Titania carbon nanotube composites for enhanced photocatalysis

Material Information

Titania carbon nanotube composites for enhanced photocatalysis
Pyrgiotakis, Georgios
Publication Date:


Subjects / Keywords:
Anatase ( jstor )
Binding energy ( jstor )
Carbon ( jstor )
Carbon nanotubes ( jstor )
Dyes ( jstor )
Electrons ( jstor )
Nanotubes ( jstor )
Oxygen ( jstor )
Raw data ( jstor )
Semiconductors ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Georgios Pyrgiotakis. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
Resource Identifier:
658216354 ( OCLC )


This item has the following downloads:
















































































































































































































Full Text







Copyright 2006


Georgios Pyrgiotakis

I dedicate this work to my parents and sister,

and to the memory of my grandfather,

the first teacher I ever had.


There are many persons that without their critical and influential support and

guidance this work would have never been accomplished.

I would like to first and foremost thank Dr. Wolfgang Sigmund whose work

ethic, compassion, support, understanding and guidance helped me through this

project. I would also like to thank my committee members, Drs. Milz, Norton,

Sinnott and Koopman for their constructive comments. Very special thanks go to

Dr. Koopman who very closely observed the whole project and whose -ii--.- -I i i

were ahv--i-i influential. Also I would like to thank Dr. Moudgil who ahv--i-

challenged me to discover new path--,i,- in science. I would also like to thank

Dr. Rinzler for all his help regarding the nanotubes. I would like to recognize

the help of the staff of MAIC (\l Ierials Analytical Instrument Center) regarding

the characterization and the help of Maria Palazeulos regarding the Raman


There are also a lot of students who without their help I would not have

finished this work. I thank Vijay Krishna and Jue Zao for the extensive discussions

about the problems we encountered and all the people in the Sigmund group,

especially Drs. S.-W. Lee, J.-M. Cho and S.-H. Lee. A very warm thank goes to

my dear friends Amit, Junhan, Isaac and Vasana, for their support and help during

my work. Also I would like to acknowledge all the past and current members in the

group for assisting me in many v-- -i during my work.

Finally I would like to acknowledge my parents for their support through all

the rough moments of my life in the USA. Special thanks to my sister for cheering

me up all the time. And last but not least, I tank my friends all over the world

((C'!1 i, Germany, Cyprus, Greece, India, Japan, Korea, Taiwan, Turkey, UK and

USA) who constantly showed me love and support. Without them I would have

never accomplished this work. Finally I would like to thank all the people that

worked towards the discovery and perfection of coffee, my ultimate support through

my doctoral.




LIST OF TABLES ...............

LIST OF FIGURES ................................

ABSTRACT ...................

1 INTRODUCTION .............



Photocatalysis and Titania .....
Carbon Nanotubes .........
Objectives ............


2.1 Structure of Titania .............
2.1.1 Anatase ...............
2.1.2 Rutile ...............
2.2 Electronic Properties of Titania ......
2.2.1 Anatase ...............
2.2.2 Rutile ...............
2.3 Semiconductor Photocatalysis ...
2.3.1 Basic Principles .. .........
2.3.2 Enhancement of Photocatalysis .
2.4 Applications of Photocatalysis .......
2.4.1 Environmental Applications ..
2.4.2 Photovoltaic Cell .. ........

PROPERTIES .. ...............



3.1 Bonding, Structure and Physics of Single-Wall Carbon Nanotubes
3.1.1 Bonding in Carbon Materials .. ..............
3.1.2 Structure and Notation .. ................
3.1.3 Symmetries and Vibrational Frequencies .. ........
3.2 Electronic Properties of SWNT and MWNT .. ..........
3.2.1 Electronic Properties of SW NT .. .............
3.2.2 Electronic properties of MWNT .. ............

3.3 Carbon Nanotubes Growth Mechanisms . . 34
3.3.1 Arc Discharge . . ......... .... 35
3.3.2 CVD: Thermal CVD, PE-CVD . . ..... 35

AND CHARACTERIZATION) .................. ..... 37

4.1 Design Parameters ................... .... 37
4.2 Nanotube Selection and Preparation ....... ........ 39
4.2.1 Materials Selection .................. .... 40
4.2.2 Purification and Dispersion . . . 40
4.2.3 C('i '.terization of the Functionalized MWNTs . 41
4.3 Sol-Gel Coating ............... ........ .. 50
4.3.1 Precursor Selection ..... ........ ... . 51
4.3.2 Coating Model ................ ... ... .. 54
4.3.3 Long MW NTs ................ ... ... .. .. 55
4.3.4 Short MW NTs ................ ... ... .. 58
4.4 'C!i o 'Iterization of the Composites .. ............ 59
4.4.1 Short ANTs: TEM, XPS, BET . . ....... 61
4.4.2 Long ANTs: TEM, XPS, BET . . ...... 62


5.1 Experimental Setup, Materials and Procedures . .... 69
5.1.1 Experimental Setup .................. .... 69
5.1.2 Dye Selection .................. ..... .. 70
5.1.3 Experimental Procedure .................. .. 72
5.2 Theory for the Photocatalytic Degradation of Dyes . ... 74
5.3 Parameters that Influence the Photocatalytic Reaction . 76
5.3.1 pH ............... ........... .. 77
5.3.2 Initial Dye Concentration .... . . 77
5.3.3 Intensity of the Radiation ..... ......... .. 79
5.3.4 Solids L .. .ii. /Surface Area .............. .. 79
5.4 Experiments ........ .. ... ........... .. 81
5.4.1 Titania Nanoparticles and Carbon Nanotubes . ... 83
5.4.2 Long ANTs: Photocatalysis under UV Light . ... 87
5.4.3 Long ANTs: Photocatalysis under Visible Light . 88
5.4.4 Long ANTs: Post UV Activity, Photocatalysis in Dark 92
5.4.5 Short Nanotubes: Photocatalysis under UV . ... 92
5.5 Conclusion .................. ............ .. 94


6.1 Raman Spectroscopy of the Carbon Nanotubes . .... 97

6.1.1 General Theory of Raman Spectroscopy of Carbon Nan-
otubes ..... .. .. .. .. ..... 98
6.1.2 Basic Raman Lines for Carbon Nanotubes . ... 100
6.2 Raman Spectroscopy of the Anatase Structure of TiO2 ...... 104
6.3 Experimental Procedures ................ .. 106
6.3.1 Sample Preparation . . ......... . 106
6.3.2 Mathematical Analysis and Manipulation . .... 107
6.4 Experimental Results ....... . . .... 111
6.4.1 Long Nanotubes after the Acid Treatment . ... 112
6.4.2 Short Nanotubes after the Acid Treatment . ... 113
6.4.3 Long Nanotubes after the Coating . . 117
6.4.4 Short Nanotubes after the Coating . . 123
6.4.5 Summary of the Raman Spectra Analysis . .... 130
6.5 X-Ray Photoelectron Spectroscopy (XPS) . . 131
6.5.1 Instrument, Sample Preparation and Mathematical Analysis 132
6.5.2 XPS of the Reference Material . . 1.33
6.5.3 XPS of the s-ANTs ................ 137
6.5.4 XPS of the f-ANTs ............. .... . 143
6.6 Summary of the XPS Analysis .............. .. 147


7.1 Conclusions ............... .......... 160
7.2 Future Work ...... .................... 160



B RAMAN PEAKS OF CNTs .................. ..... 165

REFERENCES ................... ............. 166

BIOGRAPHICAL SKETCH .................. ......... 188

Table page

4-1 The calculated initial molecular ratio for the reactions for the -CNTs 55

4-2 The calculated initial molecular ratio for the reactions regarding the
short nanotubes .................. .......... .. 57

5-1 The oxidation intermediates and their structure to be compared to
the initial dye structure in figure 52. ................. 74

5-2 Summary of the experiments performed ................ 84

5-3 Summary of the experimental results of this chapter . ..... 94

6-1 The Raman frequencies fro anatase and rutile phase of titania. The
brookite is not included here since is not a present form of TiO2
and it has in total 36 weak peaks. The notation in parenthesis is
representing the relative intensity of the peaks; w: weak; m: medium;
s: i' ir:. vs: very strong. Data are adapted from reference mate-
rial and reference .................. ......... 105

6-2 The raw fitting parameters calculated with the Levenberg-Marquardt
algorithm for the acid treated -CNTs. The graphic representation
of the results is in figure 6-6. The fit yielded X2 7.1333x 104. For
convenience at the data representation we use the symbol a2) in-
stead of F that is used in equation 6-20. ............. ..113

6-3 The raw fitting parameters calculated with the Levenberg-Marquardt
algorithm for the acid treated s-CNTs. The graphic representation
of the results is in figure 6-7. The fit yielded X2 3.9138x 10. 115

6-4 The raw fitting parameters calculated with the Levenberg-Marquardt
algorithm for the titania coated -CNTs and the titania segment of
the spectrum. The graphic representation of the results is in figure
6-9. The fit yielded X2 8.3378 x 104 ................ 117

6-5 The raw fitting parameters calculated with the Levenberg-Marquardt
algorithm for the titania coated -CNTs. The graphic representa-
tion of the results is in figure 6-6. The fit yielded X2 8.3378 x 104
For convenience at the data representation we use the symbol a)
instead of F that is used in equation 6-20. . . 121

6-6 The raw fitting parameters calculated with the Levenberg-Marquardt
algorithm for the acid treated E-CNTs. The graphic representation
of the results is in figure 6-12. The fit yielded X2 1.9924x108. 125

6-7 The raw fitting parameters calculated with the Levenberg-Marquardt
algorithm for the coated s-CNTs. The graphic representation of the
results is in figure 6-12. The fit yielded X2 1.0956x105. ..... ..127

6-8 Summary of the Raman result. Here are listed the 1 i ri" peaks and
shift both for titania and CNTs after the coating. . ... 130

6-9 Summary of the XPS peaks .................. ... 147

7-1 Electron affinity and work function for metals used to create rectify-
ing contact with titania in order to increase the photocatalytic effi-
ciency. ............... .............. .. 158

B-1 Properties of the various Raman features in graphite and SWNTs. .. 165

Figure page

2-1 The two basic titania structures. .................. 7

2-2 The electronic band structure of the two main phases of titania. .. 8

2-3 Schematic diagram representing the main photocatalysts with their
bandgap energy. In order to photo-reduce a chemical species, the
conductance band of the semiconductor must be more negative than
the reduction potential of the chemical species; to photo-oxidize a
chemical species, the potential of the valence band has to be more
positive than the oxidation potential of the chemical species. The
energies are shown for pH 0. ................ ..... 11

2-4 Schematic representation of the reactions taking place in titania. ()
Light strikes the semiconductor. () An electron-hole pair is formed.
@ Electrons and holes are migrating to the surface. @ The holes
initiate oxidation leading to CO2, Cl-H+, H20. ( The conduction
band electrons initiate reduction reactions. @ electron and holes
recombination to heat or light. ................ .. .. 13

2-5 Titania band structure (a) before and (b) after doping. The transi-
tion metals are interstitial or substitutional defects in the structure
of titania and generate trapping levels in the bandgap. . .. 14

2-6 The principles of rectifying contact between titania (Eg=3.2 eV) and
a metal with work function ( ), in this example 5 eV, greater than
the affinity (Xs) of titania. ............. ... 16

2-7 The principles of rectifying contact between anatase (a) titania (E= 3.2
eV) and and rutile (r) titania (E a3.0 eV). . 18

3-1 The 2D graphene sheets is shown with the al and a2 specifies the
chirality of the nanotube. The chiral vector, Ch, is the OA, while
the translation vector T is the OB. Also ib is the rotation angle
and 7 the translation. Those two are constitute the symmetry op-
eration R = ( I7-) . . . .. . . 225

3-2 The graphene sheet is shown with the (n, m) pair which specifies the
chiral nanotube. The pair of integer (n, m) in the figure specifies
the chiral vector Ch for carbon nanotubes, including zigzag, arm-
chair and chiral tubules. Below each pair of integer is listed the
number of distinct caps that can be joined continuously to the cylin-
drical carbon tubule denoted by (m, n) [ref]. It is also denoted the
conduction state of every chirality. ................ 28

3-3 The dispersion for graphite as calculated from equation 3-10. . 30

3-4 The dispersion energies for two different chilarities. . .... 32

4-1 SEM pictures of the two types of nanotubes. . . ...... 42

4-2 TEM images of the s-CNTs. .................. .... 43

4-3 TEM images of the -CNTs. .................. .... 44

4-4 Immediate comparison of the two different kinds of nanotubes. 45

4-5 The zeta potential for both the -CNTs (a) and s-CNTs (b). It shows
the shift of the IEP for the -CNTs (from 7 to 3.5) and the increase
at the surface charge for the s-CNTs (from -10 mV to -37 my for
ph 4). .. .. . . . .. .. . .. .. 46

4-6 The FTIR of the MWNTs after the acid treatment (only the s-CNTs
results are di-pl, i,- .1). The bands that have been identified prove
the reaction of the -COOH on the surface of the nanotubes. . 47

4-7 The differential volume and number of the s-CNTs before and after
the acid treatment. .................. .... 48

4-8 The TGA/TDA data of the s-CNTs. The peak at the 600C indi-
cates the burning temperature of the CNTs. It is observed about
i,' of the initial mass residue, which is the Fe catalyst. ...... ..49

4-9 The different Sol-Gel precursors used in this research . .... 53

4-10 Schematic diagram of the process for the coating of the -CNTs. 56

4-11 Schematic diagram of the process for the coating of the s-CNTs. 59

4-12 The TGA/TDA data of the s-ANTs. The peak at the 100C is from
the water evaporation and therefore it is accommodated by a mass
reduction. At approximately 250C the phase transition is starting
and carries on until the 500C. .................. .... 60

4-13 TEM images of the coated s-CNTs. ................. 61

4-14 TEM images of the coated -CNTs. ................. 62

4-15 The universal curve of the electrons, based on the calculations by M.
P. Seah and W. A. Dench. The curve shows the mean free path
of the electrons as function of the kinetic energy (dashed lines).
There are also experimental results that follow the same trend. The
mean free path does not depend on the material. For Mg source
the X-Ray energy is 1253.6 eV, which give a mean free path of ap-
proximately 10 A. ................... ........ 63

4-16 XRD patterns with and without the coating. . . ...... 65

4-17 XPS survey for the s-ANTs. There is a significant amount of TiO2
(16.7' Ti). There is no direct stoichiometry with the oxygen (52' .
O) since the oxygen depends on the exposed i i v-I .llgraphic orien-
tation ...... ............. .............. .. 66

4-18 XPS survey for the f-ANTs. There is a significant amount of TiO2
(1.'' Ti). Again there is no stoichiometry with the oxygen (, '-'
O). There is less TiO2 compared to the s-ANTs . ..... 67

5-1 Schematic diagram showing the basic elements of the photocatalytic
degradation chamber. .................. ..... 69

5-2 Three-dimensional structure of the Brilliant Procion Red MX-5 molecule.
As it can be seen it contains 3 benzene groups and a benzene group
with three carbon atoms replaced by nitrogen atoms (s-triazine). 71

5-3 The absorption spectrum for a 5 ppm solution of the Procion Red
MX-5B dye. .................. ......... 72

5-4 The structure of several intermediate products of the photocatalytic
reaction that show the destruction of the bonds and the size reduc-
tion of the molecules. .................... ....... 73

5-5 Comparison between the numerical solution of the Langmuir-Hinshelwood
(equation 5-1) and the approximation. The red lines represent the
approximation and the black is the numerical solution. The solid
line represents the dye concentration while the dashed represents
reaction rate. . . . . . . 75
Wt k7O0

5-6 The main parameters that influence the oxidation rate. . ... 78

5-7 The pH variation during the dye degradation. The initial value be-
tween the ANTs and Degussa P25 since the specific surface area is
different. In the first case the pH is stabilized after 10 min while in
the second case that occurs after 20 min. In both cases the stable
pH value is lower than the initial. ................. 80

5-8 The dye spectrum during the different time intervals. The three dashed
lines (513, 524 and 537 nm) are the three wavelengths that were
used for the C/Co calculation. The data were obtained from a sam-
ple of 3 mg Degussa P25 in a 50 ml of 5 ppm dye solution. ..... ..81

5-9 Investigation of the dye degradation under the UV light for two dif-
ferent dye concentrations. The UV is not having an apparent im-
pact on the dye. ............... ......... 82

5-10 The results for the experiments A-1 to A-4. .............. ..85

5-11 Collective graph of the data presented above. . . 86

5-12 Investigation of the dye adsorption on the carbon nanotubes surface.
The adsorption was not significant since it was only 5'. reduction
after 90 m in. . . . . .. .. . 87

5-13 Photocatalytic degradation of Degussa P25 and f-ANTs under UV
light of 350 nm wavelength. ................ ..... 89

5-14 The photocatalytic results of the f-ANTs and Degussa P25. The f-
ANTs clearly demonstrate photocatalytic activity with -=152.316.13
min. Degussa P25 is not demonstrating any obvious activity . 90

5-15 The dye degradation data in the dark for the f-ANTs. Degussa is not
included here since it never demonstrated behavior like such. The
data were fitted with the equation 5-9. fA ,-1.290.24 d(
The constant is 0.762.75x10-2 ................. 91

5-16 The dye degradation data in the UV light of 350 nm for the s-ANTs.
T-rANTs 177.4110.00 mins. The photocatalysis is significantly
slower that all the previous cases. ................. 93

6-1 The different Raman scattering processes for CNTs. . ... 98

6-2 Graphic representation of the i i i', Raman modes. . ... 100

6-3 Typical Raman spectra from metallic and semiconducting SWNTs.
The Radial Breathing Mode (RBM), the D Band and G Band are
the most important bands. The is denoting bands that come form
the Si substrate. Due to the distinct structure of the semiconduct-
ing nanotubes there are two additional bands M and iTOLA that
appear. ..... ........... ...... ...... 101

6-4 The G Band split and how it is related to the conductivity of the tubes. 101

6-5 Different options for the LOESS algorithm. . . 109

6-6 The -CNTs after treated with nitric acid at 140C for 10 hours. The
D Band is showing at 1312 cm-1 and the G Band at about 1594
cm-1. A very distinct split of the band can be seen with the G+ at
the 1584 cm-1 and G- at 1612 cm-1. ................ .. 114

6-7 The s-CNTs after treated with nitric acid at 100C for 6 hours. The
D Band is showing at 1305 cm-1 and the G Band at about 1586
cm-1. Although the G Band looks like it consists on to overlap-
ping peaks it still can be treated as one peak. . ..... 116

6-8 The Raman spectra of the coated long nanotubes. There are two sep-
arate regions, (i) 0-1000 cm-1 that contain the titania peaks and
(ii) 1000-1800 cm-1 that contain the carbon nanotubes peaks. The
peak identification is done later in the chapter. . .... 118

6-9 The first region from figure 6-8. There are four ini iri" peaks but only
three of them can be identified accurate. 149.56 cm-1, 628.65 cm-1
and 408.64cm-1. .................. ........ 119

6-10 The second region from figure 6-8. The D Band is at 1307 cm-1 and
the G Band is at the about 1590 cm-1. The band split still exists,
with the G- at 1579 cm-1 and the G+ at 1606 cm-1. . ... ..122

6-11 The Raman spectra of the coated short nanotubes. There are two
separate regions, (i) 0-1000 cm-1 that contain the titania peaks
and (ii) 1000-1800 cm-1 that contain the carbon nanotubes peaks.
This spectra has been obtained by the combination of two different
runs .................. ................ .. 124

6-12 The first portion of figure 6-11. There are 5 very distinctive peaks at
150 cm-1, 202 cm-1, 393 cm-1, 510 cm-1 and 633 cm-1. ..... ..126

6-13 The second portion of figure 6-11. Although the carbon peaks are
not very clear we can still see them at the 1316 cm-1 the G Band
and at the 1582 cm-1 the G Band. The G Band seems to be split-
ting in two peaks 1544 cm-1 and 1582 cm-1. The ratio between
the peaks is completely reversed but this is currently attributed to
the weak signal obtained by the s-CNTs in this case. . ... 128

6-14 The Cis peak for the reference anatase nanoparticles. The 1 i i" peak
is at the 286.4 eV that is agreement with literature and several databases. 134

6-15 The Si2p peak for the reference anatase nanoparticles. The i, i i"r
peaks are at the 98.5 eV for the Si2p1/2 and at 102.5 eV for the
Si2p3/2 which are in agreement with literature and several databases. 135

6-16 The Ols peak for the reference anatase nanoparticles. The major peaks
are at the 529.6 eV, represents the lattice oxygen, and the 531.5
eV for the surface oxygen. which are agreement in with literature
and several databases. ................ .. .... 136

6-17 The Ti2p peak for the reference anatase nanoparticles. The major
peaks are at the 458.4 eV for the Ti2p1/2 and at 464.2 eV for the
Ti2p3/2 which are in agreement with literature and several databases. 138

6-18 The Cis peak for the s-ANTs. The 1 i' Pr peak is appearing to the
284.6 eV, which is again in great agreement with literature values.
The peak at 285.9 eV is characteristic of the C-O bond while the
289.5 eV peak is attributed to C-O-Ti. . . 139

6-19 The Ols for the s-ANTs. The 1i i.r peaks are again at 530.6 eV for
the Ols for the lattice oxygen and the 532.7 eV for the surface oxy-
gen. The ratio between those two peaks reveals the surface are of
the particle. ... .. .. .. .. ... .. .. ... ... ..... 140

6-20 The Ti2p peak for the s-ANTs. The 1 i. .r peaks are at the 459.4 eV
for the Ti2p1/2 and at 465.1 eV for the Ti2p3/2. . . 142

6-21 The CIs peak for the f-ANTs. Again the ini ,i' r peak appears to be
at 284.6 eV while there is a secondary peak at 285.2 eV. This peak
is similar to the case of s-ANTs that appears to 285.9 eV. It is again
attributed to the C-O bond or C-0 bond. ........... ..144

6-22 The Ols peak for the f-ANTs. There are also two peaks observed
at 532.7 eV and at 530.9 eV. Although both are from the oxygen
the 532.7 eV is attributed to surface oxygen while the other comes
from lattice oxygen contribution. Relative to the case of s-ANTs
the surface oxygen and therefore the surface area is higher, some-
thing that was confirmed with BET as well and is in agreement
with other researchers. .................. .... 145

6-23 The Ti2p peak for the f-ANTs. The 1i, i.r peaks are at the 459.6 eV
for the Ti2p1/2 and at 465.2 eV for the Ti2p3/2 which are in signifi-
cantly shifted compared to the reference material. . ... 146

6-24 Collective representation if the XPS data regarding the coated long
carbon nanotubes. The upper row is the Ti2p and Ols peak of the
reference material and the lower row is the data obtained by the s-
ANTs. The shifts in both peaks are obvious and are summarized
in table 69. .................. ............ 149

6-25 Collective representation if the XPS data regarding the coated short
carbon nanotubes. The upper row is the Ti2p and Ols peak of the
reference material and the lower row is the data obtained by the e-
ANTs. The shifts in both peaks are obvious and are summarized
in table. .................. ............. 150

6-26 Collective representation if the XPS data regarding the coated long
and short carbon nanotubes. The upper row is the Ti2p and Ols
peak of the s-ANTs and the lower row is the data obtained by the
f-ANTs. The peaks are similar regarding the position, but are sig-
nificantly different in shape. ................ ..... 151

6-27 The Cls peak of the peak of the coated carbon nanotubes (both e-
ANTs and s-ANTs) and the reference material. The main differ-
ence between the reference material and the samples are the peaks
regarding the C-O and C=0 bonds, that are appearing only for
the s-ANTs and f-ANTs, and the peak at 289.7 eV (f-ANTs) and
289.5 eV (s-ANTs) that can be attributed to the C-O-Ti bond. 152

6-28 The Si2p peak of the peak of the coated carbon nanotubes (both e-
ANTs and s-ANTs) and the reference material. Al the peaks are
at the same energy, but the noise to signal ratio is a lot higher for
the both f-ANTs and s-ANTs. The reason for that is the thickness
of the coating. The coated MWNTs were deposited in a thicker l i. 153

6-29 Collective representation of the Raman spectra regarding the short
nanotubes before (top row) and after the coating (bottom row).
The right column is for the G band and the left column is for the
D band. ........... ...... ........ ...... 154

6-30 Collective representation of the Raman spectra regarding the long
nantubes before (top row) and after the coating (bottom row). The
right column is for the G band and the left column is for the D band. 155

6-31 Collective representation if the XPS data regarding the coated long
carbon nanotubes. The upper row is the Ti2p and Ols peak of the
reference material and the lower row is the data obtained by the s-
ANTs. The shifts in both peaks are obvious and are summarized
in table. .................. ............. 156


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Georgios Pyrgiotakis

it ,v 2006

C(' ir: Wolfgang M. Sigmund
M i i" Department: Materials Science and Engineering

Photocatalytic composites have been used for the past few decades in a wide

range of applications. The most common application is the purification of air

and water by removing toxic compounds. There is limited use however towards

biocidal applications. Despite their high efficiency, photocatalytic materials

are not comparable to the effectiveness of conventional biocidal compounds

such as chlorine and alcoholic disinfectants. On the other hand, nearly a decade

ago with the discovery of the carbon nanotubes a new vibrant scientific field

emerged. Nanotubes are unique structures of carbon that posse amazing electrical,

mechanical and thermal properties.

In this research carbon nanotubes are used as photocatalyitic enhancers. They

were coated with anatase titania to form a composite material. Two different types

of nanotubes (metallic versus non-metallic) were used and the photocatalytic ac-

tivity was measured. The metallic tubes demonstrated exceptional photocatalyitic

properties, while non-metallic tubes had low photocatalytic efficiency. The reason

for that difference was investigated and was the n i, i"r focus of this research.


The research concluded that the reasons for the high efficiency of the carbon

nanotubes were (i) the metallic nature of the tubes and (ii) the possible bond

between the titania coating and the underlying graphite l'V. rs (C-O-Ti). Since

both composites had the same indications regarding the C-O-Ti bond, the

metallic nature of the carbon nanotubes is believed to be the most dominant factor

contributing to the enhancement of the photocatalysis. The composite material

may have other potential applications such as for sensing and photovoltaic uses.


The last few decades the demand for safer environmental conditions has

increased dramatically. One reason is the constantly increasing biological threats

that can be expressed in every aspect of the daily life, ranging from cases as simple

as food bacterial contamination (E. coli and salmonella) to extremely dangerous

such as epidemic outbreaks (Ebola and SARS) and biological warfare (anthrax and

smallpox). The need for effective and efficient disinfection is driving the industry in

the development of a wide range of products. These products can be divided in to

three i ii, i" categories:

Chemical disinfectants: C(', I. Id based disinfectants are the i ii.1lily of they

have been used for the longest time. Most of them are chlorine, alcohol or

ammonium based products. They are in liquid form and therefore are limited

to surfaces. The 1i ,in, i ily are used to disinfect contaminated surfaces and not

to prevent contamination. Although their use is relatively simple and easy

they can still be dangerous if they are misused. Gasses can also be used for

the disinfection, but they are limited since they are extremely corrosive.

Radiation based disinfection: The radiation is a very effective technique since

it can immediately inactivate the ii ,in ii iy of the contaminants without

damaging the surroundings. Still however the use is limited since it usually

requires expensive equipment and under certain conditions exposure to the

used radiation can be proved dangerous.

Passive disinfectants: Passive disinfectants are characterized those that do

not require a certain application (chemicals) or operation (radiation), but

constantly purify and clean surfaces, air and water. Activated carbon filters

are probably the best known and most widely used, since they are used

widely for water and air treatment. However they do not deactivate the

contaminants so constant replacement is required. If they are not replaced

regularly they can become a source of contamination rather than disinfection


The lack of efficient passive disinfectants has led the researchers to seek solu-

tions capable to provide both capturing and inactivating of biological contaminants.

One of the most promising and rapidly emerging fields is Ih,' ..'. ,'ll-/,.:-

1.1 Photocatalysis and Titania

Pho-'...'. '-li is the type of reaction that takes place on the surface of a

certain type of material in the presence of a very specific range of radiation. There

are many materials that can display this type of reaction, but the most widely used

is titanium dioxide, TiO2, or titania. Titania in addition to the high efficiency is

cheap and environmentally safe.

There are significant limitations, however, to the application of titania since

the efficiency are not high enough or at least competitive with the results that

the chemical techniques can deliver. C'! lpter 2 will give a brief overview of the

principles of photocatalysis and specifically the catalysis on the surface of titania.

Emphasis will be given to the structure of titania and its electric properties, the

two primary reasons for the excellent photocatalytic properties. It will outline the

basic techniques that are currently used to improve the efficiency and finally will

discuss the i ii' 'r applications of titania.

1.2 Carbon Nanotubes

An also rapidly emerging field is the investigation of the properties of the

carbon nanotubes. They are a relative new material that has attracted great deal

of attention due to the unique shape and structure. Carbon nanotubes can be

visualized a graphite sheet that has been rolled seamlessly into a tube. It has been

more than a decade since the first report of nanotubes. Their unique properties,

that arise form their structure, have not yet completely understood. Probably the

most outstanding properties are the electronic. In addition, their needle-like shape

results in very high specific surface area. Both characteristics are very important to

the present research.

Although their properties are very unique and unmatchable, so far there is no

commercial application in small or large scale that takes full advantage of them.

C'! lpter 3 will explain in detail the structure and later the properties of the carbon

nanotubes. It will also give a short description of most popular methods used tod -'

for nanotubes production.

1.3 Objectives

In this research those two unique materials will be combined in the form of a

nano-composite that will deliver a high efficient photocatalyst. There are several

researchers that have already achieved it, but the composites have never been

investigated in-depth. Therefore this research has the followings objectives:

To synthesize TiO2-MWNTs composites

To evaluate the photocatalytic efficiency

To explain the behavior of the material

There are two distinct trends in combining those two materials: either in the

form of titania pellet with the nanotubes embedded, or in a more sophisticated

approach, the titania is applied as a coating on the nanotubes. In this research

the second approach was selected since it takes full advantage of the nanotube

properties, by creating a composite with a single nanotube as core.

To investigate the impact of the nanotube properties there are two different

composites synthesized. One has a pristine and highly
other has of a less ordered tubular structure. The direct comparison of those

two composites will explain the effect of the electric properties, if any, at the

photocatalytic efficiency. All the synthesis is explained in detail in chapter 4.

The photocatalytic evaluation is done via dye reduction tests. Those types of

tests are very common and are preferred since they give fast, accurate and reliable

results. A drawback of those tests is the many parameters that can impact the

results and therefore they have to be monitored while the tests are executed, but

it is something that can easily be done. Those parameters and the experiments are

discussed in detail in chapter 5.

In order to explain the behavior of the material it is critical to select tech-

niques that directly or indirectly will determine the properties of the nanotubes.

One of the most recognized techniques for that is the Raman spectroscopy. In

addition to Raman, Photoelectron Spectrometry (XPS) can be used to in-

vestigate the structure of the titania and point out structural differences that may

be related to the photocatalytic evaluation results. The complete analysis of those

two techniques, along with the necessary theory to understand Raman and XPS, is

discussed in chapter 6.

All the experimental results from chapters 4, 5 and 6 will be used to draw

conclusions on how the carbon nanotubes behave as a photocatalytic template and

what the impact of their electrical properties is on the final result.


Recently semiconductor photocatalysis has attracted a great deal of attention

since it has a wide range of applications [1, 2]. One of the most interesting mate-

rials is titania (TiO2) [3-5]. TiO2 is the material that is used here as coating on

the carbon nanotubes. It is widely available since it is used as pigment in many

applications and the production is fairly cheap [4]. Since 1972 when Fulii-1bin I et

al. reported the photocatalytic split of the water on TiO2 electrodes [6] a great deal

of research had been done to developeapplication and enhancing the properties of

titania. The applications range from photovoltaic cells to biological disinfection


One of the most popular applications is the microbial sterilization and self-

cleaning surfaces [7-15]. There are certain limitations however, coming primarily

from the electronic properties of the titania, that reduce the efficiency [4, 5]. The

'1i--. -1 breakthrough came in 1985 by Matsunaga et al. [9] where by mixing the

titania with silver particles the observed significant enhancement of the catalysis.

Since that time the range of applications has increased dramatically.

This chapter covers the basic information necessary to explain the properties of

titania. The first section is about the < i --I 1 structure and the electronic properties

of titania. Later the chapter reviews the basic principles behind photocatalysis and

the recent advances towards the improvement of the efficiency. The last part of the

chapter gives a brief overview of the most important applications of titania.

2.1 Structure of Titania

Titanium dioxide (titania) exists in principle in eight phases rutile, anatase,

brookite, columbite, baddeleyite, flourite, pyrite, and cotunnite [16]. From those

eight phases thermodynamically more stable are rutile, anatase and brookite, with

rutile to be the most stable [17, 18]. Since photocatalytic activity is demonstrated

only from rutile and anatase, the analysis will focus on those two structures only.

The columbite, baddeleyite, flourite, pyrite, and cotunnite phases can be generated

only under very high temperatures and/or pressures, which is the reason those

phases do not occur naturally [19-21], but they still possess some very interesting

properties. Cotunnite for example is the hardest piI i-i, i i--.11 M material known to

exist [16, 22].

2.1.1 Anatase

Figure 2-1(a) shows the ( i--i I1 structure of anatase. It is tetragonal with

a = b = 3.782 A and c = 9.502 A and has a D4-I41/amd symmetry. The building

block on anatase is the TiO6 which forms a slightly deformed octahedron (figure

2-1 (c)). The Ti atom that is in line with the two oxygen atoms (apical oxygen

atoms) has bond length of 1.996 A and the other four oxygen atoms (equatorial

oxygen atoms) have Ti-O bond lengths of 1.937 A. The widest angle of those two

bonds Ti-Oequatorial and Ti-Oapical is 102.308. The angle between two consecutive

equatorial bonds is 92.604or 87.394. All the bond lengths and angles given above

represent the structure at room temperature.

Anatase is an unstable structure and it transforms to rutile at approximately

800 C. While the temperature increases, the bond lengths are changing and grad-

ually the anatase turns into rutile [16, 17]. Rutile has a more compact structure

and therefore energy wise is more favorable. The transformation to rutile is an

irreversible process.

[010] [001]

4.578 A...

(a) (c) (d)

102.308 L.966

1.937 A
1.946 A

(a) (c) (d)

Figure 2-1: The two basic titania structures (a) anatase and (b) rutile. The
distorted octahedron that are shown are used to construct the (c)
anatase and (d) the rutile.

2.1.2 Rutile

Rutile has also a tetragonal structure (2-1(b)), but it is a lot more compact

compared to anatase [16, 23-25]. The tetragonal structure has a = b = 4.584 A and

c 2.953 A. It has D1-P42/mmm symmetry [16, 25, 26]. Again the building block

of the i II1 structure has an octahedral that is slightly distorted (figure 2-1(d)).

The apical oxygen atoms have Ti-O bond length of 1.983 A and the equatorial

Ti-O bond is 1.946 A. The equatorial and apical Ti-O bonds form a right angle

while the largest angle between the two equatorial bonds is 98.93.

12 -\

8 -





rx RZ r M A Z r XR zr MA Z
(a) (b)

Figure 2-2: The electronic band structure of the two main phases of titania (a)
rutile and (b) anatase [25]. The calculation is based on first principles
self consistent OLCAO.

The bond length in rutile does not change significantly with the temperature.

It is therefore thermally a stable structure and all the different phases will turn into

rutile after annealing at high temperatures for an extended period.

2.2 Electronic Properties of Titania

The electronic structure of titania has been studied both experimentally

and theoretically. Experimentally it has been probed by X-Ray photoelectron

spectroscopy [27-30] (XPS), X-Ray induced Auger electron spectroscopy [28],

Auger electron spectroscopy [28], X-Ray emission [31, 32] (XES), absorption spec-

troscopy [33, 34] (XAS), electron energy loss spectroscopy [27, 35-37] (EELS),

ultraviolet photoelectron spectroscopy (UPS) [38] and resonant ultraviolet photo-

electron spectroscopy (RUPS) [38]. The theoretical analysis has been done mainly

with total-energy calculations within the LDA using pseudopotential plane wave

formalism [23, 24, 39], as well as the more recent Hartree-Fock pseudopotential cal-

culations [40]. Recently very accurate self-consistent ab initio calculations for TiO2

have been performed. Prior to those methods the attempts to theoretically predict

the electronic structure of titania were done based on the tight-binding [41-47]

(TB) calculations and the extended Hiickel molecular orbital method [33, 36].

Certain defects in the < i--I I1 structure can impact the electric properties

of titania. Titania is an oxygen deficiency material and usually it is considered

n-type semiconductor. The Fermi-level therefore is not at a fixed value since the

production method will determine the oxygen deficiency and therefore the Fermi-

level shift. This is true for both anatase and rutile. In addition one of the most

common defects in titania is the Ti+4 substitution by Ti+3 (and often Ti+2 an

Ti+1) [48, 49], which also creates a charge imbalance that beyond for the electrical

properties, can affect spectroscopic techniques that rely on the electronic charge,

such as XPS. Those Ti cations can be generated by annealing, sputtering or

chemical reduction.

2.2.1 Anatase

Figure 2-2(b) shows the anatase band structure. The bandgap has been

experimentally measured and is 3.2 eV [50], while the theoretically determined

values can vary from 2.2 eV up to 3.89 eV [25]. Those differences are related

to the number of atoms accounted to the calculations and most important the

non-constant bond length in the < I ,-- (section 2.1.1). For this research the

experimental value of 3.2 eV, which has been repeatedly confirmed [50], will be

accepted as the bandgap energy. The width of the valence band is 4.75 eV and

the distance between the uppermost conduction band state and the lowermost

valence band state is 17.7 eV [25]. Most of the theoretical calculations show that

the bandgap is almost indirect, which is not correct. It is often attributed to the

fact that anatase has a very unstable structure [25].

Anatase also has a very high carrier mobility, 80 cm2/V s [51], (89 times faster

than rutile) [52]. Since the bandgap is 3.2 eV the main absorption peak is at 395

nm. The Hall mobility is 20 cm2/V s at room temperature [53].

2.2.2 Rutile

Figure 2-2(a) shows the electronic structure of rutile. Rutile has a bandgap

that experimentally has been measured to be 3.0 [54] and with calculations it is

1.78 eV up to 3.73 eV [23, 55]. In this case the reason for the large variation is

primarily the number of atoms accounted in the calculation and secondarily bond

length variations. The upper valence band is composed of 02p orbital and has a

width of 5.4 eV. The lower 02s band is 1.94 eV wide [30]. The separation energy

between the upper conduction band and the minimum valence band has been

measured experimentally and is 16-18 eV [30]. The lowest conduction band consists

ofn two sets of Ti3d and is 5.9 eV wide [25].

2.3 Semiconductor Photocatalysis

The term ,pi.. .... irln,.: is still under debate since strictly the term implies the

initiation of reactions in the presence of light only something that is not accurate

in the case of semiconductor photocatalysis, since in this case the presence of the

semiconductor is equally important [56]. But for the purpose of this research the

term photocatalysis will be used, and will denote the reaction that takes place on

the surface of a semiconductor in the presence of a certain range of radiation.

The first report on photocatalytic activity was by Becquerel in 1839 when

he observed voltage and electric current on a silver chloride electrode when it

is immersed in electrolyte solution in the presence of sunlight [57]. Technically

all semiconductors can di-pl wi photocatalytic properties, but usually the oxides

and compound semiconductors are demonstrating significantly better results

[5, 58, 59]. The ability of a semiconductor to undergo photocatalytic oxidation is

governed by the band energy positions of the semiconductor and redox potentials


,_ H //H2
O2J /OH2

3.0 eV 3.2 eV 3.2 eV
I 2.7 eV 2.2 eV -

TiO2 (rutile) TiO2 (anatase) SrTiO3



-4 L

Figure 2-3: Schematic diagram representing the main photocatalysts with their
bandgap energy. In order to photo-reduce a chemical species, the
conductance band of the semiconductor must be more negative than
the reduction potential of the chemical species; to photo-oxidize a
chemical species, the potential of the valence band has to be more
positive than the oxidation potential of the chemical species. The
energies are shown for pH 0.

of the acceptor species. The later is thermodynamically required to be bellow

(more positive than) the conduction band potential of the semiconductor [5, 59].

The potential level of the donor needs to be above (more negative than) the

valence band position of the semiconductor in order to donate an electron to

the vacant hole. Figure 2-3 are shows some of the most popular semiconductor

photocatalysts represented with their band energy positions. The internal energy

scale is given on the left for comparison to the Normal Hydrogen Electrode (NHE).

The positions are derived from the flat band potential in a contact to a solution

of aqueous electrolyte of pH 0 [59]. Among them TiO2 is the most popular. It is,

efficient, effective, requires shallow UV radiation, is very cheap to manufacture,


environmentally safe and easily incorporated with other materials. Since 1972 when

the ability to split the water under UV radiation was first discovered [6] there has

been great work in understanding the mechanism and the reactions that take place.

2.3.1 Basic Principles

Figure 2-4 schematically represents the steps of photocatalysis. Initially when

a photon of proper energy (hv > E,) strikes the surface of the semiconductor

it generates an electron hole pair (h+ e-). Both electron and holes, either

recombined or migrate to the surface, where, they proceed with chemical reactions.

The holes are generating [OH*] and the electrons H202. A very important factor

for those processes is the required time. Here are summarized the main reactions

and the time required for each one [4]. The required time has been measured with

laser flash photolysis [60, 61]:

('C! rge-carrier generation

TiO2 + hv h + e-, 10-15s (2-1)

('C! rge-carrier trapping

hb+ > TiVOH {> TiVOH}+ 10 x 10-9s (2-2)
ec+ > Til"OH {> Ti'OH}, 100 x 10-12s (2-3)
hb+ > TiJV > TiI', 10 x 10-9s (2-4)

('! i rge-carrier recombination

eb + {> TilvOH*}+ > TilVOH, 100 x 10-9s (2-5)
h+ + {> TiI'OH} > TilVOH, 10 x 10-9s (2-6)

Oxidation or reduction

{> TiVOH'}+ + Redo0 > TiVOH + Red'+, 100 x 10-9s (2-7)
et + Ox > TiIVOH + Ox*+, 10-3s (2-8)

According to the above proposed mechanism the overall quantum efficiency

depends on two 1i i. Pr types of reactions, the carrier recombination and the

Vacuum Vacuum

C.B. Xs ,s C.B. Xs ,s

} Trap levels
Ef ----.E. .-- E ----. E .

V.B. V.B.

(a) (b)

Figure 2-5: Titania band structure (a) before and (b) after doping. The transition
metals are interstitial or substitutional defects in the structure of
titania and generate trapping levels in the bandgap.

Since 1972 there has been extensive work towards all three types of photocatalytic

enhancement with the titania/semiconductor and titania/metal coupling more

dominant since they are easier to achieve.

Doping of titania. A great deal of work has been done the last few decades

to dope titania with transition metals, N [62] and C [63, 64]. In general transition

metals are incorporated in to the structure of titania and occupy substitutional

or interstitial positions. It is a very common defect in the case of semiconductors

since it generates trap levels in the bandgap. Figure 2-5(a) shows the electronic

structure of titania before the doping. After the doping (figure 2-5(b)) the bandgap

has been modified with the addition of the trapping levels. The trap levels are

usually located slightly below the lower edge of the conduction band and usually

are in a form of a narrow band.

There are several advantages to this modification. Before the modification

the required photon energy had to satisfy the condition hv > E,. After the

modification the required energy is going to be hv > (E, Et) where Et is the

lower edge of the trapping level band. In addition the electrons that are excited at

those levels are trapped, and the holes have sufficient time for [OH*] generation.

Even in the case that hv > E, and the electron is excited to the conduction band,

then during the de-excitation process the electron is going to be transitioned from

the conduction band to the trap levels and then to the valence band which again

retards the recombination and therefore increases the overall efficiency.

The most common transition metals used are Fe+3, Cr+3 and Cu+2. Fe+3

doping of titania has been shown to increase the quantum efficiency for the

reduction of N2 [65-67] and methylviologen [65] and to inhibit the electron hole

recombination [60, 61, 68]. In the case of phenol degradation Scalfani et al. [66]

and Palmisano et al. [69] reported that Fe+3 had little effect on the efficiency.

Enhanced photoreactivity for water splitting and N2 reduction have been reported

with Cr+3 [69-72] doping while other reports mention the opposite result. Negative

effects have been also reported with the Mo and V doping, while Gratzel and Howe

reported inhibition of electron hole recombination. Finally Karakitsou and Verykios

noted a positive effect on the efficiency by doping of titania with cations of higher

valency than Ti+4 [73]. Butler and Davis [74] and Fujihira et al. [75] reported that

Cu+ can also inhibit recombination.

Coupling with a metal. In photocatalysis the addition of metals can affect

the overall efficiency of the semiconductor by changing the semiconductor surface

properties. The addition of metal which is not chemically bonded to the TiO2 can

selectively enhance the generation of holes by scavenging away the electrons. The

enhancement of the photocatalyis by metal was first observed using the Pt/TiO2

system [76, 77] by increasing the split of H20 to H2 and 02. In particular cases the

addition of metal can affect the reaction products.

Figure 2-6 demonstrates the effect on titania band structure when titania is

coupled with a metal. In general when a semiconductor that has work function 4,







. Eg EVnt





Figure 2-6:

The principles of rectifying contact between titania (Eg=3.2 eV) and a
metal with work function (/.. ), in this example 5 eV, greater than the
affinity (Xs) of titania. (a) Before the contact, and (b) after the
contact, where a barrier is formed to prevent the electrons of crossing
back to the semiconductor. The ELt is the Fermi level if titania is an
intrinsic semiconductor and Ef is the Fermi level as an oxygen deficient

is compared with a metal with work function of '-, > s the Fermi level of the
semiconductor, El, is higher than the Fermi level of the metal E' (figure 2-6(a)).
So when the two materials are brought in contact (figure 2-6(b)) there will be


s j
01 XS

1 T
0s Xs

. 1 1

electrons flowing from the semiconductor to the metal until the two Fermi energy

levels come to equilibrium. The electrons transition will generate an excess of

positive charge that creates an upward band bending. This bending creates a small

barrier (in the order of 0.1 eV) that excited electrons can cross and be transported

to the metal. From the moment the electrons migrate to the metal it is not possible

to cross back since the barrier for this action is larger and therefore the electrons

will remain in the metal. 1

The earliest work on titania metal was the Pt/TiO2 electrode for the split

of water [76, 77]. Currently the most effective metal/TiO2 interface is achieved

by colloidal suspension [78]. It was found that in the case of Pt/TiO2 system

the Pt particles are gathered in the form of clusters on the surface of TiO2 [79].

Other metals have also been investigated. Ag has been found to increase the

efficiency [80]. Other transition metals such as Cr+3 negatively modify the surface

by creating recombination sites. Although in principle all metals can be used, noble

metals are preferred since they have higher work function and better conductivity.

In all cases high solids loading will affect the kinetics of the system, the light

distribution and eventually decrease the overall efficiency [81].

Coupling with a semiconductor. Coupling a semiconductor with a pho-

tocatalyst is a very interesting way of assisting the photocatalysis. Figure 2-7

demonstrates the principles of the TiO2 coupling with another semiconductor. In

this example as titania is considered the anatase phase, while the the semiconduc-

tor is the rutile phase. When two semiconductors are brought together, as in the

previous case, the Fermi levels tend to balance so electrons are flowing from the

semiconductor with the highest Fermi level to the semiconductor with the lowest.

1 According to quantum mechanics there is a finite possibility that the electrons
can cross back, but the number of the electrons that can do that is insignificant.


C.B. (





C.B. X
XS 0





Figure 2-7:

The principles of rectifying contact between anatase (a) titania
(E~ 3.2 eV) and and rutile (r) titania (E =3.0 eV). (a) Before the
contact and (b) after the contact, where a barrier is forming to prevent
the electrons created in anatase crossing to the rutile. On the other
hand holes created into anatase can migrate to rutile. So the couple of
anatase-rutile is creating and effective electron-hole separation.

This charge transfer will create an excess of positive charge to the semiconductor

that had the highest Fermi level and an excess of negative charge to the semicon-

ductor that had the lowest energy (figure 2-7(b)). By light illumination, e- h+

pairs are generated in both semiconductors. The barrier that forms separates the




1 I-_


electrons in the conduction band, but at the valence band the holes are free to

move and based on the energy diagram they move from the semiconductor with the

larger gap to the one with the smaller. In this case the composite material is acting

as a charge separator. The holes are gathered in the rutile where they create an

excess of holes, and despite the fact that the recombination is still the main process

the excess of holes will be enough to photo-oxidize the organic molecules.

In addition semiconductors can be used as a hole or electron injector. In order

to achieve optimum results a candidate semiconductor has to satisfy the following


Have a proper band-gap

Have a proper position of the Fermi energy level

Have proper relative position of the conduction and valence band to the

vacuum level.

The combination of the bandgap and Fermi level will determine if there are holes

or electrons that will be injected and towards which direction. Thus in order for

two coupled a semiconductor with titania in order to enhance the photocatalysis,

the semiconductor has to have very specific properties. This is the reason that this

technique, despite its simplicity, ease of manufacturing and very promising results,

is not very widely applied. Systems that have been developed are the TiO2/CdS

[82], TiO2/RuO2 [83] and Anatase-TiO2/Rutile-TiO2 [52, 84]. The last one is a

system commercially available from Degussa, known as Aeroxide P25, and is the

most powerful commercial, particulate, photocatalytic system [84]. The excellent

and uniform properties have established it as benchmark material to compare

photocatalytic efficiencies.

2.4 Applications of Photocatalysis

In this section are reviewed the main applications of the photocatalytic

systems that have been described above. The most popular uses are in environ-

mental application and photovoltaic cells. There are other applications such as

anti-fog coating and pigments in paints, but since they do not utilize the electrical

properties of titania, they are not going to be explained here.

2.4.1 Environmental Applications

During the last few decades the environmental applications of TiO2 have

attracted a great deal of attention since titania can be the base of low maintenance

systems. So far they mainly focus on water and air treatment and the objectives

are primarily the removal of organic contaminants [4, 85-87] and secondarily

biocidal applications [3, 8, 9, 11, 14, 88]. Although the systems can equally target

biological contaminants the effectiveness is less or equal to other competitive

technologies (chemical disinfection). So the biological applications, although they

are unique and ini. I ii are not widely utilized.

Several reactors configuration have been developed for the most effective

removal of the coil ~iiiii I~1-[ [-91]. One of the most popular configurations,

mainly for experimental application, is the slurry reactor, where the water is

mixed and agitated with titania particles under the presence of UV radiation. The

main advantage of this configuration is the high surface area that allows faster

processing. The main disadvantage is the separation of the particles after the

reaction, which is a very tedious process. They can be separated by filtration,

centrifugation, coagulation and flocculation [86, 92, 93]. Recently magnetic core

has been used to assist the dispersion and recollection of the particles [94]. An

alternative to the slurry reaction is the flat bed reactor where the particles are

immobilized on a ceramic membrane [95]. The efficiency is lower compared to

the slurry reaction due to the lower surface area, but the system does not need

any kind of separation, which adds to the overall efficiency. Recently in order to

increase the surface area of the titania the particles have been coated on tubes [95],

glass beads [96], fiber or woven glass [97].

2.4.2 Photovoltaic Cell

Solar cells have been used the past few decades with great success in small

devices. In 1991 Gratzel and Oregan [98] reported a high efficiency solar cell based

on TiO2. The titania used in those cells is usually dye sensitized [99-101].

The basic titania cell consists of a sandwich of a TiO2, sensitizing dye,

electrolyte and the catalyst between two conductive transparent electrodes. The

substrate usually used for this application is a standard transparent electron

conductor (TEC) glass with high optical transmission and low resistance. Titania

is an excellent material to be used as base since it carries a good combination of

optical and electrical properties. The dye is required to absorb the sunlight and

inject electrons into titania with almost 10t' '. efficiency. The oxidized dye molecule

is then reduced by the redox electrolyte. The electrolyte itself is then reduced at

the counter electrode. The cycle excitation-oxidation-reduction is then repeated.

Dye sensitized solar cells (DSSCs) continue to attract much attention as

viable systems for conversion of solar energy [102]. A titania cell that is sensitized

by a RuN3 dye achieves the highest efficiency. The best efficiency reproted is

10'-. [102]. Retartation of the recombination can further increase the efficiency

of the cell. The properties of these films depend on the phase, morphology and

preparation method that were used. There are a wide variety of techniques that

those films are synthesized. Traditional techniques include CVD, aerosol pyrolysis,

electrodeposition and sol-gel processing [100]. Most of them lead to amorphous,

partially i i -- 11i. .1 or fully
is still considered the best material, but recently brookite was reported to be

successfully used as the electrode material.


These processes are expected to be sensitive to the i -1I I structure, size and

morphology of the exposed lattice planes as it was shown, as well as to the bandgap

and to the flat band potentials. Solar cell photopotential is especially sensitive to

the nature of the semiconductor surface that determines largely the occurrence of

reverse reactions (i.e., recombination). The best actual solar cells work with the

I2/I- (or Br2/Br-) couple, because of a slow kinetics for I2 reduction on SnO and

especially on TiO2 surfaces.


Carbon nanotubes have been discovered by Iiiii i, [103] in 1991 and since

their discovery they have attracted a great deal of attention due to the exceptional

electronic [104], thermal and mechanical properties [105]. Ii'ii: i reported the

creation of multiwall carbon nanotubes (\IWNT) with outer diameter up to 55

A and inner diameter down to 23 A. Since that time extensive theoretical and

experimental research for the past decade has led to the creation of a rapidly

developing research field. In 1993 Bethune et al. [106] reported the discovery of

the singlewall nanotubes (SWNT). The very small diameter of the single nanotubes

and the very big length makes them behaving as quantum wires, giving them

very interesting properties. Due to the fact that the SWNT usually contain a

small number of carbon atoms (usually < 102), they have attracted almost all

the theoretical work. They possess some remarkable electronic, mechanical and

thermal properties that are related mainly to their diameter and chirality. Since the

nanotubes are the photocatalytic template, this chapter will give a general overview

of their unique electrical properties. Initially these properties will be described for

the SWNT that have been more intensively studied and understood. Later some of

the concepts will be expanded to include the MWNTs. Focus will also be given to

the physics of the nanotubes and especially the structure and how the structure is

related to the electric properties and the Raman active vibrational modes. The last

part of this chapter will discuses and compare the several production methods of

nanotubes and how those methods eventually will effect their properties.

3.1 Bonding, Structure and Physics of Single-Wall Carbon Nanotubes

To understand the structure of the nanotubes it is critical to review the

different bond structures of carbon. Explaining the physical properties of the

single and multi wall carbon nanotubes it is required to derive certain geometric

relation and explain the basic notation used for the NTs. It is important also to

describe several symmetries of the tubes, and how they correlate to the vibrational

frequencies. Those frequencies are crucial for explaining in chapter 6 in this

document bonding and electronic behavior.

3.1.1 Bonding in Carbon Materials

A carbon atom has six electrons from where the first two are occupying the

Is state and the other four are at the sp px and pz or sp2 and pz or sp3 hybridized

orbitals depending on the structure. The sp3 orbital is used for example at the

diamond structure, resulting three dimensional interlocking structure that is

responsible for the extreme hard nature of diamond [107]. In graphite, the three

outer shell electrons occupy the three sp2 orbitals, that is coplanar, and form

three in-plane bonds (a bond) and one out-of-plane bond with the pz (T bond)

orbital that is perpendicular to the a bond plane[108]. This results in 1]i, i. v, ,Ib1

structured carbon sheet (graphene sheet). The graphene sheets are held together by

van der Waals forces. The a bond in the sp2 orbital is 0.14 nm long and has energy

of 420 kcal/mol, where in sp3 it is 0.15 nm and has energy of 360 kcal/mol. It is

obvious that the graphite sheet is stronger in the plane direction that diamond.

Since the carbon nanotubes are rolled graphene sheets the bonding is essen-

tially sp2. However, due to the curvature of the tube, the a and 7 bonds are going

to be re-hybridized. The new structure push a bonds out of the plane, all at the

same direction (towards the center of the tube). To compromise the charge shift

the 7 bond will be de-localized to the direction outside the tube. This configuration

will make the tubes mechanically stronger and electrically and thermally more

Figure 3-1: The 2D graphene sheets is shown with the a, and a2 specifies the
chirality of the nanotube. The chiral vector, Ch, is the OA, while the
translation vector T is the OB. Also ib is the rotation angle and r the
translation. Those two are constitute the symmetry operation
R= (Tr).

conducting than graphite. The flexibility of the a bond allows the incorporation

of topological defects, such as pentagons or heptagons, that allow the formation of

caps, bend, toroidal or helical tubes [109].

The fullerenes C60 are made of 20 hexagons and 12 pentagons [110]. The

bonding is also sp2, although due to the high curvature it resembles sp3. This

unique structure gives to the fullerenes a very interesting set of properties.

