Study of Gas-Phase Reactions in Tio Production and Functionalised Thin Films

Study of Gas-Phase Reactions in TiO2
Production and Functionalised Thin Films

EngD Thesis
Ben Blackburn
Supervisors
Prof. Ivan Parkin
Prof. Claire Carmalt
Disclaimer
I Ben Blackburn confirm that the work presented in this thesis is my own. Where information has
been derived from other sources, I confirm that this has been indicated in the thesis.
Signed: ..................................................................................
Acknowledgements
This project was funded by contributions from Huntsman Pigments, the EPSRC as well as the UCL
EngDoc Center.
I would like to thank Professors Ivan Parkin and Claire Carmalt for their guidance and support
throughout this project, as well as Dr. Stephen Sutcliffe and Andrew Brown at Huntsman Pigments
and Additives for providing and industrial insight to the work done.
In addition thanks are owed to Dr. Jared Crane for his assistance in handling and analysis of TiCl4
samples and inorganic synthesis. Dr. Caroline Knapp, Dr. Peter Marchand, Dr. David Pugh for
assistance in determining molecular structures using X-ray crystallography, Dr. Kristopher Page
for assistance in procuring reagents, Dr. Steve Firth for instruction in Raman Spectrosocopy, Dr.
Andreas Kafizas for assistance with X-ray diffraction analysis, Dr. Michael Parkes and Dr. Kirsty Karu
for assistance mass spectroscopy, Dr. Michael Powell, Dr. Carlos Sotelo-Vazquez, Dr. Raul Quesada-
Cabrera for their instruction in testing the thermochromic nature of thin films determining solar
modulation and Dean Brett, Charalampos Drosos for their contributions to the deposition of thin
films.
This thesis is dedicated to Amy, whose love, friendship and patience has made the last four years
easier in every possible way.
iii
Abstract
Titanium dioxide (TiO2) is a pigment used in the whitening of millions of products from paint to
foodstuffs. Produced via purification of ores using the Chloride Process, the mined TiO2 ore is
chlorinated along with all of the impurities. The resulting TiCl4 is a liquid at room temperature and
can be separated from its impurities using distillation. V 2O5, an impurity present in this ore is
converted to vanadium oxytrichloride (VOCl3) which is too similar in boiling point to TiCl4 to
separate via distillation and instead is removed via addition of an oil, which preferably binds to
VOCl3.
This work, sponsored by Huntsman Pigments and Additives, seeks to better understand this
reaction, as well as other aspects of the Chloride Process. This is achieved by analysing industrial
samples as well as producing lab scale models for the process and investigating inorganic synthesis
of analogous compounds [TiCl4{CH2(CO2CH(CH3)2}2] [1a], [(µ-O)(TiCl3{CH2(CO2CH(CH3)2}2)]
[1b], [TiCl4{CH2(CO2CH(C6H5)}2] [2], [TiCl4{C2H4(CO2CH3)}2] [3], VOCl3+glycerol tribenzoate [4],
[VOCl2(acac)] [5], [{VOCl2(CH2(CO2Et)2}4] [6] and [VOCl2{C2H4(CO2Et)2}]n [7] that were characterised
via single crystal X-ray diffraction analysis (with the exception of 4)
Alongside this study, multiple side projects were undertaken that sought to use some of the novel
materials produced in materials synthesis, due to the useful nature of both TiO2 and VO2 in window
coating applications, the abundance of oxygen in the products making them good contenders for
metal oxide precursors. This included the deposition of TiO2 /SiO2 thin films for photocatalytic self
cleaning applications, VO2 thin films for energy saving windows and VC/VN particles for mechanical
hardening applications.
In addition, a new system for the in-situ study of high temperature gas phase inorganic reactions
via mass spectroscopy was developed and used to study the reactivity of gaseous TiCl4 and VOCl3,
as well as several other species including vanadium chloride, titanium isopropoxide and butyltin
trichloride considered to be of greater interest to the field of thin film deposition.
iv
Contents
Disclaimer ii
iii
Acknowledgements iv
Abstract v
Contents x
List Of Figures xviii
List Of Tables xiv
List Of Schemes xx
List Of
Abbreviations
1. Introduction 1
1.1. Project Overview 1

1.2. Titanium Dioxide 2
1.2.1. Structure and Phases 2
1.2.2. Uses of TiO2 3
1.3. Titanium Dioxide Production 6
1.3.1. The Sulfate Process 7
1.3.2. The Chloride Process 9
1.3.3 Issues with the Tioxide Production Process 14
1.4 Titanium(IV) Tetrachloride 16
2. Analysis of Samples from Industrial TiO₂ Production 19
19
2.1. NMR study of Industrial TiCl4 Samples
2.1.1. Experimental 19
2.1.1.1. Making an NMR sample 19
2.1.1.2. Distilling Titanium Chloride Samples 21
2.1.2. Results and Discussion 22
2.1.2.1. Initial Study 22
2.1.2.2. Comparison of Recent and Older TiCl4 Samples 24
2.1.2.3. Lab-distilled Primary Titanium Tetrachloride Samples 26
2.1.2.3a. The First 20 ml of Distillate 26
2.1.2.3b. The Second 60 ml of Distillate 26
2.1.2.3c. The Third 20 ml of Distillate 29
2.1.2.4. Investigation of Aging in Distilled Fractions 30
2.1.3. Conclusion 32
v
2.2. Analysis of the Black Precipitate in the Tetra Samples 35
2.2.2. Results and Discussion 37
2.2.2.1. NMR spectroscopy 37
2.2.2.2. Toluene Washings 37
2.2.2.3. Mass Spectroscopy 38
2.2.2.4. Infrared Spectroscopy 39
2.2.2.5. Energy dispersive X-ray Spectroscopy 41
2.2.2.6. X-ray Photoelectron Spectroscopy 43
2.2.3. Conclusion 44
3. Reaction of Titanium Chloride and Vanadium Oxytrichloride 47

with Diester and Triester Molecules as Analogues for Oil
3.1. Introduction 47
3.2. Experimental 50
3.2.1. Synthesis of tetrachloro(diisopropyl malonate)-titanium(IV) [1] 50
3.2.2. Synthesis of tetrachloro(dibenzyl malonate)-titanium(IV) [2] 51
3.2.3. Synthesis of tetrachloro(diethyl succinate)-titanium(IV) [3] 51
3.2.4. Synthesis of complex of titanium(IV) chloride with glycerol tribenzoate [4] 52
3.2.5. Synthesis of dichloro(oxo)(2, 4-Pentanedione) vanadium(V) [5] 53
3.2.6. Synthesis of dichloro(oxo)(diethyl malonate) vanadium(V) [6] 53
3.2.7. Synthesis of VOCl3 Bis-isopropyl malonate Complex [7] 54
3.2.8. Synthesis of dichloro(oxo)(diethyl succinate) vanadium(V) [8] 54
3.2.9. Synthesis Of Glycerol tribenzoate VOCl3 [9] 54
3.2.10. Crystallography 55
3.3. Results and Discussion 56
3.3.1. Interaction of TiCl4 with Diester Ligands 56
3.3.1.1. Tetrachloro(diisopropyl malonate)-titanium(IV) [1a] 57
3.3.1.2. Bis-trichloro-µ-oxo-(diisopropyl malonate)-titanium(IV) [1b] 61
3.3.1.3. Tetrachloro(dibenzyl malonate)-titanium(IV) [2] 63
3.3.1.4. Tetrachloro(diethyl succinate) titanium(IV) [3] 66
69
3.3.2. Interaction of TiCl4 with Triglyceride Analogue Glycerol Tribenzoate [4]
3.3.3. Interaction of Glycerol Tribenzoate and Diesters With VOCl3 73
3.3.3.1. Dichloro(oxo) (2,4-Pentanedione) vanadium(V) [5] 74
3.3.3.2. Dichloro(oxo) (diethyl malonate) vanadium(V) [6] 78
3.3.3.3. Dichloro(oxo)(diethyl succinate) vanadium (V) [7] 83
3.3.3.4. Reaction of VOCl3 with Other Species 88
3.4. Conclusion 91
vi
4. Vanadium Removal Flow System 93
4.2.1. Soya Bean Oil Experiments 98
4.2.2. Analysis of Naphthenic Oils 99
4.3.1. Soya Bean Oil Analysis 101
4.3.1.1. TiCl4 + Soya Oil 101
4.3.1.2. VOCl3 + Soya Oil 107
4.3.1.3. TiCl4 + 1000 ppm VOCl3 + Soya Oil 108
4.3.1.4. TiCl4 + 40,000 ppm VOCl3 + Soya Oil 109

4.3.2. Naphthenic Oil Analysis 110
4.3.2.1. NMR of the Naphthenic Oils 112
4.3.2.2. Flow Cell Experiments 113
4.3.2.3. NMR Experiments with Flavex 977s 116
4.4. Conclusion 118
5. Using Mass Spectrometry to Study Gas Phase Reactions in 120

Chemical Vapour Deposition

5.1.1. Atmospheric Pressure CVD 120
5.1.2. Using Mass Spectrometry to Study CVD 124
5.1.3. Mass Spectrometry 126
5.2.1. Preliminary Experiments 134
5.2.1.1. Toluene Test Run 134
5.2.1.2. Purge Flow Test 135
5.2.2. Precursor Deposition 136
5.2.2.1. Ethyl Acetate 139
5.2.2.2. Titanium (IV) Tetrachloride 139
5.2.2.3. Vanadium (V) Oxytrichloride 140
5.2.2.4. Titanium(IV) Isopropoxide 141
5.2.2.5. Butyltin (IV) Trichloride 141
5.2.2.6. Vanadium(IV) Chloride 142
5.3.1. Ethyl Acetate 143
5.3.2. Titanium Tetrachloride 145
5.3.2.1. Mass Spectra Analysis 145
5.3.2.2 Thin Film Analysis 148
i) X-ray diffraction 148
ii) Scanning Electron Microscopy 148
vii
5.3.3. Vanadium Oxytrichloride 150
i) X-ray Diffraction 152
5.3.4. Butyltin Trichloride 157
5.3.5. Titanium Isopropoxide 162
5.3.5.1. Mass Spectrometry Analysis 163
5.3.5.2. Thin Film Analysis 167
5.3.6. Vanadium Tetrachloride 169
5.3.6.1. Mass Spectrometry Analysis 170
5.3.6.2. Thin Film Analysis 174
ii) X-ray Photoelectron Spectroscopy 175
iii) Scanning Electron Microscopy 177
5.4. Conclusion 179
6. Aerosol Assisted CVD of Photocatalytically Active TiO₂/SiO₂ 181

Microfractal Thin Films

6.1.1. Aerosol Assisted CVD 181
6.1.2. TiO2-SiO2 Composite Films 184
6.2.1. Synthesis of [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)]2 [10] 187
6.2.2. Reaction of TiCl4 with trimethylsilyl acetate to produce [11] 187
6.2.3. Reaction of TiCl4 with silicon tetraacetate to produce [Ti3O2Cl3(CH3CO2)5] 188
[12]
6.2.4. Reaction of TiCl4 with bis-trimehylsilylmalonate to produce [13] 188
6.2.5. Deposition of Thin Films 189
6.2.6. Analytical Methods 190
6.2.7. Photocatalytic Testing 191
6.3.1. Reaction of TiCl4 with methyl trimethylsilylacetate to give 192
[Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)] [10]
6.3.2. Reaction of TiCl4 with trimethylsilylacetate to give [11] 195
6.3.3. Reaction of TiCl4 with silicon tetraacetate to give [Ti3O2Cl3(CH3CO2)5 [12] 196
6.3.4. Reaction of TiCl4 with bis-trimethylsilylmalonate to give [13] 200

6.3.5. Characterisation Thin Films 201
viii
6.3.6. Water Contact Angle Measurements 208
6.3.7. Photocatalytic Testing 210
6.4. Conclusion 212
7. Derivatives of VOCl3 and Their Uses in Materials Synthesis 214
7.1 Introduction 214

7.1.1. VO2 Thin Films 214
7.1.2. Synthesis of vanadium carbide and vanadium nitride 217
220
7.2. Formation of VO2 (M) thin films and VOx Nanoparticles Using the Single
Source Precursor [{VOCl2(CH2(COOEt2)}4] (6)
7.2.2. Deposition of Thin Films 220
7.2.3. Tungsten Doping 221
7.2.4. Synthesis of Nanoparticles 221
7.2.5. Annealing to produce VC + VN powders 222
7.2.6. Results and DiscusSiOn 222
7.2.7. Film deposition 222
7.2.8. VOx Nanoparticle Synthesis and Characterisation 231
238
7.3. Using adducts of VOCl3 and 2,4 pentadione, diethyl malonate and
diethyl succinate as precursors for Vanadium Nitride and Carbide
7.3.1.2. Synthesis of vanadium nitride and carbide powders 238
7.3.2 Results and Discussion 239
7.3.2.1. Synthesis And Characterisation Of vanadium nitride Powders 239
7.3.3. Synthesis and characterisation of vanadium carbide powders 242
7.4. Conclusions 244
8. Conclusions 245
9. References 248
ix
List of Figures
Figure 1: Crystal structures of the three polymorphs of TiO2. 3
Figure 2: Breakdown of TiO2 use in 2003 compared with 2013 4
Figure 3: Price of rutile ore a) and anatase ore b) from 2012 to 2015 and retail price of
TiO2 from 2010 to 2.015 (c) 7
Figure 4: Simplified schematic representations of the sulfate and chloride processes. In
the chloride process, all of the chemical purification takes place between the
‘chlorination’ and ‘oxidation’ steps. 9
Figure 5: Rutile TiO2 ore before and after the purification via the chloride process, as
well as the intermediate TiCl4 samples before and after purification 13
Figure 6: a) Rutile TiO2 ore before37 and after38 the purification via the Chloride
Process, as well as the intermediate TiCl4 samples before and after purification. b) The
structure of the vegetable oil,39 used to separate the vanadium oxychloride that gives
the intermediate TiCl4 its yellow colouring. 14
Figure 7: Collected black residue in the furnace, following cleaning. 15
Figure 8: Different morphologies of TiO2 and composites that have been deposited from
TiCl4 using CVD 17
Figure 9: apparatus used to collect NMR samples of TiCl4 20
Figure 10: Schematic of TiCl4 distillation with packed column 21
Figure 11: Comparison of the 1H NMR spectra of primary TiCl4 undistilled (black), plant
distilled at Greatham (blue) and laboratory distilled (red) 23
Figure 12: ¹H NMR spectra of Primary TiCl₄ samples produced approximately 15
months apart, with the shifts between 0.6 and 0.0 ppm enhanced (inset) 25
Figure 13: ¹H NMR spectra of Plant distilled TiCl4 samples produced approximately 15
months apart, with the shifts between 0.6 and 0.0 ppm enhanced (inset) , with the
2011 sample enhanced in order to see much less abundant peaks that are near the
baseline 25
Figure 14: ¹H NMR spectra of all three fractions of Distilled Primary TiCl4 using an
unpacked column 27
Figure 15: ¹H NMR spectra of all three fractions of Distilled Primary TiCl4 using a
packed column 27
Figure 16: Comparison of ¹H NMR peak pattern thought to relate to siloxane rings,
distilled primary (second fraction) in packed and unpacked distillation columns 28
Figure 17: Shifts present in the ¹H NMR spectra of the final fractions of Distilled
Primary TiCl4 indicating the present of an Isopropyl group, attached to an electron
withdrawing group 30
Figure 18: ¹H NMR spectra of the first and second fractions of distillate collected from
distilling 110 ml of TiCl4, using a packed column, collected two months apart 31
x
Figure 19: Structures of four potential siloxane species forming within the TiCl4
samples, all with boiling points between 100 and 134 ˚C, and all with NMR spectra that
would only give highly upfield peaks.58–61 33
Figure 20: Structures of four potential chlorinated methyl silane species forming
within the TiCl₄ samples, all with boiling points between 80 and 134 ˚C,62–65 and all
with NMR spectra that would only give highly upfield peaks. 34
Figure 21: Mass spectrum of a sample of DCM that has been shaken with the black
solid. Structure and simulated spectrum of plasticizer molecule bis 2-
ethylhexylphthalate inset 39
Figure 22: Infrared Spectrum of the black solid, as prepared using a pressed KBr disc 40
Figure 23: Compound peak in the O1s environment form the XPS spectrum of the
black solid, showing organic oxygen (dotted + dashed), H2O (dotted) and TiO2 (dashed) 44
Figure 24: proposed stepwise process for the formation of the precipitate, consisting
of TiO2 particles with bis 2-ethylhexyl phthalate molecules coordinating to the surface. 46
Figure 25: Ester species used as analogues for the triglyceride backbone of oil in the
reaction with TiCl4 and VOCl3 48
Figure 26: Reaction of diethyl malonate and titanium (IV) chloride, observed by both
Sobota et al. and Kakkonen et al.78,79 49
Figure 27: NMR spectra of bis-isopropyl malonate and the product of its reaction with
TiCl4 [1a], all peaks exhibiting a downfield shift indicating coordination to the titanium
metal centre 58
Figure 28: ORTEP representation of the crystal structure of tetrachloro(diisopropyl
malonate)-titanium(IV) [1a], with thermal ellipsoids at the 50% probability level.
Hydrogen omitted for clarity. 58
Figure 29: ORTEP representation of 1b with thermal ellipsoids at the 50% probability
level. Hydrogen atoms omitted for clarity 62
Figure 30: ¹H NMR spectra of dibenzyl malonate and the product of its reaction with
TiCl4 [2], all peaks exhibiting a downfield shift indicating coordination to the titanium
metal centre 64
Figure 31: ORTEP representation of tetrachloro (dibenzyl malonate)-titanium(IV) [2]
with thermal ellipsoids at the 50% probability level. Hydrogen atoms omitted for
clarity. 65
Figure 32: NMR spectra of diethyl succinate and the product of its reaction with TiCl4
[3], all peaks exhibiting a downfield shift indicating coordination to the titanium metal
centre 67
Figure 33: ORTEP representation of the crystal structure of tetrachloro (diethyl
succinate)-titanium(IV) [3] with thermal ellipsoids at the 50% probability level.
Hydrogen atoms omitted for clarity 68
Figure 34: ¹H NMR spectra of glycerol benzoate and the product of its reaction with
TiCl4 [4], all peaks exhibiting a downfield shift indicating coordination to the titanium
metal centre 70
xi
Figure 35: ORTEP representation of 4 as determined by single crystal X-ray diffraction,
with thermal ellipsoids at the 50% probability level. Hydrogen atoms omitted for
clarity 71
Figure 36: NMR spectra of 2,4 pentadione and the product of its reaction with VOCl 3
[5], The signals relating to the non-conjugated variant of acac are completely removed
in the spectrum of 5, indicating that coordinating to the VOCl 3 is responsible for loss of
a proton. Both peaks exhibiting a downfield shift indicating coordination to the
vanadium metal center 76
Figure 37: ORTEP representation of crystal structure of [VOCl2(acac)] [5] with thermal
ellipsoids at the 50% probability level. Hydrogen omitted for clarity. 77
Figure 38: NMR spectra of diethyl malonate and the product of its reaction with VOCl 3
[6], all peaks exhibiting a downfield shift and experiencing broadening indicating
coordination to the vanadium metal centre 80
Figure 39: ORTEP representation of crystal structure of [{VOCl2(CH2(COOEt2)}4] [6] with
thermal ellipsoids at the 50% probability level. Three of the four diethyl malonate
molecules and all hydrogen atoms omitted for clarity. 81
Figure 40: ¹H NMR spectra of diethyl succinate and the product of its reaction with
VOCl3 [7], all peaks exhibiting a downfield shift and experiencing broadening indicating
coordination to the vanadium metal centre 86
Figure 41: ORTEP representation of crystal structure of [VOCl2{C2H4(CO2Et)2}]n (7) with
thermal ellipsoids at the 50% probability level. Hydrogen omitted for clarity 87
Figure 42: ¹H NMR spectra of bis-isopropyl malonate and the product of its reaction
with VOCl3 [8], all peaks exhibiting a downfield shift and experiencing broadening
indicating coordination to the vanadium metal centre 89
Figure 43: ¹H NMR spectra of bis-isopropyl malonate and the product of its reaction
with VOCl3 [9], all peaks exhibiting a downfield shift and experiencing broadening
indicating coordination to the vanadium metal centre 89
Figure 44: Schematic of a Soxhlet extractor106 94
Figure 45: Gaseous metal chloride flowcell from a modified Soxhlet extractor 96
Figure 46: ¹H NMR spectra of the starting material and the contents of flasks 1, 2 and
the Soxhlet extractor from experiment in which gaseous TiCl4 was reacted in soya oil.
Spectra taken in CDCl₃. The figure shows the region between 3.5 and 5.5 ppm 102
Figure 47: ¹H NMR COSY spectrum of the contents of flask 1, from experiment in
which gaseous TiCl4 was reacted in soya oil. Spectra taken in CDCl₃. The figure shows
the region between 3.5 and 5.5 ppm 102
Figure 48: Proposed general structure of the main NMR product of the reaction of
TiCl4 and soya bean oil in the flow cell 103
xii
Figure 49: Comparison of the spectrum of the contents of flask 1 produced from
putting TiCl4 through 4 cycles with soya bean oil, compared with the simulated NMR
spectrum of 1, 3 dichloro propan-2-acetate.The simulation does not factor in long
range coupling of the 4H peak which was added manually. Simulation carried out
using NMRDB NMR predictor software 107–109 104
Figure 50: Suggested mechanism for the formation of the 1, 3 dichloro.2 acetyl
propane species observed in the NMR of the flow-cell distillate 105
Figure 51: ¹H NMR spectra of the starting material and the contents of flask 2 and the
Soxhlet extractor from experiment 2, in which gaseous VOCl3 was reacted in soya oil.
Spectra taken in CDCl₃. The figure shows the region between 0.6 and 5.5 ppm 107
the Soxhlet extractor from experiment in which gaseous TiCl4 with 1000 ppm VOCl3
was reacted in soya oil. Spectra taken in CDCl₃. The figure shows the region between
3.5 and 5.5 ppm 108
the Soxhlet extractor from experiment 4 in which gaseous 40,000 ppm VOCl 3 In TiCl4
was reacted in soya oil. Spectra taken in CDCl₃. The figure shows the region between
3.5 and 5.5 ppm 109
Figure 54: Example structures of naphthenic acids, derived from the hydrolysis of
naphthenic oil. 111
Figure 55: Comparison of the ¹H NMR spectra of four naphthenic oils: Flavex 977s
(blue), Ergon Hygold L550 (red), Gravex 926 (green) and Edelex 926s (purple). Spectra
taken in CDCl₃. The figure shows the region between 0.8 and 2.4 ppm 112
Figure 56: Pure TiCl4 and Flavex 977(s) after 4 cycles, contents of Soxhlet adaptor. TiCl4
appears to break down the naphthenic, which then carries over, staining the distillate
red 114
Figure 57: Gravex 926 with TiCl4 during first cycle. The red distillate is seen to gather
on top off the clean TiCl4 that starts off in the Soxhlet, before falling though when it
fills. 115
Figure 58: 25 ml TiCl4 and 1ml VOCl3 with Flavex 977(s). Flow system is stained; solid in
Soxhlet suggests that whatever carries over continues reacting. 115
the Soxhlet extractor from experiment in which gaseous TiCl4 was reacted in
naphthenic Flavex 977(s) oil Spectra taken in CDCl₃. The figure shows the region
between 0.0 and 5.4 ppm 116
1H
Figure 60: Comparison of the NMR spectra of Flavex 977s compared with the
contents of the Soxhlet extractor after three cycles of TiCl4 (red), VOCl3 (green), TiCl4 +
40,000 ppm VOCl3 (purple) and toluene (yellow) . Spectra taken in CDCl₃. The figure
shows the region between 0.8 and 2.4 ppm 117
Figure 61: Schematic of basic model of film growth in APCVD 121
Figure 62: Diagram of a quadrupole mass analyser 129
Figure 63: Schematic of the triple filter analysers used in the Hiden HPR60 129
xiii
Figure 64: Schematic representation of entire mass spectrometry CVD setup 132
Figure 65: Annotated photograph of mass spectrometer CVD setup 132
Figure 66: Schematic of CVD apparatus and mass spectrometer setup 133
Figure 67: Intensity of the mass spec peak at m/z - 91 representing the molecular mass
of toluene plotted alongside the flow rate through the toluene bubbler. 135
Figure 68: Intensity of the mass spec peak at m/z - 91 representing the molecular
mass of toluene plotted alongside the flow rate of the purge flow 136
Figure 69: Assigned mass spectrum of ethyl acetate as passed through an atmospheric
pressure chemical vapour deposition reactor at 600 ˚C (a) and comparison of ethyl
acetate fragmentation patterns over the temperature range of 300 to 600 ˚C (b) 144
Figure 70: Mass spectrum of titanium(IV) chloride as passed through an atmospheric
pressure chemical vapour deposition reactor at a temperature of 600 ˚C 146
Figure 71: Mass spectrum of titanium(IV) chloride and ethyl acetate as passed through
an atmospheric pressure chemical vapour deposition reactor at a temperature of 600
˚C, with an enhanced image of the peaks between m/z - 200 and 330 , showing the
newly formed adduct peaks 147
Figure 72: X-ray diffraction pattern of the films deposited from TiCl4 and ethyl acetate
onto float glass at 300 and 600 ˚C using the APCVD / mass spectrometry apparatus,
alongside an anatase TiO2 standard.186 149
Figure 73: SEM image of the film deposited from TiCl4 and ethyl acetate onto float
glass at 600 ˚C using the APCVD / mass spectrometry apparatus 150
Figure 74: Mass spectrum of vanadium(V) oxytrichloride as passed through an
atmospheric pressure chemical vapour deposition reactor at a temperature of 600 ˚C 151
Figure 75: Mass spectrum of vanadium(V) oxytrichloride and ethyl acetate as passed
through an atmospheric pressure chemical vapour deposition reactor at a
temperature of 600 ˚C, with an enhanced image of the peaks between m/z - 200 and
330 , showing the newly formed adduct peaks 152
Figure 76: X-ray diffraction pattern of the films deposited from VOCl3 and ethyl acetate
onto float glass at 600 ˚C (top) and 300 ˚C (bottom) using the APCVD / mass
spectrometry apparatus, alongside standards for monoclinic (PDF 72-0514)188 and
tetragonal (PDF 80-0690) VO2.189 154
Figure 77: SEM images of the film deposited from VOCl3 and ethyl acetate onto float
glass at 300 (a) and 600 ˚C (b) using the APCVD / mass spectrometry apparatus 156
Figure 78: Mass spectrum of butyltin trichloride as passed through an atmospheric
Figure 79: Mass spectrum of butyltin trichloride and ethyl acetate as passed through
an atmospheric pressure chemical vapour deposition reactor at a temperature of 600
˚C and 300 ˚C, with an enhanced image of the peaks between m/z - 250 and 500,
showing the newly formed adduct peaks 159
xiv
Figure 80: X-ray diffraction pattern of the film deposited from BuSnCl3 and ethyl
acetate onto float glass at 600 ˚C using the APCVD / mass spectrometry apparatus,
alongside a rutile SnO2 standard 161
Figure 81: SEM image of the film deposited from TiCl4 and ethyl acetate onto float
glass at 600 ˚C using the APCVD / mass spectrometry apparatus 162
Figure 82: Mass spectrum of titanium isopropoxide as passed through an atmospheric
Figure 83: Possible fragmentation points on the TTIP parent ion with the formulas and
masses of the fragments denoted in red 164
Figure 85: Proposed formation of an oxo-bridge between two titanium isopropoxide
molecules at high temperatures 166
Figure 86: Proposed mechanism for the loss of the elimination of an isopropanol
molecule from TTIP and its dimer at high temperatures. 167
Figure 87: X-ray diffraction pattern of the films deposited from TiCl4 and ethyl acetate
onto float glass at 300 and 600 ˚C using the APCVD / mass spectrometry apparatus,
alongside an anatase TiO2 standard 168
Figure 88: SEM images of the film deposited from titanium isopropoxide onto float
glass at 300 (a) and 600 ˚C (b) using the APCVD / mass spectrometry apparatus 168
Figure 90: Mass spectra of APCVD depositions of VCl4 at 600 ˚C with flows of ethyl
acetate at 0.2 L min-1 (top) and 0.05 L min-1 (bottom) 172
Figure 91: Mass spectra of APCVD depositions of VCl4 at 600 ˚C with flows of ethyl
acetate at 0.6 L min-1 with peaks in the range of m/z - 180 to 300 enhanced. 173
Figure 92: X-ray diffraction pattern of the films deposited from VCl4 and ethyl acetate
onto float glass at 600 ˚C using the APCVD / mass spectrometry apparatus with ethyl
acetate flow rates of 0.6 L min-1(top), 0.2 L min-1 (middle) and 0.05 L min-1 (bottom)
and alongside standards for monoclinic (PDF 72-0514) and tetragonal (PDF 80-0690)
VO2. 175
Figure 93: Detailed XPS scans of: a) V 2p of film deposited from 0.2 Lmin-1VCl4 and 0.6
L min-1 ethyl acetate and b) V 2p of film deposited from 0.2 Lmin -1VCl4 and 0.6 L min-1,
V(IV) (dotted line) vs. V(V) (dashed line) 176
Figure 94: SEM images of the film deposited from a 0.6L min-1 flow of VCl4 onto float
glass with 0.6 L min-1 (a), 0.2 L min-1 (b) and 0.05 L min-1 (c) flows of ethyl acetate using
the APCVD / mass spectrometry apparatus 178
Figure 95: Schematic diagram of the AACVD apparatus 182
Figure 96: Schematic representation of various deposition pathways possible in
AACVD 183
xv
Figure 97: Silicon containing ligands reacted with TiCl4 used in the synthesis of 8 (top
left), 9 (top right) 10 (bottom left) and 11 (bottom right) 186
Figure 98: ¹H NMR spectra of methyl trimethylsilylacetate and the product of its
reaction with TiCl4 [10], all peaks exhibiting a downfield shift indicating coordination to
the titanium metal centre 193
Figure 99: ORTEP representation of crystal structure of 10 with thermal ellipsoids at
the 50% probability level. Hydrogen omitted for clarity 194
Figure 100: ¹H NMR spectra of trimethylsilylacetate and the product of its reaction
with TiCl4 [11], showing a series of new peaks suggesting that multiple products form 196
Figure 101: ¹H NMR spectra of trimethylsilylacetate and the product of its reaction
with TiCl4 [12], showing a series of new peaks suggesting that multiple products form 197
Figure 102: ORTEP representation of crystal structure of 12 with thermal ellipsoids at
the 50% probability level. Hydrogen omitted for clarity 199
Figure 103: SEM images of the microstructure of film deposited from
[Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)] (8) at 500 ˚C at magnifications of x200 (a), 1,200 (b),
x15,000 (c) and 2,600 x (d) with addition of ratios of titanium to silicon at different
points in the film, as measured by EDX spectroscopy. x15,000 magnified images of
films deposited at 550˚C and 450˚C (e) + (f) respectively. 202
Figure 104: XRD patterns of films deposited from AACVD of [Ti{SiMe3CH2CO2Me}Cl3(µ-
Cl)] [10] at various temperatures compared with that of anatase TiO2 (bottom) 204
Figure 105: Raman Spectra of films deposited from AACVD of

[Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)] (1) at various temperatures compared with that of
anatase TiO2 (bottom) 205
Figure 106: Overlay of the depth profile XPS peaks for Si (2p) with darkening colour
representing deeper penetration below the films surface a), Atomic percentages of
titanium and silicon in the Composite films as measured using depth profile XPS b) 206
Figure 107: UV/Vis Absorption spectra of films deposited from AACVD of

[Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)] (10) at various temperatures compared with That of
anatase TiO2 206
Figure 108: Tauc plots of films deposited from AACVD of [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)]
(1) at various temperatures used to estimate the band gap of these films. 207
Figure 109: Comparison of the behaviour of water droplets on thin films of TiO2 and
the TiO2-SiO2 composite deposited at 450 °C, 500 °C and 550 °C before UV irradiation
(a, c, e and g respectively) and after UV irradiation. (b, d, f, h respectively) 209
Figure 110: Bar graph showing the Formal Quantum Efficiency (FQE) of films
deposited from AACVD of [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)] (1) at various temperatures
compared with That of anatase TiO2 and plain glass, as measured by the UV catalysed
decomposition of the stearic acid on the films surface. 210
Figure 111: Overlay of the FT-IR spectra of the stearic acid on the surface of the 500 ˚C
film over a two hour period of UV irradiation with spectra taken every 20 minutes 212
xvi
Figure 112: Diagram of the behaviour of visible and near infrared radiation upon
contact with a VO2(M) coated window, at below (left) and above (right) the MST. 215
Figure 113: SEM images of the VO2 film deposited from [6] in toluene at 550 °C using
a 2% O2 in N2 carrier gas with no dopant (a), 2.8% [W(OEt)6] (b) and 5.0% [W(OPh)6]
(c) 226
Figure 114: XRD patterns of the VO2 film deposited from [6] in toluene at 550 °C using
a 2% O2 in N2 carrier gas, both with and without 5.0% [W(OPh)6] 227
Figure 115: XPS scans of: a) V 2p3/2 of a VO2 (M) thin film from the AACVD of
[{VOCl2(CH2(COOEt2)}4] [6] at 550 ˚C (showing V(IV) at lower binding energy (thin solid
line) vs. V(V) (dotted line)), b) V 2p3/2 of a W-doped VO2 (M) thin film (film 12) at 550 ˚C
and c) corresponding W4f scan 227
Figure 116: Variable temperature UV/Vis/NIR spectra showing the change in optical
properties of thin films of undoped and W-doped VO2 thin films deposited by AACVD
of synthesised vanadium precursor solutions. Solar modulation values (ΔT sol) for each
film have been included in the Figure. a) VO2 (M) film deposited at 540 ˚C, b) 550 ˚C, c)
560 ˚C and d) W-doped VO2 (M) thin film deposited at 550 ˚C 229
Figure 117: TEM images and EDS spectra of sample A, pre-annealing (a and b) and
post annealing (d and e). C and G are the EDS spectra of the pre-annealed and post
annealed sample 1 respectively. Also shown are TEM images of samples B, C and H (g,
h and i repectively) 234
Figure 118: PXRD diffraction patterns for as synthesised powders, a) and b) and
following annealing treatment, c) and d). λ= 0.7093 Å. 235
Figure 119: XPS scans of: a) V 2p of Sample 1, VO2 nanomaterials before annealing and
b) after annealing showing V(IV) at lower binding energy (thin solid line) vs. V(V)
(dotted line). 236
Figure 120: Composite figure for the VN sample derived from complex 2. a) an EDS
spectrum demonstrating the presence of vanadium and nitrogen with little carbon
present. Copper emanated from the copper mesh TEM grid. b) Fitted N1s XPS
spectrum showing the regions assigned as VN and VC. c) and d) show XRD patterns of
VN derived from complex 2 compared to a VN ICSD standard (22321). e) is an HRTEM
image of a VN crystal with the <111> plane of VN highlighted. 239
Figure 121: Composite Figure for the VC sample derived from 2. a) s an EDS spectrum
demonstrating the presence of vanadium and high levels of carbon. Copper emanated
from the copper mesh TEM grid. b) Fitted V 2p XPS spectrum showing the regions
assigned as VC and VO. c) and d) show XRD patterns of VC derived from 6 242
xvii
List of Tables
Table 1: Complete list of industrial TiCl4 samples analysed 20
Table 2: Atomic percentages of the elements present in the black solid as
determined by EDS 42
Table 3: Crystallographic Data and Selected bond lengths angles for the crystal
structure of 1a the product of reaction TiCl4 and bis isopropyl malonate 59
structure of 1b the product of reaction TiCl4 and bis isopropyl malonate and
subsequent air exposure 62
structure of 2 the product of reaction TiCl4 dibenzyl malonate 65
structure of 3 the product of reaction TiC4 diethyl succinate 68
structure of 4 the product of reaction TiCl4 and glycerol tribenzoate 71
structure of 5 the product of reaction VOCl3 and 2,4 pentadione 77
structure of 1a the product of reaction VOCl3 and diethyl malonate 81
structure of 1a the product of reaction VOCl3 and diethyl succinate 87
Table 11: Key experiments involving soya oil carried out using the flow system 99
Table 12: Experiments run through the flow cell apparatus using naphthenic oils 100
Table 13: Comparison of acidity of naphthenic oils and soya bean oil 111
Table 14: List of precursors and conditions used in the mass spectrometry analysis of
CVD experiment 136
Table 15: Approximate vapour pressure and molar flow rates of VCl4 and ethyl
acetate during mass spectrometry APCVD depositions 171
structure of 1a the product of reaction TiCl4 and methyltrimethylsilyl acetate [10] 194
structure of 1a the product of reaction TiCl4 and silicon tetraacetate [12] 199
Table 18: Contact angles of before and after UV irradiation 209
Table 19: AACVD depositions carried out using [VOCl₂(DEM)]₄ [6] with and without
the presence of tungsten dopants 224
Table 20: Compiled reaction conditions for the formation of VOx nanoparticles using
[6] (*No reaction, non-colloidal blue solution formed. ǂAddition of 1,4-
tetradecanediol) 232
Table 21: EDS at% ratios of vanadium to oxygen in nanomaterial samples. Bracketed
quantities are those in the annealed spectra 235
xviii
List of Schemes
Scheme 1: Reaction of TiCl4 with bis-isopropyl malonate to give tetrachloro(bis-

isopropylmalonate)-titanium(IV) [1a] 57
Scheme 2: Reaction of 1a with atmospheric water to give bis-trichloro-µ-oxo-
(diisopropyl malonate)-titanium(IV) [1b], facilitated by the elimination of HCl 61
Scheme 3: reaction of TiCl4 with glycerol tribenzoate give tetrachloro(dibenzyl
malonate)-titanium(IV) [2] 64
Scheme 4: reaction of TiCl4 with glycerol tribenzoate give tetrachloro(diethyl
succinate)-titanium(IV) [3] 67
Scheme 5: Schematic of the synthesis of compound 4, formed by the reaction of
TiCl4 in hexane with glycerol tribenzoate 70
VOCl3 in hexane with diethyl malonate 76
VOCl3 in hexane with diethyl malonate 80
VOCl3 in hexane with diethyl succinate. 86
Scheme 9: Schematic of the decomposition of TTIP at high temperatures to give
TiO2.
163
xix
List of Abbreviations
AACVD aerosol assisted chemical vapour deposition

Acac acetyl acetonate (2-4, pentadione)
ALD atomic layer deposition
APCVD atmospheric pressure chemical vapour deposition
aq aqueous
CVD chemical vapour deposition
EDX energy dispersive x-ray spectroscopy
FEG free electron gun
FT-IR Fourier transform infrared spectroscopy
GADDS general area detector diffraction system
MS mass spectroscopy
NMR nuclear magnetic resonance spectroscopy
PECVD plasma enhanced chemical vapour deposition
SEM scanning electron microscopy
TEM transmission electron microscopy
TTIP titanium isopropoxide
UV ultra-violet
UV/Vis ultraviolet/ visible spectroscopy
XPS X-ray photoelectron spectroscopy
XRD X-ray diffraction
xx
Chapter 1
Introduction
1.1. Project Overview
This project was funded by Huntsman Pigments and Additives, branch of Huntsman Chemicals
that produces titanium dioxide, a pigment used in the production of paints. TiO2 is produced
on an industrial scale by The Chloride Process, a series of steps in which TiO2 ore is converted
to TiCl4 and the impurities removed via distillation. One impurity in particular, VOCl3, is
removed via the injection of oil, which is found to react exclusively with the gaseous VOCl 3,
leaving the TiCl4 to distil into the next part of the process. This technique is widely used, but
little is known as to why the TiCl4 does not react with the oil.
This project seeks to further the understanding of the Chloride Process. The reactions of TiCl4
and VOCl3 with oils have been studied using a number of methods. Initially industrial samples
of TiCl4 taken from the Chloride Process were analysed using nuclear magnetic resonance
(NMR) spectroscopy. Following this various β-diketo esters and triesters were used as
analogues for the triglyceride backbone of the oil and reacted with both species. The products
of these reactions were crystallised, and the products used to explain the oils preferred
reactivity in the Chloride Process.
The gas phase interactions of TiCl4 and VOCl3 were then studied by building a laboratory scale
replica of the distillation phase of the Chloride Process, and the samples produced were
analysed. Following this a system for studying the contents of a high temperature vapour
1
deposition reactor was designed and built, and used to study the reactivity of TiCl4 and VOCl3
in the gas phase directly, as well as several other molecules frequently used in the vapour
deposition of thin films.
In addition to data on The Chloride Process, several new single source precursors for the
chemical vapour deposition (CVD) of VO2 and composited TiO2-SiO2 thin films were produced
as an indirect result of this project. The precursors were characterised and the resulting films
grown and were analysed.
1.2. Titanium Dioxide
Titanium Dioxide (TiO2) is a heavily studied material that has a variety of uses, with over 4.5
million tonnes produced annually.1 Titanium dioxide is mostly obtained through mining and
purification of the ore, which is carried out by one of two main processes (vida infra). In recent
years a synthetic alternative to production has been developed using titanium slag from
metallurgical processes.2
1.2.1. Structure and Phases
Titanium dioxide has three common mineral phases: Rutile, Brookite and Anatase (Figure 1).
All of these phases consist of distorted octahedra with titanium surrounded by six oxygen
atoms. The orientation of these octahedral alters depending on the phase present. The most
thermodynamically stable phase is Rutile TiO2. Anatase and Brookite convert to this phase
when heated to 915 ˚C and 730 ˚C, respectively. 3
2
Figure 1: Crystal structures of the three polymorphs of TiO2.4
Rutile TiO2 has a primitive tetragonal geometry with space group 4/m 2/m 2/m. The Rutile
unit cell consists of two TiO6 octahedra each coordinated to 6 oxygen atoms which are in turn
shared by 3 titanium atoms. The resulting greater structure appears as an array of corner
sharing octahedra.
Anatase TiO2 also has tetragonal geometry and shares its space group with rutile TiO2;
however the structures differ in that the TiO6 octahedra are linked at two corners (edge
sharing). This makes the Anatase phase less dense and hard than the Rutile (Anatase has a
Moh hardness of 5.5 vs. 6.5 For Rutile).5 Despite this, Anatase frequently finds mechanical
uses due to its lower abrasiveness compared to that of Rutile.3 Brookite TiO2 has a crystal
structure with Orthorhombic geometry and space group 2/m 2/m 2/m. Unlike the other two
polymorphs it is brown in colour.
1.2.2. Uses of TiO2
Titanium dioxide’s primary uses derive from its extremely high refractive index of 2.73 for
Rutile and 2.55 for Anatase.6 This makes it the ‘whitest white pigment’ available, zinc oxide
3
having a refractive index of 2. This has led to titanium dioxide being used in whitening of a
large number of products, with a market worth $15.17 billion in 2014. 7
Figure 2 gives a breakdown of the uses of TiO2 pigments. By far the largest portion of the TiO2
pigments market is taken up by coatings (paint), however it is also heavily used in plastics,
where it is added to the resin in order to whiten the product,8 as well as the food, makeup
and clothing industries.9 As Figure 2 shows, over the last 15 years these uses, as well as the
demand for energy and antimicrobial applications, have seen a huge increase in demand for
TiO2, with ‘other’ now responsible for over half of the annual consumption of TiO2
TiO₂ Usage 2003 TiO₂ Usage 2013

Paper
10% Other
9% Paint
28%
Plastic Paint Other

Plastic
24% 57% 56%
10%
Paper
6%
Figure 2: Breakdown of TiO2 use in 20031 compared with 2013.10
The refractive index of titania is found to relate to the particle size, as well as the polymorph
present. For instance when TiO2 is encased in a polymer resin with a refractive index of 1.5,
then the refractive index is found to increase with particle size up to 0.19 μm, at which point
there is a drop off.11 This is important in the later stages of TiO2 production. In addition to
pigmentation, TiO2 has found use in a wide variety of applications, as described below.
Self-Cleaning Coatings. Titanium Dioxide’s self-cleaning capabilities are well studied. The
material exhibits super-hydrophilicity as well as photocatalysing the breakdown of dirt.12 This
4
process occurs as a result of photoexcitation. TiO2 has a band gap of 3.2 eV for the Anatase
phase,13 putting it in the UV region of the Electromagnetic spectrum. When the material
comes into contact with UV light, electrons are excited to the conductance band, leaving
positive holes. Some of these charge carriers migrate to the surface of the material, where
the holes reduce the organic material on the surface and the electrons combine with
atmospheric oxygen to form superoxide radicals, which aggressively break down organic
materials.14
Antibacterial Coatings: The light induced generation of the superoxide radicals is also
responsible titanium dioxide’s effectiveness in killing bacteria. This is believed to be due to a
mechanism known as lipid peroxidation,15 in which TiO2 catalyses the formation of radicals,
which go on to remove a proton from an unsaturated fatty acid chain. The organic radical then
reacts with dioxygen, forming a peroxide radical that removes another proton, propagating a
chain reaction. This chain of reactions is believed to prevent the cell membrane from
functioning, leading to cell death. Nanostructured TiO2 films have been shown to kill 99.9% of
bacteria in light conditions typical of hospitals.16
Photovoltaic Devices: Titanium dioxide is frequently utilised in the field of solar power
generation, most prominently in the field of dye sensitized solar cells. First developed by
Gratzel and Oregan in 1991,17 these are electric circuits with wide band gap semiconductor
anodes, that have been sensitized by dye molecules that are excited by visible light, allowing
them to conduct in the visible region. TiO2 has a wide band gap of 3.2 eV and is commonly
used as an anode material.
5
1.3. Titanium Dioxide Production
Industrial titanium dioxide production consists of two main possible processes for purifying
titanium dioxide ore, the Chloride Process and the Sulfate Process. There are multiple
different types of TiO2 ore available, with each process using different ores. The three most
commonly used types in the Chloride Process are rutile, anatase and ilmenite. Rutile and
anatase are mainly used in the Chloride Process, with rutile considered the best option due
to its higher purity.18 At the beginning of this project the price of these ores had increased
greatly year-on-year for some time, with Rutile TiO2 ore priced at $3000/tonne in 2012.
Ilmenite (FeTiO3) mainly used in the Sulfate Process but also sporadically in the Chloride
Process and is considerably cheaper at $285/tonne,19. In the intervening years, however, the
cost of titanium containing mineral has fallen dramatically (Figure 3).
There are also ‘slags’ available for each of the processes,20 which are mineral ores with greater
percentages of Fe2O3. These are cheaper but harder to work with due to the high volume of
solid iron compounds damaging waste disposal relative to rutile.21
6
Figure 3: Price of rutile ore a) and anatase ore b) from 2012 to 2015 and retail price of TiO2 from
2010 to 2015 (c),22
1.3.1. The Sulfate Process
The Sulfate Process was the first major method for producing TiO2 from its ores on a large
scale (Figure 4A).23 The dried, ground ilmenite ore is ‘digested’ using sulfuric acid to separate
7
the FeTiO3 into separate sulfates. This results in a solid ‘cake’ which is leached using dilute
sulfuric acid to produce a viscous liquid, which contains predominantly TiOSO 4 and FeSO4.24
The liquid is cooled, allowing some FeSO4 to be removed by precipitation of iron sulfate
heptahydrate. With the solid particulates removed, the TiOSO4 is hydrolysed to TiO(OH)2 by
heating with dilute sulfuric acid. The material is then filtered and redissolved in water
repeatedly and a bleaching agent, such as aluminium is added.25 At the hydrolysis stages of
the Sulfate Process it is usual to add seed solution, consisting of a suspension of TiO2 particles
that can alter the size and shape of the titanium dioxide particles that are produced.26 Dopants
can also be added during the process to affect the size of the particles formed (usually
aluminium oxide or zinc oxide).27 In the final stage of the Sulfate Process the TiO(OH)2 is
calcinated at 1000 ˚C in a rotary kiln, to produce TiO2 of the correct particle size and
distribution.
The Sulfate Process is technically the simpler of the two, and has a far cheaper feedstock,
however it has several drawbacks. The sulfuric acid must be recycled which is difficult, or
neutralized, which is costly.25 Also there is the environmental issue of disposing of large
amounts of highly acidic waste product. Since the innovation of the Chloride Process, the
Sulfate process has been in a state of decline.3
8
Figure 4: Simplified schematic representations of the Sulfate and Chloride Processes.28 In the
Chloride Process, all of the chemical purification takes place between the ‘chlorination’ and
‘oxidation’ steps.
1.3.2. The Chloride Process
The Chloride Process was first developed in the 1950s by Dupont as an alternative to the
Sulfate process (Figure 4B). It involves taking impure TiO2 ore and converting it to titanium
tetrachloride. Impurities are then removed via condensation and distillation, following which
the titanium tetrachloride is reverted back to TiO2 (Figure 5). This is the process employed by
Tioxide and is thus investigated in this project. The process itself can be separated into 6
stages as described below and is shown in Figure 6.29
9
i) Chlorination: The ore is fed into the chlorinator with chlorine gas and coke. The
chlorinator acts as a pressure vessel, with the reactants heated to 900 - 1100 ˚C. This leads to
the following chlorination reaction:
TiO2(s) + 3/2 C(s) + 2 Cl2(g) → TiCl4(g) + CO(g) + ½ CO2(g)
The coke is added in order to facilitate the chlorination of the metal by acting as a reducing
agent, forming CO, which extracts oxygen from the TiO2, increasing the rate of the
chlorination.3 This stage also leads to the chlorination of all of the metal oxide impurities in
the ore, most prominently:
2 VO2(s) + 3 Cl2(g) → 2 VOCl3(g) + O2
2 Fe2O3(s) + 6 Cl2(g) → 4 FeCl3(g) + 3 O2
2 Al2O3(s) + 3 Cl2(g) → 2 AlCl3(g) + 3 O2
SiO2(s) + 2 Cl2(g) → SiCl4(g) + 2 O2
Although not shown in the above equations, the coke functions a reducing agent in all of the
above reactions in addition to the chlorination of TiO2. These are all carried through to the
quench tower with the TiCl4 and the carbon dioxide.
ii) Removal of Solid Impurities: when the gaseous chlorides exit the chlorinator they are
met with a stream of liquid TiCl4 from further along the process which cools the gas from
1000 ˚C to below 230 ˚C. The lower temperature causes many of the chloride impurities, most
importantly FeCl2 to condense out of the gas stream as solids. The stream of metal chloride
gaseous material is pushed with nitrogen through a dirt cyclone (a cooled pipe with sloped
10
walls that forces the gas into a cyclone that pushed the solid into the walls via centrifugal
force). The tangential angle causes the solid to fall along the walls into a ‘dirt handling’ tank
where the waste is neutralized, whilst the lower boiling chlorides including TiCl4 exit through
the top of the cyclone, still in their gaseous form.
iii) Gas Cleaning: At this point, the only remaining chlorides in the TiCl4 gas stream should
be SiCl4, VOCl3 and AlCl3. These are problematic to remove due to their having similar boiling
points to TiCl4. The gas cleaning tower enables the removal of the VOCl3 by injection of a small
amount of mineral or vegetable oil, which complexes to the VOCl3.30 This leads to an increase
in the boiling point of VOCl3 enabling it to be removed via distillation.31 The gas cleaning tower
also injects a small amount of water and sometimes sodium chloride into the gaseous TiCl4
flow. This is thought to to remove the AlCl3 by converting it to AlOCl which is insoluble in
TiCl4.32 The reaction is believed to take place in two stages, with a titanium oxychloride
intermediate that facilitates the oxidation of the aluminium, as shown below.33 The small
amount of water added leads to partial oxidation of the TiCl4 to TiOCl2, which exists as a highly
unstable coordination complex that rapidly reacts with the AlCl3 due to the entropically
favourable products. 34
TiCl4(g)(excess) + H2O(g) → TiOCl2.2HCl(l)
TiOCl2.2HCl(l) + AlCl3(l) → TiCl4(g) + AlOCl(s) + HCl(g)
The injection of water takes place in the gaseous phase. The high temperature in the tower
(230 ˚C) leads to corrosion by the hot acid produced. Furthermore, to ensure complete AlCl3
removal an excess of water to AlCl3 is required, meaning that some TiOCl2 remains after this
stage, however it is recycled and converted back into TiCl4 minimising loss.
11
iv) Condensation: The condensation of the TiCl4 takes place in two columns. The primary
column cools the gas flow with a stream of liquid TiCl4, pumped through the system and
cooled with a spiral heat exchanger. The majority of the TiCl4 is condensed in this column
(around 95%). Any uncondensed gas moves through to a second condenser, which
refrigerates the stream to -12 ˚C causing the remaining TiCl4 to condense and impurities that
are less soluble in the TiCl4 to crash out as solid.
v) Distillation: This stage is a simple distillation of the liquid TiCl4 to remove remaining
impurities, mostly the oil added during the gas cleaning phase, combined with the vanadium
containing impurities. The liquid is fed through to a tank that is heated using a calandria. A
calandria is a thermosiphon boiler that heats a flow of liquid via thermal transfer from a series
of tubes heated with high pressure steam.35 The TiCl4 exits that calandria as a mixture of
boiling liquid and vapour. It is then circulated back into the distillation column where the
gaseous TiCl4 rises to the top of the column. The ‘pure’ TiCl4 gas stream is then condensed and
either moved into storage tanks or straight to the furnace for oxidation.
vi) Oxidation: The final stage of purification is the oxidation of the TiCl4 back to its original
form. The resulting TiO2 pigment is rutile. The liquid TiCl4 is heated using a furnace to 350 ˚C
becoming a vapour. The vapour is fed into a reactor and mixed with a flow of oxygen. The
oxygen is typically heated to a far higher temperature than the TiCl4, usually 900-1000 ˚C.36
This ensures that the reaction initiates and goes to completion as it cools. The resulting
powder is ground to a specific particle size and coated in order to produce a specific pigment
type, Further treatment allows specific products such as the Deltio® Freeflow TiO2 pigment
which is treated in order to make it flow more easily, leading to a higher throughput,
minimising dust.37
12
13
a)
b)
Figure 6: a)Rutile TiO2 ore before38 and after39 the purification via the Chloride Process, as well as the
intermediate TiCl4 samples before and after purification. b) The structure of the vegetable oil,40
used to separate the vanadium oxychloride that gives the intermediate TiCl4 its yellow colouring.
1.3.3. Issues with the Chloride Process
Although the Chloride Process is considered less problematic than the Sulfate Process in terms
of environmental damage, the lengthier process, as well as the aggressive nature of the
chlorides formed, can lead to several issues with production41 These problems can lead to
plant down time, at a cost of approximately £300,000 per day
i) Vanadium Carryover: When not enough oil is injected into the system, the vanadium
oxytrichloride is not fully removed from the TiCl4 before oxidation to TiO2. This causes the
resulting pigment to appear yellow. This product cannot be sold and is disposed of into landfill
at high cost. TiCl4 appears yellow at VOCl3 concentrations greater than 5 ppm, although most
processes aim for closer to 1-2 ppm. Addition of too much oil, however, can lead to fouling of
14
components further along the process. Much of the research relating to the Chloride Process
involves researching the oils used and how vanadium can be removed more effectively.
Figure 7: Collected black residue in the furnace, following cleaning.
ii) Fouling of Plant Components: Many of the components in the process have their life
cycles cut very short due to the build up of material on their surfaces, which damages their
effectiveness (Figure 7). The three main areas where fouling is an issue are the spiral heat
exchangers in the gas cleaning section, the calandria in the distillation tower and the furnace
in which The TiCl4 is heated before oxidation. The coating appears to be black, however due
to the air sensitive nature of TiCl4 and the enclosed nature of the components, the only way
to make it safe to collect a sample is to completely hydrolyse it, which would in turn greatly
diminish the value of any compositional data collected.
iii) Tetra Sample Aging: Throughout the process, samples of the liquid TiCl4 known as
‘tetra’ can be tapped off, from the gas cleaning, primary condensate and post distillation
points. This is an excellent way of determining where exactly a problem might be if the TiO2
produced were to look ‘dirtier’ than it otherwise should. It can also be a useful selling point
for the process, to observe the quality of the product. Generally speaking the tetra should be
15
as clear and colourless as possible, becoming more so, the further along the process, as more
of the impurities are removed. It has been observed that over time the samples taken from
the process at all points seem to ‘age’. This is seen in a gradual discolouration of the liquid,
going from colourless to yellow. This is observed in many samples of TiCl4, including those
bought commercially from firms such as Sigma Aldrich. The samples from Huntsmanare also
observed to precipitate a black solid over a period of weeks.
iv) Packing of the Distillation Column: When TiCl4 is industrially distilled, it is common to
pack the column with glass. This is thought to increase the surface area of the passage that
the gaseous TiCl4 has to navigate and ‘catch’ some of the lower boiling impurities that would
otherwise make it through the column. The packing has to be replaced at least once a year,
resulting in costly downtime and there has been some debate as to how much a role in the
separation of the impurities the column packing actually plays. One of the distillation studies
carried out in this work was repeated with and without a vertical column packed with glass
beads in order to determine how much a difference their presence makes.
1.4. Titanium(IV) Tetrachloride
For the vast majority of the Chloride Process the titanium is present in the form of titanium(IV)
tetrachloride. This is a wellstudied compound due to its unusual properties and high
reactivity. TiCl4 is one of only three transition metal halides that is a liquid at room
temperature (the others are VCl4 and SnCl4). This is because whilst most metal halides form
polymeric octahedral species such as the edge sharing octahedral structure of ZrCl 4,42
titanium tetrachloride is a tetrahedral monomeric species due to the relatively high enthalpy
of the Ti-Cl bond (494 kJmol-1)43 and lewis acidic nature of the compound making it more
16
stable than the equivalent edge sharing lattice Infrared spectroscopy shows the Ti-Cl bond to
have a length of 2.185 Å,44 which has been verified computationally.45
Titanium(IV) tetrachloride’s strong Lewis acid character is responsible for the majority of its
uses. It is the starting point for the synthesis of a number of titanium complexes in inorganic
synthesis and has been used in the formation of novel CVD precursors, such as titanium(IV)
arsenide complexes.46,47 TiCl4 itself is also frequently used in chemical vapour deposition,
usually in the deposition of TiO2 nanostructures,48 however it has also been used in
combinatorial CVD in the formation of titanium oxynitrides.49
Figure 8: Different morphologies of TiO2/composites that have been deposited from TiCl4 using CVD
In organic chemistry it is frequently used in coupling reactions, such as the condensation
reaction of alcohols and aromatics and aldol coupling reactions, for instance the Mukaiyama
aldol coupling which traps the enolate by forming a silyl enol ether.50
By far the most well known use of titanium tetrachloride is that of Ziegler Natta catalysis, a
specific class of heterogeneous catalysts that are used in the polymerization of alkenes. They
typically involve a transition metal halide, usually TiCl4 with an alkyl metal co-catalyst (AlEt3).
The resulting polymers are noted for their high linearity and taticity.51 52 Zeigler And Natta
both separately proposed a mechanism for the polymerization involving the propagation
of free radicals, however this does not explain the high level of steroselectivity.53,54,55
17
. Difficulty arises when investigating titanium tetrachloride due to the highly Lewis acidic
nature of the TiCl4 centre. The moisture sensitive nature of TiCl4 results in a hydrolysis
reaction.56
TiCl4(l) + 2H2O(g) → TiO2(s) + 4HCl(g)
This makes analysis of samples of TiCl4 and its use in synthesis challenging and requires a
moisture free atmosphere, necessitating the use of Schlenk techniques. This was a challenge
that had to be overcome in the first part of this project, which required characterizing a large
number of samples of TiCl4 taken directly from the industrial process in order to identify the
impurities that were carried through the different stages of the Chloride Process.
18
Chapter 2
Analysis of Samples from Industrial TiO2 Production
This section details work done in relation to improving the understanding of the Chloride
Process, by analysing various samples sent from the Greatham TiO2 production facility in order
to identify potential impurities within the samples that could lead to downtime or
discolouration of the product further down the production process. These samples consisted
of liquid TiCl4 samples which were analysed using NMR spectroscopy and black solid
precipitates that forms within these samples over time, which was separated from the TiCl4,
oxidised and identified using various analytical techniques.
2.1. NMR study of Industrial TiCl4 Samples
Initial experiments largely consisted of developing a means of reliably creating NMR samples
of TiCl4 samples and then investigating the products present in different fractions from the
distillation of those samples.
2.1.1. Experimental
2.1.1.1. Making an NMR sample
NMR spectra were obtained for a number of TiCl4 samples collected from the Chloride process
at different points and different times. The highly air sensitive nature of the samples meant
that they had to be made up under an atmosphere of nitrogen using Schlenk techniques. This
was achieved using the apparatus displayed in Figure 9.
19
Experiment Sample Point in Process Year
1 Primary condensate 2012
Initial Study 2 Plant distilled 2012

