was
attributable to low sensitivity of the XRD and the possibilities of subsurface oxygen ab-
sorbed in the bulk of the Ag crystal during chemisorption. Scholten [34] did not report any
aggrement between the two methods.
The in uence of particle dimension and size distribution factors limits the accuracy to
30 % in XRD [22]. Very poor correlation between image analysis and laser di raction exists.
Generally, in the literature, XRD did not provide good agreement with the other techniques
for particle size estimation [27{29, 31, 69, 71]. Chemisorption showed more aggrement with
TEM [27, 29, 31, 71].
In Regalbuto et al.[31] an attempt was made to explain the discrepancy in chemisorption
and XRD crystal size determination by XPS for Pt/WO3/SiO2 catalysts. This attempt was
qualitative rather than quantitative in nature. Fung [49] recognized XPS as a desirable
technique that can be used in the determination of the size of supported metal particles
notably in the region in which XRD is generally insensitive, below 3-5 nm [22, 27, 49, 61, 72].
XPS sees only the outer surface area of a particle - the projected area as the outer surface
atom contribute to the majority of the XPS signal. Using the intensity ratios eliminates
most of the complexities of reproducing the exact geometry and morphology since these
properties are not a determining characteristics of the intensity ratios. The degree of surface
atom dispersion, percentage active particle loading, the projected surface area of the particle,
and the speci c surface area of the support a ect the intensity ratios.
XPS is described as a surface weighted means of calculating the average crystallite
diameter. Surface weighted is more meaningful with regards to surface analysis in contrast
to number and volume weight [70]. Fung disputed the surface weighted claim by proposing
that XPS give something between a surface and volume average size, since the surface atoms
and atoms below the surface contribute signi cantly to the intensity of the XPS signal. This
is a valid claim when atoms are smaller than 10 nm, as XPS sees the top 10 nm of the
surface.
24
For a Ag particle in 20 wt% Ag/TiO2, where d = 6.9 nm and  Ag=1.38 nm. The
particle would be 6:9=1:38 = 5  thick. Since in the most ideal case, XPS sees signal up to 5
 in thickness, XPS would only see the surface atoms i.e., the photoelectron intensity count
would arise only from one atomic layer. For smaller particles photoelectron intensity would
arise form more than one layer.
Monolayer Perspective/Approach
6.9 nm
 = 1:38 nm
Figure 2.12: Layers of a representative silver particle
25
Chapter 3
Experimental Methodology
3.1 Introduction
Surface characterization is of crucial importance to the advancement of adsorbent tech-
nology. In this process, surface electrons are energized by an external source and are emitted
for analysis. The emitted electrons arise from the top few atomic layers of the specimen
and are a direct representation of the species at the interface. Information received from
the surface species can elucidate properties such as: composition, size, dispersion, reactivity,
selectivity, active site, etc.
The two main techniques utilized in this thesis are XPS and oxygen chemisorption.
The following sections brie y describe the theory involved in the application of these two
techniques.
3.2 X-ray Photoelectron Spectroscopy
X-ray photoelectron spectroscopy (XPS) is the most widely used surface sensitive char-
acterization technique based on the photoelectric e ect[38]. XPS give information about the
chemical environment of a sample based on binding energy variation. When the surface of
a sample is irradiated using an X-ray source, energy is transferred from the photons to the
atoms on the surface of a sample. If the photon energy is enough to overcome the attractive
forces binding the electron in its orbital, the electron will be emitted so that the atom can
regain an energetically stable environment. The emitted electron, photoelectron, powered
by the photon energy and kinetic energy are detected by an electron analyzer. Using the
26
principle of conservation of energy,
h = Ek +Eb + (3.1)
where h is the energy of the photon, Ek is the kinetic energy, Eb is the binding energy and
 is the work function.
The intensity of the photoelectronic peak for a sample as described by Penn [73] is
I = I0n  ( )D( ) (3.2)
where I0 is the X-ray  ux, n is the density of atoms,  is the photoelectron cross-section,
 ( ) is the mean free path of energy ( ), D( ) the fraction of electrons detected by the
analyzer. More complicated forms of this equation exist as described by Briggs and Seah
[74] which include more instrumental and morphological parameters; however, this form
of the equation has a  10 % uncertainty which is deemed acceptable given the degree of
di culty in acquiring all the di erent parameters. This form of the equation is referred to
as the ? rst principle model? (FPM).
The experimental intensity, I,which corresponds to the area under the graph of the XPS
peak, is used to determine the atomic concentration, Cx , the of the element in the sample
[75]:
A/Cx  ( )D( ) (3.3)
An empirical method using sensitivity factors can also be used to estimate Cx[76]:
Cx =
Ix
SFxX
i
Ii
SFi
(3.4)
where SFx is the atomic sensitivity factor for element x. The SF is directly proportional to
the cross-section, the inelastic mean free path for the photoelectron core, level and several
27
parameters that are dependent on the experimental conditions. Thus, all instrumental factors
are grouped into the SF, which is given by Wagner et al. [77]:
SF =   yAT (3.5)
where  is the photoelectric cross-section for a particular transition cm2,  is the angular
frequency factor, y is e ciency of production, A is the sample illuminated area, T is the
detector e ciency, and  is the mean free path of photoelectrons. A list of sensitivity factors
are reported in the literature for a large number of elements using the F 1s line [77].
This method is however prone to lower accuracy than XPS measurement intensity de-
rived from FPM [78]. Wertheim [75] states that there is no possibility of de ning universally
accepted applicable sensitivity factors. This is because the area of the main line that depends
on the fraction of events accompanied by the multi-electron process varies among atoms and
materials. Also, several issues arise for the transferability of SF between instruments. In
this study, the FPM was applied to determine the atomic concentration.
3.2.1 Experimental
The XPS system consists of non-monochromatic Leybold-Heraeus LHS-10 spectrometer
with a dual anode Al/Mg X-ray source and hemispherical electron energy analyzer (HSA).
XPS measurements were taken with an X-ray source typically requiring between 100 W
and 350 W of power. The system typically yielded a 1.2 eV FWHM for the Au(4f7=2)
photoelectron peak of a gold foil using a Mg K anode at a workable spectrometer pass
energy and count rate. The measurement uncertainty was recorded as 0.2 eV. The samples
were mounted on carbon, high-vacuum compatible, double-sided adhesive tape. Sorbent
samples were introduced into the load lock and degassed to pressures of ca. 1  10 6 Torr.
Subsequently, the samples were inserted into the main chamber where measurements were
taken at a residual pressure of ca. 1  10 7 Torr. Al K (h =1486.6 eV) or Mg K (h 
28
=1253.6 eV) x-ray source was used at spectrometer operating conditions of 10 mA and 50 eV.
XPS data were  tted using the XPSPEAK41 program. Sample charging was compensated
for by adjusting XPS settings to the C 1s peak at 284.6 eV. The samples were assessed
to ensure that the shoulder was not the result of artifacts due to system contamination.
Atomic ratios were derived using the analysis of the elemental peak areas and following well
established methods for XPS data analysis [42, 46, 73{75, 77, 79]. Peak deconvolution was
performed using a nonlinear Shirley baseline and a combination of Gaussian and Lorentzian
type curves.
3.3 Oxygen Chemisorption
Chemisorption is an analytical adsorption technique (widely used in heterogenous catal-
ysis) that is based on chemical interactions occurring between the the adsorbate gas and the
exposed surface of a catalytic material. Chemisorption is used to reveal information about
a variety of surface properties. These include but are not limited to: the active metal sur-
face area, dispersion of the active metal, surface energy, reducibility and oxidizability of
the catalytic material, and crystallite size. In this work, information from selective oxygen
chemisorption was used speci cally to determine the silver crystallite size of the Ag/TiO2
adsorbent. Oxygen gas was used in preference to other gases such as CO and H2 as it
most readily adsorbed to form a well-de ned monolayer coverage (with the least challenges)
[28, 80{82].
Selective oxygen chemisorption was used to obtain the active metal surface area at
conditions which supported the formation of a monolayer on the surface of the exposed
metal atoms. The number of surface metal atoms is obtained from measuring the amount of
chemisorbed gas. The amount of oxygen chemisorbed based on the formation of a monolayer
was used with the known adsorption stoichiometry between the adsorbate gas and the metal
[28, 35, 36, 67, 83]. The metal surface area is given by:
29
ASA = NmSAmNA (3.6)
where ASA is the active surface area of the metal atoms (m2/g), Nm is the oxygen monolayer
uptake (moles/g), S is the adsorption stoichiometry, Am is the cross-sectional area of the
active metal surface atom ( A/ Ag atom), and NA is Avogadro?s number (atom/moles).
Metal crystallite size was determined by estimating the volume from the mass and the
density of the supported metal, using theASA, and by assuming a geometry of the crystallite.
The geometries assumed were (i) spherical, (ii) hemispherical, and (iii) cubic. The average
crystallite size is given by:
d = LfASA  
m
(3.7)
where d is the particle diameter, L is the percent metal loading,  m is the density of the
supported metal, and f is the particle shape correction factor (f = 6 for spherical and
hemispherical particles and f = 5 for cubic particles see Section C.3).
3.3.1 Experimental
The Quantachrome Autosorb-1 was used to perform oxygen chemisorption using that
static chemisorption technique [82]. Three pretreatment steps were carried out for oxygen
chemisorption:  rstly, 5 mg of the adsorbent pellets were heated to 150 C followed by an
evacuation at 2.99  10 9 Torr for 30 minutes to remove the moisture and contaminants
from the surface. Secondly, hydrogen reduction at 300 C and 760 Torr for 60 minutes to
provide reducible oxygen uptake. Thirdly, evacuation at 300 C and 2.99  10 9 Torr for 60
minutes for the removal of physisorbed hydrogen. Oxygen uptake was recorded at 170 C after
saturation of the surface was reached. The e uent gases were continuously monitored with
a mass spectrometer. The Brunaur-Emmett-Teller (BET) surface areas were measured from
nitrogen gas adsorption-desortpion at 77 K. The average sizes of the dispersed crystallites
were derived from the active surface area.
30
3.4 Electron Paramagnetic Resonance
Electron Paramagnetic Resonance (EPR) is a spectroscopic technique that is used to
detect chemical species that have unpaired electrons. It is the measurement and interpreta-
tion of the energy di erences between the atomic or molecular states and is used to increase
knowledge about the structure of molecules and probe the adsorbate-surface interactions by
analyzing a molecular absorption spectra which arise when the frequency and the amount
of the electromagnetic radiation is changed as it passes through a sample.
EPR is based on the interaction of electromagnetic radiation with magnetic moments
arising from electrons. In EPR the intensity, the number of absorption lines and the line
position is used to identify di erent nuclei. Intensity is purely a property of the electron,but
can be correlated to a metal when the metal content is known.
Line positions in EPR are stated in g values which is a function of the ratio of the
frequency to the resonant  eld. The relationship describing the absorption of microwave
energy between two spin states is
 E = h = gbB
where:
 E is the energy di erence between the two spin states
h is Planck?s constant
 is the microwave frequency
g is the Zeeman splitting factor
b is the Bohr magneton
B is the applied magnetic  eld.
EPR is used as a complementary technique to characterize the adsorbate surface system,
information can only be obtained if the system is paramagnetic. Thus, in the AgTiO2
system we can only gain direct information on the Ag2+ paramagnetic species.
31
3.4.1 Experimental
The EPR system consists of a Bruker EMX-6/1 X-band EPR spectrometer composed
of: an EMX 1/3 console, an ER 041 X61 bridge microwave bridge with built-in microwave
frequency counter, an ER070 magnet, and an ER-410410st standard universal rectangu-
lar angular cavity. Data acquisition was performed with the software supplied by Bruker
(WINEPR acquisition program), data manipulation (determination of g-values, subtraction,
base lining, integration and conversion to ASCII  les for use with Origin or Microsoft Excel)
was done with the WINEPR program version 2.11.
Experimental
Ag=TiO2 samples were prepared by the method described previously. Prepared samples
were cooled with liquid nitrogen to 77 K in a  nger dewar. Spectra of the as-prepared sorbent
were recorded with a  eld modulation frequency of 100 kHz, a modulation amplitude of 10
mT, a frequency of 9.37 GHz, and a power supply of 2 mW.
32
Chapter 4
Experimental Results
This chapter details the results of the di erent techniques used to characterize the
physiochemical properties of the Ag=TiO2 adsorbent system. XPS and oxygen chemisorption
were the principal techniques employed in this study. XRD data was not presented in the
study due to problems determining particle size at low weight percent loadings.
4.1 XPS Analysis
The XPS survey spectra revealed signi cant compositions of the following elements: Ag,
Ti, O, C. The measurement sequence and procedure for all samples were the same. A 20 eV
and a 50 eV spectra were recorded for all samples in the following sequence: C(1s), Survey,
Ag(3d), Ti(2p), O(1s), C(1s). The spectra were referenced to the adventitious C 1s peak at
284.6 eV. Variation in spectrometer intensity was accounted for by taking the intensity of
the C(1s) peak at the beginning and end of each sequence.
The XPS spectra in Figure 4.1 show the experimental data of the Ag 3d doublets of
Ag=TiO2 adsorbents ranging in silver loading from 1 wt% - 20 wt%. As expected, the studies
indicated that the Ag content on the support increased with Ag loading; also, as is reported
in the literature, the Ag 3d doublet maintained its characteristic features: a peak separation
of 6.0 eV and an area ratio of 2:3 [76, 77, 85]
The Ag signal intensity was below the detection limit of the XPS at the 1 wt% loading;
a weak signal was detected at the 2 wt% loading and from there the Ag peak intensity
increased in prominence up to 20 wt% loading. The spectra showed a slight shift toward
a lower binding energy (BE) of the Ag 3d doublet peaks at the 12 wt% loading and 20
wt% loading. A shift in the binding energy is a result of a change in the chemical bonding
33
1 wt%
2 wt%
4 wt%
8 wt%
12 wt%
20 wt%
373.6
3d3/2
367.6
3d5/2
Ag
360370380
Binding Energy (eV)
Intensit
y(a.u.)