3.1.2 Structure and Notation

A SWNT can be thought of as a graphene sheet rolled seamlessly in a cylinder

[111]. It usually has 10-40 carbon atoms in circumference and is capped. The

direction that the graphene sheet is rolled is called chirality and it is specified by

the chiral vector Ch (figure 3-1). The honeycomb structure is described by the

vectors al and a2 and all the vectors can be described as a linear combination of

those two vectors. Ch can be defined as (figure 3-1)

Ch nal + ma (n, m) (3-1)

which often is denoted with the (n, m) symbol. A very important variable is the

angle 0 which is the angle of the chiral vector with the al direction [112]. The al

direction is called ..:/..i Consequently nanotubes rolled to that chiral direction

are called ..':.ij [113]. There are many possible directions that the graphene sheet

can be rolled with different properties (figure 3-2). The direction that has 0 = 300

is called armchair [112]. All the other nanotubes for which 0 < 0 < 300 are called

chiral. For angles 0 > 300 and 0 < 0 rotational symmetry rules apply. The tube

diameter dt can be written in terms of the integers (n, m) as:

dt C h 1Vacc (m2 nmn2)12 (3-2)

where ace is the nearest neighbor C-C distance (1.42 A in graphite). From the

geometry in figure 3-1the cos 0 and sin 0 can be calculated,

*Vm 2n+ m
sinO =- cos 0 = 2n (3-3)
2/2 + + n22/ 2v/m2 +nm +n2

Consequently the chiral angle 0 is

0 tan-1 [ 3m- (3-4)
m + 2n

The (dt, 0) pair can completely describe the nanotubes and occasionally it is used

as an alternative to the (n, m). The translation vector T is another important

vector, which on the nanotube denotes the longitudinal direction and is vertical to

the Ch (Ch T = 0). It is defined as

T = tiai + t2a2 (t, 2)


where the coefficients t1 and t2 are related to the n and m by

(2m + n) (2n + m)
ti = t2 = -(3-6)
dR dR

where dR is the greater common divisor of (2n + m, 2m + n) and is given by

dR d, if n m is not a multiple of 3d, (3-7)
3d, if n m is a multiple of 3d

where d is the greatest common division of (n, m). The magnitude of T is IT| =

T = 3Ch/dR. As unit cell of the nanotube is defined the area delineated by

the vectors T and Ch. So for instance in figure 3-1 the unit cell is defined by the

OBB'A parallelogram. The number of hexagons, N, contained within a unit cell of

a nanotube is determined by the integers (n, m) and is given by

N 22 + (3-8)

where dR is defined by equation 3-7. The carbon nanotubes are usually capped.

The cap can be thought of as a fullerene (C60) that has been bisected at the

equator. So for example if the C60 is bisected normal to a five fold symmetry axis

then that cap is suitable for armchair tube, while if it is bisected normal to the

3 fold symmetry axis then the resulting cap is suitable for a zigzag tube [112].

Since there are many diameters there are many different caps that can fit them

[112, 114]. Figure 3-2 shows several rolling directions and based on those direction

the number of distinct caps that can be joining them seamlessly.

3.1.3 Symmetries and Vibrational Frequencies

A very general way to simplify the analysis is to assume that the nanotubes

have very big length compared to the diameter and therefore ignore the caps. In

general we can distinguish two 1i i, ri" types of symmetric groups, symmorphic or

non-symmorphic. The zigzag ((n, 0) tubes) and armchair ((n, n) tubes) belong to

Figure 3-2: The graphene sheet is shown with the (n, m) pair which specifies the
chiral nanotube. The pair of integer (n, m) in the figure specifies the
chiral vector Ch for carbon nanotubes, including zigzag, armchair and
chiral tubules. Below each pair of integer is listed the number of
distinct caps that can be joined continuously to the cylindrical carbon
tubule denoted by (m, n) [ref]. It is also denoted the conduction state
of every chirality.

the first group while the other chiral belong to the second. The basic difference

that in the case of symmorphic the translational (7) and rotational (T) operation

(both shown in figure 3-1) can each be executed independently, while for the

non-symmorphic this is not true.

The complete analysis is very complicated and is beyond the scope of this

research. Briefly here will be mentioned the very basic principles. Due to their

high complexity the chiral tubes are not going to be included in the analysis. From

equation 3-8 it can be calculated that for certain structures the N can be very

large. For example for the (30, 15) N = 210 [103, 115]. The symmetries for those

structures are very complicated [114]. For zigzag (n, 0) and armchair (n, n) are less

complicated. The (n, n) and (n, 0) the symmetry groups can be described by Dih

or Dnd, that are even or odd, respectively.

The symmorphic symmetries usually have relative small area of 1D unit cell

(Ch T), therefore the number of phonon branches or number of electronic energy
bands are small. On the contrary for the chiral tubes that number is very big, since

the area of the 1D cell is large. For the zigzag tubes (n, 0) there are 4 x 3n = 12n

degrees of freedom with 60 phonon branches, having symmetry types (for n odd,

and thus Dnd symmetry) [114]:

vib 3A + 3A + 3A2 + 3A + 3 32u (3-9)

+ 6E1g + 6E1, + 6E2g + 6E2u

+ .. + 6E[(n-1)/2]g + 6E[(n-1)/2]u

From those only 7 are non-vanishing modes that are infrared active and 15 that

are Raman active, but they are not all detectable. It was found that increasing

the diameter of the zigzag tubes the number of active modes does not increase.

This concept can be proved for armchair and chiral tubes, since it is a symmetry

imposed result. In chapter 6 are explained the 1 i ri Raman lines that can be


3.2 Electronic Properties of SWNT and MWNT

3.2.1 Electronic Properties of SWNT

Their unique electronic properties are attributed to the different quantum

confinement of electrons. We can see three different directions that based on the

geometry it will result in, or not confinement. (i) In the radial direction, electrons

are confined by the mono-lriv- thickness of the graphene sheet. (ii) Around the

circumference of the nanotube, periodic boundary conditions come into pl ,iv. As

seen in the previous section the radius, therefore the boundary conditions, depends

on the (n, m) configuration. For example, for a (5, 5) the radius, dt, is 6.78 A, for

a (10, 0) it is 7.83 A [115]. So the circumference boundary conditions vary even

for tubes that are at the same category (armchair or zig zag). (iii) Finally the

2 ,
L 2a


2r kx
2a 0


Figure 3-3: The dispersion for graphite as calculated from equation 3-10.

direction parallel to the axis (T direction), since it is considered infinite there is no


Because of this 1D quantum confinement, the electrons can only propagate

along the nanotube axis designated by the vector T, and so their wavevectors k

point towards this direction. The resulting number of one-dimensional conduction

and valence bands effectively depends on the standing waves that are set up around

the circumference of the nanotube. These simple ideas can be used to calculate the

dispersion relations of the one-dimensional bands, which link wavevector to energy,

from the well known dispersion relation in a graphene sheet.

In the simplest model [113, 116, 117], the electronic properties of a nanotube

derived from the dispersion relation of a graphite sheets with wave vectors (kx, ky):

E(k, ky) 70 1 + 4cos 2vk cos ( +4 cos2 (2 (3-10)

where o7 is the neighbor-hopping parameter (usually 7o = 2.5 3.2 eV, [113, 116

118]) and a is the lattice constant a = 2.46 A. Figure 3-4 shows the plot of this

dispersion relation.

When the periodic boundary conditions are imposed along the tube circum-

ference (C direction) the k = (k, ky) is quantized along that direction. It has to

satisfy the condition k C = 27q, where q is an integer. For the armchair (n, n) this
translates to
m 27r
kj (m 1,..., N) with N 5 (3-11)

replacing this value in equation 3-10, and simplifying ky with k we get

E (- ) 7 1 4cos ( cos (2) +4cos2 (2a (3-12)

where -T < ka < 7 and m = 1,..., 5 in which k is one-dimensional vector along

the axis of the tube (T direction). The plus and minus signs are denoting the
unfolded and folded energy bands, respectively.

Similarly for the case of the zigzag tubes we get the relation

m 27r
S N, a (m= 1,...,Ny) with N = 9 (3-13)

The energy dispersion relation in this case is calculated to be

zV3ka m 2 ( ka\
E`9(k) io1 4 cos cos) + 4cos2 (3-14)

where / < ka < K and m 1,..., 9 in which k is one-dimensional vector

along the axis of the tube (T direction). In addition according to the circumference

k k
(a) (b)

Figure 3-4: The dispersion energies for (a) armchair and (b) zigzag semiconductor
as are calculated from equations 3-14 and 3-12. The different
branches have been labeled according to [116].

direction boundary condition in order to have metallic tubes;

(n- m) =3q (3-15)

That means that one third of the different nanotubes structures is metallic and two

thirds are semiconducting. Figure 3-2, shows the conductivity states for different

chiralities. For semiconducting tubes the band-gap (E,) is [119-121]

E, = 2dccd70 (3-16)

So far for this approach the only weakness is that it did not account the re-

hybridization of the a 7r orbital due to the curvature. This effect can be included

in other approaches such as the first principle calculation ab-initio [122-125]. In

this new approach it is proved that for small diameter tubes (< 1.5 nm) a band

gap opens that is about 0.02 eV for non-armchair nanotubes, that still satisfies the
condition 3-15 [126]. However this phenomenon dissipates fast for larger diameters

tubes. Therefore the graphite model can be used as a good approach to describe

the SWNT with different chiralities. STA\i studies have confirmed the accuracy

of the model [123, 126] and also the existence of the small band-gap predicted by

ab-initio calculations [126].

It has been experimentally confirmed that a SWNT [127], a SWNT rope

[128] and a MWNT [129] behave like a quantum wire intrinsically. The conduc-

tance is given by

a ~[ 2 -- M (3-17)

where jo = (2e2/h) = (12.9 kQ)-1 is quantized conductance. M is an apparent

number of conducting channels, that includes all the possible interactions, such

us electron-electron coupling, inter-tube coupling effects. For example for a

SWNT that value is 2. In a SWNT there are also impurities, structural defects,

coupling with the substrate that will further reduce the conductivity. Therefore the

experimental data have large variations from the predicted values, but they follow

the same trend.

The most important information that the graphite model can predict is the

density of states (DOS) [130-132]. According to that model the density of state

p(e) is
4 2 +0
p(C) /-3 > g(, gC (3-18)
l V37oa
S for ec > |em
g(e) = v~ (3-19)
0 for Ie < 1m
13q n + ml oa (320)
1d (3-20)

Calculations based on this model predict again that the armchair and zigzag

configurations have a continuous DOS while for the chiral a small band gap exists

[119, 133, 134]. Figure 3-2 shows the different directions that the graphene sheets

can be rolled and it is denoted if the tube is metallic or semiconducting.

3.2.2 Electronic properties of MWNT

It has already has mentioned in the previous chapter that the MWNT behave

as a wire with the conductance to follow the simple relation [129];

a= cA, [ M (3-21)

For the case of the MWNT the value of M is significantly i._-. -r than for the

SWNT to account for more conducting channels. In addition the rimultil i-r

structure increases the probability to have armchair or zigzag tubes that will

increase the conductivity. While the diameter is increasing the electrons on the

tube are less confined and the electron distribution resembles more the structure

of graphite. this is due to the re-hybridization of the a and 7r orbital, that is less

intense and the tubular structure approaches more the graphite structure. This is

obvious from equation 3-17 where while the tube diameter increases the energy

gap is diminishing even for the semiconducting tubes. So in general MWNT are

in their 1i, i i ly conducting and behave as nanowires. But there are still chances

that the tubes will be semiconducting, depending alv--,v- on the arrangement of the

tubes certain defects and < i I ,ll;;iiy.

3.3 Carbon Nanotubes Growth Mechanisms

There are two basic commercially available methods for producing carbon

nanotubes. The arc discharge and the C'! iii, I. Vapor Deposition (CVD). Both

have advantages and disadvantages that can be directly related to the properties

of the tubes. Generally speaking the two methods are competing at the quantity

versus quality, where CVD is designated for quantity and arc discharge is for


3.3.1 Arc Discharge

In general carbon nanotubes that are produced with carbon vapor that

is being created by the arc discharge, have fewer defects compared to other

techniques. The reason for that is the high growth process temperature that

ensures perfect annealing that eliminates most of the defects. The MWNT that

are produced via arc discharge are perfectly straight. The fewer defects have an

immediate dramatic impact on the tube properties such as, electric and mechanical.

One of the main disadvantages is the limited yield that this method has. Besides

the low yield it is a highly time consuming process. So in general if a a high yield

of nanotubes is required this method is not recommended, on the contrary if more

defined, and better properties is required then arc discharge is a very good solution


The most common set-up for arc-discharge two graphite electrodes of diameter

6-12 mm, that are kept in distance of 1-4 mm in a chamber that is filled with He-

lium. DC current operates the two electrodes. DC current and Helium are the two

factors that immediately influence the yield. While the positive electrode (anode)

is consumed a cylindrical slag is being deposited on the cathode. The alignment

of the electrodes does not effect the MWNTs but can effect the properties of the

single wall tubes [135].

3.3.2 CVD: Thermal CVD, PE-CVD

Since the application field of the nanotubes is growing the demand for higher

yield production methods is also growing. One of the most promising techniques

is the C'! i ii dl Vapor Deposition (CVD). It has a large knowledge base since it is

been used extensively in electronic applications for the last few decades.

The nanotubes that are CVD grown have a lot of structural defects due to the

low synthesis temperature during the growth process. An approach to improve this

is annealing the tubes, which will reduce the defects but in no case will have the

same results as the Arc-discharge [135].

The apparatus for CVD grown nanotubes is simple, which is also reducing a

lot of the cost of the production. In a quartz tube with very precise temperature

control, a substrate is placed in carbon containing gases, such as CO, CH4 or

higher order hydrocarbon, are flown in. To assist the reaction often a thermal

source is used, such us IR lamp (Thermal CVD) [135-137] or plasma (PE-CVD)

[138]. The growth rates can be controled precisely and can go from a few nm/min

up to 5 pm/min. In addition metal catalyst can further assist the yield. One of the

'i.-.- -1 advantages of CVD is the ability to grow on a patterned substrate, which is

desirable for microelectronic applications. The purification of the tubes in this case

is a necessity since they contain metal catalyst and different amorphous carbon

structures. There are many v-,v to purify the tubes; hydrothermal treatment [139],

H20-plasma oxidation [140], acid oxidation [141], dispersion and separation by

micro-filtration [142] and high-performance liquid chromatography [143].


In the previous two chapters the main properties of titania and the carbon

nanotubes were reviewed. This chapter describes the process of combining those

two materials. There are many possible combinations, but in this research the

objective is to apply the titania in the form of a thin coating on the surface of

the MWNTs in order to maximize the contact between the two materials. There

are certain design parameters that have to be satisfied in order to obtain the

optimum results. The first section explains those parameters and following that

are explained the materials selection and preparation. Later a small introduction

to the Sol-Gel chemistry is given and based on that, the choice of chemicals and

precursors is explained. Finally fundamental characterization will follow to provide

arguments for the satisfaction or not of the design parameters and in what extend

it was achieved. The actual photocatalytic efficiency as well as the detailed study

of the interface between the MWNTs and the titania will be discussed in separate

chapters later since they are the main focus of this research.

4.1 Design Parameters

As stated in the introduction the purpose of this work is to combine those

materials and their properties to produce a highly efficient photocatalytic particle.

The main objective is to synthesize a thin coating of titania to cover the surface of

the MWNTs. The process has to satisfy certain criteria.

The coating has to be the anatase phase of titania: As seen in previous

chapter 2 anatase is the most photocatalytic active phase of titania. That

phase is also thermally very unstable and therefore obtaining anatase is a

non-trivial process with many parameters.

Thin coating will result better photocatalytic performance: The whole

photocatalytic process takes place in a thin 1. -r of about 10 nm. If any

electron hole pair is generated in regions deeper than that, it is going to

recombine before it reaches the surface. In addition increasing the coating

thickness will result lighter color (since the coating will be less transparent)

and therefore the particle will absorb less light.

The coating has to be chemically bonded to the MWNTs: If the coating is

not chemically bonded on the surface of the MWNTs it is possible that it will

flake off. The coated nanotubes will have high tendency to coagulate since

the size is big enough to induce van der Waals forces. Therefore prolonged

sonication will be required to successfully disperse them, which might damage

loosely attached coating.

Individual MWNTs have to be coated: MWNTs have very high affinity into

coagulating. The hydrophobic nature of the tubes will also intensify the

phenomenon of coagulation especially when the solvent is water. In order to

maximize the surface area it is required to minimize the number of MWNTs

.. -.-! ii-. i Ies and separate the bundles.

The number of free titania particles have to be kept minimum: Sol-Gel

is a process that balances between transport phenomena and reaction rate.

Ideally in order to achieve the coating the precursor molecules have to be

transported to the surface of the MWNTs and only after the anchoring they

should react. This balance can be controlled by reaction parameters such as

temperature and pH. However, regardless the values of those parameters there

is alv- i a finite possibility of free anatase particles formation.

With those requirements in mind two distinct set of particles will be syn-

thesized. The first one will be consist of an arc discharge MWNT core and the

next one will consist of a CVD grown MWNT core. As described in the previous

chapter (section 3.3) the difference in the tube production can affect the electrical

properties of the carbon nanotubes. So the purpose of using those two different

nanotubes will be to examine the effect of the electrical properties of the tubes on

the photocatalytic activity. The CVD carbon nanotubes have been mechanically

and chemically shortened, which will result in a dramatic increase of the defects

on the surface of the tubes. The short nanotubes in addition will provide other

advantages. The high aspect ratio of the carbon nanotubes results in a particle

that interacts easily with molecules, but raises issues when is it used to deactivate

objects of comparable size such as spores and bacteria. Bacteria have very compli-

cated surfaces, that usually have fibrils of several pm length that can interfere and

prevent the coated tubes to reach the surface. In addition the spherical shape of

the spores does not allow the use of the whole available surface of the nanotubes.

So reducing the length of the MWNTs will result shorter in particles. Large scale

production of short nanotubes averagee < 1 pm), cannot be achieved with neither

arc discharge method, nor with CVD. They have to be shortened with chemi-

cally assisted mechanical grinding. The short MWNTs will be occasionally called

s-CNTs and the long MWNTs will be called -CNTs.

4.2 Nanotube Selection and Preparation

The carbon nanotubes have to be properly modified to satisfy some of the

coating requirements. They have to be individually suspended, easily dispersed

in solvents and favor the anchoring of the precursor molecules. It is also critical

to characterize the tubes before the coating in terms of i --i lliiiIy and struc-

ture, something that can be used to explain differences in terms of the electrical


4.2.1 Materials Selection

Two different nanotubes were tested as photocatalytic template. The long

nanotubes were ordered from Alfa-Aecar (stock number: 42886) in soot form. The

CVD nanotubes were ordered from NanoMat (product number: 1236YJS) and

were delivered in powder form. According to the manufacturer the tubes were

shortened in a ball mill in a highly acid environment (nitric and sulfuric acid 1:3).

MWNTs from other manufacturer (Iljin Nanotech) were tried, but did not behave

desirably so they were not used. In addition highly conductive activated carbon

from Degussa was used, again with no desirable results.

4.2.2 Purification and Dispersion

The arc discharge nanotubes were obtained in the form of soot. In the soot

along with carbon nanotubes there were many other forms of carbon such as,

carbon fibers, fullerenes and amorphous carbon. Similarly is the situation for the

CVD grown nanotubes. In addition there is residue from the catalyst (in this case

Fe). In order to coat them they have to be purified and dispersed. Since most of

the impurities are carbon nature they can be easily oxidized by acid.

The main route was the same for both materials. The tubes were dispersed in

highly concentrated HNO3 (, :'-. or 10N). The arc discharge nanotubes were in soot

form, so initially the soot was ground with molder and pestle to fine powder. After

that 50 mg of this powder was mixed in 200 ml of the nitric acid. The solution was

sonicated for 3 hours to further disperse the powder. The solution was refluxed

in an oil bath at 140C for 10 h. Then the heat was turned of and the solution

was left for additional 3 h until the temperature drops below 30C. The solution

was then centrifuged and the excess nitric acid was removed. Triple washing with

di-ionized water followed.

The CVD nanotubes were already in powder form and therefore was no

need for grinding. In addition since they were already treated with acid for the

shortening there is no need for extensive purification, but still the acid treatment

is required for dispersion purposes. As previously 50 mg of tubes were dispersed

in 200 ml of HNO3 and sonicated for 3 h. After that the tubes were refluxed again

in oil bath of 100C for 6h and afterwards the solution was cooled down to 30C.

Again the nitric acid was removed with centrifuge, and the tubes were washed with

ethanol three times.

In all cases the nanotubes were not removed from the solvent. During the

purification process there was a ,II' weight reduction. So for the coating process

are left about 30 mg. This value was estimated, by drying and weighing the

remaining nanotubes.

4.2.3 Characterization of the Functionalized MWNTs

The characterization of the tubes was performed with SEM (FEG-SEM JEOL

JSM-6335F), TEM (JEOL TEM 2010F), FTIR (Nicolet MAGNA 760 Bench), Zeta

Potential measurements (Brookhaven ZetaPlus), particle sizing (Coulter Multisizer

III) and thermal gravitational analysis (N\ 1. -I STA 449C Jupiter). The SEM

(figure 4-1) reveals roughly the general characteristic of the tubes. The s-CNTs,

figure 4-1 (b), appear more pure since they have undergone the acid treatment

twice, but they are not straight. On the contrary the L-CNTs, figure 4-1 (a),

are straight. In both cases the tubes appear to be pure and there are no obvious

impurities. After the acid treatment the tubes appear purified with no obvious

impurities (at the order of 3 nm) (figure 4-3(d)) and HR-TEM shows the graphene

1~. rs and the cap of the tubes. The TEM images showed an average diameter of

about 20 nm.

Figure 4-2 shows the TEM images of s-CNT before and after the acid treat-

ment. Before the treatment the tubes appear tangled (a) with m nlr carbon

impurities on their surface (b). The acid removed most of the impurities and the

main features of the tubes such as the cavity are visible. The purity of the tubes

Figure 4-2: TEM images of the s-CNTs. (a) A. 1.i.-_, i .! of s-CNTs. (b) High
magnification of untreated s-CNTs where the impurities around the
tube are visible. (c) Purified s-CNTs where there are almost no
impurities present. It can be seen that they are not straight and that
they have been damaged. (d) Magnification of the treated s-CNTs
where the inner cavity is visible and the outer surface is almost
completely free from impurities. From the images it can also be
concluded that the average diameter is 20 nm.

is demonstrated clearly in figure 4-2(c) where the tubes although are shown to be

.. -i negated they are free of impurities. It is also concluded that the tubes have an



100 nm

.. ": hI -
; -' i. .*: ',;

* ,I O N

-* -.. !.* *, '' -

50 nm

Figure 4-3:

TEM images of the f-CNTs. (a) Agglomerate of f-CNTs before the
acid treatment. It is hold together by the carbon impurities. (b) Single
f-CNT covered by the carbon impurities. (c) After the acid treatment
a bundle on nanotubes. It is also visible some residue of the acid
treatment by products. (d) f-CNTs after the treatment, where most of
the surface carbon impurities have been removed. Again from this
image we can see that the average f-CNT diameter is about 15 nm.

average diameter of 15 nm, which is in agreement with the manufacturer specifi-

cations (10-20 nm with average 15 nm). Similar results can be derived from figure

4-3 for the E-CNTs. Since they are arc discharge (1' .1' by weight MWNTs) they

have more impurities than the short. In figure 4-3(a) the .,-.-regates have big pieces


Figure 4-4: Immediate comparison of the two different kinds of nanotubes. The
images (a) and (b) are for the f-CNTs and the (c) and (d) for the
s-CNTs. The end of the s-CNTs is usually open due to the catalyst
(c), while the end of the -CNTs are capped (a). In addition the
s-CNTs have damaged and not well defined walls (d), while the
-CNTs are very well defined and straight.

of the carbon impurities and in a characteristic picture of an individual tube (figure

4-3(b)) shows the surface to be covered in segments of the amorphous carbon





-50- --


-100 *, I I i i
1 2 3 4 5 6 7 8 9 10 11 12

---- After the acid treatment
-10 -o-- Before the acid treatment



-70 i I\ I
1 2 3 4 5 6 7 8 9

Figure 4-5: The zeta potential for both the f-CNTs (a) and s-CNTs (b). It shows
the shift of the IEP for the f-CNTs (from 7 to 3.5) and the increase at
the surface charge for the s-CNTs (from -10 mV to -37 my for ph 4).


results are d1 ). The bands that have been identified prove the
-r o f-H

Figure 4-6: The FTIR of the MWNTs after the acid treatment (only the s-CNTs
results are di-!, iv, ,t). The bands that have been identified prove the

solvent creationn of the -COOH on the surface of the nanotubes.anol).

Finally the dimeasurement compare the ison of electric point (IEP) and surfaces the charge it isfrence on
nneesary to carie ithIR e Ts aes after the acid treated the s-C ) ad (). These
rHR-TEM imagesults (figure di-4 4(bi). The bands that have been identified provaphene lthe

cases are visible small the of carbonaceous impuritiesOH on the surface of the tubesnanotubes.

thatFinally the direct byproductson of the acid treatment [144, 145]. Althoughe main difference on be

the struremoved it is not necessary since it will bon the s-CNTs appear tuo bes are placed in ally

solvent (water or ethanol).

The measurement of the isoelectric point (IEP) and surface charge it is

necessary to clarify if there is any surface modification of the CNTs. The results

Diameter (gtm)

Figure 4-7:

The differential volume and number of the s-CNTs before and after the
acid treatment.

for the -CNTs (figure 4-5(a)) clearly show a shift to lower values of the IEP and

higher surface charge. The results for the s-CNTs show that there was pre-existing

surface modification, as result of the mechanical-chemical shortening, and therefore

the second treatment just increased the amount of surface charge. In both cases the

change can be attributed to the generation of functional groups on the surfaces of

the MWNTs.

Since the acid used for the functionalization was HNO3 the surface groups

that have been generated on the surface have to be -COO-. DR-FTIR is utilized

to further investigate the surface groups on the surface of the carbon nanotubes.

Figure 4-6 shows the FTIR spectra of the s-CNTs. The bands that identified are

very typical of the -COOH group (1170 C-OH, 3450 O-H and 1720 -COOH)






-- As obtained (Number)
-.......... Acid treated (Number)
- As obtained (Differential Volume)
.-.-.- Acid treated (Differential Volume)

\, I


; *


*1 ,

"""' """' """' """' """

r n





5 Q

""~" ' Y I LY YIY -___~_____~~~1~~1~1 I I I I I I I I I II

- -











Figure 4-8:

/ \\\
/-- \\

/ Impurities
/ burn out \

/ \

- -

S6.02 %
. .I .


L. . . . .. . .. . .. . . . . . .
100 200 300 400 500 600 700 800 900 1000
Temperature (OC)

The TGA/TDA data of the s-CNTs. The peak at the 600C indicates
the burning temperature of the CNTs. It is observed about i'. of the
initial mass residue, which is the Fe catalyst.

[146, 147]. The other bands are characteristic of the carbon nanotubes (1460

C-H, 1640 C C, 2850 C-H and 2970 C-H) [148, 149]. The band at 3450 O-H

is not proportional to the 1170 (C-OH) and 1720 (COOH) but this is due to the

atmospheric humidity. Similar results are obtained for the FTIR of the -CNTs and

they are in good agreement with the literature [146-149].

The final characterization was done by the Coulter Particle size analyzer. The

Coulter is utilizing a laser beam and with light scattering calculates the size of

the particles. The theory that is used at the Coulter instruments is similar to that

used at the Zeta Plus that was used for the measurement of the zeta potential. A

in i ri assumption is that the particles are spherical or can be assumed as spherical.

This is completely wrong for the case of the carbon nanotubes, which are high




1 >




aspect ratio particles (1 : 150 for the f-CNTs and 2 : 70 for the s-CNTs). In

addition since the limit of the instrument is 40 nm what is detected are mainly the

..--,1,ii i iIes and not the individual tubes. However, the instrument can be still

be used to showcase change in the dispersion of the CNTs. Due to the high aspect

ratio of the -CNTs the results cannot be considered accurate, and figure 4-7

shows only the s-CNTs case. The differential volume results are usually considered

more representative and according to the graph there is one order of magnitude

reduction in the diameter after the acid treatment. In both cases (zeta potential

measurements and particle measurements) the results are only used for qualitative


In addition Thermo-Gravitational Analysis (TGA) showed that the -CNTs

are starting to burn at approximately 700C while the s-CNTs are burning at

approximately 600C and they have i'. weight residue that was identified as Fe2O3

which came for the catalyst used during the production (figure 4-8). That was in

agreement with manufacturer statements.

So from this section we can conclude that the two types of carbon nanotubes

used in this research are different regarding the overall structure. Although both

have a concentric tube structure and the characteristic cavity in the center, the

two types are different in quality; the -CNTs are very straight and have very

well defined structure, while the s-CNTs type has damaged walls as result of the

production method and the chemical mechanical shortening. In addition the acid

treatment was proved enough to remove carbon nature impurities and to cause

surface modification to stabilize the tubes, either by increasing the surface charge

(s-CNTs) and by shifting the IEP (e-ANTs).

4.3 Sol-Gel Coating

The Sol-Gel [150] route is a very common and validated way to produce thin

coatings of amorphous and < i-i 1 11iii. materials. For the titania there is a great

deal of attention to this method since the size of the produced particles can be

very accurately controlled and therefore nanosized particles can be easily produced

with very high yield and reproducibility [151, 152]. So for this research Sol-Gel

is the most appropriate method for the generating anatase titania coating on the

MWNTs. This section explains the materials selection and describes the process

that was followed to obtain the TiO2 coating.

4.3.1 Precursor Selection

There is are numerous different methods to produce anatase titania via the

Sol-Gel route. The precursors can be either organometallics or salts. The molecules

will undergo a variety of reactions that will result a three dimensional molecular

network. A common example is the h'lyd .. Ii-- and condensation reactions of metal

alkoxides to form larger metal oxide i --I i1- An alkoxide has an organic group

bonded to a negatively charged oxygen atom; when this oxygen is also bonded to a

metal it is called metal alkoxide. During the hydrolysis [153, 154] all or some of the

organic chains are replaced by the -OH groups.

M (OR), + H20 HO M (OR),_ + ROH +... M (OH), + nROH (4-1)

During condensation reaction [153, 154], the M(OH), are reacting to produce the

metal oxide.

(HO),_, M OH + HO M (OH),_, (HO),_ M 0 M (OH),_ + H20 (4-2)

Or alternatively the condensation can occur from the intermediates of the reaction

4-1 [150].

(RO),_ M -OH + HO M (OR),_ (RO)_ M M (OR),_ +H20 (4-3)

(RO),_ M OH+ RO M (OR) ,_, (RO),_, M 0 M (OR),_, + ROH (4-4)

where M with valence n is the metal and the R are the organic chains. The

reaction is progressing with the h'dl.h .1i -i and the condensation of all the -OR

groups of the (RO),_, M O M (OR),_, to result in the three dimensional

network. In the case of titania this reaction will produce the TiO6 octahedral,

which is the structural element of the anatase and rutile.

One of the factors that can determine the reaction rate is the length of the

organic chain. Usually increase in chain length will result in slower reaction rate.

The chain length is directly related to the mobility of the molecule. In addition the

three dimensional structure and complexity of the molecule will also effect the reac-

tion. More complex structures such as titanium bis-ammonium-lactato-di-hydroxide

(TALH) are less reactive. Significant differences in the reaction have been reported

even in the case of titanium isopropoxide (Ti CH < CH3 [155 158]
and titanium propoxide (Ti (-0 C3H7)4) [159, 160].

There is also the case of the salts that can be used such as titanium tetra-

chloride TiCl4 [161-164] and titanium sulphate Ti2(SO4)3 [165-167]. Titanium

tetrachloride can be directly hydrolyzed to yield the rutile phase of the TiO2

TiCl4 + H20 -- Ti (OH)4 + 4HC1 (Endothermic) (4-5)

Afterwards the reaction progresses similarly to the reaction 4-2. It can also be

used for the production of metal alkoxides that later can be hydrolyzed to produce


TiCl4 + 4ROH -- Ti (OR)4 + 4HC1 (4-6)

The titanium sulfate has more complicated structure and the reaction proceeds as

Ti2 (SO4)3 + 8H20 -- 2Ti (OH)4 + 3H2SO4 + H2



Figure 4-9:

The different Sol-Gel precursors used in this research. (a) titanium
ethoxide (Ti(OC2H5)4), (b) titanium isoproxide (Ti(OC3H)4), (c)
titanium butoxide (Ti(OC4H9)4), (d)titanium
bis-ammonium-lactato-dihydroxide ([CH3CH(O) CO2NH4 2Ti(OH)2,
(e) titanium sulphate (Ti2(S04)3, (f) titanium tetrachloride.

In this research there were five different precursors used; titanium ethoxide

(Ti(-O-C2H5)4) titanium ispropoxide (Ti -0 CH < ), titanium
butoxide (Ti(-O-C4H9)4), TALH, ((CH3CH(O)CO2 )2Ti(OH)2(NH4)2) and

titanium sulphate (Ti2(SO4)3) (figure 4-9).

Several precursors were tried for every case of MWNTs. Initial conditions and

precursor were selected based on the literature. For the titanium sulfate from Lee

et al. [168], for titanium isopropoxide, ethoxide and butoxide from Jitianu et al.

[169] and finally for TALH from Lee et al. [170].The results were judged based on

the repeatability, the coverage of the coating and the number of free particles. The

surface coverage and the free particle formation were checked with the TEM.

4.3.2 Coating Model

To estimate the amount of anatase required to coat the tubes a coating model

has to be developed. A uniform coating of approximately 5 nm will give the

optimum results. The nanotubes have a diameter of 20 nm and average length of 2

pm. The optimum coating will be around 5 nm thick. So

Vnatase = 5 nm 2 7r x 10 nm x 2 pm (4-8)

2 7r 5 x 10-9 10 x 10-9 2 x 10-6 m3 (4-9)

S6.28 x 10-22 m3 (4-10)

Respectively the volume of a nanotube is

V-CNTs -= x (10 nm)2 x 2 pm 0.6 x 10-21 3 (4-11)

The average density of the tubes (PCNT) is 1.1 g/cm3. So 1 mg of MWNTs will
10 3g1.1 g
contain 0.6x1213 2 x 1012. So for every mg of -CNTs the required volume of

anatase is Vtae = 2 x 1012 6 x 10-22 m3 12 x 10-10 m3 1.2 x 10-3 cm3

of anatase. The density of the anatase (panatase) is 3.89 g/cm3, which translates to

approximately 4.67 x 10-4 g or 0.467 mg of anatase or 5.84x10-6 mol for every mg

of CNTs. In all cases minor adjustments were required to minimize the formation

of the free titania particles. In general the quantity that was used was less than

the estimated. The 1n i, r difference between short and long tubes is the length

(which does not effect the coating model) and the diameter (R_-CNT > Rs-CNT) SO


Table 4-1: The calculated initial molecular ratio for the reactions for the -CNTs

Precursors -CNTs (mg) Solvent Precursor (pl) H20 (pl)
Ti(OC2H5)4 30 mg Ethanol 300 ml 36.7 (N/A) 11.68
Ti(OC3H7)4 30 mg Ethanol 300 ml 51.8 (44.0) 11.68
Ti(OC4H9)4 30 mg Ethanol 300 ml 59.6 (N/A) 11.68
Ti2(S04)31 30 mg Water 300 ml 102.5 (106.0) N/A
1Solution of 45'. wt Ti2(SO4)3 in dilute sulfuric acid

the same model can be used for both types of MWNTs with some modification. If

m-CNT is the anatase required to coat 1 mg of t-CNTs then the amount required

anatase RCNT anatase
for 1 mg of s-CNTs is
The equivalent volume of the MWNT can be considered as a sphere of

radius RgNT = CNT/2. The volume is calculated to be VCNT (RCNT)3

4.2 x 10-18 m3 = 4.2 x 10-12 cm3. So the total equivalent volume of 1 mg MWNTs

occupy is V = 4.2 x 10-12 cm3 2 x 1012 ,8.2 ml. Therefore to ensure that the

30 mg of -CNTs (252 ml total volume) are not in contact during the coating the

tubes are suspended in 300 ml of solvent (water of 99.9' .' pure ethanol).

4.3.3 Long MWNTs

Based on the coating model the table 4-1 is constructed. Those values are the

starting values for the Sol-Gel chemistry. In parenthesis are listed the quantities

that are eventually proved to have the best results (based on surface coverage and

number of free particles). After the final washing the -CNTs suspension (30 mg

of -CNTs in 300 ml of water) were placed in a three way 300 ml flask. The flask

was placed in an oil bath at 40C and was refluxed under constant stirring speed.

After the temperature was stabilized the pH was fixed at ~3 with 0.1N HNO3. The

precursor (Ti2S(04)3) was injected and the reaction was carried for 1 hour. The

solution was divided into six 50 ml centrifuge tubes and was washed 3 times. The

composite was then let to dry at 40C for two d-,

Acid treatment
at 140C in 10 N
HNO3 for 10 h

Figure 4-10: Schematic diagram of the process for the coating of the -CNTs.

The experiment was repeated with the Ti(OC3H9)4. For this case the nan-

otubes after the functionalization were washed with ethanol. The final solution

(30 mg of tubes and 250 ml of ethanol) placed again in a flask and refluxed at

40C under constant stirring until the temperature was stabilized at 40C. The

appropriate amount of water was added and the pH was fixed at ~3 with 0.1N

HNO3. The isopropoxide was placed in another beaker with 50 ml of ethanol and

was stirred for 10 mins. This was done to dissolve it so it will be less viscous and

Table 4-2: The calculated initial molecular ratio for the reactions regarding the
short nanotubes
Precursors f-CNTs (mg) Solvent Precursor (pl) H20 (pl)
Ti(OC2H5)4 30 mg Ethanol 300 ml 48.9 (40) 11.68
Ti(OC3H7)4 30 mg Ethanol 300 ml 69.1 (58) 11.68
Ti(OC4H9)4 30 mg Ethanol 300 ml 79.4 (62) 11.68
Ti2(S04)31 30 mg Water 300 ml 136.7 (140) N/A
1Solution of 45'. wt Ti2(SO4)3 in dilute sulfuric acid

less reactive. Then it was slowly injected into the flask to react for 30 min. The

process follows as before, triple washing and drying. The same experiment was

repeated again under nitrogen atmosphere. After the pH was fixed as previously

before nitrogen was let to flow in the container for 1 h (50 cc/min) and then the

isopropoxide solution was injected. Again the reaction was carried out for 30 min.

Then the same washing and drying steps followed. The nitrogen atmosphere did

not significantly affected the reaction results.

The TGA and XRD (figure 4-16) analysis showed that heat treatment at

500C with ramping rate 10K/min will completely transform the TiO2 to anatase.

The titanium ethoxide and titanium butoxide failed completely to achieve

coating in various conditions and therefore they were not used, although there is

a report of successfully using them to coat MWNTs [169]. The TALH was also

used, by following the method be Lee et al.[168] but the final result gave strongly

..-.-1 i., i I, ed particles.

From this part it is concluded that among all the precursors the most appro-

priate for the -CNTs is mainly the Ti2(SO4)3. The titanium isopropoxide although

it also yield good results, it was not consistent. From this point onwards as coated

-CNTs will be considered the tubes that have been coated with Ti2(SO4)3 as

precursor (f-ANTs). Figure 4-10 summarizes the coating process.

4.3.4 Short MWNTs

For the s-CNTs a table similar to the -CNTs case is constructed (table 4-2).

The synthesis procedures for every precursor are identical to the previous so are

not going to be described again. The only difference is the pH that was fixed at

approximately 4. Again the optimum conditions for the i--i l11 I iii!: were found

to be at 500C for 3 h with ramping temperature of 10 K/min (figures 4-16 and


On the contrary to the previous section and the -CNTs the precursor that

shows the best results are the metal alkoxides. There is not a standalone reason for

that, but probably it is related to the different isoelectric points. The TALH was

not used for the s-CNTs.

Among the metal alkoxides the titanium isopropoxide di-! li-t 1 the most

stable performance (consistency, repeatability) and best result (number of free

particles). The titanium ethoxide was successful but it showed high sensitivity

to the pH, with sharp transitions from coated to uncoated nanotubes. On the

contrary the isopropoxide and butoxide were more stable in regards to the pH.

Butoxide, however, has high viscosity and slower reaction rate since it has a longer

organic chain. Therefore the isopropoxide was preferred for the short nanotubes.

Overall the coating of the short nanotubes seemed to be easier and more stable,

since the surface of an individual tube was significantly smaller than the surface

of the -CNTs. Additional advantage to this was the surface charge that for the

case of the s-CNTs it was higher (greater absolute value of the zeta potential)

for the selected pH value. Both kind of MWNTs showed better dispersion in the

ethanol compared to water and since the for the organic precursors the ethanol was

preferred as solvent, in the case of the s-CNTs are expected less coated bundles.

Acid treatment
at 100C in 10 N
HNO3 for 6 h

Addition of
precursor solution:
Ti(OC3H7)4 in 50 ml

Figure 4-11: Schematic diagram of the process for the coating of the s-CNTs.

4.4 Characterization of the Composites

The very basic characterization of the composite material was done with XPS


Philips APD 3720). The XRD and TGA/DTA will determine the crystal structure

and the required time for the heat treatment. XRD will also yield information

for the grain size via the Scherrer equation. This result is important not only for

the photocatalysis, but for the interpretation of certain spectra such as XPS and

Raman. Since the XRD has low detection limit, in order to determine the i 1- I I

structure just particles were synthesized following the same process as the one

that the coating produced (figures 4-10 and 4-11). Figure 4-16 (I) shows the

-0.2 X 1.0

S\ 0.8

\ 0.6
-- 0.4


-0.4 .. . .. .. 0
0 50 100 150 200 250 300 350 400 450 500
Temperature (OC)

Figure 4-12: The TGA/TDA data of the s-ANTs. The peak at the 100C is from
the water evaporation and therefore it is accommodated by a mass
reduction. At approximately 250C the phase transition is starting
and carries on until the 500C.

results of the XRD of the coated tubes and figure 4-16 (II) shows the results of the

synthesized particles.

The TEM will confirm the coating uniformity and quality. The BET will de-

termine the specific surface area of the material (m2/g). This is critical since higher

surface area means more efficient photocatalysis. This value will be necessary for

the photocatalytic degradation tests that will be preformed on the same specific

surface area base.

In addition the XPS survey will show the composition of the material. The

detailed analysis of the peaks will be discussed in a separate chapter.

Figure 4-13: TEM images of the coated s-CNTs. (a) The coating is approximately
6 nm thick with very intense variation. (b) There are cases that there
are big particles nucleated on the surface of the nanotubes. That is in
agreement with the BET results that showed specific surface area of
183 m2/g.

4.4.1 Short ANTs: TEM, XPS, BET

The TEM of the short nanotubes revealed a coating with large variation in the

thickness ranging for 3 to 10 nm with average value of 6 nm (figure 4-13 (a)). It

also observed that there were spots that the coating was not complete and there

were uncoated regions on the surface of the nanotubes. Figure 4-17 shows the XPS

results that confirm the titania coating. In addition the XRD (figure 4-16(I), line

(b)) confirms the anatase phase, while there is no indication the of rutile phase.

The Scherrer's formula will be used to estimate the grain size [171]

drain c (4-12)
B cos 0

where 0 is the Bragg' s angle, A the wavelength (1.54 A), K is a constant (K

2 (I)2 = 0.93) [171-173], B is the half value breadth of the most intense peak.
The grain size according to this calculation is 5 nm (figure 4-16, line (b)). This is

in agreement with the TEM results. The same calculations for the (figure 4-16(11),

Figure 4-14: TEM images of the coated -CNTs. (a) The TEM images reveal
coating of approximately 4 nm and it is uniform. (b) There are cases
that there are big particles nucleated on the surface of the nanotubes.
That is in agreement with the BET results that showed specific
surface area of 172 m2/g.

line (b)) showed 53 nm average grain size. Although they were produced under the

same conditions they have different grain sizes that attributed to the presence of

the nanotubes.

Finally the BET revealed a surface are of 183 m2/g. The high surface area

is due to the needle like shape of the nanotubes and the rough surface that the

Sol-Gel chemistry generated.

4.4.2 Long ANTs: TEM, XPS, BET

The TEM showed a very uniform coating 4-14(a) of approximately 4 nm

thick. In contrast with the s-ANTs the coating is very uniform and has very small

variance (3 to 5 nm). Again there were cases of partially coated tubes, but less

compared to the s-ANTs. Again the XPS confirmed the elements of Ti, O and C.

Based on the elemental concentrations the amount of titania is about 1"' This

value however is not considered accurate since the XPS is very sensitive to the

thickness of the 1 Vr -. The electrons depending on their energy can travel only

o Ag Fe
A Al A Ge
\ o Au Mo
v Be v Ni
,,o C Se

0 *
0 '

A 0 0
0 0V O. < -

I, I I i II ,[ I I ,I I, I I
5 10 20 50 100 200 500 1000 2000
Kinetic energy (eV)

The universal curve of the electrons, based on the calculations by M.
P. Seah and W. A. Dench [174]. The curve shows the mean free path
of the electrons as function of the kinetic energy (dashed lines). There
are also experimental results that follow the same trend. The mean
free path does not depend on the material. For Mg source the X-Ray
energy is 1253.6 eV, which give a mean free path of approximately 10


Figure 4-15:

a certain distance in the material, regardless what the material is (figure 4-15).

The detected electrons are coming for only the few top nm [174]. The s-ANTs

have thicker coating and therefore the elemental analysis is not representative

composition. The XRD confirmed the anatase (figure 4-16(11), line (a)). According

to the Scherrer formula (equation 4-12) the grain size is 5 nm (figure 4-16(1), line

(a)). This is slightly contrasting the TEM result that was 4 nm. This is attributed

to the fact that the signal of the carbon nanotubes overpowered the signal of

titania and therefore the calculation is not considered exact but just a rough

estimate. The grain size that was calculated based on the XRD pattern from figure

4-16(11) (just the synthesized particles), line (a) is 23 nm.

I 1 111111 1 I 1 111111 1 I 1 111111 1


Finally the BET gave a surface area of 172 m2/g. This is in agreement with

the expectations based on the TEM images and the respective result for the s-

ANTs. The lower value of the surface area is attributed to the smoother surface

that the Ti2(SO4)3 yielded. In case of the f-ANTs there are less free particles as

result of the coating process (figure 4-10).




(111) (211)
(004) (105) (220)
(103) 120) (210)

30 35 40 45 50
20 (degrees)


20 25 30 35 40 45 50
20 (degrees)


Figure 4-16:

(204) (301)
(002) (112) (107:
13 (220) (116) (220)(311)
(213)1 (221) 1 .1 (320

55 60 65 70

55 60 65 70 75

XRD patterns with and without the coating. (I) XRD patterns of the
nanotubes with the coating. (II) The XRD pattern of the particles
prepared by the same Sol-Gel method as the coating on the
nanotubes. (a) Titanium sulfate (e-ANTs) and (b) titanium
isopropoxide (s-ANTs). The solid lines denote the peaks for anatase
(black line) and rutile (light gray) with the relative intensities.


Ev/step:0.5 eV, Time/step: 30 ms, Sweeps: 10
Source: Mg, Pass Energy:89.45 eV, Work Function: 4.36 eV



O Is 52.0%

Ti 2s

Ti 2p 16.7%

C Is 31.2%

Ti 3p


Figure 4-17:

900 800 700 600 500 400
Binding Energy (eV)

XPS survey for the s-ANTs. There is a significant amount of TiO2 (16.7' Ti). There is no direct stoichiometry
with the oxygen (5'"-. 0) since the oxygen depends on the exposed i i -I i11. .graphic orientation.

Ev/step:0.5 eV, Time/step: 30 ms, Sweeps: 10
Source: Mg, Pass Energy:89.45 eV, Work Function: 4.36 eV

C Is 91.0%



Figure 4-18:



O Is 5.8%

Ti 2s

Ti 2p 1.2%

900 800 700 600 500 400 300 200
Binding Energy (eV)

XPS survey for the e-ANTs. There is a significant amount of TiO2 (1.-"'
with the oxygen (r. -'- O). There is less TiO2 compared to the s-ANTs.

STi). Again th(

Si 2p 2.0%


ire is no stoichiometry


This chapter describes the series of experiments that were performed to

evaluate the photocatalytic efficiency of the synthesized particles. The method

used for this purpose is dye degradation, where a dye is being photocatalyitcally

degraded and its concentration is being monitored as function of time [175-179].

This technique was selected over the biocidal tests since it is fast, accurate and

depends primarily on the type and properties of particles and not on particle

interactions. Other methods that could have been used, such as spore or bacteria

inactivation, have many, not fully controlled, variables that can alter the results

[180, 181].

In the case of the biocidal test the length of the particles is comparable to the

diameter of the target bacteria or spores. This will affect the kinetics of the system

and the interaction between the particles and the bacteria by inducing steric forces

and occasionally electrostatic effects. In addition, the temperature and the pH

that can vary significantly during the experiments can dramatically affect the

behavior of the spores or bacteria. Especially for the spores, temperature increase

will trigger germination that will transform them into bacteria, making them more

vulnerable to the photocatalytic destruction. Biocidal tests are also time consuming

and require a highly specialized lab. So although the particle has been designed

primarily for biological applications, the biocidal tests are not an accurate way

to measure and compare the properties. Thus the dye degradation test was used

as a quick way to validate the photocatalytic properties of the particles, which

Figure 5-1: Schematic diagram showing the basic elements of the photocatalytic
degradation chamber.

are directly related to the structure and the electronic properties of the different


In the following sections, the experimental setup is described, followed by the

theory of the dye degradation and the parameters that can influence the results.

Subsequently the experimental results and finally some general conclusions are


5.1 Experimental Setup, Materials and Procedures

5.1.1 Experimental Setup

Figure 5 1 shows a sketch of the experimental setup (photocatalytic reaction

chamber). The whole structure consists of a light-insulating chamber where the

interior is black to absorb any scattered radiation. At the top of the chamber

is a 5W fan to maintain the temperature below 30C. Inside the chamber are a

magnetic stirrer of variable speed and four UV lamps arranged over the stirrer

(figure 5-1). Depending on the test different lamps have been used:

UV 350 nm four fluorescence lamps of 350 nm peak wavelength and 8W power

each that in the current configuration gave 20 W/m2

UV 305 nm four fluorescence lamps of 350 nm peak wavelength and 8W power

each that in the current configuration gave 20 W/m2

Visible light two halogen lamps of light radiation and 100W power each that in

the current configuration gave 50 W/m2 that have built-in UV filter.

For all the different lamps the intensity was monitored as function of time.

It was found that the intensity increases with time for the first 30 min. After

this time has elapsed the intensity is stabilized at the power output given above.

Thus the lamps are alv--, given a head start of minimum 30 min before the

experiment starts. Under those conditions a test with water demonstrated that

the temperature is maintained almost constant at approximately 250C with 1 to 2

degrees variation in one hour. Temperature is also a factor that can influence the

results, but not in a significant manner.

5.1.2 Dye Selection

In the literature there are many types of dyes used for this application. For

the present experiments the Brilliant Procion Red MX-5B (C19H13C12N6Na2O7S2)

was used [176, 182]. The color of the dye is magenta and absorbs strongly in the

510-540 nm (Figure 5-3). Figure 5-2 shows the molecular structure of the dye.

The presence of the three benzene and one s-triazine rings makes the dye more

resistant to degradation compared to other dyes with fewer rings, even for low

concentrations [183]. This is very critical since fast degradation means that the

system will not be fully stabilized (pH, temperature) before the degradation is over.

Very slow degradation however will give sufficient time for water evaporation that

will alter the dye concentration. An additional advantage is the existence of both

negatively (SO42) and positively (Na+, NH+) charged chemical groups that will

induced adsorption on positively and negatively sites respectively.

Figure 5-2: Three-dimensional structure of the Brilliant Procion Red MX-5
molecule. As it can be seen it contains 3 benzene groups and a benzene
group with three carbon atoms replaced by nitrogen atoms (s-triazine).

Brilliant Procion Red MX-5B is one of the dyes that has been extensively

studied and the degradation byproducts are known [176, 185]. However in this

research there is no need to study the dye in this extend since all the necessary

information is available from the literature [176]. Table 5-1 shows the different

intermediates of the reaction in the order they appear in the solution during

degradation. The photocatalytic reaction proceeds in three steps. In the first step

the most active bonds are hydroxylated. Those bonds include the C-N bond linked

to the benzene ring or the naphthalene ring and the C-S bond of sulfonate group

linked to the naphthalene ring or the benzene ring, to form organic acids with or

without hydroxyl groups and the related ions (SOc- and NHZ). In the second step,

the groups linked to the triazine ring are replaced by hydroxyl to yield cyanuric

acid, as in the case of the s-triazines herbicides, and the related ions (SO3, Cl-).

At the same time the aromatic acids produced from the first step subsequently

hydroxylated and led to the cleavage of aromatic rings to from aliphatic groups.

400 450 500 550 600
Wavelength (nm)

Figure 5-3: The absorption spectrum for a 5 ppm solution of the Procion Red
MX-5B dye.

The third step involves a further oxidation of the aliphatic acids to produce CO2

and water. Those steps are summarized in table 5-1 and figure 5-4 represents a

visualization of the degradation.

5.1.3 Experimental Procedure

Initially a mixture of dye solution and the particles that are being evaluated

are sonicated for 20 mins. Following the sonication the particles are placed in

the dark chamber (figure 5-1). While the solution is exposed to UV light, three

samples are obtained every certain time intervals, in 1.5 ml cuvettes. The cuvettes

were left for 2 d- i,- for the particles to settle. The dye concentration was measured

via UV-VIS spectroscopy and the reaction constant was estimated based on the

Langmuir-Hinshelwood theory. Since the particles tested here are nanosized, even

after 2 d- i,- there will still be suspended particles. Those particles can scatter or

-hydroxy-propanedioic acid

p-Hydroxy-cinnamic acid

-tricarboxylic acid

Malic acid

-benzeneacetic acid

carboxylic acid

Propanedioic acid

Butenedioic acid

2-Hydroxy-benzoic acid

cyanuric acid

Propanoic acid

Oxalic acid

Figure 5-4: The structure of several intermediate products of the photocatalytic
reaction that show the destruction of the bonds and the size reduction
of the molecules.


Table 5-1: The oxidation intermediates and their structure to be compared to the
initial dye structure in figure 5-2. Adapted from reference [184].

Step Photo-oxidation intermediates
p-Hydroxy-phenyl-3-hydroxy-propanedioic acid
3-Hydroxy-benzeneacetic acid
Step-1 2-Hydroxy-benzoic acid
p-Hydroxy-cinnamic acid
1,2-Benzenedicarboxylic acid
C i-,- i uri.- acid
1-Propene-1,2,3-tricarboxylic acid
Propanedioic acid
SPropanoic acid
Malic acid
Butenedioic acid
Oxalic acid
Acetic acid
Aliphatic compounds to CO2 and H20
tep minerals (S, Na)

absorb the light, which will alter the obtained spectrum. So for every experiment

a water solution with particle concentration equal to the ongoing experiment is

prepared. This solution is also left for 2 d i- and the obtained spectrum is used as


5.2 Theory for the Photocatalytic Degradation of Dyes

Most experimental results agree that the rate of photocatalytic oxidation of

dyes can be approximated with the Langmuir-Hinshelwood (L-H) model [175

178, 180-185]. The model assumes that the rate will depend on the adsorption

of the dye molecule on the TiO2 particle and the oxidation reaction. So if it is

assumed that k, is the reaction constant and K the adsorption constant then

according to the L-H kinetics model the oxidation rate is:

dC kKC
r dt K (5-1)
dt 1 +KC

0.6 \ --0.004
u Q / n

0.4 --0.006

0.2 --0.008

0.0 -1 -0.010
0 20 40 60 80 100 120 140
Time (min)
1.0 0..

0.8 --0.02

0.6 / --0.04

0.4 -0.06

0.2 // --0.08

0.0 0.1
0 5 10 15 20 25 30 35 40
Time (min)

Figure 5-5: Comparison between the numerical solution of the
Langmuir-Hinshelwood (equation 5-1) and the approximation. The
red lines represent the approximation and the black is the numerical
solution. The solid line represents the dye concentration while the
dashed represents reaction rate (d Figure (a) is for large
concentrations (k=0.1, K 1=l, Co-10) and figure (b) is for small
concentrations (k=0.1, K=, Co=0.1).

In equation 5-1 k and r is mg/ mnin and K is in 1/mg where C is the dye concen-

tration in mg/I. This model is non linear but it can be further simplified:

In + ) k (52)
KCo \Co Co Co

where C0 is the initial dye concentration. With the assumption that Co -- 0, then

IKCn (C-) > -- 1) and equation 5-2 simplifies to:

In = -Kkt (5-3)

which yields a simple exponential decay:

C (t) Co-kKt (5-4)

C (t) = Coe-kpp. (5-5)

C (t) Coe-- (5-6)

Figure 5-5 shows a comparison of the approach for two different dye concentra-

tions. It is apparent that in the case of the low concentration (figure 5-5(a)) the

agreement between the exponential approach and the exact numerical solution is

very good, while for the case of the high concentration the difference is significant.

It has to be underlined that in the Langmuir-Hinshelwood model is assumed for

single reaction (AB # A + B), which is not true for the case of the dye degra-

dation. As described before for this certain dye there are a lot more reactions

involved during the degradation. In this case it is just assumed that the k refers to

the slowest reaction.

5.3 Parameters that Influence the Photocatalytic Reaction

There are many parameters that can affect the reaction rate. The major pa-

rameters are the pH, the initial dye concentration, the solids loading and radiation

intensity. There also other parameters such stirring speed and temperature with

minor effect at the reaction rate.