Primary condensate
3 2012
(lab distilled)
2013
4 Primary condensate
(new)
2011
5 Primary condensate
(2 yrs old)
Sample Aging
2013
6 Plant distilled
(new)
2011
7 Plant distilled
(2 yrs old)
Primary condensate 2013
8
1st 20 ml (packed) (new)
9
2nd 60 ml (packed) (new)
Fractional Primary condensate 2013
10
Distillation of 3rd 20 ml (packed) (new)
Primary Primary condensate 2013
11
Condensate 1st 20 ml (packed) (new)
12
2nd 60 ml (packed) (new)
13
3rd 20 ml (packed) (new)
Figure 9: apparatus used to
collect NMR samples of TiCl4 Table 1: Complete list of industrial TiCl4 samples analysed
The sample was made by using a canula to move 3 drops of the TiCl4 into a three necked round
bottom flask, under a flow of nitrogen. The flask was equipped with an elongated neck piece,
capped with a subaseal and a release needle, with a screw-top NMR tube inside.
20
Figure 10: Schematic of TiCl4 distillation with packed column
To the TiCl4, 2 ml of CDCl3 (Sigma Aldrich, 99.8 atom% D, anhydrous, dried over activated
molecular sieves) was injected into the flask through a side arm using a plastic syringe and
deposable needle. The solution was then taken into a second syringe via a needle through the
top subaseal, before being injected directly into the NMR tube. The tube was then screwed
shut under a flow of nitrogen and removed. Samples were collected in this manner for primary
condensate, plant distilled and laboratory distilled TiCl4, as well as the distilled fractions of
plant distilled and non-distilled TiCl4.
2.1.1.2. Distilling Titanium Chloride Samples
Two distillations were carried out following the collection of the initial NMR data: one for the
TiCl4 that had passed through the distillation column at the site in Greatham and one that had
21
not. In each case an air sensitive distillation was carried out by heating a 110 ml sample of
TiCl4 in a two necked round bottom flask to 120 ˚C using an isomantle, under a flow of nitrogen
(Figure 10). The gaseous TiCl4 and volatile impurities passed up a 10 cm column into a cold
water condenser.
At the bottom of the condenser was a three pronged distillation receiver, with two 25 ml flasks
and one 100 ml flask. These are marked at the 20 ml and 60 ml levels respectively. Upon filling
the first flask to 20 ml the join was turned allowing the second to fill to 60 ml and finally the
third to 20 ml. The flasks were then stoppered and NMR samples immediately made up (vide
infra).
2.1.2. Results and Discussion
2.1.2.1. Initial Study
NMRs taken were from the primary condensate and the distilled portion of the plant (Figure
9). Another sample was made by distilling the primary condensate in the lab under nitrogen.
The samples used were all over one year old and were highly discoloured with black
precipitate in all except for the lab distilled primary condensate.
The spectra showed several collections of peaks that can be divided into three categories
according to how their relative intensity and therefore the concentration of the source
molecule can be reduced by distilling the sample: (a) those removable by plant or lab
distillation (b), those removable by lab distillation only and (c) those not removable by either
plant or lab distillation.
a) The peaks removed by plant distillation have chemical shifts between 0.8 and
1.5 ppm. Although these peaks were dramatically reduced by distillation, they appear
22
in all three spectra. They are likely the result of the oil breaking down into alkyl species
that boil lower than TiCl4
b) The peaks that can be removed by laboratory distillation fall between 0.4 and
0.6 ppm. These are likely caused by silane compounds present in the material.
These occur as a result of SiCl4 reacting with the oil species.
1
Figure 11: Comparison of the H NMR spectra of primary TiCl4 undistilled (black), plant distilled at
Greatham (blue) and laboratory distilled (red)
c) The relative intensity of peaks at 0.65 ppm and 0.75 ppm were increased
by the distillation process, indicating that they were not affected by distillation
as removal of other impurities would lead to an increase in their relative
concentration.
23
Due to the large number of species present it would be impossible to reliably define
the peak integrals and very difficult to ascertain exact structures of the species
present. Interestingly, none of the unbroadened peaks in the spectra display any
coupling, suggesting a number of methyl species. On the other hand, some of the peaks
appeared rather broad in some instances, which may mask coupling
2.1.2.2. Comparison of Recent and Older TiCl4 Samples
Following the initial testing, ‘fresh’ TiCl4 samples were obtained from Huntsman.
Samples were once again taken from the primary condensation, and post-distillation
parts of the process. NMR spectra were taken of each sample and compared with the
equivalent 15 month old samples (Figures 12 + 13).
All samples saw a dramatic increase in the number and relative intensity of the ‘silane
peaks’ between 0.0 ppm and 0.7 ppm in the newer samples over the equivalent older
one. The only exception to this was the two peaks at 0.64 ppm and 0.73 ppm, both of
which were far more prominent in the older samples. There are a number of possible
reasons for this occurrence. It is probable that the silane species react slowly over time
within the TiCl4 samples, resulting in the observed difference in peak patterns. This study
cannot confirm this however as the samples were taken from the plant at different
points in time meaning that the different patterns could be due to a change in the
manufacturing process or the starting material. The firm were unable to confirm this.
To confirm that there is a slow reaction within the TiCl4 samples, NMR spectra were
collected of several samples two months after they arrived. The spectra collected show
changes in the peak patterns.
24
1
Figure 12: H NMR spectra of Primary TiCl4 samples produced approximately 15 months
apart, with the shifts between 0.6 and 0.0 ppm enhanced (inset).
1
Figure 13: H NMR spectra of Plant distilled TiCl4 samples produced approximately 15
months apart, with the shifts between 0.6 and 0.0 ppm enhanced (inset), with the 2011
sample enhanced in order to see much less abundant peaks that are near the baseline.
25
2.1.2.3. Lab-distilled Primary Titanium Tetrachloride Samples
With the study of samples taken directly from the plant complete, the next step was to
simulate the distillation part of the column in the laboratory setting. 110 ml samples of
primary tetra were distilled at 140 ˚C, with three samples collected from each: a) The first
20 ml to distil over, b) The following 60 ml of distillate, c) The final 20 ml. NMR spectra were
obtained from all three fractions (Figure 14) . The experiment was repeated, with the column
packed with 6 mm diameter glass beads, to emulate the packing in one of the distillation
columns at the Greatham plant (Figure 15).
2.1.2.3a. The First 20 ml of Distillate
The first fraction showed a large amount of peak broadening not seen in the other two. This
implies that some paramagnetic species with low boiling points were brought over early in
the distillation, mostly likely a vanadyl species vanadium (IV) (although it should be noted
that both samples were colourless when tested, so the concentration of vanadyl species
would be below 5 ppm). The majority of peaks seen were downfield from 0.7 ppm, which is
indicative of methylsilyl species. The packed column’s distillate was observed to give a less
broadened spectrum, implying that the packing may be able to retard distillation of VOCl3.
2.1.2.3b. The Second 60 ml of Distillate
The spectra appeared similar to the first 20 ml although far less broadened. This
suggests that all of the paramagnetic species have already distilled over and are
therefore all lower boiling than the TiCl4 itself. The vanadium is likely causing a
broadening through weak, labile coordinating to some of the molecules of
the alkyl species distilling over.
26
1
Figure 14: H NMR spectra of all three fractions of Distilled Primary
TiCl4 using an unpacked column
1
Figure 15: H NMR spectra of all three fractions of Distilled Primary
TiCl4 using a packed column
27
There were no discernible differences between the two peak patterns for the packed and
unpacked column however there are differences in the relative intensities of some of these
peaks. A set of peaks between 0.38 ppm and 0.49 ppm, appears to consist of two related
pairs of peaks, at 0.42 and 0.49 ppm and 0.39 and 0.47 ppm (Figure 16).
Figure 16: Comparison of 1H NMR peak pattern thought to relate to siloxane rings, distilled primary
(second fraction) in packed and unpacked distillation columns
28
It is possible that the peaks relate to the same group in the axial and equatorial positions on
a siloxane ring. A study by Pelletier et al. showed similar patterns with methyl and hydrogen
substituted cyclosiloxanes.57
The intensities of these peaks in relation to each other remain roughly the same over the two
spectra, increasing and diminishing by the same amount. Pair 1 is two singlets, whereas the
other pair appears as two multiplets, with a small coupling constant. The difference in the
shifts between these two ‘pairs’ peaks is too great to be down to J-coupling.
2.1.2.3c. The Third 20 ml of Distillate
Rather than the multitude of singlets seen in the earlier samples, the majority of the peak
patterns in the final distillate showed splitting, especially the more upfield peaks. This is
expected as these patterns likely relate to larger alkyl groups, whereas before they related to
smaller methyl species. These groups would be heavier and therefore would distil over after
the methyl species. Further downfield there are two shifts: a doublet at 1.60 ppm and a
septet at 5.15 ppm, with an integral ratio of 6:1. This pattern resembles the well known
pattern of isopropyl alcohol.58 Figure 17 is an enhanced image of these two environments.
The septet for the central carbon’s proton is shifted further downfield, indicating the
presence of an ester or acyl-chloride group. This pattern is by far the most prominent in the
spectra and is likely a product of the breakdown of the soya oil at high temperatures. Samples
of plant distilled TiCl4 do not feature this pattern at all. Therefore it can be inferred that the
final fraction is likely to bear the least resemblance to the actual distilled product of the three
studied.
29
1
Figure 17: Shifts present in the H NMR spectra of the final fractions of Distilled Primary TiCl4
indicating the present of an Isopropyl group, attached to an electron withdrawing group
2.1.2.4. Investigation of Aging in Distilled Fractions
Inspection of the fractionated products two months after the initial study was carried out
revealed a surprising result. The first and third fractions in both cases turned from colourless
to dark yellow as expected, however the middle fractions remained relatively colourless. All
samples were stored in the same type of container so this difference is not environmental.
The implication of this is that there are multiple species within the distilled tetra that lead to
the discolouration/precipitation. Retaking the 1H NMR spectra of the first and second
fractions from the packed column indicated that this could be a result of reactions of the
silane compounds. The broad hump in the 1H NMR of the first fraction is no longer present,
and the intensities peaks from 0.4 - 0.5ppm have altered relative to each other. The spectra
are compared in Figure 18. The emergence of doublets with a shift of 1.5 - 1.7 ppm is also
observed, however it is possible that this is due to the breakdown of the
30
1
Figure 18: H NMR spectra of the first and second fractions of distillate collected from
distilling 110 ml of TiCl4, using a packed column, collected two months apart
syringe used to make up the 1H NMR sample. Conversely, the 1H NMR spectra of the second
fraction, which had remained colourless, were almost identical, with both peak patterns and
integrals unchanged. It should be noted that for the sake of completeness, 13C NMR were
recorded for allsamples, however the higher concentrations of materials required to obtain a
signal and the numerous low concentration species present, meant that they possesses very
little useful information about the species.
31
2.1.3. Conclusion
This study of industrial samples of TiCl4 suggests strongly that there are a large number of
silicon based impurities within the samples, due to the multitude of peaks in the 0.7 ppm -
0.0 ppm region, which is usually too upfield for alkyl protons without the upfield directing
shielding character of the silane coordinated to the carbon. These peaks were largely absent
in the oldest samples, which comparatively had relatively simple NMR patterns. There are
several possible explanations for this. The silyl compounds with the very downfield shifts
could be intermediates that over a long period will react to form a smaller group of more
energetically favourable species. Another option is that they gradually form insoluble
molecules that are part of the precipitate observed, no longer appearing in the NMR. Finally
they could be caused by the use of different starting material, or a change in the process.
Components such as the distillation calandria foul quickly and must be replaced or cleaned,
so it is possible that the different stages of fouling could affect the contaminants in the TiCl4
samples that are removed.
Whilst the number of shifts present in the NMR makes characterising individual contaminant
species impossible, general structures representing groups of silanes can be inferred. The
silanes must have a boiling point close to that of TiCl4 itself or it would not make it through
distillation. If the species were too high boiling it would be recycled to the sludge tank, too
low and it would exit as waste gas. The lack of splitting in the samples suggests relatively small
alkyls. This information allows for a few general impurity structures to be proposed.
Methylcyclosiloxanes: This group consists of ring systems of alternating silicon and oxygen,
with methyl groups or chlorine bound to the silicon. Hexamethyl4 cyclosiloxane (Figure 19)
has an identical boiling point to TiCl459 and a mass spectroscopy study commissioned by
32
Tioxide found cyclosiloxanes present in samples of the material that fouls the calandria,
however the samples used to generate them were completely hydrolysed so it was not
possible to say for sure that they were present in the TiCl4 during the process. The peak
patterns between 0.3 and 0.5 ppm resemble those of chlorosiloxanes,60 with the peaks with
matching integrals representing different structural isomers of the same ring system. The
singlets may represent the hexamethyl material and the triplets the ethyl equivalents which
would present themselves as peaks in the 1.0- 1.5 region, as seen in the first 20 ml of the
more aged sample.
Figure 19: Structures of four potential siloxane species forming within the TiCl4 samples, all with
boiling points between 100 and 134 ˚C, and all with 1H NMR spectra that would only give highly
upfield peaks.61–64
33
Methylchlorosilanes: A far wider range of compounds, usually involving a single silicon atom
or two silicon atoms bound directly or with a methyl bridge, with methyl/chlorine
substituents. These are the most probable source of the very downfield peaks between 0.0
and 0.3 ppm. Many molecules in this group have boiling points similar to that of TiCl4, such
as 1-Chloro pentamethyldisilane and dichlorosilyltrimethylsilane, both of which boil at
136˚C.65,66
Figure 20: Structures of four potential chlorinated methyl silane species forming within the TiCl4
samples, all with boiling points between 80 and 134 ˚C,65–68 and all with NMR spectra that would
only give highly upfield peaks.
34
In addition to those species, a great number of similar species exist with boiling points lower
than 136 ˚C that would likely be found in the first few ml of distillate. These usually resemble
a chlorinated version of tetramethylsilane such at Chloromethyl(dimethyl)silane and
dichloro(chloromethyl)methylsilane, both shown in Figure 20, which have boiling points of
88 and 121 ˚C respectively.
2.2 Analysis of the Black Precipitate in the Tetra Samples
Each of the samples of TiCl4 was found to precipitate out a black solid over a period of time.
This occurred in both industrial samples and the samples bought from chemical suppliers The
analysis of the black precipitate found in the stored samples of TiCl4 was carried out primarily
in order to discover what it consists of and how it forms. With this information, the aim was
to confirm whether or not its formation was related to the discolouration of TiCl4 samples
and if it could be analogous to the solids that foul the process components.
The black solid in three samples of TiCl4 were isolated. The samples analyzed were from the
primary condensation tank, the distillation step and the third fraction from the lab distillation
(with packing). These samples were chosen due to the high volume of black solid present, as
well as being representative of some of the problem areas of the plant, for instance the
primary tetra would be analogous to what runs through the calandria and the distilled tetra
would be similar in content to that which causes fouling of the furnace in the production
process. The primary and distilled samples were both eighteen months old at the time of
testing and the lab distilled three months old.
35
2.2.1. Experimental
The tetra samples were stored under air in glass jars. The TiCl4 was removed from the jars as
fully as possible, leaving the black solid in the jar. This was done by flushing the jar with
nitrogen before removing the majority of the TiCl4 with a filter canula with a separate flask.
The remaining solid was then washed by adding 20 ml of hexane, forming a suspension with
the precipitate. This was then stirred for one hour whilst heating to 40 ˚C before the hexane
was removed into a separate flask via a filter canula. This was repeated twice more, with all
three of the washings entering into the same flask. Following the washing, each sample was
left under vacuum for 6 hours in order to remove any remaining hexane and TiCl4. When the
drying process was completed, the jars that had held the distilled and primary tetra held
roughly 0.25 g of grey solid and the lab distillation fraction jar held roughly 0.1 g. The samples
were then stored under air in the jars they came in.
The samples were analyzed using 1H NMR spectroscopy was carried out using a 600 Hz Bruker
Avance 600 Cryo III spectrometer, and X-ray Photoelectron spectroscopy (XPS) carried out
using a Thermo ScientificTM K-alphaTM spectrometer, with monochromated Al Kα radiation, a
dual beam charge compensation system and constant pass energy of 50 eV. Survey scans
were collected in the energy range of 0 – 1200 eV.
Mass Spectra were obtained using chemical ionization with a methane reagent gas with the
Finnigan Thermo MAT900xp spectrometer. XPS measurements were carried out on a VG
ESCALAB 220i XL instrument using focused (400 lm spot) monochromatic Al Ka radiation at a
pass energy of 20 eV. Scans were acquired with steps of 100 meV. EDX analysis was carried
out on carbon coated samples on a Philips XL30 ESEM instrument. FT-IR spectra were
obtained using an FT-IR Shimadzu 8700 spectrometer over a range of 400-3500 cm-1.
36
2.2.2.1. NMR spectroscopy
The black solid was largely insoluble in most NMR solvents. Samples were made by NMR
spectra of the black precipitate were obtained by placing a spatula tip full of solid into a small
sample tube and shaking with 5 ml of CDCl3 for 30 minutes. Although some solid remained,
some appeared to have dissolved. The spectra for all three solid species were extremely
similar. The most prominent peak in all three spectra appears at 1.45 ppm. This is similar to
the position of the diminished peak in the filtered TiCl4, further implying that the black solid
consists at least partially of an alkyl species.
2.2.2.2 Toluene Washings
The samples were washed repeatedly using hexane, however initially it was planned that
toluene would be used. In an initial test with an older sample of gas cleaning tetra TiCl4
removal was carried out in the way described above and the material washed three times
with dry toluene. It was observed that the solid became much lighter in colour following the
washings. Conversely the toluene took on a faint purple grey colour. This suggests that the
supposedly black solid impurities are in fact coated in a substance that gives them the
black/grey appearance, with a lighter core. As the samples were stored under air rather than
nitrogen due to the nature of their containers, it is a reasonable assumption that this could
be solid TiO2 with an organic coating.
2.2.2.3. Mass Spectroscopy
37
Mass spectrum were obtained for the three solid samples. Appendix 5 shows the mass
spectra of the plant distilled solid sample. The spectra all share two notable characteristics.
Firstly, the prominent peaks that appear in all three spectra: 391, 338, 279, 167, 149 and 113
m/z can all be attributed to the molecule bis 2-ethylhexyl phthalate (Figure 21).69
This molecule is a common mass spec impurity due to its use as a plasticizer for materials
such as PVC. Plastics can contain between 1 and 40%, bis 2-ethylhexyl phthalate depending
on their purpose.70 The mass spectrum peaks can be assigned to fragments of the molecule.
The only exception to this is the peak at 338 m/z. This peak is most likely attributable to
euracamide,71 a very common slip agent, added to plastics in order to reduce surface friction
and improve slip. It is also a well-known mass spec contaminant. The other important trait
to emerge from the mass spectra of the solid is the pattern of peaks every 14 m/z up to
around 800 m/z. Intervals of 14 m/z are a strong indicator of long alkyl chains, such as
polymers. This is fitting with the broad peak present in the proton NMR, which is similar in
position and shape to that of a polyethylene backbone of a polymeric species. 72
Both of the distilled samples showed this pattern, despite the fact that the distillation process
would remove any high molecular weight polymeric species. This implies that they are
entering the sample at another point in the process.
38
Figure 21: Mass spectrum of a sample of DCM that has been shaken with the black solid. Structure
and simulated spectrum of plasticizer molecule bis 2-ethylhexylphthalate inset.
2.2.2.4. Infrared Spectroscopy
IR spectra were taken of all three solid samples (Figure 22). They were prepared by grinding
a spatula tip of the solid with 0.2 g of KBr and then pressing into a disc using a dye. The
spectra of all three materials were very similar. The most noticeable feature of all three
spectra is the presence of a large broad peak between 3000 cm-1 and 3500 cm-1, which
suggests the presence of water. This confirms that the material in question is hygroscopic.
39
The spectra also has several peaks that could be due to the presence of the plasticizer that
was picked up in the mass spectroscopy experiments, namely the peak at 2932 cm-1 which
represent the stretching vibrations of ethylene C-H bonds. The peak at 1641 cm-1 is in the
correct position for the aromatic C=C bond, however the broadened shape and high relative
intensity suggest that it is not the only bond vibrating at that frequency. This could be due to
the ester group in the phthalate, although this shift appears at a lower wavenumber than
normally expected for a phthalate C=O asymmetric stretching, which usually occurs at just
above 1700 cm-1.
Figure 22: Infrared Spectrum of the black solid, as prepared using a pressed KBr disc.
40
This could be explained by the carbonyl groups co-coordinating to a metal species, which
would remove electron density from the bond, lowering its vibrational frequency. This could
explain how the phthalate is part of the precipitate and not seen in the solution NMR. Further
relevant peaks were found in the fingerprint region. The large peak at 1431cm-1 is
characteristic of the symmetric stretching of the C=O bond in the ester group. As with the
asymmetry stretch it is shifted from its expected position (1400 cm-1) in a way that suggests
that the group is coordinated to a transition metal species. The two broad bands between
1000 and 1300 cm-1 are characteristic of the C-O stretch in an ester group. The spectra also
feature two distinct broad humps at the low frequency end of the spectrum, between 500
and 800 cm-1. These represent the vibrational frequency of the Ti-O bonds in a lattice. The
broad hump at 400 cm-1 could also be representative of TiCl4 however due to the large
amount of interference and the fact that only half of the peak is visible this is not certain.
2.2.2.5. Energy Dispersive X-ray Spectroscopy (EDX)
The elemental composition of the three samples was studied using EDX and is shown in Table
2. The values shown in Table 2 are compositions of area scans of dimensions approximately
30 x 30 µm. The major components of each sample were carbon which makes up 40-55% of
the species present, and oxygen which makes up 25-40% of the material present in the
precipitate samples of primary condensate, plant distilled and lab distilled (3rd fraction)
tetra.
41
Table 2: Atomic percentages of the elements present in the black
solid as determined by EDS
Atomic%
Element Primary Plant Lab Distilled 3rd

Condensate Distilled Fraction
CK 46.16 55.19 39.66
OK 31.99 26.01 40.46
Cl K 11.24 8.85 7.24
Ti K 10.29 9.07 12.1
Each sample also has a significant amount of chlorine and titanium. The chlorine is likely in
the form of residual Ti-Cl bonds exiting within the solid, most likely as partially oxidized
TiOCl2.2(H2O) within the solid. However this cannot be the only source of titanium as the
chlorine would only account for around 10 - 20% of the atomic titanium present. The rest can
be assumed to be in the oxidized form. This implies that some of the oxygen is present in the
form of TiO2. Once again however this does not cover all of the atomic species present. The
remaining titanium in each sample only accounts for 25 -50% of the oxygen. The remaining
oxygen could be present in the form of the bis 2-ethylhexyl phthalate plasticizer picked up in
the mass spectroscopy experiments.
The plasticizer bis 2-ethylhexyl phthalate is 17% oxygen. Assuming that all the carbon comes
from the plasticizer, which it almost certainly does not, this still only accounts for two
42
combined only account for around 35% oxygen content in the older samples and 60% and
the lab distilled. The actual value is likely to be lower as mass spectroscopy suggests the
presence of a polymeric species in the material as well. The remaining oxygen most likely
comes from the water present in the solid as observed by the NMR and IR spectroscopy.
2.2.2.6. X-ray Photoelectron Spectroscopy
X-ray photoelectron spectroscopy was carried out on the solid from the primary condensate
TiCl4. The resulting spectra (Figure 23) showed the expected strong environments for the
carbon and oxygen, measuring the atomic percentages at 46.8% and 28.93% respectively.
This combined total is higher than the EDX, with substantially lower proportion of the oxygen.
The atom% of titanium is lower than that predicted by EDX at only 6.3%. The binding energies
of the Ti 2p3/2 and 2p1/2 peaks at 458.78 eV and 464.61 eV respectively indicate the presence
of Ti4+ exclusively,57, 58 as expected for the TiCl4 and TiO2 present. The atom% of Si is also
lower at 4.13%
The ratio of Ti : Cl is roughly the same in XPS and EDAX, which indicates that they are present
together at the same point in the solid, lending credence to the theory that the chlorine exists
as a small amount of unreacted TiCl4. The reason for the lower overall percentages of the
Titanium and chlorine could be the result of the XPS sampling the surface of the material.
The XPS penetrates less deeply into the material and therefore pick up more of the organic
surface coating. This also explains why the atom% of oxygen is lower as less TiO2 is being
sampled at the expense of the surface organics. The oxygen peak at 531.99 eV is seen to be
a compound of four different oxygen environments which account for the organic oxygen,
the oxygen in the water and the TiO2 present.59, 60
43
Figure 23: Compound peak in the O1s environment form the XPS spectrum of the black solid,
showing organic oxygen (dotted + dashed), H2O (dotted) and TiO2 (dashed).
2.2.3. Conclusion
Analysis of the solid confirms the presence of the bis 2-ethylhexyl phthalate plasticizer
molecule. This almost certainly came from the lid of the jars in which the TiCl4 is stored. The
process is gradual which explains why EDX analysis showed a greater carbon content for the
older samples when compared to the new ones. Infrared spectroscopy showed a peak
pattern that matches that expected from bis 2-ethylhexyl phthalate; however the peak that
is thought to represent the carbonyl group is shifted to a lower frequency than expected.
This can be explained by the carbonyl coordinating to a metal centre, co-ordination chemistry
of phthalates has been shown to cause this carbonyl peak shift.61
44
A schematic shown in Figure 24 shows a simplified, proposed mechanism for the formation
of the black solid. Exposure to the air in the jar (these samples are collected in open air from
valves on the side of the process), causes the reaction of TiCl4 with small amounts of H2O
trapped inside the jar, leading to the formation of the solid TiO2. The corrosion of the lid,
possibly due to the release of the HCl, leads to the leaching of bis 2-ethylhexyl phthalate into
the solution, which then binds to the surface of the TiO2 particles.
Phthalates have been shown in the literature to co-ordinate to the surface of titania62 and
have been used in the surface characterization of TiO2 nanoparticles.63 The mass spectra
confirms the presence of the phthalate, however it is not seen in any of the NMR spectra of
the liquid TiCl4 samples. This could be due to the breakdown of the phthalate in solution by
the TiO2, a common UV catalyzed reaction.
To simulate the conditions for the formation of the black precipitate in different samples of
TiCl4, shavings of one of the sample caps were placed in to TiCl4 (Aldrich 99.9%) under N2 and
left in a cupboard. After one week the shavings had discolored but there was no colour
change or precipitate growth in the TiCl4. The flask was filled with air, sealed and left in
darkness. After one week, a black precipitate was observed in the bottom of the flask. When
this was repeated without the plastic in the sample, a small amount of much lighter solid
formed.
The mass spec also strongly suggests the presence of long chain alkyl species as seen by the
repeating peak pattern every 14 m/z up to 600 m/z. There are a number of explanations for
this occurrence. The corrosion of the lid could also leach polymer chains into the solution;
however the lid is predominantly PVC, which would give a different mass pattern. Alkyls of
45
that molecular weight would not be able to move through either the lab or plant distillation
columns so they cannot be carried over from the oil added in the process.
Figure
24: Proposed stepwise process for the formation of the precipitate, consisting of TiO2 particles with
bis 2-ethylhexyl phthalate molecules coordinating to the surface.
A possible explanation is the formation of a Zeigler Natta type polymerization system in the
liquid TiCl4. Small alkyl molecules from the breakdown of the oil could make it through the
distillation column, or from the UV catalyzed breakdown of the phthalate molecules.
Traditional Ziegler Natta catalysis uses a heterogeneous system such as the TiO2 particles.
TiCl4 at the surface of these particles could act as a source for the polymerization of the
polymeric species in solution. Furthermore, phthalates are widely used as electron donors in
Zeigler Natta catalysis, encouraging polymerization, as well as controlling the nature of the
polymer produced.64 The TiCl4 would be activated by Lewis basic materials, possibly the
silanes and chlorosilanes present, which have been shown to interact with the
alkylaluminium co-catalysts in a way that affect the polymerization of alkyls.
46
Chapter 3
Reaction of Titanium Chloride and Vanadium