Figure 4.1: XPS spectra of the Ag=TiO2 Ag(3d3=2;3d5=2) peaks
of the atoms in a solid. Several factors such as valence electron density, lattice potential,
work function, and atomic relaxation energy are reported to in uence a shift in BE [85, 86].
If ionic charge or oxidation state were the only factors to be considered, then it could be
concluded that Ag was reduced. However, the anomalous nature of the BE of Ag oxides
compared to that of Ag metal is such that there is a negative shift in BE from Ag0 to
Ag1+, 2+ [87, 88]. Normally in the literature, most metal to metal oxides systems exhibit
positive binding energy shifts. Changes in oxidation state of Ag could not be determined
due to the slight di erences in BE values using the (3d5=2) peak binding energies Ag0 (368.2
eV), Ag1+ (367.7 eV), and Ag2+ (367.4 eV) [76, 85, 87]. Additionally, this determination
was beyond the sensitivity of the non-monochromatic LHS-10 XPS system which had an
34
uncertainty of  0.2 eV. Table 4.1 shows the values of the BE of Ag(3d5=2) peak in Ag,
Ag2O, and AgO found in literature.
Table 4.1: BE of Ag(3d5=2) Peak in Ag, Ag2O, and AgO
Compound Binding Energy (eV) Reference
Ag 368.1  0.1 Sch on [87]
Ag 368.0 Gaarenstrom [85]
Ag 367.9 Wagner [76]
Ag 367.9  0.2 This work
Ag2O 367.7  0.2 Sch on [87]
Ag2O 367.6 Gaarenstrom [85]
AgO 367.8 Wagner [76]
AgO 367.4  0.2 Sch on [87]
AgO 367.2 Gaarenstrom [85]
The decomposition of the Ag 3d photoemission spectra is shown in Figure 4.2. Peak
deconvolution was accomplished using the XPSPeak41 software. Photoelectron lines have
a Lorentzian shape corresponding to the lifetime of the core hole that is created; if all the
features were ideal a strictly Lorentzian curve would be used. However, a combination of
Gaussian-Lorentzian curve was used to account for non-idealities in the system; instrumental
factors such as the width of the x-ray line and the analyzer resolution. This mixed curve
ratio was employed for the entire analysis. Symetrical line shapes were used for all elements.
Several parameter constraints were employed in consideration of known physical constants
of Ag in order to obey the chemistry. These parameters were: the peak symmetry - de ned
by the p-subshell, a doublet separation of 6.0  0.2 eV, and the relative intensities for p
electrons, 1:2. A proper background was also necessary to obey the chemistry and physics
involved in the system. A blend of the Shirley and linear backgrounds was used to satisfy
the 1:2 ratio of the p doublet. The Gaussian-Lorentzian curve ratio, the relative intensity
2p doublet area ratio, and the peak separation were all  xed. The FWHM of the peaks were
constrained within a range. The use of the physical constants and constraints should reduce
ambiguity, elucidate underlying peaks, and eliminate the introduction of biases that would
35
360370380
2 wt. %
4 wt. %
8 wt. %
12 wt. %
20 wt. %
Binding Energy (eV)
Intensit
y(a.u.)
Figure 4.2: Deconvolution of XPS spectra of the Ag=TiO2 Ag(3d3=2;3d5=2) peaks
result in a loss of the contribution of other peaks. The peaks were  tted with a fair degree
of certainty because of these physical constants and constraints .
The prominent shoulders displayed in the 20 wt% loading were attributed to the oc-
currence of a second doublet component. The Ag silver at the 12 wt% loading showed an
anomaly in that the BE of the Ag 3d peak displayed the highest BE of all the sample ana-
lyzed. This occurrence was independently veri ed [89]. The 4, 12 and 20 wt% loading were
decomposed into two doublets. The multiple occurrence and growth of the second doublet
suggest that this is not an artifact resulting from system contamination. The second dou-
blet suggests the occurrence of multiple species of AgxOx. Table 4.2 shows the BE and the
36
FWHM values for the components of the XPS Ag 3d spectra. The BE of the silver oxides
decrease in the order Agm { Ag2O { AgO while the FWHM increases and the peaks get
broader. The  rst Ag component is assigned to Ag2O with Ag 3d5=2 located at 365.6  0.2
eV as is reported in literature (see Table 4.1). The FWHM of this component decreases
with increasing weight loading. The second component Ag peak is assigned to AgO. AgO is
reported to have two chemically inequivalent silver atoms with chemical formula AgIAgIIIO2.
AgI has two oxygen atoms and AgIII with four oxygen atoms [87].
Oxygen displays distinctive features which are characteristic to metallic silver and
AgxOx compounds; Ag0 has a peak at 532.2 eV, Ag1+ has two peaks at 529.0 eV and
530.4 eV, and Ag2+ has two peak at 528.4 eV and 530.3 eV [87]. The analyses of the Ag
3d and O 1s peaks Figures 4.3 and 4.2 lead to the conclusion that the peak occurring at
approximately 529.5 eV can be assigned to TiO2. There is most likely a small contribution
from Ag2O and AgO, but that contribution could not be extracted from this data with great
 delity. However, it can be concluded that the oxygen peaks show the presence of multiple
species. The change from one component to two component in the oxygen peak could be due
to continuously changing oxygen absorbed on Ag2O to metallic Ag noted in metal-oxygen
systems reported in Sch on [87]. Previous TPR investigations revealed that Ag is present in
its oxide state as Ag1+ with minor concentrations present as Ag2+ [21].
37
T able 4.2: FWHM and BE of Ag-3d P ea ks
Binding Energy (eV)
Ag = TiO 2 Comp onen t 1 Comp onen t 2
Loading Ag(3d 3 = 2 ) FWHM Ag(3d 5 = 2 ) FWHM Ag(3d 3 = 2 ) FWHM Ag(3d 5 = 2 ) FWHM
1 wt% UD UD UD UD UD UD UD UD
2 wt% 373.9 2.3 367.9 2.3 UD UD UD UD
4 wt% 373.5 2.1 367.5 2.1 370.8 2.1 365.1 2.1
8 wt% 373.7 2.1 367.6 2.1 UD UD UD UD
12 wt% 374.0 2.0 368.0 2.0 371.3 2.0 365.3 2.0
20 wt% 373.7 1.9 367.7 1.9 371.3 2.3 365.3 2.3
UD - undetectable
y uncertain t y  0.2 eV
38
525530535
4 wt.%
8 wt. %
12 wt. %
20 wt. %
TiO2
?
?
Binding Energy (eV)
Intensit
y(a.u.)
z - indicates multiple oxygen species.
Figure 4.3: Oxygen 1s spectra of various loading of Ag=TiO2 adsorbents
39
4.1.1 Atomic Ratios
Atomic ratios derived from the XPS data are listed in Table 4.3. These ratio were
calculated using a simpli ed version of the FPM model - Equation 3.3. Quantitative mea-
surements had 10% uncertainty. On analysis of the data, it was found that Ti(2p3=2) was a
more suitable reference due to its stability. The BE of the carbon 1s peak shifted because of
the existence of major secondary carbon species/component. For comparison, atomic ratios
referenced to carbon are listed in Table B.2.
Table 4.3: Atomic Ratio of Surface Species on Ag=TiO2 Adsorbent Pellets
Ag=TiO2 Loading
Element 1wt% 2wt% 4wt% 8wt% 12wt% 20wt%
C-1s 0.25 0.31 0.21 0.19 0.19 0.22
O-1s 0.52 0.51 0.53 0.52 0.51 0.46
TiO2-2p 0.23 0.17 0.23 0.23 0.23 0.22
Ag-3d UD 0.01 0.03 0.06 0.08 0.10
Theoretical Agy 0.008 0.015 0.03 0.06 0.09 0.15
Total 1.00 1.00 1.00 1.00 1.00 1.00
Stoichiometric Ratios
O/Ti 2.25 2.90 2.26 2.20 2.26 2.14
Ag/O | 0.02 0.06 0.14 0.16 0.22
Ag/Ti | 0.06 0.14 0.28 0.35 0.48
z - uncertainty  10%
UD - undetectable
y - calculated based on the Ag deposited on specimen.
The atomic ratios of TiO2 are fairly constant for all the weight loadings, except for the
2 wt% loading. This shows that the titania support was not appreciably obscured by the
silver deposited. Supporting quantitative evidence is shown in Appendix C.1 The deviation
in the 2 wt% sample can be explained by the increased levels of carbon which suggested
carbon contamination. Similarly, the atomic ratios of oxygen are fairly constant across the
various loadings, except for the 20 wt% sample where it is hypothesized that reduction took
place. The O/Ti ratios did vary at the 2 and 20 wt% loadings. It is believed that deviated
40
occurred at the 2 wt% due to carbon contamination and at 20 wt% as a result of some
chemical structural change.
4.1.2 Dispersion
Information regarding the dispersion and particle size of silver particles on titanium
oxide support was extracted using the XPS. XPS has been well established as a technique
for studying dispersion of metal on supported systems [33, 39{44, 50, 66, 90{93]. Figure 4.19
shows the intensities of the active particles relative to the support versus the mole fraction
as detected by the XPS for various loadings of silver particles. The XPS intensity ratio for
Ag=TiO2 increases with additional loadings of silver. There is a linear relationship observed
at lower loadings of less than 0.03 mole fraction (<4 wt% ) which suggests that silver is
highly dispersed on the titanium oxide support up to a certain critical value. At this critical
value, theorized to be between 4 and 5 wt% , increased silver loading deviates from the linear
trend and levels o smoothly. This points to the growth of silver particles. Initially, particles
are likely to be of uniform distribution and are well dispersed on the surface of the support
until they reach a critical loading where the particles begin to grow in size. Similar trends
have been observed in the literature [40, 93, 95]. Briggs [40] observed the peak areas of metal
catalysts supported on silica versus the wt% metal loading in M=SiO2 catalyst1 and Park
[93] observed the XPS Cr 2p/Al 2p intensity ratios. This theory is also supported by the
fact that sulfur capacity for the Ag=TiO2 adsorbent increases with weight loading up to a
critical value, between 4 and 5 wt% after which increased loading does not increase sulfur
capacity [1].
The dashed line in Figure 4.4 exempli es the trend that would result if the silver-
to-titania intensity ratios exhibited a strong linear dependence with loading; however, the
experimental values, as indicated by the solid line, did not show this dependence. This
dependence was calculated assuming all the silver was atomically dispersed and every atom
1the identity of M was not disclosed
41
was accessible (best case scenario). The 2, 4, and 8 wt% Ag=TiO2 loadings conformed to
uniform dispersion as revealed by the values in Table 4.4. At the 12 wt% the experimental
value falls slightly below the calculated values and at 20 wt%, the adsorbent with the highest
loading, the value falls signi cantly below linear dispersion line. It is important to note that
high and uniform dispersion is important to gain the greatest access to the active atoms of
the adsorbent.
0 2 4 6 8 10 12 14 16 18 20
0
0.1
0.2
0.3
0.4
0.5
0.6
Ag Loading on TiO 2 (wt %)
IAg
/I
TiO
2
Ag /TiO 2 ratio
Figure 4.4: Ag/Ti ratio as a function of weight loading of Ag
Table 4.4: Silver-Titania Intensity Ratios - Experimental vs Calculated
Element Ratio Ag=TiO2 Loading
IAg=ITi 0wt% 1wt% 2wt% 4wt% 8wt% 12wt% 20wt%
Experimental 0 UD 0.06 0.14 0.28 0.35 0.48
Calculated 0 0.04 0.07 0.14 0.28 0.42 0.70
z - uncertainty  10%
UD - undetectable
y - Assuming no nucleation and uniform dispersion
 - Calculated values are based on a linear line through 2 & 4 wt % loading
42
Several calculations were performed to investigate the silver surface area i.e. the acces-
sible silver atoms, covering the titania support. In these calculations, the following assump-
tions were made:
 The Ag crystals are an ordered packing of particles that are uniformly and equally
dispersed on the TiO2 support.
 The shape of the Ag particle was (i) spherical and (ii) hemispherical (ii) cubic, see
Figure 4.5.
 The Ag particles occupy the intersection of points of a square matrix on the support.
 All Ag particles occupy a position on the surface.
 The surface shielded by the base of a single hemisphere is the same as that shielded
by a sphere.
Figure 4.5 gives a pictorial description of some of the assumptions (not drawn to scale).
Detailed calculations can be seen in Appendix C.1. These calculations employed the results
of oxygen chemisorption particle size determination to derive surface coverage. It show that
maximum surface coverage of spherical and hemispherical shaped particles were 2.76 % and
5.52 % respectively,2 of the total surface area of the support. Thus, supporting the XPS
results3 that showed there was minimal shadowing of the TiO2 surface area by the silver
particles. Table C.1 summarizes the calculated surface coverage results for 4, 8, 12 and 20
% wt Ag loading. XPS was used to gain a more accurate estimation of particle size.
4.1.3 Particle Size
In addition to dispersion, the properties of a catalyst are to a large extent dependent on
particle size. However, the atomic size di cult to predict. Di culties arise in compounds
2Ag deposited at the 20 wt% loading produced the largest particles.
3See atomic ratios - section 4.1.1
43
(a) Spheres
(b) Hemispheres
(c) Cubes
Figure 4.5: Isometric View - Particles of di erent geometries - Not dawn to scale
44
(a) Spheres
(b) Hemispheres
(c) Cubes
Figure 4.6: Top View - Particles of di erent geometries - Not dawn to scale
45
because of nearest neighbors interactions and the resulting perturbation on the electron
charge cloud. Despite these di culties, a best estimation is a useful starting point. Particle
size was derived using the techniques, chemisorption and XPS. The underlying assumptions
associated with each technique was outlined in Chapter 2.