5.3.1 pH

The pH is one of the most important parameters that influence photocatalytic

reactions. The pH can impact both the particles stabilization and the actual

reaction [180, 181]. Depending on the isoelectric point the particles will induce

coagulation that will significantly reduce the surface area of the particles. For

titania the isoelectric point ranges from 5 to 7. Therefore for pH values between

5.0 and 7.0 the photocatalytic reaction rate will be reduced. For pH values >7 and

< 5 the colloidal stability is optimum. In addition the surface charge impacts the

way the dye adsorbs on the titania particles. This is especially important for the

case of azo dyes, such as the one used here, since the have many polar groups. The

charged molecules (positively charged S and Na atoms) can be adsorbed well on the

surface with negative charge (in the case of titania means pH>7).

The pH can directly affect the reaction. A high pH will increase the amount of

OH-, and vise versa. In this reaction there are three steps with multiple reactions

within each step. Slight variations of the pH can have a significant impact on some

of the reactions that will immediately effect the overall reaction. It is obvious

that there is not a specific trend for the pH, since it depends on the dye and its

byproducts. So et al. however have investigated the pH effect of the Procion Red

MX-5B, and the results are in figure 5-6(a) [186]. There is approximately a 4(0-.

variation at the reaction rate when the pH increases from 2 to 10.

5.3.2 Initial Dye Concentration

As it was already discussed smaller concentrations are more suitable for the

first order decay since it approaches more the simple exponential. However there

is a more physical dependence of the reaction rate to the dye concentration. While

the initial dye concentration increases it will increase the probability of a dye


Figure 5-6:

4 6 8 10 0 10 20 30 40
pH Co (ppm)
(a) (b)

1/2 100 nm

-1 pm

10 20 30 40 0 0.1 0.2 0.3 0
Light Intensity (W/m2) 4 (wt%)
(c) (d)

The main parameters that influence the oxidation rate. (a) pH
variation, obtained from reference [180], for the Brilliant Procion Red
MX-5B (b) as function of the initial dye concentration (c) as function
of the light intensity (d) as function of the surface area (data
calculated for Degussa P25).

molecule adsorbing on the surface and consequently leading to photocatalytic

degradation. Thus the reaction rate will increase. However, if the dye concentra-

tion increases further the solution will become darker resulting UV shielding and

therefore the rate will decrease. The increasing of the dye concentration, will also

increase the amount of adsorbed dye molecules on the surface of the particles,

which will reduce the available OH- sites and therefore reduce the [OH*] gener-

ation. So initially the reaction rate is increasing (figure 5-6(b)) almost linearly,

until it reaches a maximum and afterwards it decreasing almost exponentially. The

graph in figure 5-6(b) has been derived both with theoretical and experimental

data. The observed maximum, for the dye currently used is about 5 ppm. An addi-

tional advantage for using this concentration is that, as seen from the graph, small

variations (52 ppm) around this value do not have any impact on the reaction

rate Co C- max o).

5.3.3 Intensity of the Radiation

The light intensity is another parameter that can affect the reaction. It

is expected that low intensities (0 to 20 W/m2) will excite fewer electrons and

therefore the overall reaction rate will be low. While increasing the light intensity

the reaction rate will increase, till it reaches a maximum value and level out. The

way the light intensity influences the reaction rate cannot be derived directly from

first principles, but Ollis et al. [187] after reviewing several studies concluded that

three distinct regions can be delineated (figure 5-6(c)). (i) For low light intensities

the reaction rate increases proportionally to the light intensity (oc I). (ii) At

intermediate light intensities and beyond a certain value (approximately 20 W/m2)

the rate intensity is proportional to the square root of the light intensity (oc VI)

and (iii) at higher intensities the light intensity does not have an impact on the

reaction rate.

5.3.4 Solids Loading/Surface Area

Many researchers have reported the effect of the solids loading on the pho-

tocatalytic efficiency [187-189]. It is, however, more valid hypothesis to assume

that the reaction constant depends on the available surface area and not the solids

loading. Generally increasing the number of particles (and consequently the avail-

able surface) the sites for adsorption and OH* generation will also increase and

therefore the overall reaction rate will increase. At higher solids 1.. "li.- however,

there are other factors that come into pl iv, such as more rapid coagulation of the

particles and UV light shielding, that will eventually impede the reaction rate, until

it reaches a plateau [188, 189].

0 10 20 30 40 50
Time (min)

Figure 5-7: The pH variation during the dye degradation. The initial value
between the ANTs and Degussa P25 since the specific surface area is
different. In the first case the pH is stabilized after 10 min while in the
second case that occurs after 20 min. In both cases the stable pH value
is lower than the initial.

The solids loading ) is correlated to the surface area per solution volume )s

with the equation:

Os [-] [ ](5- 7)
S100ml pR 100PlR
The relation between Os and 0 is linear, but Os is also inversely proportional to

the particle radius R. So for the same solids loading the particle radius has a

tremendous impact on the reaction rate (figure 5-6(d)). So to avoid variations due

to surface area changes the experiments will be conducted on the same surface area

basis unless it is otherwise stated.

S/I I I \ -- Omin

524 (nm)


400 450 500 550 600 650
Wavelength (nm)

Figure 5 8: The dye spectrum during the different time intervals. The three dashed
lines (513, 524 and 537 nm) are the three wavelengths that were used
for the C/Co calculation. The data were obtained from a sample of 3
mg Degussa P25 in a 50 ml of 5 ppm dye solution.

5.4 Experiments

For all the experiments the parameters discussed above (pH, initial dye

concentration, radiation intensity and solids loading) were either kept constant or

monitored to ensure the accuracy of the result. The dye concentration was al-bi-,

kept at 5 ppm, the light intensity of the UV lamps was 20 W/m2 (50 W/m2 for the

visible radiation) and the pH was monitored during the experiments. Figure 5 7

shows the pH variation during the photocatalytic degradation. The stabilization

occurred, relatively fast, in 20 min for Degussa P25 and 10 min for the ANTs (both

short and long). The maximum difference between the reaction rates, due to the

different pH value (4.58 versus 5.64) will be only in the order of 10'-. Other minor

parameters such as temperature and stirring speed were assumed insignificant

Full Text




Copyright2006 by GeorgiosPyrgiotakis




ACKNOWLEDGMENTS Therearemanypersonsthatwithouttheircriticalandinrue ntialsupportand guidancethisworkwouldhaveneverbeenaccomplished. IwouldliketorstandforemostthankDr.WolfgangSigmundw hosework ethic,compassion,support,understandingandguidancehe lpedmethroughthis project.Iwouldalsoliketothankmycommitteemembers,Drs .Milz,Norton, SinnottandKoopmanfortheirconstructivecomments.Verys pecialthanksgoto Dr.Koopmanwhoverycloselyobservedthewholeprojectandw hosesuggestions werealwaysinruential.AlsoIwouldliketothankDr.Moudgi lwhoalways challengedmetodiscovernewpathwaysinscience.Iwouldal soliketothank Dr.Rinzlerforallhishelpregardingthenanotubes.Iwould liketorecognize thehelpofthestaofMAIC(MaterialsAnalyticalInstrumen tCenter)regarding thecharacterizationandthehelpofMariaPalazeulosregar dingtheRaman Spectroscopy. TherearealsoalotofstudentswhowithouttheirhelpIwould nothave nishedthiswork.IthankVijayKrishnaandJueZaofortheex tensivediscussions abouttheproblemsweencounteredandallthepeopleintheSi gmundgroup, especiallyDrs.S.-W.Lee,J.-M.ChoandS.-H.Lee.Averywar mthankgoesto mydearfriendsAmit,Junhan,IsaacandVasana,fortheirsup portandhelpduring mywork.AlsoIwouldliketoacknowledgeallthepastandcurr entmembersinthe groupforassistingmeinmanywaysduringmywork. FinallyIwouldliketoacknowledgemyparentsfortheirsupp ortthroughall theroughmomentsofmylifeintheUSA.Specialthankstomysi sterforcheering meupallthetime.Andlastbutnotleast,Itankmyfriendsall overtheworld iv


(China,Germany,Cyprus,Greece,India,Japan,Korea,Taiw an,Turkey,UKand USA)whoconstantlyshowedmeloveandsupport.Withoutthem Iwouldhave neveraccomplishedthiswork.FinallyIwouldliketothanka llthepeoplethat workedtowardsthediscoveryandperfectionofcoee,myult imatesupportthrough mydoctoral. v


TABLEOFCONTENTS page ACKNOWLEDGMENTS .............................iv LISTOFTABLES .................................ix LISTOFFIGURES ................................xi ABSTRACT ....................................xviii 1INTRODUCTION ..............................1 1.1PhotocatalysisandTitania ......................2 1.2CarbonNanotubes ..........................2 1.3Objectives ...............................3 2PHOTOCATALYSISONTiO 2 (TITANIA)SURFACEPRINCIPLES ANDAPPLICATIONS ..........................5 2.1StructureofTitania ..........................6 2.1.1Anatase ............................6 2.1.2Rutile ..............................7 2.2ElectronicPropertiesofTitania ...................8 2.2.1Anatase ............................9 2.2.2Rutile ..............................10 2.3SemiconductorPhotocatalysis ....................10 2.3.1BasicPrinciples ........................12 2.3.2EnhancementofPhotocatalysis ................13 2.4ApplicationsofPhotocatalysis ....................20 2.4.1EnvironmentalApplications .................20 2.4.2PhotovoltaicCell .......................21 3CARBONNANOTUBES(CNTs):STRUCTUREANDELECTRICAL PROPERTIES ...............................23 3.1Bonding,StructureandPhysicsofSingle-WallCarbonNa notubes 24 3.1.1BondinginCarbonMaterials .................24 3.1.2StructureandNotation ....................25 3.1.3SymmetriesandVibrationalFrequencies ..........27 3.2ElectronicPropertiesofSWNTandMWNT ............29 3.2.1ElectronicPropertiesofSWNT ................29 3.2.2ElectronicpropertiesofMWNT ...............34 vi


3.3CarbonNanotubesGrowthMechanisms ...............34 3.3.1ArcDischarge .........................35 3.3.2CVD:ThermalCVD,PE-CVD ................35 4ANATASECOATEDCARBONNANOTUBES(ANTs):SYNTHESIS ANDCHARACTERIZATION) ......................37 4.1DesignParameters ..........................37 4.2NanotubeSelectionandPreparation .................39 4.2.1MaterialsSelection ......................40 4.2.2PuricationandDispersion ..................40 4.2.3CharacterizationoftheFunctionalizedMWNTs ......41 4.3 Sol-Gel Coating ............................50 4.3.1PrecursorSelection ......................51 4.3.2CoatingModel .........................54 4.3.3LongMWNTs .........................55 4.3.4ShortMWNTs .........................58 4.4CharacterizationoftheComposites .................59 4.4.1ShortANTs:TEM,XPS,BET ................61 4.4.2LongANTs:TEM,XPS,BET ................62 5PHOTOCATALYTICEVALUATIONOFTHESYNTHESIZEDPARTICLESWITHDYEDEGRADATIONTESTS .............68 5.1ExperimentalSetup,MaterialsandProcedures ...........69 5.1.1ExperimentalSetup ......................69 5.1.2DyeSelection .........................70 5.1.3ExperimentalProcedure ...................72 5.2TheoryforthePhotocatalyticDegradationofDyes ........74 5.3ParametersthatInruencethePhotocatalyticReaction ......76 5.3.1pH ...............................77 5.3.2InitialDyeConcentration ...................77 5.3.3IntensityoftheRadiation ...................79 5.3.4SolidsLoading/SurfaceArea .................79 5.4Experiments ..............................81 5.4.1TitaniaNanoparticlesandCarbonNanotubes ........83 5.4.2LongANTs:PhotocatalysisunderUVLight ........87 5.4.3LongANTs:PhotocatalysisunderVisibleLight ......88 5.4.4LongANTs:PostUVActivity,PhotocatalysisinDark ..92 5.4.5ShortNanotubes:PhotocatalysisunderUV .........92 5.5Conclusion ...............................94 6SPECTROSCOPICTECHNIQUESTOEXPLAINTHEPHOTOCATALYTICEFFICIENCYOFTHEANTs. .................96 6.1RamanSpectroscopyoftheCarbonNanotubes ...........97 vii


6.1.1GeneralTheoryofRamanSpectroscopyofCarbonNanotubes ............................98 6.1.2BasicRamanLinesforCarbonNanotubes .........100 6.2RamanSpectroscopyoftheAnataseStructureofTiO 2 ......104 6.3ExperimentalProcedures .......................106 6.3.1SamplePreparation ......................106 6.3.2MathematicalAnalysisandManipulation ..........107 6.4ExperimentalResults .........................111 6.4.1LongNanotubesaftertheAcidTreatment .........112 6.4.2ShortNanotubesaftertheAcidTreatment .........113 6.4.3LongNanotubesaftertheCoating ..............117 6.4.4ShortNanotubesaftertheCoating ..............123 6.4.5SummaryoftheRamanSpectraAnalysis ..........130 6.5X-RayPhotoelectronSpectroscopy(XPS) .............131 6.5.1Instrument,SamplePreparationandMathematicalAna lysis 132 6.5.2XPSoftheReferenceMaterial ................133 6.5.3XPSofthe s -ANTs ......................137 6.5.4XPSofthe ` -ANTs ......................143 6.6SummaryoftheXPSAnalysis ....................147 7CONCLUSIONSANDFUTUREWORK ..................157 7.1Conclusions ..............................160 7.2FutureWork ..............................160 APPENDICESAMATHEMATICAALGORITHMUSEDFORTHELOESSMETHOD .162 BRAMANPEAKSOFCNTs .........................165 REFERENCES ...................................166 BIOGRAPHICALSKETCH ............................188 viii


LISTOFTABLES Table page 4{1Thecalculatedinitialmolecularratioforthereaction sforthe ` -CNTs 55 4{2Thecalculatedinitialmolecularratioforthereaction sregardingthe shortnanotubes .............................57 5{1Theoxidationintermediatesandtheirstructuretobeco mparedto theinitialdyestructureingure5{2. .................74 5{2Summaryoftheexperimentsperformed .................84 5{3Summaryoftheexperimentalresultsofthischapter. ..........94 6{1TheRamanfrequenciesfroanataseandrutilephaseoftit ania.The brookiteisnotincludedheresinceisnotapresentformofTi O 2 andithasintotal36weakpeaks.Thenotationinparenthesis is representingtherelativeintensityofthepeaks;w:weak;m :medium; s:strong;vs:verystrong.Dataareadaptedfromreferencem aterialandreference ............................105 6{2TherawttingparameterscalculatedwiththeLevenberg -Marquardt algorithmfortheacidtreated ` -CNTs.Thegraphicrepresentation oftheresultsisingure6{6.Thetyielded 2 =7.1333 10 4 .For convenienceatthedatarepresentationweusethesymbol a (2)L insteadofthatisusedinequation6 20. ...............113 6{3TherawttingparameterscalculatedwiththeLevenberg -Marquardt algorithmfortheacidtreated s -CNTs.Thegraphicrepresentation oftheresultsisingure6{7.Thetyielded 2 =3.9138 10 7 ...115 6{4TherawttingparameterscalculatedwiththeLevenberg -Marquardt algorithmforthetitaniacoated ` -CNTsandthetitaniasegmentof thespectrum.Thegraphicrepresentationoftheresultsisi ngure 6{9.Thetyielded 2 =8 : 3378 10 4 ................117 6{5TherawttingparameterscalculatedwiththeLevenberg -Marquardt algorithmforthetitaniacoated ` -CNTs.Thegraphicrepresentationoftheresultsisingure6{6.Thetyielded 2 =8 : 3378 10 4 Forconvenienceatthedatarepresentationweusethesymbol a (2)L insteadofthatisusedinequation6 20. ..............121 ix


6{6TherawttingparameterscalculatedwiththeLevenberg -Marquardt algorithmfortheacidtreated ` -CNTs.Thegraphicrepresentation oftheresultsisingure6{12.Thetyielded 2 =1.9924 10 8 ...125 6{7TherawttingparameterscalculatedwiththeLevenberg -Marquardt algorithmforthecoated s -CNTs.Thegraphicrepresentationofthe resultsisingure6{12.Thetyielded 2 =1.0956 10 5 ......127 6{8SummaryoftheRamanresult.Herearelistedthemajorpea ksand shiftbothfortitaniaandCNTsafterthecoating. ..........130 6{9SummaryoftheXPSpeaks ........................147 7{1Electronanityandworkfunctionformetalsusedtocrea terectifyingcontactwithtitaniainordertoincreasethephotocatal yticeciency. ..................................158 B{1PropertiesofthevariousRamanfeaturesingraphiteand SWNTs. ..165 x


LISTOFFIGURES Figure page 2{1Thetwobasictitaniastructures. .....................7 2{2Theelectronicbandstructureofthetwomainphasesofti tania. ...8 2{3Schematicdiagramrepresentingthemainphotocatalyst swiththeir bandgapenergy.Inordertophoto-reduceachemicalspecies ,the conductancebandofthesemiconductormustbemorenegative than thereductionpotentialofthechemicalspecies;tophoto-o xidizea chemicalspecies,thepotentialofthevalencebandhastobe more positivethantheoxidationpotentialofthechemicalspeci es.The energiesareshownforpH0. ......................11 2{4Schematicrepresentationofthereactionstakingplace intitania. 1OLightstrikesthesemiconductor. 2OAnelectron-holepairisformed. 3OElectronsandholesaremigratingtothesurface. 4OTheholes initiateoxidationleadingtoCO 2 ,Cl H + ,H 2 O. 5OTheconduction bandelectronsinitiatereductionreactions. 6Oelectronandholes recombinationtoheatorlight. ....................13 2{5Titaniabandstructure(a)beforeand(b)afterdoping.T hetransitionmetalsareinterstitialorsubstitutionaldefectsint hestructure oftitaniaandgeneratetrappinglevelsinthebandgap. .......14 2{6Theprinciplesofrectifyingcontactbetweentitania(E g =3.2eV)and ametalwithworkfunction( m ),inthisexample5eV,greaterthan theanity( s )oftitania. .......................16 2{7Theprinciplesofrectifyingcontactbetweenanatase( )titania(E g =3.2 eV)andandrutile( r )titania(E rg =3.0eV). .............18 3{1The2Dgraphenesheetsisshownwiththe a 1 and a 2 speciesthe chiralityofthenanotube.Thechiralvector, C h ,isthe OA ,while thetranslationvector T isthe OB .Also istherotationangle and thetranslation.Thosetwoareconstitutethesymmetryoperation R =( j ). ...........................25 xi


3{2Thegraphenesheetisshownwiththe( n;m )pairwhichspeciesthe chiralnanotube.Thepairofinteger( n;m )inthegurespecies thechiralvector C h forcarbonnanotubes,includingzigzag,armchairandchiraltubules.Beloweachpairofintegerisliste dthe numberofdistinctcapsthatcanbejoinedcontinuouslytoth ecylindricalcarbontubuledenotedby( m;n )[ref].Itisalsodenotedthe conductionstateofeverychirality. ..................28 3{3Thedispersionforgraphiteascalculatedfromequation 3 10. ....30 3{4Thedispersionenergiesfortwodierentchilarities. ..........32 4{1SEMpicturesofthetwotypesofnanotubes. ..............42 4{2TEMimagesofthe s -CNTs. .......................43 4{3TEMimagesofthe ` -CNTs. .......................44 4{4Immediatecomparisonofthetwodierentkindsofnanotu bes. ...45 4{5Thezetapotentialforboththe ` -CNTs(a)and s -CNTs(b).Itshows theshiftoftheIEPforthe ` -CNTs(from7to3.5)andtheincrease atthesurfacechargeforthe s -CNTs(from-10mVto-37mvfor ph4). ..................................46 4{6TheFTIRoftheMWNTsaftertheacidtreatment(onlythe s -CNTs resultsaredisplayed).Thebandsthathavebeenidentiedp rove thereactionofthe COOHonthesurfaceofthenanotubes. ....47 4{7Thedierentialvolumeandnumberofthe s -CNTsbeforeandafter theacidtreatment. ...........................48 4{8TheTGA/TDAdataofthe s -CNTs.Thepeakatthe600‰indicatestheburningtemperatureoftheCNTs.Itisobservedabo ut 6%oftheinitialmassresidue,whichistheFecatalyst. .......49 4{9Thedierent Sol-Gel precursorsusedinthisresearch. .........53 4{10Schematicdiagramoftheprocessforthecoatingofthe ` -CNTs. ...56 4{11Schematicdiagramoftheprocessforthecoatingofthe s -CNTs. ...59 4{12TheTGA/TDAdataofthe s -ANTs.Thepeakatthe100‰isfrom thewaterevaporationandthereforeitisaccommodatedbyam ass reduction.Atapproximately250‰thephasetransitionisstarting andcarriesonuntilthe500‰. .....................60 4{13TEMimagesofthecoated s -CNTs. ...................61 4{14TEMimagesofthecoated ` -CNTs. ...................62 xii


4{15Theuniversalcurveoftheelectrons,basedonthecalcu lationsbyM. P.SeahandW.A.Dench.Thecurveshowsthemeanfreepathoftheelectronsasfunctionofthekineticenergy(dashedli nes). Therearealsoexperimentalresultsthatfollowthesametre nd.The meanfreepathdoesnotdependonthematerial.ForMgsourcetheX-Rayenergyis1253.6eV,whichgiveameanfreepathofap proximately10 A. ...........................63 4{16XRDpatternswithandwithoutthecoating. ..............65 4{17XPSsurveyforthe s -ANTs.ThereisasignicantamountofTiO 2 (16.7%Ti).Thereisnodirectstoichiometrywiththeoxygen (52% O)sincetheoxygendependsontheexposedcrystallographic orientation. ..................................66 4{18XPSsurveyforthe ` -ANTs.ThereisasignicantamountofTiO 2 (1.2%Ti).Againthereisnostoichiometrywiththeoxygen(5 .8% O).ThereislessTiO 2 comparedtothe s -ANTs. ...........67 5{1Schematicdiagramshowingthebasicelementsofthephot ocatalytic degradationchamber. .........................69 5{2Three-dimensionalstructureoftheBrilliantProcionR edMX-5molecule. Asitcanbeseenitcontains3benzenegroupsandabenzenegro up withthreecarbonatomsreplacedbynitrogenatoms(s-triaz ine). ..71 5{3Theabsorptionspectrumfora5ppmsolutionoftheProcio nRed MX-5Bdye. ...............................72 5{4Thestructureofseveralintermediateproductsoftheph otocatalytic reactionthatshowthedestructionofthebondsandthesizer eductionofthemolecules. ..........................73 5{5ComparisonbetweenthenumericalsolutionoftheLangmu ir-Hinshelwood (equation5 1)andtheapproximation.Theredlinesrepresentthe approximationandtheblackisthenumericalsolution.Thes olid linerepresentsthedyeconcentrationwhilethedashedrepr esents reactionrate d dt C C 0 .........................75 5{6Themainparametersthatinruencetheoxidationrate. ........78 5{7ThepHvariationduringthedyedegradation.Theinitial valuebetweentheANTsandDegussaP25sincethespecicsurfacearea is dierent.IntherstcasethepHisstabilizedafter10minwh ilein thesecondcasethatoccursafter20min.Inbothcasesthesta ble pHvalueislowerthantheinitial. ...................80 xiii


5{8Thedyespectrumduringthedierenttimeintervals.The threedashed lines(513,524and537nm)arethethreewavelengthsthatwer e usedforthe C=C 0 calculation.Thedatawereobtainedfromasampleof3mgDegussaP25ina50mlof5ppmdyesolution. .....81 5{9InvestigationofthedyedegradationundertheUVlightf ortwodifferentdyeconcentrations.TheUVisnothavinganapparenti mpactonthedye. ............................82 5{10TheresultsfortheexperimentsA-1toA-4. ...............85 5{11Collectivegraphofthedatapresentedabove. ..............86 5{12Investigationofthedyeadsorptiononthecarbonnanot ubessurface. Theadsorptionwasnotsignicantsinceitwasonly5%reduct ion after90min. ..............................87 5{13PhotocatalyticdegradationofDegussaP25and ` -ANTsunderUV lightof350nmwavelength. ......................89 5{14Thephotocatalyticresultsofthe ` -ANTsandDegussaP25.The ` ANTsclearlydemonstratephotocatalyticactivitywith =152.31 6.13 min.DegussaP25isnotdemonstratinganyobviousactivity. ....90 5{15Thedyedegradationdatainthedarkforthe ` -ANTs.Degussaisnot includedheresinceitneverdemonstratedbehaviorlikesuc h.The datawerettedwiththeequation5 9. DARK ` ANTs =1.29 0.24days. Theconstantis0.76 2.75 10 2 ...................91 5{16ThedyedegradationdataintheUVlightof350nmforthe s -ANTs. UV s ANTs =177.41 10.00mins.Thephotocatalysisissignicantly slowerthatallthepreviouscases. ...................93 6{1ThedierentRamanscatteringprocessesforCNTs. ..........98 6{2GraphicrepresentationofthemajorRamanmodes. .........100 6{3TypicalRamanspectrafrommetallicandsemiconducting SWNTs. TheRadialBreathingMode(RBM),theDBandandGBandarethemostimportantbands.The*isdenotingbandsthatcomefo rm theSisubstrate.Duetothedistinctstructureofthesemico nductingnanotubestherearetwoadditionalbandsMandiTOLAthatappear. .................................101 6{4TheGBandsplitandhowitisrelatedtotheconductivityo fthetubes. 101 6{5DierentoptionsfortheLOESSalgorithm. ...............109 xiv


6{6The ` -CNTsaftertreatedwithnitricacidat140‰for10hours.The DBandisshowingat1312cm 1 andtheGBandatabout1594 cm 1 .AverydistinctsplitofthebandcanbeseenwiththeG + at the1584cm 1 andG at1612cm 1 .................114 6{7The s -CNTsaftertreatedwithnitricacidat100‰for6hours.The DBandisshowingat1305cm 1 andtheGBandatabout1586 cm 1 .AlthoughtheGBandlookslikeitconsistsontooverlappingpeaksitstillcanbetreatedasonepeak. ............116 6{8TheRamanspectraofthecoatedlongnanotubes.Thereare twoseparateregions,(i)0-1000cm 1 thatcontainthetitaniapeaksand (ii)1000-1800cm 1thatcontainthecarbonnanotubespeaks.The peakidenticationisdonelaterinthechapter. ...........118 6{9Therstregionfromgure6{8.Therearefourmajorpeaks butonly threeofthemcanbeidentiedaccurate.149.56cm 1 ,628.65cm 1 and408.64cm 1 ............................119 6{10Thesecondregionfromgure6{8.TheDBandisat1307cm 1 and theGBandisattheabout1590cm 1 .Thebandsplitstillexists, withtheG at1579cm 1 andtheG + at1606cm 1 ........122 6{11TheRamanspectraofthecoatedshortnanotubes.Therea retwo separateregions,(i)0 1000cm 1 thatcontainthetitaniapeaks and(ii)1000 1800cm 1thatcontainthecarbonnanotubespeaks. Thisspectrahasbeenobtainedbythecombinationoftwodie rent runs. ...................................124 6{12Therstportionofgure6{11.Thereare5verydistinct ivepeaksat 150cm 1 ,202cm 1 ,393cm 1 ,510cm 1 and633cm 1 ......126 6{13Thesecondportionofgure6{11.Althoughthecarbonpe aksare notveryclearwecanstillseethematthe1316cm 1 theGBand andatthe1582cm 1 theGBand.TheGBandseemstobesplittingintwopeaks1544cm 1 and1582cm 1 .Theratiobetween thepeaksiscompletelyreversedbutthisiscurrentlyattri butedto theweaksignalobtainedbythe s -CNTsinthiscase. ........128 6{14TheC1speakforthereferenceanatasenanoparticles.T hemajorpeak isatthe286.4eVthatisagreementwithliteratureandsever aldatabases. 134 6{15TheSi2ppeakforthereferenceanatasenanoparticles. Themajor peaksareatthe98.5eVfortheSi2p 1 = 2 andat102.5eVforthe Si2p 3 = 2 whichareinagreementwithliteratureandseveraldatabase s. 135 xv


6{16TheO1speakforthereferenceanatasenanoparticles.T hemajorpeaks areatthe529.6eV,representsthelatticeoxygen,andthe53 1.5 eVforthesurfaceoxygen.whichareagreementinwithlitera ture andseveraldatabases. .........................136 6{17TheTi2ppeakforthereferenceanatasenanoparticles. Themajor peaksareatthe458.4eVfortheTi2p 1 = 2 andat464.2eVforthe Ti2p 3 = 2 whichareinagreementwithliteratureandseveraldatabase s. 138 6{18TheC1speakforthe s -ANTs.Themajorpeakisappearingtothe 284.6eV,whichisagainingreatagreementwithliteraturev alues. Thepeakat285.9eVischaracteristicoftheC Obondwhilethe 289.5eVpeakisattributedtoC O Ti. ...............139 6{19TheO1sforthe s -ANTs.Themajorpeaksareagainat530.6eVfor theO1sforthelatticeoxygenandthe532.7eVforthesurface oxygen.Theratiobetweenthosetwopeaksrevealsthesurfacear eof theparticle. ...............................140 6{20TheTi2ppeakforthe s -ANTs.Themajorpeaksareatthe459.4eV fortheTi2p 1 = 2 andat465.1eVfortheTi2p 3 = 2 ...........142 6{21TheC1speakforthe ` -ANTs.Againthemajorpeakappearstobe at284.6eVwhilethereisasecondarypeakat285.2eV.Thispe ak issimilartothecaseof s -ANTsthatappearsto285.9eV.Itisagain attributedtotheC ObondorC=Obond. .............144 6{22TheO1speakforthe ` -ANTs.Therearealsotwopeaksobserved at532.7eVandat530.9eV.Althoughbotharefromtheoxygenthe532.7eVisattributedtosurfaceoxygenwhiletheotherc omes fromlatticeoxygencontribution.Relativetothecaseof s -ANTs thesurfaceoxygenandthereforethesurfaceareaishigher, somethingthatwasconrmedwithBETaswellandisinagreementwithotherresearchers. .........................145 6{23TheTi2ppeakforthe ` -ANTs.Themajorpeaksareatthe459.6eV fortheTi2p 1 = 2 andat465.2eVfortheTi2p 3 = 2 whichareinsignicantlyshiftedcomparedtothereferencematerial. ..........146 6{24CollectiverepresentationiftheXPSdataregardingth ecoatedlong carbonnanotubes.TheupperrowistheTi2pandO1speakofthereferencematerialandthelowerrowisthedataobtainedbyt he s ANTs.Theshiftsinbothpeaksareobviousandaresummarizedintable6{9. ...............................149 xvi


6{25CollectiverepresentationiftheXPSdataregardingth ecoatedshort carbonnanotubes.TheupperrowistheTi2pandO1speakofthereferencematerialandthelowerrowisthedataobtainedbyt he ` ANTs.Theshiftsinbothpeaksareobviousandaresummarizedintable. .................................150 6{26CollectiverepresentationiftheXPSdataregardingth ecoatedlong andshortcarbonnanotubes.TheupperrowistheTi2pandO1speakofthe s -ANTsandthelowerrowisthedataobtainedbythe ` -ANTs.Thepeaksaresimilarregardingtheposition,butare signicantlydierentinshape. ......................151 6{27TheC1speakofthepeakofthecoatedcarbonnanotubes(b oth ` ANTsand s -ANTs)andthereferencematerial.Themaindierencebetweenthereferencematerialandthesamplesarethep eaks regardingtheC OandC=Obonds,thatareappearingonlyfor the s -ANTsand ` -ANTs,andthepeakat289.7eV( ` -ANTs)and 289.5eV( s -ANTs)thatcanbeattributedtotheC O Tibond. .152 6{28TheSi2ppeakofthepeakofthecoatedcarbonnanotubes( both ` ANTsand s -ANTs)andthereferencematerial.Althepeaksare atthesameenergy,butthenoisetosignalratioisalothighe rfor theboth ` -ANTsand s -ANTs.Thereasonforthatisthethickness ofthecoating.ThecoatedMWNTsweredepositedinathickerl ayer. 153 6{29CollectiverepresentationoftheRamanspectraregard ingtheshort nanotubesbefore(toprow)andafterthecoating(bottomrow ). TherightcolumnisfortheGbandandtheleftcolumnisfortheDband. .................................154 6{30CollectiverepresentationoftheRamanspectraregard ingthelong nantubesbefore(toprow)andafterthecoating(bottomrow) .The rightcolumnisfortheGbandandtheleftcolumnisfortheDba nd. 155 6{31CollectiverepresentationiftheXPSdataregardingth ecoatedlong carbonnanotubes.TheupperrowistheTi2pandO1speakofthereferencematerialandthelowerrowisthedataobtainedbyt he s ANTs.Theshiftsinbothpeaksareobviousandaresummarizedintable. .................................156 xvii


AbstractofDissertationPresentedtotheGraduateSchool oftheUniversityofFloridainPartialFulllmentofthe RequirementsfortheDegreeofDoctorofPhilosophy TITANIACARBONNANOTUBECOMPOSITESFORENHANCED PHOTOCATALYSIS By GeorgiosPyrgiotakis May2006 Chair:WolfgangM.SigmundMajorDepartment:MaterialsScienceandEngineering Photocatalyticcompositeshavebeenusedforthepastfewde cadesinawide rangeofapplications.Themostcommonapplicationisthepu ricationofair andwaterbyremovingtoxiccompounds.Thereislimiteduseh owevertowards biocidalapplications.Despitetheirhigheciency,photo catalyticmaterials arenotcomparabletotheeectivenessofconventionalbioc idalcompounds suchaschlorineandalcoholicdisinfectants.Ontheotherh and,nearlyadecade agowiththediscoveryofthecarbonnanotubesanewvibrants cienticeld emerged.Nanotubesareuniquestructuresofcarbonthatpos seamazingelectrical, mechanicalandthermalproperties. Inthisresearchcarbonnanotubesareusedasphotocatalyit icenhancers.They werecoatedwithanatasetitaniatoformacompositemateria l.Twodierenttypes ofnanotubes(metallicversusnon-metallic)wereusedandt hephotocatalyticactivitywasmeasured.Themetallictubesdemonstratedexcep tionalphotocatalyitic properties,whilenon-metallictubeshadlowphotocatalyt iceciency.Thereason forthatdierencewasinvestigatedandwasthemajorfocuso fthisresearch. xviii


Theresearchconcludedthatthereasonsforthehighecienc yofthecarbon nanotubeswere(i)themetallicnatureofthetubesand(ii)t hepossiblebond betweenthetitaniacoatingandtheunderlyinggraphitelay ers(C O Ti).Since bothcompositeshadthesameindicationsregardingtheC O Tibond,the metallicnatureofthecarbonnanotubesisbelievedtobethe mostdominantfactor contributingtotheenhancementofthephotocatalysis.The compositematerial mayhaveotherpotentialapplicationssuchasforsensingan dphotovoltaicuses. xix


CHAPTER1 INTRODUCTION Thelastfewdecadesthedemandforsaferenvironmentalcond itionshas increaseddramatically.Onereasonistheconstantlyincre asingbiologicalthreats thatcanbeexpressedineveryaspectofthedailylife,rangi ngfromcasesassimple asfoodbacterialcontamination( E.coli and salmonella )toextremelydangerous suchasepidemicoutbreaks( Ebola and SARS )andbiologicalwarfare( anthrax and smallpox ).Theneedforeectiveandecientdisinfectionisdriving theindustryin thedevelopmentofawiderangeofproducts.Theseproductsc anbedividedinto threemajorcategories:Chemicaldisinfectants: Chemicalbaseddisinfectantsarethemajorityofthey havebeenusedforthelongesttime.Mostofthemarechlorine ,alcoholor ammoniumbasedproducts.Theyareinliquidformandtherefo rearelimited tosurfaces.Themajorityareusedtodisinfectcontaminate dsurfacesandnot topreventcontamination.Althoughtheiruseisrelatively simpleandeasy theycanstillbedangerousiftheyaremisused.Gassescanal sobeusedfor thedisinfection,buttheyarelimitedsincetheyareextrem elycorrosive. Radiationbaseddisinfection: Theradiationisaveryeectivetechniquesince itcanimmediatelyinactivatethemajorityofthecontamina ntswithout damagingthesurroundings.Stillhowevertheuseislimited sinceitusually requiresexpensiveequipmentandundercertainconditions exposuretothe usedradiationcanbeproveddangerous. Passivedisinfectants: Passivedisinfectantsarecharacterizedthosethatdo notrequireacertainapplication(chemicals)oroperation (radiation),but constantlypurifyandcleansurfaces,airandwater.Activa tedcarbonlters 1


2 areprobablythebestknownandmostwidelyused,sincetheya reused widelyforwaterandairtreatment.Howevertheydonotdeact ivatethe contaminantssoconstantreplacementisrequired.Iftheya renotreplaced regularlytheycanbecomeasourceofcontaminationrathert handisinfection medium. Thelackofecientpassivedisinfectantshasledtheresear cherstoseeksolutionscapabletoprovidebothcapturingandinactivatingof biologicalcontaminants. Oneofthemostpromisingandrapidlyemergingeldsis photocatalysis 1.1PhotocatalysisandTitania Photocatalysis isthetypeofreactionthattakesplaceonthesurfaceofa certaintypeofmaterialinthepresenceofaveryspecicran geofradiation.There aremanymaterialsthatcandisplaythistypeofreaction,bu tthemostwidelyused istitaniumdioxide,TiO 2 ,or titania .Titaniainadditiontothehigheciencyis cheapandenvironmentallysafe. Therearesignicantlimitations,however,totheapplicat ionoftitaniasince theeciencyarenothighenoughoratleastcompetitivewith theresultsthat thechemicaltechniquescandeliver.Chapter2willgiveabr iefoverviewofthe principlesofphotocatalysisandspecicallythecatalysi sonthesurfaceoftitania. Emphasiswillbegiventothestructureoftitaniaanditsele ctricproperties,the twoprimaryreasonsfortheexcellentphotocatalyticprope rties.Itwilloutlinethe basictechniquesthatarecurrentlyusedtoimprovetheeci encyandnallywill discussthemajorapplicationsoftitania. 1.2CarbonNanotubes Analsorapidlyemergingeldistheinvestigationofthepro pertiesofthe carbonnanotubes .Theyarearelativenewmaterialthathasattractedgreatde al ofattentionduetotheuniqueshapeandstructure.Carbonna notubescanbe visualizedagraphitesheetthathasbeenrolledseamlessly intoatube.Ithasbeen


3 morethanadecadesincetherstreportofnanotubes.Theiru niqueproperties, thatariseformtheirstructure,havenotyetcompletelyund erstood.Probablythe mostoutstandingpropertiesaretheelectronic.Inadditio n,theirneedle-likeshape resultsinveryhighspecicsurfacearea.Bothcharacteris ticsareveryimportantto thepresentresearch. Althoughtheirpropertiesareveryuniqueandunmatchable, sofarthereisno commercialapplicationinsmallorlargescalethattakesfu lladvantageofthem. Chapter3willexplainindetailthestructureandlaterthep ropertiesofthecarbon nanotubes.Itwillalsogiveashortdescriptionofmostpopu larmethodsusedtoday fornanotubesproduction. 1.3Objectives Inthisresearchthosetwouniquematerialswillbecombined intheformofa nano-compositethatwilldeliverahighecientphotocatal yst.Thereareseveral researchersthathavealreadyachievedit,butthecomposit eshaveneverbeen investigatedin-depth.Thereforethisresearchhasthefol lowingsobjectives:ˆTosynthesizeTiO 2 -MWNTscompositesˆToevaluatethephotocatalyticeciencyˆToexplainthebehaviorofthematerial Therearetwodistincttrendsincombiningthosetwomateria ls:eitherinthe formoftitaniapelletwiththenanotubesembedded,orinamo resophisticated approach,thetitaniaisappliedasacoatingonthenanotube s.Inthisresearch thesecondapproachwasselectedsinceittakesfulladvanta geofthenanotube properties,bycreatingacompositewithasinglenanotubea score. Toinvestigatetheimpactofthenanotubepropertiestherea retwodierent compositessynthesized.Onehasapristineandhighlycryst allinecoreandthe otherhasofalessorderedtubularstructure.Thedirectcom parisonofthose


4 twocompositeswillexplaintheeectoftheelectricproper ties,ifany,atthe photocatalyticeciency.Allthesynthesisisexplainedin detailinchapter4. Thephotocatalyticevaluationisdone via dyereductiontests.Thosetypesof testsareverycommonandarepreferredsincetheygivefast, accurateandreliable results.Adrawbackofthosetestsisthemanyparameterstha tcanimpactthe resultsandthereforetheyhavetobemonitoredwhilethetes tsareexecuted,but itissomethingthatcaneasilybedone.Thoseparametersand theexperimentsare discussedindetailinchapter5. Inordertoexplainthebehaviorofthematerialitiscritica ltoselecttechniquesthatdirectlyorindirectlywilldeterminetheprope rtiesofthenanotubes. Oneofthemostrecognizedtechniquesforthatisthe Raman spectroscopy.In additiontoRaman, X-RayPhotoelectronSpectrometry (XPS)canbeusedtoinvestigatethestructureofthetitaniaandpointoutstructu raldierencesthatmay berelatedtothephotocatalyticevaluationresults.Theco mpleteanalysisofthose twotechniques,alongwiththenecessarytheorytoundersta ndRamanandXPS,is discussedinchapter6. Alltheexperimentalresultsfromchapters4,5and6willbeu sedtodraw conclusionsonhowthecarbonnanotubesbehaveasaphotocat alytictemplateand whattheimpactoftheirelectricalpropertiesisonthenal result.


CHAPTER2 PHOTOCATALYSISONTiO 2 (TITANIA)SURFACE:PRINCIPLESAND APPLICATIONS Recentlysemiconductorphotocatalysishasattractedagre atdealofattention sinceithasawiderangeofapplications[1,2].Oneofthemos tinterestingmaterialsistitania(TiO 2 )[3{5].TiO 2 isthematerialthatisusedhereascoatingon thecarbonnanotubes.Itiswidelyavailablesinceitisused aspigmentinmany applicationsandtheproductionisfairlycheap[4].Since1 972whenFujishimaet al.reportedthephotocatalyticsplitofthewateronTiO 2 electrodes[6]agreatdeal ofresearchhadbeendonetodevelopeapplicationandenhanc ingthepropertiesof titania.Theapplicationsrangefromphotovoltaiccellsto biologicaldisinfection [3{5]. Oneofthemostpopularapplicationsisthemicrobialsteril izationandselfcleaningsurfaces[7{15].Therearecertainlimitationsho wever,comingprimarily fromtheelectronicpropertiesofthetitania,thatreducet heeciency[4,5].The biggestbreakthroughcamein1985byMatsunagaetal.[9]whe rebymixingthe titaniawithsilverparticlestheobservedsignicantenha ncementofthecatalysis. Sincethattimetherangeofapplicationshasincreaseddram atically. Thischaptercoversthebasicinformationnecessarytoexpl ainthepropertiesof titania.Therstsectionisaboutthecrystalstructureand theelectronicproperties oftitania.Laterthechapterreviewsthebasicprinciplesb ehindphotocatalysisand therecentadvancestowardstheimprovementoftheeciency .Thelastpartofthe chaptergivesabriefoverviewofthemostimportantapplica tionsoftitania. 5


6 2.1StructureofTitania Titaniumdioxide(titania)existsinprincipleineightpha sesrutile,anatase, brookite,columbite,baddeleyite,rourite,pyrite,andco tunnite[16].Fromthose eightphasesthermodynamicallymorestablearerutile,ana taseandbrookite,with rutiletobethemoststable[17,18].Sincephotocatalytica ctivityisdemonstrated onlyfromrutileandanatase,theanalysiswillfocusonthos etwostructuresonly. Thecolumbite,baddeleyite,rourite,pyrite,andcotunnit ephasescanbegenerated onlyunderveryhightemperaturesand/orpressures,whichi sthereasonthose phasesdonotoccurnaturally[19{21],buttheystillposses ssomeveryinteresting properties.Cotunniteforexampleisthehardestpolycryst allinematerialknownto exist[16,22].2.1.1Anatase Figure 2{1 (a)showsthecrystalstructureofanatase.Itistetragonal with a = b =3 : 782 Aand c =9 : 502 AandhasaD 194 h -I4 1 /amdsymmetry.Thebuilding blockonanataseistheTiO 6 whichformsaslightlydeformedoctahedron(gure 2{1 (c)).TheTiatomthatisinlinewiththetwooxygenatoms(api caloxygen atoms)hasbondlengthof1.996 Aandtheotherfouroxygenatoms(equatorial oxygenatoms)haveTi Obondlengthsof1.937 A.Thewidestangleofthosetwo bondsTi O equatorial andTi O apical is102.308.Theanglebetweentwoconsecutive equatorialbondsis92.604or87.394.Allthebondlengthsandanglesgivenabove representthestructureatroomtemperature. Anataseisanunstablestructureandittransformstorutile atapproximately 800‰.Whilethetemperatureincreases,thebondlengthsarechan gingandgraduallytheanataseturnsintorutile[16,17].Rutilehasamor ecompactstructure andthereforeenergywiseismorefavorable.Thetransforma tiontorutileisan irreversibleprocess.


7 (a) (b) n n (c) r r r r r (d) Figure2{1:Thetwobasictitaniastructures(a)anataseand (b)rutile.The distortedoctahedronthatareshownareusedtoconstructth e(c) anataseand(d)therutile. 2.1.2Rutile Rutilehasalsoatetragonalstructure( 2{1 (b)),butitisalotmorecompact comparedtoanatase[16,23{25].Thetetragonalstructureh as a = b =4 : 584 Aand c =2 : 953 A.IthasD 154 h -P4 2 /mmmsymmetry[16,25,26].Againthebuildingblock ofthecrystalstructurehasanoctahedralthatisslightlyd istorted(gure 2{1 (d)). TheapicaloxygenatomshaveTi Obondlengthof1.983 Aandtheequatorial Ti Obondis1.946 A.TheequatorialandapicalTi Obondsformarightangle whilethelargestanglebetweenthetwoequatorialbondsis9 8.93.


8 (a)(b) Figure2{2:Theelectronicbandstructureofthetwomainpha sesoftitania(a) rutileand(b)anatase[25].Thecalculationisbasedonrst principles selfconsistentOLCAO. Thebondlengthinrutiledoesnotchangesignicantlywitht hetemperature. Itisthereforethermallyastablestructureandallthedie rentphaseswillturninto rutileafterannealingathightemperaturesforanextended period. 2.2ElectronicPropertiesofTitania Theelectronicstructureoftitaniahasbeenstudiedbothex perimentally andtheoretically.ExperimentallyithasbeenprobedbyX-R ayphotoelectron spectroscopy[27{30](XPS),X-RayinducedAugerelectrons pectroscopy[28], Augerelectronspectroscopy[28],X-Rayemission[31,32]( XES),absorptionspectroscopy[33,34](XAS),electronenergylossspectroscopy [27,35{37](EELS), ultravioletphotoelectronspectroscopy(UPS)[38]andres onantultravioletphotoelectronspectroscopy(RUPS)[38].Thetheoreticalanalys ishasbeendonemainly withtotal-energycalculationswithintheLDAusingpseudo potentialplanewave


9 formalism[23,24,39],aswellasthemorerecentHartree-Fo ckpseudopotentialcalculations[40].Recentlyveryaccurateself-consistent abinitio calculationsforTiO 2 havebeenperformed.Priortothosemethodstheattemptstot heoreticallypredict theelectronicstructureoftitaniaweredonebasedontheti ght-binding[41{47] (TB)calculationsandtheextendedHuckelmolecularorbit almethod[33,36]. Certaindefectsinthecrystalstructurecanimpacttheelec tricproperties oftitania.Titaniaisanoxygendeciencymaterialandusua llyitisconsidered n -typesemiconductor.TheFermi-levelthereforeisnotata xedvaluesincethe productionmethodwilldeterminetheoxygendeciencyandt hereforetheFermilevelshift.Thisistrueforbothanataseandrutile.Inaddi tiononeofthemost commondefectsintitaniaistheTi +4 substitutionbyTi +3 (andoftenTi +2 an Ti +1 )[48,49],whichalsocreatesachargeimbalancethatbeyond fortheelectrical properties,canaectspectroscopictechniquesthatrelyo ntheelectroniccharge, suchasXPS.ThoseTicationscanbegeneratedbyannealing,s putteringor chemicalreduction.2.2.1Anatase Figure 2{2 (b)showstheanatasebandstructure.Thebandgaphasbeen experimentallymeasuredandis3.2eV[50],whilethetheore ticallydetermined valuescanvaryfrom2.2eVupto3.89eV[25].Thosedierence sarerelated tothenumberofatomsaccountedtothecalculationsandmost importantthe non-constantbondlengthinthecrystal(section 2.1.1 ).Forthisresearchthe experimentalvalueof3.2eV,whichhasbeenrepeatedlycon rmed[50],willbe acceptedasthebandgapenergy.Thewidthofthevalenceband is4.75eVand thedistancebetweentheuppermostconductionbandstatean dthelowermost valencebandstateis17.7eV[25].Mostofthetheoreticalca lculationsshowthat thebandgapisalmostindirect,whichisnotcorrect.Itisof tenattributedtothe factthatanatasehasaveryunstablestructure[25].


10 Anatasealsohasaveryhighcarriermobility,80cm 2 /Vs[51],(89timesfaster thanrutile)[52].Sincethebandgapis3.2eVthemainabsorp tionpeakisat395 nm.TheHallmobilityis20cm 2 /Vsatroomtemperature[53]. 2.2.2Rutile Figure 2{2 (a)showstheelectronicstructureofrutile.Rutilehasaba ndgap thatexperimentallyhasbeenmeasuredtobe3.0[54]andwith calculationsitis 1.78eVupto3.73eV[23,55].Inthiscasethereasonforthela rgevariationis primarilythenumberofatomsaccountedinthecalculationa ndsecondarilybond lengthvariations.TheuppervalencebandiscomposedofO2p orbitalandhasa widthof5.4eV.ThelowerO2sbandis1.94eVwide[30].Thesep arationenergy betweentheupperconductionbandandtheminimumvalenceba ndhasbeen measuredexperimentallyandis16-18eV[30].Thelowestcon ductionbandconsists ofntwosetsofTi3dandis5.9eVwide[25]. 2.3SemiconductorPhotocatalysis Theterm photocatalysis isstillunderdebatesincestrictlythetermimpliesthe initiationofreactionsinthepresenceoflightonlysometh ingthatisnotaccurate inthecaseofsemiconductorphotocatalysis,sinceinthisc asethepresenceofthe semiconductorisequallyimportant[56].Butforthepurpos eofthisresearchthe termphotocatalysiswillbeused,andwilldenotethereacti onthattakesplaceon thesurfaceofasemiconductorinthepresenceofacertainra ngeofradiation. TherstreportonphotocatalyticactivitywasbyBecquerel in1839when heobservedvoltageandelectriccurrentonasilverchlorid eelectrodewhenit isimmersedinelectrolytesolutioninthepresenceofsunli ght[57].Technically allsemiconductorscandisplayphotocatalyticproperties ,butusuallytheoxides andcompoundsemiconductorsaredemonstratingsignicant lybetterresults [5,58,59].Theabilityofasemiconductortoundergophotoc atalyticoxidationis governedbythebandenergypositionsofthesemiconductora ndredoxpotentials


11 Figure2{3:Schematicdiagramrepresentingthemainphotoc atalystswiththeir bandgapenergy.Inordertophoto-reduceachemicalspecies ,the conductancebandofthesemiconductormustbemorenegative than thereductionpotentialofthechemicalspecies;tophoto-o xidizea chemicalspecies,thepotentialofthevalencebandhastobe more positivethantheoxidationpotentialofthechemicalspeci es.The energiesareshownforpH0. oftheacceptorspecies.Thelateristhermodynamicallyreq uiredtobebellow (morepositivethan)theconductionbandpotentialofthese miconductor[5,59]. Thepotentiallevelofthedonorneedstobeabove(morenegat ivethan)the valencebandpositionofthesemiconductorinordertodonat eanelectronto thevacanthole.Figure 2{3 areshowssomeofthemostpopularsemiconductor photocatalystsrepresentedwiththeirbandenergypositio ns.Theinternalenergy scaleisgivenontheleftforcomparisontotheNormalHydrog enElectrode(NHE). Thepositionsarederivedfromtheratbandpotentialinacon tacttoasolution ofaqueouselectrolyteofpH0[59].AmongthemTiO 2 isthemostpopular.Itis, ecient,eective,requiresshallowUVradiation,isveryc heaptomanufacture,


12 environmentallysafeandeasilyincorporatedwithotherma terials.Since1972when theabilitytosplitthewaterunderUVradiationwasrstdis covered[6]therehas beengreatworkinunderstandingthemechanismandthereact ionsthattakeplace. 2.3.1BasicPrinciples Figure 2{4 schematicallyrepresentsthestepsofphotocatalysis.Ini tiallywhen aphotonofproperenergy( h E g )strikesthesurfaceofthesemiconductor itgeneratesanelectronholepair( h + e ).Bothelectronandholes,either recombinedormigratetothesurface,where,theyproceedwi thchemicalreactions. Theholesaregenerating[OH ]andtheelectronsH 2 O 2 .Averyimportantfactor forthoseprocessesistherequiredtime.Herearesummarize dthemainreactions andthetimerequiredforeachone[4].Therequiredtimehasb eenmeasuredwith laserrashphotolysis[60,61]:ˆCharge-carriergeneration TiO 2 + h h +vb + e cb ; 10 15 s(2 1)ˆCharge-carriertrapping h +vb + > Ti IV OH > Ti IV OH + ; 10 10 9 s(2 2) e cb + > Ti IV OH > Ti III OH ; 100 10 12 s(2 3) h +vb + > Ti IV > Ti III ; 10 10 9 s(2 4)ˆCharge-carrierrecombination e cb + > Ti IV OH + > Ti IV OH ; 100 10 9 s(2 5) h +vb + > Ti III OH > Ti IV OH ; 10 10 9 s(2 6)ˆOxidationorreduction > Ti IV OH + +Red 0 > Ti IV OH+Red + ; 100 10 9 s(2 7) e tr +Ox > Ti IV OH+Ox + ; 10 3 s(2 8) Accordingtotheaboveproposedmechanismtheoverallquant umeciency dependsontwomajortypesofreactions,thecarrierrecombi nationandthe


13 h + e Ox Ox red 0 red + CO 2 ,Cl H + ,H 2 O 1 2 3 3 4 5 6 Figure2{4:Schematicrepresentationofthereactionstaki ngplaceintitania. 1OLightstrikesthesemiconductor. 2OAnelectron-holepairisformed. 3OElectronsandholesaremigratingtothesurface. 4OTheholesinitiate oxidationleadingtoCO 2 ,Cl H + ,H 2 O. 5OTheconductionband electronsinitiatereductionreactions. 6Oelectronandholes recombinationtoheatorlight. [OH ]/H 2 O 2 generation.Thedominantreactionistherecombinationoft hee and h + (1ns)followedbythereductionreaction(10ns)andoxidati on(1ms).Since therecombinationisalsoassistedbythelocalizedcrystal defects,theremaining carriersarenotenoughforanecientphotocatalyticreact ion. 2.3.2EnhancementofPhotocatalysis Itisnecessarytoenhancethephotocatalyticeciencyofti taniatoobtain amoreeectivematerial.Time-wisetheoxidationcomingfr omtheholesisthe fastestdegradingreaction[60].Itisreasonabletherefor etofavorthisreaction overthereductionreactioninitiatedbytheelectrons.Sin cethemechanismthat isresponsibleforthereducedeciencyistherecombinatio nbetweenthe h + and e allthepreviousresearchhasfocusedoneitherscavengingt heelectronsaway fromthesystemtopreventrecombination,orjustretarding therecombinationso theholeswillgenerate[OH ][4,5,58,59].Namelythebestknownwaysarethe dopingoftitania,thecouplingwithametalandthecoupling ofasemiconductor.