Oxytrichloride with Diester and Triester Molecules as
Analogues for Oil
3.1. Introduction
During the Chloride Process, titanium dioxide ore is commonly purified using a fluid
bed reactor which is heated to over 1000 ˚C whilst a stream of concentrated chlorine
gas is passed over the ore. This converts the all solid TiO2 present into gaseous TiCl4.73
The ores used are heavily contaminated with various other metal oxide species. These
contaminants, most notably Fe2O3, V2O5, Al2O3 and SiO2 are chlorinated along with the
TiO2, forming gaseous chlorinated species that rise out of the fluid bed reactor along
with the TiCl4. Purification of the TiCl4 is carried out using a series of temperature
changes and distillations, described in detail in the introductory chapter of this thesis,
before being oxidised back to TiO2.74
Vanadium oxytrichloride (VOCl3) represents the most challenging impurity to remove
(with the exception of SiCl4 which is not removed due to its lower concentration in the
ore and lower boiling point along with the relative inertness and whiteness of SiO2
meaning that its presence is not detrimental to the final product). The p
resence of over 5 ppm of VOCl3 in the TiCl4 distillate can be extremely damaging to the
final product due to the intense yellow colour of vanadium oxide, due to the presence
of Vanadium (V) centres, ruining the white titania pigment. VOCl3 has a similar boiling
47
point to TiCl4, (126 ˚C and 136 ˚C respectively) and is therefore extremely challenging
to separate effectively using a large scale industrial distilling column without somehow
increasing the boiling point of the VOCl3.75 This is traditionally achieved via addition of
mineral oils, the active component of which are triglyceride backbones and
unsaturated carbon chains of lengths C15-18 (Figure 25).76 Although this process has
been used for decades, little is understood about why the oil preferentially reacts with
the VOCl3 rather than the TiCl4.77
In order to improve the understanding of the Chloride Process on a molecular level,
substituted esters were reacted with titanium chloride and vanadium oxytrichloride.
The esters used were diethyl malonate, diethyl succinate, dibenzyl malonate, bis-
isopropyl malonate and glycerol tribenzoate. Each molecule used was a diester,
featuring two carbonyl groups, with the exception of glycerol tribenzoate which
features three carbonyl groups. Glycerol tribenzoate was considered by far the most
similar to the triglyceride backbone of the oil. The diester ligands were used to study
the effects of the length of the hydrocarbon backbone separating the carbonyl groups,
as well as the size of the side groups of the products formed.
Figure 25: Ester species used as analogues for the triglyceride backbone of oil in the reaction
with TiCl4 and VOCl3
48
The simple structures of the diester species were also beneficial in producing single
crystals of the reaction products. Due to TiCl4 having a prominent role in Zeigler Natta
catalysis and the use of diesters as electron donor species,78 several studies have been
carried out on the reactivity of TiCl4 with bidentate carbonyl ligands. Maier et al. found
that co-ordination of 3,3-dimethyl 2,4-pentadione to TiCl4, resulted in a six coordinate
octahedral complex, with both carbonyl oxygen atoms coordinating to the metal
centre without the displacement of the chloride ligands.79
Conversely a similar study found that unsubstituted 2,4-pentadione (acacH) displaces
a chloride atom, giving the dimeric [TiCl2(acac)µ-Cl]2 as a product. The monomeric
equivalent was prepared by carrying out the reaction in tetrahydrofuran (THF) as
opposed to CH2Cl2, with the THF acting as an Lewis base stabilising the monomeric
species.80
Complexes of TiCl4 with diesters have been studied previously using diethyl
malonate.81 Like with acac, diethyl malonate acts as a bidentate ligand with the
titanium coordinating to both carbonyl oxygen atoms. However, unlike acac, all of the
chloride ligands were retained. The crystal structure showed that the ligand molecule
was unchanged and the complexes adopted a distorted octahedral geometry (Figure
26).
Figure 26: Reaction of diethyl malonate and titanium(IV) chloride, observed by both Sobota
et al. and Kakkonen et al.81,82
49
3.2. Experimental
All reactions were carried out under argon obtained from BOC using standard Schlenk
techniques. All Solvents were dried over activated alumina by the Grubbs method
using anhydrous engineering equipment, such that the water concentration was 5 – 10
ppm.83 Anhydrous TiCl4 and malonate starting materials were purchased from Sigma
Aldrich; all were used without further purification. 1H and 13C{1H} NMR spectra were
obtained on a Bruker AV-600 Mz spectrometer, operating at 295 K and 600.13 MHz
(1H). Signals are reported relative to SiMe4 (δ = 0.00 ppm) and the following
abbreviations are used s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet), b
(broad). CDCl3 was obtained from GOSS Scientific and was dried and degassed twice
over activated 3 Å molecular sieves.
3.2.1. Synthesis of tetrachloro(diisopropyl malonate)-titanium(IV) [1a]
Diisopropyl malonate (0.5 cm3, 3.3 mmol) was added dropwise to TiCl4 (1.0 cm3, 9.1
mmol) in 50 cm3 of hexane and stirred under argon for 2 h. A yellow precipitate was
immediately formed but the mixture was stirred for a further 2 h to ensure no further
reaction. The solvent and excess TiCl4 were removed via filtration and the precipitate
was filtered and washed three times with 20 cm 3 hexane before drying in vacuo to
afford complex 1a in good yield (0.8 g, 63%). Some of the product was re-dissolved in
5 cm3 of dichloromethane and layered with 15 cm3 of hexane. After 48 hours small
crystals formed. The molecular structure of 1a was determined by X-ray
crystallography. 1H NMR (CDCl3): δ 5.46 (b, 2H, -CH), 3.95 (b, 2H, -CH2), 1.47 (d, 12H, -
CH3, J = 6.20Hz). 13C{1H} NMR (CDCl3): δ 21.7 (-CH3), 38.0 (-CH2), 54.8 (-OCH2), 172
50
(C=O). Elemental analysis calculated for TiCl4C9O4H16: C, 28.60; H, 4.27. Found: C, 26.03;
H, 3.45.
When 1a was crystallised, two types of crystals were observed, block type and plate
type. Carrying out single crystal XRD on both types revealed that a second product had
formed (1b), a dimeric species consisting of two titanium centres linked by a bridging
oxygen, each titanium centre coordinated to one malonate ligand. The additional
oxygen that entered the system is likely responsible for the discrepancy between
calculated atomic percentage and that found through elemental analysis.
3.2.2. Synthesis of tetrachloro(dibenzyl malonate)-titanium(IV) [2]
Dibenzyl malonate (1.0 cm3, 3.5 mmol) was added dropwise to TiCl4 (1.0 cm3, 9.1 mmol) in 50
cm3 of hexane. A yellow precipitate was observed immediately and the reaction was stirred
under argon for 2 h. The solvent and excess TiCl4 were removed via filtration and the
precipitate was filtered and washed three times with 20 cm 3 hexane before drying in vacuo
to afford 2. A good yield was recovered (1.2 g, 77%). Yellow crystals were again formed from
a solution in CH2Cl2 layered with hexane. The molecular structure of 2 was determined by X-
ray crystallography. 1H NMR (CDCl3): δ 7.3 – 7.5 (m, 10H, Ph), 5.5 (b, 4H, -CH2), 4.0 (b, 2H, -
CH2). 13C {1H} NMR (CDCl3): δ 38.0 (-CH2), 73.0 (-OCH2), 129-130 (m, C6, -Ph). Elemental
analysis calculated for TiCl4C15O4H12: C, 43.08; H, 3.40. Found: C, 43.24; H, 3.40.
3.2.3. Synthesis of tetrachloro(diethyl succinate)-titanium(IV) [3]
Diethyl succinate (0.5 cm3, 3.3 mmol) was added dropwise to TiCl4 (2.0 cm3, 18.2 mmol)
in 50 cm3 of hexane and stirred under argon for 2 h. A yellow precipitate formed
immediately, the reaction was stirred for 2 h to ensure completion. The yellow
precipitate was filtered off and washed three times with 10 cm 3 hexane and dried in
51
vacuo, to afford complex 3 in good yield (0.9 g, 75%). The product was re-dissolved in
5 cm3 dichloromethane and layered with 10 cm 3 hexane. Small crystals formed
overnight allowing the molecular structure of 3 to be determined by X-ray
crystallography. 1H NMR (CDCl3): δ 4.61 (q, 4H, -CH2, J = 14.3 Hz), 3.08 (s, 4H, -CH2),
1.46 (t, 6H, -CH3, J = 14 Hz). 13C {1H} NMR (CDCl3): δ 14.1 (-CH3), 29.2 (-CH2CH2), 66.2 (-
OCH2), 179 (C=O). Elemental analysis calculated for TiCl4C8O4H14: C, 26.4; H, 3.88.
Found: C, 26.41; H, 4.02
3.2.4. Synthesis of complex of titanium(IV) chloride with glycerol tribenzoate [4]
Glycerol tribenzoate (0.5 g, 1.2 mmol) was added gradually to TiCl4 (2.0 cm3, 18.2
mmol) in 50 cm3 of hexane and stirred under argon for 2 h. An excess of TiCl4 was used
as its removal after reaction completion is facile. Gradual precipitation of a yellow
product was observed. The solvent and excess TiCl4 were removed via filtration and
the precipitate was filtered and washed three times with 20 cm 3 hexane before drying
in vacuo to afford 4 in good yield (1.6 g 75%). Crystals suitable for X-ray diffraction
were obtained by dissolving the yellow product in a minimum amount of
dichloromethane, cooling to -10 ᵒC, and leaving for one month. 1H NMR (CDCl3): δ 8.27
(d, 2H, o-CH, J = 7.7 Hz), 8.17 (d, 4H, o-CH, J = 7.7 Hz), 7.69 (t, 2H,-CH2, J = 4.8 Hz, 12.3
Hz). 2H, p-CH J = 7.7 Hz), 7.65 (t, 1H, p-CH, J = 7.7 Hz), 7.53 (t, 2H, m-CH, J = 7.7 Hz),
7.50 (t, 1H, m-CH, J = 7.7 Hz), 5.82 (quin, 1H, -OCH, J = 5.1 Hz), 4.98 (dd, 2H, -CH2, J =
4.8 Hz, 12.3 Hz), 4.75 (dd, 4H, 13C{1H} NMR (CDCl3): δ 63.5 (CH2), 72.2 (CH), 129 (o-
CH), 131 (p-CH), 135 (m-CH), 168 (C=O). Elemental analysis calculated for
Ti4Cl16C48O12H46: C, 36.64; H, 2.95. Found: C, 37.6; H, 2.98.
3.2.5. Synthesis of Dichloro(oxo)(2, 4-Pentanedione) vanadium(V) [5]
52
2,4-pentadione (acacH) (0.5 cm3, 4.87 mmol) was added dropwise to VOCl3 (2 cm3,
21.1 mmol) in 30 cm3 of hexane and stirred under nitrogen for 2 h. A dark precipitate
formed immediately, giving a thick suspension that appeared black. The precipitate
was filtered and washed with hexane and dried in vacuo. This afforded the dark red
complex 5 in good yield (1.1 g, 75%). 1H and 13C NMR were carried out on this product
under nitrogen. Crystals suitable for single crystal X-ray diffraction were grown by
layering a saturated solution of 5 in dichloromethane with hexane. Large green crystals
formed over the course of 2 days and were analysed. 1H NMR (CDCl3): δ 2.39 (s, 6H,
CH3), 6.13 (s, 1H, CH). 13C {1H} NMR (CDCl3): δ 26.4 (CH3), 105 (CH), 193 (C=O).
Elemental analysis calculated for VO3Cl2C5H7: C, 25.34; H, 2.98. Found: C, 25.80; H,
3.02.
3.2.6. Synthesis of Dichloro(oxo)(diethyl malonate) vanadium(V) [6]
Diethyl malonate (0.5 cm3, 3.3 mmol) was added dropwise to VOCl 3 (2 cm3, 21.1 mmol)
in 50 cm3 of hexane and stirred under nitrogen for 2 h. An excess of VOCl 3 was used to
ensure completion as unreacted VOCl3 is facile to remove from the reaction. A very
dark precipitate was formed in a dark red solution. The precipitate was filtered and
washed three times with 20 cm3 hexane and dried in vacuo to afford 6 in good yield
(0.9 g, 78%). 1H and 13C NMR were carried out on this product under nitrogen. Some
of the product was re-dissolved in 5 cm3 dichloromethane and layered with 15 cm 3
hexane. Small crystals formed over approximately a week. 1H NMR (CDCl3): δ 1.34 (t,
6H, -CH3, J = 7.25 Hz, 3.35 (s, 2H, -CH2), 4.18 (q, 4H, -CH2, J = 7.25 Hz). 13C{1H} NMR
(CDCl3): δ 14.2 (CH3), 55.9 (CH2), 63.6 (CH2CH3), 163 (C=O). NMR peaks were highly
53
broadened due to presence of paramagnetic vanadium species. Elemental analysis
calculated for V4Cl8C28O20H48: C, 28.21; H, 4.06. Found: C, 26.38; H, 3.78.
3.2.7. Synthesis of VOCl3 Bis-isopropyl malonate Complex [7]
Bis-isopropyl malonate (0.5 cm3, 2.7 mmol) was added dropwise to VOCl3 (2 cm3, 21.1
mmol) in 30 cm3 of hexane and stirred under nitrogen for 2 h. A dark red precipitate [7] was
formed in a dark solution, the precipitate was filtered and washed 3 times with 20 cm3
hexane. The final washing was coloured red suggesting the product is partially soluble in
hexane. The sample was dried in vacuo. 1H NMR (CDCl3): δ 5.14 (b, 1H, -CH), 3.37 (b, 2H, -
CH(CH3)2 ), 1.30 (b,12H, -CH3) 13C {1H} NMR (CDCl3): δ 21.5 (CH3), 54 (CH2), 73 (CH2), 129 (C =
0).
3.2.8. Synthesis of Dichloro(oxo)(diethyl succinate) vanadium(V) [8]
Di-ethyl succinate (0.5 cm3, 3.3 mmol) was added dropwise to VOCl3 (2 cm3, 21.1 mmol) in
40 cm3 of hexane and stirred under nitrogen for 2 h. A dark red solution formed
immediately but no precipitate was observed. The product was dried in vacuo leading to the
removal of the solvent and VOCl3, leaving a viscous red liquid product (0.8 g, 90%). 1H and
13C NMR were carried out on this product under nitrogen. The flask containing the liquid
product was left on its side for two weeks. Small green crystals were observed in the flask.
1H NMR (CDCl3): δ 1.21 (b, 6H, -CH3), 2.59 (b, 4H, -CH2CH2), 4.12 (b, 4H, -CH2CH3). 13C {1H}
NMR (CDCl3): δ 14.3 (CH3), 29.2 (CH2CH2), 60.9 (CH2CH3), 173 (C = 0).
3.2.9. Synthesis Of Glycerol tribenzoate VOCl3 [9]
2 cm3 of VOCl3 (21.1 mmol) in 10 cm3 of toluene, was added dropwise to a rapidly stirred
suspension of glycerol tribenzoate (0.5 g, 1.2 mmol) in toluene (30 cm3 )using a canula. The
mixture was refluxed under nitrogen for 15 h. A red-brown solution was removed by filtration
to yield and a brown precipitate. This precipitate was washed and dried in vacuo and 1H and
54
13C NMR were obtained. The compound was dissolved in CH2Cl2 and layered with hexane in
an attempt to grow crystals, however none formed after multiple attempts. NMR (CDCl 3): δ
8.05 (b, 9H, (o-CH)), 7.5 (b, 16H, p-CH and m-CH) 5.7 (b, 1H, -OCH), 4.8 (b, 4H, -CH2), 4.12
(broad, 2H, (O-CH3). 13C {1H} NMR (CDCl3): 63.30 (CH2), 71.32 (CH), 128 (o-CH), 130 (p-CH), 134
(m-CH), 166 (C=O). Elemental analysis calculated for VCl2C8O5H14: C, 30.79; H, 4.52. Found: C,
29.83; H, 4.49.
It should be noted that although elemental analysis results for carbon and hydrogen
were in come cases, outside of the 0.4% tolerance, this was expected due to the
exteremly air sensitive nature of these compounds,
3.2.10. Crystallography
Suitable crystals were selected and mounted on a nylon loop, datasets of 1a and 1b
were collected on a Rigaku AFC12 goniometer equipped with an enhanced sensitivity
(HG) Saturn724+ detector mounted at the window of an FR-E+ SuperBright
molybdenum rotating anode generator (λ1 = 0.71073 Å) with VHF Varimax optics (100
µm focus). Crystals of compound 1b were of poor quality and did not diffract well,
leading to a low-quality dataset. Although the atomic connectivity is unambiguous, the
high R factors (R1 = 0.1153, wR2 = 0.3054) and low precision on C–C bond lengths
preclude accurate comparison of bond lengths and angles. The diffraction pattern for
3 was obtained using a Rigaku R-Axis Spider diffractmeter including curved Fujifilm
image plate and a graphite monochromated sealed tube Mo generator (λ1 = 0.71073
Å). The datasets of 2, 4, 5, 6, and 8 were collected on a SuperNova, Dual, Cu,
Atlas diffractometer. The crystal was kept at 150 K during data collection. 1a, 1b, and
3 were solved using cell determination, data collection, data reduction, cell refinement
55
and absorption correction were carried out using CrystalClear-SM Expert 3.1 b18.84
Structure solution and refinement were carried out using WinGX and software
packages.85 H atoms attached to C atoms were placed in geometrically assigned
positions, with C-H distances of 0.95 Å (CH), 0.98 Å (CH3) or 0.99 Å (CH2) and refined
using a riding model, with Uiso(H) = 1.2Ueq(C) (CH, CH2) or 1.5Ueq(C) (CH3). enCIFer was
used to sprepare CIFs for publication.86 Compounds 2, 4, 5, 6, and 8 were solved using
Olex2,87 the structure was solved with the olex2.solve88 structure solution program
using Charge Flipping and refined with the ShelXL 89 refinement package using Least
Squares minimisation. The crystal structures were solved by Dr. David Pugh and Dr.
Peter Marchand in the case of 1a, 1b and 3 and Dr. Caroline Knapp in the case of 2, 4,
5, 6, and 8.
3.3. Results and Discussion
The diesters bis-isopropyl malonate, dibenzyl malonate, diethyl succinate and the
triester glycerol benzoate were all added to solutions of TiCl4 in hexane. In each case a
bright yellow precipitate was observed to form immediately. After the solution was
stirred for two hours the precipitate was separated via filtration and washed with
moisture free hexane three times. All of the reactions gave yields above 60%. The
products were analysed using NMR spectroscopy and elemental analysis. Single
crystals for X-ray diffraction analysis were obtained by layering saturated solutions of
products in CH2Cl2 with hexane in the case of 1-3 and cooling a saturated solution in
CH2Cl2 in the case of 4.
All four complexes adopt a pseudo octahedral geometry with two carbonyl oxygen
atoms coordinating to the titanium centre without the displacement of the chlorine
56
atoms. The octahedral geometry of the complex is distorted, largely due to the
chelating nature of the ligands, which forces the O(1)-Ti(1)-O(2) bond to narrow to
80.23(4) ° as the carbon carbon bonds are generally longer than the titanium oxygen
bonds. Although chelation is the dominant force in the narrowing of the angle, the
effect is exacerbated due to the electronegative nature and greater atomic radii of the
chlorine, compared with the carbonyl oxygen. Chlorine has a Van der Waals radius of
1.75 Å whereas the Van der Waals radius of oxygen is 1.52 Å,90 leading to the bond
angles between the chlorine ligands to be wider than those between the oxygens.
Forcing the less electronegative carbonyl oxygen atoms closer together.
3.3.1. Interaction of TiCl4 with Diester Ligands
In addition to the triester species, the interactions of the dieter molecules (diisopropyl
malonate, dibenzyl malonate and diethyl succinate) with TiCl4 were also studied. Whilst
less indicative of the active species in the oils used for vanadium removal, these
molecules still provide information regarding the potential reactivity of TiCl4 with ester
functionality. These species were advantageous in that their simplicity allowed for
single crystals to be obtained easily. The resulting NMR spectra were far less complex
and easy to rationalize. As only two ester groups are present the species behave in the
same way as the glycerol tribenzoate species without the third carbonyl species
facilitating dimerisation. This may be of relevance as it is likely that the long alkyl chains
of the oil could sterically hinder dimerisation observed in 4, making it less favourable.
57
Scheme 1: Reaction of TiCl4 with bis-isopropyl malonate to give tetrachloro(bis-
isopropylmalonate)-titanium(IV) [1a]
Figure 27: 1H NMR spectra of bis-isopropyl malonate and the product of its reaction with
TiCl4 [1a], all peaks exhibiting a downfield shift indicating coordination to the titanium metal
centre
58
Figure 28: ORTEP representation of the crystal structure of tetrachloro(diisopropyl
malonate)-titanium(IV) [1a], with thermal ellipsoids at the 50% probability level. Hydrogen
omitted for clarity.
Table 3: Crystallographic Data and Selected bond lengths angles for the crystal structure of 1a the product of
reaction TiCl4 and bis isopropyl malonate
Structural Data Selected Bond Angles ° Selected Bond Lengths Å

crystal system Monoclinic C(4)-O(2)-Ti(1) 129.89(9) Cl(3)-Ti(1) 2.2312(6)
space group P21/c C(6)-O(1)-Ti(1) 131.45(9) Cl(4)-Ti(1) 2.2519(5)
A 6.6655(14) Å O(1)-Ti(1)-O(2) 80.23(4) O(2)-Ti(1) 2.1032(10)
B 12.954(3) Å O(1)-Ti(1)-Cl(3) 170.85(3) O(1)-Ti(1) 2.1120(10)
C 18.983(5) Å O(1)-Ti(1)-Cl(4) 84.68(3) C(4)-O(2) 1.2340(15)
Α 90.000° Cl(1)-Ti(1)-Cl(4) 97.21(2) Cl(1)-Ti(1) 2.2322(5)
Β 90.032(4)° O(1)-Ti(1)-Cl(1) 90.40(3) Cl(2)-Ti(1) 2.3414(5)
Γ 90.000° Cl(1)-Ti(1)-Cl(3) 98.44(2) C(6)-O(1) 1.2294(16)
R1 0.023 O(1)-Ti(1)-Cl(2) 83.26(3) C(4)-C(5) 1.506(2)
wR2 0.0569 Cl(3)-Ti(1)-Cl(2) 93.86(2) C(7)-C(8) 1.4996(18)
3.3.1.1. Tetrachloro(diisopropyl malonate)-titanium(IV) [1a]
Diisopropyl malonate was reacted with an excess of titanium(IV) chloride (Scheme 1)
at room temperature, yeilding yellow crystals after layering in CH2Cl2 with hexane.
NMR spectroscopy indicates a simple 1:1 coordination complex of the ligand and TiCl4
(Figure 27). The 1H NMR signal of the two protons of the central carbon atom between
the two ester groups has shifted by 0.66 ppm downfield in comparison to the free
ligand. Likewise the signal of the isopropyl CH group had shifted by 0.42 ppm and the
signal of the four isopropyl methyl groups have shifted by 0.25 ppm when compared
to the uncoordinated ligand. The NMR confirms there is no loss of a proton from the
central carbon, as is customary with the similar ligand acac, implying that the ester
groups are able to mitigate the formation of a charged complex, instead the product
59
remains covalent. This is confirmed by the crystal structure which shows no shortening
of the C(4)-C(5) bond compared with other C-C bonds (table 3).
The structure of 1a was confirmed by single crystal XRD (Figure 28). The malonate
ligand forms a six membered ring structure with the titanium centre, datively bound
to the two carbonyls, linked in turn by a carbon chain. The C(4)-O(2)-Ti(1) angle of the
six membered ring species is considerably wider (129.89(9)°) than that expected for a
flat hexagon (120°).
Comparatively the O-Ti-O bond angle observed for this complex is 80.23(4)°, narrower
than 90° due to the electronegativity of the chlorine atoms forcing the carbonyls closer
together. This, combined with the differing length of the Ti-O bonds (~ 2.1Å) and C-O
bonds (~ 1.2Å) leads to considerable ring puckering. Sobota et al. observed similar
angles when carrying out the related reaction with TiCl4 and diethyl malonate.16 The
length of the bond between the titanium and the carbonyl oxygen on the diisopropyl
malonate is 2.123(3) Å. This is far closer to the dative Ti-O bond length of [TiCl4(THF)2]
91,92 than that of a covalent Ti-O bond27 and is due to the coordination of the bis-
isopropyl malonate being dative, via a lone pair on each the carbonyl oxygen atoms.
3.3.1.2. Bis-trichloro-µ-oxo-(diisopropyl malonate)-titanium(IV) [1b]
This dimeric species was produced during the same synthesis as 1a described above. The
reaction produced yellow crystals, after workup. Two noticeably different types of crystal
were observed, both of which were characterized by single crystal X-ray diffraction. While all
of the other compounds described in this thesis form as a result of simple addition of the
ligand to the tetrahedral TiCl4 with no displacement of chlorine, 1b contains titanium centres
linked by an oxygen atom, facilitated by loss of chlorine (Figure 29).
60
This was likely a result of a small amount of air entering the vessel, which lead to reaction of
the product with water in which the addition of oxygen was facilitated by the formation of
HCl gas. This effect has been reported with similar titanium chloride complexes, leading to
oxo bridged dimers and tetramers (Scheme 2). 93 The bond distances between the bridging
oxygen and the titanium centres was 1.741(7) and 1.863(7) Å, showing the bonding to be
covalent in nature.28
Scheme 2: Reaction of 1a with atmospheric water to give bis-trichloro-µ-oxo-(diisopropyl

malonate)-titanium(IV) [1b], facilitated by the elimination of HCl.
The bridging oxygen stabilised the system and similar bridging compounds have been
reported before as easier to crystallise. 94 This is likely a result of the system forming
multiple π-bonding interactions between the filled p orbitals of the bridging oxygen
and the empty d orbitals of the two titanium centres, interactions which are stronger
than two Ti-Cl bonds that are replaced. The species is likely the result of there being a
limited oxygen supply, with further oxidation occurring over time should the flask have
been opened fully.
61
Figure 29. ORTEP representation of 1b with thermal ellipsoids at the 50% probability level. Hydrogen
atoms omitted for clarity
Table 4: Crystallographic Data and Selected bond lengths angles for the crystal structure of 1b the
product of reaction TiCl4 and bis isopropyl malonate and subsequent air exposure
Selected Bond
Structural Data Selected Bond Angles ° Lengths Å
crystal system triclinic C(4)-O(2)-Ti(1) 133.0(7) Cl(3)-Ti(1) 2.230(4)
space group P21/c C(6)-O(1)-Ti(1) 131.2(8) O(2)-Ti(1) 2.136(8)
a 18.335(16) Å O(1)-Ti(1)-O(2) 80.0(3) O(1)-Ti(1) 2.129(8)
b 12.436(12) Å O(1)-Ti(1)-Cl(3) 167.3(3) C(4)-O(2) 1.225(14)
c 13.683(11) Å O(1)-Ti(1)-Cl(1) 92.3(2) Cl(1)-Ti(1) 2.222(4)
α 90.000° Cl(1)-Ti(1)-Cl(3) 100.31(14) Cl(2)-Ti(1) 2.306(4)
β 102.474(14) ° O(1)-Ti(1)-Cl(2) 81.7(2) C(6)-O(1) 1.233(14)
γ 90.000° Cl(3)-Ti(1)-Cl(2) 96.22(16) Ti(2)-Cl(5) 2.373(3)
R1 0.1153 O(6)-Ti(2)-Cl(6) 168.9(2) O(3)-Ti(2) 1.741(7)
wR2 0.3054 Ti(2)-O(1)-Ti(1) 177.2(5) O(3)-Ti(1) 1.863(7)
The bond angles and distances in the two bridged titanium centres of 1b are similar to
those reported previously. The bridging Ti-O-Ti bond angle being 177.3(5)°, which is
similar to other oxo-bridged titanium chloride dimers with alkoxide ligands that are
less sterically hindering.94,95 In the presence of bulkier ligands, more acute oxo bridge
62
angles have been reported.93 The formation of the dimer demonstrates how
air/moisture sensitive this system is and how oxygen can stabilise the product.
3.3.1.3. Tetrachloro(dibenzyl malonate)-titanium(IV) [2]
Dibenzyl malonate was reacted with an excess of titanium(IV) chloride at room
temperature (Scheme 3), which yielded yellow crystals after layering in
dichloromethane with hexane. NMR spectroscopy revealed that the ligand
coordinated to the metal centre, in a manner similar to compound 1a. The 1H NMR
spectra of the product shows the signal of the two central protons to shift 0.50 ppm
downfield from the signal of the free ligand. Signals from the two CH2 groups adjacent
to the ester oxygen shifted by 0.32 ppm (Figure 30).
Finally the benzyl proton signal, a single broad peak in the spectrum of the
uncoordinated ligand is shifted to reveal two of the overlapping phenyl group proton
signals. The ortho protons shifted downfield by 0.9 ppm, the meta and para by only 0.5
ppm. As described for the diisopropyl malonate, the complex features a puckered six
membered ring (Figure 31).
Like compound 1a, the chelating nature of the ligand as well as the repulsion of the
four chlorine ligands resulted in the subsequent compression of the O-Ti-O bond angle
to 79.33(19)°. As with the diisopropyl malonate complex, the Ti-O bond length of
2.122(5) Å suggests that the bond between the titanium and the carbonyl is dative in
nature (Figure 31).
63
dibenzyl malonate
Scheme 3: reaction of TiCl4 with Glycerol tribenzoate give tetrachloro(dibenzyl malonate)-

titanium(IV) [2].
Figure 30: 1H NMR spectra of dibenzyl malonate and the product of its reaction with TiCl4 [2],
all peaks exhibiting a downfield shift indicating coordination to the titanium metal centre.
64
Figure 31 : ORTEP representation of tetrachloro (dibenzyl malonate)-titanium(IV) [2] with
thermal ellipsoids at the 50% probability level. Hydrogen atoms omitted for clarity.
Table 5: Crystallographic Data and Selected bond lengths angles for the crystal structure of 2 the product of
reaction TiCl4 dibenzyl malonate

crystal system monoclinic C(4)-O(2)-Ti(1) 134.6(5) Cl(3)-Ti(1) 2.224(4)
space group P21 C(6)-O(1)-Ti(1) 134.4(4) Cl(4)-Ti(1) 2.303(2)
a 16.3215(4) Å O(1)-Ti(1)-O(2) 79.33(19) O(2)-Ti(1) 2.115(5)
b 9.5645(2) Å O(1)-Ti(1)-Cl(3) 169.39(16) O(1)-Ti(1) 2.122(5)
c 19.7379(4) Å O(1)-Ti(1)-Cl(4) 82.63(16) C(4)-O(2) 1.221(8)
α 90.000° Cl(1)-Ti(1)-Cl(4) 95.47(8) Cl(1)-Ti(1) 2.213(2)
β 90.069(2) ° O(1)-Ti(1)-Cl(1) 90.59(15) Cl(2)-Ti(1) 2.292(2)
γ 90.000° Cl(1)-Ti(1)-Cl(3) 99.96(9) C(6)-O(1) 1.224(8)
R1 0.0849 O(1)-Ti(1)-Cl(2) 84.78(16) C(4)-C(5) 1.489(12)
wR2 0.2263 Cl(3)-Ti(1)-Cl(2) 95.28(8) C(3)-C(2) 1.482(13)
65
3.3.1.4. Tetrachloro(diethyl succinate) titanium(IV) [3]
Diethyl succinate was reacted with an excess of titanium(IV) chloride in hexane at
room temperature, which yielded yellow crystals after dichloromethane/hexane
layering. 1H NMR once again revealed that the ligand had coordinated to the metal
centre unchanged, forming a simple 1:1 adduct (Figure 32). This was confirmed by
single crystal X-ray diffraction.
Like other complexes synthesised in this work, this structure contains a ring system
formed by the bidentate coordination of the ligand to the metal centre. In this case
however the ring is seven membered. As with the compounds 1a and 2 the ring is
puckered, taking on the boat chair conformer, and distorted by the presence of the
chloride ligands to give an acute Ti-O-Ti bond angle of 86.26(6)° (Figure 33, Table 6).
This is wider than that of the malonate complex 80.23(4)° , due to the extra C-C bond
making a less restrictive chelating complex.
This is to my knowledge the first time that a succinate ligand has been coordinated to
titanium chloride. Due to the increased rigidity and shorter C-C bond length resulting
from the double bond, the malonate structure is more strained, with a Ti-O-Ti bond
angle of 81.2°.
As with the structure reported here for compound 2a, due to the dative nature of the
coordination between the malonate ligand and the Ti centre the metal oxygen distance
is longer (2.1101(10) Å) than for other compounds of this type.
66
diethyl succinate
Scheme 4: reaction of TiCl4 with diethyl succinate to give tetrachloro(diethyl succinate)-