In the literature, there are several instances where Ag has been shown to have various
geometries when deposited on various support including titania [29, 67, 96]. Thus, three
di erent geometries were explored. Table 4.6 lists the average crystallite size determined
using oxygen chemisorption with the corresponding weight percent loading of Ag for spheri-
cal, hemispherical, and cubic geometries. These results are also shown graphically in Figure
4.5. The graph in Figure 4.7 indicates that average crystallite size increased linearly with
Ag loading. Note that on this graph the crystal size approximations of the spherical and
hemispherical particles geometries were equal. The crystallite size dimensions were derived
using the active surface area (ASA) values reported in Table 5 in Nair et al [1] using Egn
3.7. A discrepancy was found in the values reported at the 20 wt% loading. The ASA value
corresponded to a value of 8.0 nm (using Egn 3.7) instead of the reported 6.9 nm value.
Both values were considered in the analysis.
In Figures 4.7 and 4.8, a line of best  t was drawn on the graphs which could be
predictive of the trend on which these results are based. It was noteworthy that the line of
best  t fell within the the error bars using the reported crystallite size and the crystallite
size calculated due to the reported ASA value at the 20 wt% loading. Thus, the accurate
value could not be determined using regression analysis. This analysis showed that assuming
chemisorption is correct there is not much di erence in assuming a spherical, hemispherical,
or cubic geometry.
46
4 6 8 10 12 14 16 18 20
3
4
5
6
7
8
?
?
Ag Loading (wt %)
Avg
Crystal
Size
(nm)
for
O2
Chemisorption
cubic
spherical
hemispherical
Table 4.5: Trend in average crystal size for Ag particles on TiO2 determined by O2
chemisorption -(i) spherical, (ii) hemispherical and, (iii) cubic models
Table 4.6: Crystal Size due to Geometry
Metal Active Surface Average Crystal Size (nm)
Loading (wt%) Area (m2/g) Sphere Hemisphere Cube
4 6.69 3.4 3.4 2.9
8 10.70 4.1 4.1 3.6
12 12.05 5.3 5.3 4.7
20z 14.31 8.0 8.0 6.7
20 16.50 6.9 6.9 5.8
z - dimension calculated based on reported ASA value.
 - dimension reported in Nair [1].
47
4 6 8 10 12 14 16 18 20
3
4
5
6
7
Ag Loading (wt %)
Avg
Crystal
Size
(nm)
for
O2
Chemisorption
spherical and hemispherical
cubic
Figure 4.7: Trend in Ag average crystallite size in reported in Nair [1]
4 6 8 10 12 14 16 18 20
3
4
5
6
7
Ag Loading (wt %)
Avg
Crystal
Size
(nm)
for
O2
Chemisorption
spherical and hemispherical
cubic
Figure 4.8: Trend in Ag average crystallite size (dimensions calculated based on reported
ASA value Nair[1])
48
In this section, particle size was determined using the method outlined in Kuipers et al.
[91]. In this method a general model is derived that allowed the quanti cation of particle
dimension using XPS signal received from the surface/volume ratios of supported phase.
XPS intensity ratios determined by surface-to-volume ratios are independent of particle
shape [29, 91].
Using the relationship from Kiupers et al. - Equation 1, the signal ratio IAg=ITiO2 :
IAg
ITiO2 =
I1Ag
I1TiO2  
 Ag [1 exp( tAg= Ag)]
1  Ag [1 exp( tAg= TiO2)] (4.1)
where I is the intensity, 1 superscript indicates the signal ratio of the pure compound,
and  is the inelastic mean free path (IMFP),  is geometric surface coverage, parallel to
the normals of the underlying support, t is the layer thickness. I1Ag is estimated for Ag(111)
surface plane, see Section C.3. I1TiO2 is estimated for TiO2(001) surface plane, see Section C.4.
Several surface common planes of TiO2 vary within 0.2 of the reported atomic spacing, and
spherical particles of diameter 2RAg = 3tAg, where R is the radius, with half the coverage of
the layers[97, 98]. XPS cannot distinguish between supported layers of thickness and surface
coverage [91]. This fact was used to  nd the unknown value  Ag from the relationship:
 Ag tAg = cAg 
Ag TiO2 (1 cAg)
(4.2)
where  is the density , c is the concentration,  is the speci c surface area of the support.
The following estimation were employed in various models.
Case 1: Estimate IAg/ITiO2 using known particle size values from chemisorption.
1. Use Equation 4.2 then apply Equation 4.1-Using the surface coverage estimation from
Table C.1- (this table used the chemisortpion data to estimate surface coverage using (i)
spherical (ii) hemispherical and (iii) cubic model), the thickness tAg, was found. Next,
this value was used in Equation 4.1 to  nd IAg/ITiO2. For sample data see Appendix
C.5.
49
4 6 8 10 12 14 16 18 20
0
0.02
0.04
0.06
0.08
0.1
0.12
Ag Loading(wt %)
IAg
/ITi
cubic
spherical
hemispherical
Figure 4.9: Case 1 - Estimation of IAg/ITiO2 using known particle size values from
chemisorption using (i) spherical, (ii) hemispherical, and (iii) cubic models.
Table 4.7: IAg/ITiO2 Experimental versus Theoretical -Kuipers - Case 1
IAg/ITiO2
Ag Loading Experimental Theoretical
wt % Spherical Hemispherical Cubic
4 0.14 0.022 0.047 0.000
8 0.28 0.036 0.078 0.022
12 0.35 0.040 0.088 0.025
20 0.48 0.050 0.111 0.031
50
Three models for particles were assumed (spheres, hemispheres and cubes), the theo-
retical IAg=ITiO2 ratio was calculated for the various surface areas. The result showed
that the IAg=ITiO2 ratio increased with increased surface area. The hemispheres had
the greatest surface coverage area hence the greatest predicted IAg=ITiO2 ratio. The
three geometric models produced the same basic shape in the graph in Figure 4.9, the
corresponding values can be seen in Table 4.7.
2. Use relationship 2RAg = 3tAg to get  Ag, then apply Equation 4.1.
4 6 8 10 12 14 16 18 20
0.03
0.04
0.05
0.06
0.07
0.08
0.09
Ag Loading(wt %)
IAg
/ITi
Figure 4.10: Case 2 - Estimation of IAg/ITiO2 using thickness calculated from chemisorption
dimensions.
The IAg/ITiO2 theoretical values derived using case 2 was lower than the experimental
values by a factor of 4 -7, increasing from 4 - 20 wt% loading. The surface coverage from
case 1 was calculated based on the chemisorption diameters to calculate the thickness and in
case 2, the chemisorption diameters were used directly in using the relationship 2RAg = 3tAg
to calculate the thickness.
51
Table 4.8: IAg/ITiO2 Experimental versus Theoretical - Case 2
Ag Loading IAg/ITiO2
Experimental Theoretical
wt% Spherical Cubic
4 0.14 0.032 0.040
8 0.28 0.053 0.061
12 0.35 0.058 0.067
20 0.48 0.068 0.083
This method of determining theoretical values for IAg/ITiO2 does not produce a cor-
relation with experimental values. Since the surface coverage values were estimated using
chemisorption values, an improvement in this method would be to  nd an independent source
of surface coverage estimation.
Case 3: Estimate particle size values from XPS data. Use the values from XPS data
for IAg/ITiO2 in Equation 4.1, 4.2, and the relationship 2RAg = 3tAg, to derive the particle
size.
The plot of the average particle size predicted from the experimental XPS intensity
ratios Figure 4.11 displayed a linear relationship. This linear relationship was expected in
these predictions as a curve would be indicative of particle growth. Particle growth is not
accounted for by the model because XPS see mono distribution, thus when are large particles
the accuracy of particle size determination by XPS decreases. This estimation is unrealistic
as 0.7  A is in the subatomic region. The atomic radius of silver is 1.75  A thus estimating a
particle diameter to be less than the atomic radius is incorrect.
The method outlined in Kuipers et al. describes a homogenous mixture of silver particles
with the titania support, not the silver dispersed on the surface of the titania. This accounts
for the low calculated crystallite size predictions. These results showed that Ag is not
homogeneously dispersed throughout the support.
52
2 4 6 8 10 12 14 16 18 20
0
0.03
0.06
0.14
0.23
0.43
Ag Loading (wt %)
Avg
Crystal
Size
(nm)
Avg Crystal Size (nm)
Figure 4.11: Case 3- Particle size estimation using XPS experimental data
Table 4.9: Particle Size Dimensions - Oxygen Chemisorption versus XPS
Ag Loading Particle Size (nm)
wt % Oxygen Chemisorption XPS
4 3.4 0.07
8 4.1 0.15
12 5.3 0.24
20 6.9 0.47
53
4.2 Evaluation of Particle Size using XPS Experimental Data
The intensity of the Ag photoelectron peak from  rst principles is described by:
IAg = I0n Ag Ag( )D( )
Z t
0
e t= dt (4.3)
where I0 is the X-ray  ux, n is the density of Ag atoms,  is the photoelectron cross-section
of Ag, f( ) is a function of energy,  ( ) is the mean free path of energy of Ag ( ), D( ) the
fraction of electrons detected by the analyzer. The Equation 4.3 will be further simpli ed
to:
IAg = B0 Ag Ag( )NAg
Z t
0
e t= dt
| {z }
growthterm
(4.4)
where B0 is a constant which includes the X-ray  ux, transmission e ciency, accepted
angles, analyzer e ciency, all morphological and instrumental parameters which can be
reasonably assumed constant given uniform instrumental collection parameters, such as pass
energy and detection range during the sample analysis. NAg is the number of atoms/g,
Rt
0 e
 t= describes the cumulative probability with which elctrons are ejected form the sample
at depth t. It also describes the growth of the particles above atomic dispersion which was
assumed equal to 1, when particle growth is absent.
Similarly, ITiO2 photoelectron peak will be described by:
ITiO2 = B0 TiO2 TiO2NTiO2
Z 1
0
e t= dt (4.5)
So,
IAg
ITiO2 =
B0 Ag AgNAgRt0 e t= dt
B0 TiO2 TiO2NTiO2Rt=10 e t= dt (4.6)
54
Apply Equation 4.6 to the Ag/TiO2 system,
IAg
ITiO2 =
B0 Ag AgNAgRt0 e t= dt
B0 TiO2 TiO2NTiO2R10 e t= dt (4.7)
It is theorized that at low loading all atoms are accessible on the surface of the TiO2.
Thus at the linear section of the graph IAg=ITi vs Ag Loading (Figure 4.13) atoms were
assumed to be on the surface of the support. Also, the XPS signal arising from the support
will be constant as there was no diminution in the XPS signal of TiO2 from the increased
loading, between 0{20 wt% Ag on the surface of the TiO2.
IAg
ITiO2 =
  B0 Ag AgNAg     
  :1Rt=0
0 e
t= dt
  B0 TiO2 TiO2NTiO2
Z 1
0
e t= dt
| {z }
constant
=  Ag AgNAg 
TiO2 TiO2NTiO2
(4.8)
NTiO2 = N0TiO2
Z t=10
0
e t= dt (4.9)
NTiO2 was calculated at t =1. In XPS in nite depth t =1 was assumed to be
 t = 1000, at this value no photoelectron should be detected according to the governing
equations. An atomic layer was modeled to comprise of the total number of TiO2 atoms on
a square matrix of area 150 m2, see schematic representation and photoelectron intensity
probability function in Figure 4.12. Figure 4.12 (a) shows a simpli ed representation of the
layers of atoms of TiO2 and the graph in Figure 4.12 (b) shows the probability that an
electron will escape as a function of depth corresponding to  .
N0TiO2 = 1 1021
 = 18.5  A
55
Number of Layers = 1000
TiO2 (100) Interatomic spacing = 3.79  A
NTiO2 = 5:40 1021
The resulting theoretical IAg/ITiO2 is listed in Table 4.10.
3.79 ?A
2?
(a) Layers of representative atoms
0 1 2 3 4 50
0.2
0.4
0.6
0.8
1
?
Intensit
yprobabilit
y
(b) Photoelectron intensity probability function
Figure 4.12: Schematic of the TiO2 support
In Figure 4.13 the theoretical and experimental IAg/ITiO2 are plotted with respect to
wt% loading of Ag on TiO2. The solid lines with the diamond and oplus markers represent
the Ag/TiO2 ratio observed and the theoretical calculated values of the Ag/TiO2 ratio
respectively. In general all the theoretical calculated values are lower than the experimental
value; this suggests there is an error arising from one of the physical parameters. The
theoretical calculated values are the best case scenario and such the experimental line should
fall below the theoretical line. The two most likely sources of error lie in the calculation of the
mean free path and the estimation of N0 in Equation 4.9. The mean free path was calculated
from the semi-empirical formula developed by Penn [73]. The Penn formula assumes that
the atoms and molecules are homogeneously distribute throughout the material and that
56
0 2 4 6 8 10 12 14 16 18 20
0
0.1
0.2
0.3
0.4
0.5
0.6
?=18.5 ?A
Ag Loading on TiO 2 (wt %)
IAg
/I
TiO
2
Exp erimen tal Ag /TiO 2
Theoretical Ag /TiO 2
Figure 4.13: Ag/Ti ratio as a function of weight loading of Ag
Table 4.10: IAg/ITiO2 -Experimental versus Theoretical
Ag IAg/ITiO2
Loading (wt %) Experimental Theoretical
2 0.06 0.04
4 0.14 0.07
8 0.28 0.14
12 0.35 0.21
20 0.48 0.36
57
0 2 4 6 8 10 12 14 16 18 20
0
0.1
0.2
0.3
0.4
0.5
0.6
?=18.5 ?A
?=12 ?A
Ag Loading on TiO 2 (wt %)
IAg
/I
TiO
2
Exp erimen tal Ag /TiO 2
Theoretical Ag /TiO 2
Corrected Ag /TiO 2
Figure 4.14: Ag/Ti ratio as a function of weight loading of Ag
the material is 100% dense. While this is not the case, the value will be used as a starting
approximation. The value of N0 based on the interatomic spacing of the TiO2 of the (100)
plane.