14 C.B. E g s s V.B. E f E g Vacuum C.B. E g s s V.B. g Traplevels E f E g Vacuum (a) (b) Figure2{5:Titaniabandstructure(a)beforeand(b)afterd oping.Thetransition metalsareinterstitialorsubstitutionaldefectsinthest ructureof titaniaandgeneratetrappinglevelsinthebandgap. Since1972therehasbeenextensiveworktowardsallthreety pesofphotocatalytic enhancementwiththetitania/semiconductorandtitania/m etalcouplingmore dominantsincetheyareeasiertoachieve. Dopingoftitania. Agreatdealofworkhasbeendonethelastfewdecades todopetitaniawithtransitionmetals,N[62]andC[63,64]. Ingeneraltransition metalsareincorporatedintothestructureoftitaniaandoc cupysubstitutional orinterstitialpositions.Itisaverycommondefectinthec aseofsemiconductors sinceitgeneratestraplevelsinthebandgap.Figure 2{5 (a)showstheelectronic structureoftitaniabeforethedoping.Afterthedoping(g ure 2{5 (b))thebandgap hasbeenmodiedwiththeadditionofthetrappinglevels.Th etraplevelsare usuallylocatedslightlybelowtheloweredgeoftheconduct ionbandandusually areinaformofanarrowband. Thereareseveraladvantagestothismodication.Beforeth emodication therequiredphotonenergyhadtosatisfythecondition h E g .Afterthe modicationtherequiredenergyisgoingtobe h ( E g E t )where E t isthe


15 loweredgeofthetrappinglevelband.Inadditiontheelectr onsthatareexcitedat thoselevelsaretrapped,andtheholeshavesucienttimefo r[OH ]generation. Eveninthecasethat h E g andtheelectronisexcitedtotheconductionband, thenduringthede-excitationprocesstheelectronisgoing tobetransitionedfrom theconductionbandtothetraplevelsandthentothevalence bandwhichagain retardstherecombinationandthereforeincreasestheover alleciency. ThemostcommontransitionmetalsusedareFe +3 ,Cr +3 andCu +2 .Fe +3 dopingoftitaniahasbeenshowntoincreasethequantumeci encyforthe reductionofN 2 [65{67]andmethylviologen[65]andtoinhibittheelectron hole recombination[60,61,68].Inthecaseofphenoldegradatio nScalfanietal.[66] andPalmisanoetal.[69]reportedthatFe +3 hadlittleeectontheeciency. EnhancedphotoreactivityforwatersplittingandN 2 reductionhavebeenreported withCr +3 [69{72]dopingwhileotherreportsmentiontheoppositeres ult.Negative eectshavebeenalsoreportedwiththeMoandVdoping,while GratzelandHowe reportedinhibitionofelectronholerecombination.Final lyKarakitsouandVerykios notedapositiveeectontheeciencybydopingoftitaniawi thcationsofhigher valencythanTi +4 [73].ButlerandDavis[74]andFujihiraetal.[75]reported that Cu + canalsoinhibitrecombination. Couplingwithametal. Inphotocatalysistheadditionofmetalscanaect theoveralleciencyofthesemiconductorbychangingthese miconductorsurface properties.Theadditionofmetalwhichisnotchemicallybo ndedtotheTiO 2 can selectivelyenhancethegenerationofholesbyscavenginga waytheelectrons.The enhancementofthephotocatalyisbymetalwasrstobserved usingthePt/TiO 2 system[76,77]byincreasingthesplitofH 2 OtoH 2 andO 2 .Inparticularcasesthe additionofmetalcanaectthereactionproducts. Figure 2{6 demonstratestheeectontitaniabandstructurewhentitan iais coupledwithametal.Ingeneralwhenasemiconductorthatha sworkfunction s


16 E f Vacuum m C.B. s s V.B. E intf E f E g V.B. Vacuum (a) E f Vacuum C.B. V.B. Vacuum s s E g m (b) Figure2{6:Theprinciplesofrectifyingcontactbetweenti tania(E g =3.2eV)anda metalwithworkfunction( m ),inthisexample5eV,greaterthanthe anity( s )oftitania.(a)Beforethecontact,and(b)afterthe contact,whereabarrierisformedtopreventtheelectronso fcrossing backtothesemiconductor.TheE intf istheFermileveliftitaniaisan intrinsicsemiconductorandE f istheFermilevelasanoxygendecient material. iscomparedwithametalwithworkfunctionof m > s theFermilevelofthe semiconductor, E s f ,ishigherthantheFermilevelofthemetal E m f (gure 2{6 (a)). Sowhenthetwomaterialsarebroughtincontact(gure 2{6 (b))therewillbe


17 electronsrowingfromthesemiconductortothemetaluntilt hetwoFermienergy levelscometoequilibrium.Theelectronstransitionwillg enerateanexcessof positivechargethatcreatesanupwardbandbending.Thisbe ndingcreatesasmall barrier(intheorderof0.1eV)thatexcitedelectronscancr ossandbetransported tothemetal.Fromthemomenttheelectronsmigratetothemet alitisnotpossible tocrossbacksincethebarrierforthisactionislargerandt hereforetheelectrons willremaininthemetal. 1 TheearliestworkontitaniametalwasthePt/TiO 2 electrodeforthesplit ofwater[76,77].Currentlythemosteectivemetal/TiO 2 interfaceisachieved bycolloidalsuspension[78].Itwasfoundthatinthecaseof Pt/TiO 2 system thePtparticlesaregatheredintheformofclustersonthesu rfaceofTiO 2 [79]. Othermetalshavealsobeeninvestigated.Aghasbeenfoundt oincreasethe eciency[80].OthertransitionmetalssuchasCr +3 negativelymodifythesurface bycreatingrecombinationsites.Althoughinprincipleall metalscanbeused,noble metalsarepreferredsincetheyhavehigherworkfunctionan dbetterconductivity. Inallcaseshighsolidsloadingwillaectthekineticsofth esystem,thelight distributionandeventuallydecreasetheoveralleciency [81]. Couplingwithasemiconductor. Couplingasemiconductorwithaphotocatalystisaveryinterestingwayofassistingthephotoc atalysis.Figure 2{7 demonstratestheprinciplesoftheTiO 2 couplingwithanothersemiconductor.In thisexampleastitaniaisconsideredtheanatasephase,whi lethethesemiconductoristherutilephase.Whentwosemiconductorsarebrought together,asinthe previouscase,theFermilevelstendtobalancesoelectrons arerowingfromthe semiconductorwiththehighestFermileveltothesemicondu ctorwiththelowest. 1 Accordingtoquantummechanicsthereisanitepossibility thattheelectrons cancrossback,butthenumberoftheelectronsthatcandotha tisinsignicant.


18 C.B. s s V.B. E f E g Vacuum C.B. rs rs V.B. E rf E rg Vacuum (a) C.B. E g s s V.B. E f E g Vacuum rs rs E rg C.B. (b) Figure2{7:Theprinciplesofrectifyingcontactbetweenan atase( )titania (E g =3.2eV)andandrutile( r )titania(E rg =3.0eV).(a)Beforethe contactand(b)afterthecontact,whereabarrierisforming toprevent theelectronscreatedinanatasecrossingtotherutile.Ont heother handholescreatedintoanatasecanmigratetorutile.Sothe coupleof anatase-rutileiscreatingandeectiveelectron-holesep aration. Thischargetransferwillcreateanexcessofpositivecharg etothesemiconductor thathadthehighestFermilevelandanexcessofnegativecha rgetothesemiconductorthathadthelowestenergy(gure 2{7 (b)).Bylightillumination, e h + pairsaregeneratedinbothsemiconductors.Thebarriertha tformsseparatesthe


19 electronsintheconductionband,butatthevalencebandthe holesarefreeto moveandbasedontheenergydiagramtheymovefromthesemico nductorwiththe largergaptotheonewiththesmaller.Inthiscasethecompos itematerialisacting asachargeseparator.Theholesaregatheredintherutilewh eretheycreatean excessofholes,anddespitethefactthattherecombination isstillthemainprocess theexcessofholeswillbeenoughtophoto-oxidizetheorgan icmolecules. Inadditionsemiconductorscanbeusedasaholeorelectroni njector.Inorder toachieveoptimumresultsacandidatesemiconductorhasto satisfythefollowing criteria.ˆHaveaproperband-gapˆHaveaproperpositionoftheFermienergylevelˆHaveproperrelativepositionoftheconductionandvalence bandtothe vacuumlevel. ThecombinationofthebandgapandFermilevelwilldetermin eifthereareholes orelectronsthatwillbeinjectedandtowardswhichdirecti on.Thusinorderfor twocoupledasemiconductorwithtitaniainordertoenhance thephotocatalysis, thesemiconductorhastohaveveryspecicproperties.This isthereasonthatthis technique,despiteitssimplicity,easeofmanufacturinga ndverypromisingresults, isnotverywidelyapplied.Systemsthathavebeendeveloped aretheTiO 2 /CdS [82],TiO 2 /RuO 2 [83]and Anatase -TiO 2 / Rutile -TiO 2 [52,84].Thelastoneisa systemcommerciallyavailablefromDegussa,knownasAerox ideP25,andisthe mostpowerfulcommercial,particulate,photocatalyticsy stem[84].Theexcellent anduniformpropertieshaveestablisheditasbenchmarkmat erialtocompare photocatalyticeciencies.


20 2.4ApplicationsofPhotocatalysis Inthissectionarereviewedthemainapplicationsofthepho tocatalytic systemsthathavebeendescribedabove.Themostpopularuse sareinenvironmentalapplicationandphotovoltaiccells.Thereareother applicationssuchas anti-fogcoatingandpigmentsinpaints,butsincetheydono tutilizetheelectrical propertiesoftitania,theyarenotgoingtobeexplainedher e. 2.4.1EnvironmentalApplications Duringthelastfewdecadestheenvironmentalapplications ofTiO 2 have attractedagreatdealofattentionsincetitaniacanbetheb aseoflowmaintenance systems.Sofartheymainlyfocusonwaterandairtreatmenta ndtheobjectives areprimarilytheremovaloforganiccontaminants[4,85{87 ]andsecondarily biocidalapplications[3,8,9,11,14,88].Althoughthesys temscanequallytarget biologicalcontaminantstheeectivenessislessorequalt oothercompetitive technologies(chemicaldisinfection).Sothebiologicala pplications,althoughthey areuniqueandinteresting,arenotwidelyutilized. Severalreactorscongurationhavebeendevelopedforthem osteective removalofthecontaminants[89{91].Oneofthemostpopular congurations, mainlyforexperimentalapplication,istheslurryreactor ,wherethewateris mixedandagitatedwithtitaniaparticlesunderthepresenc eofUVradiation.The mainadvantageofthiscongurationisthehighsurfacearea thatallowsfaster processing.Themaindisadvantageistheseparationofthep articlesafterthe reaction,whichisaverytediousprocess.Theycanbesepara tedbyltration, centrifugation,coagulationandrocculation[86,92,93]. Recentlymagneticcore hasbeenusedtoassistthedispersionandrecollectionofth eparticles[94].An alternativetotheslurryreactionistheratbedreactorwhe retheparticlesare immobilizedonaceramicmembrane[95].Theeciencyislowe rcomparedto theslurryreactionduetothelowersurfacearea,butthesys temdoesnotneed


21 anykindofseparation,whichaddstotheoveralleciency.R ecentlyinorderto increasethesurfaceareaofthetitaniatheparticleshaveb eencoatedontubes[95], glassbeads[96],berorwovenglass[97].2.4.2PhotovoltaicCell Solarcellshavebeenusedthepastfewdecadeswithgreatsuc cessinsmall devices.In1991GratzelandOregan[98]reportedahighec iencysolarcellbased onTiO 2 .Thetitaniausedinthosecellsisusuallydyesensitized[9 9{101]. ThebasictitaniacellconsistsofasandwichofaTiO 2 ,sensitizingdye, electrolyteandthecatalystbetweentwoconductivetransp arentelectrodes.The substrateusuallyusedforthisapplicationisastandardtr ansparentelectron conductor(TEC)glasswithhighopticaltransmissionandlo wresistance.Titania isanexcellentmaterialtobeusedasbasesinceitcarriesag oodcombinationof opticalandelectricalproperties.Thedyeisrequiredtoab sorbthesunlightand injectelectronsintotitaniawithalmost100%eciency.Th eoxidizeddyemolecule isthenreducedbytheredoxelectrolyte.Theelectrolyteit selfisthenreducedat thecounterelectrode.Thecycleexcitation-oxidation-re ductionisthenrepeated. Dyesensitizedsolarcells(DSSCs)continuetoattractmuch attentionas viablesystemsforconversionofsolarenergy[102].Atitan iacellthatissensitized byaRuN3dyeachievesthehighesteciency.Thebestecienc yreprotedis 10%[102].Retartationoftherecombinationcanfurtherinc reasetheeciency ofthecell.Thepropertiesoftheselmsdependonthephase, morphologyand preparationmethodthatwereused.Thereareawidevarietyo ftechniquesthat thoselmsaresynthesized.Traditionaltechniquesinclud eCVD,aerosolpyrolysis, electrodepositionandsol-gelprocessing[100].Mostofth emleadtoamorphous, partiallycrystallizedorfullycrystallizedanatase.For theDSSCanataseTiO 2 isstillconsideredthebestmaterial,butrecentlybrookit ewasreportedtobe successfullyusedastheelectrodematerial.


22 Theseprocessesareexpectedtobesensitivetothecrystals tructure,sizeand morphologyoftheexposedlatticeplanesasitwasshown,asw ellastothebandgap andtotheratbandpotentials.Solarcellphotopotentialis especiallysensitiveto thenatureofthesemiconductorsurfacethatdetermineslar gelytheoccurrenceof reversereactions(i.e.,recombination).Thebestactuals olarcellsworkwiththe I 2 /I (orBr 2 /Br )couple,becauseofaslowkineticsforI 2 reductiononSnOand especiallyonTiO 2 surfaces.


CHAPTER3 CARBONNANOTUBES(CNTs):STRUCTUREANDELECTRICAL PROPERTIESOVERVIEW CarbonnanotubeshavebeendiscoveredbyIijima[103]in199 1andsince theirdiscoverytheyhaveattractedagreatdealofattentio nduetotheexceptional electronic[104],thermalandmechanicalproperties[105] .Iijimareportedthe creationofmultiwallcarbonnanotubes(MWNT)withouterdi ameterupto55 Aandinnerdiameterdownto23 A.Sincethattimeextensivetheoreticaland experimentalresearchforthepastdecadehasledtothecrea tionofarapidly developingresearcheld.In1993Bethuneetal.[106]repor tedthediscoveryof thesinglewallnanotubes(SWNT).Theverysmalldiameterof thesinglenanotubes andtheverybiglengthmakesthembehavingasquantumwires, givingthem veryinterestingproperties.DuetothefactthattheSWNTus uallycontaina smallnumberofcarbonatoms(usually < 10 2 ),theyhaveattractedalmostall thetheoreticalwork.Theypossesssomeremarkableelectro nic,mechanicaland thermalpropertiesthatarerelatedmainlytotheirdiamete randchirality.Sincethe nanotubesarethephotocatalytictemplate,thischapterwi llgiveageneraloverview oftheiruniqueelectricalproperties.Initiallythesepro pertieswillbedescribedfor theSWNTthathavebeenmoreintensivelystudiedandunderst ood.Latersomeof theconceptswillbeexpandedtoincludetheMWNTs.Focuswil lalsobegivento thephysicsofthenanotubesandespeciallythestructurean dhowthestructureis relatedtotheelectricpropertiesandtheRamanactivevibr ationalmodes.Thelast partofthischapterwilldiscusesandcomparetheseveralpr oductionmethodsof nanotubesandhowthosemethodseventuallywilleecttheir properties. 23


24 3.1Bonding,StructureandPhysicsofSingle-WallCarbonNa notubes Tounderstandthestructureofthenanotubesitiscriticalt oreviewthe dierentbondstructuresofcarbon.Explainingthephysica lpropertiesofthe singleandmultiwallcarbonnanotubesitisrequiredtoderi vecertaingeometric relationandexplainthebasicnotationusedfortheNTs.Iti simportantalsoto describeseveralsymmetriesofthetubes,andhowtheycorre latetothevibrational frequencies.Thosefrequenciesarecrucialforexplaining inchapter6inthis documentbondingandelectronicbehavior.3.1.1BondinginCarbonMaterials Acarbonatomhassixelectronsfromwherethersttwoareocc upyingthe 1 s stateandtheotherfourareatthe spp x and p z or sp 2 and p z or sp 3 hybridized orbitalsdependingonthestructure.The sp 3 orbitalisusedforexampleatthe diamondstructure,resultingthreedimensionalinterlock ingstructurethatis responsiblefortheextremehardnatureofdiamond[107].In graphite,thethree outershellelectronsoccupythethree sp 2 orbitals,thatiscoplanar,andform threein-planebonds( bond)andoneout-of-planebondwiththe p z ( bond) orbitalthatisperpendiculartothe bondplane[108].Thisresultsinhoneycomb structuredcarbonsheet(graphenesheet).Thegrapheneshe etsareheldtogetherby vanderWaalsforces.The bondinthe sp 2 orbitalis0.14nmlongandhasenergy of420kcal/mol,wherein sp 3 itis0.15nmandhasenergyof360kcal/mol.Itis obviousthatthegraphitesheetisstrongerintheplanedire ctionthatdiamond. Sincethecarbonnanotubesarerolledgraphenesheetsthebo ndingisessentially sp 2 .However,duetothecurvatureofthetube,the and bondsaregoing tobere-hybridized.Thenewstructurepush bondsoutoftheplane,allatthe samedirection(towardsthecenterofthetube).Tocompromi sethechargeshift the bondwillbede-localizedtothedirectionoutsidethetube. Thisconguration willmakethetubesmechanicallystrongerandelectrically andthermallymore


25 Figure3{1:The2Dgraphenesheetsisshownwiththe a 1 and a 2 speciesthe chiralityofthenanotube.Thechiralvector, C h ,isthe OA ,whilethe translationvector T isthe OB .Also istherotationangleand the translation.Thosetwoareconstitutethesymmetryoperati on R =( j ). conductingthangraphite.Therexibilityofthe bondallowstheincorporation oftopologicaldefects,suchaspentagonsorheptagons,tha tallowtheformationof caps,bend,toroidalorhelicaltubes[109]. Thefullerenes C 60 aremadeof20hexagonsand12pentagons[110].The bondingisalso sp 2 ,althoughduetothehighcurvatureitresembles sp 3 .This uniquestructuregivestothefullerenesaveryinteresting setofproperties. 3.1.2StructureandNotation ASWNTcanbethoughtofasagraphenesheetrolledseamlessly inacylinder [111].Itusuallyhas10-40carbonatomsincircumferencean discapped.The directionthatthegraphenesheetisrollediscalledchiral ityanditisspeciedby thechiralvector C h (gure 3{1 ).Thehoneycombstructureisdescribedbythe


26 vectors a 1 and a 2 andallthevectorscanbedescribedasalinearcombinationo f thosetwovectors. C h canbedenedas(gure 3{1 ) C h = n a 1 + m a 2 ( n;m )(3 1) whichoftenisdenotedwiththe( n;m )symbol.Averyimportantvariableisthe angle whichistheangleofthechiralvectorwiththe a 1 direction[112].The a 1 directioniscalled zigzag .Consequentlynanotubesrolledtothatchiraldirection arecalled zigzag [113].Therearemanypossibledirectionsthatthegraphene sheet canberolledwithdierentproperties(gure 3{2 ).Thedirectionthathas =30 iscalled armchair [112].Alltheothernanotubesforwhich0 << 30 arecalled chiral.Forangles > 30 and < 0 rotationalsymmetryrulesapply.Thetube diameter d t canbewrittenintermsoftheintegers( n;m )as: d t = j C h j = 1 p 3 a CC m 2 + nm + n 2 1 = 2 (3 2) where a CC isthenearestneighborC Cdistance(1.42 Aingraphite).Fromthe geometryingure 3{1 thecos andsin canbecalculated, sin = p 3 m 2 p m 2 + nm + n 2 ; cos = 2 n + m 2 p m 2 + nm + n 2 (3 3) Consequentlythechiralangle is =tan 1 p 3 m m +2 n # (3 4) The( d t ; )paircancompletelydescribethenanotubesandoccasional lyitisused asanalternativetothe( n;m ).Thetranslationvector T isanotherimportant vector,whichonthenanotubedenotesthelongitudinaldire ctionandisverticalto the C h ( C h T =0).Itisdenedas T = t 1 a 1 + t 2 a 2 ( t 1 ;t 2 )(3 5)


27 wherethecoecients t 1 and t 2 arerelatedtothe n and m by t 1 = (2 m + n ) d R ;t 2 = (2 n + m ) d R (3 6) where d R isthegreatercommondivisorof(2 n + m; 2 m + n )andisgivenby d R = 8><>: d; if n m isnotamultipleof3 d 3 d; if n m isamultipleof3 d (3 7) where d isthegreatestcommondivisionof( n;m ).Themagnitudeof T is j T j = T = p 3 C h =d R .Asunitcellofthenanotubeisdenedtheareadelineatedby thevectors T and C h .Soforinstanceingure 3{1 theunitcellisdenedbythe OBB'Aparallelogram.Thenumberofhexagons, N ,containedwithinaunitcellof ananotubeisdeterminedbytheintegers( n;m )andisgivenby N =2 ( m 2 + n 2 + nm ) d R (3 8) where d R isdenedbyequation 3 7 .Thecarbonnanotubesareusuallycapped. Thecapcanbethoughtofasafullerene( C 60 )thathasbeenbisectedatthe equator.Soforexampleifthe C 60 isbisectednormaltoavefoldsymmetryaxis thenthatcapissuitableforarmchairtube,whileifitisbis ectednormaltothe 3foldsymmetryaxisthentheresultingcapissuitableforaz igzagtube[112]. Sincetherearemanydiameterstherearemanydierentcapst hatcantthem [112,114].Figure 3{2 showsseveralrollingdirectionsandbasedonthosedirecti on thenumberofdistinctcapsthatcanbejoiningthemseamless ly. 3.1.3SymmetriesandVibrationalFrequencies Averygeneralwaytosimplifytheanalysisistoassumethatt henanotubes haveverybiglengthcomparedtothediameterandthereforei gnorethecaps.In generalwecandistinguishtwomajortypesofsymmetricgrou ps,symmorphicor non-symmorphic.Thezigzag(( n; 0)tubes)andarmchair(( n;n )tubes)belongto


28 " # $ % & ( ) " # # $ $ % % & & ' + # $ % & ( ) " $ # % # & # # ( # ) # # # % $ & $ $ ( $ ) $ $ $ " & % % ( % ) % % & ( & ) & & ( ) ( ( ) ( # $ % & ( . / 0 1 2 2 3 2 / 2 1 4 5 2 1 4 3 1 0 4 Figure3{2:Thegraphenesheetisshownwiththe( n;m )pairwhichspeciesthe chiralnanotube.Thepairofinteger( n;m )inthegurespeciesthe chiralvector C h forcarbonnanotubes,includingzigzag,armchairand chiraltubules.Beloweachpairofintegerislistedthenumb erof distinctcapsthatcanbejoinedcontinuouslytothecylindr icalcarbon tubuledenotedby( m;n )[ref].Itisalsodenotedtheconductionstate ofeverychirality. therstgroupwhiletheotherchiralbelongtothesecond.Th ebasicdierence thatinthecaseofsymmorphicthetranslational( )androtational()operation (bothshowningure 3{1 )caneachbeexecutedindependently,whileforthe non-symmorphicthisisnottrue. Thecompleteanalysisisverycomplicatedandisbeyondthes copeofthis research.Brieryherewillbementionedtheverybasicprinc iples.Duetotheir highcomplexitythechiraltubesarenotgoingtobeincluded intheanalysis.From equation 3 8 itcanbecalculatedthatforcertainstructuresthe N canbevery large.Forexampleforthe(30 ; 15) N =210[103,115].Thesymmetriesforthose structuresareverycomplicated[114].Forzigzag( n; 0)andarmchair( n;n )areless complicated.The( n;n )and( n; 0)thesymmetrygroupscanbedescribedby D nh or D nd ,thatareevenorodd,respectively.


29 Thesymmorphicsymmetriesusuallyhaverelativesmallarea of1Dunitcell ( C h T ),thereforethenumberofphononbranchesornumberofelect ronicenergy bandsaresmall.Onthecontraryforthechiraltubesthatnum berisverybig,since theareaofthe1Dcellislarge.Forthezigzagtubes( n; 0)thereare4 3 n =12 n degreesoffreedomwith60phononbranches,havingsymmetry types(for n odd, andthus D nd symmetry)[114]: vibn =3 A 1 g +3 A 1 u +3 A 2 g +3 A 2 u (3 9) +6 E 1 g +6 E 1 u +6 E 2 g +6 E 2 u + +6 E [( n 1) = 2] g +6 E [( n 1) = 2] u Fromthoseonly7arenon-vanishingmodesthatareinfrareda ctiveand15that areRamanactive,buttheyarenotalldetectable.Itwasfoun dthatincreasing thediameterofthezigzagtubesthenumberofactivemodesdo esnotincrease. Thisconceptcanbeprovedforarmchairandchiraltubes,sin ceitisasymmetry imposedresult.Inchapter6areexplainedthemajorRamanli nesthatcanbe detected. 3.2ElectronicPropertiesofSWNTandMWNT 3.2.1ElectronicPropertiesofSWNT Theiruniqueelectronicpropertiesareattributedtothedi erentquantum connementofelectrons.Wecanseethreedierentdirectio nsthatbasedonthe geometryitwillresultin,ornotconnement.(i)Intheradi aldirection,electrons areconnedbythemono-layerthicknessofthegrapheneshee t.(ii)Aroundthe circumferenceofthenanotube,periodicboundaryconditio nscomeintoplay.As seenintheprevioussectiontheradius,thereforethebound aryconditions,depends onthe( n;m )conguration.Forexample,fora(5 ; 5)theradius, d t ,is6.78 A,for a(10 ; 0)itis7.83 A[115].Sothecircumferenceboundaryconditionsvaryeven fortubesthatareatthesamecategory(armchairorzigzag). (iii)Finallythe


30 p a p 2 a 0 p 2 a p a k x p a p 2 a 0 p 2 a p a k y 20 10 0 10 20 E H eV L 2 a 0 p 2 a p a k y Figure3{3:Thedispersionforgraphiteascalculatedfrome quation 3 10 directionparalleltotheaxis( T direction),sinceitisconsideredinnitethereisno connement. Becauseofthis1Dquantumconnement,theelectronscanonl ypropagate alongthenanotubeaxisdesignatedbythevector T ,andsotheirwavevectors k pointtowardsthisdirection.Theresultingnumberofone-d imensionalconduction andvalencebandseectivelydependsonthestandingwavest hataresetuparound thecircumferenceofthenanotube.Thesesimpleideascanbe usedtocalculatethe dispersionrelationsoftheone-dimensionalbands,whichl inkwavevectortoenergy, fromthewellknowndispersionrelationinagraphenesheet.


31 Inthesimplestmodel[113,116,117],theelectronicproper tiesofananotube derivedfromthedispersionrelationofagraphitesheetswi thwavevectors( k x ;k y ): E ( k x ;k y )= r 0 ( 1+4cos p 3 k x a 2 cos k y a 2 +4cos 2 k y a 2 ) 1 = 2 (3 10) where r 0 istheneighbor-hoppingparameter(usually r 0 =2 : 5 3 : 2eV,[113,116{ 118])and a isthelatticeconstant a =2 : 46 A.Figure 3{4 showstheplotofthis dispersionrelation. Whentheperiodicboundaryconditionsareimposedalongthe tubecircumference( C direction)the k =( k x ;k y )isquantizedalongthatdirection.Ithasto satisfythecondition k C =2 q ,whereqisaninteger.Forthearmchair( n;n )this translatesto k m x = m N x 2 p 3 a ( m =1 ;:::;N x )with N x =5(3 11) replacingthisvalueinequation 3 10 ,andsimplifying k y with k weget E arm m ( k )= r 0 1 4cos m 5 cos ka 2 +4cos 2 ka 2 1 = 2 (3 12) where

32 G X €3 €2 €1 0 1 2 3 kE( k )/ g 0A 1g + E 1g + E 2g + A 1g E 1g E 2g A 1u E 1u E 1u + A 1u + E 1u + E 2u + X G €3 €2 €1 0 1 2 3 kE( k )/ g 0A 1g + E 1g + E 2g + E 3g + E 4g + A 1g E 1g E 4g E 2g E 3u E 3g A 1u + E 1u + E 2u + E 3u + E 4u + A 1u E 1u E 4u E 2u (a)(b) Figure3{4:Thedispersionenergiesfor(a)armchairand(b) zigzagsemiconductor asarecalculatedfromequations 3 14 and 3 12 .Thedierent brancheshavebeenlabeledaccordingto[116]. directionboundaryconditioninordertohavemetallictube s; ( n m )=3 q (3 15) Thatmeansthatonethirdofthedierentnanotubesstructur esismetallicandtwo thirdsaresemiconducting.Figure 3{2 ,showstheconductivitystatesfordierent chiralities.Forsemiconductingtubestheband-gap( E g )is[119{121] E g =2 d CC r 0 d t (3 16) Sofarforthisapproachtheonlyweaknessisthatitdidnotac counttherehybridizationofthe orbitalduetothecurvature.Thiseectcanbeincluded inotherapproachessuchastherstprinciplecalculation ab-initio [122{125].In thisnewapproachitisprovedthatforsmalldiametertubes( < 1 : 5nm)aband


33 gapopensthatisabout0.02eVfornon-armchairnanotubes,t hatstillsatisesthe condition 3 15 [126].Howeverthisphenomenondissipatesfastforlargerd iameters tubes.Thereforethegraphitemodelcanbeusedasagoodappr oachtodescribe theSWNTwithdierentchiralities.STMstudieshaveconrm edtheaccuracy ofthemodel[123,126]andalsotheexistenceofthesmallban d-gappredictedby ab-initiocalculations[126]. IthasbeenexperimentallyconrmedthataSWNT[127],aSWNT rope [128]andaMWNT[129]behavelikeaquantumwireintrinsical ly.Theconductanceisgivenby = 0 M = 2 e 2 h M (3 17) where 0 =(2 e 2 =h )=(12 : 9kn) 1 isquantizedconductance. M isanapparent numberofconductingchannels,thatincludesallthepossib leinteractions,such uselectron-electroncoupling,inter-tubecouplingeect s.Forexamplefora SWNTthatvalueis2.InaSWNTtherearealsoimpurities,stru cturaldefects, couplingwiththesubstratethatwillfurtherreducethecon ductivity.Thereforethe experimentaldatahavelargevariationsfromthepredicted values,buttheyfollow thesametrend. Themostimportantinformationthatthegraphitemodelcanp redictisthe densityofstates(DOS)[130{132].Accordingtothatmodelt hedensityofstate ( )is ( )= 4 l 2 p 3 r 0 a + 1 X m = 1 g ( ; m )(3 18) where, g ( )= 8><>: j j p 2 2m for j j > j m j 0for j j < j m j (3 19) and j j = j 3 q n + m j r 0 a p 3 d t (3 20)


34 Calculationsbasedonthismodelpredictagainthatthearmc hairandzigzag congurationshaveacontinuousDOSwhileforthechiralasm allbandgapexists [119,133,134].Figure 3{2 showsthedierentdirectionsthatthegraphenesheets canberolledanditisdenotedifthetubeismetallicorsemic onducting. 3.2.2ElectronicpropertiesofMWNT Ithasalreadyhasmentionedinthepreviouschapterthatthe MWNTbehave asawirewiththeconductancetofollowthesimplerelation[ 129]; = 0 M = 2 e 2 h M (3 21) ForthecaseoftheMWNTthevalueof M issignicantlybiggerthanforthe SWNTtoaccountformoreconductingchannels.Inadditionth emultilayer structureincreasestheprobabilitytohavearmchairorzig zagtubesthatwill increasetheconductivity.Whilethediameterisincreasin gtheelectronsonthe tubearelessconnedandtheelectrondistributionresembl esmorethestructure ofgraphite.thisisduetothere-hybridizationofthe and orbital,thatisless intenseandthetubularstructureapproachesmorethegraph itestructure.Thisis obviousfromequation 3 17 wherewhilethetubediameterincreasestheenergy gapisdiminishingevenforthesemiconductingtubes.Soing eneralMWNTare intheirmajorityconductingandbehaveasnanowires.Butth erearestillchances thatthetubeswillbesemiconducting,dependingalwaysont hearrangementofthe tubescertaindefectsandcrystallinity. 3.3CarbonNanotubesGrowthMechanisms Therearetwobasiccommerciallyavailablemethodsforprod ucingcarbon nanotubes.ThearcdischargeandtheChemicalVaporDeposit ion(CVD).Both haveadvantagesanddisadvantagesthatcanbedirectlyrela tedtotheproperties ofthetubes.Generallyspeakingthetwomethodsarecompeti ngatthequantity


35 versusquality,whereCVDisdesignatedforquantityandarc dischargeisfor quality.3.3.1ArcDischarge Ingeneralcarbonnanotubesthatareproducedwithcarbonva porthat isbeingcreatedbythearcdischarge,havefewerdefectscom paredtoother techniques.Thereasonforthatisthehighgrowthprocesste mperaturethat ensuresperfectannealingthateliminatesmostofthedefec ts.TheMWNTthat areproduced via arcdischargeareperfectlystraight.Thefewerdefectshav ean immediatedramaticimpactonthetubepropertiessuchas,el ectricandmechanical. Oneofthemaindisadvantagesisthelimitedyieldthatthism ethodhas.Besides thelowyielditisahighlytimeconsumingprocess.Soingene ralifaahighyield ofnanotubesisrequiredthismethodisnotrecommended,ont hecontraryifmore dened,andbetterpropertiesisrequiredthenarcdischarg eisaverygoodsolution [135]. Themostcommonset-upforarc-dischargetwographiteelect rodesofdiameter 6-12mm,thatarekeptindistanceof1-4mminachamberthatis lledwithHelium.DCcurrentoperatesthetwoelectrodes.DCcurrentand Heliumarethetwo factorsthatimmediatelyinruencetheyield.Whiletheposi tiveelectrode(anode) isconsumedacylindricalslagisbeingdepositedonthecath ode.Thealignment oftheelectrodesdoesnoteecttheMWNTsbutcaneectthepr opertiesofthe singlewalltubes[135].3.3.2CVD:ThermalCVD,PE-CVD Sincetheapplicationeldofthenanotubesisgrowingthede mandforhigher yieldproductionmethodsisalsogrowing.Oneofthemostpro misingtechniques istheChemicalVaporDeposition(CVD).Ithasalargeknowle dgebasesinceitis beenusedextensivelyinelectronicapplicationsforthela stfewdecades.


36 ThenanotubesthatareCVDgrownhavealotofstructuraldefe ctsduetothe lowsynthesistemperatureduringthegrowthprocess.Anapp roachtoimprovethis isannealingthetubes,whichwillreducethedefectsbutinn ocasewillhavethe sameresultsastheArc-discharge[135]. TheapparatusforCVDgrownnanotubesissimple,whichisals oreducinga lotofthecostoftheproduction.Inaquartztubewithverypr ecisetemperature control,asubstrateisplacedincarboncontaininggases,s uchasCO,CH 4 or higherorderhydrocarbon,arerownin.Toassistthereactio noftenathermal sourceisused,suchusIRlamp(ThermalCVD)[135{137]orpla sma(PE-CVD) [138].Thegrowthratescanbecontroledpreciselyandcango fromafewnm/min upto5 m/min.Inadditionmetalcatalystcanfurtherassisttheyie ld.Oneofthe biggestadvantagesofCVDistheabilitytogrowonapatterne dsubstrate,whichis desirableformicroelectronicapplications.Thepuricat ionofthetubesinthiscase isanecessitysincetheycontainmetalcatalystanddieren tamorphouscarbon structures.Therearemanywaystopurifythetubes;hydroth ermaltreatment[139], H 2 O-plasmaoxidation[140],acidoxidation[141],dispersio nandseparationby micro-ltration[142]andhigh-performanceliquidchroma tography[143].


CHAPTER4 ANATASECOATEDCARBONNANOTUBES(ANTs):SYNTHESISAND CHARACTERIZATION) Intheprevioustwochaptersthemainpropertiesoftitaniaa ndthecarbon nanotubeswerereviewed.Thischapterdescribestheproces sofcombiningthose twomaterials.Therearemanypossiblecombinations,butin thisresearchthe objectiveistoapplythetitaniaintheformofathincoating onthesurfaceof theMWNTsinordertomaximizethecontactbetweenthetwomat erials.There arecertaindesignparametersthathavetobesatisedinord ertoobtainthe optimumresults.Therstsectionexplainsthoseparameter sandfollowingthat areexplainedthematerialsselectionandpreparation.Lat erasmallintroduction tothe Sol-Gel chemistryisgivenandbasedonthat,thechoiceofchemicals and precursorsisexplained.Finallyfundamentalcharacteriz ationwillfollowtoprovide argumentsforthesatisfactionornotofthedesignparamete rsandinwhatextend itwasachieved.Theactualphotocatalyticeciencyaswell asthedetailedstudy oftheinterfacebetweentheMWNTsandthetitaniawillbedis cussedinseparate chapterslatersincetheyarethemainfocusofthisresearch 4.1DesignParameters Asstatedintheintroductionthepurposeofthisworkistoco mbinethose materialsandtheirpropertiestoproduceahighlyecientp hotocatalyticparticle. Themainobjectiveistosynthesizeathincoatingoftitania tocoverthesurfaceof theMWNTs.Theprocesshastosatisfycertaincriteria.Thecoatinghastobetheanatasephaseoftitania: Asseeninprevious chapter2anataseisthemostphotocatalyticactivephaseof titania.That 37


38 phaseisalsothermallyveryunstableandthereforeobtaini nganataseisa non-trivialprocesswithmanyparameters. Thincoatingwillresultbetterphotocatalyticperformanc e: Thewhole photocatalyticprocesstakesplaceinathinlayerofabout1 0nm.Ifany electronholepairisgeneratedinregionsdeeperthanthat, itisgoingto recombinebeforeitreachesthesurface.Inadditionincrea singthecoating thicknesswillresultlightercolor(sincethecoatingwill belesstransparent) andthereforetheparticlewillabsorblesslight. ThecoatinghastobechemicallybondedtotheMWNTs: Ifthecoatingis notchemicallybondedonthesurfaceoftheMWNTsitispossib lethatitwill rakeo.Thecoatednanotubeswillhavehightendencytocoag ulatesince thesizeisbigenoughtoinducevanderWaalsforces.Therefo reprolonged sonicationwillberequiredtosuccessfullydispersethem, whichmightdamage looselyattachedcoating. IndividualMWNTshavetobecoated: MWNTshaveveryhighanityinto coagulating.Thehydrophobicnatureofthetubeswillalsoi ntensifythe phenomenonofcoagulationespeciallywhenthesolventiswa ter.Inorderto maximizethesurfaceareaitisrequiredtominimizethenumb erofMWNTs agglomeratesandseparatethebundles. Thenumberoffreetitaniaparticleshavetobekeptminimum: Sol-Gel isaprocessthatbalancesbetweentransportphenomenaandr eactionrate. Ideallyinordertoachievethecoatingtheprecursormolecu leshavetobe transportedtothesurfaceoftheMWNTsandonlyaftertheanc horingthey shouldreact.Thisbalancecanbecontrolledbyreactionpar ameterssuchas temperatureandpH.However,regardlessthevaluesofthose parametersthere isalwaysanitepossibilityoffreeanataseparticlesform ation.


39 Withthoserequirementsinmindtwodistinctsetofparticle swillbesynthesized.TherstonewillbeconsistofanarcdischargeMWN Tcoreandthe nextonewillconsistofaCVDgrownMWNTcore.Asdescribedin theprevious chapter(section 3.3 )thedierenceinthetubeproductioncanaecttheelectric al propertiesofthecarbonnanotubes.Sothepurposeofusingt hosetwodierent nanotubeswillbetoexaminetheeectoftheelectricalprop ertiesofthetubeson thephotocatalyticactivity.TheCVDcarbonnanotubeshave beenmechanically andchemicallyshortened,whichwillresultinadramaticin creaseofthedefects onthesurfaceofthetubes.Theshortnanotubesinadditionw illprovideother advantages.Thehighaspectratioofthecarbonnanotubesre sultsinaparticle thatinteractseasilywithmolecules,butraisesissueswhe nisitusedtodeactivate objectsofcomparablesizesuchassporesandbacteria.Bact eriahaveverycomplicatedsurfaces,thatusuallyhavebrilsofseveral mlengththatcaninterfereand preventthecoatedtubestoreachthesurface.Inadditionth esphericalshapeof thesporesdoesnotallowtheuseofthewholeavailablesurfa ceofthenanotubes. SoreducingthelengthoftheMWNTswillresultshorterinpar ticles.Largescale productionofshortnanotubes( d average < 1 m),cannotbeachievedwithneither arcdischargemethod,norwithCVD.Theyhavetobeshortened withchemicallyassistedmechanicalgrinding.TheshortMWNTswillbe occasionallycalled s -CNTsandthelongMWNTswillbecalled ` -CNTs. 4.2NanotubeSelectionandPreparation Thecarbonnanotubeshavetobeproperlymodiedtosatisfys omeofthe coatingrequirements.Theyhavetobeindividuallysuspend ed,easilydispersed insolventsandfavortheanchoringoftheprecursormolecul es.Itisalsocritical tocharacterizethetubesbeforethecoatingintermsofcrys tallinityandstructure,somethingthatcanbeusedtoexplaindierencesinter msoftheelectrical properties.


40 4.2.1MaterialsSelection Twodierentnanotubesweretestedasphotocatalytictempl ate.Thelong nanotubeswereorderedfromAlfa-Aecar(stocknumber:4288 6)insootform.The CVDnanotubeswereorderedfromNanoMat(productnumber:12 36YJS)and weredeliveredinpowderform.Accordingtothemanufacture rthetubeswere shortenedinaballmillinahighlyacidenvironment(nitric andsulfuricacid1:3). MWNTsfromothermanufacturer(IljinNanotech)weretried, butdidnotbehave desirablysotheywerenotused.Inadditionhighlyconducti veactivatedcarbon fromDegussawasused,againwithnodesirableresults.4.2.2PuricationandDispersion Thearcdischargenanotubeswereobtainedintheformofsoot .Inthesoot alongwithcarbonnanotubesthereweremanyotherformsofca rbonsuchas, carbonbers,fullerenesandamorphouscarbon.Similarlyi sthesituationforthe CVDgrownnanotubes.Inadditionthereisresiduefromtheca talyst(inthiscase Fe).Inordertocoatthemtheyhavetobepuriedanddisperse d.Sincemostof theimpuritiesarecarbonnaturetheycanbeeasilyoxidized byacid. Themainroutewasthesameforbothmaterials.Thetubeswere dispersedin highlyconcentratedHNO 3 (63%or10N).Thearcdischargenanotubeswereinsoot form,soinitiallythesootwasgroundwithmolderandpestle tonepowder.After that50mgofthispowderwasmixedin200mlofthenitricacid. Thesolutionwas sonicatedfor3hourstofurtherdispersethepowder.Thesol utionwasreruxed inanoilbathat140‰for10h.Thentheheatwasturnedofandthesolution wasleftforadditional3huntilthetemperaturedropsbelow 30‰.Thesolution wasthencentrifugedandtheexcessnitricacidwasremoved. Triplewashingwith di-ionizedwaterfollowed. TheCVDnanotubeswerealreadyinpowderformandthereforew asno needforgrinding.Inadditionsincetheywerealreadytreat edwithacidforthe


41 shorteningthereisnoneedforextensivepurication,buts tilltheacidtreatment isrequiredfordispersionpurposes.Aspreviously50mgoft ubesweredispersed in200mlofHNO 3 andsonicatedfor3h.Afterthatthetubeswerereruxedagain inoilbathof100‰for6handafterwardsthesolutionwascooleddownto30‰. Againthenitricacidwasremovedwithcentrifuge,andthetu beswerewashedwith ethanolthreetimes. Inallcasesthenanotubeswerenotremovedfromthesolvent. Duringthe puricationprocesstherewasa40%weightreduction.Sofor thecoatingprocess areleftabout30mg.Thisvaluewasestimated,bydryingandw eighingthe remainingnanotubes.4.2.3CharacterizationoftheFunctionalizedMWNTs ThecharacterizationofthetubeswasperformedwithSEM(FE G-SEMJEOL JSM-6335F),TEM(JEOLTEM2010F),FTIR(NicoletMAGNA760Be nch),Zeta Potentialmeasurements(BrookhavenZetaPlus),particles izing(CoulterMultisizer III)andthermalgravitationalanalysis(NetzschSTA449CJ upiter).TheSEM (gure 4{1 )revealsroughlythegeneralcharacteristicofthetubes.T he s -CNTs, gure 4{1 (b),appearmorepuresincetheyhaveundergonetheacidtrea tment twice,buttheyarenotstraight.Onthecontrarythe ` -CNTs,gure 4{1 (a), arestraight.Inbothcasesthetubesappeartobepureandthe rearenoobvious impurities.Aftertheacidtreatmentthetubesappearpuri edwithnoobvious impurities(attheorderof3nm)(gure 4{3 (d))andHR-TEMshowsthegraphene layersandthecapofthetubes.TheTEMimagesshowedanavera gediameterof about20nm. Figure 4{2 showstheTEMimagesof s -CNTbeforeandaftertheacidtreatment.Beforethetreatmentthetubesappeartangled(a)with manycarbon impuritiesontheirsurface(b).Theacidremovedmostofthe impuritiesandthe mainfeaturesofthetubessuchasthecavityarevisible.The purityofthetubes


42 (a) (b) Figure4{1:SEMpicturesofthetwotypesofnanotubes.(a)Th elongMWNTs (averagelengh1 m).(b)TheshortMWNTs(averagelengh100nm).


43 (a) (b) (c) (d) Figure4{2:TEMimagesofthe s -CNTs.(a)Agglomerateof s -CNTs.(b)High magnicationofuntreated s -CNTswheretheimpuritiesaroundthe tubearevisible.(c)Puried s -CNTswheretherearealmostno impuritiespresent.Itcanbeseenthattheyarenotstraight andthat theyhavebeendamanged.(d)Magnicationofthetreated s -CNTs wheretheinnercavityisvisibleandtheoutersurfaceisalm ost completelyfreefromimpurities.Fromtheimagesitcanalso be concludedthattheaveragediameteris20nm. isdemonstratedclearlyingure 4{2 (c)wherethetubesalthoughareshowntobe aggregatedtheyarefreeofimpurities.Itisalsoconcluded thatthetubeshavean


44 (a) (b) (c) (d) Figure4{3:TEMimagesofthe ` -CNTs.(a)Agglomerateof ` -CNTsbeforethe acidtreatment.Itisholdtogetherbythecarbonimpurities .(b)Single ` -CNTcoveredbythecarbonimpurities.(c)Aftertheacidtre atment abundleonnanotubes.Itisalsovisiblesomeresidueofthea cid treatmentbyproducts.(d) ` -CNTsafterthetreatment,wheremostof thesurfacecarbonimpuritieshavebeenremoved.Againfrom this imagewecanseethattheaverage ` -CNTdiameterisabout15nm. averagediameterof15nm,whichisinagreementwiththemanu facturerspecications(10-20nmwithaverage15nm).Similarresultscanbe derivedfromgure 4{3 forthe ` -CNTs.Sincetheyarearcdischarge(60%byweightMWNTs)the y havemoreimpuritiesthantheshort.Ingure 4{3 (a)theaggregateshavebigpieces


45 (a) (b) (c) (d) Figure4{4:Immediatecomparisonofthetwodierentkindso fnanotubes.The images(a)and(b)areforthe ` -CNTsandthe(c)and(d)forthe s -CNTs.Theendofthe s -CNTsisusuallyopenduetothecatalyst (c),whiletheendofthe ` -CNTsarecapped(a).Inadditionthe s -CNTshavedamagedandnotwelldenedwalls(d),whilethe ` -CNTsareverywelldenedandstraight. ofthecarbonimpuritiesandinacharacteristicpictureofa nindividualtube(gure 4{3 (b))showsthesurfacetobecoveredinsegmentsoftheamorph ouscarbon impurities.


46 12345678910 11 12 €100 €75 €50 €25 0 25 50 75 100 pHZeta Potential (mV) After the acid treatment Before the acid treatment (a) 123456789 €70 €60 €50 €40 €30 €20 €10 0 10 pHZeta Potential (mV) After the acid treatment Before the acid treatment (b) Figure4{5:Thezetapotentialforboththe ` -CNTs(a)and s -CNTs(b).Itshows theshiftoftheIEPforthe ` -CNTs(from7to3.5)andtheincreaseat thesurfacechargeforthe s -CNTs(from-10mVto-37mvforph4).


47 5001000150020002500300035004000 Wavelength (cm 1 )Reflactance (a.u.) COO CH COH Figure4{6:TheFTIRoftheMWNTsaftertheacidtreatment(on lythe s -CNTs resultsaredisplayed).Thebandsthathavebeenidentiedp rovethe reactionofthe COOHonthesurfaceofthenanotubes. Finallythedirectcomparisonofthenanotubesfocusesthem aindierenceon thestructureofthetubewalls.Inadditionthe s -CNTsappeartobeoccasionally openended,whilethe ` -CNTsareinthemajoritycapped( 4{4 (c)and(a)).The HR-TEMimages(gure 4{4 (b)and(d))showclearlywelldenedgraphenelayers forthe ` -CNTswhilethegraphenelayersforthe s -CNTsappeardamaged.Inall casesarevisiblesmalllayersofcarbonaceousimpuritieso nthesurfaceofthetubes thataredirectbyproductsoftheacidtreatment[144,145]. Althoughtheycanbe removeditisnotnecessarysinceitwillbedissolvewhenthe tubesareplacedina solvent(waterorethanol). Themeasurementoftheisoelectricpoint(IEP)andsurfacec hargeitis necessarytoclarifyifthereisanysurfacemodicationoft heCNTs.Theresults


48 0.01 0.1 1101001000 0 5 10 15 20 0 5 10 15 Diameter ( m m)Number (%) Differential Volume (%) As obtained (Number) Acid treated (Number) As obtained (Differential Volume) Acid treated (Differential Volume) Figure4{7:Thedierentialvolumeandnumberofthe s -CNTsbeforeandafterthe acidtreatment. forthe ` -CNTs(gure 4{5 (a))clearlyshowashifttolowervaluesoftheIEPand highersurfacecharge.Theresultsforthe s -CNTsshowthattherewaspre-existing surfacemodication,asresultofthemechanical-chemical shortening,andtherefore thesecondtreatmentjustincreasedtheamountofsurfacech arge.Inbothcasesthe changecanbeattributedtothegenerationoffunctionalgro upsonthesurfacesof theMWNTs. SincetheacidusedforthefunctionalizationwasHNO 3 thesurfacegroups thathavebeengeneratedonthesurfacehavetobe COO .DR-FTIRisutilized tofurtherinvestigatethesurfacegroupsonthesurfaceoft hecarbonnanotubes. Figure 4{6 showstheFTIRspectraofthe s -CNTs.Thebandsthatidentiedare verytypicalofthe COOHgroup(1170C OH,3450O Hand1720 COOH)


49 100200300400500600700800900 1000 0 10 20 30 40 50 60 70 80 90 100 €1 €0.5 0 0.5 1 1.5 2 2.5 3 Temperature ( o C)TGA (%) DTA ( m V) 6.02 % Impurities burn out Figure4{8:TheTGA/TDAdataofthe s -CNTs.Thepeakatthe600‰indicates theburningtemperatureoftheCNTs.Itisobservedabout6%o fthe initialmassresidue,whichistheFecatalyst. [146,147].Theotherbandsarecharacteristicofthecarbon nanotubes(1460 C H,1640C=C,2850C Hand2970C H)[148,149].Thebandat3450O H isnotproportionaltothe1170(C OH)and1720(COOH)butthisisduetothe atmospherichumidity.Similarresultsareobtainedforthe FTIRofthe ` -CNTsand theyareingoodagreementwiththeliterature[146{149]. ThenalcharacterizationwasdonebytheCoulterParticles izeanalyzer.The Coulterisutilizingalaserbeamandwithlightscatteringc alculatesthesizeof theparticles.ThetheorythatisusedattheCoulterinstrum entsissimilartothat usedattheZetaPlusthatwasusedforthemeasurementofthez etapotential.A majorassumptionisthattheparticlesaresphericalorcanb eassumedasspherical. Thisiscompletelywrongforthecaseofthecarbonnanotubes ,whicharehigh


50 aspectratioparticles(1:150forthe ` -CNTsand2:70forthe s -CNTs).In additionsincethelimitoftheinstrumentis40nmwhatisdet ectedaremainlythe agglomeratesandnottheindividualtubes.However,theins trumentcanbestill beusedtoshowcasechangeinthedispersionoftheCNTs.Duet othehighaspect ratioofthe ` -CNTstheresultscannotbeconsideredaccurate,andgure 4{7 showsonlythe s -CNTscase.Thedierentialvolumeresultsareusuallycons idered morerepresentativeandaccordingtothegraphthereisoneo rderofmagnitude reductioninthediameteraftertheacidtreatment.Inbothc ases(zetapotential measurementsandparticlemeasurements)theresultsareon lyusedforqualitative purposes. InadditionThermo-GravitationalAnalysis(TGA)showedth atthe ` -CNTs arestartingtoburnatapproximately700‰whilethe s -CNTsareburningat approximately600‰andtheyhave6%weightresiduethatwasidentiedasFe 2 O 3 whichcameforthecatalystusedduringtheproduction(gur e 4{8 ).Thatwasin agreementwithmanufacturerstatements. Sofromthissectionwecanconcludethatthetwotypesofcarb onnanotubes usedinthisresearcharedierentregardingtheoverallstr ucture.Althoughboth haveaconcentrictubestructureandthecharacteristiccav ityinthecenter,the twotypesaredierentinquality;the ` -CNTsareverystraightandhavevery welldenedstructure,whilethe s -CNTstypehasdamagedwallsasresultofthe productionmethodandthechemicalmechanicalshortening. Inadditiontheacid treatmentwasprovedenoughtoremovecarbonnatureimpurit iesandtocause surfacemodicationtostabilizethetubes,eitherbyincre asingthesurfacecharge ( s -CNTs)andbyshiftingtheIEP( ` -ANTs). 4.3 Sol-Gel Coating The Sol-Gel [150]routeisaverycommonandvalidatedwaytoproducethin coatingsofamorphousandcrystallinematerials.Fortheti taniathereisagreat


51 dealofattentiontothismethodsincethesizeoftheproduce dparticlescanbe veryaccuratelycontrolledandthereforenanosizedpartic lescanbeeasilyproduced withveryhighyieldandreproducibility[151,152].Sofort hisresearch Sol-Gel isthemostappropriatemethodforthegeneratinganataseti taniacoatingonthe MWNTs.Thissectionexplainsthematerialsselectionandde scribestheprocess thatwasfollowedtoobtaintheTiO 2 coating. 4.3.1PrecursorSelection Thereisarenumerousdierentmethodstoproduceanataseti tania via the Sol-Gel route.Theprecursorscanbeeitherorganometallicsorsalt s.Themolecules willundergoavarietyofreactionsthatwillresultathreed imensionalmolecular network.Acommonexampleisthehydrolysisandcondensatio nreactionsofmetal alkoxidestoformlargermetaloxidecrystals.Analkoxideh asanorganicgroup bondedtoanegativelychargedoxygenatom;whenthisoxygen isalsobondedtoa metalitiscalledmetalalkoxide.Duringthehydrolysis[15 3,154]allorsomeofthe organicchainsarereplacedbythe OHgroups. M(OR) n +H 2 O HO M(OR) n 1 +ROH+ ::: M(OH) n + n ROH(4 1) Duringcondensationreaction[153,154],theM(OH) n arereactingtoproducethe metaloxide. (HO) n 1 M OH+HO M(OH) n 1 (HO) n 1 M O M(OH) n 1 +H 2 O(4 2) Oralternativelythecondensationcanoccurfromtheinterm ediatesofthereaction 4 1 [150]. (RO) n 1 M OH+HO M(OR) n 1 (RO) n 1 M O M(OR) n 1 +H 2 O(4 3) (RO) n 1 M OH+RO M(OR) n 1 (RO) n 1 M O M(OR) n 1 +ROH(4 4)


52 whereMwithvalence n isthemetalandtheRaretheorganicchains.The reactionisprogressingwiththehydrolysisandthecondens ationofallthe OR groupsofthe(RO) n 1 M O M(OR) n 1 toresultinthethreedimensional network.Inthecaseoftitaniathisreactionwillproduceth eTiO 6 octahedral, whichisthestructuralelementoftheanataseandrutile. Oneofthefactorsthatcandeterminethereactionrateisthe lengthofthe organicchain.Usuallyincreaseinchainlengthwillresult inslowerreactionrate. Thechainlengthisdirectlyrelatedtothemobilityofthemo lecule.Inadditionthe threedimensionalstructureandcomplexityofthemolecule willalsoeectthereaction.Morecomplexstructuressuchastitaniumbis-ammoniu m-lactato-di-hydroxide (TALH)arelessreactive.Signicantdierencesinthereac tionhavebeenreported eveninthecaseoftitaniumisopropoxide(Ti 0B@ O CH < CH 3 CH 3 1CA 4 )[155{158] andtitaniumpropoxide(Ti( O C 3 H 7 ) 4 )[159,160]. Thereisalsothecaseofthesaltsthatcanbeusedsuchastita niumtetrachlorideTiCl 4 [161{164]andtitaniumsulphateTi 2 (SO 4 ) 3 [165{167].Titanium tetrachloridecanbedirectlyhydrolyzedtoyieldtherutil ephaseoftheTiO 2 TiCl 4 +H 2 O Ti(OH) 4 +4HCl(Endothermic)(4 5) Afterwardsthereactionprogressessimilarlytothereacti on 4 2 .Itcanalsobe usedfortheproductionofmetalalkoxidesthatlatercanbeh ydrolyzedtoproduce TiO 2 TiCl 4 +4ROH Ti(OR) 4 +4HCl(4 6) Thetitaniumsulfatehasmorecomplicatedstructureandthe reactionproceedsas Ti 2 (SO 4 ) 3 +8H 2 O 2Ti(OH) 4 +3H 2 SO 4 +H 2 (4 7)


53 (a)(b) (c)(d) (e)(f) Figure4{9:Thedierent Sol-Gel precursorsusedinthisresearch.(a)titanium ethoxide(Ti(OC 2 H 5 ) 4 ),(b)titaniumisoproxide(Ti(OC 3 H 7 ) 4 ),(c) titaniumbutoxide(Ti(OC 4 H 9 ) 4 ),(d)titanium bis-ammonium-lactato-dihydroxide([CH 3 CH(O )CO 2 NH 4 ] 2 Ti(OH) 2 (e)titaniumsulphate(Ti 2 (SO 4 ) 3 ,(f)titaniumtetrachloride. Inthisresearchtherewerevedierentprecursorsused;ti taniumethoxide (Ti( O C 2 H 5 ) 4 )titaniumispropoxide(Ti 0B@ O CH < CH 3 CH 3 1CA 4 ),titanium butoxide(Ti( O C 4 H 9 ) 4 ),TALH,((CH 3 CH(O )CO 2 =) 2 Ti(OH) 2 (NH 4 ) 2 )and titaniumsulphate(Ti 2 (SO 4 ) 3 )(gure 4{9 ).