titanium(IV) [3].
Figure 32: 1H NMR spectra of diethyl succinate and the product of its reaction with TiCl4 [3],
67
Figure 33: ORTEP representation of the crystal structure of tetrachloro (diethyl succinate)-
titanium(IV) [3] with thermal ellipsoids at the 50% probability level. Hydrogen atoms omitted
for clarity.
Table 6: Crystallographic Data and Selected bond lengths angles for the crystal structure of 3 the product of
reaction TiCl4 diethyl succinate

crystal system monoclinic C(4)-O(2)-Ti(1) 136.55(10) Cl(3)-Ti(1) 2.2379(4)
space group C 2/c C(6)-O(1)-Ti(1) 136.55(10) Cl(4)-Ti(1) 2.3070(4)
a 16.6592(11) Å O(1)-Ti(1)-O(2) 86.26(6) O(2)-Ti(1) 2.123(2)
b 8.5418(5) Å O(1)-Ti(1)-Cl(3) 173.87(3) O(1)-Ti(1) 2.1101(10)
c 10.3708(7) Å O(1)-Ti(1)-Cl(4) 83.57(3) C(4)-O(2) 1.2408(18)
α 90.000(5) ° Cl(1)-Ti(1)-Cl(4) 95.985(15) Cl(1)-Ti(1) 2.2379(4)
β 98.836(3)° O(1)-Ti(1)-Cl(1) 88.39(3) Cl(2)-Ti(1) 2.3070(4)
γ 90.000(5)° Cl(1)-Ti(1)-Cl(3) 97.10(2) C(6)-O(1) 1.2408(18)
R1 0.0244 O(1)-Ti(1)-Cl(2) 86.39(3)
wR2 0.0669 Cl(3)-Ti(1)-Cl(2) 95.985(15)
68
3.3.2. Interaction of TiCl4 with Triglyceride Analogue Glycerol Tribenzoate [4]
By using diesters and triesters as model compounds for the interaction of oils with
TiCl4, their ability to bind other metal chlorides, while leaving TiCl4 can be better
understood. By knowing the portion of the oil molecule responsible for the removal
and how it happens, the most effective oil or combination of oils can be found, leading
to improvements in the efficiency of the Chloride Process.
Glycerol tribenzoate is a triester, meaning that of all the ligands in this study, it is the
most representative of the glycerol group of the oil used for purification of TiCl4 in TiO2
production. Rather than unsaturated hydrocarbon chains however, in this case each
ester group is linked to a benzene ring. This simpler molecule serves as a close
analogue to the structure and reactivity of the glyceride group of the oils used in the
Chloride Process, however the absence of long alkyl chains makes the characterisation
of the product, as well as the growth of single crystals, more facile. Glycerol tribenzoate
was reacted with excess TiCl4 under Schlenk conditions in hexane (Scheme 5). An
excess of TiCl4 was used as the product’s lack of solubility in hexane means that the
excess reagent could easily be removed via filtration. This is in fitting with the Chloride
Process, where the oil is added stoiciometrically in far lower amounts. After two hours
a yellow product was observed, which was washed with hexane and dried in vacuo.
Crystals suitable for single X-ray crystallography were grown by making a concentrated
solution of 4 in dichloromethane and cooling to -1 °C for 1 month.
The 1H NMR spectrum for unreacted glycerol tribenzoate shows two sets of doublet of
doublets around 4.7 ppm. These represent the inequivalent CH2 environments situated on
69
the ends of the molecule’s central propyl carbon chain, which couple to each other as well as
the proton on the central carbon. Upon reaction of TiCl4 with the glycerol.
Scheme 5: Schematic of the synthesis of compound 4, formed by the reaction of TiCl4 in hexane with
glycerol tribenzoate.
Figure 34: 1H NMR spectra of glycerol benzoate and the product of its reaction with TiCl4 [4],
70
Figure 35: ORTEP representation of 4 as determined by single crystal X-ray diffraction, with
thermal ellipsoids at the 50% probability level. Hydrogen atoms omitted for clarity.
Table 7: Crystallographic Data and Selected bond lengths angles for the crystal structure of 4 the
product of reaction TiCl4 and glycerol tribenzoate.
Selected Bond
crystal system Triclinic C(4)-O(2)-Ti(1) 164.3(2) Cl(3)-Ti(1) 2.2258(9)
space group P-1 C(6)-O(1)-Ti(1) 170.7(2) Cl(4)-Ti(1) 2.2756(9)
a 9.9955(3) Å O(1)-Ti(1)-O(2) 83.59(8) O(2)-Ti(1) 2.124(2)
b 13.9772(3) Å O(1)-Ti(1)-Cl(3) 177.38(7) O(1)-Ti(1) 2.101(2)
c 14.4644(4) Å O(1)-Ti(1)-Cl(4) 87.98(6) C(4)-O(2) 1.228(3)
α 71.868(2)° Cl(1)-Ti(1)-Cl(4) 94.63(3) Cl(1)-Ti(1) 2.2234(8)
β 76.307(2)° O(1)-Ti(1)-Cl(1) 93.79(6) Cl(2)-Ti(1) 2.2938(9)
γ 77.073(2)° Cl(1)-Ti(1)-Cl(3) 98.32(4) C(6)-O(1) 1.227(3)
R1 0.0381 O(1)-Ti(1)-Cl(2) 84.49(7) Ti(2)-Cl(5) 2.5002(8)
wR2 0.0977 Cl(3)-Ti(1)-Cl(2) 95.38(3) Ti(2)-Cl(7) 2.2328(9)
O(6)-Ti(2)-Cl(7) 171.73(7) Ti(2)-O(6) 2.020(2)
O(6)-Ti(2)-Cl(6) 90.13(7)
O(6)-Ti(2)-Cl(5i) 84.62(7)
Cl(5)-Ti(2)-Cl(8) 167.70(4)
Cl(5i)-Ti(2)-Cl(8) 92.11(3)
Cl(5i)-Ti(2)-Cl(5) 78.41(3)
71
tribenzoate one of these peaks was shifted upfield by 0.2 ppm whereas the other only
shifts by 0.1 ppm, in comparison to the free ligand.
The 13C {1H} NMR shows a shift for the signal of the carbonyl, upfield from 174 to 168
ppm compared to the free ligand, which is due to the formation of a dative bond to
the titanium centre reducing the shielding around the carbon (Figure 34).
The crystal structure of 4 shows two different environments for the TiCl4 bound to the
glycerol tribenzoate. In one of these environments titanium is bound to four chlorine
atoms and also to two carboxyl oxygen atoms of the same glycerol tribenzoate (Figure
35). The bond distances and angles observed around this titanium centre which adopts
a distorted octahedral coordination geometry, are similar to those observed for the
diester complexes 1-3.
The third carbonyl oxygen of the glycerol benzoate molecule, rather than coordinating
to the same titanium centre, which would be sterically unfavourable, instead
coordinates to a second TiCl4 molecule. This leads to an electron deficient, 5 coordinate
TiCl4 centre. This species would be highly unstable as a monomer and thus dimerises.
A two chloro-bridged dimeric species is formed, with the two central titanium atoms
taking on edge sharing octahedral geometry. The bridging Ti-Cl bond distance of
2.5002(8) Å is longer than the terminal Ti-Cl bond distances observed to be on average
2.244(16) Å (Table 7). This is unsurprising as the electron density of the bridging
chloride is shared between two metal centres rather than just one in the case of the
terminal chlorine.
Furthermore among the terminal Ti-Cl bonds, it is observed that those trans to the
ester carbonyl are notably shorter than the equivalent cis-chlorines. This is a result of
72
the carbonyl oxygen being less electron withdrawing than the chlorine and thereby
increasing the electron density around the chlorine opposite. For example Ti(1)-Cl(1)
(trans) and Ti(1)-Cl(2) (cis) are 2.2234(8) and 2.2938(9) Ǻ respectively, a difference of
0.0704 Ǻ. This is also the case for products 1-3, all of which show a shortening of the
Ti-Cl bond trans to the ester carbonyl of 0.7–1.0 ppm, due to the trans influence of the
coordinating carbonyl group.
The formation of chloro-bridged bimetallic titanium species is relatively commonplace.
Wu et al. reacted 2-propanol and TiCl4 to produce a Ti2Cl8 dimer.96 What is of note in
the case of this species is that despite the complexity of the molecule formed, there is
no elimination of any chloride species as a result of the reaction. Only addition
followed by dimerisation of the resulting bimetallic species was observed. This is of
interest in relation to the Chloride Process as the TiCl4 is still intact within the molecule,
and therefore at the high temperatures used in industrial synthesis, this coordination
would be reversible.
3.3.3. Interaction of Glycerol Tribenzoate and Diesters With VOCl3
Excess vanadium(V) oxytrichloride was reacted with the bidentate dicarbonyl ligands;
2,4-pentadione, diethyl malonate, bis-isopropyl malonate, and diethyl succinate and
glycerol tribenzoate to yield compounds 5, 6, 7, 8 and 9 in anhydrous hexane under an
atmosphere of nitrogen. Repeated attempts to grow single crystals of the resulting
compounds were made, of which only 5, 6 and 7 were successful. These crystals
revealed that the reactions had yielded a monomeric species, an oxo-bridged tetramer
and a one dimensional coordination polymer, respectively.
73
In each case the loss of chlorine from the VOCl3 was observed. This is significant as
having lost a chlorine atom from the metal centre, dissociation of the ligand would not
be enough to restore the staring material. This makes the reaction less reversible than
those observed with titanium in complexes 1- 4, when dissociation of the ligand would
leave TiCl4 completely intact.
In the production of TiO2 via the Chloride Process, the addition of oil is carried out at
around 200 °C, and hence all of the metal chlorides present are in the vapour phase.
At this temperature, the dative coordination observed for titanium would likely be
more reversible as the extreme conditions would make the ligand more labile. The
reactions with VOCl3, explain the preference with which the glyceride-containing oils
are able to remove VOCl3 from a gaseous TiCl4 stream.
3.3.3.1. Dichloro(oxo) (2,4-Pentanedione) Vanadium(V) [VOCl2(acac)] [5]
Excess VOCl3 was reacted with 2,4-pentadione (acacH) and stirred for 2 hours under a
nitrogen environment. A dark precipitate in a dark red solution formed immediately
but the mixture was stirred for 2 hours to ensure full reaction. The dark purple product,
[VOCl2(acac)] (5), was dissolved in dichloromethane and layered with hexane. After a
week, small black crystals were observed which were suitable for single crystal X-ray
crystallography. The crystal structure for compound 5 is shown in Figure 37 along with
selected bond lengths and angles in Table 5.
1H NMR of 5 showed a peak at 2.39 ppm corresponding to the methyl groups on the
acac ligand, which was upfield to the same peak, observed at 2.2 ppm, in
uncoordinated acacH (Figure 36). Two further peaks were observed at 3.5 ppm and 5.4
74
ppm in uncoordinated acacH, which represent the H atoms bonded to the central
bridging carbon. These normally equivalent protons are responsible for two different
chemical shifts because one proton can freely dissociate, giving a conjugated system.
When two protons are bound to the carbon atom the chemical shift is 3.5 ppm,
however upon disassociation of one of the protons, shielding around the remaining
proton increases, leading to a downfield shift to 5.4 ppm. This is a very useful tool in
determining coordination. In the 1H NMR spectrum of 5 only one further peak (in
addition to the peak at 2.39 ppm) was observed at 6.13 ppm. The integral of this peak
has a 6:1 ratio with the peak corresponding to the two methyl groups at 2.39 ppm. This
provides clear evidence that one proton from the acac ligand has been lost on
coordination to the vanadium centre with a concordant upfield chemical shift of 0.7
ppm. This suggests that the reaction of VOCl 3 with acacH proceeds via loss of HCl, as
shown in Scheme 6.
The structure of compound 5 shows that the acac ligand is bound to the vanadium
centre via the carbonyl oxygen atoms (Figure 37). The V-O bond lengths are 1.918(3)Å
for O(2) and 1.904(3) Å for O(3). These are shorter than typical dative V-O bond
distances of ~2.1 Å, as described for 6 and 7 (vide infra) and is the result of the
increased electron density given from the carbonyl oxygen. A proton from C(3) was
lost in the reaction and a delocalised system is therefore created between O(3), C(2),
C(3), C(4) and O(2) (Table 8).
75
Scheme 6: Schematic of the synthesis of compound 5, formed by the reaction of VOCl3 in hexane
with diethyl malonate
Figure 36: 1H NMR spectra of 2,4 pentadione and the product of its reaction with VOCl 3 [5],
The signals relating to the non-conjugated variant of acac are completely removed in the
spectrum of 5, indicating that coordinating to the VOCl 3 is responsible for loss of a proton.
Both peaks exhibiting a downfield shift indicating coordination to the vanadium metal centre.
76
Figure 37: ORTEP representation of crystal structure of [VOCl 2(acac)] [5] with thermal
ellipsoids at the 50% probability level. Hydrogen omitted for clarity.
Table 8: Crystallographic Data and Selected bond lengths angles for the crystal structure of 5 the
product of reaction VOCl3 and 2,4 pentadione.
Selected Bond
crystal
system monoclinic Cl2 - V1 - Cl1 93.45(3) V1 - Cl1 2.2880(8)
space group P21/c O1 - V1 - Cl1 99.58(9) V1 - Cl2 2.2275(9)
a 15.3393(8) Å O1 - V1 - Cl2 99.94(9) V1 - O1 1.569(2)
b 9.7939(5) Å O1 - V1 - O2 100.11(10) V1 - O2 1.903(2)
c 8.6842(5) Å O1 - V1 - O3 101.11(10) V1 - O3 1.918(2)
α 90 O2 - V1 - Cl2 159.64(7) O2 - C2 1.290(4)
β 94.664(5) O2 - V1 - O3 84.62(9) O3 - C4 1.283(4)
γ 73.746(5) O2 - V1 - Cl1 86.68(8) C2 - C3 1.398(4)
R Factor 3.21% O3 - V1 - Cl2 88.12(7) C3 - C4 1.398(4)
This results in the ligand having an overall negative charge, which is stabilised by the
reaction with the VOCl3 resulting in a stronger bond to the vanadium than seen in
complexes of VOCl3 with the diesters, diethyl malonate and diethyl succinate (vide
infra). This is evidenced in the crystal structure of 5 in which the C2-C3 and C3-C4 bonds
between the carbonyl groups appear shortened (1.398(5) Å and 1.382(5) Å
77
respectively).22 This shows that the central carbon has been deprotonated leading to
a conjugated 6 membered ring system as is frequently seen with coordination
compounds of acac.23
Compound 5 is five coordinate, which is common for vanadium species, particularly
when a V=O bond is present. This is also seen for [VO(acac) 2], a five coordinate species
in which the 4 carbonyl oxygen atoms take an almost square planar geometry with the
double bond perpendicular.24 In the case of compound 5, the geometry is similar, with
bond angles between the oxygen atoms in the acac O(2)–V(1)-O(3) at 84.59(12)°,
whereas the angle between the chlorine atoms and the equivalent oxygen atoms on
the acac are wider, at 86.68(8)° and 88.12(8)°, respectively for O(2)-V(1)-Cl(1) and O(3)-
V(1)-Cl(2). The angle between the two chlorines (Cl(1)-V(1)-Cl(2)) is widest at 93.47(4)°.
This is due to the increased repulsion between the chlorine atoms due to their greater
electron density.
3.3.3.2. Dichloro(oxo) (diethyl malonate) Vanadium(V) [{VOCl2(CH2(COOEt2)}4] [6]
Diethyl malonate was reacted with an excess of VOCl 3 at room temperature, which
yielded dark red crystals of [{VOCl2(CH2(COOEt2)}4] [6] after layering a concentrated
solution of 6 in dichloromethane with hexane (Scheme 7).The 1H NMR spectra of
compound 6 shows that peaks corresponding to the coordinated ligand have shifted
downfield compared with that of the unreacted malonate starting material (Figure 38).
The protons on the central carbon showed the greatest shift, implying they are most
heavily deshielded by the presence of the vanadium, with the signal shifting downfield
to 3.5 ppm from 3.2 ppm for the unreacted ligand. NMR spectroscopy displayed strong
78
peak broadening may suggest the presence of a paramagnetic vanadium species,
however the crystallographic data (vide infra) would suggest otherwise, as the
malonate has not been deprotonated as with the acac, implying a redox exchange has
not occurred. It is more likely that there is a degree of ligand exchange happening
between the vanadium atoms in solutions, resulting in the loss of peak resolution.
Single crystal X-ray diffraction analysis of the red crystals revealed that the tetrameric
complex [{VOCl2(CH2(COOEt))2}]4 6 had formed, as shown in Figure 39. The crystal
structure of compound 6 is a tetramer linked by four bridging oxygen atoms. From each
of the bridged vanadium centres a chlorine atom has been lost and the coordination
sphere filled with the bridging V-O bond. Around each vanadium atom there are 4 V-O
bonds. The two longest bonds are the dative bonds to the carbonyl groups of the
diethyl malonate. These are similar to the equivalent Ti-O bond lengths (2.112(4) Å and
2.102(4) Å) observed for the same malonate ligand coordinating to titanium
tetrachloride.97 In the case of the TiCl4 however, all four chlorides are retained, giving
a monomeric species. The C(2)-C(3) bond lengths of the central carbon chain are
equivalent to those of the side group C(6)-C(7), (1.496(9) &1.498(7) Å respectivelty)
suggest that no deprotonation as occurred (as with 1-3 but unlike with VO(acac)2).
The other two V-O bond lengths represent the bridging oxygen bonds, this shows that
the oxygen atom does not sit equidistant between the two vanadium atoms. The
shorter of the two has a length of 1.623(3) Å, close to that expected for a V=O bond,
for example 1.595(5) Å observed in VOCl3.25 The other bridging V-O bond distance was
considerably longer at 2.036(3) Å, resembling the dative coordination of the malonate.
This shows that the original V=O bond in the VOCl 3 has not been broken but datively
79
coordinates to another vanadium atom. This structure has been reported previously
by Sobota et al. formed via reaction of [V2(μ-Cl)2Cl2(MeCO2Et)2] with diethylmalonate to
give [V2(µ-Cl)2Cl2(MeCO2Et)4], which was exposed to air to give 6. 98
with diethyl malonate
Figure 38: 1H NMR spectra of diethyl malonate and the product of its reaction with VOCl3 [6],
all peaks exhibiting a downfield shift and experiencing broadening indicating coordination to
the vanadium metal centre
80
Figure 39: ORTEP representation of crystal structure of [{VOCl 2(CH2(COOEt2)}4] [6] with
thermal ellipsoids at the 50% probability level. Three of the four diethyl malonate molecules
and all hydrogen atoms omitted for clarity.
structure of 6 the product of reaction VOCl3 and diethyl malonate
Selected Bond Selected Bond
Structural Data Angles ° Lengths Å
crystal
system Triclinic O1–V1–Cl2 89.07(10) V1 - O1 2.077(3)
space group P-1 O4–V1–O1 83.64(13) V1 - O2 2.148(3)
a 12.516(5) Å O1–V1–O2 80.58(12) V1 - O3 1.623(3)
b 13.338(5) Å O3–V1–O1 93.47(14) V1 - O4 2.036(3)
c 17.573(5) Å O3–V1–O2 171.21(15) V4 - O3 2.027(3)
α 85.751(5) Cl1–V1–Cl2 93.66(5) O2 - C5 1.222(6)
β 84.006(5) V1–O3–V4 172.1(2) C3 - C4 1.498(7)
γ 90.000(5) O4–V1–O2 78.95(12) C4 - C5 1.492(7)
R Factor 5.51% C3–O1–V1 133.1(3) C6 - C7 1.496(9)
81
In this work however, a direct route to 6 is reported from VOCl3 and diethyl malonate.
Despite 6 being the same the structure reported by Sobota et al., with similar R factors
(4.94% cf. 5.51%), the crystallographic data differs, which is likely a result of collection
temperature (283-303 K cf. 150 K). This in turn has resulted in significant differences in
the unit cell dimensions (Sobota structure has the unit cell dimensions: a = 12.655(4)
Å, b = 13.735(3) Å, c = 18.141(3) Å, compared with: a = 12.5230(2) Å, b = 13.3895(2) Å,
c = 17.5946(3) Å, reported here). The low temperature collection reported here has
been submitted to the CCDC since there are noticeable differences in the majority of
the bond lengths and angles of the two structures and we consider this of interest to
the field. For example the V-O bond distance between the vanadium and the carbonyl
groups on the malonate is 2.177(6) Å in the previously reported structure, but
considerably shorter in this case at only 1.904(3) Å.
Dimeric oxo-bridged vanadium complexes have been produced in a similar way by
reaction of VOCl3 and pinacol. In this case one of the oxygen atoms of each pinacol
ligand is coordinated to both vanadium centres, acting as bridges, with the other
oxygen coordinating to one.99 The V-O bond lengths observed for the bridging oxygen
atoms are 2.0211(11) Å, resembling the dative coordination of the malonate, rather
than a direct oxo bridge. This is likely due to the electron decent nature of the
vanadium centre. The dative V-O bond length of 2.077(3) Å in 6 is still shorter than the
V-Cl bonds at 2.3031(13) Å. This shows how the coordination sphere of the oxygen is
important in these reactions, and until the oxygen is bound to two other elements it
will be reactive when subjected to an electron poor metal centre.
82
In compound 6, each of the vanadium atoms are coordinated to two chlorine atoms,
having lost one on coordination to the ligand. The lengths of the C-C bonds between
the two carbonyl groups are 1.492(7) and 1.496(9) Å which indicates that both are
single bonds.22 This implies that the coordination to the vanadium centre has not lead
to the loss of a proton from the central carbon atom resulting in the evolution of HCl
as would be expected and observed in the formation of compound 5. Had this occurred
the C-C bond lengths would be shortened to around 1.38 Å due to the increased
electronic density resulting from the aromaticity of the resulting species. Therefore,
the vanadium centre has been reduced to V(IV) from V(V) with concurrent oxidation
of chloride ligands to elemental chlorine. This has been observed previously in the
reaction of VOCl3 with 2-ethoxy ethanol, which was shown to coordinate in a bidentate
fashion to the vanadium whilst maintaining the ethanoic proton. 28 This reaction was
found to result in the reversible loss of chlorine ligands via oxidation to Cl2. Vapours
from the reaction were found to test positive for Cl2 gas.100
3.3.3.3. Dichloro(oxo)(diethyl succinate) Vanadium (V) [VOCl2{C2H4(CO2Et)2}]n [7]
Excess VOCl3 was reacted with diethyl succinate in hexane and stirred for 2 hours under
nitrogen. The reaction mixture turned from orange to dark red over 5 minutes and was
left to stir for 2 hours to ensure a full reaction. The reaction yielded a dark red oily
product (Scheme 8). The flask was left under nitrogen for one month, during which
time, green crystals of compound 7 were formed which were suitable for single crystal
X-ray crystallography. The NMR spectrum of 7 showed customary peak broadening
associated with coordination to a paramagnetic vanadium species (Figure 40).
83
The diethyl succinate acts as a bidentate ligand in 7, with each carbonyl binding to a
different metal centre rather than chelating as was observed with the malonate ligand
in compound 6. Likewise, each vanadium centre forms bonds with two different ester
groups thus forming a polymer chain. This polymer chain is classed as a one
dimensional coordination polymer. Coordination polymers are metal-ligand complexes
that extend "indefinitely" into one, two or three dimensions via covalent metal-ligand
bonding. Coordination polymers are also known as metal organic frameworks
(MOFs).101,102 One dimensional coordination polymers are of particular interest for
their electronic properties, for use as nanowires.103,104
The first non-cluster vanadium coordination polymer was synthesised by Zhang et al.
in 2001, with the synthesis of [VO(dod) 2]X2 (X = Cl, Br; dod = 1,4-
diazoniabicyclo[2,2,2]octane-1,4-diacetate].105 There have been other reported 1-
dimensional coordination polymers containing vanadium, one with O,O,N-dichelating
ligands106 and a further from the reaction of vanadium sulfate with 1,3-aryl-linked bis-
β -diketones in aqueous alcohol and sodium acetate followed by 4,4-bipyridine, which
stabilises formation of a polymeric species.107 Furthermore, succinate has been used
as a bridging ligand in a cobalt based MOF, forming a similar one dimensional chain,
stabilised with benzadine ligands.108
This is the first mononuclear 1-dimensional vanadyl coordination polymer to be
isolated as a single crystal and the structure probed by X-ray diffraction. The presence
of the two V-Cl bonds makes the polymer extremely moisture sensitive. The vanadium
centre is similar to that in 6 in that it is five coordinate, bound to three oxygen atoms
and two chlorine atoms. The complexes differ in the way that the ligand is oriented. In
84
compound 5, the carbonyl oxygen atoms are cis to each other since they are from the
same acac ligand. In compound 7, the carbonyl oxygen atoms originate from two
different molecules and are positioned axial to each other, rather than equatorial, as
in the case for 5.
This is reflected in the O(1)-V(1)-O(21) bond angle of 167.72(6)˚ in 7, whereas in
compound 5 the equivalent O-V-O angle between the carbonyl oxygen atoms is
84.59(12)˚ (Figure 41).
Structure 7 differs from the related cobalt polymer synthesised by Roy et al.,34 since
the O-M-O bond angle was observed to be 180°. This is due to the electronegativity of
the two chloride ligands, resulting in the Cl(1)-V(1)-Cl(2) bond angle being 131.62(3)°,
as opposed to the expected 120° for a five coordinate geometry. Regardless of this the
chain is linear due to the positioning of the succinate bridge.
The V-O bond lengths in 7 for the carbonyl oxygen atom are 2.0483(13) Å and
2.0370(13) Å respectively, which resembles the dative coordinative bonds observed
with the malonate ligand in 6. It is likely that VOCl3 forms this polymer because of the
two bridging carbon atoms between the ester groups. When there is one bridging
carbon, as is the case with the malonate and the acac, both of the carbonyl oxygen
atoms lie on the same side of the chain. In the case of the succinate ligand there are
two bridging carbon atoms, the bond between which can rotate freely, allowing the
carbonyls to lie on opposite sides due to steric hindrance. This appears to be the more
favoured conformation of the ligand, most likely due to the presence of the relatively
large, electron rich chlorine centres.
85
with diethyl succinate.
Figure 40: 1H NMR spectra of diethyl succinate and the product of its reaction with VOCl 3 [7],
all peaks exhibiting a downfield shift and experiencing broadening indicating coordination to
the vanadium metal centre.
86
Figure 41: ORTEP representation of crystal structure of [VOCl2{C2H4(CO2Et)2}]n (7) with
thermal ellipsoids at the 50% probability level. Hydrogen omitted for clarity.
Table 10: Crystallographic Data and Selected bond lengths angles for the crystal structure of 7
the product of reaction VOCl3 and diethyl succinate.
Selected Bond Selected Bond
Structural Data Lengths Å Angles °
crystal system orthorhombic Cl2 - V1 - Cl1 131.62(3) V1 - Cl1 2.2854(6)
space group P212121 O3 - V1 - Cl1 113.26(6) V1 - Cl2 2.2776(6)
a 8.3503(3) Å O3 - V1 - O1 93.40(7) V1 - O3 1.5783(14)
b 13.5046(4) Å O3 - V1 - O21 98.62(6) V1 - O1 2.0483(13)
c 15.2041(5) Å O1 - V1 - Cl2 85.88(4) V1 - O21 2.0370(13)

α 90.000(5) O2¹ - V1 - Cl2 86.74(4) C3 - C4 1.487(3)
β 90.000(5) C3 - O1 - V1 142.20(13) O1 - C3 1.228(2)
γ 90.000(5) C3 - C4 - C5 113.60(16) O4 - C3 1.312(2)