A correction was made by assuming that at Ag loading less or equal to 4 wt% every
a signal from TiO2 molecule is sampled. Thus, the ultimate signal will be acquired and
experimental values scan only be less than or equal to this value. The dotted line in Figure
4.14 shows the corrected theoretical gradient. Using this gradient, a value of 12.0  A was
derived for the mean free path. The mean free path is highly surface dependent, thus
considering the amorphous nature, surface roughness, and the porosity of the adsorbent, the
derived mean free path can be used as a working basis approximation.
58
4.3 Particle Growth Estimation
Predicted XPS Intensity Ratio due to Particle Growth
If an assumption is made that the Ag atoms are all uniformly dispersed on the support,
then an estimation can be made to locate the point beyond which the Ag particles begin
to increase in size based on a comparison of the experimental IAg/ITiO2 and the calculated
IAg/ITiO2 assuming 100% accessibility of Ag atoms.
0 2 4 6 8 10 12 14 16 18 20
0
0.1
0.2
0.3
0.4
0.5
0.6
Ag Loading on TiO 2 (wt %)
IAg
/I
Ti
Ag /Ti ratio
Figure 4.15: Ag/Ti ratio as a function of weight loading of Ag
Table 4.11: Particle Growth - after 8 wt% Loading
Ag IAg/ITiO2 Predicted IAg/ITiO2
Loading (wt%) Experimental due to Particle Growth
8 0.28 |
12 0.35 0.39  0.04
20 0.48 0.56  0.07
59
The data displayed in Figure 4.15 suggests that all the Ag atoms are accessible up to
8 wt% loading after which the crystallites growth occurs. This is signi ed by the departure
from the linear behavior demonstrated from 0 wt% up to 8 wt%. Another linear trend is
displayed from 8 wt% to 20 wt% which indicates that the particles form crystallites that
are uniform in size and increase in occurrence with increased loading up to 20 wt%. The
IAg/ITiO2 experimental values form 0 { 8 wt% loadings lie on the on a line which typically
indicates uniform dispersion. Since non-linear behavior starts after 8 wt% loading, it was
assumed that nucleation occurred beyond this point. Table 4.11 lists the predicted IAg/ITiO2
values due to particle growth. The predicted values at 12 wt% fell at the lower end of
the uncertainty range while at the 20 wt% loading the predicted value was just outside the
uncertainty range.
0 2 4 6 8 10 12 14 16 18 20
0
0.1
0.2
0.3
0.4
0.5
0.6
Ag Loading on TiO 2 (wt %)
IAg
/I
Ti
Ag /Ti ratio
Figure 4.16: Ag/Ti ratio as a function of weight loading of Ag
Table 4.12 lists the predicted IAg/ITiO2 values when particle growth is assumed to after
4 wt% loading. These values are well within the uncertainty range. The results listed in
60
Tables 4.11 and 4.12 indicates that arguably the Ag particle growth could be somewhere
between 4 wt% than 8 wt%. Also, since the sulfur heterocyclic adsorption selectively has
the greatest capacity at about the 4 wt% loading a more con dent assertion is that particle
growth begins at this point.
Table 4.12: Particle Growth- after 4 wt% Loading
Ag IAg/ITiO2 Predicted IAg/ITiO2
Loading (wt%) Experimental due to Particle Growth
4 0.14 |
8 0.28 0.24  0.03
12 0.35 0.33  0.04
20 0.48 0.46  0.05
Predicted Particle Size
The dimension change of the crystallites were found by equating the particle growth
term as seen in Equation 4.4 to the IAg/ITiO2 experimental values. Particle growth was
assumed to begin after 4 wt% loading.
Table 4.13: XPS Ag Crystallite Growth Estimation
Ag Crystallite No. of atoms Total No. of
Loading (wt%) Size (nm) on Edge Atoms
8 0.71 2 8
12 0.84 3 27
20 1.11 4 64
The  nal dimensions can be calculated by accounting for the initial atomic size plus
the growth dimension. One Ag atomic radius is equivalent to approximately 0.35 nm. If a
crystallite at the 8 wt% loading contains 2 atoms on an edge of a cube then the total number
of atoms contained would be 8 atoms. The subsequent size at 8 { 20 wt% loadings are listed
in Table 4.13.
The diagrams in Figure 4.17 are a representation of the model surface of the adsorbent
at the various weight loading that have the characteristics that are consistent with the
61
TiOtwo.inferior Surface
2 wt % Ag/TiOtwo.inferior
4 wt % Ag/TiOtwo.inferior
Defects
TiOtwo.inferior Molecule
Ag Atom
Figure 4.17: Schematic representation of the most probable morphological con guration
of Ag=TiO2. 0 { 4 wt% Ag loading.
62
8 wt % Ag/TiOtwo.inferior
12 wt % Ag/TiOtwo.inferior
20 wt % Ag/TiOtwo.inferior
Defects
TiOtwo.inferior Molecule
Ag Atom
Ag Particle
Figure 4.18: Schematic representation of the most probable morphological con guration
of Ag=TiO28{ 20 wt% Ag loading.
63
experimental results of the XPS. At the 2 and 4 wt% loading it is believed that the atoms
are of the same size but increase in occurrence as the weight loading of silver increase. This
theory is due to the fact that 0, 2, and 4 wt% loadings line on the same line segment. At
 4 wt% loading the Ag crystallites occupy surface defects on the titanium surface and
all the atoms are settled on the surface. When the crystallites settle into the high energy
sites they will have a lower free energy and thus be more stable. It is theorized that all the
surface defected high energy sites are populated at about the 4 wt% loading and thus further
increase in loading results in particle growth.
At the 8 wt% loading the silver coverage is about 0.28 ML (refer to Figure 4.19), hence
it can be assumed that silver atoms can still occupy the available surface on the titanium
dioxide. However, there are no more defects available on the titanium sites that can hold
the individual silver atoms strongly. At the 8, 12, and 20 wt% loadings the crystal grow
progressively bigger as revealed by the particle size estimations.
64
Assuming Ag was 100 % dispersed, the monolayer thickness was calculated. Figure 4.19
shows the mole fraction of Ag detected and the mole fraction of Ag deposited with respect
to monolayer thickness and percent weight loading. The  gure indicated the e ectiveness of
the XPS at detected the actual number of moles of Ag deposited.
0 2 4 6 8 10 12 14 16 18 20
0
0.008
0.015
0.03
0.06
0.08
0.09
0.1
0.15
Ag Loading (wt %)
Mole
Fraction
of
Ag/
gof
Ag
/TiO
2
Ag Detected
Ag Deposited
0 0.07 0.14 0.28 0.44 0.81
ML Loading
Ag Detected
Figure 4.19: Monolayer detected versus moles deposited of Ag
65
4.4 EPR Analysis
Previous work using EPR, revealed that only  0.1 % of the total Ag was present as
Ag2+ on the 4 wt% Ag=TiO2 sorbent [21]. The shift towards lower B.E in the XPS spectra
of the Ag 3d peaks from the 4 wt% to the 20 wt% loading of Ag=TiO2 (see  gure 4.2)
prompted complementary investigation by EPR. Speci cally, the EPR signal for nanoparti-
cles of metallic Ag that may be evident at about 3250G and 3350 G [99].
The 2, 4, 8, 20 wt% loadings of Ag=TiO2 was analyzed in an attempt to resolve the EPR
line for Ag0 and Ag2+ species. The results were inconclusive with regards to the presence of
Ag0 line. EPR spectral lines at about 3000G and 3600 G were except due to Ag0 [100{102].
The Ag signal could not be resolved from the samples with sites found in literature [100{102].
The EPR signal for the Ag2+ species is most easily identi ed in the 4 wt% loading. This
result correlates with other  ndings [1] which leads to the conclusion that 4 wt% Ag=TiO2
has special properties that enhances its capacity for sulfur removal. In the 8 and 20 wt%
loadings the Ag line maybe present in the 3250 - 3400 G region but it is unobservable due
to overlapping with the Ti EPR line- the EPR line has been broaden line in this region.
This broadening is in uenced by the presence of NO bound to the bulk titania [103, 104].
Resolution of the Ag signal in the Ag/titania sorbent is di cult because the EPR signal
due to trapped nitrate and Ag/titania sorbent are identical. Previous work demonstrated
that there is a small concentration of NO bound to bulk TiO2 in the Ag=TiO2 sorbent
[104]. The EPR measurement temperature was varied in an e ort to distinguish the spectral
contributions. Spectra was taken at 77 K and 298 K.
Preliminary XAFS measurement [89] indicated the presence of a metallic contribution.
Why does XAFS indicate the presence of metallic silver when EPR does not? The answer lies
the intrinsic manner in which the techniques evaluates the presence of metallic Ag. Maybe
XAFS is more sensitive to the presence of metallic Ag than EPR.
66
2 wt.% Ag/TiO2
4 wt.% Ag/TiO2
8 wt.% Ag/TiO2
20 wt.% Ag/TiO2
Ag signal region
2500 2700 2900 3100 3300 3500 3700 3900
Magnetic Field (G)
Intensit
y(a.u.)
Figure 4.20: EPR spectra of the Ag=TiO2 at 77 K
67
20 wt.% AgTiO2
Ag signal
not observed
2500 2700 2900 3100 3300 3500 3700 3900
Magnetic Field (G)
Intensit
y(a.u.)
Figure 4.21: EPR spectra of the Ag=TiO2 at 298 K
68
Chapter 5
Conclusion
Fundamental studies aimed at acquiring a greater understanding of desulfurization ma-
terials were performed. The salient  ndings of this study are stated below.
The surface of the titania support from an XPS perspective is not appreciably shadowed
by the Ag particles up to 20 wt% loading. The maximum area occupied by silver particle
is less than approximately 6% of the total surface area of the titania assuming that the
crystallites are evenly spaced of spherical, hemispherical or cubic geometries. Homogenous
and mono dispersal of Ag on the support would result in coverage up to 0.81 ML at 20 wt%.
A functional relationship exists between the dispersed phase, the size of the silver particle
and the observed XPS intensity. Ag/TiO2 adsorbents display mono dispersion up to the 4
wt% loading. This corresponds to hypothesis that the silver atoms populate defects on the
titanium support up to this speci c weight loading. This 4 wt% loading is the critical loading
baseline after which the atoms begin to grow into bigger crystallites.
The characteristic property of sulfur selectivity is a combination of the interaction of
the silver oxide, the population of titania defects and the size of the silver crystallites. XPS
result indicated the presence of Ag in its oxide form. The XPS oxygen peak due assigned
to Ag1+ was the most dominant suggesting that Ag2O was in the majority.XPS results also
indicated that the activity of Ag/TiO2 is most likely optimal not only for a speci c size but
also when the defects on the titania support are populated.
Oxygen chemisorption should not be used as the sole means of determining the particle
size of Ag particles; the inherent assumptions may lead to inaccuracies in estimation. The
combination of XPS and chemisorption measurements are a more accurate indication of the
actual crystallite size. Oxygen chemisorption over-estimated the particle size for Ag particles
69
in the Ag/TiO2 system studied. XPS provided a more accurate representation of particle
size. The 2-10  A particle size range estimated from XPS is supported by the fact that the
Ag particle size could not be determined from XRD measurements, thus suggesting than the
particle sizes are less than 5 nm.
70
Chapter 6
Future Work
The ultimate goal of this project is to gain crucial information about the surface proper-
ties responsible for the Ag=TiO2 adsorbent?s capacity to desulfurize liquid phase high sulfur
logistic fuels. This knowledge is necessary to enhance performance and develop new sorbents.
The following are suggestions of future studies that can be done to contribute to a complete
understanding of the sorbent.
Isolation of Ag0 species by EPR. Implement a stepwise approach by trying a di erent
support to isolate and identify the Ag species. The support used should not be paramag-
netically active. Investigate the existence of Ag0 at 473 K signal that was reported in Wang
and Weh [102], which should be absent at room temperature due to sample exposure to air.
Evaluation of TiO2 native oxygen uptake. Account for oxygen uptake using oxygen
chemisorption on the native or nitrated TiO2 surface and re-evaluate crystallite size approx-
imations.
Temperature programmed analysis. Exploration of the desorption kinetics, surface
reaction mechanisms, nature of the surface intermediates and surface acidity. Investigate
the acidity of Ag=TiO2 adsorbents by TPD/TPRS using probe molecules such as ammonia
and pyridine that provide selective interaction with the surface sites. Ammonia can be used
to indicate strong and weak Br nsted and Lewis sites while pyridine can be to show strong
and weak Br nsted and strong and weak Lewis sites, thereby enabling the discernment of
the weak Br nsted site [105{108].
71
Temperature programmed XPS of the chemisorption assumptions and conditions. This
would provide a better understanding of oxygen chemisorption when used with silver parti-
cles.
Investigation of the adsorption process and assessment of the removal pathway.
The introduction of selective refractory heterocyclic species, such as thiophene and benzoth-
iophene and in TPD/TPRS will enable the assessment of speci city of various heterocyclic
compounds. Since it is known that the surface of the Ag strongly in uences the sorbent?s
a nity for the refractory heterocyclic compounds. It is reported that Ag forms step edges
at which the sulfur binds on Ag surfaces [109, 110] thus, it is likely that it is the structure
of the edges that facilitate the ?lock? for these big refractory compounds.