54 SeveralprecursorsweretriedforeverycaseofMWNTs.Initi alconditionsand precursorwereselectedbasedontheliterature.Forthetit aniumsulfatefromLee etal.[168],fortitaniumisopropoxide,ethoxideandbutox idefromJitianuetal. [169]andnallyforTALHfromLeeetal.[170].Theresultswe rejudgedbasedon therepeatability,thecoverageofthecoatingandthenumbe roffreeparticles.The surfacecoverageandthefreeparticleformationwerecheck edwiththeTEM. 4.3.2CoatingModel Toestimatetheamountofanataserequiredtocoatthetubesa coatingmodel hastobedeveloped.Auniformcoatingofapproximately5nmw illgivethe optimumresults.Thenanotubeshaveadiameterof20nmandav eragelengthof2 m.Theoptimumcoatingwillbearound5nmthick.So V anatase =5nm2 10nm 2 m(4 8) =2 5 10 9 10 10 9 2 10 6 m 3 (4 9) =6 : 28 10 22 m 3 (4 10) Respectivelythevolumeofananotubeis V ` CNTs = (10nm) 2 2 m 0 : 6 10 21 m 3 (4 11) Theaveragedensityofthetubes( CNT )is1.1g/cm 3 .So1mgofMWNTswill contain 10 3 g= 1 : 1 g cm 3 0 : 6 10 21 m 3 2 10 12 .Soforeverymgof ` -CNTstherequiredvolumeof anataseis V total anatase =2 10 12 6 10 22 m 3 =12 10 10 m 3 =1 : 2 10 3 cm 3 ofanatase.Thedensityoftheanatase( anatase )is3.89g/cm 3 ,whichtranslatesto approximately4 : 67 10 4 gor0.467mgofanataseor5.84 10 6 molforeverymg ofCNTs.Inallcasesminoradjustmentswererequiredtomini mizetheformation ofthefreetitaniaparticles.Ingeneralthequantitythatw asusedwaslessthan theestimated.Themajordierencebetweenshortandlongtu besisthelength (whichdoesnoteectthecoatingmodel)andthediameter( R ` CNT >R s CNT )so


55 Table4{1:Thecalculatedinitialmolecularratioforthere actionsforthe ` -CNTs Precursors ` -CNTs(mg)SolventPrecursor( l)H 2 O( l) Ti(OC 2 H 5 ) 4 30mgEthanol300ml36.7(N/A)11.68 Ti(OC 3 H 7 ) 4 30mgEthanol300ml51.8(44.0)11.68 Ti(OC 4 H 9 ) 4 30mgEthanol300ml59.6(N/A)11.68 Ti 2 (SO 4 ) 3 1 30mgWater300ml102.5(106.0)N/A 1 Solutionof45%wtTi 2 (SO 4 ) 3 indilutesulfuricacid thesamemodelcanbeusedforbothtypesofMWNTswithsomemod ication.If m ` CNT anatase istheanataserequiredtocoat1mgof ` -CNTsthentheamountrequired for1mgof s -CNTsis m s CNT anatase = R ` CNT R s CNT m ` CNT anatase TheequivalentvolumeoftheMWNTcanbeconsideredasaspher eof radius R G CNT = l CNT = 2.Thevolumeiscalculatedtobe V G CNT = 4 3 R G CNT 3 = 4 : 2 10 18 m 3 =4 : 2 10 12 cm 3 .Sothetotalequivalentvolumeof1mgMWNTs occupyis V =4 : 2 10 12 cm 3 2 10 12 8.2ml.Thereforetoensurethatthe 30mgof ` -CNTs(252mltotalvolume)arenotincontactduringthecoat ingthe tubesaresuspendedin300mlofsolvent(waterof99.99%pure ethanol). 4.3.3LongMWNTs Basedonthecoatingmodelthetable 4{1 isconstructed.Thosevaluesarethe startingvaluesforthe Sol-Gel chemistry.Inparenthesisarelistedthequantities thatareeventuallyprovedtohavethebestresults(basedon surfacecoverageand numberoffreeparticles).Afterthenalwashingthe ` -CNTssuspension(30mg of ` -CNTsin300mlofwater)wereplacedinathreeway300mlrask. Therask wasplacedinanoilbathat40‰andwasreruxedunderconstantstirringspeed. AfterthetemperaturewasstabilizedthepHwasxedat 3with0.1NHNO 3 .The precursor(Ti 2 S(O 4 ) 3 )wasinjectedandthereactionwascarriedfor1hour.The solutionwasdividedintosix50mlcentrifugetubesandwasw ashed3times.The compositewasthenlettodryat40‰fortwodays.


56 Grindingofthe soot Sonicationin 200mlHNO 3 (10N)for3h Acidtreatment at140‰in10N HNO 3 for10h Triplewashwith d.i.water Reruxedat40‰for1h Dispersingin300ml ofd.i.water Additionof precursorsolution Ti 2 (SO 4 ) 3 pHat3 Washingwith d.i.water Dryingat40‰fortwodays Figure4{10:Schematicdiagramoftheprocessforthecoatin gofthe ` -CNTs. TheexperimentwasrepeatedwiththeTi(OC 3 H 9 ) 4 .Forthiscasethenanotubesafterthefunctionalizationwerewashedwithethano l.Thenalsolution (30mgoftubesand250mlofethanol)placedagaininaraskand reruxedat 40‰underconstantstirringuntilthetemperaturewasstabiliz edat40‰.The appropriateamountofwaterwasaddedandthepHwasxedat 3with0.1N HNO 3 .Theisopropoxidewasplacedinanotherbeakerwith50mlofe thanoland wasstirredfor10mins.Thiswasdonetodissolveitsoitwill belessviscousand


57 Table4{2:Thecalculatedinitialmolecularratioforthere actionsregardingthe shortnanotubes Precursors ` -CNTs(mg)SolventPrecursor( l)H 2 O( l) Ti(OC 2 H 5 ) 4 30mgEthanol300ml48.9(40)11.68 Ti(OC 3 H 7 ) 4 30mgEthanol300ml69.1(58)11.68 Ti(OC 4 H 9 ) 4 30mgEthanol300ml79.4(62)11.68 Ti 2 (SO 4 ) 3 1 30mgWater300ml136.7(140)N/A 1 Solutionof45%wtTi 2 (SO 4 ) 3 indilutesulfuricacid lessreactive.Thenitwasslowlyinjectedintotherasktore actfor30min.The processfollowsasbefore,triplewashinganddrying.Thesa meexperimentwas repeatedagainundernitrogenatmosphere.AfterthepHwas xedaspreviously beforenitrogenwaslettorowinthecontainerfor1h(50cc/m in)andthenthe isopropoxidesolutionwasinjected.Againthereactionwas carriedoutfor30min. Thenthesamewashinganddryingstepsfollowed.Thenitroge natmospheredid notsignicantlyaectedthereactionresults. TheTGAandXRD(gure 4{16 )analysisshowedthatheattreatmentat 500‰withrampingrate10K/minwillcompletelytransformtheTiO 2 toanatase. Thetitaniumethoxideandtitaniumbutoxidefailedcomplet elytoachieve coatinginvariousconditionsandthereforetheywerenotus ed,althoughthereis areportofsuccessfullyusingthemtocoatMWNTs[169].TheT ALHwasalso used,byfollowingthemethodbeLeeetal.[168]butthenalr esultgavestrongly agglomeratedparticles. Fromthispartitisconcludedthatamongalltheprecursorst hemostappropriateforthe ` -CNTsismainlytheTi 2 (SO 4 ) 3 .Thetitaniumisopropoxidealthough italsoyieldgoodresults,itwasnotconsistent.Fromthisp ointonwardsascoated ` -CNTswillbeconsideredthetubesthathavebeencoatedwith Ti 2 (SO 4 ) 3 as precursor( ` -ANTs).Figure 4{10 summarizesthecoatingprocess.


58 4.3.4ShortMWNTs Forthe s -CNTsatablesimilartothe ` -CNTscaseisconstructed(table 4{2 ). Thesynthesisproceduresforeveryprecursorareidentical totheprevioussoare notgoingtobedescribedagain.TheonlydierenceisthepHt hatwasxedat approximately4.Againtheoptimumconditionsforthecryst allizationwerefound tobeat500‰for3hwithrampingtemperatureof10K/min(gures 4{16 and 4{12 ). Onthecontrarytotheprevioussectionandthe ` -CNTstheprecursorthat showsthebestresultsarethemetalalkoxides.Thereisnota standalonereasonfor that,butprobablyitisrelatedtothedierentisoelectric points.TheTALHwas notusedforthe s -CNTs. Amongthemetalalkoxidesthetitaniumisopropoxidedispla yedthemost stableperformance(consistency,repeatability)andbest result(numberoffree particles).Thetitaniumethoxidewassuccessfulbutitsho wedhighsensitivity tothepH,withsharptransitionsfromcoatedtouncoatednan otubes.Onthe contrarytheisopropoxideandbutoxideweremorestableinr egardstothepH. Butoxide,however,hashighviscosityandslowerreactionr atesinceithasalonger organicchain.Thereforetheisopropoxidewaspreferredfo rtheshortnanotubes. Overallthecoatingoftheshortnanotubesseemedtobeeasie randmorestable, sincethesurfaceofanindividualtubewassignicantlysma llerthanthesurface ofthe ` -CNTs.Additionaladvantagetothiswasthesurfacecharget hatforthe caseofthe s -CNTsitwashigher(greaterabsolutevalueofthezetapoten tial) fortheselectedpHvalue.BothkindofMWNTsshowedbetterdi spersioninthe ethanolcomparedtowaterandsincethefortheorganicprecu rsorstheethanolwas preferredassolvent,inthecaseofthe s -CNTsareexpectedlesscoatedbundles.


59 Sonicationin 200mlHNO 3 (10N)for3h Acidtreatment at100‰in10N HNO 3 for6h Triplewashwith ethanol Reruxedat40‰for1h Dispersingin250mlof absoluteethanol Additionof precursorsolution: Ti(OC 3 H 7 ) 4 in50ml pHat4 Washingwith water Dryingat50‰fortwodays Figure4{11:Schematicdiagramoftheprocessforthecoatin gofthe s -CNTs. 4.4CharacterizationoftheComposites Theverybasiccharacterizationofthecompositematerialw asdonewithXPS (KRATOSXSAM800),TEM(JEOLTEM2010F),TGA/DTAandXRD(XRDPhilipsAPD3720).TheXRDandTGA/DTAwilldeterminethecry stalstructure andtherequiredtimefortheheattreatment.XRDwillalsoyi eldinformation forthegrainsizeviatheScherrerequation.Thisresultisi mportantnotonlyfor thephotocatalysis,butfortheinterpretationofcertains pectrasuchasXPSand Raman.SincetheXRDhaslowdetectionlimit,inordertodete rminethecrystal structurejustparticlesweresynthesizedfollowingthesa meprocessastheone thatthecoatingproduced(gures 4{10 and 4{11 ).Figure 4{16 (I)showsthe


60 050 100150200250300350400450500 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Temperature ( o C)TGA (mg) DTA ( m V) Figure4{12:TheTGA/TDAdataofthe s -ANTs.Thepeakatthe100‰isfrom thewaterevaporationandthereforeitisaccommodatedbyam ass reduction.Atapproximately250‰thephasetransitionisstarting andcarriesonuntilthe500‰. resultsoftheXRDofthecoatedtubesandgure 4{16 (II)showstheresultsofthe synthesizedparticles. TheTEMwillconrmthecoatinguniformityandquality.TheB ETwilldeterminethespecicsurfaceareaofthematerial(m 2 /g).Thisiscriticalsincehigher surfaceareameansmoreecientphotocatalysis.Thisvalue willbenecessaryfor thephotocatalyticdegradationteststhatwillbepreforme donthesamespecic surfaceareabase. InadditiontheXPSsurveywillshowthecompositionofthema terial.The detailedanalysisofthepeakswillbediscussedinaseparat echapter.


61 (a) (b) Figure4{13:TEMimagesofthecoated s -CNTs.(a)Thecoatingisapproximately 6nmthickwithveryintensevariation.(b)Therearecasesth atthere arebigparticlesnucleatedonthesurfaceofthenanotubes. Thatisin agreementwiththeBETresultsthatshowedspecicsurfacea reaof 183m 2 /g. 4.4.1ShortANTs:TEM,XPS,BET TheTEMoftheshortnanotubesrevealedacoatingwithlargev ariationinthe thicknessrangingfor3to10nmwithaveragevalueof6nm(gu re 4{13 (a)).It alsoobservedthattherewerespotsthatthecoatingwasnotc ompleteandthere wereuncoatedregionsonthesurfaceofthenanotubes.Figur e 4{17 showstheXPS resultsthatconrmthetitaniacoating.InadditiontheXRD (gure 4{16 (I),line (b))conrmstheanatasephase,whilethereisnoindication theofrutilephase. TheScherrer'sformulawillbeusedtoestimatethegrainsiz e[171] d grain = K B cos (4 12) where istheBragg'sangle, thewavelength(1.54 A), K isaconstant( K = 2 ln2 1 2 =0 : 93)[171{173],Bisthehalfvaluebreadthofthemostintense peak. Thegrainsizeaccordingtothiscalculationis5nm(gure 4{16 ,line(b)).Thisis inagreementwiththeTEMresults.Thesamecalculationsfor the(gure 4{16 (II),


62 (a) (b) Figure4{14:TEMimagesofthecoated ` -CNTs.(a)TheTEMimagesreveal coatingofapproximately4nmanditisuniform.(b)Thereare cases thattherearebigparticlesnucleatedonthesurfaceofthen anotubes. ThatisinagreementwiththeBETresultsthatshowedspecicsurfaceareaof172m 2 /g. line(b))showed53nmaveragegrainsize.Althoughtheywere producedunderthe sameconditionstheyhavedierentgrainsizesthatattribu tedtothepresenceof thenanotubes. FinallytheBETrevealedasurfaceareof183m 2 /g.Thehighsurfacearea isduetotheneedlelikeshapeofthenanotubesandtheroughs urfacethatthe Sol-Gel chemistrygenerated. 4.4.2LongANTs:TEM,XPS,BET TheTEMshowedaveryuniformcoating 4{14 (a)ofapproximately4nm thick.Incontrastwiththe s -ANTsthecoatingisveryuniformandhasverysmall variance(3to5nm).Againtherewerecasesofpartiallycoat edtubes,butless comparedtothe s -ANTs.AgaintheXPSconrmedtheelementsofTi,OandC. Basedontheelementalconcentrationstheamountoftitania isabout12%.This valuehoweverisnotconsideredaccuratesincetheXPSisver ysensitivetothe thicknessofthelayers.Theelectronsdependingontheiren ergycantravelonly


63 25102050 100200500 10002000 3 5 10 20 50 100 200 Kinetic energy (eV)Mean free path l ( A )Ag Al Au Be C Fe Ge Mo Ni Se Figure4{15:Theuniversalcurveoftheelectrons,basedont hecalculationsbyM. P.SeahandW.A.Dench[174].Thecurveshowsthemeanfreepat h oftheelectronsasfunctionofthekineticenergy(dashedli nes).There arealsoexperimentalresultsthatfollowthesametrend.Th emean freepathdoesnotdependonthematerial.ForMgsourcetheXRay energyis1253.6eV,whichgiveameanfreepathofapproximat ely10 A. acertaindistanceinthematerial,regardlesswhatthemate rialis(gure 4{15 ). Thedetectedelectronsarecomingforonlythefewtopnm[174 ].The s -ANTs havethickercoatingandthereforetheelementalanalysisi snotrepresentative composition.TheXRDconrmedtheanatase(gure 4{16 (II),line(a)).According totheScherrerformula(equation 4 12 )thegrainsizeis5nm(gure 4{16 (I),line (a)).ThisisslightlycontrastingtheTEMresultthatwas4n m.Thisisattributed tothefactthatthesignalofthecarbonnanotubesoverpower edthesignalof titaniaandthereforethecalculationisnotconsideredexa ctbutjustarough estimate.ThegrainsizethatwascalculatedbasedontheXRD patternfromgure 4{16 (II)(justthesynthesizedparticles),line(a)is23nm.


64 FinallytheBETgaveasurfaceareaof172m 2 /g.Thisisinagreementwith theexpectationsbasedontheTEMimagesandtherespectiver esultforthe s ANTs.Thelowervalueofthesurfaceareaisattributedtothe smoothersurface thattheTi 2 (SO 4 ) 3 yielded.Incaseofthe ` -ANTstherearelessfreeparticlesas resultofthecoatingprocess(gure 4{10 ).


65 202530354045505560657075 2 q (degrees)Counts (a.u.)(a) (b) (101) (100) (101) (103) (004) (112) (200) (111) (210) (200) (211) (105) (211) (220) (213) (204) (002) (220) (221) (116) (220) (112) (301) (320) (107) (311) (I) 202530354045505560657075 2 q (degrees)Counts (a.u.)(a) (b) (101) (100) (101) (103) (004) (112) (200) (111) (210) (200) (211) (105) (211) (220) (213) (204) (002) (220) (221) (116) (220) (112) (301) (320) (107) (II) Figure4{16:XRDpatternswithandwithoutthecoating.(I)X RDpatternsofthe nanotubeswiththecoating.(II)TheXRDpatternoftheparti cles preparedbythesame Sol-Gel methodasthecoatingonthe nanotubes.(a)Titaniumsulfate( ` -ANTs)and(b)titanium isopropoxide( s -ANTs).Thesolidlinesdenotethepeaksforanatase (blackline)andrutile(lightgray)withtherelativeinten sities.


66 0 100 200 300 400 500 600 700 800 900 1000 Binding Energy (eV)N(E)C 1s 31.2% Ti 2p 16.7% O 1s 52.0% Ev/step:0.5 eV, Time/step: 30 ms, Sweeps: 10Source: Mg, Pass Energy:89.45 eV, Work Function: 4.36 eV Ti 3p Na KVV Ti 2s O KVV Ti LVV C KVV Figure4{17:XPSsurveyforthe s -ANTs.ThereisasignicantamountofTiO 2 (16.7%Ti).Thereisnodirectstoichiometry withtheoxygen(52%O)sincetheoxygendependsontheexpose dcrystallographicorientation.


67 0 100 200 300 400 500 600 700 800 900 1000 Binding Energy (eV)N(E)C 1s 91.0% Ti 2p 1.2% O 1s 5.8% Ev/step:0.5 eV, Time/step: 30 ms, Sweeps: 10Source: Mg, Pass Energy:89.45 eV, Work Function: 4.36 eV Ti 3p Na KVV Ti 2s O KVV Ti LVV C KVV Si 2p 2.0% Figure4{18:XPSsurveyforthe ` -ANTs.ThereisasignicantamountofTiO 2 (1.2%Ti).Againthereisnostoichiometry withtheoxygen(5.8%O).ThereislessTiO 2 comparedtothe s -ANTs.


CHAPTER5 PHOTOCATALYTICEVALUATIONOFTHESYNTHESIZEDPARTICLES WITHDYEDEGRADATIONTESTS Thischapterdescribestheseriesofexperimentsthatwerep erformedto evaluatethephotocatalyticeciencyofthesynthesizedpa rticles.Themethod usedforthispurposeisdyedegradation,whereadyeisbeing photocatalyitcally degradedanditsconcentrationisbeingmonitoredasfuncti onoftime[175{179]. Thistechniquewasselectedoverthebiocidaltestssinceit isfast,accurateand dependsprimarilyonthetypeandpropertiesofparticlesan dnotonparticle interactions.Othermethodsthatcouldhavebeenused,such assporeorbacteria inactivation,havemany,notfullycontrolled,variablest hatcanaltertheresults [180,181]. Inthecaseofthebiocidaltestthelengthoftheparticlesis comparabletothe diameterofthetargetbacteriaorspores.Thiswillaectth ekineticsofthesystem andtheinteractionbetweentheparticlesandthebacteriab yinducingstericforces andoccasionallyelectrostaticeects.Inaddition,thete mperatureandthepH thatcanvarysignicantlyduringtheexperimentscandrama ticallyaectthe behaviorofthesporesorbacteria.Especiallyforthespore s,temperatureincrease willtriggergerminationthatwilltransformthemintobact eria,makingthemmore vulnerabletothephotocatalyticdestruction.Biocidalte stsarealsotimeconsuming andrequireahighlyspecializedlab.Soalthoughthepartic lehasbeendesigned primarilyforbiologicalapplications,thebiocidaltests arenotanaccurateway tomeasureandcomparetheproperties.Thusthedyedegradat iontestwasused asaquickwaytovalidatethephotocatalyticpropertiesoft heparticles,which 68


69 Figure5{1:Schematicdiagramshowingthebasicelementsof thephotocatalytic degradationchamber. aredirectlyrelatedtothestructureandtheelectronicpro pertiesofthedierent particles. Inthefollowingsections,theexperimentalsetupisdescri bed,followedbythe theoryofthedyedegradationandtheparametersthatcaninr uencetheresults. Subsequentlytheexperimentalresultsandnallysomegene ralconclusionsare derived. 5.1ExperimentalSetup,MaterialsandProcedures 5.1.1ExperimentalSetup Figure 5{1 showsasketchoftheexperimentalsetup(photocatalyticre action chamber).Thewholestructureconsistsofalight-insulati ngchamberwherethe interiorisblacktoabsorbanyscatteredradiation.Atthet opofthechamber isa5Wfantomaintainthetemperaturebelow30‰.Insidethechamberarea magneticstirrerofvariablespeedandfourUVlampsarrange doverthestirrer (gure 5{1 ).Dependingonthetestdierentlampshavebeenused:


70 UV350nm fourruorescencelampsof350nmpeakwavelengthand8Wpower eachthatinthecurrentcongurationgave20W/m 2 UV305nm fourruorescencelampsof350nmpeakwavelengthand8Wpower eachthatinthecurrentcongurationgave20W/m 2 Visiblelight twohalogenlampsoflightradiationand100Wpowereachthat in thecurrentcongurationgave50W/m 2 thathavebuilt-inUVlter. Forallthedierentlampstheintensitywasmonitoredasfun ctionoftime. Itwasfoundthattheintensityincreaseswithtimeforther st30min.After thistimehaselapsedtheintensityisstabilizedatthepowe routputgivenabove. Thusthelampsarealwaysgivenaheadstartofminimum30minb eforethe experimentstarts.Underthoseconditionsatestwithwater demonstratedthat thetemperatureismaintainedalmostconstantatapproxima tely25‰with1to2 degreesvariationinonehour.Temperatureisalsoafactort hatcaninruencethe results,butnotinasignicantmanner.5.1.2DyeSelection Intheliteraturetherearemanytypesofdyesusedforthisap plication.For thepresentexperimentstheBrilliantProcionRedMX-5B(C 19 H 13 Cl 2 N 6 Na 2 O 7 S 2 ) wasused[176,182].Thecolorofthedyeismagentaandabsorb sstronglyinthe 510-540nm(Figure 5{3 ).Figure 5{2 showsthemolecularstructureofthedye. Thepresenceofthethreebenzeneandones-triazineringsma kesthedyemore resistanttodegradationcomparedtootherdyeswithfewerr ings,evenforlow concentrations[183].Thisisverycriticalsincefastdegr adationmeansthatthe systemwillnotbefullystabilized(pH,temperature)befor ethedegradationisover. Veryslowdegradationhoweverwillgivesucienttimeforwa terevaporationthat willalterthedyeconcentration.Anadditionaladvantagei stheexistenceofboth negatively(SO 2 4 )andpositively(Na + ,NH +4 )chargedchemicalgroupsthatwill inducedadsorptiononpositivelyandnegativelysitesresp ectively.


71 Figure5{2:Three-dimensionalstructureoftheBrilliantP rocionRedMX-5 molecule.Asitcanbeseenitcontains3benzenegroupsandab enzene groupwiththreecarbonatomsreplacedbynitrogenatoms(striazine). BrilliantProcionRedMX-5Bisoneofthedyesthathasbeenex tensively studiedandthedegradationbyproductsareknown[176,185] .Howeverinthis researchthereisnoneedtostudythedyeinthisextendsince allthenecessary informationisavailablefromtheliterature[176].Table 5{1 showsthedierent intermediatesofthereactionintheordertheyappearinthe solutionduring degradation.Thephotocatalyticreactionproceedsinthre esteps.Intherststep themostactivebondsarehydroxylated.Thosebondsinclude theC Nbondlinked tothebenzeneringorthenaphthaleneringandtheC Sbondofsulfonategroup linkedtothenaphthaleneringorthebenzenering,toformor ganicacidswithor withouthydroxylgroupsandtherelatedions(SO 2 4 andNH +4 ).Inthesecondstep, thegroupslinkedtothetriazineringarereplacedbyhydrox yltoyieldcyanuric acid,asinthecaseofthes-triazinesherbicides,andthere latedions(SO 3 ,Cl ). Atthesametimethearomaticacidsproducedfromtherstste psubsequently hydroxylatedandledtothecleavageofaromaticringstofro maliphaticgroups.


72 400450500550600 Wavelength (nm)Absorption (a.u.) Figure5{3:Theabsorptionspectrumfora5ppmsolutionofth eProcionRed MX-5Bdye. Thethirdstepinvolvesafurtheroxidationofthealiphatic acidstoproduceCO 2 andwater.Thosestepsaresummarizedintable 5{1 andgure 5{4 representsa visualizationofthedegradation.5.1.3ExperimentalProcedure Initiallyamixtureofdyesolutionandtheparticlesthatar ebeingevaluated aresonicatedfor20mins.Followingthesonicationthepart iclesareplacedin thedarkchamber(gure 5{1 ).WhilethesolutionisexposedtoUVlight,three samplesareobtainedeverycertaintimeintervals,in1.5ml cuvettes.Thecuvettes wereleftfor2daysfortheparticlestosettle.Thedyeconce ntrationwasmeasured via UV-VISspectroscopyandthereactionconstantwasestimate dbasedonthe Langmuir-Hinshelwoodtheory.Sincetheparticlestestedh erearenanosized,even after2daystherewillstillbesuspendedparticles.Thosep articlescanscatteror


73 p -Hydroxy-phenyl-3-3-Hydroxy-2-Hydroxy-benzoicacid -hydroxy-propanedioicacid-benzeneaceticacid p -Hydroxy-cinnamicacid1,2-Benzenedi-cyanuricacid carboxylicacid 1-Propene-1,2,3-PropanedioicacidPropanoicacid -tricarboxylicacid MalicacidButenedioicacidOxalicacid Figure5{4:Thestructureofseveralintermediateproducts ofthephotocatalytic reactionthatshowthedestructionofthebondsandthesizer eduction ofthemolecules.


74 Table5{1:Theoxidationintermediatesandtheirstructure tobecomparedtothe initialdyestructureingure 5{2 .Adaptedfromreference[184]. Step Photo-oxidationintermediates Step-1 p -Hydroxy-phenyl-3-hydroxy-propanedioicacid 3-Hydroxy-benzeneaceticacid 2-Hydroxy-benzoicacid p -Hydroxy-cinnamicacid 1,2-Benzenedicarboxylicacid Step-2 Cyanuricacid 1-Propene-1,2,3-tricarboxylicacid Propanedioicacid Propanoicacid Malicacid Butenedioicacid Oxalicacid Aceticacid Step-3 AliphaticcompoundstoCO 2 andH 2 O minerals(S,Na) absorbthelight,whichwillaltertheobtainedspectrum.So foreveryexperiment awatersolutionwithparticleconcentrationequaltotheon goingexperimentis prepared.Thissolutionisalsoleftfor2daysandtheobtain edspectrumisusedas background. 5.2TheoryforthePhotocatalyticDegradationofDyes Mostexperimentalresultsagreethattherateofphotocatal yticoxidationof dyescanbeapproximatedwiththeLangmuir-Hinshelwood(LH)model[175{ 178,180{185].Themodelassumesthattheratewilldependon theadsorption ofthedyemoleculeontheTiO 2 particleandtheoxidationreaction.Soifitis assumedthat k ,isthereactionconstantand K theadsorptionconstantthen accordingtotheL-Hkineticsmodeltheoxidationrateis: r = dC dt = kKC 1+ KC (5 1)


75 020406080100120140 0.0 0.2 0.4 0.6 0.8 1.0 €0.010 €0.008 €0.006 €0.004 €0.002 0.000 Time (min)C/C 0r=d(C/C 0 )/dt (a) 0510152025303540 0.0 0.2 0.4 0.6 0.8 1.0 €0.1 €0.08 €0.06 €0.04 €0.02 0 Time (min)C/C 0r=d(C/C 0 )/dt (b) Figure5{5:Comparisonbetweenthenumericalsolutionofth e Langmuir-Hinshelwood(equation 5 1 )andtheapproximation.The redlinesrepresenttheapproximationandtheblackisthenu merical solution.Thesolidlinerepresentsthedyeconcentrationw hilethe dashedrepresentsreactionrate d dt C C 0 .Figure(a)isforlarge concentrations( k =0.1, K =1, C 0 =10)andgure(b)isforsmall concentrations( k =0.1, K =1, C 0 =0.1).


76 Inequation 5 1 k and r ismg/lminand K isinl/mgwhere C isthedyeconcentrationinmg/l.Thismodelisnonlinearbutitcanbefurther simplied: 1 KC 0 ln C C 0 + C C 0 1 = kt C 0 (5 2) where C 0 istheinitialdyeconcentration.Withtheassumptionthat C 0 0,then 1 KC 0 ln C C 0 C C 0 1 andequation 5 2 simpliesto: ln C C 0 = Kkt (5 3) whichyieldsasimpleexponentialdecay: C ( t )= C 0 e kKt (5 4) C ( t )= C 0 e k app: t (5 5) C ( t )= C 0 e t= (5 6) Figure 5{5 showsacomparisonoftheapproachfortwodierentdyeconce ntrations.Itisapparentthatinthecaseofthelowconcentratio n(gure 5{5 (a))the agreementbetweentheexponentialapproachandtheexactnu mericalsolutionis verygood,whileforthecaseofthehighconcentrationthedi erenceissignicant. IthastobeunderlinedthatintheLangmuir-Hinshelwoodmod elisassumedfor singlereaction( AB *) A + B ),whichisnottrueforthecaseofthedyedegradation.Asdescribedbeforeforthiscertaindyethereareal otmorereactions involvedduringthedegradation.Inthiscaseitisjustassu medthatthe k refersto theslowestreaction. 5.3ParametersthatInruencethePhotocatalyticReaction Therearemanyparametersthatcanaectthereactionrate.T hemajorparametersarethepH,theinitialdyeconcentration,thesoli dsloadingandradiation


77 intensity.Therealsootherparameterssuchstirringspeed andtemperaturewith minoreectatthereactionrate.5.3.1pH ThepHisoneofthemostimportantparametersthatinruencep hotocatalytic reactions.ThepHcanimpactboththeparticlesstabilizati onandtheactual reaction[180,181].Dependingontheisoelectricpointthe particleswillinduce coagulationthatwillsignicantlyreducethesurfacearea oftheparticles.For titaniatheisoelectricpointrangesfrom5to7.Thereforef orpHvaluesbetween 5.0and7.0thephotocatalyticreactionratewillbereduced .ForpHvalues > 7and < 5thecolloidalstabilityisoptimum.Inadditionthesurfac echargeimpactsthe waythedyeadsorbsonthetitaniaparticles.Thisisespecia llyimportantforthe caseofazodyes,suchastheoneusedhere,sincethehavemany polargroups.The chargedmolecules(positivelychargedSandNaatoms)canbe adsorbedwellonthe surfacewithnegativecharge(inthecaseoftitaniameanspH > 7). ThepHcandirectlyaectthereaction.AhighpHwillincreas etheamountof OH ,andviseversa.Inthisreactiontherearethreestepswithm ultiplereactions withineachstep.SlightvariationsofthepHcanhaveasigni cantimpactonsome ofthereactionsthatwillimmediatelyeecttheoverallrea ction.Itisobvious thatthereisnotaspecictrendforthepH,sinceitdependso nthedyeandits byproducts.Soetal.howeverhaveinvestigatedthepHeect oftheProcionRed MX-5B,andtheresultsareingure 5{6 (a)[186].Thereisapproximatelya40% variationatthereactionratewhenthepHincreasesfrom2to 10. 5.3.2InitialDyeConcentration Asitwasalreadydiscussedsmallerconcentrationsaremore suitableforthe rstorderdecaysinceitapproachesmorethesimpleexponen tial.Howeverthere isamorephysicaldependenceofthereactionratetothedyec oncentration.While theinitialdyeconcentrationincreasesitwillincreaseth eprobabilityofadye


78 246810Reaction Rate (a.u.)pH 010203040Reaction Rate (a.u.)C 0 (ppm) (a) (b) 010203040Reaction Rate (a.u.)Light Intensity (W/m 2 ) r= aI r= b I 1/2 0 Rate (a.u.)f (wt%) 1 m m 100 nm 10 m m Radius increase (c) (d) Figure5{6:Themainparametersthatinruencetheoxidation rate.(a)pH variation,obtainedfromreference[180],fortheBrillian tProcionRed MX-5B(b)asfunctionoftheinitialdyeconcentration(c)as function ofthelightintensity(d)asfunctionofthesurfacearea(da ta calculatedforDegussaP25). moleculeadsorbingonthesurfaceandconsequentlyleading tophotocatalytic degradation.Thusthereactionratewillincrease.However ,ifthedyeconcentrationincreasesfurtherthesolutionwillbecomedarkerresu ltingUVshieldingand thereforetheratewilldecrease.Theincreasingofthedyec oncentration,willalso increasetheamountofadsorbeddyemoleculesonthesurface oftheparticles, whichwillreducetheavailableOH sitesandthereforereducethe[OH ]generation.Soinitiallythereactionrateisincreasing(gure 5{6 (b))almostlinearly, untilitreachesamaximumandafterwardsitdecreasingalmo stexponentially.The graphingure 5{6 (b)hasbeenderivedbothwiththeoreticalandexperimental


79 data.Theobservedmaximum,forthedyecurrentlyusedisabo ut5ppm.Anadditionaladvantageforusingthisconcentrationisthat,asse enfromthegraph,small variations(5 2ppm)aroundthisvaluedonothaveanyimpactonthereaction rate dr dC 0 C 0 = C max =0 5.3.3IntensityoftheRadiation Thelightintensityisanotherparameterthatcanaectther eaction.It isexpectedthatlowintensities(0to20W/m 2 )willexcitefewerelectronsand thereforetheoverallreactionratewillbelow.Whileincre asingthelightintensity thereactionratewillincrease,tillitreachesamaximumva lueandlevelout.The waythelightintensityinruencesthereactionratecannotb ederiveddirectlyfrom rstprinciples,butOllisetal.[187]afterreviewingseve ralstudiesconcludedthat threedistinctregionscanbedelineated(gure 5{6 (c)).(i)Forlowlightintensities thereactionrateincreasesproportionallytothelightint ensity( / I ).(ii)At intermediatelightintensitiesandbeyondacertainvalue( approximately20W/m 2 ) therateintensityisproportionaltothesquarerootofthel ightintensity( / p I ) and(iii)athigherintensitiesthelightintensitydoesnot haveanimpactonthe reactionrate.5.3.4SolidsLoading/SurfaceArea Manyresearchershavereportedtheeectofthesolidsloadi ngonthephotocatalyticeciency[187{189].Itis,however,morevalid hypothesistoassume thatthereactionconstantdependsontheavailablesurface areaandnotthesolids loading.Generallyincreasingthenumberofparticles(and consequentlytheavailablesurface)thesitesforadsorptionandOH generationwillalsoincreaseand thereforetheoverallreactionratewillincrease.Athighe rsolidsloading,however, thereareotherfactorsthatcomeintoplay,suchasmorerapi dcoagulationofthe particlesandUVlightshielding,thatwilleventuallyimpe dethereactionrate,until itreachesaplateau[188,189].


80 01020304050 4.0 4.5 5.0 5.5 6.0 6.5 Time (min)pH Degussa P25 ANTs 4.58 5.64 Figure5{7:ThepHvariationduringthedyedegradation.The initialvalue betweentheANTsandDegussaP25sincethespecicsurfacear eais dierent.IntherstcasethepHisstabilizedafter10minwh ileinthe secondcasethatoccursafter20min.Inbothcasesthestable pHvalue islowerthantheinitial. Thesolidsloading iscorrelatedtothesurfaceareapersolutionvolume S withtheequation: S m 2 100 ml = 3 R h g 100 ml i (5 7) Therelationbetween S and islinear,but S isalsoinverselyproportionalto theparticleradius R .Soforthesamesolidsloadingtheparticleradiushasa tremendousimpactonthereactionrate(gure 5{6 (d)).Sotoavoidvariationsdue tosurfaceareachangestheexperimentswillbeconductedon thesamesurfacearea basisunlessitisotherwisestated.

PAGE 100

81 400450500550600650 Wavelength (nm)Absorption (a.u.) 524 (nm) 537 (nm) 513 (nm) 60 min 50 min 40 min 30 min 20 min 10 min 0 min Figure5{8:Thedyespectrumduringthedierenttimeinterv als.Thethreedashed lines(513,524and537nm)arethethreewavelengthsthatwer eused forthe C=C 0 calculation.Thedatawereobtainedfromasampleof3 mgDegussaP25ina50mlof5ppmdyesolution. 5.4Experiments Foralltheexperimentstheparametersdiscussedabove(pH, initialdye concentration,radiationintensityandsolidsloading)we reeitherkeptconstantor monitoredtoensuretheaccuracyoftheresult.Thedyeconce ntrationwasalways keptat5ppm,thelightintensityoftheUVlampswas20W/m 2 (50W/m 2 forthe visibleradiation)andthepHwasmonitoredduringtheexper iments.Figure 5{7 showsthepHvariationduringthephotocatalyticdegradati on.Thestabilization occurred,relativelyfast,in20minforDegussaP25and10mi nfortheANTs(both short and long ).Themaximumdierencebetweenthereactionrates,duetot he dierentpHvalue(4.58versus5.64)willbeonlyintheorder of10%.Otherminor parameterssuchastemperatureandstirringspeedwereassu medinsignicant

PAGE 101

82 0102030405060708090 0 0.2 0.4 0.6 0.8 1 Time (min)C/C 0 Dye itself (10 ppm) Dye itself (5 ppm) Figure5{9:InvestigationofthedyedegradationundertheU Vlightfortwo dierentdyeconcentrations.TheUVisnothavinganapparen timpact onthedye. andthereforetheywerejustkeptconstant.Thesimpliedap proximationofthe Langmuir-Hinshelwoodmodel(equation 5 6 )wasusedtointerprettheobtained data.Toavoidvariationsduetotheinitialdyeconcentrati oninsteadofthe C ( t ) the C ( t ) =C 0 ( / I ( t ) =I 0 )valuewasusedtoobtainthereactionrate.Forevery experiment,threesampleswerecollectedandweremeasured withthePerkin-Elmer Lambda800UV/VIS.Figure 5{8 representsaverytypicalseriesoftheobtained spectra.Thedashedlinesdenotethethreedierentwavelen gthsthatwereused. Soeverydatapoint( I ( t ) =I 0 )wastheaveragevalueof9dierentintensities.The dierentresultswerecomparedwiththeparameter whichistheinversereaction constant1 =k app: .Physicallyitistherequiredtimefor67.21%destructiono fthe dye.OneoftheconcernswasthestabilityofthedyeunderUV. Toinvestigate whetherthedyeisUVstable,twosolutionswithdierentdye concentrations(5

PAGE 102

83 ppmand10ppm)wereexposedtoUVandthedyeconcentrationwa smeasured withthemethodpreviouslydescribed.Thedyeshowedexcell entstabilityinthe UV(350nm)(gure 5{9 )whichisinagreementwithHuetal.[180,184]and Sivalingametal.[178].5.4.1TitaniaNanoparticlesandCarbonNanotubes Thissetofexperiments,willinvestigatewhetherthecarbo nnanotubescanbe usedasphotocatalysisenhancers.Anataseparticleswillb emixedwithdierent amountsofnanotubesandwilldegradethedyeunderUV.These resultswillthen becomparedwiththerespectiveresultsfromtheparticleso nly.Theparticles areanatasenanoparticles(obtainedbyAlfa-Aesar,produc tnumber:44689)with primaryparticlediameter5nm( -TiO 2 ).Sincetheparticlesareverysmallitis expectedthatthebandgapwillbelargerduetoquantumeect s.Thechangein thebandgap( E g ): E g = h 2 2 2 R 2 1 m e + 1 m h 1 : 786 e 2 R 0 : 248 E RY (5 8) where h isthePlanckconstant,Rtheparticleradius, E RY theeectiveRydberg energycalculatedtobe4 : 3 10 39 J, isthedielectricconstantofanataseTiO 2 whichis86, m e andm h aretheelectronandholemasses,respectively[190].Reddy etal.[191]calculatedthe E g ,for5and10nmparticlesandis0.2and0.1eV respectively.Sofor5nmparticlesisrequiredminimumof34 6nm 1 .Accordingto Soetal.howeverinordertoeectivelyassistthephotocata lysisarerequiredUV lampswithpeakwavelength305nm[186]. Therewereintotalfourexperimentsperformed.Table 5{2 listsallthoseexperimentswiththeamountoftheparticlesandtheresult( ).Therstexperiment 1 = hc E g =12 : 398 10 7 ) [ A]= 12 ; 398 E g [ eV ]

PAGE 103

84 Table5{2:Summaryoftheexperimentsperformed Experiment AnataseparticlesCNTs 2 ID [min] A-1 3mg0mg52 : 40 0 : 9711.2615 A-2 3mg1mg27 : 09 0 : 2664.5865 A-3 3mg2mg53 : 46 0 : 994.2614 A-4 3mg3mg83 : 03 0 : 563.9119 (A-1)isdonetoevaluatethephotocatalyticactivityofthe anataseparticles.3mg ofanataseparticlesweredispersedin50mlof5ppmdyesolut ionandwereplaced intheUVchamber.Thesameexperimentwasrepeatedagainwit htheadditionof dierentamountofcarbonnanotubes(1mg,2mgand3mg). Figure 5{10 showsthephotocatalyticdegradationresults.Theredline sdenote thettingaccordingtoequation 5 6 .Theinsertsarethelogarithmicplot.Forthe rstexperiment is54.94min(gure 5{10 (b)).When1mgofnanotubesisadded inthesolutionthetime dropsto27.54min(gure 5{10 (b))whichrepresents asignicantreductiontotheparameter by50%.Thisprovesthattheinitial hypothesisthatthenanotubescanbeusedasphotocatalytic carriertoenhancethe eciencyistrue.Howeverfurtherincreaseofthenanotubes 2mgand3mgisnot havingthesameeect(gures 5{10 (c)and 5{10 (d)respectively).Thisisattributed tothefactthatthepresenceofthehighconcentrationofnan otubesisshieldingthe UVlightandmakesthesolutiondarker.Figures 5{11 (a)and(b)showthesame resultscollectivelyforimmediatecomparison. Oneofthequestionsraisedhereiswhetherthedyeadsorbson thecarbon nanotubesinsteadofbeingdestroyedbythetitaniaparticl es.Ifthatistruethe attributionofthedyeconcentrationreductiontotheenhan cementofthephotocatalysisisincorrect.Itisnecessarythereforetoperfor mcontrolmeasurements forseveralcarbonnanotubessolidsloadingz.Figure 5{12 showstheresultsofthe controls,wherethedyeconcentrationdoesnotchangesigni cantlyduringthe experiment.Itisobserved,however,asmall,stillquestio nable,reductiontothe

PAGE 104

85 0102030405060708090 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Time (min)C/C 0 0306090 0.01 0.1 1 Time (min)ln (C/C 0 ) 0102030405060708090 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Time (min)C/C 0 0306090 0.01 0.1 1 Time (min)ln (C/C 0 )(a) (b) 0102030405060708090 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Time (min)C/C 0 0306090 0.01 0.1 1 Time (min)ln (C/C 0 ) 0102030405060708090 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Time (min)C/C 0 0306090 0.01 0.1 1 Time (min)ln (C/C 0 )(c) (d) Figure5{10:TheresultsfortheexperimentsA-1toA-4.(a)J usttheanatase particles(b)anataseparticlesandCNTstogether( m TiO 2 : m CNTs =3:1)(c)( m TiO 2 : m CNTs =3:2)(d) ( m TiO 2 : m CNTs =1:1). orderof2%in90min.Besidesifthatwasthecase,withtheadd itionofthecarbon nanotubesinconcentrationsof2and3mgwouldfurtherappea rtoincreasethe photocatalyticeciency.Sofromthisexperimentsitisacc uratetoconcludethat thecarbonnanotubescanindeedassistthephotocatalytice ciency.

PAGE 105

86 0102030405060708090 0.01 0.1 1 Time (min)C/C 0 0 mg 1 mg 2 mg 3 mg (a) 0246 0 20 40 60 80 100 CNT solids loading (mg/100 ml)1/ t52.40 min 27.09 min 53.46 min 83.03 min (b) Figure5{11:Collectivegraphofthedatapresentedabove.I ngure(a)thedata areshownwiththettingandingure(b)thebarchartdemonstratesthedierenceinthevarious

PAGE 106

87 0102030405060708090100 0 0.2 0.4 0.6 0.8 1 Time (min)C/C 0 10 ppm CNT 5 ppm CNT Figure5{12:Investigationofthedyeadsorptiononthecarb onnanotubessurface. Theadsorptionwasnotsignicantsinceitwasonly5%reduct ion after90min. 5.4.2LongANTs:PhotocatalysisunderUVLight Inthissectiontheexperimentsarepreformedtoevaluateth ephotocatalytic eciencyoftheanatasecoatedlongcarbonnanotubes( ` -ANTs).Thebenchmark materialwastheDegussaAeroxideP25fromDuPont.Thereare twotypesof experiments;samesurfaceareabasisandsamemassbasis.As itwasdiscussed previouslythemostaccuratewaytodirectlycomparethepar ticlesistokeep mostoftheparametersthatinruencethereactionrate,cons tantforbothcases. Soinordertocomplywiththisrequirementweperformtheexp erimentsonthe samesurfaceareabasis.Thiswillguaranteethattheresult sdependonlyonthe photocatalyticpropertiesofthematerialandnotthepossi blehigherspecic surfacearea.Howevertheparticleshavebeendevelopedina mannerthattheywill

PAGE 107

88 providebothhighsurfaceareaandexceptionalphotocataly ticproperties,thusthe sameexperimentswereperformedagainonthesamemassbasis Figure 5{13 (a)showstheresultsforphotocatalyticdegradationonthe same surfaceareabasis.Thelightintensityusedherewas20W/m 2 andthepeak wavelengthofthelampwas350nm.Thesurfaceareaofthe ` -ANTsmeasured172 2 /gandforDegussaP25itwasfound52m 2 /g.Sothemassratiousedforthose experimentswasapproximately m ` ANTs : m P25 =1:3.Theresultsshowthatthe ` -ANTsperformedwellunderUVcomparedtotheDegussaP25.Th e ` ANTs =19.1 min( 0.4min)comparedtoDegussaP25forwhich P25 =24.1min( 0.4min). Sincethisexperimentwasdoneonthesamesurfaceareabasis ,thisdierenceis attributedtothephotocatalyticpropertiesoftheparticl es.Itisinterestingto comparethisresulttothoseforthecarbonnanotubes/anata separticlesmixture. Thesurfaceareaoftheanataseparticleis70m 2 /g,whichmeansthatthesurface ratiobetween ` -ANTsandanataseparticlesis S anatase : S ` ANTs =2 : 14.Still, however,wenoticethatthe ` ANTs issmaller.IntheMWNTs/ TiO 2 mixture thecontactbetweentheparticlesandthenanotubesisoccur ringduetoBrownian motionanditisinstantaneous.Inthecaseofthe ` -ANTsthecontactbetweenthe coatingandthecarbonnanotubesispermanent. Figure 5{13 (b)showsthedegradationdataonthemassbasiscomparison (1mgofDegussaP25and1mgof ` -ANTs).Thephotocatalyticeciency,as expectedwassignicantlyincreasedcomparedtotheprevio usresult.Thereason forthisisthehigherspecicsurfaceareaofthe ` -ANTs. 5.4.3LongANTs:PhotocatalysisunderVisibleLight Thissetofexperimentsinvestigatestheactivityofthepar ticlesunderthe presenceofvisiblelight.Thevisiblelightsourceweretwo halogenlampsof100W eachandthetotaloutputpowerwas50W/m 2 .Accordingtothemanufacturerof thelamp,thelamptemperatureissuch,thatthespectrumcon tainsasmallportion

PAGE 108

89 0102030405060 0.0 0.4 0.8 Time (min)C/C 0ANTs Degussa P25 (a) 0102030405060 0 0.2 0.4 0.6 0.8 1 Time (min)C/C 0 ANTs Degussa P25 (b) Figure5{13:PhotocatalyticdegradationofDegussaP25and ` -ANTsunderUV lightof350nmwavelength.(a) ` -ANTsareshowntobemore eectiveindestroyingthedyewith =19.1 0.4minwhileDegussa has =24.1 0.4min.(b)Thesamemassbaseresults. =19.1 0.4 minforthe ` -ANTswhileDegussahas =72.27 1.46min.

PAGE 109

90 04080 120 0.0 0.2 0.4 0.6 0.8 1.0 Time (min)C/C 0Degussa P25 ANTs Figure5{14:Thephotocatalyticresultsofthe ` -ANTsandDegussaP25.The ` -ANTsclearlydemonstratephotocatalyticactivitywith =152.31 6.13min.DegussaP25isnotdemonstratinganyobvious activity. intheUVregion.Thelamp,however,includesabuiltinUVlt erthatblocksthe UVradiation.InadditionaUVdetector(detectsradiationf rom270to400nm) veriedthatthereisnoUVlightpresentduringtheexperime nt.Inthiscasean amountof3mg ` -ANTsparticlesand3mgofDegussaP25wereused.Usingless quantityofthe ` -ANTswillyieldverypoorresultsandthemeasuredeciency was notreliable.Accordingtothespecicsurfaceareaofthepa rticlestheDegussaP25 hadtobe9mg,whichhoweverwouldhavemadethesolutioncomp letelyopaque, resultingthehighsolidsloadingproblems,suchascoagula tionandUVshielding eects.Thereforetheexperimentswerepreformedonthesam emassbasis.

PAGE 110

91 pro Fit TRIAL version 0246810 0.6 0.7 0.8 0.9 1.0 Time (days)C/C 0 const.shoulderFigure5{15:Thedyedegradationdatainthedarkforthe ` -ANTs.Degussaisnot includedheresinceitneverdemonstratedbehaviorlikesuc h.The datawerettedwiththeequation 5 9 DARK ` ANTs =1.29 0.24days. Theconstantis0.76 2.75 10 2 Theresultsarepresentedinthesamemanneringure 5{14 .Forthe ` -ANTs VIS ` ANTs =151.2 4.7min.DegussaP25failedtodemonstrateanyphotocatalyt ic behavior( 1 ).Thisisduetothewhitecoloroftitania,whichrerectsalm ost alltherangeofthevisiblelight.Onthecontraryforthe ` -ANTs,sincethecoating isverythin(4-6nm),thecolorofthecompositeisblackandt hereforeabsorbsall thevisiblelight.Thisresultisveryimportant,sinceanew propertyemergesfor the ` -ANTs.Therangeoftheapplicationcannowbeextendedfurth ersincethe ` -ANTscanbeeasilyusedunderthevisiblelight.

PAGE 111

92 5.4.4LongANTs:PostUVActivity,PhotocatalysisinDark Thisexperimentwasdesignedandperformedafteritwasobse rvedthat cuvettescontainingUVirradiatedsamples,leftinthedark forlongperiodoftime (days)appearedtocontainnodye.Sothisexperimentintent stomeasurethepost UVirradiation. Asolutionof1mgof ` -ANTswasplacedinthedarkchamberandirradiated for13.5minwithUV(350nmpeakwavelength)andintensityof 20W/m 2 .Inthis timethedyeconcentrationhasdecreased,accordingtothee xperimentdescribedin section 5.4.2 ,by50%.Thesolutionthenwasplacedinalightinsulatedcha mber underamagneticstirring,inatightlysealedvialtopreven tanywaterevaporation. Threesamplesof1.5mlwerecollectedeverytwodaysandwere leftfordayssothe particlescouldsettle.Thesamplesweremeasuredaccordin gtotheprotocolthat wasdescribedinsection 5.1.3 Figure 5{15 showcasesthepost-UVphotocatalyticeciency.Theobserv ed datafollowstherstorderreductionasbefore(equation 5 6 ),butithastome properlymodied: C ( t )= C 0 e t= +const.(5 9) Theconstantisdenotingthatthephotocatalyticdegradati oninthedarkis terminatedaftersomeperiodtimehaselapsed.Thereisalso ashoulderatthe beginning,whichdenotesadelayofthemechanismresponsib leforthedegradation. Forthoseexperiments =1 : 29days.Thedelayisroughlyabout2dayswhilethe degradationseemstostopatapproximately75%.5.4.5ShortNanotubes:PhotocatalysisunderUV Thesamesetofexperimentsasinsection 5.4.2 wereperformedwiththe s ANTs.1mgof s -ANTsweredispersedindyesolution via sonicationandthenthey wereplacedinthereactorwithlampsof350nmpeakwavelengt handtotaloutput value20W/m 2 .Figure 5{16 showstheresultofthephotocatalyticdegradation.

PAGE 112

93 0306090 120150 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Time (min)C/C 0 0306090 120150 0.1 1 Time (min)ln (C/C 0 )Figure5{16:ThedyedegradationdataintheUVlightof350nm forthe s -ANTs. UV s ANTs =177.41 10.00mins.Thephotocatalysisissignicantly slowerthatallthepreviouscases. The s -ANTsdemonstratedphotocatalyticresultsbutverypoorco mparedtothe ` -ANTs,DegussaP25andeventhe -TiO 2 .Theinversereactionconstantwas foundtobe =177.41 10.00min. The s -ANTswereexpected,toperformequallytothe ` -ANTssincethey arebothconsistingonanatasecoatingonmultiwallcarbonn anotubes.However theresultsaredramaticallydierent.Thereisnotanappar entreasonforthat. TheXRD(gure)showedanatasecrystalstructureforbothma terialsandthe XPS(gure)surveyshowedthattheparticlesconsistonlyon titaniaandcarbon nanotubes.Thedierent sol-gel precursorsandtheslightlydierentprocesscan haveminorimpactonthenalresult,sincetheanataseisinb othcasestheonly

PAGE 113

94 Table5{3:Summaryoftheexperimentalresultsofthischapt er. Experiment Light Materials 2 ID Source [min] A-1 UV305nm 3mg -TiO 2 52 : 40 0 : 9711 : 2615 A-2 3mg -TiO 2 +1mgCNT 27 : 09 0 : 2664.5865 A-3 3mg -TiO 2 +2mgCNT 53 : 46 0 : 984.2140 A-4 3mg -TiO 2 +3mgCNT 83 : 03 0 : 563.9119 AD-1 UV350nm 1mg ` -ANTs 19 : 61 0 : 208.9373 AD-2 3mgDegussaP25 24 : 06 0 : 3116.1456 AD-3 1mg ` -ANTs 19 : 61 0 : 208.9373 AD-4 1mgDegussaP25 72 : 27 1 : 420.0012 V-1 Visible 1mg ` -ANTs 152 : 31 6 : 130.6798 V-2 3mgDegussaP25 N/AN/A D-1 Dark 1mg ` -ANTs 1 ; 858 3460.0221 SA-1 UV350nm 1mg s -ANTs 177 : 41 10 : 00 presentphase.Sothereasonhavetobesoughtonthedierenc ebetweenthetwo kindsoftubes( s -CNTsand ` -CNTs). 5.5Conclusion Inthischapteraseriesofexperimentswaspreformedtoquan tifythephotocatalyticactivityofthesynthesizedparticles.Thephotocat alyticevaluationwasdone bythedegradationstudiesoftheazodye,BrilliantProcion RedMX-5B.After reviewingtheparametersthatwillpotentiallyinruenceth eresults,theconditions weresettosolelyobtainresultsbasedonthephotocatalyti cpropertiesofthe particles(theresultsaresummarizedintable 5{3 ).Thefollowingconclusionscan bederived.ˆCarbonnanotubescanassistthephotocatalysisbyalmost50 %whenmixed withanataseparticlesin1:3massratioaccordingtoexperi mentsA-1,A-2, A-3andA-4.ˆThe ` -ANTscanfunctionbetterundertheUV(350nm)comparedto DegussaP25onthesamesurfacearebase(AD-1andAD-2)andon thesame massbase(AD-3andAD-4).