R Factor 3.40% O2¹ - V1 - O1 167.72(6) O2 - C6 1.237(2)
O5 - C6 1.312(2)
O5 - C7 1.473(2)
C4 - C5 1.530(3)
87
Each of the diester molecules were observed to coordinate datively to the vanadium
metal centre, facilitated by the loss of a chlorine ligand, as was observed for compound
6. The structures differ greatly according to the ligand, with the malonate appearing to
favour oxo-bridged oligomeric structures, whilst the less oxygen rich acac favours a
simple monomeric species. This can be related to the effectiveness of the mineral oil
as a means of removing VOCl3 in the Chloride Process as opposed to other organic
mixtures. By encouraging the formation of oligomeric species the likelihood of
vanadium crashing out of the TiCl4 as a solid or liquid in the Chloride Process rather
than passing through the distillation column is increased, leading to a cleaner final
product. Another interesting characteristic of the diester species is that both the 1H
NMR spectra and bond lengths observed suggests that there is no deprotonation of
the ligand to form a conjugated species, as is observed in the acac ligand. This is
unexpected, especially for the diethyl malonate ligand, the deprotonated state of
which is widely used in organic synthesis.35 This may relate back to the propensity for
forming oligomeric and polymeric species, which could facilitate the loss of Cl 2 gas in
the event of the dimerisation of two unstable VOCl 3L centres as a preferable reaction
pathway.28
3.3.3.4. Reaction of VOCl3 with Other Species
In addition to the above three structures, VOCl 3 was reacted with other ester species,
however it proved impossible to isolate crystals of these products. Bis-isopropyl
malonate and glycerol tribenzoate were reacted with VOCl3 to give 8 and 9
respectively. Compound 8 was formed under the same conditions as 5, 6 and 7 with
the ligand added dropwise to a solution of VOCl3 in hexane.
88
Figure 42: 1H NMR spectra of bis-isopropyl malonate and the product of its reaction with VOCl3 [8], all
peaks exhibiting a downfield shift and experiencing broadening indicating coordination to the
vanadium metal centre.
Figure 43: 1H NMR spectra of bis-isopropyl malonate and the product of its reaction with VOCl 3 [9], all
peaks exhibiting a downfield shift and experiencing broadening indicating coordination to the
vanadium metal centre.
89
This approach with glycerol tribenzoate was found to be ineffective as glycerol
tribenzoate is highly insoluble in hexane and preventing the reaction from going to
completion. The reaction was repeated in toluene and refluxed for 15 hours, at which
point a black liquid was obtained. NMR of 8 (Figure 42) reveals considerable peak
broadening in all three peaks, as well as a minor downfield directing effect relating to
the electron poor desheilding nature of the VOCl3, The most noticeable shift is, as
expected, the central proton of the malonate, which is shifted from 5.05 to 5.44 ppm.
Peaks at 1.25 and 3.31 ppm in the unreacted bis-isopropyl malonate spectra have also
shifted downfield, however this is far less pronounced.
As a crystal structure could not be obtained due to the extremely air sensitive nature
of the product, the NMR data confirms coordination of the malonate to the VOCl3. The
highly similar nature of the ligand would suggest that the mostly likely structure taken
on by 8 would be cyclic, similar to 6, with the vanadyl oxygen forming bridges between
the metal centres.
The same is true for the glycerol tribenzoate and VOCl3 complex. In this case the
broadening and downfield shifting in the NMR spectrum (Figure 43) is far more
pronounced than that seen in the bis-isopropyl malonate. The doublet peak of proton
B has shifted from 4.78 to 5.10 ppm and broadened greatly. The single proton singlet
at 5.9 ppm has also broadened to the point to where it is barely visible from the
background. The signals for the aromatic species have homogenised to form two
extremely broad peaks over the range of 7.3 to 8.3 ppm. This once again confirms
coordination of the glycerol tribenzoate to the vanadium centre, however predicting
the structure is considerably more challenging. The length of the bridging alkyl chain
90
between the carbonyls resembles that of the succinate species, and is long enough for
polymerization to be possible. However the fact that there are three carbonyl groups
would make the formation of a branched polymer species possible. The complexity of
the resulting species would explain the extreme difficulty in forming crystals.
3.4. Conclusion
All of the diesters studied coordinated to the titanium(IV) chloride in the same way, forming
complexes of an octahedral geometry, distorted by the presence of the chloride ligands. No
chlorine was lost from the TiCl4 upon formation of the complexes 1 - 3. The same is true for
glycerol tribenzoate titanium chloride complex 4, with two of the ester groups datively
coordinating via the carbonyl oxygen to a single titanium centre. This lead to a six coordiate
octahedral centre with octahedral geometry. The third ester carbonyl group coordinates to
a second titanium chloride molecule. The resulting species instability leads to dimerization.
Formation of a two choro-bridged Ti2Cl6(µ-2)Cl2 centre was observed, with both titanium
centres coordinated to a single carbonyl oxygen of a different glycerol benzoate molecule. As
with the other structures, no chlorine loss is observed.
Conversely, with VOCl3, all three molecules coordinating to the vanadium lead to the loss of
a chlorine atom. This is significant as the reaction could not be reversed by simple dissociation
of the ligand alone, which could occur for the TiCl4 complexes at the high temperatures used
in the Chloride Process.
Additionally, whilst all three diesters form monomeric species with TiCl4 to compensate for
the loss of the chlorine atom, the diester complexes of vanadium (4 + 5) form a ring and a
91
chain respectively, with the oxygen of the VOCl3 forming a dative bond with another metal
centre.
The addition of oil in the Chloride Process for the production of TiO2 is carried out at around
200 ˚C, at which point all of the metal chlorides present are in the vapour phase. At this
temperature, the dative coordination would likely be more reversible, as neither the TiCl4 nor
the esters are altered structurally by coordination. Conversely the addition of donor ligands
to VOCl3, the species removed by the oil, frequently causes the dissociation of one of the
chlorine atoms.
This suggests that even though the Lewis acidy of the TiCl4 is greater than VOCl3, the VOCl3
may be changed irreversibly by reaction with carbonyl groups in the oils, making it the
preferred species for the reaction of the triglyceride group of the oil, despite being present
in far lower concentrations in the gas stream. The group may coordinate briefly to many
titanium(IV) chloride molecules before coming into contact with one vanadium oxytrichloride
but having done so the energy barrier to reversibility would be greater and so the interaction
leads to a permanent formation of a new species which is not able to pass through the
distillation.
92
Chapter 4
Vanadium Removal Flow System
4.1. Introduction
Having analyzed the contents of various samples from different points along the Chloride
Process using NMR spectroscopy, and modeled the interaction of TiCl4 and VOCl3 with diesters
and triesters as analogues for the oils, the next step was to try and recreate the gas phase
interactions of the heated TiCl4 and VOCl3. To achieve this, a piece of glassware was designed
to mimic the distillation column of the Chloride Process, complete with the ability to cycle the
same liquid through the distillation repeatedly. Samples of TiCl4 and VOCl3 that had been
passed through this apparatus were analyzed using NMR using the same method described
in Chapter 2.
In order to investigate the behaviour of oil in the Chloride Process on a laboratory scale, the
conditions within the Chloride Process had to be emulated as closely as possible. This
involved reacting mixtures of gaseous titanium(IV) chloride and vanadium(V)
oxytrichloride with the liquid oil at over 140 ˚C under moisture free conditions. An
additional challenge lies in the fact that the industrial scale production of TiO2 is a cyclic
process. Material that does not pass through the distillation columns is not discarded but
recycled into the chlorinator, passing through the process repeatedly. This is done to minimize
waste and production of harmful by-products as well as to maximize the yield of titanium
dioxide obtained from the ore.
93
A potential problem with this method however is that byproducts, such as those formed by
the breakdown of the oil, can be recycled repeatedly and build up in the system, leading to
fouling or unwanted side reactions. To accurately recreate the addition of oil, the same
sample of TiCl4/VOCl3 needs to react with the oil repeatedly, simulating possible build-up of
breakdown products. In order to achieve this, a flow cell was designed by modifying a Soxhlet
extractor (Figure 44) with the purpose of forcing gaseous TiCl4 and VOCl3 over a small amount
of rapidly stirred oil, simulating the injection of the oil as a dispersion of liquid droplets into
the gaseous TiCl4 stream, as shown in the schematic of the process (Figure 45).
Figure 44: Schematic of a Soxhlet extractor.109
This was achieved by modifying the bottom pipe of the Soxhlet, through which the distillate
drains in the flask. To the side of this tube was connected to a thinner glass pipe (0.5 cm
diameter) that leads into a second two necked round bottom flask. From this flask, a second
0.5 cm wide pipe leads 10 cm directly upwards before connecting to the upper Soxhlet. This
94
set up effectively elongates the distillation tube of the Soxhlet to include a second round
bottom flask to which oil can be added (Figure 45).
4.2. Experimental
Before each experiment, 5 ml of the selected oil (Tables 11 and 12) was added to flask 2. The
flow cell was assembled as shown in Figure 45 and purged with nitrogen anhydrous nitrogen
from a Schlenk line for 30 minutes whilst the bottom two flasks were heated to 140 ˚C using
a silicone oil bath and a hot plate, in order to remove as much moisture from the system as
possible, so as to more accurately emulate the conditions during industrial TiCl4 removal.
The apparatus was fitted with an ice water condenser in order to ensure that all of the TiCl4
remained inside the system, not escaping through the top of the Soxhlet extractor. The
apparatus was also fitted with an oil bubbler in order to regulate the flow of nitrogen through
the system, which was kept as low as possible. Following this, the condenser was removed
from the top of the Soxhlet chamber and replaced with a rubber septum, through which 15
ml of TiCl4, VOCl3 or a mixture of the two were injected. This caused the Soxhlet to fill, leading
to roughly 10 ml falling into flask one and 5 ml remaining in the top chamber of the Soxhlet.
This was important as it prevented backflow of the hot vapour, forcing it through the system
in the correct direction, ensuring interaction with the oil in flask 2. Following the addition of
the TiCl4, VOCl3 or mixture, the rubber septum was removed and replaced with the condenser
and oil bubbler. After the addition of the metal chloride species both flasks were stirred
vigorously using magnetic stirrer bars. The movement of the metal chloride species within the
system can be described as 5 steps. Throughout the process the system was heavily insulated
using foil and thermal insulation material.
95
Figure 45: Gaseous metal chloride flowcell from a modified Soxhlet extractor.
The metal chloride species pass through the flowcell in five steps:
96
1) The titanium chloride was added via the top of the Soxhlet extractor into the flow cell
whereupon it filled the Soxhlet before dropping through into flask one. The TiCl4 was
heated to 140 ˚C using an oil bath. This lead to evaporation of the TiCl4. As some liquid
TiCl4 remain in the Soxhlet above, the path of least resistance was into flask 2.
2) Flask 2 contain roughly 5 ml of the selected oil, which was vigorously stirred at 140 ˚C.
The heating caused the oil to become far less viscous and droplets were kicked up by the
stirrer, emulating the dispersion of oil that is injected into the gas cleaning tank. The
gaseous TiCl4 interacted with the oil in flask 2 whist purely in the gaseous state as the
inlet was not submerged in the oil.
3) Following reaction with the oil, the TiCl4 and VOCl3 were distilled through the 0.5 cm
diameter glass tube, along with any volatile products that may have formed in flask 2.
4) The TiCl4 and volatile products moved into the condenser. Here their temperature was
reduced sufficiently to cause them to enter the liquid phase and drip down, filling the
Soxhlet extractor sample chamber.
5) The full Soxhlet emptied into flask one beginning the cycle over. This simulates the cyclic
nature of the Chloride Process, allowing possible byproducts to build up over multiple
cycles.
97
Whilst the experiments were carried out, a third magnetic stirrer was kept inside the Soxhlet
chamber above flask 1. This reduced the amount of liquid that was required to build up inside
before emptying into flask 1 from 10 ml to around 5 ml. This was found to improve the flow
of the gas and prevent either of the heated flasks from boiling dry.
Once the Soxhlet had filled and emptied a certain number of times the nitrogen flow was
removed and the system left to cool to room temperature, at which point three NMR samples
were taken.
- The contents of flask 1, which contained TiCl4 and any by-products built up over
several cycles.
- The contents of the Soxhlet extractor, which contained the freshly distilled TiCl4
- The contents of flask 2, which contained the oil residue, which had often solidified as
well as the vanadium oxytrichloride that had been removed from the gas stream.
4.2.1. Soya Bean Oil Experiments
The initial experiments all used soya oil as this is the oil currently used by Huntsman Pigments
and Additives in the removal of VOCl3. After some initial optimizing of the technique by
running toluene, SiCl4 and finally TiCl4 through the system, a series of experiments were
carried out. The details of these experiments are shown in Table 11. The 1000 ppm
vanadium(V) oxytrichloride dosed titanium(IV) chloride was obtained from Huntsman. The
40,000 ppm VOCl3 sample was made up in a separate Schlenk flask by adding 1 cm 3 of VOCl3
to 25 cm3 of TiCl4.
98
Table 11: Key experiments involving soya oil carried out using the flow system
EXP Chloride Oil flask No. Time Results + Observations

no. flask Cycles /hrs
1 TiCl4 soya 4 3 Oil becomes brown and viscous after

3 cycles. NMR shows chlorinated
alkene species have distilled into
Soxhlet.
2 VOCl3 soya 3 4 Oil became black and viscous during
first cycle. VOCl3 that distills into
Soxhlet identical to starting material
by NMR.
3 TiCl4 + VOCl3 soya 4 3 Oil becomes brown and viscous after
(1000ppm) 2 cycles. NMR shows chlorinated
alkenes species have distilled into
Soxhlet.
4 TiCl4 + soya 4 4 Oil becomes black and viscous after
VOCl3 1.5 cycles. 1H NMR shows no
(40,000ppm) broadening suggesting that VOCl3 has
been removed. Chloroalkane peaks
only present in oil.
5 TiCl4 - 4 3 NMR shows no reaction.
4.2.2. Analysis of Naphthenic Oils
Following the experiments with soya bean oil, a second flow cell study was carried out using
various naphthenic oils (Table 12). All of these experiments were carried out as described in
the previous section, with an NMR spectrum being obtained from each of the naphthenic
samples for comparison. After initially sampling several naphthenic oils with only TiCl4 it was
decided that Flavex 977(s) would be reacted with the other metal chloride mixtures in the
same way as the soya bean oil, as it possessed greater fluidity and was observed to become
solid less rapidly upon exposure to TiCl4.
99
Table 12: Experiments run through the flow cell apparatus using naphthenic oils
EXP Chloride Oil No. Time Results + Observations

no. flask Flask Cycles (hrs)
6 TiCl4 Flavex 4 2 Oil becomes dark red/brown, almost as
977(s) soon as TiCl4 begins boiling however
remains fluid. TiCl4 condensing in Soxhlet
dark red
7 TiCl4 EH 4 3 Oil becomes dark red/brown, almost as
L550 soon as TiCl4 begins boiling however
dark red
8 TiCl4 Gravex 4 3 Oil becomes dark red/brown, almost as
926 soon as TiCl4 begins boiling however
dark red
9 TiCl4 Edelex 3 4 Oil becomes dark red/brown, almost as
926(s) soon as TiCl4 begins boiling however
dark red
10 Toluene Flavex 4 1.5 NMR shows pure toluene
977(s)
11 TiCl4 + Flavex 4 4 Same as 8 except the oil is visibly darker
VOCl3 977(s) during the initial cycle.
(1000ppm)
12 TiCl4 + Flavex 4 3 Oil goes from colourless to black as soon
VOCl3 977(s) as it is exposed to the gaseous species.
(40,000ppm) The distilled fraction appears to stain the
apperattus orange (Figure 30).
Precipitation of black solid is observed in
distillate
13 VOCl3 Flavex 3 4 Oil initially appears less viscous, however
977(s) over the course of the cycle it gradually
becomes more solid. By the third cycle
only 4 ml of VOCL3 remains, the rest
appears to have been taken up by the oil.
100
4.3.1. Soya Bean Oil Analysis
4.3.1.1. TiCl4 + Soya Oil
TiCl4 caused the oil to gradually go from light yellow to brown, becoming more viscous over
time. After the Soxhlet had emptied and filled three times, the oil had become so thick that
the stirrer could no longer move. The TiCl4 passing into the Soxhlet was clear and colourless,
lighter in colour than the starting material, which was a light yellow due to aging. The NMR
spectra taken from flask 1 and the Soxhlet show a number of signals that were not present in
the starting material.
These peaks were not present in any of the spectra from when the cell was run without oil,
which only shows signals from the grease used on the joints. This suggests strongly that the
peaks are a result of TiCl4 breaking down the oil. Figure 46 shows enhanced images of two
major new peaks, whilst Figure 47 shows the COSY (correlation spectroscopy) spectrum of
the same two peaks. The two signals in question are a quartet of doublets at 3.95 ppm and a
pentet at 5.2 ppm. These are the strongest peaks that appear as a result of the flow system
experiment. The COSY shows that the two signals are from protons situated on adjacent
carbons. This is in keeping with the 1H NMR signal of a glycerol group, which shows both the
pentet and quartet of doublets when coordinated to TiCl4, however both were considerably
more downfield shifted due to the presence of the three highly electron withdrawing ester
groups (Figure 47). These signals strongly suggest that the glycerol group of the soya bean oil
is responsible for these signals emerging.
101
Figure 46: 1H NMR spectra of the starting material and the contents of flasks 1, 2 and the Soxhlet
extractor from experiment in which gaseous TiCl4 was reacted in soya oil. Spectra taken in CDCl3. The
figure shows the region between 3.5 and 5.5 ppm.
Figure 47:1H NMR COSY spectrum of the contents of flask 1, from experiment in which gaseous TiCl4
was reacted in soya oil, showing the region between 3.5 and 5.5 ppm. Spectra taken in CDCl3.
102
Despite this it is unlikely that the soya bean oil was able to make it though the distillation intact.
Neither the NMR spectra from the Soxhlet or flask 1 showed any trace of the unsaturated
alkene peaks at 2.1, 2.4 and 2.8 ppm that were also seen in the NMR spectrum of the soya bean
oil. Furthermore the signals have undergone an upfeild shift, with the equivalent signals in soya
bean oil appearing at 5.4 and 4.3 ppm as opposed to the peaks at 5.1 and 3.9 ppm observed in
this spectrum (Figure 46).
This evidence points to the molecules that produce these signals being the products of the
breakdown of the soya bean oil, with the fatty acid chains separating from the glycerol.
The resulting molecule must still be able to provide the signals resulting from a 1, 2, 3 propane
species whilst providing downfield shifts in the correct environments. The protons on the 1 and
3 carbons must also be equivalent as otherwise the signals from the carbons in the 1 and 2
positions of the propyl backbone would not be in the same position, as was observed in these
spectra. The species that fits best in simulated NMR is that of a dichlorinated species in which
the ester groups have been replaced by the chlorine in the 1 and 3 positions (Figure 48 and 49).
Figure 48: Proposed general structure of the main NMR product of the reaction of TiCl4 and soya bean
oil in the flow cell
103
This matches well with the product of the reaction of the TiCl4 with glycerol tribenzoate as
shown in the previous chapter. The TiCl4 was found to coordinate to the two terminal ester
groups or the central ester on two molecules forming a chloro-bridged tetra-titanium species.
It is probable therefore that this species, formed in the gas phase, or at least part of it, should
be the intermediate for the formation of the volatile molecule that would be able to distill over.
The R group represents the carbon chain that would also need to be far shorter to make
distillation possible.
Figure 49: Comparison of the spectrum of the contents of flask 1 produced from putting TiCl4 through
4 cycles with soya bean oil, compared with the simulated 1H NMR spectrum of 1, 3 dichloro propan-2-
acetate.The simulation does not factor In long range coupling of the 4H peak which was added
manually. Simulation carried out using NMRDB NMR predictor software. 110–112
The chlorination of an ester species from TiCl4 would be unfavorable due to the greater bond
enthalpy of the C-O linkage than the formed C-Cl (348 kJ mol-1 vs 328 kJ mol-1 respectively) as
well as the large energy barrier of 373kJ mol-1 required o break the TiCl4 bond. 113 However, it
104
should be noted that just because the formation of the above distillate product is unfavorable
does not mean it should be discounted. The flowcell device is designed to allow for the buildup
of marginal species over a period of hours, as was seen in the Chloride Process, with undistilled
components cycled repeatedly through the entire system. Even if a very small amount of each
5 ml of distilled TiCl4 reacts with the oil in this way, it would build up in the system, being
removed from the oil flask via distillation preventing the reaction from reaching equilibrium.
The spectrum also shows weaker signals that could be the same molecule but with different
ester groups substituted. A postulated mechanism is shown in Figure 50, with the chlorine
attacking the electron poor carbon adjacent to the ester and the resulting rearrangement
leading to fission of the C-O bond. The resulting species rearranges giving a covalent structure
similar to those observed by Barrow et al.114 It should be noted that despite the species denoted
by Barrow being bis-titanium dimers bridged by carboxyl ligands, with both carboxyl oxygen
atoms coordinating to a titanium centre, a similar reaction could occur here, with the second
stage facilitated by the presence of a second species.
Figure 50: Suggested mechanism for the formation of the 1, 3 dichloro 2 acetyl propane species
observed in the 1H NMR of the flow-cell distillate.
105
Despite the unfavourable nature of this reaction due to the Lewis acid character of the titanium
chloride, it would be irreversible and as the experiment involves cycling the same sample of
TiCl4 repeatedly, even a product that forms in minuscule amounts could be observed to build
up.
Another explanation could be that the formation of HCl in the system plays a part in the
chlorination and stabilizing the resulting TiCl4 carboxylate species. The cause of this chlorination
could either be the formation of small amounts of HCl from residual water, as well as the small
amount of water in the nitrogen stream. This would not be entirely inaccurate to the process
as before the addition of oil, a small jet of water vapour is injected in order to remove water.
4.3.1.2. VOCl3 + Soya Oil
When the flow system was used with only VOCl3 present, the oil was found to become viscous
far more rapidly, becoming black and solid just after the first Soxhlet-full had passed through.
It was observed that most of the VOCl3 appeared to have been ‘consumed’ by the oil, with only
around 5 ml flowing through the system by the third cycle. The distilled VOCl 3 in the Soxhlet
appeared unchanged. This implied that all of the soya oil had reacted with the VOCl 3 allowing
what remained to pass through the flow cell repeatedly undisturbed. This was confirmed by
the 1H NMR analysis (Figure 51). The VOCl3 in the Soxhlet adaptor showed no significant change
from the starting material, except for the removal of peaks at 0.5 – 0.7 ppm, which are the
result of unsaturated alkyl impurities. These were apparently removed by the distillation.
The NMR spectrum of the oil flask was similar to that of unreacted soya oil, except broadened
significantly, suggesting the presence of vanadium, coordinated to the oil. This is in keeping
with current theory that the VOCl3 coordinates to the double bonds and triglyceride backbone
106
irreversibly, rendering it unable to pass through the distillation column. Conversely the TiCl4
coordinates in a more reversible manner, allowing it to pass through the distillation.
These findings support the work in which TiCl4 and VOCl3 were reacted with diesters. The
reaction with the VOCl3 was found to displace chlorine, which heavily reduces the reversibility
of the reaction when compared with the TiCl4, which remains intact upon coordination, simply
adopting an octahedral moiety, maintaining all four chlorine atoms.
Figure 51: 1H NMR spectra of the starting material and the contents of flask 2 and the Soxhlet
extractor from experiment 2, in which gaseous VOCl3 was reacted in soya oil. Spectra taken in CDCl3.
The figure shows the region between 0.6 and 5.5 ppm.
4.3.1.3. TiCl4 + 1000ppm VOCl3 + Soya Oil
107
Of the two mixtures put through the flow cell, the 1000 ppm VOCl3 mixture was far more similar
to the gas stream found in the actual Chloride Process. The results are highly similar to those
obtained, using pure titanium chloride. The 1H NMR spectra of flask 1 and the Soxhlet both
show the same peaks between 3 and 6 ppm including the large quartet of doublets at 3.95 ppm
(Figure 52). The lack of any broadening in the spectra as well as the colourless nature of the
distillate suggests complete removal of the vanadium species by the oil. This is a promising
result as it shows that the flow cell is able to emulate the distillation and removal of VOCl 3 in
the plant effectively.
extractor from experiment in which gaseous TiCl4 with 1000 ppm VOCl3 was reacted in soya oil.
Spectra taken in CDCl3. The figure shows the region between 3.5 and 5.5 ppm.
4.3.1.4. TiCl4 + 40,000ppm VOCl3 + Soya Oil
108
The presence of a higher proportion of VOCl3 in the distillation mixture lead to the oil solidifying
far more rapidly, similar to when only VOCl3 was passed through the system. However the
distilled product in the Soxhlet remained colourless, suggesting successful removal of the
VOCl3. This is supported by the lack of peak broadening in any of the flask’s spectra, save the
oil suggesting that the VOCl3 had been effectively removed as with the Chloride Process itself
(Figure 53)
Figure 53:1H NMR spectra of the starting material and the contents of flasks 1, 2 and the Soxhlet
extractor from experiment 4 in which gaseous 40,000 ppm VOCl3 In TiCl4 was reacted in soya oil.
Spectra taken in CDCl3. The figure shows the region between 3.5 and 5.5 ppm.
Interestingly, this far higher dose of VOCl3 appears to prevent the carryover of the
chloroacetate formed by reaction of the oil, with the TiCl4. The signals relating to the
chlorinated glyceride species are only present in the spectra of the oil and neither of the TiCl4
samples from the Soxhlet or flask 1. This suggests that whilst the TiCl4 may still be reacting with
109
the oil as seen in the previous experiments, it is impeded by the thickening of the oil and cannot
pass through the distillation column. The lack of peak broadening in the quartet of doublets
implies that although prevented from passing over, this species has not directly coordinated to
the vanadium.
4.3.2. Naphthenic Oil Analysis
Following the successful study involving soya bean oil, the flow system was used to investigate
a range of prospective oils that can be used to separate VOCl3 from TiCl4. The purpose of this
study was less about defining the breakdown products of these species, but to investigate the
ability of these oil to separate the two transition metal species, as well as to determine whether
or not breakdown products formed by using NMR to observe them, as well as looking at the
colour of the oil entering into the Soxhlet extractor. This study was carried out at the request
of Huntsman Pigments and Additives in order to better understand their competitors method
for refining TiO2.
Naphthenic oils are produced from the naphtha fraction of crude oils. Naphthenics feature
hydrocarbon chains containing ring systems, both saturated and conjugated. The oil’s are made
up of different blends of these species in order to control characteristics such as viscosity and
lubricity. Structurally, all of these oils are therefore far more complex than soya-bean oil, which
is formed of the same basic triglyceride structure, but with only four possible long unsaturated
chains, all of which are linear An example of some of the commonly featured species in
naphthenic oils are shown in Figure 54. Viscosity, boiling range and colour are all affected by
the number and size of these groups, as well as their ratio to straight chain portions.
110
Figure 54: Example structures of naphthenic acids, derived from the hydrolysis of naphthenic oil.
Table 13: Comparison of acidity of naphthenic oils and soya bean oil.
Oil pH
Soya bean Dec 13 7
Flavex 977(s) 5.9
Ergon Hygold L500 6.3
Gravex 926 (s) 6.6
Edelex 926 6.4
This study focussed on the oil Flavex 977(s) as a previous study at Huntsman Pigments and
Additives concluded that it showed potential for the removal of VOCl3. Other oils tested were
Ergon-Hygold 550 (the worst performer in the aforementioned test, as well as Gravex 926 (s)
and Edelex 926. These were chosen to represent the ‘other end’ of the naphthenics spectrum,
being far lighter in colour and less viscous. The pH of all of the oils was tested using a pH meter.
Each of the oils were found to be more acidic than soya bean oil (Table 13).
Unfortunately it was impossible to obtain information as to the exact or even approximate
contents of each of these oils, as the companies that produce them, Shell, Nynas and Ergon
Hygold, would not respond to any enquiries made about them.
4.3.2.1. NMR of the Naphthenic Oils

111
All of the oils tested give extremely similar NMR spectra (Figure 55). This is not to say that they
are the same, simply that the number of proton environments are in the same region but
slightly different, which broadens the spectrum peaks to the point where differences cannot
be seen. All spectra showed a peak at 1 ppm representing the primary carbon at the end of the
hydrocarbon chains, a large peak at 1.5 ppm for secondary carbons and cyclic chains. This peak
is very broad at the base due to the number of different ring systems. Several other broad
humps can be seen on the side representing either more stained rings or rings with a double
bond. Finally a region between 7-8 ppm of many smaller peaks representing a small number of
conjugated ring systems can be observed.
Figure 55: Comparison of the 1H NMR spectra of four naphthenic oils: Flavex 977s (blue), Ergon Hygold
L550 (red), Gravex 926 (green) and Edelex 926s (purple). Spectra taken in CDCl3. The figure shows the
region between 0.8 and 2.4 ppm
4.3.2.2. Flow Cell Experiments
112
In a setup identical to the previous section, 20 ml of TiCl4 was pushed through the flow system
with 5ml of oil stirred vigorously in the second flask. In each case the Soxhlet was allowed to fill
four times when possible before the heater was turned off and the samples collected.
In the case of all 4 oils, regardless of the original oil colour and viscosity, interaction with the
gaseous TiCl4 caused the oil to immediately take on a dark red colour as the TiCl4 began to boil
and pass over it. The TiCl4 that enters the Soxhlet having passed though the oil appears red in
all four cases (Figure 56). The stained TiCl4 was observed to layer on top of the clean TiCl4 within
the Soxhlet extractor, not mixing at all until the first emptying (Figure 57). This red colouring
becomes more prominent with each emptying of the Soxhlet. This was never seen when soya
bean oil was used.
The cause of this red staining was most likely related to the formation of a weakly bound charge
transfer complex between the titanium chloride and the aromatic portions of the oil that had
broken away from the molecule and carried over to distillation column. This is a commonly
observed phenomenon with TiCl4, most commonly observed in its interaction with toluene and
tetrahydrofuran (THF). 115 This phenomenon is the result of the electron rich conjugated and
double bond systems within the oils donating weakly to the Lewis acidic TiCl4 centre, however
not being able to bond directly due to the Lewis basic character of these species being too
weak, as well as steric hindrance. This results in the formation of a weakly bound charge
transfer complex, usually deep orange or red in colour. These species can only exist in solution,
with solids unable to form and lower boiling aromatics still being separable by distillation. It has
been observed that solutions of TiCl4 with toluene do form small amounts of solid when left
over long period.115 Taking the red solution and refrigerating it does not result in the formation
of a solid.
113
The Flavex 977s oil appeared to be able to effectively remove VOCl3 from the gaseous TiCl4.
When the flow system was run with 15 ml of VOCl3, only 4 ml remained after three cycles, the
rest appeared to have been ‘captured’ by the Flavex 977(s). When VOCl 3 and TiCl4 were run
together, the distillate appeared dirtier, staining the pipe leading up to the Soxhlet with a black
solid. This implies that gas phase reactions continue to occur throughout the distillation,
resulting in the formation of a solid that crashes out in the liquid phase (Figure 58). When
toluene was run through the system with Flavex 977(s) no oil was found to have passed
through, implying that TiCl4 and VOCl3 were responsible for the breakdown and carry-over of
the oil fragments, not simply the heating and agitation.
Figure 56: Pure TiCl4 and Flavex 977(s) after 4 cycles, contents of Soxhlet adaptor. TiCl4 appears to
break down the naphthenic, which then carries over, staining the distillate red.
114
Figure 57: Gravex 926 with TiCl4 during first cycle. The red distillate is seen to gather on top off the
clean TiCl4 that starts off in the Soxhlet, before falling though when it fills.
Figure 58: 25ml TiCl4 and 1ml VOCl3 with Flavex 977(s). Flow system is stained; solid in Soxhlet
suggests that whatever carries over continues reacting.
115
4.3.2.3. NMR Experiments with Flavex 977s
Following the four initial experiments with TiCl4 and the four naphthenic oils, all of which
yielded extremely similar spectra, further experiments were carried out, focussed on the Flavex
977s oil, altering the contents of flask one, in order to study the effect of VOCl 3 on the oil, as
well as its ability to separate it from TiCl4.
extractor from experiment in which gaseous TiCl4 was reacted in naphthenic Flavex 977(s) oil Spectra
taken in CDCl3. The figure shows the region between 0.0 and 5.4 ppm.
Figure 59 shows the NMR spectra of the contents of the Soxhlet, Flask 1 and the oil from flask
2 after four cycles of gaseous TiCl4, as well as the unreacted oil. The familiar pattern of
diminished CH2 and enhanced CH3 was seen. The TiCl4 flask and Soxhlet flask also show traces
of chloroakanes, implying that the chlorination of double bonds seen with soya oil was still
116
occurring, but less prevalent. Several peaks were only seen in the TiCl4 flask spectra, implying
continued reaction of the oil flask after distillation.
Pure TiCl4, pure VOCl3 and TiCl4 containing 1000 ppm and 40,000 ppm of VOCl3 were put
through the flow system with Flavex 977s. The NMR spectra of the Soxhlet extractor’s contents
after each experiment are compared in Figure 60. The spectra from the Soxhlet were largely
the same for each. The peak representing chain and cyclic species was reduced in relation to
the straight chain peak. It can be inferred from this that the chains carrying over are
considerably shorter than the ones in the starting oil.
Figure 60: Comparison of the 1H NMR spectra of Flavex 977s compared with the contents of the
Soxhlet extractor after three cycles of TiCl4 (red), VOCl3 (green), TiCl4 + 40000ppm VOCl3 (purple) and
toluene (yellow) . Spectra taken in CDCl3. The figure shows the region between 0.8 and 2.4 ppm.
117
Firstly that the TiCl4 is breaking down the oil into smaller fragments, similar to the soya oil. It
could also be due to the distillation only carrying though the smaller chains, however as nothing
was seen to carry through when TiCl4 was replaced with toluene, this would still require an
interaction with the TiCl4 or VOCl3, possibly in the form of complexation. Following this it is
likely that the straight chain alkane fragments that carry through react further after distillation.
This explains the solids carrying over, as seen in the Huntsman Pigments and Additives study.
The oil appears to be efficient at vanadium removal. Even with very high vanadium
concentration in the starting material, there was not broadening in the final spectra.
Furthermore, when using only VOCl3 after three cycles, there was not enough to fill the Soxhlet,
the rest appearing to have congealed with the oil in the second flask. The first flow system
experiment carried out was that of pure TiCl4 and Flavex 977s.
4.4. Conclusion
A lab bench scale system for removal of VOCl3 from a gaseous stream of TiCl4 using liquid phase
oil was designed and built. This involved modifying a Soxhlet extractor so that the volatile
products pass through a round bottom flask and an extended column further than directly into
the sample chamber. This was done in order to analyse the side products formed during the
vanadium removal process, with TiCl4 reacting with the oil species to give unwanted volatile
products.
The results conclusively show that TiCl4 is capable of breaking down the soy oil into component
fragments capable of passing through a distillation column. The most abundant species formed
appears to be the result of the triester backbone of the oil being chlorinated at the 1, and 3
positions, which mirrors the way in which the species is observed to coordinate to TiCl4 in the
118
case of glycerol tribenzoate. This suggests that the TiCl4 is capable of chlorinating the
triglyceride. This reaction is not likely to be favorable and therefore this only happens in small
amounts, however, it is likely that this species could build up over time within the Chloride
Process due to the recycling of the undistilled fractions.
This is not how TiCl4 usually behaves with alkenes, showing that the state of the reactants in
the process is important, as well as the absence of water from the system. It has also been
demonstrated that the flow system is capable of separating VOCl3 from the TiCl4 in doses both
equal to and far higher than found in industry.
As well as soya bean oil, naphthenic oils were also reacted with TiCl4 using the flowcell. These
we also found to produce the chlorinated glyceride species, as well as forming a red charge
transfer complex with the TiCl4 which was able to distil over, staining the products and also to
continue reacting with the TiCl4 after distillation, forming a black precipitate both of which
would have been disastrous were they to occur in industrial production.
119
Chapter 5
Using Mass Spectrometry to Study Gas Phase