Additional XPS characterization. Perform XPS investigation on the adsorbent system
Ag/TiO2 using a more sensitive system than the non-monochromated LHS-10. XPS has
become even more powerful with new technological advancements: higher counting rates,
more sensitive detection limits, narrower FWHM, greater signal-to noise ratios. These new
advances have enabled acquisition of more fundamental surface chemical information. Thus,
more information can be extracted by the technique since limitations encountered were due
to the current XPS system and the technique.
Examination of the Ag Auger energy levels to distinguish the factors such as static
relaxation term in uencing chemical shifts in the Ag/TiO2 adsorbent. Comparison of the
Auger energy shifts and binding energy shift can lead to the identi cation of the Ag chemical
environment.
Also, XPS can be used as a complementary technique to TPD/TPRS study surface
acidity of Ag=TiO2. This technique has been used to study the adsorption of probe molecules
such as: benzene, pyridine, and pyrrole in literature [95, 111, 112]. Speci cally, perform alkali
treatment to modify acid-base properties to investigate surface acidity.
72
Complementary crystallite size determination. Use scanning probe microscopy (SPM)
to get an independent determination of crystallite size. SPM uses a  ne probe to image sur-
faces at the nanometer scale. This technique is not restricted by the wavelength of light or
electrons and can resolve atoms and true 3-D maps of surfaces.
Sintering study. Perform a sintering study to show the nucleation growth of the mobile
silver particles. Anneal the Ag=TiO2 samples at various temperatures for example 400, 450,
500 C, etc. Analyse the samples using XPS and plot the IAg/ITiO2 as a function of the
annealing temperature.
73
Bibliography
[1] S. Nair and B. J. Tatarchuk, \Supported Silver Adsorbents for Selective Removal of
Sulfur Species from Hydrocarbon Fuels," Fuel, vol. 89, no. 11, pp. 3218{3225, 2010.
[2] U. E. P. A. (EPA), \Fuel Sulfur Standards," 01 2009. Code of Federal Regulations
(CFR) citations: 40 CFR Part 80 Subpart H.
[3] C. A. I. for Asian Cities (CAI-Asia), \Current and Proposed Sulfur levels in Diesel in
Asia, EU and USA.," 2009.
[4] U.S. Energy Information Administration, \Light-Duty Diesel Vehicles: E ciency and
Emissions Attributes and Market Issues," tech. rep., U.S. Energy Information Admin-
istration, 2009. SR/OIAF(2009)02.
[5] \Part 80 - Regulation of Fuels and Fuel Additives," Federal Register, vol. 65, pp. 6822{
6870, February 2000.
[6] X. Ma, M. Sprague, and C. Song, \Deep Desulfurization of Gasoline by Selective
Adsorption over Nickel-Based Adsorbent for Fuel Cell Applications," Industrial &
Engineering Chemistry Research, vol. 44, no. 15, pp. 5768{5775, 2005.
[7] C. Song, \An Overview of New Approaches to Deep Desulfurization for Ultra-Clean
Gasoline, Diesel Fuel and Jet Fuel," Catalysis Today, vol. 86, no. 1-4, pp. 211 { 263,
2003. E ects of Support in Hydrotreating Catalysis for Ultra-clean Fuels.
[8] S. Velu, X. Ma, and C. Song, \Selective Adsorption for Removing Sulfur from Jet Fuel
over Zeolite-Based Adsorbents," Industrial & Engineering Chemistry Research, vol. 42,
no. 21, pp. 5293{5304, 2003.
74
[9] X. Ma, S. Velu, J. H. Kim, and C. Song, \Deep Desulfurization of Gasoline by Selective
Adsorption over Solid Adsorbents and Impact of Analytical Methods on ppm-level
Sulfur Quanti cation for Fuel Cell Applications," Applied Catalysis B: Environmental,
vol. 56, no. 1-2, pp. 137 { 147, 2005.
[10] A. J. Hernandez-Maldonaldo and R. Yang, \Desulfurization of Transportation Fuels
by Adsorption," Catalysis Reviews, vol. 46, no. 2, pp. 111{150, 2004.
[11] E. Ito and J. R. van Veen, \On novel Processes for Removing Sulphur from Re nery
Streams," Catalysis Today, vol. 116, no. 4, pp. 446 { 460, 2006.
[12] P. T. Vasudevana and J. L. G. Fierro, \A Review of Deep Hydrodesulfurization Catal-
ysis," Catalysis Reviews: Science and Engineering, vol. 38, pp. 161{188, May 1996.
[13] X. Ma, K. Sakanishi, and I. Mochida, \Hydrodesulfurization Reactivities of Various
Sulfur Compounds in Vacuum Gas Oil," Industrial & Engineering Chemistry Research,
vol. 35, no. 8, pp. 2487{2494, 1996.
[14] S. Haji and C. Erkey, \Removal of Dibenzothiophene from Model Diesel by Adsorption
on Carbon Aerogels for Fuel Cell Applications," Industrial & Engineering Chemistry
Research, vol. 42, no. 26, pp. 6933{6937, 2003.
[15] A. J. Hernndez-Maldonado and R. T. Yang, \Desulfurization of Liquid Fuels by Ad-
sorption via Complexation with Cu(I)Y and AgY Zeolites," Industrial & Engineering
Chemistry Research, vol. 42, no. 1, pp. 123{129, 2003.
[16] R. V. Siriwardane and J. A. Poston, \Characterization of Copper Oxides, Iron Oxides,
and Zinc Copper Ferrite Desulfurization Sorbents by X-ray Photoelectron Spectroscopy
and Scanning Electron Microscopy," Applied Surface Science, vol. 68, no. 1, pp. 65 {
80, 1993.
75
[17] J. H. Kim, X. Ma, A. Zhou, and C. Song, \Ultra-deep Desulfurization and Denitro-
genation of Diesel Fuel by Selective Adsorption over Three Di erent Adsorbents: A
Study on Adsorptive Selectivity and Mechanism," Catalysis Today, vol. 111, no. 1-2,
pp. 74 { 83, 2006.
[18] S. Velu, X. Ma, C. Song, M. Namazian, S. Sethuraman, and G. Venkataraman, \Desul-
furization of JP-8 Jet Fuel by Selective Adsorption over a Ni-based Adsorbent for Micro
Solid Oxide Fuel Cells," Energy & Fuels, vol. 19, pp. 1116{1125, 2005.
[19] D. L. King and L. Li, \Removal of Sulfur Components from Low Sulfur Gasoline using
Copper Exchanged Zeolite Y at Ambient Temperature," Catalysis Today, vol. 116,
no. 4, pp. 526 { 529, 2006.
[20] B.Tatarchuk, H. Yang, and S. Nair, \Silver-Based Sorbents," April 2008. US
2008/0283446 A1.
[21] A. Samokhvalov, S. Nair, E. C. Duin, and B. J. Tatarchuk, \Surface Characterization
of Ag/Titania Adsorbents," Applied Surface Science, vol. 256, no. 11, pp. 3647{3652,
2010.
[22] J. R. Anderson, Structure of Metallic Catalysts. Academic Press Inc, 1975.
[23] M. A. Wahab, Solid State Physics: Structure and Properties of Materials. Alpha
Science International Ltd, 2nd ed., 2005.
[24] C. Brooks and G. Christopher, \Measurement of the State of Metal Dispersion on
Supported Nickel Catalysts by Gas Chemisorption," Journal of Catalysis, vol. 10,
no. 3, pp. 211{223, 1968.
[25] B. G. Linsen, ed., Physical and Chemical Aspects of Adsorbents and Catalysts : Ded-
icated to J. H. de Boer on the Occasion of His Retirement from the Technological
University, Delft, the Netherlands. Academic Press, 1970.
76
[26] T. E. Whyte, \Metal Particle Size Determination of Supported Metal Catalysts,"
Catalysis Reviews, vol. 8, no. 1, pp. 117{134, 1974.
[27] D. G. Mustard and C. H. Bartholomew, \Determination of Metal Crystallite Size
and Morphology in Supported Nickel Catalysts," Journal of Catalysis, vol. 67, no. 1,
pp. 186{206, 1981.
[28] D. E. Strohmayer, G. L. Geo roy, and M. A. Vannice, \Measurement of Silver Surface
area by the H2 Titration of Chemisorbed Oxygen," Applied Catalysisreporte, vol. 7,
no. 2, pp. 189{198, 1983.
[29] S. Seyedmonir, D. E. Strohmayer, G. L. Geo roy, M. Vannice, H. W. Young, and
J. W. Linowski, \Characterization of Supported Silver Catalysts: I. Adsorption of
O2;H2;N2O; and the H2-Titration of Adsorbed Oxygen on Well-Dispersed Ag on
TiO2," Journal of Catalysis, vol. 87, no. 2, pp. 424{436, 1984.
[30] B. Kip, F. Duivenvoorden, D. Koningsberger, and R. Prins, \Determination of Metal
Particle Size of Highly Dispersed Rh, Ir, and Pt Catalysts by Hydrogen Chemisorption
and EXAFS," Journal of Catalysis, vol. 105, no. 1, pp. 26{38, 1987.
[31] J. Regalbuto, T. Fleisch, and E. Wolf, \An Integrated Study of Pt=WO3=SiO2 Cata-
lysts for the NO-CO Reaction: I. Catalyst Characterization by XRD, Chemisorption,
and XPS," Journal of Catalysis, vol. 107, no. 1, pp. 114{128, 1987.
[32] D. Farin and D. Avnir, \Crystallite Size E ects in Chemisorption on Dispersed Met-
als," Journal of Catalysis, vol. 120, no. 1, pp. 55 { 67, 1989.
[33] R. Wojcieszak, M. J. Genet, P. Eloy, P. Ruiz, and E. M. Gaigneaux, \Determination of
the Size of Supported Pd Nanoparticles by X-ray Photoelectron Spectroscopy. Compar-
ison with X-ray Di raction, Transmission Electron Microscopy, and H2 Chemisorption
Methods," The Journal of Physical Chemistry C, vol. 114, no. 39, pp. 16677{16684,
2010.
77
[34] J. Scholten, J. Konvalinka, and F. Beekman, \Reaction of Nitrous Oxide and Oxygen
with Silver Surfaces, and Application to the Determination of Free-Silver Surface Areas
of Catalysts," Journal of Catalysis, vol. 28, no. 2, pp. 209{220, 1973.
[35] K. M. Kholyavenko, M. Y. Rubanik, and N. Chernukhina, \Determination of the
Surface Area of Silver Deposited on a Carrier by Chemisorption," Kinetika i Kataliza
(Kinectics and Catalysis), vol. 5, no. 3, pp. 505{512, 1964.
[36] A. W. Czanderna, \The Adsorption of Oxygen on Silver," The Journal of Physical
Chemistry, vol. 68, no. 10, pp. 2765{2771, 1964.
[37] M. Barteau and R. Madix, The Surface Reactivity of Silver: Oxidation Reactions in:
The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis, ch. 4, pp. 95{142.
Elsevier, 1982.
[38] J. S. Brinen, \Applications of ESCA to Industrial Chemistry," Journal of Electron
Spectroscopy and Related Phenomena, vol. 5, no. 1, pp. 377{400, 1974.
[39] L. Scharpen, \The Dispersion of Platinum on Silica-Correlation of ESCA and Gas
Adsorption Data," Journal of Electron Spectroscopy and Related Phenomena, vol. 5,
no. 1, pp. 369{376, 1974.
[40] D. Briggs, \ESCA and Metal Crystallite Size/Dispersion in Catalysts," Journal of
Electron Spectroscopy and Related Phenomena, vol. 9, no. 5, pp. 487{491, 1976.
[41] R. Shalvoy and P. Reucroft, \Quantitative Analysis of ESCA Signal Intensi es from
Coprecipitated Nickel on Alumina Catalysts," Journal of Electron Spectroscopy and
Related Phenomena, vol. 12, no. 3, pp. 351{356, 1977.
[42] M. Houalla and B. Delmon, \Use of XPS to Detect Variations in Dispersion of Impreg-
nated and Ion-exchanged NiO=SiO2 Systems," Surface and Interface Analysis, vol. 3,
no. 3, pp. 103{105, 1981.
78
[43] C. Defosse, M. Houalla, A. Lycourghiotis, and F. Delannay, \Joint Use of X-Ray
Photoelectron Spectroscopy and Analytical Electron Microscopy in the Investigation
of Cobalt and Nickel Oxide Supported on Na-Doped Alumina," in New Horizons in
Catalysis Proceedings of the 7th International Congress on Catalysis (T. Seivama and
K. Tanabe, eds.), vol. 7, Part A of Studies in Surface Science and Catalysis, pp. 108{
121, Elsevier, 1981.
[44] Z. Liu, Z. Lin, H. Fan, F. Li, Q. Bao, and S. Zhang, \Dispersion of V2O5 Supported on
a TiO2 Surface by X-Ray Photoelectron Spectroscopy," Applied Physics A: Materials
Science & Processing, vol. 45, pp. 159{164, 1988.
[45] V. D. Castro, C. Furlani, M. Gargano, N. Ravasio, and M. Rossi, \XPS Study of
Copper Dispersion in CuO=Al2O3 Catalysts," Journal of Electron Spectroscopy and
Related Phenomena, vol. 52, no. 0, pp. 415{422, 1990.
[46] S. Kaliaguine, \Application of Surface Science Techniques in the Field of Zeolitic
Materials," in Recent Advances and New Horizons in Zeolite Science and Technology
(S. W. H. Chon and S.-E. Park, eds.), vol. 102 of Studies in Surface Science and
Catalysis, pp. 191{230, Elsevier, 1996.
[47] B. M. Reddy, B. Chowdhury, E. P. Reddy, and A. Fernndez, \An XPS Study of Dis-
persion and Chemical State of MoO3 on Al2O3-TiO2 Binary Oxide Support," Applied
Catalysis A: General, vol. 213, no. 2, pp. 279{288, 2001.