PAGE 114

95ˆAlsothe ` -ANTsdisplayphotocatalyticactivityundervisiblelight (V-1) althoughitissignicantlylowerthantheUV-activity.Deg ussaP25failedto demonstratesuchactivityunderthoseconditions.ˆ` -ANTsdisplayed,whatisnamedaspost-UVactivity,dyedegr adationinthe dark,afteraninitialdoseofUVradiation.ˆTheexperimentV-1wasrepeatedforthecaseforthe s -ANTs(SA-1).The resultswerecompletelydierentcomparedto ` -ANTs.SincetheXRD showedthatbothtimeswehadanatasestructureofTiO 2 meansthatthe dierencecanbeattributedtotheCNTs.

PAGE 115

CHAPTER6 SPECTROSCOPICTECHNIQUESTOEXPLAINTHEPHOTOCATALYTIC EFFICIENCYOFTHEANTs. Theuseofthecarbonnanotubesascarrierofphotocatalysth adadramatic eectontheoverallphotocatalyticeciency.Thefactthat theanatasecoated carbonnanotubeshaveperformedbetterthantheDegussaP25 onthesamesurface areabasis(section 5.4.2 )showcasedthattherehastobesomethingmorethanjust thehighspecicsurfacearea.Inordertoascertainthereas onforthisfunctionality theinvestigationhastobefocusedontotheinterfaceofthe titaniacoatingand thecarbonnanotubes.Manycharacterizationtechniquesar eavailable,butin thiscasetheyarelimitedbytheamountofcarbonthatthecom positematerial contains.Thecarbonwilloverpowertheobtainedspectraan dconsequentlythe informationcannotbeconsideredaccurate.Thiswasalread yaproblemduring thecharacterizationofthecompositeparticleswithXRD.T hisresearchtherefore willmainlyfocusontheutilizationofsurfacesensitivete chniques.Sinceinthat casethemajorityoftheinformationwillcomeonlyfromthet opfewnmofthe materialtheinformationwillregardtheTiO 2 coatingandtheinterfaceofthe CNTsandTiO 2 .TheselectedtechniqueforthispurposeisX-RayPhotoelec tron Spectrometry(XPS).InadditiontotheXPS,RamanSpectrosc opyisusedto investigatethenatureofthenanotubesandproduceinforma tionregardingthe bonds.Literature,databasesandreferencematerial(anat asenanoparticles)are usedfortheanalysisofthedata Thematerialsthatwillbeinvestigatedaretheanatasecoat edshortcarbon ( s -ANTs)nanotubesandtheanatasecoatedlongnanotubes( ` -ANTs).Theshort nanotubeshavedisplayedverypoorphotocatalyticactivit ycomparedtothelong 96

PAGE 116

97 nanotubes( 5.4.2 and 5.4.5 ).Theshorteningofthetubeswasdonewithchemicalmechanicalprocessingwhichwouldhaveasignicantimpact onthestructure ofthetubesalthoughtheymaintainedthetubularstructure andtheconcentric cylinderarrangement(gure 4{2 ).Structuralinformationforthenanotubesand theTiO 2 canbeobtainedwithRamanandmoreaccuratebonding(CNT/Ti tania coating)informationcanbeobtainedfromXPS. ThischapterwillinitiallygivethegeneraltheoryoftheRa manspectroscopy forboththecarbonnanotubesandfortheanatasephaseoftit ania.Aparameter ofmajorimportanceinRamanspectroscopyistheprotocolth atwillbeusedto analyzetheobtainedspectra(samplepreparation,spectra smoothingandpeak recognition).Thereforeaprotocolisinitiallyestablish edandalltheobtained spectraareanalyzedbasedonthis.Thelastpartofthechapt erisdealingwith theX-RayPhotoelectronSpectrometry(XPS).TheXPSwasuse dprimarilyas complimentarytechniquetoRamantoreconrmtheresults,a ndsecondarily toinvestigatethepresenceofstresseswhicharisesfromth ebondbetweenthe MWNTsandtheTiO 2 coating. 6.1RamanSpectroscopyoftheCarbonNanotubes TheRamanspectroscopyisaverypowerfulandvaluabletoolf ortheinvestigationofthecarbonnanotubesproperties[192{194].Nanot ubescanbethought asverycomplexmacromoleculeswiththousandsofcarbonato msthatwillgive risetomanyvibrationalfrequenciesthatarestronglydepe ndedonthestructure ofthetubes.Althoughthecarbonnanotubesarearelativene wmaterial,fromthe extensivestudyoftheHighlyOrderedPyrolyticGraphite(H OPG)thereissucientknowledgetostudythepropertiesofthenanotubeswit hRamanspectroscopy. Inadditioncomputersimulationsareagreatcomplimentary tooltoexplainthe Ramanspectra,sinceinthecaseofcarbonnanotubestheanal yticalcalculations areverycomplicatedanddemanding[195].

PAGE 117

98 1-phononemission 2-phonons FirstOrder SecondOrder 6 7 8 9 8 : 9 ; < = < > = ? @ ? A @ B ? C D E F E G F H H I J J K L M K N M M (a) (b) (c) Figure6{1:ThedierentRamanscatteringprocessesforCNT s.(a)Firstorder Ramanscattering(b)and(c)arethesecondorderRamanscatt ering. The k representsthemomentumoftheincidentphotonand q representstheemittedphonon.Therstrowrepresentsthei ncident resonance,andthesecondthescatteredresonance. 6.1.1GeneralTheoryofRamanSpectroscopyofCarbonNanotu bes TheRamanspectraofgraphiteandSWNTscanprovideinformat ionabout theexceptional1Dstructureofcarbonmaterials,sucharep hononandelectron distributions.Sincetheconductingstate(insulator,sem iconductorconductor)is directlyrelatedtotheelectronicstructureRamanspectra candirectlycorrelate thosepropertiestocertainpeaksandpeakshapes[196{198] .Similarly,themechanicalandthermalpropertiesarestronglycorrelatedto thephononinteractions andthephonondistribution,andthereforeRamanspectraca nprovideverydetail informationoftheSWNTsregardingthethermalandmechanic alproperties. BothRamanspectraandFTIRareinelasticscatteringofthel ight.Fora Ramanprocessandduringascatteringevent(i)anelectroni sexcitedfromthe valencetotheconductionbandbyabsorbingaphoton,(ii)th eexcitedelectron isscatteredbyemitting(orabsorbing)phonons,and(iii)t heelectronrelaxes

PAGE 118

99 tothevalencebandbyemittingaphoton.Theobservedscatte redphotonhas energy,whichissmallerthantheenergyoftheincidentphot on(whenaphononis emittedduringthede-excitation).InRamantheintensityo fthescatteredphoton ismeasuredasafunctionofthedownshiftoftheenergy(phon onemission).The downshiftismeasuredusuallyincm 1 .ThoseRamanpeaksarecalledStokeslines. Ifthesameprocessisrepeatedandthistimetheup-shiftfre quency(absorptionof phonons)isrecordedthenitiscalledanti-Stokeslines[19 9{201].Theanti-Stokes andStokeslinesaresymmetricto0cm 1 whichrepresentstheRayleighscattering. Ingeneral,however,theadsorptionofphononislesslikely tohappenandthe intensityoftheanti-StokelinesislowerthantheStokelin es. Thenumberofemittedphonons(orabsorbed)beforetherelax ationofthe latticecanbeone,canbetwo,ormore,whicharecalledoneph onon,twophonon andmulti-phononRamanprocessesrespectively.Ifthereis onlyelasticscattering, withnofrequencyshiftinvolveditcorrespondstoRayleigh scattering.Figure 6{1 showsthebasictransitionsthatgiverisetotheRamanscatt eringforcarbon nanotubes;therstrowrepresentstheincidentresonance( incidentphotonenergy isequaltothegap)andthesecondthescatteredresonance(t heemittedphoton energyisequaltothebandgap).The symbolingure 6{1 symbolizesthe energydiagramwiththeconductionandvalenceband.Figure 6{1 (a)demonstrates thesimplestrstorderRamanscattering.Onephoton( k )excitesanelectronto ahigherband,aninelasticscatteringfollowsaccompanied withtheemissionofa phonon( q )andthentherestoftheenergyisemittedinformofaphoton. The energyoftheemittedphotonthereforeis E Raman = h ( k q ).Ingures 6{1 (b) thesameprocessisdone,butinthiscaseanelasticscatteri ngisinvolved(dashed lines).ThoseareprocessesthatareknownassecondorderRa mansincethere aretwoscatteringprocessesinvolved.Case(c)isanothers econdorderRaman scatteringwheretwophononsareemittedbeforetheobserve dphotonisemitted.

PAGE 119

100 (a) (b) Figure6{2:GraphicrepresentationofthemajorRamanmodes .(a)Radial breathingmodeofasinglewallnanotube.(b)GBandgraphicf or singewallnanotube.TheDBand,sinceitinvolvestwoconseq uent vibrations,cannotberepresented. 6.1.2BasicRamanLinesforCarbonNanotubes Basedonthepreviouslyexplainedtheorythemostimportant Ramanactive bandswilltobedescribed.MostofthemappearonlyforSWNTs ,buttheresome verysignicantpeaksthatarealsopresentinthecaseofthe MWNTs[202{204]. Figure 6{3 showsatypicalspectrumforSWNTswiththemostdominant Ramanfeatures,theRadialBreathingMode(RBM),theGband, bothclassiedas rstorderprocesses,andtheDband,whichisclassiedasse condorder. TheRBMisthecoherentexpansionandcontractionofthenano tubestothe radialdirection[197,206](gure 6{2 ).TheRBMhasbeenstudiedextensivelysince itisrelatedtothediameterofthenanotubes[197]andsecon dlyonthedensityof electronicstates[206].ThisisaneasilyobservedmodeinS WNTsandincertain caseforisolateddoublewallednanotubes.Whenthesamplec onsistsofmultiwall nanotubesthenthisfrequencyusuallydiminishes.Thesefe aturesareuniquetothe carbonnanotubesandoccurwithfrequencies RBM between120and350cm 1 for

PAGE 120

101 Figure6{3:TypicalRamanspectrafrommetallicandsemicon ductingSWNTs. TheRadialBreathingMode(RBM),theDBandandGBandarethemostimportantbands.The*isdenotingbandsthatcomeformt heSi substrate.Duetothedistinctstructureofthesemiconduct ing nanotubestherearetwoadditionalbandsMandiTOLAthatapp ear. (a) (b) Figure6{4:TheGBandsplitandhowitisrelatedtotheconduc tivityofthe tubes.(a)TheGBandsplitandhowitisrelatedtotheconduct ivity ofthetubes.(b)Thedierencebetweenthe + G and G .The + G isnot changingbutthe G varieswiththediameterandfollowstheequation 6 2 [205].

PAGE 121

102 tubesrangingfrom0.7nm
PAGE 122

103 thenanotubes(metallicsemiconducting)[214].IftheG lineshapeisbroaderthan theG + oneanditisbetterapproachedbyBreit-Wigner-Fanoequati onthenit meansthatthetubesaremetallic[215]( 6{4 (a)).Thereareempiricalrelationsthat correlatethedierencebetweenthe + G and G withthediameterofthetubes. + G G = A d 2 (6 2) where A is47.7nm 2 /cm 1 [196]or45.7nm 2 /cm 1 [193]forsemiconductingand 79.5nm 2 /cm 1 [196]formetallicSWNTs(gure 6{4 (b)).Ifthesplitdoesnot appearindicatesthatthetubesarenotmetallic.Thosefeat uresoftheGBand canbegeneralizedtothecaseofthemultiwallnanotubesand forverywelldened MWNTsitcanbebetterthantheSWNTs[216]. AnotherbandwithsignicantinterestistheDBand[217].Th eDbandis oneofthesecondorderRamanscatteringandinvolveseither onephononandone elasticscattering(gure 6{1 (b))ortwophonons(gure 6{1 (c))[218,219].The frequencywheretheRamanshiftappearsfortheDBanddepend sonthelaser energy[207,220].AtypicalexampleofthisfeatureistheDB andthatshowsat 1350cm 1 andshiftsby53cm 1 ,whenthelaserenergychangesby1eV.TheD Bandshowsforamorphouscarbonalso,anditappearsatthefr equenciesbetween 1285cm 1 and1300cm 1 andtheFullWidthatHalfMaximum(FWHM)ismore than100cm 1 .Fornanotubesthisshowsatfrequenciesbetween1305cm 1 and 1350cm 1 andwithFWHMabout30-60cm 1 .Averyinterestingfeaturearises whentheDbandiscomparedtotheGBand.Theratiobetween I D and I G isa measureofthecrystallinityofthenanotubes,meaninghowp ristinethenanotubes are[205]. R = I D I G (6 3)

PAGE 123

104 Usuallywhen R 0( R< 1)thenthecrystallinityishigher.Someresearchers denethesameratioasthe R = R + 1 1 f D ( ) d! R + 1 1 f G ( ) d! (6 4) where f D ( )and f G ( )istheLorentzianoftheRamanDandGpeakrespectively. Thereareotherlesssignicantpeaksthatcangivemoredeta iledstructural information,butsincetheyareobservedonlyforSWNTsthey arenotdiscussed here.Thistheory,however,isenoughtodescribethebehavi orofthecomposite materials( ` -ANTsand s -ANTs). 6.2RamanSpectroscopyoftheAnataseStructureofTiO 2 ThegeneraltheoryoftheRamanspectroscopyissimilarfort hetitania crystals,butinthiscasethevibrationsarerepresentingc oherentlatticevibrations insteadofjustbondvibrations.TheTiO 2 canexistinanatase,rutileandbrookite, witheachstructurehavingverydistinctvibrationalfrequ encies.Asalready discussedin 2.1.1 sectionanataseistetragonal( D 19 4 h )withtwoformulaunitsper unitcellandsixRamanactivemodes( A 1 g +2 B 1 g +3 E g )[221].Rutileisalso tetragonal( D 14 4 h )andhastwounitcellandfouractivemodes( A 1 g + B 1 g + B 2 g + E g ) [222].Finallybrookiteisorthorhombic( D 15 2 h )andhaseightformulaunitsperunit cellandshows36Ramanactivemodes(9 A 1 g +9 B 1 g +9 B 2 g +9 B 3 g )[223].Table 6{1 enliststheRamanfrequenciesandtherelativeintensityof thepeaksforanatase andrutile.Theanalyticalcalculationsforthosepeaksare ingreatagreementwith experiments. Oneofthemostimportantcharacteristicsisthepeakat144c m 1 .Itwas recentlydiscoveredthatitisverysensitivetothesizeoft hegrainandtherefore thesizeoftheparticles[224].Thatsensitivitycanbeexpr essedinanasymmetricbroadeningofthepeaklineshapeandblueshift(towa rdshigherwave numbers)[225].Inaninnitycrystalthephononsarefreeto travelinanydirection beforetheyareabsorbedbackfromthelattice.Inthecaseof nano-sizedhowever

PAGE 124

105 Table6{1:TheRamanfrequenciesfroanataseandrutilephas eoftitania.The brookiteisnotincludedheresinceisnotapresentformofTi O 2 andit hasintotal36weakpeaks.Thenotationinparenthesisisrep resenting therelativeintensityofthepeaks;w:weak;m:medium;s:st rong;vs: verystrong.Dataareadaptedfromreferencematerialandre ference Anatase D 19 4 h I4 1 /amd Rutile D 14 4 h P4 2 /mnm E g 144cm 1 (vs) B 1 g 143cm 1 (w) E g 197cm 1 (w) E g 447cm 1 (s) B 1 g 399cm 1 (m) A 1 g 612cm 1 (s) A 1 g 515cm 1 (m) B 2 g 826cm 1 (w) B 1 g 519cm 1 (m) -E g 639cm 1 (m) -crystalsthephononsareconnedinaspacelessthantherequ iredforunconstrainedinteractions[226].Thecalculationsfortheline -shapechangehavetobe doneinthereciprocalspace.Inthisformulationthe I ( )isgivenbytheequation [225]: I ( )= Z B.Z. j C (0 ; q ) j 2 d 3 q [ ! ( q )] 2 + a 2L (6 5) whereB.Z.denotesthelimitsforthe1 st Brillouinzone, a L isthehalfwidthat halfmaximum, ( q )isthephonondispersioncurveand C (0 ; q )isthescattering coecientforrstorderscatteringofsphericalnanocryst alsanditcanbewritten as: j C (0 ; q ) j 2 =exp q 2 d 2 16 2 (6 6) ( q )isthedispersioncurvefortitania.Thisresultistoocomp licatedtobedirectlycalculatedbutitcanbeapproachedwiththeassumpti onthatthedispersive relationisasimplevibrationalmodeinacrystal,suchus: ( q )= 0 + [1 cos( j q a j )](6 7) The1 st BrillouinzonecanbeapproachedbytheFermisphere.Sothel imitfor theintegralinequation 6 5 are0to k f = q 2 E f m e h 2 .Withthoseassumptions

PAGE 125

106 equation 6 5 canbemodiedto: I ( )= Z r 2 E f m e h 2 0 exp q 2 d 2 16 2 2 d 3 q ( ! 0 + [1 cos( j q a j )]) 2 + a 2L (6 8) whichfurtherreducessinceweareusingtheFermispherefor the1 st Brillouinzone into: I ( )= Z r 2 E f m e h 2 0 exp q 2 d 2 16 2 2 4 q d q ( ! 0 + [1 cos( j q a j )]) 2 + a 2L (6 9) Thecalculationofthefunction I ( )isnottrivialeveninthecaseoftheequation 6 9 .Althoughtherearealotofassumptionsandsimplications ,depending ontheapproaches(Brillouinzone,dispersionrelations)t hereisanasymmetric broadeningandblueshiftthatstronglydependsonthediame teroftheparticles. Theshiftis3.2cm 1 forparticlesofaveragediameter5nmandhasanadditional thebroadeningof3cm 1 (FWHM)towardslowerenergyvalues.Thisbroadening andshiftingisrelatesonlywiththe144cm 1 lineandthesizedoesnotaectthe otherbands.Inadditionsinceoneoftheassumptionswasthe sphericalshapeof theparticles,whichisnotaccuratesincethereisnoindica tionaboutspherical titaniaparticlesonthenanotubes.Itishoweveragoodesti mationoftheorderof magnitude. 6.3ExperimentalProcedures Thissectionexplainsthebasicmethodsforpreparingandob tainingthe Ramanspectra.Aprotocolthatsummarizesallthemathemati calmodelsand manipulationthatwillbeusedtoanalyzethedatahastobees tablishedand accordingtowhichallthedatawillbeprocessed.6.3.1SamplePreparation OneofthebiggestadvantagesofRamanSpectroscopyisthefa ctthatit requiresverylittlesamplepreparation.Ineverycase5mgo fsampleweremixed with1mlofiso-propanolinaformofthinslurry.Theslurryw asplacedon

PAGE 126

107 aslideglassandleftinroomtemperaturetoevaporatetheis o-propanol.The Ramanspectrawereobtainedbythe(NicoletMAGNA760Benchw ithSpectra TechContinuumIRMicroscopeandFT-Raman)andthelaserwav elengthwas 752nm.Sincethesampleswereblackincolorthefullpowerof thelaserwas usedtomaximizetheobtainedsignal.Dierentspotsofthes amesampleand dierentsamplesofthesamematerialyieldedthesamespect ra,butwithdierent intensitiesanddierentnoiselevels.6.3.2MathematicalAnalysisandManipulation Smoothingisaverysensitivemanipulationofthedatasince oversmoothing mayresultdisappearingofsomepeaks( 6{5 (b))andunder-smoothingmayshow pseudopeaksthatmaybemisleading( 6{5 (c)).Althoughthereiscommercialsoftwareavailabletosmoothandanalyzethedata,inthisresear chmanualsmoothing andtwaspreferredsothedatamanipulationisfullycontro lled. ThealgorithmforthesmoothingwasLOESS.Thetermisderive dfromthe termlocallyweightedscatterplotsmooth.Themethodusesl ocallyweightedlinear regressiontosmooththedata.Theprocessisweightedbecau searegressionweight functionisdenedforthedatapointscontainedwithinthes pan.Inadditionto theregressionweightfunction,youcanusearobustweightf unction,whichmakes theprocessresistanttooutliers.Finally,themethodLOES Susesaquadratic polynomial.Ifitusesalinearpolynomial,itiscalledLOWE SS.Thealgorithm givestheoptionofusingallthedataoracertainsectionaro undthedatapoint ( x i ;y i ),calledspan, .Forlargespanthedatabecomesmootherandthetime requiredforthecalculationsincreasesdramatically.For aseriesofdata( x j ;y j ), where j =1 ; ;N andthepoint( x i ;y i )theprocesshastheinthesteps: 1.Thefollowingdistancesarecalculated: d i = j x k x i j ;i =1 ; 2 ; ;N (6 10)

PAGE 127

108 whichthensortedintoascendingorder. 2.Thequantityqiscalculated, q =max(Truncate( N ) ; 1)(6 11) 3.Thisisusedtocalculatethedistancescale D = d q 1 d N > 1 (6 12) Thesteps2and3haveonlycomputationalpurposesandbasica llytheywill guarantythatthesmallestdistance D willnotbesmallerthan d 1 4.Theweightedfunctionforthedatapointis: T ( u )= (1 j u j 3 ) 3 j u j 1 0 j u j 1 (6 13) andbasedonthisequationtheweightsforthedatapointsare thengivenby w i = T x i D (6 14) 5.ForLOESS,theregressionusesasecondorthirddegreepol ynomial.Theset ofpointsusedforthetareintheform( x i ;y i ;w i )Thedierencebetween weightedleastsquaresandtheregularleastsquaresisthat thefunctionthat isminimizedis F ( a 1 ;a 2 ; ;a N )= N X i =1 w i y i f ( x i ; a 1 ;a 2 ; ;a M ) i 2 (6 15) where f isthepolynomial P Ml =1 a l x l ,thatisusedforthetandcanberst, secondorthirdorder. 6.Theprocessisrepeatedforthenextpoint. Figure 6{5 demonstratetheapplicationoftheLOWESSalgorithm.Dier ent variationsintheparameter canhavedramaticeectonthenalresult,(b) under-smoothed,(c)over-smoothedand(d)nicelysmooth.I nthecaseofthe presentdatatheparameter wasvarying0.03-0.02dependingonthenoiseto signalratio.ThesoftwarethatrunthesmoothalgorithmisM athematica5.1by WorthramResearch.ThewholealgorithmisatAppendixandwa sobtainedbythe

PAGE 128

109 1000 1100 1200130014001500160017001800 Raman Shift (cm 1 )Intensity (a.u.) Raw data 1000 1100 1200130014001500160017001800 Raman Shift (cm 1 )Intensity (a.u.) Raw data a =0.03, second order (a) (b) 1000 1100 1200130014001500160017001800 Raman Shift (cm 1 )Intensity (a.u.) Raw data a =0.3, second order 1000 1100 1200130014001500160017001800 Raman Shift (cm 1 )Intensity (a.u.) Raw data a =0.09, second order (c) (d) Figure6{5:DierentoptionsfortheLOESSalgorithm.(a)Ra wdataasobtained. (b)Under-smootheddata,thatgivefalsepeaks,(c)Oversmo othed datathatsmoothoutnecessarypeaks(d)nicelysmootheddat awith allthepeaksshowingnice.Alwaysusedquadraticequationf orthet andthevariationwascomingfromthespan classnotesofDr.McQuarrieandisbasedonthealgorithmbyC leaveland[227{ 229].Itwasslightlymodiedsoitcouldhandlelargernumbe rofdatainshorter time.

PAGE 129

110 TheobtainedpeaksinmostofthecaseswerettedwiththeLor entzpeak prole.Theequationdescribingthatproleis f ( )= I 0 a L ( ! 0 ) 2 + a 2L (6 16) where I 0 isthemaximumintensity, a L isthehalfwidthathalfheight,and 0 the frequencywherethepeakappears.Althoughthe a L canbedirectlymeasuredand obtainedfromthegraphitisnotrecommendedsincethebackg roundhastobe rstsubtractedandthentheexactheightandwidthofthepea kcanbemeasured. Sointhiscasethe a L isoneofthettingparameters.Fornpeaksequation 6 16 becomes: f ( )= 1 n X i =1 I ( i ) 0 a ( i ) L ( ! ( i ) 0 ) 2 +( a ( i ) L ) 2 (6 17) ForthebackgroundofRamanspectroscopythereareseverala pproaches,but inthiscasethebestonefoundtobeasimplepolynomialequat ionthatgoesupto thethirdorder. k X i =0 a i i (6 18) where k =0 ; 1 ; 2 ; 3.Sotheequationusedtottheobtainedspectrais f ( )= 1 n X i =1 I ( i ) 0 a ( i ) L ( ! ( i ) 0 ) 2 +( a ( i ) L ) 2 | {z } LorentzPeaks + k X i =0 a i i | {z } Background (6 19) where n isthenumberofpeaksand k theorderofthepolynomialbackground correction. Asstatedintheprevioussectiontherearecaseswherethe G canbetted withtheBreit-Wigner-FanoequationwhichissimilartoLor entz,buthasan asymmetricbroadening.TheequationofBreit-Wigner-Fano is I ( )= I 0 1+ ! 0 q 2 1+ ! 0 2 (6 20)

PAGE 130

111 where I 0 istheintensity,isthehalfwidthathalfmaximum(HWHM), q isa broadeningparameterand 0 isthefrequencywheretheRamanpeakappears.In thiscasetheequationusedis: f ( )= 1 n 1 X i =1 I ( i ) 0 a ( i ) L ( ! ( i ) 0 ) 2 +( a ( i ) L ) 2 | {z } LorentzPeaks + k X i =0 a i i | {z } Background + I 0 1+ ! 0 q 2 1+ ! 0 2 | {z } Breit-Wigner-Fano (6 21) Inallcasesforthebackgroundbothsecondandthirdorderpo lynomialswereused. Basedontheparameter 2 theorderthatwasgivingthebestvaluewaskept. ThealgorithmicforthettingwastheMonte-Carlo,Levenbe rg-Marquardtand Robust,whichcameaspartofthesoftwareProFitfromQuantS oft.Inmostof thecasesallthealgorithmsgavethesameresultsatthetti ngparameterswith minordeviations.Insomecases,certainalgorithms(Monte -CarloorLevenbergMarquardt)failedtoconvergeandonlytheremainingalgori thmswereused. Besidestheobviouspeaksthetwasattemptedwithmorepeak s,toensurethat therearenotanyotherhiddenpeaks.Sowhenforexamplether earethreeobvious peaks,thetisattemptedwiththreeandinadditionfouror vebuthiddenor overlappingpeakswereneverfound. Thegraphsarerepresenteddirectlywiththesmootheddata. Theblack solidlinerepresentsthesmootheddata;theredline(darkg rey)representsthet includingthepeaksandthebackground;andnallythedashe dlinesrepresentthe dierentpeaksthathavebeenidentied. 6.4ExperimentalResults Thissectionisdividedintwosubsections.Therstonerega rdscharacterizationofthenanotubesbeforethecoatingandtheotheroneaft erthecoating.All thedatahavebeenobtainedandprocessedaccordingtothepr otocolestablishedin section 6.3

PAGE 131

112 6.4.1LongNanotubesaftertheAcidTreatment Figure 6{6 showstheRamanspectraforthelongnanotubesaftertheacid treatment.Theacidtreatmentisexpectedtodamagethesurf aceofthenanotubes, whichwillhaveaneectonthevibrationalfrequenciesofth etubes.Thelow frequenciessuchastheRBMdidnotappearsothe0 1000cm 1 regionisnot included.SincethereisasplitintheGBandthetwasattemp tedforboth line-shapesLorentz(equations 6 19 )andBreit-Wigner-Fano(equation 6 21 ). Equation 6 21 gavebettert( 2 parameter),thusthe 6 21 twaskept. Therstthingnoticeablefromgure 6{6 andtable 6{2 ,isthattheDBand appearsatthe1312cm 1 andthe a L is22cm 1 whichisaverydistinctcharacteristicthatthetubesconsistontubulararrangementof graphenesheetsand notamorphouscarbon(e.g.carbonnanowires).Thenextvery importantfeature showinginthegureistheGBand.Itshowsroughlyat1594cm 1 andhasavery distinctsplit.Breit-Wigner-Fanogavebettertresults, whichisaveryprofound characteristicofthemetallicnatureofthecarbonnanotub es.Thisisoneofthe mostimportantresultssince,itpointsoutthatthe ` -CNTsareconductiveinnature.AndbasedonthetheorythatwasdiscussedinChapter2a dditionofmetals tothephotocatalystcandramaticallyimprovetheoverallp erformance. Thenextsteptotheanalysisistocalculatetheratiosbetwe entheGBand andDBand via theequation 6 3 and 6 4 .Thisanalysiswillgivethemagnitudeofthecrystallinityofthetubes.Accordingto 6 3 (as I G 0 isconsideredthe I G 0 + I G + 0 2 )theratiois3 : 081.Forthecaseof 6 4 theequationhastobemodiedas R = R + 1 1 f D ( ) d! R + 1 1 f G + ( ) d! + R + 1 1 f G ( ) d! (6 22) Accordingtotheequation 6 22 theratiois2.624slightlylowerthantheprevious. Ingeneralthe R< 1isforverycrystallinetubesand R> 1isfortubeswith defectsonthestructure.Thisisreasonablesincetheacidt reatmenthasshownthe

PAGE 132

113 Table6{2:TherawttingparameterscalculatedwiththeLev enberg-Marquardt algorithmfortheacidtreated ` -CNTs.Thegraphicrepresentationof theresultsisingure 6{6 .Thetyielded 2 =7.1333 10 4 .For convenienceatthedatarepresentationweusethesymbol a (2)L insteadof thatisusedinequation 6 20 FittedparametersStandarddeviations Background a 0 =-824.6088 a 0 =14.7875 a 1 =1.5224 a 1 =2.1812 10 2 a 2 =-5.4635 10 4 a 2 =7.7575 10 6 DBand Lorentz I (1) 0 =863.1688 I (1) 0 =2.0660 (1) 0 =1311.4516 (1) 0 =5.2327 10 2 a (1)L =21.9585 a (1)L =9.0829 10 2 GBand(G ) BFW I (2) 0 =308.2726 I (2) 0 =2.5939 (2) 0 =1582.5923 (2) 0 =0.1729 a (2)L =15.5437 a (2)L =0.2718 q=0.0342 q =2.7345 10 6 GBand(G + ) Lorentz I (3) 0 =252.0681 I (3) 0 =3.4170 (3) 0 =1611.2235 (3) 0 =0.1401 a (3)L =9.6443 a (3)L =0.2420 destructionoftheouterwallsandtheattachmentof COOHand OHgroups. Anotherpossiblereasonforthevalueoftheratiocanbethep resenceofimpurities otherthanandacidtreatmentbyproducts.ButtheXPSproved thatthereareno otherelementsthancarbon,andcarbonimpuritiesareinthe formofthinlayer thatcannotaecttheRamanspectrainsuchamanner.Itisacc urateresultsto concludethatthe ` -CNTsareconducting,havingdistincttubularstructurewi th veryhighdensityofsurfacedefectsasaresultoftheacidtr eatment. 6.4.2ShortNanotubesaftertheAcidTreatment AgaininthiscasetheRamanSpectroscopydidnotshowntheRB Mfrequency. BesidesthefactthatRBMisnotveryeasilyidentiedinthec aseofMWNTsthese nanotubeshavedamagedstructurenotonlyfromtheacidtrea tmentbutalsofrom theshorteningprocess,whichutilizesacids(H 2 SO 4 andHNO 3 )andmechanical method.Thustheregionofinterestisfrom1000cm 1 to1800cm 1 .Fromthe shapeofthegraphthereisnotadistinctsplitattheGBandan dalthoughboth

PAGE 133

114 1000 1100 1200130014001500160017001800 Raman Shift(cm 1 )Intensity (a.u.) Raw data Peak at 1312 cm 1 Peak at 1583 cm 1 Peak at 1612 cm 1 Fit Figure6{6:The ` -CNTsaftertreatedwithnitricacidat140‰for10hours.TheD Bandisshowingat1312cm 1 andtheGBandatabout1594cm 1 .A verydistinctsplitofthebandcanbeseenwiththeG + atthe1584 cm 1 andG at1612cm 1 equations, 6 21 and 6 19 wereusedthe 6 21 failedtogiveaccurateandfurther morethe 6 19 couldnotbettedwhenthenumberofpeakswassetat3. Fromgure 6{7 andthetable 6{3 againtherstnoticeablethinghereisthe DBandthatappearsat1305cm 1 whichisthelowerlimitfortheDBandinthe caseofMWNTs.Thetgavea a DL ofabout31.3431whichisnotbroadenough toconcludethatthisisamorphouscarbon,butstillitcanbe assumedthatthe broadeningiscomingfromtheheavydamagethatthetubessu eredduetothe

PAGE 134

115 Table6{3:TherawttingparameterscalculatedwiththeLev enberg-Marquardt algorithmfortheacidtreated s -CNTs.Thegraphicrepresentationof theresultsisingure 6{7 .Thetyielded 2 =3.9138 10 7 FittedparametersStandarddeviations Background a 0 =-1.4326 10 4 a 0 =1403.8835 a 1 =35.2712 a 1 =3.2537 a 2 =-1.4479 10 2 a 2 =2.4599 10 6 a 3 =-1.5079 10 6 a 3 =6.0625 10 7 DBand Lorentz I (1) 0 =1.3329 10 4 I (1) 0 =39.1153 (1) 0 =1305.1773 (1) 0 =9.1653 10 2 a (1)L =31.3431 a (1)L =0.1725 GBand Lorentz I (2) 0 =5585.0378 I (2) 0 =40.1330 (2) 0 =1586.3196 (2) 0 =0.2058 a (2)L =29.7420 a (2)L =0.4163 shorteningprocessandtheacidtreatment.EveniftheTGAan dtheXPSsurvey showedthepresenceofiron(6.0%wt),theironalonecannota ectdirectlythe Ramanspectra. ThenextcharacteristicistheGBand,whichseemstobetheov erlappingof toodierentpeaksveryclosetogetherandalsoverybroad.A llthetalgorithms failedtorecognizetwopeakswithvariationinthesmoothin gparametersand background.Itisthereforeaccuratetoconcludethatthere areisnotadistinct splitoftheband.Fromthetable 6{3 weseethatthepeakshowsat1586cm 1 whichisexpectedfornanotubes.Theinterestingfeatureis thatthebroadening ofthatpeakis30cm 1 whichisverylargeforDBandpeak.Basedonthetting parametersobtainedfromthetable 6{3 thecalculationof R ,equations 6 3 and 6 4 ,willdeterminethequalityofthetubes.Therstapproach( equation 6 3 ) gives R =2 : 38657whichisverysmallcomparedtothepreviouscase.This giveinitiallytheimpressionthattheshortnanotubesarel essdefectiveandtheir structureismoredenedthanthelong.Butwhenthesecondap proach(equation 6 4 )isusedthenthe R =3 : 60633,whichismoreacceptableresultregardingthe

PAGE 135

116 1000 1100 1200130014001500160017001800 Raman Shift(cm 1 )Intensity (a.u.) Raw data Peak at 1305 cm 1 Peak at 1586 cm 1 Fit Figure6{7:The s -CNTsaftertreatedwithnitricacidat100‰for6hours.TheD Bandisshowingat1305cm 1 andtheGBandatabout1586cm 1 AlthoughtheGBandlookslikeitconsistsontooverlappingp eaksit stillcanbetreatedasonepeak. processinghistory.Thatshowsthatthe s -CNTshavemoredefectscomparedtothe longnanotubesdiscussedintheprevioussection. Themostimportantresultfromthesespectrahowever,remai nstheshape oftheGBand.Theabsenceofthesplit(oratleastaverydisti nctsplit)denotes theveryhighpossibilitythatthosetubeslackofconductin gproperties.Thisisa veryimportantconclusionandcanbecorrelatedtothepoorp erformanceofthe ` -ANTs.Theconclusionthattheshortnanotubeshavepoorerc onductivitycan

PAGE 136

117 Table6{4:TherawttingparameterscalculatedwiththeLev enberg-Marquardt algorithmforthetitaniacoated ` -CNTsandthetitaniasegmentofthe spectrum.Thegraphicrepresentationoftheresultsising ure 6{9 .The tyielded 2 =8 : 3378 10 4 FittedparametersStandarddeviations a 0 =-8809.5084 a 0 =2.0077 10 4 Background a 1 =9.1589 a 1 =16.7950 a 2 =2.9614 10 4 a 2 =4.6200 10 3 I (1) 0 =427.8412 I (1) 0 =5.2873 E g (1) 0 =150.1796 (1) 0 =0.1802 a (1)L =16.4892 a (1)L =0.4071 I (2) 0 =71.307 I (2) 0 =0.881 B 1 g (2) 0 =408.7834 (2) 0 =0.3522 a (2)L =36.3452 a (2)L =1.2131 I (3) 0 =85.568 I (3) 0 =1.057 E g (3) 0 =629.1235 (3) 0 =0.4801 a (3)L =22.3412 a (3)L =1.3041 beusedtoexplainthisbehavior.Smallestconductivitymea nspoorerabilityto transporttheelectronsawayfromthetitania,whichresult slessholes,consequently less[OH ]andthereforelowerphotocatalyticactivity. 6.4.3LongNanotubesaftertheCoating Figure 6{8 showsthetotalspectrumofthe ` -CNTsafterthecoating.Inthis casethecarbonnanotubeshaveathincoatingoftitaniaandt hereforeallthe carbonnanotubepeaksappearclearly.Onthecontrarytheti taniapeaksarenot veryclearandonlytheverystrongandmediumstrengthpeaks appeared.The spectracanbedividedintotworegions,0 1000cm 1 wherearethetitaniapeaks appearandthe1000-1800cm 1 wherearetheMWNTspeaksappear.Avery generalcharacteristicisthattheCNTpeaksappearverycle arandthattheyhave maintainedtheirbasicshape.Againalltheanalysiswasdon ebasedonthesame protocol(section 6.3 ). Figure 6{9 showsthespectraandtable 6{4 summarizesthetresults. Thetitaniapartisintheformofaverythincoatingandthere foretheonly

PAGE 137

118 0 200400600800 10001200140016001800 Raman Shift (cm 1 )Intensity (a.u.) Figure6{8:TheRamanspectraofthecoatedlongnanotubes.T herearetwoseparateregions,(i)0-1000cm 1 thatcontain thetitaniapeaksand(ii)1000-1800cm 1thatcontainthecarbonnanotubespeaks.Thepeakidentic ationis donelaterinthechapter.

PAGE 138

119 0 100200300400500600700800900 1000 Raman Shift ( cm 1 )Intensity (a.u.) Raw data Fit Peak at 150 cm 1 Peak at 409 cm 1 Peak at 629 cm 1 Figure6{9:Therstregionfromgure 6{8 .Therearefourmajorpeaksbutonly threeofthemcanbeidentiedaccurate.149.56cm 1 ,628.65cm 1 and 408.64cm 1 obviouspeaksis E g thanthemainpeakataround144cm 1 whichisnicelytted withaLorentzian.Thepeakhowevershowsat150cm 1 whichindicatesa6 cm 1 blueshiftcomparedtotheliteraturevalue.Therearetwoma jorreasons forthisirregularity.Thesurfaceterminationoftitaniap articlesisimposing constrainstothephonons,whichresultsamoreasymmetricp eakandblueshift. Calculationsbasedonequation 6 9 ,showthatthisshiftwilloccurforvalues of2-2.5nmparticles,whichisinagreementwiththeliterat ure[226]andmore detailedcalculations.Grainsofthissizemayexistbutiti snotthemajority,since

PAGE 139

120 theaveragegrainsizeis5nm.Consequentlywecannotconclu dethattheblue shiftcomesonlyfromthesizeoftheparticles.Inadditionf orashiftof5cm 1 accordingtoequation 6 9 weshouldobserveagreatasymmetricbroadeningofthe peaks.Howeverthisisnotthecase,sincethepeaksarenicel yttedwithjusta singleLorentzline.Thatleadstoonemorereasonfortheshi ft.ShiftsinRaman spectroscopycancomefromalterationofthesymmetry,asar esultofpossible bondingtoanon-nativeelement.Inthiscaseasithasbeenst atedinchapter4, thereare COOHand OHgroupsonthesurfaceofthenanotubes.Thosegroups areusedasanchoringspotsforthe sol-gel chemistryofthetitaniacrystals.So itispossibletohaveaTiO 2 bondintheformofC O Ti.Ramanspectroscopy providesevidenceofthatbond. Additionalproofcomesfromtheothertwopeaksthathavebee nidentied inthespectrumtheoneat399cm 1 and639cm 1 .Thosepeaksaresignicantly furtherthananyrutilepeaks(447cm 1 and612cm 1 respectively)sothereis nodoubttheybelongtoanatase.Forbothpeaksweobserveash iftthatisnot towardsthesamedirection.Morespecicallyforthe B 1 g peakisobservedablue shiftby10cm 1 (peakat409cm 1 )andforthe E g peakitisobservedaredshift by10cm 1 (peakat629cm 1 ).Thosepeaksdonotchangeduetothedimensions ofparticle,theonlyreasontheyshiftedcanbethebondingt oanonlattice element.Thereforethisargumentcanfurtherjustifythere sultfortheCNT TiO 2 bondandthattheblueshiftoftherst E g frequency(144cm 1 )doesnotcomes exclusivelyfromthesizeeect.Thefrequencieswherethep eaksappearisnot inruencedbytheTi +3 orTi +4 butonlyonthephonondistribution. Theothersegment(1000 1800cm 1)ofthegraphisaboutthenanotubes. Analysisaccordingtotheprotocolgavetheresultsthatare collectivelyrepresented intable 6{5 .AgainwenoticethepositionoftheDbandthatisat1307cm 1 and the a L is26cm 1 whichisslightlylargerthanthevaluebeforethecoating.T he

PAGE 140

121 Table6{5:TherawttingparameterscalculatedwiththeLev enberg-Marquardt algorithmforthetitaniacoated ` -CNTs.Thegraphicrepresentationof theresultsisingure 6{6 .Thetyielded 2 =8 : 3378 10 4 .For convenienceatthedatarepresentationweusethesymbol a (2)L insteadof thatisusedinequation 6 20 FittedparametersStandarddeviations Background a 0 =-1086.0823 a 0 =17.1629 a 1 =1.9385 a 1 =2.5408 10 2 a 2 =-6.6503 10 4 a 2 =9.0231 10 6 DBand Lorentz I (1) 0 =1091.2343 I (1) 0 =2.0617 (1) 0 =1307.0246 (1) 0 =4.9037 10 2 a (1)L =26.2351 a (1)L =8.9694 10 2 GBandG BFW I (2) 0 =389.9422 I (2) 0 =0.2694 (2) 0 =1579.1871 (2) 0 =0.3259 a (2)L =19.5770 a (2)L =0.3259 q=0.0546 q =1.3445 10 6 GBandG + Lorentz I (3) 0 =369.6873 I (3) 0 =5.3204 (3) 0 =1605.8443 (3) 0 =0.1502 a (3)L =12.0105 a (3)L =0.2529 signicantobservationhereisthattheDbandafterthecoat inghaveablueshift by4cm 1 .Thereasonforthatisagainthepossiblebondbetweentheti tania coatingandthenanotubes.Thebroadeningofpeakcanalsobe attributedtothe samereason,sincethewidthisdirectlycorrelatedtotheam ountofcoherencein thevibrations.Thecoatingwillconstrainthosevibration sandconsequentlywill broadentheD-band. ThenextbandistheGband,whichhasmaintainedthesplit,ac haracteristic oftheconductingnatureofthenanotubes.TheG isappearingatthe1589cm 1 andthe a L is20cm 1 whiletheG + bandisatthe1606cm 1 andthe a L is12 cm 1 .Comparedtothevaluesbeforethecoatingthatindicatesar edshiftby6 cm 1 forG andablueshiftby6cm 1 forG + .Itisinterestingtocomparethe relativeintensityofthe I G + to I G ( R = + )beforeandafterthecoating.Similarly

PAGE 141

122 1000 1100 1200130014001500160017001800 Raman Shift (cm 1 )Intensity (a.u.) Raw data Fit Peak at 1307 cm 1 Peak at 1579 cm 1 Peak at 1606 cm 1 Figure6{10:Thesecondregionfromgure 6{8 .TheDBandisat1307cm 1 and theGBandisattheabout1590cm 1 .Thebandsplitstillexists, withtheG at1579cm 1 andtheG + at1606cm 1 tothe I G =I D ratiothe R = + = R + 1 1 f G ( ) d! R + 1 1 f G + ( ) d! (6 23) beforeandafterthecoatingwillgiveameasureonhowthepea kshavechanged.So beforethecoatingthisyields R = + =1 : 97106andafterthecoating R = + =1 : 71925 whichindicatesthattherelativeintensityoftheG + toG hasincreased.As mentionedinsection 6.1.2 theG + issensitivetochargetransferthatcomesform sourcessuchasdopantaddition.Inthiscasethechangeisal sorelatedtothe titania-CNTbondthatcanresultchargetransfertotheunde rlyingnanotubes

PAGE 142

123 fromtheTiO 2 .Anotherinterestingpointisthecalculationoftheratiob etween theGandDband.Accordingtoequation 6 4 theratiois3.9637andaccording 6 3 is2.8731.Inbothcasesitissignicantlyhigherthantheva luescalculated forthecaseofthebarenanotubes(2.64and3.081respective ly).Thismeansthat thecrystalstructureofthenanotubeshavebeensignicant lydistortedduetothe possiblebondingwiththetitaniacoating. Howeveramongallthedierentchanges,themostoutstandin gisthefrequencyshift.Theshiftsaresignicantindicationoftheex istingofC O Tibonds. Insimilarcases,otherresearchershavereportedsimilarp eaksthathavebeen attributedtocertainbonds.Yakovlevetal.[230]andKamad aetal.[231]have workedwiththincoatingsoftitaniaonsilicaandreportedt heexistenceofthe Si O Tibondat950cm 1 butthisbondwasnotaccompaniedbybondshift atthetitaniaorsiliconpeaks.Thisismostlikelyduetothe factthatthelm wasthick(700nm)andthereforethebulktitaniapeaks(that appearinnormal frequencies)coveredanyshiftduetothebonding.Inthisca sewedonotobserve anypeakthatcanbedirectlyattributedtoaC O Tibond.Howeverthepeak shiftaloneisaverystrongevidenceforthatbond.6.4.4ShortNanotubesaftertheCoating Figure 6{11 showstheRamanspectraforthecaseoftheshortcoatedCNTs s ANTs.Againthisspectrumcanbedividedintotworegions;on efrom0-1000cm 1 regardingthetitaniapeaksandasecondfrom1000-1800cm 1 forthenanotubes. Oneoftheinterestingresultsisthatthetitaniapeaksarea lotmoreobviousand intensecomparedtothepeaksbefore.Thereasonforthatist hethickercoatingthe s -ANTs(6nm)haveversusthe ` -ANTs(4nm).Ramanisrelativesurfacesensitive (approximately800nm)techniqueandthereforethethickne ssofthecoatingwill impacttheresults.Thereforegure 6{11 requiredtwodierentacquisitionswith dierentsettings.Initiallythetitaniaoverpoweredthes pectrum,soasecondrun

PAGE 143

124 0 200400600800 10001200140016001800 Raman Shift (cm 1 )Intensity (a.u.) Figure6{11:TheRamanspectraofthecoatedshortnanotubes .Therearetwoseparateregions,(i)0 1000cm 1 that containthetitaniapeaksand(ii)1000 1800cm 1thatcontainthecarbonnanotubespeaks.Thisspectrahas beenobtainedbythecombinationoftwodierentruns.

PAGE 144

125 Table6{6:TherawttingparameterscalculatedwiththeLev enberg-Marquardt algorithmfortheacidtreated ` -CNTs.Thegraphicrepresentationof theresultsisingure 6{12 .Thetyielded 2 =1.9924 10 8 FittedparametersStandarddeviations Background a 0 =-824.6088 a 0 =14.7875 a 1 =82.2700 a 1 =254.8581 a 2 =-0.1334 a 2 =2.9830 10 4 a 3 =6.3907 10 5 a 3 =1.6551 10 6 E g I (1) 0 =1.3060 10 5 I (1) 0 =191.5728 (1) 0 =149.9038 (1) 0 =1.3982 10 2 a (1)L =9.6293 a (1)L =2.5634 10 2 B g I (2) 0 =6099.0402 I (2) 0 =211.8730 (2) 0 =202.3874 (2) 0 =0.2600 a (2)L =7.5055 a (2)L =0.4139 E 1 g I (3) 0 =5456.3761 I (3) 0 =138.7558 (3) 0 =392.5676 (3) 0 =0.4355 a (3)L =17.4229 a (3)L =0.7914 B 1 g A 1 g I (4) 0 =3461.7658 I (4) 0 =155.9701 (3) 0 =510.0127 (3) 0 =0.5928 a (4)L =13.3065 a (4)L =0.5928 E g I (5) 0 =1.0324 10 4 I (5) 0 =125.1647 (5) 0 =632.6918 (5) 0 =0.2443 a (5)L =20.4046 a (5)L =0.4285 wasneedtofocusontheCNTspart.Themathematicalanalysis againwasdone accordingtotheprotocoldescribedinsection 6.3 Figure 6{12 showstherstpartofthespectrumregardingthetitania.In thiscasesincethetitaniacoatingwasthicker,thepeaksar esignicantlyclearer thanbeforeandallofthepeakslistedintable 6{1 appear.Intable 6{6 the mathematicalanalysisofthosepeaks,showsagainalargebl ueshiftonthe144 cm 1 for6cm 1 .Againthisshifthastwomajorcontributions,sizeeectof thecoatingandthebondingoftitaniaonthenanotube.Regar dingtherst contributionandaccordingtochapter4theprimaryparticl esizeis4to8nm andaccordingtoequation 6 9 theshifthastobeapproximately2and3cm 1 respectively.Sointhiscasesincetheshiftis6cm 1 therehastobeanadditional

PAGE 145

126 01002003004005006007008009001000 Raman Shift (cm 1 )Intensity (a.u.) Raw data Fit Peak at 150 cm 1 Peak at 202 cm 1 Peak at 393 cm 1 Peak at 510 cm 1 Peak at 633 cm 1 Figure6{12:Therstportionofgure 6{11 .Thereare5verydistinctivepeaksat 150cm 1 ,202cm 1 ,393cm 1 ,510cm 1 and633cm 1 reasonforthepeakshift.Thatreasonagainisbondingofthe carbonnanotube andtitania.Theotherpeaksappearalsoshifted.Sothe E g isat202cm 1 shifted by5cm 1 ,the B 1 g isatthe393cm 1 shiftedby6cm 1 (blueshift),the A 1 g isat 510cm 1 shiftedby5cm 1 (blueshift)andthe E g nallyisat633shiftedby6 cm 1 (blueshift).Animportantcharacteristic,theshiftsthat arenotequaland theyarenotallatthesamedirection.Someofthemareblue( B 1 g A 1 g and E g ) andsomered( E g ).Theshiftsagain,dependsonthekindofvibrationandon howitisaectedbythebondtothenonlatticeelement,inthi scasecarbon.A veryimportantpeakisthesmallpeakattheendofthespectru m(730cm 1 )that

PAGE 146

127 Table6{7:TherawttingparameterscalculatedwiththeLev enberg-Marquardt algorithmforthecoated s -CNTs.Thegraphicrepresentationofthe resultsisingure 6{12 .Thetyielded 2 =1.0956 10 5 FittedparametersStandarddeviations Background a 0 =7561.0312 a 0 =147.6613 a 1 =-15.4746 a 1 =0.3227 a 2 =1.0866 10 2 a 2 =2.3081 10 4 a 3 =-2.5298 10 6 a 3 =5.4214 10 8 DBand I (1) 0 =152.8715 I (1) 0 =2.1243 (1) 0 =1316.0874 (1) 0 =0.5633 10 2 a (1)L =46.6845 a (1)L =1.3153 G Band Lorentz I (2) 0 =111.2193 I (2) 0 =2.7395 (2) 0 =1544.7456 (2) 0 =0.6863 a (2)L =21.3581 a (2)L =1.2052 G + Band Lorentz I (2) 0 =543.3215 I (2) 0 =3.8876 (2) 0 =1582.2848 (2) 0 =8.2552 a (2)L =10.9464 a (2)L =0.1459 10 2 wasintentionallyomittedfromthet,sinceitisnotrecogn izedasanatase,rutile, brookiteoranycarbonvibrationalmode.Itisbelievedthat itistheC O Ti bondthatisformed.Yakovlevetal.[230]mentiontheTi O Sibondat950 cm 1 .Similarlyitcanbearguedthatthispeakat750cm 1 isaTi O Xbond. Atthispointonlyextensivemathematicalcalculationscan provethevalidityof thisconceptandthereforeisnotgoingtobethemainargumen tofthesection. Onthecontrarythebondshiftisaverysolidproofandtheref oreisgoingtobe themajorargument.Thereisnotanobviousasymmetricbroad eningofthepeaks asitwasexpectedfromnanosizedparticleswhichmeansthat thepeakshiftis primarilyassociatedwiththeC O Ti.Thereasonforthatisthattheparticles arenotexpectedtobespherical,andthedimensionontherad ialdirectionofthe tubeisnotnecessaryequaltothedimensionatdirectionpar alleltothetube.That willeectthephononconnement(byapproachingabulkcrys tal).Thereforeitis accurateatthispointtoconcludethatpeakshiftisproduct oftheCbondingand notthephononconnement.

PAGE 147

128 1000 1100 1200130014001500160017001800 Raman Shift (cm 1 )Intensity (a.u.) Raw data Fit Peak at 1316 cm 1 Peak at 1544 cm 1 Peak at 1582 cm 1 Figure6{13:Thesecondportionofgure 6{11 .Althoughthecarbonpeaksarenot veryclearwecanstillseethematthe1316cm 1 theGBandandat the1582cm 1 theGBand.TheGBandseemstobesplittingintwo peaks1544cm 1 and1582cm 1 .Theratiobetweenthepeaksis completelyreversedbutthisiscurrentlyattributedtothe weaksignal obtainedbythe s -CNTsinthiscase. Ingure 6{13 andintable 6{7 aretheresultsregardingtheCNTspart,in thiscasethe s -CNTs.Thequalityoftheplotisnotverygoodsincethetitan ia layerwasrelativethicksotheemittedradiationwasnotint enseenough.Following thesameanalysisasbeforethedataweresmoothed,butthesm oothcouldnot eliminateapseudopeakthatappearedintheGBand(1544cm 1 ).Thatpeak

PAGE 148

129 althoughitcouldbeapproachedwithaLorentzpeak,butnotw ithaBreit-WignerFanopeak,yieldsvaluesforthewidth( a L )andfrequency( G )thatforaG Band arenotrealistic.Still,however,sincetheG isrelatedmainlytochargetransfer ofthatsplitcanbeattributedtopossiblebondbetweenthec oatingandtheCNT. SincetheG cannotbeapproachedwiththeBreit-Wigner-Fano 1 peakissecure toconcludethatthenanotubesarenotchangingstate(semic onducting metallic), whichisaphysicalacceptableresult.Inadditionallthepe akshaveshifted comparedtotheuncoatednanotubes.TheDBandisshowingabl ueshiftby16 cm 1 andtheGband(inthecaseafterthecoatingwillbeconsidere dtheG + )is showingredshiftby4cm 1 Averyinterestingresultinthisspectrumistheratiobetwe entheDandG Band( R ).Theratiocanbecalculatedbytheequations 6 3 and 6 4 .Equation 6 3 weobtain0.2799(assumingthat I G = I G + )andfromequation 6 4 weget 0.857487.Inbothcasesweobtainednumberssmallerthan1,w hichmeansthat thenanotubeshaveverycrystallinestructure.Thisresult cannotberepresentative ofthosetubesspeciallyaccountingtheprocessinghistory andthecoating.The coatingasinthecaseofthelongnanotubeshavetoincreaset herationinsteadof decreasingit.Inthiscasethoughsincethecoatingisveryt hick,manyvibrational modeshavebeenprevented.SinceDBandistheresultsoftwoc onsequentvibrations(aphononexchangebetweentwodissimilarcarbonatom s)itisexpectedto havereducedintensity.SooveralltheRamanspectraforthe caseofthe s -ANTs demonstratedthesameresultsasinthecaseofthe ` -ANTs.Allthenanotube andtitaniapeakswereshiftedandinadditionthereweredra maticchangesonthe 1 InthecaseofBreit-Wigner-Fanot,noneofthealgorithmsc ouldconvergetoa realisticresult.