Reactions in Chemical Vapour Deposition
5.1. Introduction
The work described in this chapter involved using a molecular beam mass spectrometer to
study the hot vapours within an atmospheric pressure CVD reactor. This was designed with the
goal of furthering the work with the flow-cell in the previous chapter by detecting and
characterising the intermediates of gas phase metal chlorides and carbonyl containing ligands,
similar to the reactions seen within this Chloride Process.
Whilst related to the work carried out studying the production of TiO2, this work is of greater
scope, contributing to the understanding of several of the most common CVD reactions studied
worldwide. Using a specialised mass spectrometer able to withstand heated corrosive gaseous
species and a custom designed mass spectrometer/ CVD adaptor apparatus the deposition of
TiO2, VO2 and SnO2 were studied from a variety of precursors.
5.1.1. Atmospheric Pressure CVD
Atmospheric pressure chemical vapour deposition (APCVD) is one of the earliest forms of
chemical vapour deposition,116 and is the building block from which many more complex
varieties arise such as atomic layer deposition (ALD)117 and plasma enhanced CVD (PECVD).118
The process involves the heating of a precursor molecule containing the metal atom for which
the film is to be comprised, often with the bonds to be present in the film pre formed.119–121
120
The vapours of these materials form a thin film on a substrate at high temperatures inside a
heated reactor vessel.
APCVD is thought to occur in 6 stages (Figure 61):122
1. Evaporation of precursors and transport of gaseous molecules in the bulk gas flow.
2. Precursor vapour enters the heated reactor leading to the formation of reactive
intermediates and by products.
3. Some of the gaseous intermediates come into contact with the surface of the heated
substrate.
4. Adsorption of precursor/reactive species onto substrate surface.
5. Reactive species diffuse across the surface to growth sites, such as imperfections in the
substrate or existing crystals of the product material. This leads to nucleation of the thin
film as the reactive species break down.
6. Gaseous by-products of the thin film formation desorb from the surface and are carried
out the reactor via the exhaust. 123
Figure 61: Schematic of basic model of film growth in APCVD122
121
APCVD has been used extensively in the deposition of a wide range of metal oxide thin films.
Films of TiO2,124–126, VO2(M),127–129, SnO2, 130–132 Fe2O3, 133 WO3, 134,135, ZnO,136–138, α-Al2O3,139,140
as well as many others. 141 Composite films of two or more of these species that combines the
advantageous properties of its components have also been fabricated. 142–144 Furthermore,
doping small amounts of metallic or main groups species into a thin film has proven highly
effective in APCVD, for instance fluorine doped tin oxide142 as a transparent conducting oxide,
N doped TiO2 as a photocatalyst 16,145 and W doped VO2 (M).146 This can either be achieved by
the introduction of a separate bubbler full of co-material/dopant, or they can be deposited
from the same vessel if the thermal conditions to generate the desired amount of precursor
vapour match closely enough.
Many of the above examples of metal oxide thin film deposition involve using metallic chloride
precursors. These are highly prized in APCVD due to their relatively low boiling point as well as
being more volatile and reactive than other transition metal species. 147 The advantages of
these precursors are however counteracted by their lack of stability in air making handling the
precursors challenging.148 Additionally, unlike with metal alkoxides, another heavily used
precursor group120,128,134,140,149 there are no preformed metal oxygen bonds. This means that
the precursor needs to be oxidised by an outside source. This can be achieved using a stream
of water,144 or air.150
Another method is to co-deposit the metal precursor with a second, oxygen-heavy organic
precursor that provides oxygen by supposedly coordinating to the metal centre and having the
excessive organic material ‘burnt off’. A common organic oxidising agent is ethyl acetate. 151
This has been used extensively in the oxidation of TiCl4,152,153 however it has also been used in
the deposition of VO2 from VCl4.154 Ethyl acetate is used in this study as the predominant
oxidising agent, due to its similarity to diethyl malonate, as the malonate themselves were
122
observed to lead to blockages.
In APCVD the concentrations of the precursors are controlled by the vapour pressures
generated. By altering the temperature of the bubbler, the amount of vapour formed is
controlled. This is determined by the Clausius-Clapeyron equation (equation 1).155
ln(P1/P2) = (ΔHvap/R)((1/T2) - (1/T1)) (1)
where:
ΔHvap= The enthalpy of vaporization of the liquid
R = The real gas content, or 8.314 J/(K × Mol).
T1 = The temperature at which the vapour pressure is known
T2 = The required vapour pressure
P1 & P2 = The vapour pressures at T1 and T2, respectively.
This equation is used to generate vapour pressure curves of the precursors in order to
determine the pressure evolved at a given temperature.156 In general the precursor vapour is
generated at roughly half of its boiling point 157 By controlling the temperature of the bubbler
the amount of each precursor reaching the reactor can be manipulated. This is especially useful
in dopant studies, wherein only a very small amount of the dopant precursor can make it to the
reactor.
APCVD offers the ability to uniformly coat large areas of substrate with a repeatability that is
hard to obtain for other CVD techniques such as aerosol assisted (AACVD). It is also highly
tuneable, with the flow rates and temperature of the bubblers controlling precursor flows
effectively.158 These factors lead to the most compelling reason to choose APCVD over other
means of depositing thin films: industrial applications. As well as simplicity and reliability the
fact that it is carried out entirely at ambient pressures means that the process involves no
123
vacuum pumps, that are expensive to procure and maintain. Cold wall and hot wall industrial
reactors for the deposition of SiO2 barrier layers onto float glass has been in existence for many
years.159
The major drawback of APCVD is the limited number of precursors available. Precursors must
be suitably volatile to generate a vapour and suitably reactive to deposit a film. This means that
for every material there are only a few options, for instance with TiO2 there are few studies
outside of TiCl4 and titanium isopropoxide. Furthermore there can be major issues with the
purity of some films,160 particularly with carbon contamination.
This study seeks to study the gas phase interactions (Figure 61), in order to determine the
intermediates that lead to the formation of films at different temperatures, by examining three
frequently deposited thin films, namely TiO2, VxOy and SnO2.
5.1.2. Using Mass Spectrometry to Study CVD
Chemical vapour deposition is a widely used means of depositing metal oxide thin films onto a
variety of substrates.161,162 Despite widespread use in the deposition of metal oxide coatings,
little is known about reactions within the gas phase during the deposition of the thin film. 122
Part of the reason for this is the challenge in taking in-situ measurement from the inside of a
chemical vapour deposition reactor, as well as the potentially damaging nature of those
vapours to analytical equipment.
Although relatively few in number, mass spectrometry studies of vapour phase deposition have
been carried out previously. Turgambaeva et al. studied the gas phase reactions of copper(I)
cyclopentadienyltriethylphosphine for the deposition of cuprous films. This was achieved by
heating a small sealed ampoule of the precursor, whilst attached to the front of a time of flight
mass spectrometer that measured the content of the evolved gas.163 Time resolved spectra
124
showed how the evolution of fragments over time as the sample was heated, similar to
thermogravimetric analysis.
Li et al. used mass spectrometry to study the deposition of silicon carbide from a tetraethyl
silane precursor using hot-wire CVD. This is a technique whereby the decomposition of the
gaseous precursor is catalysed by the heated substrate it is deposited onto, usually tungsten
wire or stainless steel.164 Measurements were obtained by placing the hot – wire apparatus
into a chamber with a time of flight (TOF) mass spectrometer. This study revealed the formation
of four membered rings of alternating silicon and carbon in the gas phase.165
Numerous mass spectrometry studies have been undertaken by Rego, Petherbridge and Tsang
et al. on the mechanism for the CVD of diamond from methane. Using a Hiden HPR-60, the
same model spectrometer as this study, the levels of CH4, C2H6, C2H4 and C2H2 and CH3 in the
gas phase were analysed. 166 Additionally the device was employed to measure the effects of
nitrogen,167 chlorine,168 and phosphine169 on the growth mechanics. Furthermore low
temperature plasmas of methane and CO2 were analysed, in order to determine the ideal
conditions for formation of the •CH3 radical, which is believed to be the major species in the
formation of diamond thin films.168
Analysis of titanium containing precursors has been carried out using mass spectrometry
previously. Rahtu et al. used a four vacuum chamber quadrupole assisted MBMS system,
sampling directly from the exhaust of a thin quartz tube hot wall reactor to examine the
formation of Ti-N-Si composite thin films from Ti(NMe2)4, NH3, and SiH4, examining the decay
of the SiH4 peak in order to construct an Arrhenius plot and in doing so propose a mechanism
for the interaction of the precursors and subsequent thin film deposition. 170
Additionally Rahtu and Matero et al. used a similar system combined with a quartz crystal mass
balance to analyse atomic layer deposition of titanium tetrachloride,171 titanium(IV)
125
isopropoxide172 and trimethyl aluminium170 with deuterated water. These studies provided
insight into the breakdown of the precursors within the gas chamber, showing how loss of
ligand is affected by the presence of water. The use of a deuterated solvent also allowed for
the differentiation of side products evolved from reaction of the precursor with the D 2O and
those evolved from the thermal decay of the precursor itself.
The main difference between the system proposed in these experiments and those discussed
above is that of versatility and scale. Most of the above studies were on heavily modified
systems, that do not resemble an unmonitored deposition, with the exception of the work on
diamond. That work whilst similar, only focused on the use of methane as a precursor, scanning
in very low mass ranges (below 100 amu). A problem with this is that spectra can be noisy in
this region. The system designed for this work seeks to emulate thin film deposition on a
laboratory scale as closely as possible, whilst still sampling continuously and safely from within
the reactor. The work also focuses on a wider range of precursors than seen in previous studies,
with multiple titanium and vanadium precursors analysed as well as additional work with tin
oxide and silicon oxide precursors carried out.
5.1.3. Mass Spectrometry
Mass spectrometry is the technique used mainly in the characterisation of the organic
materials, with uses in chemistry173 as well as widespread use in the fields of art restoration
and archaeology.174
The general principal of mass spectrometry is based around the ionization of a parent molecule,
forming a charged species, usually positive (equation 2)175
M + e- M+• + 2e- (2)
This radical cation then decays by numerous possible pathways that can be categorized into
two paths, shown in equation 3. The species either breaks down to form a radical (R) and a
126
positively charges species [EE]+ as is the case with equation 3a or the parent ion can lose a
neutrally charged species (equation 3b), with the other species remains a radical cation.176
[EE]+ + R∙ (3a)
M+•
[OE] +• + N (3b)
The sample can be ionised by a wide range of different techniques. The simplest of these is
electron ionisation (EI), as is used in this study. This involves subjecting the sample to a beam
of high energy electrons from a filament. The current applied to the filament is used to control
the energy of the electrons, which is usually set in the region of 70 eV. 175,177
Higher energy electrons lead to more fragmentation. A method of ionisation is judged ‘hard or
soft’ depending on the level of fragmentation it produces (hard being more fragmentation). EI
is considered overall to be a hard ionisation method.
Soft ionisation methods include chemical ionisation (CI), which involve creating a charged cloud
of molecules. The most common chemical ionisation agent is methane, which forms highly
unstable CH5+ and C2H5+.178 These species go on to ionise the sample. Another soft method is
electrospray ionisation (ESI), in which a solution of the material to be studied is pushed through
a thin, charged tube into a heated vacuum chamber. The solvated analyte exists as an ion in
solution. Upon entering the chamber the solvent evaporates and the now gaseous ions are
accelerated through the spectrometer. 179 These techniques are frequently used in biology in
the study of protein molecules as they allow the base structure of the macromolecule to be
maintained.
127
The charged species are separated according to their mass to charge ratio in the mass analyser
portion of the device. There are several types of mass analyser including magnetic sector,
where the charged species pass through a curved tube with a variable magnetic field applied
so that only species of a certain mass can pass through, the others colliding with the walls. 180
There are also time of flight (TOF) detectors, which separate ions by velocity.181 This method
struggles to separate ions of high mass. To counteract this most modern TOF analysers have a
reflectron, a series of high voltage ring electrodes, creating an electrostatic field at the end of
the flight tube that reflect the particles in the opposite direction, heavier molecules being able
travel further into the magnetic field before being repelled. Quadrupole spectrometers offer
advantages in terms of reproducibility and cost, however unlike TOF, quadrupoles must scan a
range of masses in succession, not all masses simultaneously. A result of this is that scans take
longer to complete the wider scan range.
In the case of the Hiden HPR-60 the spectrometer, this is fitted with a quadrupole mass
analyser. This consists of four metal rods arranged parallel in a square pattern (Figure 62) inside
a cylindrical chamber.182 These rods act as electrode with a constant DC voltage and a variable
radiofrequency (RF) voltage applied across them. Ions that pass along this path oscillate
according to this frequency however for each frequency only ions of a certain mass/ charge
ratio can pass through, with heavier or lighter species ‘spinning out’ and colliding with the
chamber walls.183 The quadrupole in the spectrometer used in this work is a triple filter
quadrupole. These devices feature a smaller quadrupole in front of and behind the main unit,
arranged in tandem (Figure 63). This is beneficial as it can increase resolution in detection of
high mass species.184 Other triple quadrupole can be used to filter ions and then break down
the filtered ions into fragments, the second quadrupole acting as a collision chamber, before
filtering again. 185
128
Figure 62: Diagram of a quadrupole mass analyser Figure 63: Schematic of the triple filter
analysers used in the Hiden HPR-60368
In this study, all mass spectrometry measurements were taken using a modified quartz tube
CVD reactor containing a heated stainless steel pipe. This pipe carried a stream of the reactive
vapour inside the system into a heated, airtight chamber attached to the front of a Hiden HPR60
molecular beam mass spectrometer (MBMS). This spectrometer has an inlet approximately
0.25 mm in diameter. This leads to three consecutive vacuum chambers of increasingly low
pressure. This focuses the gas entering the system into a beam of singular molecules that enter
the ionisation chamber. The species are then ionised using a dual source low profile electron
ionisation source, before separation by Quadrupole mass analyser. The device uses an Off-Axis
Positive and Negative Ion Pulse Counting Single Channel Electron multiplier detector with a
maximum threshold of 1000 Da. This system allows for examination of extremely low
concentrations of analyte compared to traditional spectrometers, enabling the analysis of
transition metal species that would damage the detector in larger quantities. The MBMS system
is ideal for high energy gas phase studies as upon entering the system the gaseous molecules
undergo no further reaction, not coming into contact with each other or the walls of the
detector.
129
The detector was set to positive ion RGA mode for all measurements meaning that all species
analysed are positively charged (cations or radicals). The energy of the ionizing electrons was
set to 70 eV for all experiments.
5.2. Experimental
Argon (Puresheild 99.998%) was obtained from BOC and used as supplied. Films of TiO2 were
deposited using a modified cold wall CVD reactor, onto 60 × 40 × 4 mm silicon coated float glass
substrates, from Pilkington Glass Plc. The argon carrier gas was delivered into the system using
stainless steel tubing. All of the tubing making up the rig was heated to 150 °C using heater
tape, monitored using by Pt-Rh thermocouples. The precursor was introduced into the gas
stream using stainless steel bubbler, heated to roughly half the boiling point of the precursor
using a heater jacket. Hot argon or nitrogen was passed through this bubbler, the gas stream
carrying the precursor molecules to the reactor. When used, ethyl acetate was introduced
through a second bubbler, heated to 35 °C (Figure 64). The amount of precursor was controlled
using the flow rate of argon through the bubbler. This was kept as low as possible whilst still
seeing data on the mass spec readout as to avoid dealing with unnecessary damage to the
detector.
The precursor vapour enters into the reactor, which is held at an elevated temperature,
between 300 and 600 ˚C. A stainless steel tube above the substrate carried a sample of the
reagent gas out of the reactor with the majority passing out of the rear of the reactor as
exhaust, as with regular vapour deposition. Stainless steel tube passes through a chamber
bolted to the front of the mass spectrometer, where a small amount entered the chamber
through a hole in the tube (Figure 65). Both the chamber and the tube were heated to 170 ˚C,
however due to the flow of hot air from the reactor, it is estimated the temperature inside the
tubing is far higher. A small amount of the vapour entered the spectrometer through a ¼ mm
130
hole in the front end, which then formed the molecular beam. The remainder was carried back
into the fume hood through the heated tube or a secondary exhaust built into the front
chamber (Figure 66). There is added protection for the spectrometer in the form of a weak flow
of nitrogen or argon over the spectrometer’s inlet, known as the purge flow, which can be used
to dilute the gas stream further.
In most experiments the mass spectrometer was set to continuously scan over a set range.
After collecting a background of at least 10 scans, the precursor bubbler was opened in order
to view the spectrum of the precursor only. Following this the ethyl acetate oxidising agent
bubbler was opened (if used), and the spectra of both reagents in the gas stream was observed.
Each time either of the bubblers was turned on or the flow rates altered, the mass spectrometer
would be allowed to take 10 scans if the range was between m/z- 50 - 300 and 5 scans if it was
between m/z- 50 – 500. This was decided as it allows for enough scans to identify anomalous
results, which can be the result of impurities or build-up of charge on the quadrupole, whilst
still preventing overexposure of the detector to the gas stream, which can lead to blockages
and damage.
131
Figure 64: Schematic representation of entire mass spectrometry CVD setup.
Figure 65: Annotated photograph of mass spectrometer CVD setup.
132
Figure 66: Schematic of CVD apparatus and mass spectrometer setup.
133
In the majority of experiments scanning began at 50 amu, so as to avoid sampling the carrier
gas N2 (mass = 28 amu) or argon (mass = 40 amu). Due to their extremely high concentration,
sampling these numbers for long period can age the detector extremely quickly. Furthermore
if too high a concentration of a single mass is obtained, the detector will automatically shut off
in order to prevent damage. By not sampling the highly abundant carrier gasses, the system
can be opened to allow more analyte into the system, improving the signal to noise ratio of the
spectra. For these reasons scans were only taken below 50 amu when specifically searching for
lower mass fragments.
5.2.1. Preliminary Experiments
Before using the apparatus to study thin film depositions, two proof of concept experiments in
which toluene was passed through the reactor in order to test the devices sensitivity were
carried out.
5.2.1.1. Toluene Test Run
In order to test the spectrometer’s ability to sample a mobile gas stream, toluene was heated
in a bubbler to 60 ˚C. Scans were taken between 1 and 100 amu. The bubbler was opened and
closed alternately every 10 scans, each time increasing the flow rate through the bubbler by
0.2 L min-1 from 0.2 up to 1 L min-1. The reactor was kept at 150 ˚C and the plain line flow at 6 L
min-1 over the whole experiment. Figure 67 shows the effect of the different flow rates on the
number of counts per second for the mass number corresponding to the intact toluene
molecule (91amu).
134
Toluene [M]+ peak scans (M/Z: 91 amu)
35000 1.2
Toluene flow rate L min-1

30000 1
25000
Mass: 91 Counts
0.8
20000
0.6
15000
0.4
10000
5000 0.2
0 0
20 30 40 50 60 70 80
scan number
Mass 91 Counts (toluene) Toluene Flow Rate
Figure 67: Intensity of the mass spec peak at m/z- 91 representing the molecular mass of toluene
plotted alongside the flow rate through the toluene bubbler.
This study shows that the spectrometer was able to pick up a signal of toluene across the entire
range of flow rates, as well as that the signal increases roughly proportionately with flow rate.
This is an encouraging result as toluene has a boiling point of 110 ˚C, similar to that of many
CVD precursors.
5.2.1.2. Purge Flow Test
Toluene was also used to the test the effectiveness of the purge flow system in lowering the
concentration of harmful vapour species. As before, toluene was heated in a bubbler to 60 ˚C
with the reactor kept at 150 ˚C and the plain line flow at 6 L min-1. Scans were carried out over
the range of m/z- 1 -100. Every 5 scans the purge flow rate was increased by 0.5 L min-1, from
0 – 2 L min-1. Over this range the signal was observed to fall from ~22000 to ~15000 counts, a
reduction in signal of around 30 % (Figure 68).
135
Effect Of Purge Flow Rate
25000 2.5
20000 2
Purge Flow Rate Lmin-1