[48] J. Brinen, J. Schmitt, W. Doughman, P. Achorn, L. Siegel, and W. Delgass, \X-
ray Photoelectron Spectroscopy Studies of the Rhodium on Charcoal Catalyst: II.
Dispersion as a Function of Reduction," Journal of Catalysis, vol. 40, no. 3, pp. 295{
300, 1975.
[49] S. C. Fung, \Application of XPS to the Determination of the Size of Supported Par-
ticles in a Catalyst{Model Development and its Application to Describe the Sintering
79
Behavior of a Silica-Supported Pt Film," Journal of Catalysis, vol. 58, no. 3, pp. 454{
469, 1979.
[50] S. Kaliaguine, A. Adnot, and G. Lemay, \A Model for the Quantitative Analysis of
ESCA Intensity Ratios for Supported Catalysts with Partial Surface Segregation," The
Journal of Physical Chemistry, vol. 91, no. 11, pp. 2886{2890, 1986.
[51] S. Davis, \Particle Size Information from Dispersed Phase Photoemission Intensity
Ratios," Journal of Catalysis, vol. 117, no. 2, pp. 432{446, 1989.
[52] S. Kaliaguine, A. Adnot, G. Lemay, and L. Rodrigo, \On the Quantitative Analysis
of XPS Intensity Data for Supported Catalysts with Partial Surface Segregation,"
Journal of Catalysis, vol. 118, no. 1, pp. 275{279, 1989.
[53] E. Voogt, A. Mens, O. Gijzeman, and J. Geus, \XPS Analysis of Palladium Oxide
Layers and Particles," Surface Science, vol. 350, no. 13, pp. 21{31, 1996.
[54] J. Watts, \Surface Analysis | X-Ray Photoelectron Spectroscopy," in Encyclopedia of
Analytical Science (P. Worsfold, A. Townshend, and C. Poole, eds.), vol. 8, pp. 5047{
5058, Oxford: Elsevier, 1st ed., 1995.
[55] J. E. Fulghum, \Recent Developments in High Energy and Spatial Resolution Anal-
ysis of Polymers by XPS," Journal of Electron Spectroscopy and Related Phenomena,
vol. 100, no. 13, pp. 331{355, 1999.
[56] S. Tougaard, \Surface Analysis | X-Ray Photoelectron Spectroscopy," in Encyclopedia
of Analytical Science (P. Worsfold, A. Townshend, , and C. Poole, eds.), pp. 446{456,
Oxford: Academic Press, 2nd ed., 2005.
[57] J. Walton and N. Fairley, \Characterisation of the Kratos Axis Ultra with Spherical
Mirror Analyser for XPS Imaging," Surface and Interface Analysis, vol. 38, no. 8,
pp. 1230{1235, 2006.
80
[58] J. I. Langford and A. J. C. Wilson, \Scherrer after Sixty Years: A Survey and Some
New Results in the Determination of Crystallite Size," Journal of Applied Crystallog-
raphy, vol. 11, pp. 102{113, Apr 1978.
[59] C. Suryanarayana and M. G. Norton, X-ray Di raction: A Practical Approach.
Springer, 1st ed., 1998.
[60] B. Cullity and S. Stock, Elements of X-Ray Di raction. Prentice Hall, 3rd ed., 2001.
[61] J. Angevine, W. N.Delgass, and J. C.Vartuli, \Dispersion and Uniformity of Supported
Catalyst by X-ray Photoelectron Spectroscopy," Proceedings of the 6th International
Congress on Catalysis, vol. 2, pp. 611{620, 1976.
[62] P. Georgopoulos and J. B. Cohen, \Study of Supported Platinum Catalysts by Anoma-
lous Scattering," Journal of Catalysis, vol. 92, no. 2, pp. 211{215, 1985.
[63] M. Treacy and A. Howie, \Contrast E ects in the Transmission Electron Microscopy of
Supported Crystalline Catalyst Particles," Journal of Catalysis, vol. 63, no. 1, pp. 265{
269, 1980.
[64] J. Sanders and K. Pratt, \The Relationship of Structure and Activity of NiMo Sul des
to Composition of the Precursor Oxides," Journal of Catalysis, vol. 67, no. 2, pp. 331{
347, 1981.
[65] S. Nair, \Personal Communication.".
[66] F. P. J. M. Kerkhof and J. A. Moulijn, \Quantitative Analysis of XPS Intensities for
Supported Catalysts," The Journal of Physical Chemistry, vol. 83, no. 12, pp. 1612{
1619, 1979.
[67] X. E. Verykios, F. P. Stein, and R. W. Coughlin, \In uence of Metal Crystallite Size
and Morphology on Selectivity and Activity of Ethylene Oxidation Catalyzed by Sup-
ported Silver," Journal of Catalysis, vol. 66, no. 2, pp. 368{382, 1980.
81
[68] Y. Okamoto, K. Oh-Hiraki, T. Imanaka, and S. Teranishi, \X-ray Photoelec-
tron Spectroscopic Study of Mixed Oxide Catalysts Containing Molybdenum: I.
SnO2 MoO3Catalysts," Journal of Catalysis, vol. 71, no. 1, pp. 99{110, 1981.
[69] R. Baker, E. Prestridge, and L. Murrell, \Electron Microscopy of Supported Metal
Particles: III. The Role of the Metal in an SMSI Interaction," Journal of Catalysis,
vol. 79, no. 2, pp. 348{358, 1983.
[70] J. Cohen, \X-ray Di raction Studies of Catalysts," Ultramicroscopy, vol. 34, no. 12,
pp. 41{46, 1990.
[71] X. Shen, L.-J. Garces, Y. Ding, K. Laubernds, R. P. Zerger, M. Aindow, E. J. Neth,
and S. L. Suib, \Behavior of H2 Chemisorption on Ru/TiO2 Surface and its Application
in Evaluation of Ru Particle Sizes Compared with TEM and XRD Analyses," Applied
Catalysis A: General, vol. 335, no. 2, pp. 187{195, 2008.
[72] X. Zhu, R. Birringer, U. Herr, and H. Gleiter, \X-ray Di raction Studies of the Struc-
ture of Nanometer-sized Crystalline Materials," Phys. Rev. B, vol. 35, pp. 9085{9090,
Jun 1987.
[73] D. R. Penn, \Quantitative Chemical Analysis by ESCA," Journal of Electron Spec-
troscopy and Related Phenomena, vol. 9, no. 1, pp. 29{40, 1976.
[74] D. Briggs and M. P. Seah, eds., Practical Surface Analysis: by Auger and X-ray Pho-
toelectron Spectroscopy. Chichester: Wiley, 1983.
[75] G. Wertheim, \New Method for Bulk Quantitative Analysis by ESCA," Journal of
Electron Spectroscopy and Related Phenomena, vol. 50, no. 1, pp. 31{38, 1990.
[76] C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder, and G. E.Muilenberg, Hand-
book of X-ray Photoelectron Spectroscopy. Physical Electronics Division. Perkin-Elmer
Corp., 1978.
82
[77] C. D. Wagner, L. E. Davis, M. V. Zeller, J. A. Taylor, R. H. Raymond, and L. H.
Gale, \Empirical Atomic Sensitivity Factors for Quantitative Analysis by Electron
Spectroscopy for Chemical Analysis," Surface and Interface Analysis, vol. 3, no. 5,
pp. 211{225, 1981.
[78] C. Battistoni, G. Mattogno, and E. Paparazzo, \Quantitative Surface Analysis by
XPS: A Comparison among Di erent Quantitative Approaches," Surface and Interface
Analysis, vol. 7, no. 3, pp. 117{121, 1985.
[79] S. Kumar, J. V. Ramana, V. S. Raju, J. Arunachalam, and S. Gangadharan, \De-
termination of Stoichiometry of Cadmium Telluride by XPS," Fresenius? Journal of
Analytical Chemistry, vol. 343, pp. 879{880, 1992.
[80] D. Hayward and B. Trapnell, Chemisorption. Butterworth and Co. Ltd, 2nd ed., 1964.
[81] G. Wedler, Chemisorption: An Experimental Approach. Butterworth and Co. Ltd,
1970.
[82] I. E. Wachs, ed., Characterization of Catalytic Materials. Butterworth-Heinemann,
1992.
[83] W. W. Smeltzer, E. L. Tollefson, and A. Cambron, \Adsorption of Oxygen by a Silver
Catalyst," Canadian Journal of Chemistry, vol. 34, no. 8, pp. 1046{1060, 1956.
[84] J. A. Weil, J. R. Bolton, and J. E. Wertz, Electron Paramagnetic Resonance : Ele-
mentary Theory and Practical Applications. John Wiley & Sons, Inc, 1994.
[85] S. W. Gaarenstroom and N. Winograd, \Initial and Final State E ects in the ESCA
Spectra of Cadmium and Silver Oxides," Journal of Chemical Physics, vol. 67, no. 10,
pp. 3500{3506, 1977.
83
[86] K. Kim and N. Winograd, \X-ray Photoelectron Spectroscopic Binding Energy Shifts
due to Matrix in Alloys and Small Supported Metal Particles," Chemical Physics
Letters, vol. 30, no. 1, pp. 91{95, 1975.
[87] G. Sch on, \ESCA Studies of Ag,Ag2O and AgO," Acta Chemica Scandinavica, vol. 27,
pp. 2623{2633, 1973.
[88] J. S. Hammond, S. W. Gaarenstroom, and N. Winograd, \X-ray Photoelectron Spec-
troscopic Studies of Cadmium- and Silver-Oxygen Surfaces," Analytical Chemistry,
vol. 47, no. 13, pp. 2193{2199, 1975.
[89] J. Heinzel, \EXAFS Analysis of Supported Ag for Liquid-Phase Logistics Fuel Desul-
furization.".
[90] F. Delannay, C. Defosse, B. Delmon, P. G. Menon, and G. F. Froment, \Chloriding of
Pt Al2O3 Catalysts. Studies by Transmission Electron Microscopy and X-ray Pho-
toelectron Spectroscopy," Industrial & Engineering Chemistry Product Research and
Development, vol. 19, no. 4, pp. 537{541, 1980.
[91] H. P. C. E. Kuipers, H. C. E. Van Leuven, and W. M. Visser, \The Characterization of
Heterogeneous Catalysts by XPS based on Geometrical Probability. 1: Monometallic
Catalysts," Surface and Interface Analysis, vol. 8, no. 6, pp. 235{242, 1986.
[92] A. Meijers, A. de Jong, L. van Gruijthuijsen, and J. Niemantsverdriet, \Preparation of
Zirconium Oxide on Silica and Characterization by X-ray Photoelectron Spectroscopy,
Secondary Ion Mass Spectrometry, Temperature Programmed Oxidation and Infra-red
Spectroscopy," Applied Catalysis, vol. 70, no. 1, pp. 53{71, 1991.
[93] P. W. Park and J. S. Ledford, \Characterization and CH4 Oxidation Activity of
Cr=Al2O3 Catalysts," Langmuir, vol. 13, no. 10, pp. 2726{2730, 1997.
84
[94] G. C. Smith, Surface Analysis by Electron Spectroscopy: Measurement and Interpreta-
tion (Updates in Applied Physics and Electrical Technology). Springer, 1994.
[95] P. O. Scokart, A. Amin, C. Defosse, and P. G. Rouxhet, \Direct Probing of the Sur-
face Properties of Alkali-Treated Aluminas by Infrared and X-ray Photoelectron Spec-
troscopy," The Journal of Physical Chemistry, vol. 85, no. 10, pp. 1406{1412, 1981.
[96] J. Hamilton and P. Logel, \Nucleation and Growth of Ag and Pd on Amorphous
Carbon by Vapor Deposition," Thin Solid Films, vol. 16, no. 1, pp. 49{63, 1973.
[97] K. D. Sattler, ed., Handbook of Nanophysics: Functional Nanomaterials. CRC Press,
2010.
[98] D. Zhang, X. Yang, J. Zhu, Y. Zhang, P. Zhang, and G. Li, \Graphite-like Carbon De-
posited Anatase TiO2 Single Crystals as E cient Visible-light Photocatalysts," Jour-
nal of Sol-Gel Science and Technology, vol. 58, pp. 594{601, 2011.
[99] M. Danilczuk, A. Lund, J. Sadlo, H. Yamada, and J. Michalik, \Conduction Electron
Spin Resonance of Small Silver Particles," Spectrochimica Acta Part A: Molecular and
Biomolecular Spectroscopy, vol. 63, no. 1, pp. 189{191, 2006.
[100] B. Bales and L. Kevan, \EPR of Ag0 Site Conversion in Gamma-irradiated Frozen
AgNO3 Ices," Chemical Physics Letters, vol. 3, no. 7, pp. 484{487, 1969.
[101] G. E. Holmberg, W. P. Unruh, and R. J. Friauf, \An ESR and ENDOR Study of the
Ag0 Center in KCl and NaCl," Phys. Rev. B, vol. 13, pp. 983{992, 1976.
[102] Y.-P. Wang and C.-T. Yeh, \Electron Paramagnetic Resonance Study of the Inter-
actions of Oxygen with Silver/Titania," Journal of the Chemical Society, Faraday
Transactions, vol. 87, pp. 345{348, 1991.
[103] S. Livraghi, M. Paganini, M. Chiesa, and E. Giamello, \Trapped Molecular Species in
N-doped TiO2," Research on Chemical Intermediates, vol. 33, pp. 739{747, 2007.
85
[104] A. Samokhvalov, E. C. Duin, S. Nair, M. Bowman, Z. Davis, and B. J. Tatarchuk,
\Study of the Surface Chemical Reactions of Thiophene with Ag/Titania by the
Complementary Temperature-Programmed Electron Spin Resonance, Temperature-
Programmed Desorption, and X-ray Photoelectron Spectroscopy: Adsorption, Des-
orption, and Sorbent Regeneration Mechanisms," The Journal of Physical Chemistry
C, vol. 114, no. 9, pp. 4075{4085, 2010.