PAGE 149

130 Table6{8:SummaryoftheRamanresult.Herearelistedthema jorpeaksand shiftbothfortitaniaandCNTsafterthecoating. ` -ANTs s -ANTs Titania ` -CNTs Titania s -ANTs BandShift BandShift BandShift BandShift E g +5cm 1 DBand 5cm 1 E g +5cm 1 DBand 5cm 1 E g G Band 4cm 1 E g +5cm 1 G BandAppeared B 1 g +10cm 1 G + Band 6cm 1 B 1 g 6cm 1 G + Band 4cm 1 A 1 g A 1 g B 1 g 10cm 1 B 1 g 9cm 1 E g E g 6cm 1 shapeoftheCNTspeakthatconcludethatthereisabondbetwe enthetitania coatingandtheCNTs.6.4.5SummaryoftheRamanSpectraAnalysis Table 6{8 showscollectivelytheresultsoftheRamanspectroscopy.T hemost interestingresultcomeswhenthespectrabeforeandafterc oatingarecompared, allthepeaks(bothtitaniaandCNTs)weresignicantlyshif ted.Thesecond importantresultisthatallthepeakshavedierentshiftno tonlyinmagnitude, butindirectiontoo.Thisbasicallyeliminatesthefacttha ttheshiftcanoccurred duetoamiss-calibrationoftheinstrument.Thefactthatth etwocompletely dierentparticleswithdierentphotocatalyticproperti esdisplayedsimilarresults inregardstothebondinginformation,leadstoanotherreas onforthedierencein thephotocatalyticeciency.Thatreasoncanbelocatedtot hesplitoftheGBand thatoccursonlyinthecaseofthe ` -CNTs(excellentphotocatalyticproperties)and notforthecaseof s -CNTs(poorphotocatalyticproperties).Thesplitwasnoto nly veryobviouswiththetwopeakstohavealmostsimilarintens ity,buttheG was ttedbetterwiththeBreit-Wigner-Fanolineshapecompare dtoLorentz.Sofrom thespectroscopicanalysiswecanconcludethat:ˆThereisabondbetweenthetitaniacoatingandthecarbonnan otubes

PAGE 150

131ˆAndthatthe ` -ANTsthatperformedbetteratthephotocatalyticevaluati on consistonmetalliccarbonnanotubes,whilethe s -ANTsconsistofnonmetallicpropertiescarbonnanotubes. AtthispointitisobviousthattheC O Tibondexists,butinordertoreconrm thatresultinthefollowingsectionX-Rayphotoelectronsp ectroscopyisperformed. 6.5X-RayPhotoelectronSpectroscopy(XPS) Inthissectionweareusingthephotoelectronspectroscopy toconrmthe resultsoftheRamanspectroscopyregardingthebondinginf ormation.XPSisalso averysurfacesensitivetechniquesoitwillgiveinformati onfortheanatasecrystal andtheinterface.InXPStheemittedX-raysejectacoreelec tron.Thiselectron's energyis E k = E X-Ray E b where E b isthebindingenergyoftheelectron.Sincethe E X-Ray isverywelldenedandtheenergyoftheemittedphotoelectr oncanbevery preciselymeasuredthe E b isknownwithveryhighaccuracy.Thebindingenergyof theelectrononaverysimpliedmodelis: E b = k 2 e 4 m e 2 h 2 Z eff n 2 (6 24) Where Z eff istheeectivenucleuscharge,aftertheelectronscloudpa rtially shieldthenucleus.Inthecaseofthebondofoneelementtoan othertheelectron distributionwillimmediatelyimpacttheeectivechargea ndthereforethebinding energywillbechanged.InXPSspectrumthischangecanbesee nintwoways; Peakshift: Themajorbindingpeakwillshift,sincethe Z eff ischanging.The changein Z eff cancomefrompossiblebondtoadierentelementortobond stressduetocrystalconnement.Thechemicalshiftdepend sontheamount ofstressorthenumberofbondstothedierentelement.Chem icalshiftsto higherenergiesareattributedtobondstoelementsthatatt racttheelectron cloudandthereforeincreasingthe Z eff whichaccordingtoequation 6 24 will increasethebindingenergy.Shiftstolowerenergieswills imilarlymeanthat

PAGE 151

132 theelementthatthepeakiscomingfrom,isattractingtheel ectroniccloud andthe Z eff issmallerthereforetheenergyshiftstolowerbondingener gies. Extrapeak: Theoriginofthispeakisthesameasthechemicalshiftbut,i n thiscasenotalltheatomsarebondedtootherelementssothe initialpeak remainsandjustanextrapeakappears,atslightlydierent energy. TheXPSspectrumoftitaniahasbeenstudiedextensivelyalr eadyandthereisa largeliteraturereferencelibraryaboutit.Themajorpeak saretheOxygenpeak O1sthathasamajorpeakat531.5eV.Thereisasecondarypeak ataround527529eV.Thatpeakisattributedtolatticeoxygen,whilethe rstoneisattributed tosurfaceoxygen.Thelatticeoxygenisverysensitivetoth esizeofthecrystal grain.SoinordertoinvestigatetheXPSfornanosizedparti clesitisrecommended touseareferencematerialwiththesamesizetodetermineth eexactpositionof thelatticepeak.ThenextpeaksthataresignicantfortheX PSisthetitanium peakTi2p 1 = 2 andTi2p 3 = 2 .TheTi2p 1 = 2 peakappearsat464.2eVandisvery preciseasitisingoodagreementwithliteraturedatabase. TheTi2p 3 = 2 isat 458-459eV[232{236].Sharpandintensepeaksisagoodindic ationthattheTiO 2 consistonlyonTi +4 .InadditioninthecaseofTi +3 andaccordingtoequation 6 24 the Z eff willbereducedwhichwillshiftthebindingenergytolowere nergies. Soforthisstudyweusedthreedierentsamples,the5nmpart icles( -TiO 2 ),the ` -ANTsand s -ANTs. 6.5.1Instrument,SamplePreparationandMathematicalAna lysis TheinstrumentusedforthisstudyistheKratosAnalyticalS urfaceAnalyzer XSAM800.Foreveryspectrumtwodierentsampleswereprepa redandthepeaks werecomparedtoensurethattheresultsareaccurate.Thesa mplepreparationwas similartothesamplepreparationfollowedfortheRamanspe ctroscopy.Thinslurry waspreparedbymixing5mgofparticlesand1mlofiso-propan ol.Theslurrywas placedona1cm 1cmsiliconwafer(crystallographicplane(100))andleftd ry.

PAGE 152

133 TheenergyiscalibratedusuallywiththeCarbon1speak.Int hiscasein additiontothattheSilicon1speakwillalsousedforcalibr ation.Thereasonis thatsincethecarbonnanotubesarebondedtothetitaniathe bindingenergyof carbon(C1s)mighthavebeenshifted. Thecommercialsoftwarethatcamewiththeinstrumentwasus edtosmooth thedata.ThepeakswerettedwithGausslineshape; I ( E )= I 0 p 2 e ( E E 0 ) 2 2 2 (6 25) where I 0 istheintensity( N ( E )), E isthebindingenergy E 0 thebindingenergy wherethepeakappearsand aparameterrelatedwiththewidthofthepeak ( FWHM = p 2 ).Incertaincasesinordertotthetailofthepeaksweare usingtheVoigtpeak,whichisamixofLorentzandGausspeak: I ( E )= r 2 I 0 s l s g VT E E 0 ) s g p 2 ; s l s g p 2 (6 26) where VT ( y;x )= y Z + 1 1 e t 2 y 2 +( x t ) 2 dt (6 27) again E isthebindingenergy, E 0 istheenergywherethepeakappears, s g and s l is parametersofthet.ThedefaultequationistheGaussifnot stateddierent.The talgorithmsusedaretheLevenberg-Marquardt,Monte-Car loandRobust.The peakrecognitionwasdonebasedontheliteratureandonthec ommercialsoftware. 6.5.2XPSoftheReferenceMaterial TherstmaterialanalyzedwithXPSistheanatasenanopowde r( -TiO 2 ). Sincethesizeis5nmitisexpectedthatthelatticeoxygenwi llhaveaslightshift thatcomesfromthesizeconstrainwhilethesurfaceoxygenw illnotbeaected. Figure 6{14 isshowsthecarbonpeak.Thecarbonisnotpartofthemateria l composition,butcomesfromtheatmosphereanditisexpecte dtobepresent ineveryXPSsample.Thecarbonpeakat284.6eVisthetypical C1speakand

PAGE 153

134 280 282 284 286 288 290 292 294 296 298 300 Binding Energy (eV)N(E) Raw Data Fit Peak at 284.6 eV Peak at 288.4 eV 284.6 eV 288.4 eV Figure6{14:TheC1speakforthereferenceanatasenanopart icles.Themajor peakisatthe286.4eVthatisagreementwithliteratureands everal databases. representselementalcarbon.Thereisasecondarysmallerp eakthatispresent at288.4eV,whichisalsotypicalpeakforcarboncontaminat edsamples[237{ 239].Thenoisetosignalratioisrelativehigh,whichisexp ectedforcarbonof thisnature.Ithastobenotedthatthesecondarypeakcannot besatisfactory approachedbyanyoftheGauss(equation 6 25 )ortheVoigt(equation 6 26 ) peaks,sincethenoisetosignalratioishigh. ThenextimportantpeakistheSi2ppeakandisdisplayeding ure 6{15 .It comesfromthesubstrateandithasalsobeenusedtocalibrat ethemeasurements.

PAGE 154

135 96 98 100 102 104 106 108 Binding Energy (eV)N(E) Raw Data Peak Fit Peaks at 98.5 eV Peak at 102.5 eV 102.5 eV 98.5 eV Figure6{15:TheSi2ppeakforthereferenceanatasenanopar ticles.Themajor peaksareatthe98.5eVfortheSi2p 1 = 2 andat102.5eVforthe Si2p 3 = 2 whichareinagreementwithliteratureandseveraldatabase s. Thetwopeaksareinaagreementwithdatabasevalues.Therei ssomeslightshift, whichisattributedtotheformationofathinoxidelayeront opofthewafer.The doublecalibrationwasdonetore-ensurethattheTiO 2 peaksarecorrectlylabeled andlocated. ThenextpeakthatisresolvedfromXPSistheO1s(gure 6{16 ).Thenoise tosignalratioisverylowandthereforettingisverywell. ForO1sitisexpected onlyonepeaksinceitisasingleenergylevel(nospin-orbit alcoupling).Howeverin thiscasethepeakappearssplit.Therstpeakobservedat52 9.6eV,representsthe

PAGE 155

136 528 530 532 534 536 538 540 542 Binding Energy (eV)N(E) Raw data Peak fitting Peak at 529.6 eV Peak at 531.6 eV 529.6 eV 531.5 eV Figure6{16:TheO1speakforthereferenceanatasenanopart icles.Themajor peaksareatthe529.6eV,representsthelatticeoxygen,and the531.5 eVforthesurfaceoxygen.whichareagreementinwithlitera tureand severaldatabases. latticeoxygen.Itisshiftedslightlycomparedtodatabase values,butthereason forthatisthesize,whichis5nm.Theotherpeakappearsat53 1.5eVandis attributedtothesurfaceoxygen.Inthiscasetheenergyish ighercomparedtothe bulksincethereareopenbonds.Inadditionsincethemateri alisnano-sized,it hashighsurfacearea,thereisalotofsurfaceoxygenandthe reforetheintensity ofthepeakishigheraswell.Ithasbeenarguedthattheratio betweenthetwo peakscanbecorrelatedtothesurfaceareaofthematerial[2 34].Howeverthisis

PAGE 156

137 notabsolutelycorrectsincetheratiobetweenbulkandsurf aceoxygendependson thecrystalorientation.Certaincrystallographicorient ationsarericheronoxygen. Sotherelativeratioofthetwopeakswillbeafunctionofbot hthesurfacearea andthecrystallographicorientation.Thosepeakswheret tedverywellwiththe Gaussianlineshape.Fromthetresultswecancalculatethe ratioofsurfaceto bulkoxygen(RSB) RSB= I Surface I Bulk (6 28) Thatratioforthereferencematerialisestimatedtobe0.83 07. Thelastpeakisthetitaniumpeak(gure 6{17 ).Therearetwopeaksfor titaniumthe458.4eVfortheTi2p 1 = 2 andat464.2eVfortheTi2p 3 = 2 ,bothingood agreementwiththedatabase.Againthettingwasexcellent withtheGaussian lineshape.6.5.3XPSofthe s -ANTs Thenextsampleanalyzed via XPSisthe s -ANTs.TheC1swillbeusedas calibrationsincetheSi2pisveryweakandisinsecuretobeu sedforcalibration(a summaryofallthepeaksisgivenattheendofthechapter).In additionthe284.6 eVisaverycharacteristicpeakofcarbonbasedmaterials.A mongallthecarbon peaksthe284.6eVwillalwayshavethehighestintensitysin ceitisgeneratedby elementalcarbon.Secondarypeakswillrepresentotherstr ucturessuchasbonds andfunctionalgroups. ExaminingtheC1speakofthecoatedshortcarbonnanotubes( gure 6{18 )it isobservedthattherearetwoverydistinctpeaksat284.6eV and285.9eV.The 284.6eVisthepeakthatisbeingattributedprimarilytothe elementalcarbonsecondarilytothegraphitestructure.Inadditionthereisave ryintensepeakatthe 285.9eV.Thatpeakcanoccurfortwomajorsreasons.Oneisth eC=O, COOH and OHbonds[240{242]andtheotheristhepresenceofnitridesg roups,suchus NH 2 ,thatareattachedonabenzenering.Thesecondoptionaltho ughitsounds

PAGE 157

138 455 458 461 464 467 470 Binding Energy (eV) Raw Data Peaks Fit Peak at 458.4 eV Peak at 464.2 eV 458.4 eV 464.2 eVN(E)Figure6{17:TheTi2ppeakforthereferenceanatasenanopar ticles.Themajor peaksareatthe458.4eVfortheTi2p 1 = 2 andat464.2eVforthe Ti2p 3 = 2 whichareinagreementwithliteratureandseveraldatabase s. reasonable(HNO 3 wasusedforthepurication),isnotacceptablesincethesu rvey ofthesampledidnotrevealanynitrogen.Soconsequentlyth epeakhastobe attributedtoC=O, COOHand OHbonds.Thosegroupsareexpectedtobe presentaftertheacidtreatmentofthenanotubesaspartoft he COOHgroups thathavebeenformedonthesurfaceandareresponsiblefort hestabilizationof thenanotubesinasuspension.TheywerealsoconrmedbyFTI R(gure 4{6 ). Anotherveryinterestingpeakthatappearsinthespectrais theoneat289.7 eV.Thispeakisfarformtheelementalcarbonregionanditha stobeduetothe

PAGE 158

139 280 282 284 286 288 290 292 294 296 298 300 Binding Energy (eV)N(E) Raw data Fit Peak at 284.6 eV Peak at 285.9 eV 284.6 eV 285.9 eV Figure6{18:TheC1speakforthe s -ANTs.Themajorpeakisappearingtothe 284.6eV,whichisagainingreatagreementwithliteraturev alues. Thepeakat285.9eVischaracteristicoftheC Obondwhilethe 289.5eVpeakisattributedtoC O Ti. bondofcarbontoanotherelement.Intheliteraturetherear emanyreferences forthispeakmostofareaboutruoritebondeddirectlytocar bon[243]andsome metalslikeNaandLithatarealsodirectlybondedtocarbon[ 244].Therearesome referencesthatreportthispeakasanoxygenbond.Howevern oneoftheprevious reasonscangiveasatisfactoryexplanation.Sinceformthe Ramananalysisthere wasstrongevidenceofthebondoftheMWNTstotheTiO 2 itisbelievedthatthis peakhasthesameorigin.Sinceatthesameregionusuallyrep ortedtheC Metal

PAGE 159

140 527 529 531 533 535 537 539 541 Binding Energy (eV)N(E) Raw data Fit Peaks at 530.7 eV Peaks at 532.6 eV 530.6 eV 532.7 eV Figure6{19:TheO1sforthe s -ANTs.Themajorpeaksareagainat530.6eVfor theO1sforthelatticeoxygenandthe532.7eVforthesurfaceoxygen.Theratiobetweenthosetwopeaksrevealsthesurfac eareof theparticle. bondsasuitableoptionforthebondistheTi Cbond,whichhoweverappearsat 281.3eV.Asalreadydiscussed,ingure 6{12 therewasapeakat730cm 1 and itwasattributedtoC O Tibond.Soatthispoint,thereareevidences,strong enough,toattributethepeakat289.7eVtoapossibleC O Tibond.Again verydetailedanalyticalworkcouldprovetheconcept,whic hhoweverisbeyondthe purposeofthisresearch.

PAGE 160

141 ThenextimportantpeakisO1s(gure 6{19 ).Theshapeofthepeakis signicantlydierentcomparedtothelineshapeoftherefe rencematerial(gure 6{16 ).Itlookslikeasinglepeak.Therearetwodierentpossibi litiestotthat peak.Oneistoassumethatthereisonlyonepeakat530.6eVan dattributeit tothelatticeoxygen.Inthiscasethettingwillbedonewit htheVoigtequation toincludetheasymmetricbroadening.Thisapproachfailed togivereliable results.Theotherapproachistostartwiththeassumptiont hatthebroadening comesfromasecondoverlappingpeakwithlowerintensity,t hesurfaceoxygen. Thatismorereasonableapproachandyieldsnicet.Theback groundwasnot ttedproperlybutthatisbecauseweassumedpolynomialbac kgroundwhere inXPSitcanbemorecomplicated(Shirley).Thepeakappears at532.7eV.If weestimatethe I Surface =I bulk ratioisfoundtobe0.2526whereinthecaseofthe referencematerialitwas0.8307.Thismeansthatthenanopa rticleshavemore surfaceoxygen,somethingthatcontradictswiththeBETspe cicsurfacearea measurements,whichgavehighersurfaceareaforthe s -ANTs.Thiscontradiction howevercanbeexplainedonthecrystallographicorientati on,whichinthecaseof the s -ANTscanhaveorientationtoexposethesurfaceslessoxyge nrich. Themostimportantresulthowever,istheshift(comparedto thereference materialvalues)thathasoccurredforbothpeaks.Therstp eakat530.6eV, regardingthelatticeoxygen,isshiftedby+1 : 0eV(originalvalue529.6eV)and thesecondpeakthatisat532.7eV,surfaceoxygen,hasbeens hiftedby+1.1eV (originalvalueat531.5eV).Thisshiftagaincanbeattribu tedtobondingtoanon nativeelementwhichinthiscaseisC.Anothersourceofthis shiftcouldbethe dimensionsthatare6nm,whichislargeenoughtoeliminaten anosizedeects.But inthiscasetheshiftwouldoccurtowardslowerenergies.So itissafetoconclude thattheshiftisduetothebondbetweenTiO 2 andMWNTs.Thoseoxygenpeaks canhaveasignicantcontributionfromtheoxygenthatcome sfromthinlayerof

PAGE 161

142 455 458 461 464 467 470 Binding Energy (eV)N(E) Raw data Fit Peak at 459.4 eV Peak at 465.1 eV 459.4 eV 465.1 eV Figure6{20:TheTi2ppeakforthe s -ANTs.Themajorpeaksareatthe459.4eV fortheTi2p 1 = 2 andat465.1eVfortheTi2p 3 = 2 SiO 2 .Howeverthepeaksofthesiliconareveryweakandthecontri butionofthe SiO 2 oxygen,ifany,canbeneglected. Thenextpeakistheonefromtitanium(gure 6{20 ).Therearetwopeaks thatappearintitaniumandareatenergies459.4eVfortheTi 2p 1 = 2 andtheother 465.1eVTi2p 3 = 2 .Comparingthosepeakswiththepeaksatthereferencemater ial bothappearshifted.TheTi2p 1 = 2 isshiftedby1.0eVandtheTi2p 3 = 2 isshiftedby 1.1eV.Theshiftisagainsignicantandsincefortitaniump eaktheliteraturedoes notreportanysizeeectsthentheonlyreasonforthepeaksh iftisthebondtothe

PAGE 162

143 underlyinggraphite.SofromtheanalysisoftheXPSspectra fortheshortcarbon nanotubestherearestrongevidencesthatthereisbondbetw eenthetitaniaand theMWNTs.AndsincetheverycharacteristicpeakofC Tiisnotpresent,the bondshouldbeC O Ti. 6.5.4XPSofthe ` -ANTs InthissectionweexaminetheXPSspectrafromthe ` -ANTs.Againthe C1speakwasusedtocalibratethespectrasincetheSipeakis veryweak.The MWNTsusedinthissamplearedierentaswellasthetitaniap recursor.Butsince structurallythenalresultisnotverydierentitcanbeex pectedthatthetwo spectrawillbesimilar. StartingagainfromtheC1s(gure 6{21 )peakweseethemaingraphitepeak at284.6eV.Sincethemajoranalysisofthispeakisthesamea sinthecaseof the s -ANTs,onlythemajordierenceswillbeanalyzed.Inthisca sethepeak at285.2eVisslightlyshiftedcomparedtothepreviouscase (285.9eV).This hastodohoweverwiththeamountof COOHanditthereforeisrelatedto thetreatmentofthetubes.The s -CNTshavebeentreatedwithsulfuricacidin additiontothenitricacid.The ` -CNTsweretreatedonlywiththenitricacid. Thereforeitwasexpectedforthatpeaktobelessintensecom paredtothe s ANTs.Sincethescaleisinarbitraryunitsthepeakscannotb edirectlycompared buttherelativeheighttothemaincarbonpeakcan.Inthecas eofthe s -ANTs thatratiois I 1Cs =I C=0, COOH =1.1026andforthe ` -ANTsthatratiobecomes I 1Cs =I C=0, COOH =0.3273.Thatratiocanberelateddirectlytothenumberoft he COOHgroups,anditisaverystrongevidencethatthelongert ubeshaveless carboxylicgroupsonthesurface.Thenextimportantpeakis theonethatshows at289.7eV.Thisisalmostatthesamepositionasthepeaktha tintheprevious sectionwasattributedtotheC O Tipeak.Thepeakhereisbroaderthanbefore andalotlessintense.However,sincethereductionof COOHwasfollowedbythe

PAGE 163

144 280 285 290 295 300 Binding Energy (eV)N(E) Raw data Fit Peak 284.6 eV Peak 285.2 eV Peak 289.7 eV 286.4 eV 285.2 eV 289.7 eV Figure6{21:TheC1speakforthe ` -ANTs.Againthemajorpeakappearstobeat 284.6eVwhilethereisasecondarypeakat285.2eV.Thispeak is similartothecaseof s -ANTsthatappearsto285.9eV.Itisagain attributedtotheC ObondorC=Obond. 289.7eVpeakreductionitcanbeassumedthatthosetwopeaks arecloselyrelated. Soitisagainsafetoconcludethatthe289.7eVisindeedapea kthatcomesfrom theC O Tibond. Thefollowingtwopeaksareforthetitaniumandoxygen.Sinc e,asmentioned before,thetitaniainthissampleislessthanthe s -ANTstheintensityofthepeaks arelowerthanbeforeandthatcanbeseenfromthenoisetosig nalratio,which ishigher(gure 6{22 ).Butstillsomeimportantfeaturesarerecognizable.The

PAGE 164

145 528 530 532 534 536 538 540 542 Binding Energy (eV)N(E) Raw data Fit Peak at 530.9 eV Peak at 532.7 eV 530.9 eV 532.7 eV Figure6{22:TheO1speakforthe ` -ANTs.Therearealsotwopeaksobservedat 532.7eVandat530.9eV.Althoughbotharefromtheoxygenthe532.7eVisattributedtosurfaceoxygenwhiletheothercome sfrom latticeoxygencontribution.Relativetothecaseof s -ANTsthe surfaceoxygenandthereforethesurfaceareaishigher,som ething thatwasconrmedwithBETaswellandisinagreementwithoth er researchers. peaksinthiscasearealsosignicantlyshiftedcomparedto thereferencematerial. Thepeakthatcomesfromthesurfaceoxygenisat532.7eVloca tedatthesame energyasthesurfaceoxygenpeakforthe s -ANTs.Thesecondpeak,regarding thelatticeoxygen,isatthe530.9eVandisveryclosewheret herespectivepeak forthe s -ANTsis(530.6eV).AgainthecontributionoftheSiO 2 inthisspectrum

PAGE 165

146 455 458 461 464 467 470 Binding Energy (eV)N(E) Raw data Peak Fit Peak 459.6 eV Peak 465.3 eV 459.6 eV 465.3 eV Figure6{23:TheTi2ppeakforthe ` -ANTs.Themajorpeaksareatthe459.6eV fortheTi2p 1 = 2 andat465.2eVfortheTi2p 3 = 2 whicharein signicantlyshiftedcomparedtothereferencematerial. isnegligiblesotheintensityofthepeaksisattributedalm ostexclusivelyfromthe TiO 2 peaks.Theotherveryimportantresultistherelativeinten sityofthetwo peaks.Theratio I Surface =I bulk ratioisfoundtobe1.5636whereinthecaseofthe referencematerialitwas0.8307andinthecaseofthe s -ANTsthatwas0.2526. Thatisinagreementwithresearchersthatreportthatamong severalprecursors theTi 2 (SO 4 ) 3 yieldshighersurfacearea.Therelativehighnoisetosigna lratiodid notallowforgoodtofthebackgroundbutthepeakswerevery nicelyttedwith

PAGE 166

147 Table6{9:SummaryoftheXPSpeaks Peak Reference s -ANTs ` -ANTs Peak[eV] Peak[eV]Shift[eV] Peak[eV]Shift[eV] C1s Graphite 284.6 284.60.0 284.60.0 C1s C O ,C=O N/A 285.9C1s C O Ti N/A 289.5O1s Surface 531.5 532.71.1 532.71.1 O1s Bulk 529.6 530.61 530.91.3 Ti2p 1 = 2 458.4 459.41. 459.61.2 Ti2p 3 = 2 464.2 465.10.9 465.31.1 theGauss.Stillthemainresultofthosepeaksremainsthesh iftofthepeaksto higherenergies. ThenalpeakisagaintheTi2p(gure 6{23 ).Themajorpeaksareatthe 459.6eVfortheTi2p 1 = 2 andat465.2eVfortheTi2p 3 = 2 thatareveryclosetothe respectivevaluesofthe s -ANTs(459.4eVand465.1eVrespectively).Againitis obviousthatthenoiseisslightlyincreasedcomparedtothe referencematerialand the s -ANTsduetotherelativelessamountoftitaniainthesample .Butoverall thepeakshifts,1.2eVfortheTi2p 1 = 2 and1eVfortheTi2p 3 = 2 ,isdenotingagain thatthereisabondbetweenTiO 2 andMWNT. 6.6SummaryoftheXPSAnalysis ThelastsectionofthischapterwasdevotedintheXPSanalys isofboth the ` -ANTsand s -ANTs.TheXPSconrmedtheresultsoftheRaman.Allthe peaksshoweddisplacementcomparedtothereferencemateri al(table 6{9 ).Since thegrainsizewasnotsignicantlydierentthoseshiftsca nbeattributedtothe bondoftitaniaonthecarbonnanotube.Inadditiontotheshi ft,anewpeak thatisnotexplainedreasonablefromthedatabases,appear edatapproximately 289.7eV.ItisnotaccuratetoattributethispeaktotheC O Tibond.The combination,howeverofRamanandXPScanleadtosuchaconcl usion,whichcan bebackedupfromtheoreticalcalculations.Wecansafelyco ncludethereforethat theTiO 2 coatingisbondedtotheMWNTs.Furthermorethebondisinthe form

PAGE 167

148 ofC O Tibond.Thisisdirectlyrelatedtotheproductionprocess. The COOH and OHgroupshavebeensuccessfullyusedasanchoringpointsdu ringthe sol-gel process.

PAGE 168

149 Tip2 O1s ReferenceMaterial 528 530 532 534 536 538 540 542 Binding Energy (eV)N(E) Raw data Peak fitting Peak at 529.6 eV Peak at 531.6 eV 529.6 eV 531.5 eV 455 458 461 464 467 470 Binding Energy (eV) Raw Data Peaks Fit Peak at 458.4 eV Peak at 464.2 eV 458.4 eV 464.2 eVN(E)LongANTs 528 530 532 534 536 538 540 542 Binding Energy (eV)N(E) Raw data Fit Peak at 530.9 eV Peak at 532.7 eV 530.9 eV 532.7 eV 455 458 461 464 467 470 Binding Energy (eV)N(E) Raw data Peak Fit Peak 459.6 eV Peak 465.3 eV 459.6 eV 465.3 eV Figure6{24:CollectiverepresentationiftheXPSdatarega rdingthecoatedlong carbonnanotubes.TheupperrowistheTi2pandO1speakofthereferencematerialandthelowerrowisthedataobtainedbyt he s -ANTs.Theshiftsinbothpeaksareobviousandaresummarize din table 6{9

PAGE 169

150 Tip2 O1s ReferenceMaterial 528 530 532 534 536 538 540 542 Binding Energy (eV)N(E) Raw data Peak fitting Peak at 529.6 eV Peak at 531.6 eV 529.6 eV 531.5 eV 455 458 461 464 467 470 Binding Energy (eV) Raw Data Peaks Fit Peak at 458.4 eV Peak at 464.2 eV 458.4 eV 464.2 eVN(E)ShortANTs 527 529 531 533 535 537 539 541 Binding Energy (eV)N(E) Raw data Fit Peaks at 530.7 eV Peaks at 532.6 eV 530.6 eV 532.7 eV 455 458 461 464 467 470 Binding Energy (eV)N(E) Raw data Fit Peak at 459.4 eV Peak at 465.1 eV 459.4 eV 465.1 eV Figure6{25:CollectiverepresentationiftheXPSdatarega rdingthecoatedshort carbonnanotubes.TheupperrowistheTi2pandO1speakofthereferencematerialandthelowerrowisthedataobtainedbyt he ` -ANTs.Theshiftsinbothpeaksareobviousandaresummarize din table.

PAGE 170

151 Tip2 O1s ShortANTs 527 529 531 533 535 537 539 541 Binding Energy (eV)N(E) Raw data Fit Peaks at 530.7 eV Peaks at 532.6 eV 530.6 eV 532.7 eV 455 458 461 464 467 470 Binding Energy (eV)N(E) Raw data Fit Peak at 459.4 eV Peak at 465.1 eV 459.4 eV 465.1 eV LongANTs 528 530 532 534 536 538 540 542 Binding Energy (eV)N(E) Raw data Fit Peak at 530.9 eV Peak at 532.7 eV 530.9 eV 532.7 eV 455 458 461 464 467 470 Binding Energy (eV)N(E) Raw data Peak Fit Peak 459.6 eV Peak 465.3 eV 459.6 eV 465.3 eV Figure6{26:CollectiverepresentationiftheXPSdatarega rdingthecoatedlong andshortcarbonnanotubes.TheupperrowistheTi2pandO1sp eak ofthe s -ANTsandthelowerrowisthedataobtainedbythe ` -ANTs. Thepeaksaresimilarregardingtheposition,butaresigni cantly dierentinshape.

PAGE 171

152 C1s ReferenceMaterial 280 282 284 286 288 290 292 294 296 298 300 Binding Energy (eV)N(E) Raw Data Fit Peak at 284.6 eV Peak at 288.4 eV 284.6 eV 288.4 eV 280 285 290 295 300 Binding Energy (eV)N(E) Raw data Fit Peak 284.6 eV Peak 285.2 eV Peak 289.7 eV 286.4 eV 285.2 eV 289.7 eV ShortANTs 280 282 284 286 288 290 292 294 296 298 300 Binding Energy (eV)N(E) Raw Data Fit Peak at 284.6 eV Peak at 288.4 eV 284.6 eV 288.4 eV LongANTs 280 282 284 286 288 290 292 294 296 298 300 Binding Energy (eV)N(E) Raw data Fit Peak at 284.6 eV Peak at 285.9 eV 284.6 eV 285.9 eV Figure6{27:TheC1speakofthepeakofthecoatedcarbonnano tubes(both ` -ANTsand s -ANTs)andthereferencematerial.Themaindierence betweenthereferencematerialandthesamplesarethepeaksregardingtheC OandC=Obonds,thatareappearingonlyforthe s -ANTsand ` -ANTs,andthepeakat289.7eV( ` -ANTs)and289.5 eV( s -ANTs)thatcanbeattributedtotheC O Tibond.

PAGE 172

153 Si2p ReferenceMaterial 96 98 100 102 104 106 108 Binding Energy (eV)N(E) Raw Data Peak Fit Peaks at 98.5 eV Peak at 102.5 eV 102.5 eV 98.5 eV 96 98 100 102 104 106 108 Binding Energy (eV)N(E) Raw data 98.6 eV 102.4 eV ShortANTs 96 98 100 102 104 106 108 Binding Energy (eV)N(E) Raw Data Peak Fit Peaks at 98.5 eV Peak at 102.5 eV 102.5 eV 98.5 eV LongANTs 96 98 100 102 104 106 108 Binding Energy (eV)N(E) Raw data 98.6 eV 102.5 eV Figure6{28:TheSi2ppeakofthepeakofthecoatedcarbonnan otubes(both ` -ANTsand s -ANTs)andthereferencematerial.Althepeaksareat thesameenergy,butthenoisetosignalratioisalothigherf orthe both ` -ANTsand s -ANTs.Thereasonforthatisthethicknessofthe coating.ThecoatedMWNTsweredepositedinathickerlayer.

PAGE 173

154 GBand DBand AcidTreated s -CNTs 12001220124012601280130013201340136013801400 Raman Shift(cm 1 )Intensity (a.u.) 15001520154015601580160016201640166016801700 Raman Shift(cm 1 )Intensity (a.u.) CNTsSegmentof s -ANTs 12001220124012601280130013201340136013801400 Raman Shift (cm 1 )Intensity (a.u.) 15001520154015601580160016201640166016801700 Raman Shift (cm 1 )Intensity (a.u.) Figure6{29:CollectiverepresentationoftheRamanspectr aregardingtheshort nanotubesbefore(toprow)andafterthecoating(bottomrow ).The rightcolumnisfortheGbandandtheleftcolumnisfortheDba nd.

PAGE 174

155 GBand DBand AcidTreated ` -CNTs 12001220124012601280130013201340136013801400 Raman Shift(cm 1 )Intensity (a.u.) 15001520154015601580160016201640166016801700 Raman Shift(cm 1 )Intensity (a.u.) CNTsSegmentof ` -ANTs 12001220124012601280130013201340136013801400 Raman Shift (cm 1 )Intensity (a.u.) 15001520154015601580160016201640166016801700 Raman Shift (cm 1 )Intensity (a.u.) Figure6{30:CollectiverepresentationoftheRamanspectr aregardingthelong nantubesbefore(toprow)andafterthecoating(bottomrow) .The rightcolumnisfortheGbandandtheleftcolumnisfortheDba nd.

PAGE 175

156 ` -ANTs s -ANTs ReferenceMaterial 0 100200300400500600700800900 1000 Raman Shift (cm 1 )Intensity (a.u.) E g 148 cm 1 E g 197 cm 1 B 1g 399 cm 1 A 1g 515 cm 1 E g 639 cm 1 0 100200300400500600700800900 1000 Raman Shift (cm 1 )Intensity (a.u.) E g 148 cm 1 E g 197 cm 1 B 1g 399 cm 1 A 1g 515 cm 1 E g 639 cm 1 CNTsSegmentof ` -ANTs 01002003004005006007008009001000 Raman Shift (cm 1 )Intensity (a.u.) Raw data Fit Peak at 150 cm 1 Peak at 202 cm 1 Peak at 393 cm 1 Peak at 510 cm 1 Peak at 633 cm 1 01002003004005006007008009001000 Raman Shift ( cm 1 )Intensity (a.u.) Raw data Fit Peak at 150 cm 1 Peak at 409 cm 1 Peak at 629 cm 1 Figure6{31:CollectiverepresentationiftheXPSdatarega rdingthecoatedlong carbonnanotubes.TheupperrowistheTi2pandO1speakofthereferencematerialandthelowerrowisthedataobtainedbyt he s -ANTs.Theshiftsinbothpeaksareobviousandaresummarize din table.

PAGE 176

CHAPTER7 CONCLUSIONSANDFUTUREWORK Themainobjectiveofthisresearch,asoutlinedintheintro duction,isto combinethetwodierentmaterials,MWNTsandTiO 2 ,inonecompositethat willdeliverhighphotocatalyticeciency.Thisnewcompos itewilltakeadvantage oftheexcellentelectronicpropertiesandhighspecicsur facearea.Ingeneral photocatalysiscanbeimprovedbytheincreasingthesurfac earea,orbyimproving the[OH ].Thelaterisdirectlycorrelatedtotherateatwhichthe e and h + are generatedandrecombined.Thisratecanbemathematicallye xpressedas quant. / k CT k CT + k R (7 1) where k CT isthechargetransferrate,andthe k R istherecombinationrate.So byminimizingtherecombinationrate( k R 0)thequantumeciencywill increase(lim k R 0 k CT k CT + k R =1).AsseeninChapter5,ifshieldingandcoagulation phenomenaareneglected,theeciencydependencetosurfac eareaisjustalinear relationship. surf. / S (7 2) Theoveralleciencywillbe tot. / k CT k CT + k R S (7 3) Inordertomaximizetheoveralleciencyitisnecessarytom inimizetherecombinationrateandincreasethesurfacearea.Thewaystominimi zetherecombination ratehavebeenalreadyexplainedandtheyare;theincorpora tionoftransition metals(Cu +3 ,Cr +3 andFe +3 ),NorCinthecrystalstructureoftitaniaandthe 157

PAGE 177

158 Table7{1:Electronanityandworkfunctionformetalsused tocreaterectifying contactwithtitaniainordertoincreasethephotocatalyti ceciency. ElementworkFunction( )[eV]ElectronAnity( )[eV] Pt5.552.128 Au5.382.309 Ag4.631.302 Al4.170.441 C(amorphous)5.001.263 C 60 7.742.780(2.650 0.020) y SWNT(9,0) 5.10SWNT(5,5) ? 4.7802.840-2.660 MWNT4.80-5.05y Experimentalvalue Conductingzig-zag ? Conductingarmchair couplingwithametal.Accordingtothetheoryofphotocatal ysis,workfunctionis acriticalparametertothecreationoftherectifyingconta ct.Table 7{1 compares theworkfunctionofthenanotubestotheworkfunctionofoth ertraditionalmetals amongwhicharePtandAu,bothusedtoimprovephotocatalysi s.Carbonnanotubesarestandingthecomparisonverywell,sincetheyare slightlybellowAu. Thereforetheutilizationofcarbonnanotubesasthecoreof thephotocatalytic compositeisexpectedtoenhancethephotocatalysissincei thastheabilityto increasetheeciencybybothmethodsmentionedearlier,hi ghspecicareaand metallicproperties. Thephotocatalyticdegradationexperimentsthatwerecarr iedout(chapter5) demonstratedthevalidityofthisassumption.Theaddition of1mgofnanotubes inthesolutionof3mgofanatasenanoparticles,enhancedth eeciencybynearly doublingthereactionrate.Furthermorethe ` -ANTsshowedexceptionalphotocatalyticpropertiescomparedtotheMWNTs/TiO 2 nanoparticlesmixtureandeven comparedtoDegussaP25.Howeverthe s -ANTsdisplayedpoorphotocatalytic activity.

PAGE 178

159 Duringthesynthesisofthetitaniacoatedcarbonnanotubes COOHgroups weregeneratedonthesurfaceofthetubesbytheacidtreatme nt.Thosegroupsact laterasanchoringpointsforthe sol-gel precursors.Thisisthefundamentalreason whyabondbetweenthecarbonnanotubesandthetitaniacoati ngisformed.The bondwasconrmedbyRamanspectroscopy,whichindicatedas ignicantshiftof thetitaniaandnanotubespeaks,andbyXPS,whichalsodispl ayedpeakshiftsin additiontoanewpeakat289 : 6 0 : 1eVwhichisattributedtheC O Ti.Since bothtechniquesshowedtheexistenceofthebondbetweenTiO 2 andMWNTsitis accuratetoconcludethatthisbondexistintheformofC O Ti. Thecharacterizationofthenanotubesinchapter4,beforet hecoating, revealedthatbothtypesofnanotubes(longandshort)havea concentricstructure, butthe s -CNTshadsignicantlymoredamagedstructure,whichwilla ect primarilytheirelectricproperties.Inchapter6theRaman spectraveriedthis hypothesis.TheGBanddidnotdemonstrateadistinctsplit, whichisaverydirect indicationfortheabsenceofmetallicproperties.The ` -CNTsonthecontrary,not onlyshowedthattheyhavewelldenedstructure,butinaddi tiontheGBand splitwasverydistinct.TheG bandwasbetterapproximatedwiththeBreitWigner-Fanopeakmodelwhichfurtherjustiesthevalidity thisargumentabout themetallicnatureofthenanotubes. Themostimportantexperimentalresultsofthisworkcanbes ummarizedat thefollowingpoints:ˆTheMWNTscanenhancedthephotocatalysisbehaviorˆTheTiO 2 coatingwasbondedontheMWNTsˆThe ` -CNTsweremetallicwhilethe s -CNTsdidnothaveanyindicationof similarproperties.

PAGE 179

160 7.1Conclusions Accordingtothepreviousdiscussionthefollowingconclus ionscanbesummarized.ˆThehighworkfunctionofthenanotubesandtheconductingpr opertiesisthe mainreasonthatthenantubescanassistthephotocatalysis whentheyarein colloidalsuspension.ˆApplyingtheTiO 2 ascoatingonthecarbonnanotubes,yieldveryhigh photocatalyticeciency.Thisisduetothebond(C O Ti)thatiscreated betweentheMWNTsandtheTiO 2 .Thebondmakestheunderlinedcarbon atomsdopantstothestructureoftitania.ˆThemetallicnatureofthecarbonnanotubesismorecritical thanthebond. Bothsamplespreparedandtestedhere( ` -ANTsand s -ANTs)displayed thesameevidencesfortheC O Tibond.Howeverthe s -ANTsdidnot hadconductingproperties,andthereforetheyhadverypoor photocatalytic activity.ˆOverallcarbonintheformofcarbonnanotubescanbeaverypr omisingway toenhancethephotocatalyis.Forthistohappen,thecarbon nanotubesmust beverywelldenedwithdistinctstructureandgoodelectri calproperties. 7.2FutureWork Theconceptsexplainedandinvestigatedinthisresearchar ebasedalmost exclusivelyonexperimentalresultsItisthereforenecess arytoinvestigatethemain principlesontheoreticalbase.OneofthemishowtheXPSpea kswillshiftand wheretheC O Tipeakwillappear.Toderivethisinformationitisrequire dto knowtheelectronicstructureofthecompositematerial,so methingthatcurrently canbederivedwithcomputersimulations.Inadditionthedi rectionandthe amountoftheshiftsinthetitaniaandMWNTspeaksandtheapp earanceofthe C O TiintheRamanspectraneedstobetheoreticallyexplained.

PAGE 180

161 Fromanexperimentalperspectivetheresultsfromthisdiss ertationcanbe appliedinmanyways.Sincetitaniahassuchawiderangeofus es,thisresearchcan bethefoundationformanyapplications.Themostimmediate workthatcanbe done,istotestthesecompositeparticlesonawiderangeofb acteria,sporesand otherbiologicalcontaminants,andexaminetheinteractio ns.Anotherapplicationis tocombinethelargeknowledgebaseregardingthesolarcell applicationoftitania toproducecellwithveryhigh,energyconversion.Inamoree ngineeringapproach, waystomassproducetheproductandcommercializetheprodu ctcanbesought. Thishastobedone,however,inrespecttotherecentlyraise dpotentialissues aboutthetoxicityofthenanotubes.

PAGE 181

APPENDIXA MATHEMATICAALGORITHMUSEDFORTHELOESSMETHOD Needs["Statistics`ContinuousDistributions`"];DataRange[x_]:={Min[x],Max[x]}LoessFit[(x_)?NumberQ,data_,\[Alpha]_:0.75,\[Lambda ]_:1]:= WLSFit[data,LoessWts[x,data,\[Alpha]],\[Lambda],x] LoessFit[(x_)?VectorQ,data_,\[Alpha]_:0.75,\[Lambda ]_:1]:= Table[LoessFit[x[[i]],data,\[Alpha],\[Lambda]],{i,L ength[x]}] SLPlot[fits_,res_,p_:0.5,\[Alpha]_:1]:= Module[{a,f,r,data2,s,lines}, data2=Sort[Transpose[{fits,res}],First[#1]All,Axes->False,Frame->Tr ue, FrameLabel->{"fit","Abs[res]^p"},PlotStyle->{PointSize[0.02],RGBColor[0,0,1]},Epilog->{RGBColor[0,1,0],Thickness[0.02],lines}];BW Plot[a-s]] RobustLoessFit::MaximumReached="ThemaximumnumberofiterationsoftheIRWLSalgorithm,asspecifiedbytheoptionMaxIterations,hasbeenreachedand without convergenceofthealgorithm.YoucouldtryincreasingMaxIterations.";RobustLoessFit[data_,\[Alpha]_:0.75,\[Lambda]_:1,(o pts___)?OptionQ]:= Module[{x,y,\[Delta],res,rsum=0,data2,iter=0,rprev= 1, r=Table[1,{Length[data]}]}, maxiter=MaxIterations/.{opts}/.Options[RobustLoessF it]; data2=Sort[data,First[#1]0.001, res=Table[y[[i]]-WLSFit[data,\[Delta][[i]]*r,\[Lamb da],x[[i]]], {i,1,Length[x]}];r=BiSquare[res/(6*Median[Abs[res]] )]; 162

PAGE 182

163 rprev=rsum;rsum=Abs[Plus@@r;]]; If[iter==maxiter,Message[RobustLoessFit::MaximumRea ched]]; y-res]Options[RobustLoessFit]={MaxIterations->25}LoessSummary[data_,\[Alpha]_:0.75,\[Lambda]_:1]:= With[{res=LoessResiduals[data,\[Alpha],\[Lambda]]}, {res->res,\[Sigma]->Sqrt[Plus@@(res^2)/Length[res]] \[Mu]->1.199999999999999*(\[Lambda]+1)/\[Alpha]}] LoessResiduals[data_,\[Alpha]_:0.75,\[Lambda]_:1]:= Last[Transpose[data]]-LoessFit[First[Transpose[data ]],data,\[Alpha], \[Lambda]] RobustLoessPlot[data_,\[Alpha]_:0.6,\[Lambda]_:1,op ts___]:= Module[{x,y,fits,data2,lines}, data2=Sort[data,First[#1]All,Axes->False,Frame->Tr ue, PlotStyle->{PointSize[0.02],RGBColor[0,0,1]},Epilog->{RGBColor[0,1,0],Thickness[0.02],lines},opt s]] LoessPlot[data_,\[Alpha]_:0.75,\[Lambda]_:1,numvalu es_:30,opts___]:= With[{x=EquispaceVector[First[Transpose[data]],numv alues]}, ListPlot[data,opts,PlotStyle->{PointSize[0.05]}, PlotRange->ScaleRectangle[data],Frame->True,Axes->F alse, Epilog->{Thickness[0.02],RGBColor[0,1,1], Line[Transpose[{x,LoessFit[x,data,\[Alpha],\[Lambda ]]}]]},opts]] WLSFit[data_,wts_,ldegree_:1,x_]:= Fit[Transpose[(wts*#1&)/@ Join[{Table[1,{Length[data]}]},Transpose[data]]], Join[{u},Table[v^i,{i,ldegree}]],{u,v}]/.{u->1,v->x } LoessWts[x_,data_,\[Alpha]_]:= Tricube[(x-First[Transpose[data]])/LoessDistance[x, data,\[Alpha]]] Tricube=Compile[{{x,_Real,1}},If[Abs[#]<1,(1-#^2)^2 ,0]&/@x]; Bisquare=Compile[{{x,_Real,1}}, If[Abs[#]<1,(1-Abs[#]^3)^3,0]&/@x]; LoessDistance[x_,data_,\[Alpha]_]:=

PAGE 183

164 Module[{A=Max[1,\[Alpha]],X=First[Transpose[data]], q}, q=Min[Length[X],Ceiling[\[Alpha]*Length[X]]];A*Sort [Abs[X-x]][[q]]] EquispaceVector[(x_)?VectorQ,numvalues_:30]:= Range[Min[x],Max[x],N[(Max[x]-Min[x])/numvalues]] ScaleRectangle[data_]:= With[{x=Transpose[data]},(AddEps[#1]&)/@ {DataRange[First[x]],DataRange[Last[x]]}] AddEps[{xlo_,xhi_}]:= With[{\[Epsilon]=(xhi-xlo)*0.05},{xlo-\[Epsilon],xh i+\[Epsilon]}] BWPlot[data_]:= Module[{datadim,k,datapts,whiskers,box,outsidepts,m edpt,dmax,dmin, coldata,Q1,Q3,uplim,dnlim,outside,jitter,drange,eps ilon, boxwidth=0.4},datadim=Dimensions[data];k=If[Length[datadim]==2,datadim[[2]],1];datapts=outsidepts=whiskers=box=medpt=dmax=dmin={};Do[coldata=If[k==1,data,Column[data,i]]; datapts= Join[datapts,Transpose[{coldata,Table[i,{Length[col data]}]}]]; medpt=Join[medpt,{PointSize[0.04], Point[{Quantile[coldata,0.5],i}]}]; Q1=Quantile[coldata,0.25];Q3=Quantile[coldata,0.75] ; box=Join[box,{RGBColor[0.690207,0.7685929999999999, 0.870602], Polygon[{{Q1,i-boxwidth},{Q1,i+boxwidth}, {Q3,i+boxwidth},{Q3,i-boxwidth}}]}]; step=1.5*(Q3-Q1);uplim=Q3+step;dnlim=Q1-step;upadj=Max[Select[coldata,#1<=uplim&]];dnadj=Min[Select[coldata,#1>=dnlim&]];whiskers= Join[whiskers,{Thickness[0.005],Line[{{dnadj,i},{Q1 ,i}}], Line[{{Q3,i},{upadj,i}}]}]; outside= Join[Select[coldata,#1>uplim&],Select[coldata,#1{Automatic,None}, PlotStyle->{AbsolutePointSize[0]},PlotRange->{drang e,{0,k+1}}, Axes->{True,False},Epilog->{outsidepts,box,medpt,wh iskers}]]

PAGE 184

APPENDIXB RAMANPEAKSOFCNTs TableB{1:PropertiesofthevariousRamanfeaturesingraph iteandSWNTs. Name (cm 1 )Resonanced /d E L iTA288DR1129LA453DR1216RBM248/ d t SR0 IFM 750DR2 220 oTO860DR10iFM + 960DR2180 D1350DR153LO1450DR10BWF1550SR0G1582SR 0 M 1732DR2 26 M + 1755DR20 iTOLA1950DR2230G'2700DR21062LO2900DR202G3180DR20 Adaptedfrom[245]. 165

PAGE 185

REFERENCES [1]G.Ertl,\Heterogeneouscatalysisontheatomicscale," TheJapanChemical JournalForum ,vol.1,no.1,pp.33{45,2001. [2]J.Liqiang,S.Xiaojun,S.Jing,C.Weimin,X.Zili,D.Yao guo,andF.Honggang,\Reviewofsurfacephotovoltagespectraofnano-size dsemiconductor anditsapplicationsinheterogeneousphotocatalysis," SolarEnergyMaterials andSolarCells ,vol.79,no.2,pp.133{151,2003. [3]A.Fujishima,T.N.Rao,andD.A.Tryk,\Titaniumdioxide photocatalysis," JournalofPhotochemistryandPhotobiologyC:Photochemis tryReviews ,vol. 1,no.1,pp.1{21,2000. [4]M.R.Homann,S.T.Martin,W.Y.Choi,andD.W.Bahnemann ,\Environmentalapplicationsofsemiconductorphotocatalysis, ChemicalReviews vol.95,no.1,pp.69{96,1995. [5]A.L.Linsebigler,G.Q.Lu,andJ.T.Yates,\Photocataly sisonTiO 2 surfaces-principles,mechanisms,andselectedresults," ChemicalReviews vol.95,no.3,pp.735{758,1995. [6]A.FujishimaandK.Honda,\Electrochemicalphotolysis ofwaterata semiconductorelectrode," Nature ,vol.238,no.5358,pp.37{40,1972. [7]K.P.Kuhn,I.F.Chaberny,K.Massholder,M.Stickler,V. W.Benz,H.G. Sonntag,andL.Erdinger,\Disinfectionofsurfacesbyphot ocatalytic oxidationwithtitaniumdioxideanduvalight," Chemosphere ,vol.53,no.1, pp.71{77,2003. [8]T.Matsunaga,Y.Namba,andT.Nakajima,\Electrochemic alsterilizationof microbial-cells," BioelectrochemistryandBioenergetics ,vol.13,no.4-6,pp. 393{400,1984. [9]T.Matsunaga,R.Tomoda,T.Nakajima,andH.Wake,\Photo electrochemicalsterilizationofmicrobial-cellsbysemiconduct orpowders," Fems MicrobiologyLetters ,vol.29,no.1-2,pp.211{214,1985. [10]A.G.RinconandC.Pulgarin,\Photocatalyticalinacti vationof E.Coli : eectof(continuous-intermittent)lightintensityandof (suspended-xed) TiO 2 concentration," AppliedCatalysisB-Environmental ,vol.44,no.3,pp. 263{284,2003. 166

PAGE 186

167 [11]A.G.RinconandC.Pulgarin,\Bactericidalactionofil luminatedTiO 2 onpureescherichiacoliandnaturalbacterialconsortia:p ost-irradiation eventsinthedarkandassessmentoftheeectivedisinfecti ontime," Applied CatalysisB-Environmental ,vol.49,no.2,pp.99{112,2004. [12]A.G.RinconandC.Pulgarin,\EectofpH,inorganicion s,organicmatter andH 2 O 2 on E-Coli K12photocatalyticinactivationbyTiO 2 -Implications insolarwaterdisinfection," AppliedCatalysisB-Environmental ,vol.51,no. 4,pp.283{302,2004. [13]A.G.RinconandC.Pulgarin,\Fieldsolare-coliinacti vationinthe absenceandpresenceofTiO 2 :isUVsolardoseanappropriateparameterfor standardizationofwatersolardisinfection?," SolarEnergy ,vol.77,no.5,pp. 635{648,2004. [14]A.G.Rincon,C.Pulgarin,N.Adler,andP.Peringer,\In teractionbetween E.Coli inactivationandDBP-precursors-dihydroxybenzeneisome rs-inthe photocatalyticprocessofdrinking-waterdisinfectionwi thTiO 2 ," Journalof PhotochemistryandPhotobiologya-Chemistry ,vol.139,no.2-3,pp.233{241, 2001. [15]T.Sato,Y.Koizumi,andM.Taya,\Photocatalyticdeact ivationofairborne microbialcellsonTiO 2 -loadedplate," BiochemicalEngineeringJournal ,vol. 14,no.2,pp.149{152,2003. [16]J.Muscat,V.Swamy,andN.M.Harrison,\First-princip lescalculationsof thephasestabilityofTiO 2 ," PhysicalReviewB ,vol.65,no.22,pp.{,2002. [17]T.MitsuhashiandO.J.Kleppa,\Transformationenthal piesoftheTiO 2 polymorphs," JournaloftheAmericanCeramicSociety ,vol.62,no.7-8,pp. 356{357,1979. [18]A.Navrotsk,J.C.Jamieson,andO.J.Kleppa,\Enthalpy oftransformation ofahigh-pressurepolymorphoftitaniumdioxidetorutilem odication," Science ,vol.158,no.3799,pp.388{&,1967. [19]H.Arashi,\Raman-spectroscopicstudyofthepressure -inducedphasetransitioninTiO 2 ," JournalofPhysicsandChemistryofSolids ,vol.53,no. 3,pp.355{359,1992. [20]J.F.Mammone,M.Nicol,andS.K.Sharma,\Raman-spectr aofTiO 2 II,TiO 2 -III,SnO 2 ,andGeO 2 athigh-pressure," JournalofPhysicsand ChemistryofSolids ,vol.42,no.5,pp.379{384,1981. [21]J.F.Mammone,S.K.Sharma,andM.Nicol,\Raman-studyo frutile(TiO 2 ) athigh-pressures," SolidStateCommunications ,vol.34,no.10,pp.799{802, 1980.

PAGE 187

168 [22]L.S.Dubrovinsky,N.A.Dubrovinskaia,V.Swamy,J.Mus cat,N.M. Harrison,R.Ahuja,B.Holm,andB.Johansson,\Materialssc ience-the hardestknownoxide," Nature ,vol.410,no.6829,pp.653{654,2001. [23]K.M.GlassfordandJ.R.Chelikowsky,\Optical-proper tiesoftitaniumdioxideintherutilestructure," PhysicalReviewB ,vol.45,no.7,pp. 3874{3877,1992. [24]K.M.Glassford,N.Troullier,J.L.Martins,andJ.R.Ch elikowsky,\Electronicandstructural-propertiesofTiO 2 intherutilestructure," SolidState Communications ,vol.76,no.5,pp.635{638,1990. [25]S.D.MoandW.Y.Ching,\Electronicandoptical-proper tiesof3phasesof titanium-dioxide-rutile,anatase,andbrookite," PhysicalReviewB ,vol.51, no.19,pp.13023{13032,1995. [26]M.Calatayud,P.Mori-Sanchez,A.Beltran,A.M.Pendas ,E.Francisco, J.Andres,andJ.M.Recio,\Quantum-mechanicalanalysisof theequation ofstateofanataseTiO 2 ," PhysicalReviewB ,vol.64,no.18,pp.184113:1{ 184113:4,2001. [27]W.Gopel,J.A.Anderson,D.Frankel,M.Jaehnig,K.Phi llips,J.A.Schafer, andG.Rocker,\Surface-defectsofTiO 2 (110)-acombinedXPS,XAESand ELSstudy," SurfaceScience ,vol.139,no.2-3,pp.333{346,1984. [28]S.HufnerandG.K.Wertheim,\Core-electronsplitting sandhyperne eldsintransition-metalcompounds," PhysicalReviewB ,vol.7,no.6,pp. 2333{2336,1973. [29]S.P.Kowalczyk,L.Ley,F.R.Mcfeely,andD.A.Shirley, \X-Rayphotoelectronspectroscopicstudiesofelectronic-structur eoftransition-metal diruorides," PhysicalReviewB ,vol.15,no.10,pp.4997{5002,1977. [30]S.P.Kowalczyk,F.R.Mcfeely,L.Ley,V.T.Gritsyna,an dD.A.Shirley, \Electronic-structureofSrTiO 3 andsomesimplerelatedoxides(MgO,Al 2 O 3 SrO,TiO 2 )," SolidStateCommunications ,vol.23,no.3,pp.161{169,1977. [31]D.W.Fischer,\X-Raybandspectraandmolecular-orbit alstructureofrutile TiO 2 ," PhysicalReviewB ,vol.5,no.11,pp.4219{&,1972. [32]K.Tsutsumi,O.Aita,andK.Ichikawa,\X-RayTiKspectr aandband structuresofoxidesoftitanium," PhysicalReviewB ,vol.15,no.10,pp. 4638{4643,1977. [33]L.A.GrunesandJ.J.Ritsko,\Valenceandcoreexcitati on-spectrainK,Rb, andCsalkali-metalstage-1intercalatedgraphite," PhysicalReviewB ,vol.28, no.6,pp.3439{3446,1983.