[105] H. G. Karge and V. Dondur, \Investigation of the Distribution of Acidity in Zeo-
lites by Temperature-Programmed Desorption of Probe Molecules. I. Dealuminated
Mordenites," The Journal of Physical Chemistry, vol. 94, no. 2, pp. 765{772, 1990.
[106] C. U. I. Odenbrand, J. G. M. Brandin, and G. Busca, \Surface Acidity of Silica-Titania
Mixed Oxides.," ChemInform, vol. 23, no. 36, pp. no{no, 1992.
[107] A. Auroux, R. Monaci, E. Rombi, V. Solinas, A. Sorrentino, and E. Santacesaria,
\Acid Sites Investigation of Simple and Mixed Cxides by TPD and Microcalorimetric
techniques," Thermochimica Acta, vol. 379, no. 1-2, pp. 227 { 231, 2001.
[108] F. Arena, R. Dario, and A. Parmaliana, \A Characterization Study of the Surface
Acidity of Solid Catalysts by Temperature Programmed Methods," Applied Catalysis
A: General, vol. 170, no. 1, pp. 127 { 137, 1998.
[109] P. A. Thiel, M. Shen, D.-J. Liu, and J. W. Evans, \Adsorbate-Enhanced Transport of
Metals on Metal Surfaces: Oxygen and Sulfur on Coinage Metals," Journal of Vacuum
Science & Technology A: Vacuum, Surfaces, and Films, vol. 28, no. 6, pp. 1285{1298,
2010.
[110] M. Shen, C. Jenks, J. Evans, and P. Thiel, \How Sulfur Controls Nucleation of Ag
Islands on Ag(111)," Topics in Catalysis, vol. 54, pp. 83{89, 2011.
86
[111] R. Borade, A. Sayari, A. Adnot, and S. Kaliaguine, \Characterization of Acidity in
ZSM-5 Zeolites: An X-ray Photoelectron and IR Spectroscopy Study," The Journal of
Physical Chemistry, vol. 94, no. 15, pp. 5989{5994, 1990.
[112] C. Defoss e, P. Canesson, P. Rouxhet, and B. Delmon, \Surface Characterization of
Silica-Aluminas by Photoelectron Spectroscopy," Journal of Catalysis, vol. 51, no. 2,
pp. 269{277, 1977.
[113] JabRef Development Team, JabRef. JabRef Development Team, 2010.
[114] G. Somorjai and Y. Li, Introduction to Surface Chemistry and Catalysis. John Wiley
& Sons Limited, 2nd ed., 2010.
[115] R. J. Culbertson, L. C. Feldman, P. J. Silverman, and H. Boehm, \Epitaxy of Au on
Ag(111) Studied by High-Energy Ion Scattering," Phys. Rev. Lett., vol. 47, pp. 657{
660, 1981.
[116] D. R. Lide, ed., CRC Handbook of Chemistry and Physics. CRC Press, 86 ed., 2005.
[117] J. H. Sco eld, \Hartree-Slater Subshell Photoionization Cross-sections at 1254 and
1487 eV," Journal of Electron Spectroscopy and Related Phenomena, vol. 8, no. 2,
pp. 129{137, 1976.
[118] C. Powell, \Attenuation Lengths of Low-Energy Electrons in Solids," Surface Science,
vol. 44, no. 1, pp. 29{46, 1974.
[119] G. G. Fuentes, E. Elizalde, F. Yubero, and J. M. Sanz, \Electron Inelastic Mean Free
Path for Ti, TiC, TiN and TiO2 as Determined by Quantitative Re ection Electron
Energy-Loss Spectroscopy," Surface and Interface Analysis, vol. 33, no. 3, pp. 230{237,
2002.
87
Appendices
88
Appendix A
Physical Properties
Useful Silver Properties
Ag metallic radius 1.45  A
Ag1+ ionic radius 1.26  A
Ag2+ ionic radius 1.01  A
Ag atomic weight 107.867 g/mol
Ag density 1.05 107 g/m3
Ag NN interatomic distance 2.889  A
Ag(111) fcc surface spacing [114, 115] 2.35  0.1  A
Ag(110) fcc surface spacing [114] 1.33  0.04  A
Ag fcc lattice parameter [116] 4.08  A
Ag active surface atom cross-sectional area [1] 8.6960  A2/Ag atom
Ag photoemission cross-section Ag 3d5=2 Al-K [117] 1.45 10 23 m2
10.66 13600 barns
Ag photoemission cross-section Ag 3d3=2 Al-K [117] 1.00 10 23 m2
7.38 13600 barns
Ag photoemission cross-section Ag 3d5=2 Mg-K [117] 2.37 10 23 m2
Mean free path (1118 eV) Ag 3d5=2 Al-K ,  [73] 1.38  10 9 m
89
Useful Titanium Properties
TiO2 molecular weight 79.9 g/mol
Ti density 4.23 106 g/cm3
Ti photoemission cross-section Ti 2p1=2 Al-K [117] 3.66 10 24 m2
2.69 13600 barns
Ti photoemission cross-section Ti 2p3=2 Al-K [117] 7.10 10 24 m2
5.22 13600 barns
Mean free path (1118 eV) Ti 2p1=2 Al-K ,  [73] 1.62  10 9 m
90
Other Useful Properties
Al-K line width [74] 0.85 eV
Al-K line x-ray energy 1486.6 eV
Mg-K x-ray energy 1253.6 eV
Mg-K line width[74] 0.70 eV
Bohr radius,a0, 0.529  A
91
Appendix B
Sample Calculations
B.1 Calculating the Ag Mole Fraction in Silver-Titania Adsorbent
Physical Data - TiO2
Molecular Weight, MM TiO2: 79.9 g/mol
Density  ; TiO2: 4.23 g/cm3
SA TiO2: 150 m2/g
Total pore volume TiO2: 0.38 cm2/g
MM AgNO3: 169.87 g/mol
Sorbent Preparation for 4 wt % Ag=TiO2: 5.8 mL of 1M AgNO3 solution is deposited on
15 g TiO2
There are 0.0058 moles of AgNO3 in a 5.8 mL 1M AgNO3 solution
# of moles of TiO2 = 15 g 1 mol79:9 g=mol = 0:1877 moles
XAgNO3 = 0:0058 moles AgNO30:0058 moles AgNO
3 + 0:1877 moles TiO2
= 0:03
XTiO2 = 0:1877 moles TiO20:0058 moles AgNO
3 + 0:1877 moles TiO2
= 0:97
Table B.1: Atomic Ratio Ag=TiO2 Sorbent
Loading (wt %) 1 2 4 8 12 20
TiO2 0.992 0.985 0.97 0.94 0.91 0.85
Ag 0.008 0.015 0.03 0.06 0.09 0.15
92
Table B.2: Atomic Ratio of Surface Species on Ag=TiO2 Adsorbent Pellets
Ag=TiO2 Loading
Element 1wt% 2wt% 4wt% 8wt% 12wt% 20wt%
C-1s 0.24 0.32 0.20 0.21 0.21 0.25
O-1s 0.56 0.50 0.55 0.52 0.50 0.43
TiO2-2p 0.21 0.16 0.23 0.20 0.22 0.22
Ag-3d UD 0.01 0.03 0.07 0.07 0.11
Theoretical Agy 0.008 0.015 0.03 0.06 0.09 0.15
Total 1.00 1.00 1.00 1.00 1.00 1.00
Stoichiometric Ratios
O/Ti 2.67 3.21 2.41 2.55 2.32 1.95
Ag/O | 0.02 0.05 0.13 0.14 0.25
Ag/Ti | 0.07 0.13 0.34 0.32 0.49
z - uncertainty  10%
UD - undetectable
y - calculated based on the Ag deposited on specimen.
93
B.2 Calculating Mean Free Path of Titanium Dioxide - Penn Method
The mean free path was calculated for the compound TiO2 using the method described
in Penn [73],
 T =  =[a(ln +b)] (B.1)
where  T is the total mean free path due to the valence and core electrons,  is excitation
energy in eV, and a and b are electronic concentration functions of  e. Al K (1486.6 eV)
x-ray excitation source was used.
 e = 6:02 1023 Z =A (B.2)
where  e is density of valence electrons, cm 3, Z is # of valence electrons,  is concen-
tration of solid, g cm 3, and A is molecular weight, g mol 1.
For TiO2, Z = 4 + 2 6 = 16,  = 4:23 g cm 3, A = 79:9 g mol 1.
 e = 6:02 10
23  16 3:9
79:9 = 5:10 10
23 cm 3
rs =
 3
4 
1
ne
 1
3 1
a0 (B.3)
rs =
 3
4 
1
5:10 1023
 1
3 1
0:529 10 8 = 1:47
For rs = 1:47, using Figure 1 in Penn[73] av = 15:75 and bv = 2:75.
From Penn [73], Table 1, for a Ti atom N is the number of core electrons in the highest core
level, N= 6, and the excitation energy  E, of the highest core level,  E = 66.7 eV, for an
O atom, N= 0 has no core level of importance. Thus,
ac = 3:92 102( N=A E) (B.4)
94
ac = 3:92 102
 4:23 6
79:9 66:7
 
= 1:87
bc = ln( E=4) (B.5)
bc = ln(66:7=4) = 2:81
a = av +ac = 17:62
b = (avbv +acbc)=(av +ac) (B.6)
b = (15:75  2:75 + 1:87  2:81)(15:75 + 1:87) = 2:76
 v =  =[av(ln +bv)]
 v = 1486:6=[15:75(ln 1486:6 2:75)] = 2:07  A
 c =  =[ac(ln +bc)]
 c = 1486:6=[1:87(ln 1486:6 2:81)] = 176:88  A
 T =   1v +  1c (B.7)
 T = 18:5  1  A
This value for  TiO2 at 1486.6 eV was found to be in good agreement with other values
found in the literature [119]. Also, as expected,  TiO2 < Ti due electrons not being as tightly
bound in TiO2 as it is in Ti metal  Ti= 21.4  A [73].
95
Appendix C
Crystal Size Approximation
C.1 Quantitative Estimation of Surface Atoms
How far apart are the Ag atoms in the TiO2 surface?
Surface area of TiO2: 150 m2/g
Assumption: Ag atoms are well-dispersed and equally spaced.
For a TiO2 surface area loaded with 20 wt% Ag, the SA due to TiO2 is:
= 150 m2=g 0:80 g = 120 m2
Assuming area is a square, side length of square = p120 m2 11:0 m
Figure C.1: Square matrix with evenly spaced spherical particles
Number of silver atoms in 0.20 g of Ag:
= 0:20 g Ag107:87 g=mol  6:022 1023 atoms=mol = 1:12 1021 atoms
96
Number of atoms on a square matrix:
=p1:12 1021 = 3:34 1010 atoms
Separation distance of Ag atoms:
= 11:0 m3:34 1010 atoms = 3:27  A = 0:327 nm
Using Oxygen Chemisorption data and assumptions:
 of Ag : 10.5 g=cm3 =1:05 107g=m3
At 20 wt%, the avg crystallite size = 6.9 nm
Assuming spherical shape, Volume of sphere, d = 6.9 nm
V= 43 r3 = 16 d3
V= 16 (69 A)3 = 1:72 10 25 m3
Mass of atoms in a 69  A crystal :
= density of Ag Volume of sphere
= 1:05 107g=m3 1:72 10 25 m3
= 1:81 10 18 g
Number of atoms of Ag in a 69  A crystal :
= 1:81 10
 18 g Ag
Atomic Weight of Ag g=mol  NA atoms=mol
= 1:81 10
 18 g Ag
107:87 Ag g=mol  6:022 10
23 atoms=mol = 1:01 104 atoms
97
Number of Ag crystallites:
= 1:12 10
21 atoms
1:01 104 atoms = 1:11 10
17 crystallites
Assuming one crystallite is equivalent to 6.9 nm
Number of crystallites on a square matrix:
=p1:11 1017 = 3:33 108 crystallites
Considering the shadowing e ect
Cross-sectional area of a sphere, d = 6.9 nm:
Ac=  r2 =  d
2
4
=  (6:9 nm)
2
4 = 3:74 10
 17m2
Total surface area of Ag crystallites for 20 wt % loading:
= Ac of a crystallite Total # of crystallites
= 3:74 10 17m2 1:11 1017 crystallites
= 4:15 m2
Percentage of surface area Ag occupies
= 4:15 m
2
150 m2  100 = 2:8 %
98
Separation distance of crystallites on a side of the square l= 11 m, assuming center to
center separation:
= 11 m3:33 108 crystallites
= 3:29 10 8 m=crystallite = 329  A between each crystallite
Assuming hemispherical shape
Figure C.2: Square matrix with evenly spaced hemispherical particles
Volume of hemisphere, d = 6.9 nm
V= 23 r3 = 112 d3
V= 112 (69 A)3 = 8:60 10 26 m3
Mass of atoms in a 69  A crystal :
= 1:05 107g=m3 8:60 10 26 m3
= 9:03 10 19 g
99
Number of atoms of Ag in a 69  A crystal :
= 9:03 10
 18 g Ag
107:87 g=mol  6:022 10
23 atoms=mol = 5:04 103 atoms
Number of Ag crystallites:
= 1:12 10
21 atoms
5:04 103 atoms = 2:22 10
17 crystallites
Number of crystallites on a square matrix:
=p2:22 1017 = 4:71 108 crystallites
Cross-sectional area of a hemisphere, d = 6.9 nm:
Ac=  r2 =  d
2
4
=  (6:9 nm)
2
4 = 3:74 10
 17m2
Total surface area of Ag crystallites for 20 wt% loading:
= Ac of one crystallites Total # of crystallites
= 3:74 10 17m2 2:21 1017 crystallites
= 8:28 m2
Percentage of surface area Ag occupies
= 8:28 m
2
150 m2  100 = 5:5 %
100
Separation distance of crystallites on a side of the square l= 11 m, assuming center to
center separation:
= 11 m4:71 108 crystallites
= 2:33 10 8 m=crystallite = 233  A between each crystallite
Assuming cubic shape
Figure C.3: Square matrix with evenly spaced cubic particles
Side edge of cube s, s = 6.9 nm ()
V = s3
V = (69 A)3 = 3:29 10 25 m3
Mass of atoms in a 69  A crystal :
= 1:05 107g=m3 3:29 10 25 m3
= 3:45 10 18 g
101
Number of atoms of Ag in a 69  A crystal :
= 3:45 10
 18 g Ag
107:87 g=mol  6:022 10
23 atoms=mol = 1:93 104 atoms
Number of Ag crystallites:
= 1:12 10
21 atoms
1:926 104 atoms = 5:80 10
16 crystallites
Number of crystallites on a square matrix:
=p5:80 1016 = 2:41 108 crystallites
Cross-sectional area of a cube, s = 6.9 nm:
Ac = s2
= (6:9 nm)2 = 4:76 10 17m2
Total surface area of Ag crystallites for 20 wt % loading:
= Ac of one crystallites Total # of crystallites
= 4:76 10 17m2 5:80 1016 crystallites
= 2:76 m2
Percentage of surface area Ag occupies
= 2:76 m
2
150 m2  100 = 1:8 %
102
Separation distance of crystallites on a side of the square l= 11 m, assuming center to
center separation:
= 11 m2:41 108 crystallites
= 4:56 10 8 m=crystallite = 456  A between each crystallite
69 ?
329 ?
329 ?
Figure C.4: Schematic representation of 20 wt% Ag particles (particles size derived from
chemisorption) on TiO2 surface .
103
C.2 Monolayer Estimation
1 ML Ag(111)= 1:14 1015 atoms=cm2 (see Section 2.2)
Number of silver atoms in 0.04 g of Ag:
= 0:04 g Ag107:87 g=mol  NA
= 0:04 g Ag107:87 g=mol  6:022 1023 atoms=mol
= 2:23 1020 atoms
Thus, there are 2:23 1020 atoms in 1 g of 4 wt% Ag/TiO2.
Number of atoms needed to form 1ML of Ag on 144 m2 on TiO2:
= 1:14 1019 atoms=m2 144 m2
= 1:64 1021 atoms
Thus, ML formed on 4 wt% Ag/TiO2 assuming homogenous distribution is:
= 2:23 10
20
1:64 1021
= 0:131ML
xA working basis approximation for polycrystalline surface is to assume that the surface is
formed from equal proportions of the main low index planes. The (100), (110), and (111)
surfaces are the so-called low index surfaces of a cubic crystal system. The number of
surface atoms per unit area of Ag polycrystalline surface is 1:14 1019 atoms=m2.
104
T able C.1: Surface Co v erage Estimation
Ag = TiO 2 Loading
Data 4 wt% 8 wt% 12 wt% 20 wt% Units
Chemisorption Avg C rystal Size (diameter) 3.4 4.1 5.3 6.9 nm
S.A due to TiO 2 144 138 132 120 m 2
Side length of square 12.0 11.7 11.5 11.0 m
No. of silv e r atoms in Ag Loading 2.23  10 20 4.47  10 20 6.70  10 20 1.12  10 21 atoms
No. of atoms on a square matrix 1.4 9  10 10 2.11  10 10 2.59  10 10 3.34  10 10 atoms
Separation distance of Ag atoms 1 8.030  10  10 5.559  10  10 4.439  10  10 3.278  10  10 m/atom
Spherical Shap e
V olume of sphere ( V = 1 = 6   R 3 ) 2.06  10  26 3.61  10  26 7.80  10  26 1.72E-25 m 3
Mass of atoms in crystallite 2.16  10  19 3.79  10  19 8.18  10  19 1.81  10  18 g
No. of atoms in crystallite 1.21  10 3 2.12  10 3 4.57  10 3 1.01  10 4 atoms
No. of crystallite 1.85  10 17 2.11  10 17 1.47  10 17 1.11  10 17 crystallites
No. of crystallites on square matri x 4.30  10 8 4.59  10 8 3.83  10 8 3.33  10 8 crystallites
Cross sectional area of sphere 9.08  10  18 1.32  10  17 2.21  10  17 3.74  10  17 m 2
T otal SA of Ag crystallites 1.68 2.7 9 3.23 4.14 m 2
105
T able C.1: Surface Co v erage Estimation
Ag = TiO 2 Loading
Data 4 wt% 8 wt% 12 wt% 20 wt% Units
% SA of Ag atoms 1.12 1.86 2.1 6 2.76 %
Separation distance of crystallite 2.79  10  8 2.56  10  8 3.00  10  8 3.29  10  8 m/crystallite
Hemispherical Shap e
V olume of hemisphere ( V = 1 = 12   d 3 ) 1.03  10  26 1.80  10  26 3.90  10  26 8.60  10  26 m 3
Mass of atoms in crystallite 1.08  10  19 1.89  10  19 4.09  10  19 9.03  10  19 g
No. of atoms in crystallite 6.03  10 2 1.06  10 3 2.28  10 3 5.04  10 3 atoms
No. of crystallite 3.70  10 17 4.22  10 17 2.93  10 17 2.21  10 17 crystallites
No. of crystallites on square matrix 6.08  10 8 6.50  10 8 5.41  10 8 4.71  10 8 crystallites
Cross sectional area of hemisphere 9.08  10  18 1.32  10  17 2.21  10  17 3.74  10  17 m 2
T otal SA of Ag crystallites 3.36 5.57 6.47 8.28 m 2
% SA of Ag atoms 2.24 3.72 4. 31 5.52 %
Separation distance of crystallite 1.97  10  8 1.81  10  8 2.12  10  8 2.33  10  8 m/crystallite
Cubic Shap e
V olume of cub e 2 V = d 3 3.93  10  26 6.89  10  26 1.49  10  25 3.29  10  25 m 3
Mass of atoms in crystallite 4.13  10  19 7.24  10  19 1.56  10  18 3.45  10  18 g
106
T able C.1: Surface Co v erage Estimation
Ag = TiO 2 Loading
Data 4 wt% 8 wt% 12 wt% 20 wt% Units
No. of ato ms in crystallite 2.304  10 3 4.040  10 3 8.727  10 3 1.926  10 4 atoms
No. of cry stallite 9.69  10 16 1.11  10 17 7.68  10 16 5.80  10 16 crystallites
No. of cr ystallites on square matrix 3.11  10 8 3.32  10 8 2.77  10 8 2.41  10 8 crystallites
Cross sectional area of cub e 1.156  10  17 1.681  10  17 2.809  10  17 4.761E  10  17 m 2
T otal SA of Ag crystallites 1.12 1.86 2.16 2.76 m 2
% SA of Ag A toms 0.747 1.24 1. 44 1.84 %
Separation distance of crystallite 3.85  10  8 3.53  10  8 4.15  10  8 4.55  10  8 m/crystallite
Equiv ale n t ML Co v era ge 0.131 0.261 0.392 0.653 ML
1 Separation estimation du e to w eigh t loading and siz e of sample area
2 dimension of one side of cubic particle
107
C.3 Chemisorption Particle Geometry Analysis
Cubic crystallites with  ve sides exposed, the surface-to-volume (S=V) ratio of crystallite is:
ST
VT =
N5d2
Nd3
= 5d (C.1)
where ST is the total metal surface area, VT is the total volume of metal crystallites, N is
the number of crystallites, and d is particle dimension. Rearranging Eqn C.1
d = 5VS
Substituing  = Wm/VT
d = 5WmS (C.2)
where  is the density of the metal, and Wm is the percent weight loading of the metal.
Since the average crystallite size is based arbitrarily on a chosen geometry the deviation
from the mean of the crystallite size from the geometries was noted in Table C.2
Table C.2: Crystallite Geometry Comparison
Geometry S=V ratio Shape factor
Cube 5=d 5
Sphere 6=d 6
Hemisphere 6=d 6
108
C.4 Theoretical XPS Intensity Calculation
Table C.3: Ag(111) Intensity Estimation
Interatomic Spacing1 Atomic Spacing Intensity Cum. Intensity Mean Free Path2
 A  
Surface 0.00 1.00 1.00
1 2.35 0.84 1.84
2 4.70 0.71 2.55
3 7.05 0.60 3.15
4 9.40 0.51 3.66
5 11.75 0.43 4.09
6 14.10 0.36 4.45 1
7 16.45 0.30 4.75
8 18.80 0.26 5.01
9 21.15 0.22 5.22
10 23.50 0.18 5.41
11 25.85 0.15 5.56
12 28.20 0.13 5.69 2
13 30.55 0.11 5.80
14 32.90 0.09 5.89
15 35.25 0.08 5.97
16 37.60 0.07 6.03
17 39.95 0.06 6.09
18 42.30 0.05 6.14 3
19 44.65 0.04 6.17
20 47.00 0.03 6.21
21 49.35 0.03 6.24
109
Table C.3: Ag(111) Intensity Estimation
Interatomic Spacing1 Atomic Spacing Intensity Cum. Intensity Mean Free Path2
 A  
22 51.70 0.02 6.26
23 54.05 0.02 6.28
24 56.40 0.02 6.30 4
25 58.75 0.014 6.31
26 61.10 0.012 6.32
27 63.45 0.010 6.33
28 65.80 0.008 6.34
29 68.15 0.007 6.35
30 70.50 0.006 6.35 5
40 94.00 0.001 6.38
50 118.0 2.00 10 4 6.39
100 1175 4.02 10 8 6.39
500 1175 4.11 10 37 6.39
1000 2350 1.11 10 74 6.39
1Ag (111) interatomic surface spacing 2.35  A[114]
2Ag mean free path 13.8 A. [73]
110
Table C.4: Ti(001) Intensity Estimation
Interatomic Spacing1 Atomic Spacing Intensity Cum. Intensity Mean Free Path2
 A  
Surface 0.00 1.00 1.00
1 3.79 0.74 1.74
2 7.58 0.55 2.30
3 11.37 0.41 2.71
4 15.16 0.31 3.01
5 18.95 0.23 3.24 1
6 22.74 0.17 3.41
7 26.53 0.13 3.54
8 30.32 0.09 3.63
9 34.11 0.07 3.70
10 37.90 0.05 3.75 2
11 41.69 0.04 3.79
12 45.48 0.03 3.82
13 49.27 0.02 3.84
14 53.06 0.02 3.86
15 56.85 0.01 3.87 3
16 60.64 0.01 3.88
17 64.43 0.01 3.88
18 68.22 4.85 10 3 3.89
19 72.01 3.60 10 3 3.89
20 75.80 2.68 10 3 3.89 4
21 79.59 1.99 10 3 3.90
22 83.38 1.48 10 3 3.90
111
Table C.4: Ti(001) Intensity Estimation
Interatomic Spacing1 Atomic Spacing Intensity Cum. Intensity Mean Free Path2
 A  
23 87.17 1.10 10 3 3.90
24 90.96 8.20 10 4 3.90
25 94.75 6.10 10 4 3.90 5
30 113.70 1.39 10 4 3.90
50 189.0 3.72 10 7 3.90
100 379.37 1.38 10 13 3.90
500 1175 5.06 10 65 3.90
1000 2350 2.56 10 129 3.90
1TiO2(001) interatomic surface spacing 3.79  A[97, 98]
2TiO2 mean free path 18.5  A. see Section B.2
112
Table C.5: IAg/ITiO2 Theoretical Estimation-Kuipers Case-1
Ag Loading Concentration Surface Area Coverage tk IAg=ITiO2
wt% moles Ag m2 m
Spherical Model
4 0.03 0.0112 1.75 10 9 0.022
8 0.06 0.0186 2.18 10 9 0.036
12 0.09 0.0216 2.91 10 9 0.040
20 0.15 0.0276 4.06 10 9 0.050
Hemispherical Model
4 0.03 0.0224 8.77 10 10 0.047
8 0.06 0.0372 1.09 10 9 0.078
12 0.09 0.0431 1.46 10 9 0.088
20 0.15 0.0552 2.03 10 9 0.111
Cubic Model
4 0.03 0.00747 2.63 10 9 0.000
8 0.06 0.0124 3.27 10 9 0.022
12 0.09 0.0144 4.36 10 9 0.025
20 0.15 0.0184 6.09 10 9 0.031
113
Table C.6: IAg/ITiO2 Theoretical Estimation-Kuipers Case 2
Ag Loading Concentration Diameter Surface Area Coverage tk IAg=ITiO2
wt% moles Ag m m2 m
4 0.03 3.40 10 9 1.13 10 9 0.0163 0.0332
8 0.06 4.10 10 9 1.37 10 9 0.0262 0.0053
12 0.09 5.30 10 9 1.77 10 9 0.0294 0.0583
20 0.15 6.90 10 9 2.30 10 9 0.0352 0.0681
114
T able C.7: I Ag /I TiO 2 Theoretical Estimation-Kuip ers C ase 3
Ag Loading Concen tration I Ag = I TiO2 Surface Area Co v erage t k Diameter Dia meter
wt% moles Ag m 2 m m nm
1 0.008
2 0.015 0.06 0.96 1.02  10  11 3.05  10  11 0.03
4 0.03 0.14 0.92 2.18  10  11 6.55  10  11 0.07
8 0.06 0.31 0.85 5.13  10  11 1.53  10  10 0.15
12 0.09 0.35 0.83 8.16  10  11 2.45  10  10 0.24
20 0.15 0.48 0.78 1.56  10  10 4.66  10  10 0.47115