PAGE 188

169 [34]G.VanderLaan,\PolaronicsatellitesinX-Ray-absorp tionspectra," Physical ReviewB ,vol.41,no.17,pp.12366{12368,1990. [35]R.Brydson,H.Sauer,W.Engel,J.M.Thomas,E.Zeitler, N.Kosugi,and H.Kuroda,\Electron-EnergyLossandX-RayAbsorption-Spe ctroscopy ofrutileandanatase-atestofstructuralsensitivity," JournalofPhysicsCondensedMatter ,vol.1,no.4,pp.797{812,1989. [36]L.A.Grunes,R.D.Leapman,C.N.Wilker,R.Homann,and A.B.Kunz, \Oxygen-knear-edgene-structure-anelectron-energy-l ossinvestigation withcomparisonstonewtheoryforselected3dtransition-m etaloxides," PhysicalReviewB ,vol.25,no.12,pp.7157{7173,1982. [37]M.H.Mohamed,H.R.Sadeghi,andV.E.Henrich,\Electro n-energy-loss studyoftheTiO 2 (110)surface," PhysicalReviewB ,vol.37,no.14,pp. 8417{8423,1988. [38]R.H.TaitandR.V.Kasowski,\Ultravioletphotoemissi onandlowenergy-electrondiractionstudiesofTiO 2 (rutile)(001)and(110)surfaces," PhysicalReviewB ,vol.20,no.12,pp.5178{5191,1979. [39]D.C.AllanandM.P.Teter,\Localdensityapproximatio ntotalenergy calculationsforsilicaandtitaniastructureanddefects, Journalofthe AmericanCeramicSociety ,vol.73,no.11,pp.3247{3250,1990. [40]B.Silvi,N.Fourati,R.Nada,andC.R.A.Catlow,\Pseud opotential periodichartree-fockstudyofrutileTiO 2 ," JournalofPhysicsandChemistry ofSolids ,vol.52,no.8,pp.1005{1009,1991. [41]N.Daude,C.Gout,andC.Jouanin,\Electronicband-str uctureoftitaniumdioxide," PhysicalReviewB ,vol.15,no.6,pp.3229{3235,1977. [42]S.MunnixandM.Schmeits,\Electronicdensitiesofsta tesofdefect-free TiO 2 (110)andTiO 2 (001)surfaces," PhysicalReviewB ,vol.28,no.12,pp. 7342{7345,1983. [43]S.MunnixandM.Schmeits,\Electronic-structureofid ealTiO 2 (110), TiO 2 (001),andTiO 2 (100)surfaces," PhysicalReviewB ,vol.30,no.4,pp. 2202{2211,1984. [44]S.MunnixandM.Schmeits,\Electronic-structureofox ygenvacanciesin TiO 2 ," SolidStateCommunications ,vol.50,no.12,pp.1087{1089,1984. [45]S.MunnixandM.Schmeits,\Originofdefectstatesonth esurfaceofTiO 2 ," PhysicalReviewB ,vol.31,no.6,pp.3369{3371,1985. [46]K.Vos,\RerectanceandelectrorerectanceofTiO 2 single-crystals.2. assignmenttoelectronic-energylevels," JournalofPhysicsC-SolidState Physics ,vol.10,no.19,pp.3917{3939,1977.

PAGE 189

170 [47]K.VosandH.J.Krusemeyer,\Rerectanceandelectrorer ectanceofTiO 2 single-crystals.1.optical-spectra," JournalofPhysicsC-SolidStatePhysics vol.10,no.19,pp.3893{3915,1977. [48]R.L.Kurtz,R.Stockbauer,T.E.Madey,E.Roman,andJ.L .Desegovia, \SynchrotronradiationstudiesofH 2 OadsorptiononTiO 2 (110)," Surface Science ,vol.218,no.1,pp.178{200,1989. [49]J.M.PAN,B.L.Maschho,U.Diebold,andT.E.Madey,\In teractionof water,oxygen,andhydrogenwithTiO 2 (110)surfaceshavingdierentdefect densities," JournalofVacuumScience&Technologya-VacuumSurfacesan d Films ,vol.10,no.4,pp.2470{2476,1992. [50]H.Tang,H.Berger,P.E.Schmid,F.Levy,andG.Burri,\P hotoluminescenceinTiO 2 anatasesingle-crystals," SolidStateCommunications ,vol.87, no.9,pp.847{850,1993. [51]N.Kopidakis,E.A.Schi,N.-G.Park,J.vandeLagemaat ,andA.J.Frank, \Ambipolardiusionofphotocarriersinelectrolyte-lle d,nanoporousTiO 2 ," JournalofPhysicalChemostry ,vol.104,no.16,pp.3930{3936.,2000. [52]D.C.Hurum,K.A.Gray,T.Rajh,andM.C.Thurnauer,\Rec ombination pathwaysinthedegussap25formulationofTiO 2 :Surfaceversuslattice mechanisms," JournalofPhysicalChemistryB ,vol.109,no.2,pp.977{980, 2005. [53]Q.Fan,B.McQuillin,A.K.Ray,TurnerM.L.,andSeddonA .B.,\High density,non-porousanatasetitaniathinlmsfordeviceap plications," JournalofPhysicsD:AppliedPhysics ,vol.33,pp.2683{2686,(2000. [54]J.Pascual,J.Camassel,andH.Mathieu,\Piezo-spectr oscopicinvestigationofnatureofsubsidiaryvalencebandextremainTiO 2 ," SolidState Communications ,vol.28,no.4,pp.239{241,1978. [55]B.Poumellec,R.Cortes,G.Tourillon,andJ.Berthon,\ Angular-dependence oftheti-kedgeinrutileTiO 2 ," PhysicaStatusSolidiB-BasicResearch ,vol. 164,no.1,pp.319{326,1991. [56]P.Suppan, ChemistryandLight ,SocietyofChemistry,London,1994. [57]E.Becquerel,\M^cmoiresurleseetselectriquespro duitssousl'inruencedes rayonssolaires," ComptesRendusdel'Acad(C)miedesSciences ,vol.9,pp. 561{567,1839. [58]M.A.FoxandM.T.Dulay,\Accelerationofsecondarydar kreactionsof intermediatesderivedfromadsorbeddyesonirradiatedTiO 2 powders," JournalofPhotochemistryandPhotobiologya-Chemistry ,vol.98,no.1-2,pp. 91{101,1996.

PAGE 190

171 [59]A.MillsandS.Lehunte,\Anoverviewofsemiconductorp hotocatalysis," JournalofPhotochemistryandPhotobiologyA-Chemistry ,vol.108,no.1,pp. 1{35,1997. [60]S.T.Martin,H.Herrmann,W.Y.Choi,andM.R.Homann,\ Timeresolvedmicrowaveconductivity.1.TiO 2 photoreactivityandsizequantization," JournaloftheChemicalSociety-FaradayTransactions ,vol.90,no.21, pp.3315{3322,1994. [61]S.T.Martin,H.Herrmann,andM.R.Homann,\Time-reso lvedmicrowave conductivity.2.quantum-sizedTiO 2 andtheeectofadsorbatesand light-intensityoncharge-carrierdynamics," JournaloftheChemicalSocietyFaradayTransactions ,vol.90,no.21,pp.3323{3330,1994. [62]S.U.M.Khan,M.Al-Shahry,andW.B.Ingler,\Ecientph otochemical watersplittingbyachemicallymodiedn-TiO 2 ," Science ,vol.297,no.5590, pp.2243{2245,2002. [63]H.Irie,Y.Watanabe,andK.Hashimoto,\Carbon-dopeda nataseTiO 2 powdersasavisible-lightsensitivephotocatalyst," ChemistryLetters ,vol.32, no.8,pp.772{773,2003. [64]S.SakthivelandH.Kisch,\Daylightphotocatalysisby carbon-modied titaniumdioxide," AngewandteChemie-InternationalEdition ,vol.42,no.40, pp.4908{4911,2003. [65]J.Moser,M.Gratzel,andR.Gallay,\Inhibitionofelec tron-holerecombinationinsubstitutionallydopedcolloidalsemiconductorcr ystallites," Helvetica ChimicaActa ,vol.70,no.6,pp.1596{1604,1987. [66]A.Sclafani,L.Palmisano,andM.Schiavello,\Phenola ndnitrophenols photodegradationcarriedoutusingaqueousTiO 2 anatasedispersions," AbstractsofPapersoftheAmericanChemicalSociety ,vol.203,pp.132{ ENVR,1992. [67]J.Soria,J.C.Conesa,V.Augugliaro,L.Palmisano,M.S chiavello,and A.Sclafani,\Dinitrogenphotoreductiontoammoniaoverti tanium-dioxide powdersdopedwithferricions," JournalofPhysicalChemistry ,vol.95,no. 1,pp.274{282,1991. [68]W.Y.Choi,A.Termin,andM.R.Homann,\Eectsofmetal -iondopants onthephotocatalyticreactivityofquantum-sizedTiO 2 particles," AngewandteChemie-InternationalEditioninEnglish ,vol.33,no.10,pp.1091{ 1092,1994.

PAGE 191

172 [69]L.Palmisano,V.Augugliaro,A.Sclafani,andM.Schiav ello,\Activityof chromium-ion-dopedtitaniaforthedinitrogenphotoreduc tiontoammonia andforthephenolphotodegradation," JournalofPhysicalChemistry ,vol. 92,no.23,pp.6710{6713,1988. [70]J.M.Herrmann,J.Disdier,andP.Pichat,\Eectofchro miumdopingon theelectricalandcatalyticpropertiesofpowdertitaniau nderuvandvisible illumination," ChemicalPhysicsLetters ,vol.108,no.6,pp.618{622,1984. [71]W.MU,J.M.Herrmann,andP.Pichat,\Room-temperature photocatalytic oxidationofliquidcyclohexaneintocyclohexanoneoverne atandmodied TiO 2 ," CatalysisLetters ,vol.3,no.1,pp.73{84,1989. [72]B.Sun,E.P.Reddy,andP.G.Smirniotis,\Eectofthecr 6+concentration incr-incorporatedTiO 2 -loadedmcm-41catalystsforvisiblelightphotocatalysis," AppliedCatalysisB-Environmental ,vol.57,no.2,pp.139{149, 2005. [73]K.E.KarakitsouandX.E.Verykios,\Eectsofalterval entcationdopingof TiO 2 onitsperformanceasaphotocatalystforwatercleavage," Journalof PhysicalChemistry ,vol.97,no.6,pp.1184{1189,1993. [74]E.C.ButlerandA.P.Davis,\Photocatalyticoxidation inaqueoustitaniumdioxidesuspensions-theinruenceofdissolvedtransition -metals," Journal ofPhotochemistryandPhotobiologya-Chemistry ,vol.70,no.3,pp.273{283, 1993. [75]M.Fujihira,Y.Satoh,andT.Osa,\Heterogeneousphoto catalyticreactions onsemiconductor-materials3:eectofpHandCu +2 ionsonthephoto-fenton typereaction," BulletinoftheChemicalSocietyofJapan ,vol.55,no.3,pp. 666{671,1982. [76]S.SatoandJ.M.White,\Photo-decompositionofwatero verp-TiO 2 catalysts," ChemicalPhysicsLetters ,vol.72,no.1,pp.83{86,1980. [77]S.SatoandJ.M.White,\Photoassistedreactionofco2w ithh2oonptTiO 2 ," AbstractsofPapersoftheAmericanChemicalSociety ,vol.179,no. MAR,pp.75{COLL,1980. [78]D.Bockelmann,R.Goslich,andD.Bahnemann, InsolarThermalEnergy Utilization ,vol.6,SpringlerVerlagGmbH,Heidelberg,1992. [79]P.Pichat,M.N.Mozzanega,J.Disdier,andJ.M.Herrman n,\Ptcontent andtemperatureeectsonthephotocatalytich-2productio nfromaliphaticalcoholsoverptTiO 2 ," NouveauJournalDeChimie-NewJournalof Chemistry ,vol.6,no.11,pp.559{564,1982.

PAGE 192

173 [80]A.Sclafani,M.N.Mozzanega,andP.Pichat,\Eectofsi lverdepositsonthe photocatalyticactivityoftitanium-dioxidesamplesfort hedehydrogenation oroxidationof2-propanol," JournalofPhotochemistryandPhotobiology a-Chemistry ,vol.59,no.2,pp.181{189,1991. [81]D.Bahnemann,D.Bockelmann,andR.Goslich,\Mechanis ticstudies ofwaterdetoxicationinilluminatedTiO 2 suspensions," SolarEnergy Materials ,vol.24,no.1-4,pp.564{583,1991. [82]K.R.Gopidas,M.Bohorquez,andP.V.Kamat,\Photoelec trochemistry insemiconductorparticulatesystems.16.photophysicala ndphotochemical aspectsofcoupledsemiconductors-charge-transferproce ssesincolloidal cds-TiO 2 andcds-agisystems," JournalofPhysicalChemistry ,vol.94,no. 16,pp.6435{6440,1990. [83]D.Duonghong,E.Borgarello,andM.Gratzel,\Dynamics oflight-induced watercleavageincolloidalsystems," JournaloftheAmericanChemical Society ,vol.103,no.16,pp.4685{4690,1981. [84]B.SunandP.G.Smirniotis,\Interactionofanataseand rutileTiO 2 particles inaqueousphotooxidation," CatalysisToday ,vol.88,no.1-2,pp.49{59, 2003. [85]R.J.Davis,J.L.Gainer,G.Oneal,andI.W.Wu,\Photoca talyticdecolorizationofwaste-waterdyes," WaterEnvironmentResearch ,vol.66,no.1, pp.50{53,1994. [86]L.A.DibbleandG.B.Raupp,\Fluidized-bedphotocatal yticoxidationof trichloroethyleneincontaminatedairstreams," EnvironmentalScience& Technology ,vol.26,no.3,pp.492{495,1992. [87]B.C.FaustandM.R.Homann,\Photoinducedreductived issolutionof alpha-fe2o3bybisulte," EnvironmentalScience&Technology ,vol.20,no.9, pp.943{948,1986. [88]S.H.LeeandW.M.Sigmund,\Inactivationofbacteriale ndosporesby photocatalyticnanocomposites.," AbstractsofPapersoftheAmerican ChemicalSociety ,vol.227,pp.U1275{U1276,2004. [89]E.R.Carraway,A.J.Homan,andM.R.Homann,\Photoca talytic oxidationoforganic-acidsonquantum-sizedsemiconducto rcolloids," EnvironmentalScience&Technology ,vol.28,no.5,pp.786{793,1994. [90]C.Kormann,D.W.Bahnemann,andM.R.Homann,\Photoly sisof chloroformandotherorganic-moleculesinaqueousTiO 2 suspensions," EnvironmentalScience&Technology ,vol.25,no.3,pp.494{500,1991.

PAGE 193

174 [91]G.MillsandM.R.Homann,\Photocatalyticdegradatio nofpentachlorophenolonTiO 2 particles-identicationofintermediatesand mechanismofreaction," EnvironmentalScience&Technology ,vol.27,no.8, pp.1681{1689,1993. [92]G.Chester,M.Anderson,H.Read,andS.Esplugas,\Ajac ketedannular membranephotocatalyticreactorforwaste-watertreatmen t-degradation offormic-acidandatrazine," JournalofPhotochemistryandPhotobiology a-Chemistry ,vol.71,no.3,pp.291{297,1993. [93]K.Hofstadler,R.Bauer,S.Novalic,andG.Heisler,\Ne wreactordesignfor photocatalyticwaste-watertreatmentwithTiO 2 immobilizedonfused-silica glass-bers-photomineralizationof4-chlorophenol," EnvironmentalScience &Technology ,vol.28,no.4,pp.670{674,1994. [94]S.W.Lee,J.Drwiega,D.Mazyck,andW.M.Sigmund,\Synt hesisand characterizationofhardmagneticcompositephotocatalys t(bariumferrite silicatitania)anditsphotoactivity.," AbstractsofPapersoftheAmerican ChemicalSociety ,vol.227,pp.U1246{U1246,2004. [95]R.W.Matthews,M.Abdullah,andG.K.C.Low,\Photocata lyticoxidation fortotalorganic-carbonanalysis," AnalyticaChimicaActa ,vol.233,no.2, pp.171{179,1990. [96]R.W.Matthews,\Solar-electricwater-puricationus ingphotocatalytic oxidationwithTiO 2 asastationaryphase," SolarEnergy ,vol.38,no.6,pp. 405{413,1987. [97]R.S.Henderson,R.J.Dawson,M.D.Hasino,G.Azuelos, S.Ahmad, A.Sernaangel,M.Blecher,T.P.Gorringe,B.C.Robertson,a ndD.H. Wright,\Acylindricaldriftchamberforradiativemuon-ca ptureexperiments attriumf," IeeeTransactionsonNuclearScience ,vol.37,no.3,pp.1116{ 1119,1990. [98]B.OreganandM.Gratzel,\Alow-cost,high-eciencyso lar-cellbasedon dye-sensitizedcolloidalTiO 2 lms," Nature ,vol.353,no.6346,pp.737{740, 1991. [99]J.R.Durrant,S.A.Haque,andE.Palomares,\Towardsop timisation ofelectrontransferprocessesindyesensitisedsolarcell s," Coordination ChemistryReviews ,vol.248,no.13-14,pp.1247{1257,2004. [100]A.J.Frank,N.Kopidakis,andJvandeLagemaat,\Elect ronsinnanostructuredTiO 2 solarcells:transport,recombinationandphotovoltaicpr operties," CoordinationChemistryReviews ,vol.248,pp.11651179,2004.

PAGE 194

175 [101]H.Tang,K.Prasad,R.Sanjines,andF.Levy,\TiO 2 anatasethin-lmsas gassensors," SensorsandActuatorsB-Chemical ,vol.26,no.1-3,pp.71{75, 1995. [102]K.Schwarzburg,R.Ernstorfer,S.Felber,andF.Willi g,\Primaryandnal chargeseparationinthenano-structureddye-sensitizede lectrochemicalsolar cell," CoordinationChemistryReviews ,vol.248,no.13-14,pp.1259{1270, 2004. [103]S.Iijima,\Helicalmicrotubulesofgraphiticcarbon ," Nature ,vol.354,no. 6348,pp.56{58,1991. [104]J.W.MintwireandC.T.White,\Electronicandstructu ralpropertiesof carbonnanotubes," Carbon ,vol.33,no.7,pp.893{902,1995. [105]R.S.RuoandD.C.Lorents,\Mechanicalandthermal-p ropertiesofcarbon nanotubes," Carbon ,vol.33,no.7,pp.925{930,1995. [106]D.S.Bethune,C.H.Kiang,M.S.Devries,G.Gorman,R.S avoy,J.Vazquez, andR.Beyers,\Cobalt-catalyzedgrowthofcarbonnanotube swithsingleatomic-layerwalls," Nature ,vol.363,no.6430,pp.605{607,1993. [107]M.A.Prelas,G.Popovici,andL.K.Bigelow, HandbookofIntustrialDiamondsandDiamondFilms, ,MarcelDekker,NewYork,1997. [108]B.TKelly, PhysicsofGraphite ,AppliedScience,London,1981. [109]S.IharaandS.Itoh,\Helicallycoiledandtoroidalca geformsofgraphite carbon," Carbon ,vol.33,no.7,pp.931{939,1995. [110]H.W.Kroto,J.R.Heath,S.C.Obrien,R.F.Curl,andR.E .Smalley,\C 60 -Buckminsterfullerene," Nature ,vol.318,no.6042,pp.162{163,1985. [111]M.S.Dresselhaus,G.Dresselhaus,andP.C.Eklund, ScienceofFullerenes andCarbonNanotubes ,AcademicPress,NewYork,1996. [112]M.S.Dresselhaus,G.Dresselhaus,andR.Saito,\Carb on-bersbasedonc 60 andtheirsymmetry," PhysicalReviewB ,vol.45,no.11,pp.6234{6242,1992. [113]R.Saito,M.Fujita,G.Dresselhaus,andM.S.Dresselh aus,\Electronicstructureofgraphenetubulesbasedonc 60 ," PhysicalReviewB ,vol.46,no. 3,pp.1804{1811,1992. [114]P.C.Eklund,J.M.Holden,andR.A.Jishi,\Vibrationa l-modesofcarbon nanotubes-spectroscopyandtheory," Carbon ,vol.33,no.7,pp.959{972, 1995.

PAGE 195

176 [115]R.A.Jishi,M.S.Dresselhaus,andG.Dresselhaus,\Sy mmetryproperties ofchiralcarbonnanotubes," PhysicalReviewB ,vol.47,no.24,pp.16671{ 16674,1993. [116]N.Hamada,S.Sawada,andA.Oshiyama,\Newone-dimens ionalconductors -graphiticmicrotubules," PhysicalReviewLetters ,vol.68,no.10,pp. 1579{1581,1992. [117]J.W.Mintmire,B.I.Dunlap,andC.T.White,\Arefulle renetubules metallic," PhysicalReviewLetters ,vol.68,no.5,pp.631{634,1992. [118]S.J.Tans,M.H.Devoret,H.J.Dai,A.Thess,R.E.Small ey,L.J.Geerligs, andC.Dekker,\Individualsingle-wallcarbonnanotubesas quantumwires," Nature ,vol.386,no.6624,pp.474{477,1997. [119]T.W.Odom,J.L.Huang,P.Kim,andC.M.Lieber,\Atomic structureand electronicpropertiesofsingle-walledcarbonnanotubes, Nature ,vol.391,no. 6662,pp.62{64,1998. [120]J.W.G.Wildoer,A.vanOudenaarden,C.J.P.M.Harmans ,andH.van Kempen,\Scanningtunnelingmicroscopetipasapositionab lecontact: Probingajosephson-junctionarrayatsubkelvintemperatu res," Journal ofVacuumScience&TechnologyB ,vol.16,no.5,pp.2837{2840,1998, 129WWTimesCited:1CitedReferencesCount:8. [121]H.YorikawaandS.Muramatsu,\Chirality-dependence ofenergygapsof semiconductingnanotubules," SolidStateCommunications ,vol.94,no.6,pp. 435{437,1995. [122]X.Blase,L.X.Benedict,E.L.Shirley,andS.G.Louie, \Hybridization eectsandmetallicityinsmallradiuscarbonnanotubes," PhysicalReview Letters ,vol.72,no.12,pp.1878{1881,1994. [123]M.Ouyang,J.L.Huang,C.L.Cheung,andC.M.Lieber,\E nergygapsin "metallic"single-walledcarbonnanotubes," Science ,vol.292,no.5517,pp. 702{705,2001. [124]C.T.White,D.H.Robertson,andJ.W.Mintmire,\Helic alandrotational symmetriesofnanoscalegraphitictubules," PhysicalReviewB ,vol.47,no.9, pp.5485{5488,1993. [125]H.YorikawaandS.Muramatsu,\Theoreticalstudyofel ectronicdensityof statesforcarbonnanotubules," SyntheticMetals ,vol.82,no.1,pp.17{21, 1996. [126]M.Ouyang,J.L.Huang,C.L.Cheung,andC.M.Lieber,\A tomically resolvedsingle-walledcarbonnanotubeintramolecularju nctions," Science vol.291,no.5501,pp.97{100,2001.

PAGE 196

177 [127]M.Bockrath,D.H.Cobden,P.L.McEuen,N.G.Chopra,A. Zettl, A.Thess,andR.E.Smalley,\Single-electrontransportinr opesofcarbonnanotubes," Science ,vol.275,no.5308,pp.1922{1925,1997. [128]S.J.Tans,M.H.Devoret,H.J.Dai,A.Thess,R.E.Small ey,L.J.Geerligs, andC.Dekker,\Individualsingle-wallcarbonnanotubesas quantumwires," Nature ,vol.386,no.6624,pp.474{477,1997. [129]S.Frank,P.Poncharal,Z.L.Wang,andW.A.deHeer,\Ca rbonnanotube quantumresistors," Science ,vol.280,no.5370,pp.1744{1746,1998. [130]M.Freitag,Y.Martin,J.A.Misewich,R.Martel,andP. H.Avouris, \Photoconductivityofsinglecarbonnanotubes," NanoLetters ,vol.3,no.8, pp.1067{1071,2003. [131]A.HagenandT.Hertel,\Quantitativeanalysisofopti calspectrafrom individualsingle-wallcarbonnanotubes," NanoLetters ,vol.3,no.3,pp. 383{388,2003. [132]J.A.Misewich,R.Martel,P.Avouris,J.C.Tsang,S.He inze,andJ.Terso, \Electricallyinducedopticalemissionfromacarbonnanot ubefet," Science vol.300,no.5620,pp.783{786,2003. [133]J.W.G.Wildoer,A.vanOudenaarden,C.J.P.M.Harmans ,andH.van Kempen,\Scanningtunnelingmicroscopetipasapositionab lecontact: Probingajosephson-junctionarrayatsubkelvintemperatu res," Journalof VacuumScience&TechnologyB ,vol.16,no.5,pp.2837{2840,1998. [134]J.W.G.Wildoer,L.C.Venema,A.G.Rinzler,R.E.Small ey,and C.Dekker,\Electronicstructureofatomicallyresolvedca rbonnanotubes," Nature ,vol.391,no.6662,pp.59{62,1998. [135]M.Meyyappan,Ed., CarbonNanotubesScienceandApplications ,CRC Press,MoettField,CA,2005. [136]L.Delzeit,B.Chen,A.Cassell,R.Stevens,C.Nguyen, andM.Meyyappan, \Multilayeredmetalcatalystsforcontrollingthedensity ofsingle-walled carbonnanotubegrowth," ChemicalPhysicsLetters ,vol.348,no.5-6,pp. 368{374,2001. [137]H.Kind,J.M.Bonard,L.Forro,K.Kern,K.Hernadi,L.O .Nilsson, andL.Schlapbach,\Printinggel-likecatalystsforthedir ectedgrowthof multiwallcarbonnanotubes," Langmuir ,vol.16,no.17,pp.6877{6883,2000. [138]M.A.LiebermanandL.A.J., Principlesofplasmadischargesandmaterials processing ,JonWileyandSons,Inc.,NewYork,1994.

PAGE 197

178 [139]K.Tohji,H.Takahashi,Y.Shinoda,N.Shimizu,B.Jeya devan,I.Matsuoka, Y.Saito,A.Kasuya,S.Ito,andY.Nishina,\Puricationpro cedurefor single-wallednanotubes," JournalofPhysicalChemistryB ,vol.101,pp. 1974,1997. [140]S.M.HuangandL.M.Dai,\Plasmaetchingforpuricati onandcontrolled openingofalignedcarbonnanotubes," JournalofPhysicalChemistryB ,vol. 106,no.14,pp.3543{3545,2002. [141]I.W.Chiang,B.E.Brinson,R.E.Smalley,J.L.Margrav e,andR.H.Hauge, \Puricationandcharacterizationofsingle-wallcarbonn anotubes," Journal ofPhysicalChemistryB ,vol.105,no.6,pp.1157{1161,2001. [142]S.Bandow,A.M.Rao,K.A.Williams,A.Thess,R.E.Smal ley,andP.C. Eklund,\Puricationofsingle-wallcarbonnanotubesbymi croltration," JournalofPhysicalChemistryB ,vol.101,no.44,pp.8839{8842,1997. [143]S.Niyogi,H.Hu,M.A.Hamon,P.Bhowmik,B.Zhao,S.M.R ozenzhak, J.Chen,M.E.Itkis,M.S.Meier,andR.C.Haddon,\Chromatog raphic puricationofsolublesingle-walledcarbonnanotubes(sswnts)," Journalof theAmericanChemicalSociety ,vol.123,no.4,pp.733{734,2001. [144]A.G.Rinzler,J.Liu,H.Dai,P.Nikolaev,C.B.Human, F.J.RodriguezMacias,P.J.Boul,A.H.Lu,D.Heymann,D.T.Colbert,R.S.Le e,J.E. Fischer,A.M.Rao,P.C.Eklund,andR.E.Smalley,\Large-sc alepuricationofsingle-wallcarbonnanotubes:process,product,an dcharacterization," AppliedPhysicsa-MaterialsScience&Processing ,vol.67,no.1,pp.29{37, 1998. [145]K.B.Shelimov,R.O.Esenaliev,A.G.Rinzler,C.B.Hu man,andR.E. Smalley,\Puricationofsingle-wallcarbonnanotubesbyu ltrasonically assistedltration," ChemicalPhysicsLetters ,vol.282,no.5-6,pp.429{434, 1998. [146]M.Burghard,\Electronicandvibrationalproperties ofchemicallymodied single-wallcarbonnanotubes," SurfaceScienceReports ,vol.58,no.1-4,pp. 1{109,2005. [147]J.Cui,W.Wang,Y.You,C.Liu,andP.Wang,\Functional izationof multiwalledcarbonnanotubesbyreversibleadditionfragm entationchaintransferpolymerization," Polymer ,vol.45,no.26,pp.8717{8721,2004. [148]A.Misra,P.K.Tyagi,M.K.Singh,andD.S.Misra,\Ftir studiesof nitrogendopedcarbonnanotubes," DiamondandRelatedMaterials ,vol.In Press,CorrectedProof,1996.

PAGE 198

179 [149]C.Zhao,L.Ji,H.Liu,G.Hu,S.Zhang,M.Yang,andZ.Yan g,\Functionalizedcarbonnanotubescontainingisocyanategroups," JournalofSolidState Chemistry ,vol.177,no.12,pp.4394{4398,2004. [150]L.L.HenchandJ.K.West,\Thesol-gelprocess," ChemicalReviews ,vol. 90,no.1,pp.33{72,1990. [151]K.Baiju,C.P.Sibu,K.Rajesh,P.K.Pillai,P.Mukunda n,K.G.K.Warrier, andW.Wunderlich,\Anaqueoussol-gelroutetosynthesizen anosized lanthanadopedtitaniahavinganincreasedanatasephasest abilityfor photocatalyticapplication," MaterialsChemistryandPhysics ,vol.90,no.1, pp.123{127,2005. [152]T.Sreethawong,Y.Suzuki,andS.Yoshikawa,\Synthes is,characterization, andphotocatalyticactivityforhydrogenevolutionofnano crystallinemesoporoustitaniapreparedbysurfactant-assistedtemplatin gsolagelprocess," JournalofSolidStateChemistry ,vol.178,pp.329A338,2005. [153]R.L.Ballard,J.P.Williams,J.M.Njus,B.R.Kiland,a ndM.D.Soucek, \Inorganicaorganichybridcoatingswithmixedmetaloxi des," European PolymerJournal ,vol.37,no.2,pp.381{398,2001. [154]R.A.SailerandM.D.Soucek,\Oxidizingalkydceramer s," Progressin OrganicCoatings ,vol.33,no.1,pp.36{43,1998. [155]J.Aarik,A.Aidla,T.Uustare,M.Ritala,andM.Leskel a,\Titanium isopropoxideasaprecursorforatomiclayerdeposition:ch aracterizationof titaniumdioxidegrowthprocess," AppliedSurfaceScience ,vol.161,no.3-4, pp.385{395,2000. [156]W.Cho,J.-w.Park,andC.-S.Ha,\Directsynthesisoft itanium-modied hybridperiodicmesoporousorganosilicasfrom1,2-bis(tr iethoxysilyl)ethane andtitaniumisopropoxide," MaterialsLetters ,vol.58,no.27-28,pp.3551{ 3554,2004. [157]U.Qureshi,T.D.Manning,C.Blackman,andI.P.Parkin ,\Composite thermochromicthinlms:(TiO 2 )-(vo2)preparedfromtitaniumisopropoxide, vocl3andwater," Polyhedron ,vol.25,no.2,pp.334{338,2006. [158]A.Rodriguez-DelgadoandE.Y.X.Chen,\Ligandexchan geandabstraction reactionsinvolvingtitaniumisopropoxidewithtris(pent aruorophenyl)borane and-alane:ramicationsforring-openingpolymerization ofpropyleneoxide," InorganicaChimicaActa ,vol.357,no.13,pp.3911{3919,2004. [159]I.Arcon,B.Malic,M.Kosec,andA.Kodre,\Exafsstudy ofpztsols," MaterialsResearchBulletin ,vol.38,no.15,pp.1901{1906,2003.

PAGE 199

180 [160]M.P.Feth,A.Weber,R.Merkle,U.Reinohl,andH.Berta gnolli,\Exafs andx-raydiractionstudiesonsol-gelpreparedzirconium titaniumoxides," JournalofNon-CrystallineSolids ,vol.298,no.1,pp.43{52,2002. [161]A.A.Aly,\Tripleself-condensationoffusedcycloal kanonylparacyclophanes promotedbytitaniumtetrachlorideandtriethylamine," TetrahedronLetters vol.46,no.3,pp.443{446,2005. [162]T.KapiasandR.F.Griths,\Accidentalreleasesofti taniumtetrachloride (ticl4)inthecontextofmajorhazards{spillbehaviourusi ngreactpool," JournalofHazardousMaterials ,vol.119,no.1-3,pp.41{52,2005. [163]K.Makino,Y.Henmi,M.Terasawa,O.Hara,andY.Hamada ,\Remarkable eectsoftitaniumtetrachlorideindiastereoselectiveaz adiels-aldercycloaddition:synthesisof(s)-piperazicacid," TetrahedronLetters ,vol.46,no.4,pp. 555{558,2005. [164]J.Yuan,M.Chen,J.Shi,andW.Shangguan,\Preparatio nsandphotocatalytichydrogenevolutionofn-dopedTiO 2 fromureaandtitanium tetrachloride," InternationalJournalofHydrogenEnergy ,vol.InPress, CorrectedProof,2006. [165]G.Chen,G.Luo,X.Yang,Y.Sun,andJ.Wang,\Anatase-T iO 2 nanoparticlepreparationwithamicro-mixingtechniqueandits photocatalytic performance," MaterialsScienceandEngineeringA ,vol.380,no.1-2,pp. 320{325,2004. [166]Y.Wei,R.Wu,andY.Zhang,\Preparationofmonodisper sedsphericalTiO 2 powderbyforcedhydrolysisofti(so4)2solution," MaterialsLetters ,vol.41, no.3,pp.101{103,1999. [167]J.Zhao,Z.Wang,L.Wang,H.Yang,andM.Zhao,\Thesynt hesisand characterizationofTiO 2 /wollastonitecomposite," MaterialsLetters ,vol.37, no.3,pp.149{155,1998. [168]S.W.LeeandW.M.Sigmund,\FormationofanataseTiO 2 nanoparticleson carbonnanotubes," ChemicalCommunications ,,no.6,pp.780{781,2003. [169]A.Jitianu,T.Cacciaguerra,R.Benoit,S.Delpeux,F. Beguin,andS.Bonnamy,\Synthesisandcharacterizationofcarbonnanotubes -TiO 2 nanocomposites," Carbon ,vol.42,no.5-6,pp.1147{1151,2004. [170]S.H.LeeandW.M.Sigmund,\Inactivationofbacterial endosporesby photocatalyticnanocomposites.," AbstractsofPapersoftheAmerican ChemicalSociety ,vol.227,pp.U1275{U1276,2004. [171]A.L.Paterson,\ThescherrerformulaofX-Rayparticl esizedetermination," PhysicalReview ,vol.56,pp.978{982,1939.

PAGE 200

181 [172]MurdockC.C.,\TheformoftheX-Raydiractionbandsf orregularcrystals ofcolloidalsize," PhysicalReview ,vol.35,no.1,pp.18{23,1930. [173]F.W.Jones,\Themeasurementofparticlesizebythexraymethod," ProceedingsoftheRoyalSocietyofLondon.SeriesA,Mathem aticaland PhysicalSciences ,vol.166,no.924,pp.16{43,1938. [174]M.P.SeahandW.A.Dench,\Quantitativeelectronspec troscopyof surfaces:Astandarddatabaseforelectroninelasticmeanf reepathsin solids," SurfaceandInterfaceAnalysis ,vol.1,no.1,pp.2{11,1979. [175]K.DoushitaandT.Kawahara,\Evaluationofphotocata lyticactivitybydye decomposition," JournalofSol-GelScienceandTechnology ,vol.22,no.1-2, pp.91{98,2001. [176]H.Lachheb,E.Puzenat,A.Houas,M.Ksibi,E.Elaloui, C.Guillard,and J.M.Herrmann,\Photocatalyticdegradationofvarioustyp esofdyes (alizarins,croceinorangeg,methylred,congored,methyl eneblue)inwater byuv-irradiatedtitania," AppliedCatalysisB-Environmental ,vol.39,no.1, pp.75{90,2002. [177]I.Mora-SeroandJ.Bisquert,\Fermilevelofsurfaces tatesinTiO 2 nanoparticles," NanoLetters ,vol.3,no.7,pp.945{949,2003. [178]G.Sivalingam,K.Nagaveni,M.S.Hegde,andG.Madras, \Photocatalytic degradationofvariousdyesbycombustionsynthesizednano anataseTiO 2 ," AppliedCatalysisB-Environmental ,vol.45,no.1,pp.23{38,2003. [179]T.Stergiopoulos,M.C.Bernard,A.H.L.Go,andP.Fal aras,\Resonance micro-Ramanspectrophotoelectrochemistryonnanocrysta llineTiO 2 thinlm electrodessensitizedbyRu(II)complexes," CoordinationChemistryReviews vol.248,no.13-14,pp.1407{1420,2004. [180]C.Hu,J.C.Yu,Z.P.Hao,andP.K.Wong,\Eectsofacidi tyandinorganic ionsonthephotocatalyticdegradationofdierentazodyes ," Applied CatalysisB-Environmental ,vol.46,no.1,pp.35{47,2003. [181]I.K.KonstantinouandT.A.Albanis,\TiO 2 -assistedphotocatalyticdegradationofazodyesinaqueoussolution:kineticandmechanis ticinvestigations -AReview," AppliedCatalysisB-Environmental ,vol.49,no.1,pp.1{14, 2004. [182]C.Galindo,P.Jacques,andA.Kalt,\Photodegradatio noftheaminoazobenzeneacidOrange52bythreeadvancedoxidationproces ses:UV/H 2 O 2 UV/TiO 2 andVIS/TiO 2 -Comparativemechanisticandkineticinvestigations," JournalofPhotochemistryandPhotobiologya-Chemistry ,vol.130,no. 1,pp.35{47,2000.

PAGE 201

182 [183]C.Guillard,H.Lachheb,A.Houas,M.Ksibi,E.Elaloui ,andJ.M.Herrmann,\Inruenceofchemicalstructureofdyes,ofpHandofi norganicsalts ontheirphotocatalyticdegradationbyTiO 2 comparisonoftheeciencyof powderandsupportedTiO 2 ," JournalofPhotochemistryandPhotobiology A-Chemistry ,vol.158,no.1,pp.27{36,2003. [184]C.Hu,J.C.Yu,Z.P.Hao,andP.K.Wong,\Photocatalyti cdegradationof triazine-containingazodyesinaqueousTiO 2 suspensions," AppliedCatalysis B-Environmental ,vol.42,no.1,pp.47{55,2003. [185]A.Houas,H.Lachheb,M.Ksibi,E.Elaloui,C.Guillard ,andJ.M.Herrmann,\Photocatalyticdegradationpathwayofmethyleneb lueinwater," AppliedCatalysisB-Environmental ,vol.31,no.2,pp.145{157,2001. [186]C.M.So,M.Y.Cheng,J.C.Yu,andP.K.Wong,\Degradati onofazodye ProcionRedMX-5Bbyphotocatalyticoxidation," Chemosphere ,vol.46,no. 6,pp.905{912,2002. [187]D.F.Ollis,E.Pelizzetti,andN.Serpone,\Photocata lyzeddestructionof watercontaminants," EnvironmentalScience&Technology ,vol.25,no.9,pp. 1522{1529,1991. [188]S.Sakthivel,B.Neppolian,M.V.Shankar,B.Arabindo o,M.Palanichamy, andV.Murugesan,\Solarphotocatalyticdegradationofazo dye:comparison ofphotocatalyticeciencyofZnOandTiO 2 ," SolarEnergyMaterialsand SolarCells ,vol.77,no.1,pp.65{82,2003. [189]M.SaquibandM.Muneer,\TiO 2 -mediatedphotocatalyticdegradationof atriphenylmethanedye(gentianviolet),inaqueoussuspen sions," Dyesand Pigments ,vol.56,no.1,pp.37{49,2003. [190]Y.WangandN.Herron,\Nanometer-sizedsemiconducto rclusters-materials synthesis,quantumsizeeects,andphotophysicalpropert ies," Journalof PhysicalChemistry ,vol.95,no.2,pp.525{532,1991. [191]K.M.Reddy,S.V.Manorama,andA.R.Reddy,\Bandgapst udieson anatasetitaniumdioxidenanoparticles," MaterialsChemistryandPhysics vol.78,no.1,pp.239{245,2003. [192]T.BelinandR.Epron,\Characterizationmethodsofca rbonnanotubes: AReview," MaterialsScienceandEngineeringB-SolidStateMaterials for AdvancedTechnology ,vol.119,no.2,pp.105{118,2005. [193]M.S.Dresselhaus,G.Dresselhaus,A.Jorio,A.G.Souz a,andR.Saito, \Ramanspectroscopyonisolatedsinglewallcarbonnanotub es," Carbon ,vol. 40,no.12,pp.2043{2061,2002.

PAGE 202

183 [194]M.S.Dresselhaus,G.Dresselhaus,R.Saito,andA.Jor io,\Ramanspectroscopyofcarbonnanotubes," PhysicsReports-ReviewSectionofPhysics Letters ,vol.409,no.2,pp.47{99,2005. [195]R.Saito,G.Dresselhaus,andM.S.Dresselhaus, PhysicalPropertiesof CarbonNanotubes ,ImperialCollegePress,London,1998. [196]A.Jorio,M.A.Pimenta,A.G.Souza,R.Saito,G.Dresse lhaus,andM.S. Dresselhaus,\Characterizingcarbonnanotubesampleswit hresonance Ramanscattering," NewJournalofPhysics ,vol.5,pp.139.1{139.17,2003. [197]A.Jorio,R.Saito,J.H.Hafner,C.M.Lieber,M.Hunter ,T.McClure, G.Dresselhaus,andM.S.Dresselhaus,\Structural(n,m)de termination ofisolatedsingle-wallcarbonnanotubesbyresonantRaman scattering," PhysicalReviewLetters ,vol.86,no.6,pp.1118{1121,2001. [198]M.A.Pimenta,A.Marucci,S.A.Empedocles,M.G.Bawen di,E.B. Hanlon,A.M.Rao,P.C.Eklund,R.E.Smalley,G.Dresselhaus ,andM.S. Dresselhaus,\Ramanmodesofmetalliccarbonnanotubes," PhysicalReview B ,vol.58,no.24,pp.16016{16019,1998. [199]L.G.Cancado,M.A.Pimenta,R.Saito,A.Jorio,L.O.L adeira, A.Grueneis,A.G.Souza,G.Dresselhaus,andM.S.Dresselha us,\Stokes andanti-StokesdoubleresonanceRamanscatteringintwo-d imensional graphite," PhysicalReviewB ,vol.66,no.3,pp.035415.1{035415.5,2002. [200]A.G.Souza,S.G.Chou,G.G.Samsonidze,G.Dresselhau s,M.S.Dresselhaus,L.An,J.Liu,A.K.Swan,M.S.Unlu,B.B.Goldberg,A .Jorio, A.Gruneis,andR.Saito,\Stokesandanti-StokesRamanspec traofsmalldiameterisolatedcarbonnanotubes," PhysicalReviewB ,vol.69,no.11,pp. 115428.1{115428.8,2004. [201]A.G.Souza,A.Jorio,J.H.Hafner,C.M.Lieber,R.Sait o,M.A.Pimenta, G.Dresselhaus,andM.S.Dresselhaus,\Electronictransit ionenergyE-iifor anisolated(n,m)single-wallcarbonnanotubeobtainedbya nti-Stokes/Stokes resonantRamanintensityratio," PhysicalReviewB ,vol.6324,no.24,pp.{241404,2001. [202]M.S.Dresselhaus,G.Dresselhaus,A.Jorio,A.G.Souz a,M.A.Pimenta, andR.Saito,\SinglenanotubeRamanspectroscopy," AccountsofChemical Research ,vol.35,no.12,pp.1070{1078,2002. [203]M.S.Dresselhaus,G.Dresselhaus,A.Jorio,A.G.Souz a,G.G.Samsonidze, andR.Saito,\Scienceandapplicationsofsingle-nanotube Ramanspectroscopy," JournalofNanoscienceandNanotechnology ,vol.3,no.1-2,pp. 19{37,2003.

PAGE 203

184 [204]D.S.KnightandW.B.White,\Characterizationofdiam ondlmsby Raman-spectroscopy," JournalofMaterialsResearch ,vol.4,no.2,pp. 385{393,1989. [205]A.Jorio,A.G.Souza,G.Dresselhaus,M.S.Dresselhau s,A.K.Swan,M.S. Unlu,B.B.Goldberg,M.A.Pimenta,J.H.Hafner,C.M.Lieber ,and R.Saito,\G-BandresonantRamanstudyof62isolatedsingle -wallcarbon nanotubes," PhysicalReviewB ,vol.65,no.15,pp.155412.1{155412.9,2002. [206]A.Jorio,A.G.Souza,G.Dresselhaus,M.S.Dresselhau s,R.Saito,J.H. Hafner,C.M.Lieber,F.M.Matinaga,M.S.S.Dantas,andM.A. Pimenta, \Jointdensityofelectronicstatesforoneisolatedsingle -wallcarbonnanotube studiedbyresonantRamanscattering," PhysicalReviewB ,vol.63,no.24, pp.245416.1{245416.6,2001. [207]M.Milnera,J.Kurti,M.Hulman,andH.Kuzmany,\Perio dicresonance excitationandintertubeinteractionfromquasicontinuou sdistributed helicitiesinsingle-wallcarbonnanotubes," PhysicalReviewLetters ,vol.84, no.6,pp.1324{1327,2000. [208]M.S.DresselhausandG.Dresselhaus,\Intercalation compoundsof graphite," AdvancesinPhysics ,vol.30,no.2,pp.139{326,1981. [209]E.Enoki,M.Endo,andM.Suzuki, GraphiteIntercalationCompundsand Applications ,OxfordUniversityPress,NewYork,2003. [210]A.Jorio,M.A.Pimenta,A.G.Souza,G.G.Samsonidze,A .K.Swan,M.S. Unlu,B.B.Goldberg,R.Saito,G.Dresselhaus,andM.S.Dres selhaus, \ResonanceRamanspectraofcarbonnanotubesbycross-pola rizedlight," PhysicalReviewLetters ,vol.90,no.10,pp.107403.1{107403.4,2003. [211]P.Corio,A.Jorio,N.Demir,andM.S.Dresselhaus,\Sp ectro-electrochemical studiesofsinglewallcarbonnanotubeslms," ChemicalPhysicsLetters ,vol. 392,no.4-6,pp.396{402,2004. [212]P.Corio,P.S.Santos,V.W.Brar,G.G.Samsonidze,S.G .Chou,andM.S. Dresselhaus,\PotentialdependentsurfaceRamanspectros copyofsinglewall carbonnanotubelmsonplatinumelectrodes," ChemicalPhysicsLetters vol.370,no.5-6,pp.675{682,2003. [213]A.G.Souza,A.Jorio,G.G.Samsonidze,G.Dresselhaus ,R.Saito,and M.S.Dresselhaus,\Ramanspectroscopyforprobingchemica lly/physically inducedphenomenaincarbonnanotubes," Nanotechnology ,vol.14,no.10, pp.1130{1139,2003. [214]O.DubayandG.Kresse,\Accuratedensityfunctionalc alculationsforthe phonondispersionrelationsofgraphitelayerandcarbonna notubes," Physical ReviewB ,vol.69,no.8,pp.035401,2004.

PAGE 204

185 [215]S.D.M.Brown,A.Jorio,P.Corio,M.S.Dresselhaus,G. Dresselhaus, R.Saito,andK.Kneipp,\OriginoftheBreit-Wigner-Fanoli neshapeofthe tangentialG-Bandfeatureofmetalliccarbonnanotubes," PhysicalReviewB vol.63,no.15,pp.155414.1{155414.8,2001. [216]A.M.Rao,A.Jorio,M.A.Pimenta,M.S.S.Dantas,R.Sai to,G.Dresselhaus,andM.S.Dresselhaus,\Commenton'PolarizedRama nstudyof alignedmultiwalledcarbonnanotubes'-reply," PhysicalReviewLetters ,vol. 85,no.16,pp.3545{3545,2000. [217]M.S.DresselhausandP.C.Eklund,\Phononsincarbonn anotubes," AdvancesinPhysics ,vol.49,no.6,pp.705{814,2000. [218]C.ThomsenandS.Reich,\DoubleresonantRamanscatte ringingraphite," PhysicalReviewLetters ,vol.85,no.24,pp.5214{5217,2000. [219]F.TuinstraandJ.L.Koenig,\Ramanspectrumofgraphi te," Journalof ChemicalPhysics ,vol.53,no.3,pp.1126{1130,1970. [220]J.Kurti,V.Zolyomi,A.Gruneis,andH.Kuzmany,\D oubleresonant RamanphenomenaenhancedbyvanHovesingularitiesinsingl e-wallcarbon nanotubes," PhysicalReviewB ,vol.65,no.16,pp.165433.1{165433.9,2002. [221]T.Ohsaka,F.Izumi,andY.Fujiki,\Raman-spectrumof anatase,TiO 2 ," JournalofRamanSpectroscopy ,vol.7,no.6,pp.321{324,1978. [222]A.Chaves,R.S.Katiyar,andS.P.S.Porto,\Coupledmo deswitha1 symmetryintetragonalBaTiO 3 ," PhysicalReviewB ,vol.10,no.8,pp. 3522{3533,1974. [223]G.A.Tompsett,G.A.Bowmaker,R.P.Cooney,J.B.Metso n,K.A. Rodgers,andJ.M.Seakins,\Theraman-spectrumofbrookite ,TiO 2 (pbca, z=8)," JournalofRamanSpectroscopy ,vol.26,no.1,pp.57{62,1995. [224]W.F.Zhang,Y.L.He,M.S.Zhang,Z.Yin,andQ.Chen,\Ra man scatteringstudyonanataseTiO 2 nanocrystals," JournalofPhysicsDAppliedPhysics ,vol.33,no.8,pp.912{916,2000. [225]BersaniD.,P.P.Lottici,andX.-Z.Ding,\Phononcon nementeectsinthe ramanscatteringbyTiO 2 nanocrystals," AppliedPhysicsLetters ,vol.72,pp. 73{75,1998. [226]H.C.Choi,Y.M.Jung,andS.B.Kim,\Sizeeectsinther amanspectraof TiO 2 nanoparticles," VibrationalSpectroscopy ,vol.37,no.1,pp.33{38,2005. [227]W.S.Cleveland,\Visualandcomputationalconsidera tionsinsmoothing scatterplotsbylocallyweightedtting,"1978.

PAGE 205

186 [228]W.S.Cleveland,\Robustlocallyweightedttingands moothingscatterplots," JournaloftheAmericanStatisticalAssociation ,vol.74,pp.829{836, 1979. [229]W.S.Cleveland,\LOWESS:Aprogramforsmoothingscat terplotsbyrobust locallyweightedtting," TheAmericanStatistician ,vol.35,pp.54,1981. [230]V.V.Yakovlev,G.Scarel,C.R.Aita,andS.Mochizuki, \Short-rangeorder inultrathinlmtitaniumdioxidestudiedbyramanspectros copy," Applied PhysicsLetters ,vol.76,no.9,pp.1107{1109,2000. [231]T.Kamada,M.Kitagawa,M.Shibuya,andT.Hirao,\Stru ctureand propertiesofsilicontitanium-oxidelmspreparedbyplas ma-enhanced chemicalvapor-depositionmethod," JapaneseJournalofAppliedPhysics Part1-RegularPapersShortNotes&ReviewPapers ,vol.30,no.12B,pp. 3594{3596,1991. [232]G.Cappelletti,C.L.Bianchi,andS.Ardizzone,\XPSs tudyofthesurfactant lmadsorbedontogrowingtitaniananoparticles," AppliedSurfaceScience vol.InPress,CorrectedProof,2006. [233]M.P.Casaletto,L.Lisi,G.Mattogno,P.Patrono,andG .Ruoppolo,\An XPSstudyoftitania-supportedvanadylphosphatecatalyst sfortheoxidative dehydrogenationofethane," AppliedCatalysisA:General ,vol.267,no.1-2, pp.157{164,2004. [234]H.Jensen,A.Soloviev,Z.Li,andE.G.Sogaard,\XPSan dftirinvestigation ofthesurfacepropertiesofdierentpreparedtitanianano -powders," Applied SurfaceScience ,vol.246,no.1-3,pp.239{249,2005. [235]N.J.Price,J.B.Reitz,R.J.Madix,andE.I.Solomon,\ Asynchrotron XPSstudyofthevanadia-titaniasystemasamodelformonola yeroxide catalysts," JournalofElectronSpectroscopyandRelatedPhenomena ,vol. 98-99,pp.257{266,1999. [236]G.Silversmit,G.DeDoncker,andR.DeGryse,\Aminera lTiO 2 (001) anatasecrystalexaminedbyXPS," SurfaceScienceSpectra ,vol.9,pp.21{29, 2002. [237]P.K.ChuandL.Li,\Characterizationofamorphousand nanocrystalline carbonlms," MaterialsChemistryandPhysics ,vol.96,no.2-3,pp.253{277, 2006. [238]L.Huang,H.Jiang,J.Zhang,Z.Zhang,andP.Zhang,\Sy nthesisofcopper nanoparticlescontainingdiamond-likecarbonlmsbyelec trochemical method," ElectrochemistryCommunications ,vol.8,no.2,pp.262{266,2006.

PAGE 206

187 [239]W.Lisowski,E.G.Keim,A.H.J.vandenBerg,andM.A.Sm ithers, \Thermaldesorptionofdeuteriumfrommodiedcarbonnanot ubesandits correlationtothemicrostructure," Carbon ,vol.44,no.5,pp.974{982,2006. [240]Z.Konya,I.Vesselenyi,J.Kiss,A.Farkas,A.Oszko,a ndI.Kiricsi,\XPS studyofmultiwallcarbonnanotubesynthesisonni-,v-,and ni,v-zsm-5 catalysts," AppliedCatalysisA:General ,vol.260,no.1,pp.55{61,2004. [241]T.I.T.Okpalugo,P.Papakonstantinou,H.Murphy,J.M cLaughlin,and N.M.D.Brown,\HighresolutionXPScharacterizationofche micalfunctionalisedmwcntsandswcnts," Carbon ,vol.43,no.1,pp.153{161,2005. [242]S.Point,T.Minea,B.Bouchet-Fabre,A.Granier,andG .Turban,\XPS andnexafscharacterisationofplasmadepositedverticall yalignedn-doped mwcnt," DiamondandRelatedMaterials ,vol.14,no.3-7,pp.891{895,2005. [243]E.Ech-chamikh,A.Essafti,Y.Ijdiyaou,andM.Azizan ,\XPSstudy ofamorphouscarbonnitride(a-c:n)thinlmsdepositedbyr eactiverf sputtering," SolarEnergyMaterialsandSolarCells ,vol.InPress,Corrected Proof,2006. [244]S.S.Roy,R.McCann,P.Papakonstantinou,P.Maguire, andJ.A.McLaughlin,\Thestructureofamorphouscarbonnitridelmsusinga combinedstudy ofnexafs,XPSandramanspectroscopies," ThinSolidFilms ,vol.482,no. 1-2,pp.145{150,2005. [245]M.S.Dresselhaus,G.Dresselhaus,R.Saito,andA.Jor io,\Ramanspectroscopyofcarbonnanotubes," PhysicsReports ,vol.409,pp.4799,2005.

PAGE 207

BIOGRAPHICALSKETCH GeorgiosPyrgiotakiswasborninHeraklion,Greece,in1977 .In2000he graduatedwithaB.S.degreeinphysicsfromUniversityofCr ete.In2003he earnedhisM.S.fromUniversityofFloridainmaterialsscie nceandengineering.He enjoyscookingandmixingmusic. 188