Mass 91 counts
15000 1.5
10000 1
5000 0.5
0 0
110 115 120 125 130 135 140 145
Scan number
Mass 91 (toluene) counts Purge Flow Rate
Figure 68: Intensity of the mass spec peak at m/z- 91 representing the molecular mass of toluene
plotted alongside the flow rate of the purge flow.
5.2.2. Precursor Deposition
Table 14 shows the experiments carried out using the CVD mass spectrometry apparatus that
successfully produced spectra of the precursors. Initially the reaction of titanium and vanadium
oxytrichloride with ethyl acetate was studied. Ethyl acetate was chosen as it is a common
oxidising agent in CVD, with a low boiling point allowing it to be easily manipulated. Attempts
were made to use diethyl malonate to which would have been a better fit with the
crystallography study, however blockages within the tubing were found to form quickly upon
opening the bubbler in both cases. Ethyl acetate can be thought of as half of a diethyl malonate
molecule. Although lacking the chelating properties of diethyl malonate, the increased
reactivity of such was the likely cause of the blockages.
136
Following the study with TiCl4 and VOCl3, titanium(IV) isopropoxide (TTIP) were studied, along
with butyltin trichloride, a common SnO2 precursor.
Finally a study was carried out investigating the reactivity of VCl4 with ethyl acetate. This
common vanadium oxide precursor has been found to be extremely sensitive to the presence
of an oxidising agent, with the formation of thermochromic VO2 occurring with roughly equal
mass flow rates of the precursor and ethyl acetate. Increasing the flow of ethyl acetate has
been found to result in growth of V2O5 films. Using mass spectrometry, the different gas phase
species forming as the ratio of ethyl acetate to VCl4 changes has been analysed in order to
determine a possible the cause of this sensitivity.
137
Table 14
Table 14: List of precursors and conditions used in the mass spectrometry analysis of CVD experiment.
Number Precursor Reactor Scan Bubbler Bubbler Plain Ethyl Observations
Temp (˚C) Range Temp Flow Line acetate
(˚C) rate (L Flow bubbler
min ⁻¹) Rate (L flow rate (L
min ⁻¹) min ⁻¹)
1 Ethyl Acetate 300, 400, 1-100 35 0.4 6 n/a Strong peaks visible immediately. No film deposited no
500 +600 discernible difference between spectra at different
temperatures
2 Titanium 300 + 600 50 - 300 60 0.4 2 0 Spectrum of TiCl4 fully visible
tetrachloride
3 Titanium 300 + 600 50 - 500 60 0.4 2 0.4 At 600 ˚C Coordination peaks of ethyl acetate to TiCl4 visible.
tetrachloride Only TiCl4 and ethyl acetate peaks seen at 300 ˚C
4 Vanadium 300 + 600 50 - 300 60 0.4 2 0 Spectrum of VOCl3 fully visible
Oxytrichloride
5 Vanadium 300 + 600 50 - 300 60 0.4 2 0.4 Coordination peaks of ethyl acetate to vanadium oxytrichloride
Oxytrichloride visible at 300 ˚C and 600 ˚C
6 Titanium(IV) 300 + 600 50 - 300 110 0.2 6 0 Major issues with blockages above 300 ˚C, little difference
isopropoxide between temperatures
7 Butyltin 300 + 600 50 - 500 120 0.4 6 0 Spectrum of BuSnCl3 fully visible
Trichloride
8 Butyltin 300 + 600 50 - 500 120 0.4 6 0.4 Coordination peaks of ethyl acetate to BuSnCl3 visible at 300 ˚C
Trichloride and 600 ˚C
9 Vanadium 300 + 600 50 - 500 35 0.4 2 0 Spectrum of VCl4 fully visible
Tetrachloride
10 Vanadium 550 50 - 500 35 0.4 2 0.05 Spectrum of VCl4 fully visible. Small peaks match those of VOCl3
Tetrachloride
11 Vanadium 550 50 - 500 35 0.4 2 0.2 Spectrum of VCl4 fully visible. Small peaks match those of VOCl3
Tetrachloride stronger than in 10
12 Vanadium 550 50 - 500 35 0.4 2 0.6 Spectrum of VCl4 fully visible. VOCl3 peaks now more intense
Tetrachloride that VCl4 peaks of similar mass/charge ratio. Intermediate VCl4 +
ethyl acetate complex observed
138
5.2.2.1. Ethyl Acetate
Ethyl acetate (anhydrous 99.8%, Sigma Aldrich) was placed in a stainless steel bubbler equipped
with a heater jacket and a thermocouple. This bubbler was connected to the CVD system. The
bubbler was heated to 35 ˚C whilst the mass spectrometer scanned the range of 1-100 mass
numbers. The bubbler was opened and a flow of nitrogen (0.6 L min-1) transported the gaseous
ethyl acetate to the plain line through an 80 ˚C 10 mm stainless steel tube. Once inside the plain
line the vapour joined a 6 L min-1 flow of nitrogen heated to 150 ˚C, carrying it into the reactor.
From the reactor the sample was transported to the mass spectrometer via the 20 mm
sampling tube and analysed, with the excess vapour transported back into the fume hood
through the exhaust tubes. Following the opening of the bubbler, 10 scans were taken of the
reagent gas before closing the bubbler and increasing the reactor temperature by 100 ˚C. This
was repeated 4 times, from 300 up to 600˚C.
5.2.2.2. Titanium Tetrachloride
Titanium tetrachloride (anhydrous 99%, Sigma Aldrich) was placed in a stainless steel bubbler
(B1) equipped with a heater jacket and a thermocouple that was then connected to the CVD
apparatus. The bubbler was heated to 60 ˚C. Ethyl acetate (anhydrous 99.8%, Sigma Aldrich)
was added to a second identical bubbler (B2), attached in the same way as the first and heated
to 35 ˚C. The mass spectrometer scanned the range of 50 to 500 mass numbers. The TiCl4
bubbler was opened and a flow of nitrogen (0.4 L min-1) transported the gaseous TiCl4 to the
plain line through a 150 ˚C, 10 mm stainless steel tube. Once inside the plain line the gaseous
TiCl4 joined a 2 L min-1 flow of nitrogen heated to 150 ˚C, carrying the precursor into the reactor,
which was heated to 600 ˚C. A sample of the atmosphere within the reactor was transported
to the mass spectrometer via the 20 mm sampling tube and analysed, with the excess vapour
transported back into the fume hood through the exhaust tubes. Five scans were taken of the
139
TiCl4 vapour before B2 was opened as well, releasing a stream of ethyl acetate carried by a 0.4
L min-1 stream of N2 into the plain line, where it mixed with the TiCl4 before entering the reactor.
Five scans were obtained of the TiCl4 / ethyl acetate mixture before both bubblers were closed.
The reactor temperature was lowered to 300 ˚C and the atmosphere in the detector allowed
to return to its previous state before the process was repeated. Only five scans were obtained
for each set of conditions as the highly reactive nature of the TiCl4 lead to rapid oxidation to
solid TiO2. This lead to the formation of blockages, which would then clog the mass
spectrometer inlet, preventing further scans.
5.2.2.3. Vanadium Oxytrichloride
Vanadium oxytrichloride (99%, Sigma Aldrich) was placed into bubbler B1 and attached to the
CVD apparatus. The bubbler was heated to 60 ˚C. Ethyl acetate (anhydrous 99.8%, Sigma
Aldrich) was added to B2, attached and heated to 35 ˚C. The reactor vessel was heated to 600
˚C. The mass spectrometer scanned the range of 50 to 500 mass numbers. B1 was opened first,
allowing a stream of gaseous VOCl3, to be carried through the 150 ˚C, 10 mm stainless steel
tubing with a flow of N2 (0.4 L min-1) towards the plain line, joining a 2 L min-1 nitrogen gas
stream at 150 ˚C. The precursor passed into the reactor where a sample of the atmosphere
within was transported to the mass spectrometer via the heated 20 mm sampling tube. The
excess reagent gas passed back into the fume hood through the exhaust tubes. VOCl 3 was
judged to be far slower to block the spectrometer than TiCl4 allowing for 10 scans to be
obtained. Following this, ethyl acetate was released into the gas stream from B2 (heated to 35
˚C , with a gas flow of 0.4 L min-1). Ten more scans were taken before both bubblers were shut
off. The reactor temperature was lowered to 300 ˚C and the signals given by the previous
experiment observed to dissipate until the spectrum resembled that of the carrier gas only. At
140
this point the process was repeated, capturing 10 scans of VOCl3 and VOCl3 / ethyl acetate at
the lower temperature.
5.2.2.4. Titanium(IV) Isopropoxide
Titanium(IV) isopropoxide (TTIP) (97%, Sigma Aldrich) was placed in B1 and connected to the
CVD system at which point it was heated to 110 ˚C. The reactor temperature was set to 300 ˚C
and the mass spectrometer scanned the range of 50 to 300 mass numbers. B1 was opened and
a 0.4 L min-1 stream of nitrogen carried the gaseous titanium(IV) isopropoxide through stainless
steel tubing heated to 150 ˚C to the plain line, heated to the same temperature with a 6 L min -
1 flowing towards the reactor. A sample of the atmosphere within the reactor was transported
to the mass spectrometer via the 20 mm sampling tube and analysed, with the excess vapour
transported back into the fume hood through the exhaust tubes. Five scans were taken of the
TTIP vapour before B1 was closed and the reactor temperature increased by 300 ˚C to 600 ˚C.
The process was repeated with five scans of TTIP vapour collected at 600 ˚C.
5.2.2.5. Butyltin Trichloride
Butyltin trichloride (95%, Sigma Aldrich) was placed into bubbler B1 and the vessel attached to
the line of the CVD system, sealing it shut. The bubbler was heated to 120 ˚C. Ethyl acetate
(anhydrous 99.8%, Sigma Aldrich) was added to B2, attached and heated to 35 ˚C. The reactor
vessel was heated to 600 ˚C. The mass spectrometer was set to scan the range of 50 to 500
mass numbers. B1 was opened first, allowing a stream of gaseous BuSnCl3 , to be pass through
the 150 ˚C, 10 mm stainless steel tubing with a flow of N2 (0.4 L min-1) towards the plain line,
mixing with the 2 L min-1 plain line flow of nitrogen at 150 ˚C. The precursor passed into the
reactor where a sample of the atmosphere within was transported to the mass spectrometer
via the 20 mm stainless steel sampling tube, heated to 150 ˚C. The excess reagent gas passed
back into the fume hood through the exhaust tubes. Unlike the titanium precursors, BuSnCl 3
141
was found not to block the detector at all, allowing for more numerous scans over a longer
range. After ten scans, ethyl acetate was released into the gas stream from B2 as described in
the method for the TiCl4 and VOCl3 depositions. 10 more scans were taken before both
bubblers were shut off. The reactor temperature was lowered to 300 ˚C and the signals given
by the previous experiment observed to dissipate until the spectrum resembled that of the
carrier gas only. At this point the process was repeated.
5.2.2.6. Vanadium(IV) Chloride
The study of VCl4 consisted of four different depositions. In each, VCl4 was placed into bubbler
B1, connected to the APCVD system and heated to 35 ˚C. A 0.6 L min-1 flow of N2 pushed the
VCl4 through the 10 mm stainless steel tubing into the plain line, which were heated to 100 and
150 ˚C respectively. The gaseous VCl4 flowed into the reactor which was set to 550 ˚C from
which some of the atmosphere within passed along the 20 mm stainless steel sampling tube
into the mass spectrometer front chamber. This atmosphere was sampled by the mass
spectrometer and the remaining gas passed harmlessly back into the fume hood through the
exhaust pipes.
This process was repeated three times, between which the reactor was switched off and the
substrate replaced. For the subsequent scans, in addition to the VCl4 vapour stream, a stream
of ethyl acetate from B2, heated at 30 ˚C was also used. The flow of nitrogen through B2 was
varied between each deposition, altering the mass flow of ethyl acetate. The first of these three
depositions used a flow of 0.6 L min-1 ethyl acetate, roughly 2 x the mass flow of VCl4 the second
and third depositions used flows of 0.2 L min -1 and 0.05 L min -1, roughly equal and half the
mass flow of VCl4, respectively. As with the first experiment five scans were obtained at 550 ˚C,
with both the VCl4 and ethyl acetate bubblers open.
142
A Hiden HPR-60 quadrupole mass spectrometer was connected to a CVD reactor in order to
sample the gaseous species within during atmospheric pressure vapour deposition reactions.
Depositions using TiCl4, VOCl3, VCl4, SnBuCl3, TTIP and were carried out, and the atmosphere
within the reactor sampled at different temperatures and with different levels of the oxidising
agent ethyl acetate in the atmosphere to determine the gas phase intermediates that form
during high temperature deposition reactions.
5.3.1. Ethyl Acetate
Initial depositions used ethyl acetate without a metal oxide precursor. This was in order to
establish a ‘baseline’ as ethyl acetate’s role as an oxygen donor meant that it would be present
in the spectra of all further depositions. Furthermore, due to ethyl acetate’s interesting
fragmentation pattern, its mass spectrum is well studied and provided a good means of
establishing the effectiveness of the mass spectrometer at picking up low mass signals
spectrometer.
Ethyl acetate has a simple but telling set of peaks at m/z 88(M+), 70, 61, 45, 31 and 18. All of
these are clearly visible in the spectra obtained using the CVD MS apparatus (Figure 69a). This
combined with its interesting fragmentation pattern sees ethyl acetate frequently used as an
example in mass spectrometry training courses. Rather than simple fragmentation, ethyl
acetate undergoes condensation of water resulting in the loss of the ethyl oxygen and
subsequent rearrangement to form a methyl vinyl ketone species. The formation of this is not
fully understood.
143
Figure 69: Assigned mass spectrum of ethyl acetate as passed through an atmospheric pressure
chemical vapour deposition reactor at 600 ˚C (a) and comparison of ethyl acetate fragmentation
patterns over the temperature range of 300 to 600 ˚C (b).
144
The spectrum of the ethyl acetate was observed to be very stable to temperature. The spectra
remained identical with variation in reactor temperature from 300 to 600 ˚C in 100 ˚C
increments (Figure 69b).
5.3.2. Titanium Tetrachloride
5.3.2.1. Mass Spectra Analysis
In the first instance titanium chloride was passed through the reactor without the presence of
ethyl acetate at 600 ˚C. The spectrum recorded matches well to that taken for TiCl4 first taken
by Hogg in 1954 that features in the NIST Webbook.186
The four major peaks at m/z 189, 154, 118, and 83 accounts for titanium(IV) chloride having
lost 0, 1, 2 and 3 chlorine atoms respectively. In each case the isotope pattern for the correct
number of chlorines was observed, with the peaks representing the presence of one and two
mass 37 chlorine atoms increasing in size as more chlorine is present (Figure 70).
An additional peak was visible at m/z 225, which represents a titanium atom with five chlorine
ligands present. This is most likely caused by the extremely high energies within the ionisation
chamber, giving discarded chlorine radicals enough energy to react with the parent molecule,
allowing for the formation of the highly energetically unfavourable TiCl5. Although surprising,
side reactions involving fragments and the parent ion are relatively common in mass
spectrometry, (although usually in high energy ionisation sources) and are not relevant to this
study.
145
Figure 70: Mass spectrum of titanium(IV) chloride as passed through an atmospheric pressure
chemical vapour deposition reactor at a temperature of 600 ˚C.
When the experiment was repeated with a flow of ethyl acetate joining that of TiCl4, the fumes
passing out of the reactor, as well as the two exhaust tubes of the mass spectrometer chamber,
were observed to be far thicker and denser. The inlet to the spectrometer was found to block
more rapidly as well, demonstrating that the presence of the oxygen rich organic species
facilitates TiO2 formation. The spectra collected (shown in Figure 71), contained a number of
new peaks, representing intermediates formed via coordination of the ethyl acetate to the
TiCl4.
146
Figure 71: Mass spectrum of titanium(IV) chloride and ethyl acetate as passed through an atmospheric
pressure chemical vapour deposition reactor at a temperature of 600 ˚C, with an enhanced image of
the peaks between m/z 200 and 330, showing the newly formed adduct peaks.
Spectra collected for the same experiment at 300 ˚C do not show these peaks, suggesting that
the formation of these intermediate species is conducive to the formation of TiO2 films.
147
The peak at m/z 314 possesses the isotope pattern of a trichloride, and is m/z- 159 heavier than
the TiCl3 ion. This suggests that the species present is likely to be a TiCl3 centre complete with
two ethyl acetate molecules coordinated to the titanium centre, one of which is missing a
methyl group, a common fragmentation pathway for the molecule.
This suggests that the TiCl4 is able to coordinate to one ethyl acetate without loss of chlorine,
with subsequent coordination by ethyl acetate molecules leading to displacement of a chloride
ion in the gas phase. It is however plausible that the tetrachlorotitanium bis-ethylacetate
species exists, but is simply too unstable to make it to the spectrometer intact. The remaining
peaks at m/z 279, 244 and 209 represent the same species following the loss of a second third
and fourth chlorine, respectively, as confirmed by the clear isotope patterns. Their presence as
single charge cations demonstrates that this loss of chlorine is taking place at the ionisation
source following formation of a radical cation parent ion, as with the TiCl4 molecule.
5.3.2.2. Thin Film Analysis
The film deposited from TiCl4 and ethyl acetate at 600 ˚C appeared largely transparent,
displaying the birefringence common with titanium dioxide films. A dark, hazy patch formed
immediately before the position of the gas sampling tube, which is likely the result of the
obstruction leading to a build-up of product in that area. The deposition at 300 ˚C gave no film,
instead yielding a very thin grey powder, easily removed from the substrate.
i) X-ray diffraction
The X-ray diffraction pattern of the film deposited at 600 ˚C from TiCl4 and ethyl acetate
revealed that the film consisted of TiO2 in the anatase phase (Figure 72). X-ray diffraction of the
material deposited at 300 ˚C gave a largely amorphous pattern, several small signals suggesting
the faint presence of anatase TiO2.
148
Figure 72: X-ray diffraction pattern of the films deposited from TiCl4 and ethyl acetate onto float glass
at 300 and 600 ˚C using the APCVD / mass spectrometry apparatus, alongside an anatase TiO2
standard.187
ii) Scanning Electron Microscopy
SEM images were obtained for the films deposited from TiCl4 at 600 and 300 ˚C. The film
deposited at 600 ˚C revealed a dense microstructure consisting of a mixture of thin plates
roughly 200 nm in diameter and protruding rods of approximately 50 nm in diameter (Figure
73). Imaging of the powdery substance deposited at 300 ˚C shows a substrate sparsely
populated with roughly square particulates of around 100 nm in diameter.
149
Figure 73: SEM image of the film deposited from TiCl4 and ethyl acetate onto float glass at 600 ˚C using
the APCVD / mass spectrometry apparatus.
SEM and XRD analysis confirms the formation of TiO2 thin films at elevated temperatures when
TiCl4 was deposited with an oxidising agent, as is well documented in the literature. 151 The
pronounced difference in the mass spectra taken during depositions at 300 ˚C and 600 ˚C, as
well as the resulting films, is strong evidence that the formation of the TiO2 on the substrate
surfaces relates to the formation of the gas phase [TiCl3(CH3COOC2H5)2] intermediate observed
in the mass spectrum of the reactor atmosphere.
5.3.3. Vanadium Oxytrichloride
5.3.3.1. Mass Spectrometry Analysis
Scanning the atmosphere when only VOCl3 was present in the gas stream alongside the carrier
gas shows that the precursor predominantly makes it through to the detector intact (Figure
74). The peaks at m/z 172, 137, 102 and 67 represent the intact parent ion and the loss of one,
150
two and three chlorine ligands respectively. This is confirmed by the very clear isotopic patterns
of the chlorine for all three peaks. The signals at m/z 102 and 86 represent the parent molecule
minus the double bound oxygen, as well as the loss of one and two chlorines,respectively. These
species appear considerably less abundant than that of those only missing chlorine, likely due
to the energy required to break the V=O double bond.
Figure 74: Mass spectrum of vanadium(V) oxytrichloride as passed through an atmospheric pressure
Similarly to TiCl4, when an equivalent amount of ethyl acetate was run through the reactor
alongside the VOCl3 stream at the same temperature, new peaks at a mass greater than that of
VOCl3 were observed relating to the products of gas phase reactions (Figure 75).
151
Figure 75: Mass spectrum of vanadium(V) oxytrichloride and ethyl acetate as passed through an
atmospheric pressure chemical vapour deposition reactor at a temperature of 600 ˚C, with an
enhanced image of the peaks between m/z- 200 - 330, showing the newly formed adduct peaks.
As with the TiCl4 experiment, the new peaks relate to the VOCl3 coordinating to the ethyl
acetate in the gas phase. In this case only two peaks were visible, both of which were the result
of ethyl acetate coordinating to the vanadium, presumably via the carbonyl oxygen, displacing
a chlorine atom on the vanadium. This has occurred twice in the species represented by the
m/z 278 peak and in the species represented by the peak at m/z - 225. This is confirmed by the
isotope patterns of the chlorine in both cases.
This presents an interesting difference in the reactivity of the TiCl4 and VOCl3, previously alluded
to in this thesis. In the case of TiCl4, the peak at m/z- 314 represents the TiCl4 having coordinated
to two ethyl acetate molecules with the loss of one chloride. It can be inferred from this that,
although perhaps unstable, the TiCl4 is capable of coordinating to a single ethyl acetate without
152
the loss of any chlorine. In the case of VOCl3 in the case of each new species involves
coordination to an ethyl acetate molecule comes at the expense of a chlorine ligand.
This pattern relates closely to that seen with the diester species reacting with TiCl4 and VOCl3
where the coordination of a diester ligand to the titanium saw no loss of chlorine, making it far
more reversible than the same reaction with VOCl3 which would always result in the
displacement of chlorine.
Furthermore, unlike TiCl4, the formation of the intermediate species with VOCl3 was visible at
both 300 ˚C and 600 ˚C, with the ratio of intermediates to unreacted VOCl3 in fact higher at 300
˚C. This suggests that gas phase reactions of the ethyl acetate and VOCl3 either possess a lower
activation energy barrier than the equivalent reaction with TiCl4, or have greater stability ty in
the gas phase and able to make it to the detector at the lower temperature, rather than simply
dissociate, both of which fit well with the chemistry observed in the chloride process.
Both depositions at 600 ˚C and 300 ˚C yielded films. The film deposited at 600 ˚C was completely
opaque, with a matte dark blue/black appearance and complete coverage of the substrate. The
film deposited at 300 ˚C was lighter in colour with translucent brown areas around the edges
likely due to the temperature being lower, leading to incomplete oxidation of the precursor to
V2O5. The film appeared only on the back half of the substrate. The films were both very thick,
likely the result of the very long deposition time taken to obtain 10 scans.
i) X-Ray Diffraction
XRD patterns were obtained for both the films deposited at 300 and 600 ˚C (Figure 76). Both
films appear to be predominantly VO2, however the different temperatures appear to result in
the formation of different polymorphs. This is an interesting development as the vanadium is
entering the reactor in the 5+ oxidation state, implying that the ethyl aceate, whilst providing
153
Figure 76: X-ray diffraction pattern of the films deposited from VOCl3 and ethyl acetate onto float glass
at 600 ˚C (top) and 300 ˚C (bottom) using the APCVD / mass spectrometry apparatus, alongside
standards for monoclinic (PDF 72-0514)188 and tetragonal (PDF 80-0690) VO2.189
oxygen is not an oxidising agent. It is plausible that it could be attributable to the donation of
electron into the d empty shell by the dative coordination of the acetate, or it could be a more
complex chain or reactions occurring at the surface of the substrate.
At 600 ˚C, the monoclinic polymorph appear to be the dominant phase, with strong reflections
of the (001), (200) and (211) planes. Although there is less work done on the use of VOCl 3 as a
precursor for the deposition of VO2 than for VCl4, due to the vanadium being in the +5 oxidation
state, this is in keeping with literature findings of depositions with VOCl3 and water.190 At the
lower temperature of 300 ˚C, the dominant phase appears to be the tetragonal polymorph of
VO2. Although both patterns are fairly broadened, the lower temperature film gave far weaker
reflections. This is reflected in SEM images taken of the film (vide infra), in which the higher
temperature films appear to have a far more crystalline morphology.
154
Scanning electron micrographs of the films reveal a striking difference in the morphologies of
the deposited films with deposition temperature. At both temperatures, the films were thick
and dark in appearance due to the length of depositions time. This is reflected in the
morphologies with both films featuring densely packed particles.
At both temperatures, the films consist of elongated particles of between 1 and 2 μm in length
with no preferred orientation (Figure 77a). At 600 ˚C the film consists of fairly uniform, rod-like
structures roughly 100 nm wide. These crystallites have flattened edges, indicating preferred
growth of certain faces. Formation of vanadium oxide nanorods has been observed in several
cases, including previously with APCVD of VOCl3. 191–193 The particles formed at 300 ˚C were
considerably wider and less uniform with rounded edges and diameters ranging from 200 to
600 nm (Figure 77b).
155
Figure 77: SEM images of the film deposited from VOCl3 and ethyl acetate onto float glass at 300 (a)
and 600 ˚C (b) using the APCVD / mass spectrometry apparatus.
156
5.3.4. Butyltin Trichloride
Unlike the previous two species, the presence of the butyl group on the butyltin trichloride
makes the spectrum far more complex. Furthermore, isotope patterns of the chlorine ligands
are far more unclear due to the overlapping of peaks from which one or two protons are lost
creating complex multiplets (Figure 78). Nevertheless as there is only one alkyl group present
the fragments are relatively facile to deduce.
Figure 78: Mass spectrum of butyltin trichloride as passed through an atmospheric pressure chemical
vapour deposition reactor at a temperature of 600 ˚C.
157
The parent ion peak was not seen in this spectrum. Instead the highest molecular weight peaks
has a signal at of m/z 252, which relates to the parent ion having lost a [CH3CH2]+ from the butyl
species.
This peak overlaps with another set of peaks at around m/z 245, resulting from the parent ion
having lost a chlorine ligand. The peaks at m/z 210 and 177 (extremely faint) represent the
subsequent loss of two more chlorine atoms. Of these three species the only one with a
relatively clear isotope pattern is that at m/z 210. Peaks at m/z 225, 190 and 155 likewise
represent the parent species having lost the butyl group and subsequently losing a chlorine
ligand, giving [SnCl3]+∙, [SnCl2]+∙ and [SnCl]+∙ species respectively. These are presumably radical
species as otherwise the multiple charges would cause them to appear at far lower mass
numbers.
When gaseous ethyl acetate was flowed through the reactor chamber alongside the BuSnCl3, A
host of new peaks, formed from reaction of the two species were visible in both the scans
collected at 300 and 600 ˚C (Figure 79). Similarly to the titanium chloride, the highest mass peak
at m/z 423 relates to that of BuSnCl3, having coordinated to two ethyl acetate ligands, leading
to the loss of a single chloride. The peak at m/z 335 represents the same molecule following
the loss of one of those ethyl acetate molecules.
The ethyl acetate also appears to be able to displace the butyl group, with an equilvalent pair
of peaks at m/z 401 and 313, generated by the coordination of two ethyl acetate ligands and
the loss of a butyl group and the subsequent loss of a one ethyl acetate in the ion source, as
well as a peak at m/z 366 relating to the loss of one chlorine and one butyl group following the
coordination of two ethyl acetate groups.
158
Figure 79: Mass spectrum of Butyltin trichloride and ethyl acetate as passed through an atmospheric
pressure chemical vapour deposition reactor at a temperature of 600 ˚C and 300 ˚C, with an enhanced
image of the peaks between m/z- 250 and 500, showing the newly formed adduct peaks.
The fact that these adduct peaks appear in both the low and high temperature runs is hard to
explain as BuSnCl3 will only deposit tin oxide at above 400 ˚C,194 implying that the formation of
the intermediates is not enough to guarantee film deposition. It should be noted that all peaks
159
relating to unreacted SnBuCl3 have greater relative intensity in the depositions at 300 ˚C,
whereas the products of the ethyl acetate and SnBuCl3 have a higher intensity in the 600 ˚C
deposition, suggesting that a greater proportion of the SnBuCl3 is reacting with the ethyl
acetate at higher temperatures. It is possible to consider that at 300 ˚C the intermediates are
capable of forming; however this does not mean that they have enough energy to fully oxidise.
This could explain why despite most of the intermediate peaks in the spectrum taken at 300
˚C are roughly half the relative intensity of those taken at 600 ˚C, the peaks at m/z 401 and 423,
representing the fully intact intermediates, are the same intensity as those in the high
temperature and low temperature spectrum. This increase in relative height could be the result
of a failure to react in the gas phase inside the reactor, leading to a greater concentration on
the ensuing mass spectrum. This experiment was repeated, with the flow rate of ethyl acetate
lowered from 0.6 L min-1 to 0.2 L min-1, resulting in an excess of BuSnCl3 entering the reactor.
This yielded the same result, with intermediates visible in both the higher and lower
temperature spectra.
5.3.4.2. Thin film Analysis
The depositions with butyltin trichloride and ethyl acetate at 600 ˚C yielded a colourless
transparent thin film, with slight haze just before the entrance to the sampling tube.
Depositions at 300 ˚C appeared to yield no visible deposition onto the substrate.
i) X-ray Diffraction Analysis
The XRD pattern of the film deposited from 0.6 L min-1 BuSnCl3 and 0.6 L min-1 ethyl acetate at
600 ˚C are shown in Figure 80. The XRD patterns of the films confirm the deposition of rutile
SnO2, as compared to a standard. The peak for the (101) miller plane is present although that
plane appears heavily disfavoured by the particles making up the film. The deposition under
160
the same conditions at 300 ˚C shows no visible peaks, suggesting that had a film formed it was
highly amorphous.
Figure 80: X-ray diffraction pattern of the film deposited from BuSnCl3 and ethyl acetate onto float
glass at 600 ˚C using the APCVD / mass spectrometry apparatus, alongside a rutile SnO2 standard. 195
SEM imaging of the film deposited at 600 ˚C shows a leaf-like microstructure consisting of
flattened, pointed particles, packed tightly together on the substrate (Figure 81). This is a
similar morphology to fluorine doped tin oxide reported in the literature, as grown from
BuSnCl3 from atmospheric pressure and aerosol assisted CVD.196,197
The substrate from the equivalent experiment at 300 ˚C was also imaged by SEM, revealing a
sparse scattering of sphere like particulates of what is likely amorphous SnO2.
161
Figure 81: SEM image of the film deposited from TiCl4 and ethyl acetate onto float glass at 600 ˚C using
the APCVD / mass spectrometry apparatus.
5.3.5. Titanium Isopropoxide
Depositions were carried out using titanium isopropoxide (TTIP) at 600 ˚C and 300 ˚C, both
yielding thin films. The film deposited at 600 ˚C was far thicker and more adherent, appearing
transparent with a circular area of haziness at the point of the substrate positioned in front of
the sampling tube. This film passed the Scotch tape test and was observed to be fairly resistant
to surface abrasion.
The film deposited at 300 ˚C was far more powdery and easily wiped from the surface by
touching. The opaque film was white in colour as is customary with TiO2.
Ethyl acetate was not added as an oxygen source in these experiments as it is not commonly
used in conjunction with TTIP depositions in the literature, as the precursor molecule already
contains four Ti-O bonds, making oxygen plentiful enough to deposit TiO2 without an additional
162
oxygen source. This was observed in the depositions, with TTIP depositions blocking the
machine far more rapidly than any other precursor.
The deposition carried out at 300 ˚C gave the expected spectra of TTIP, with no additional gas
phase interactions visible (Figure 82). In this case the TiO2 forming is simply the result of the
alkyl side groups burning off to leave TiO2. (Scheme 9)
Scheme 9: Schematic of the decomposition of TTIP at high temperatures to give TiO2. 198
Figure 82: Mass spectrum of titanium isopropoxide as passed through an atmospheric pressure
Despite being far more stable than TiCl4 and VOCl3 the molecular ion peak was extremely weak,
instead the dominant peak of the spectrum was the one corresponding to the loss of a single
methyl group from one of the isopropoxide ligands. This is a phenomenon commonly seen in
163
the ionisation chambers of EI mass spectrometers,199 suggesting that the spectrum seen here
is a very close approximation to putting TTIP through an EI instrument directly. This spectrum
is similar to one taken from an atomic layer deposition reactor at 250 ˚C, the major difference
being the signal at m/z 247 which could not be assigned to a fragmentation peak of the TTIP
molecular ion. The following fragmentation is complex although the major peaks have been
assigned based on the idea that there are three bond at which fragmentation could occur: the
Ti-O bond, the O-C bond and the C-C bonds, resulting in the loss of a methyl (mass = 15 amu),
propyl (mass = 43 amu ) or isopropyl (mass = 59 amu) group respectively (Figure 83).
Figure 83: Possible fragmentation points on the TTIP parent ion with the formulas and masses of the
fragments denoted in red
The deposition at 600 ˚C gave an extremely different pattern to that observed at 300 ˚C. In this
case a totally new set of peaks, many above the mass of the parent ion were observed (Figure
84). The initial assumption when observing the repeating pattern of 60 amu decreases in the
fragment masses, was that the product was simply a polymer consisting of removed
isopropoxide ligands potentially polypropylene glycol.
164
However the difference in mass between the fragments would then be m/z 58, as a proton
would have to be lost from either end of the monomer.200 The formation of an oxo-bridged
dimer species in titanium isopropoxide has been proposed as a possible mechanism for the
formation of nanoparticles via the solvothermal synthesis of TiO2 from TTIP and
benzylamine.201
It is thought that the high temperature leads to the loss of the propyl group leaving an unstable
(OiPr)3TiOH. In this case it appears as though a dioxo bridge has formed between the two
titanium centres (Figure 85) with electron density shared over the two oxygen atoms forming
a four membered ring.
165
Figure 85: Proposed formation of an oxo-bridge between two titanium isopropoxide molecules at high
temperatures.201
The formation of the two oxo bridges without the loss of the isopropyl ligands would be highly
disfavoured due to the maximum oxidation state of titanium being 4+, therefore this is likely
not the whole picture. Two bridging oxygen atoms could potentially be hydroxyl ligands
themselves, both of which could easily lose a proton in the mass spectrometer ionisation
chamber. The mass of this species would be 228, which is a peak present in the spectrum taken
at 300, however it is masked by a cluster of peaks bringing another fragment molecule so it is
unknown if it relates to this species also. This would explain how such a species would be stable.
A small peak at m/z 484 could explain this, however it is too weak to be seen as proof of this
species forming. If this is indeed the case, deprotonation of the bridging species could be a
viable path for the loss of the isopropanol fragments.
Following the formation of the dimer species, fragments are lost of m/z - 60 . As with the
polymer this is not simply a case of heterolytic fission causing loss of the ligand, as would occur
in an ionisation chamber, as the fragments would have m/z - 59. Whilst 1 mass unit below the
expected value is common due to subsequent proton loss, one heavier is harder to explain. The
isopropoxide ligand group must be abstracting a proton as it leaves, forming isopropanol. This
is corroborated by the presence of a large peak at m/z 60 in the spectrum, however it cannot
be ruled out that the ligand would coordinate to a proton later.
The most likely pathway for the loss of isopropanol from the ligand is via the abstraction of a
proton from another isopropyl ligand (Figure 86). The central proton would be the most likely
166
provider of a proton due to the tertiary nature of the carbon providing greater stability for the
negative charge was well a being adjacent to the oxygen. The oxygen from one isopropoxide
abstracts the central proton from the adjacent ligand, resulting in the formation of an acetone
group which remains coordinated to the metal centre.
Figure 86: Proposed mechanism for the loss of the elimination of an isopropanol molecule from TTIP
and its dimer at high temperatures.
This reaction can be assumed to happen into the gas phase in the CVD reactor, as these peaks
were not observed during the deposition at 300 ˚C. Further reactivity occurs in the loss of more
isopropanol molecules in this way, with the dimer eventually splitting apart. The lowest main
fragment featuring titanium involves coordination to an acetone molecule and an oxygen atom.
Should the dimer species be forming inside the CVD reactor it is likely a precursor for the
deposition of crystalline TiO2 at higher temperatures.
i) X-ray Diffraction Analysis
As with the films deposited from TiCl4 and ethyl acetate the film deposited at 600 ˚C was
considerably more crystalline than the film deposited at 300 ˚C (Figure 87). In both cases, the
film appears to be anatase phase TiO2, with the (200), (211) and (204) miller planes present in
167
both films. The usually dominant (101) plane however appears to be disfavoured at the lower
temperatures.
(10
160
1)
140
TTIP 600˚C
120
TTIP 300 ˚C
intesity / a.u.
100
anatase TiO₂ standard
80
(211
(200 )
60
) (204
)
40
20
10 20 30 40 50 60
2θ/˚
Figure 87: X-ray diffraction pattern of the films deposited from TiCl4 and ethyl acetate onto float glass
at 300 and 600 ˚C using the APCVD / mass spectrometry apparatus, alongside an anatase TiO2
standard.187
SEM micrographs were taken of the films deposited from TTIP at 300 and 600 ˚C (Figure 88). At
300 ˚C the film appears to display more pores and a disordered microstructure constructed of
small grains of approximately 100 nm in diameter forming pseudo dendritic protrusions.
168
Figure 88: SEM images of the film deposited from titanium isopropoxide onto float glass at 300 (a) and
600 ˚C (b) using the APCVD / mass spectrometry apparatus.
The film deposited at 600 ˚C consisted of flattened diamond shaped crystallites roughly 100 nm
in width and between 200 and 300 nm in width, in a disordered arrangement with some
perpendicular to the substrate an others lying flat.
5.3.6. Vanadium Tetrachloride
Thin films were deposited using the APCVD mass spectrometer apparatus VCl4 and ethyl acetate
as a precursors. Four depositions were carried out, the first with only a 0.2 L min-1 flow of VCl4
and the following three with 0.6 L min-1, 0.2 L min-1 and 0.05 L min-1, giving approximate ratios
of 2.5:1, 1:1 and 1:4 ethyl acetate to vanadium chloride respectively. Upon removal of the thin
films from the reactor, films deposited with a 0.6 L min-1 flow of ethyl acetate appeared pale
yellow and opaque. The film was fairly adherent to the substrate surface, resisting the Scotch
tape test. The film deposited from a 0.2 L min-1 flow of ethyl acetate was transparent and light
orange in colour, strongly resembling those deposited from VOCl3 in section 5.3.3. The film
deposited from the lowest flow rate of ethyl acetate was opaque and dark red, with no
adherence to the substrate. Upon exposure to the atmosphere the film changed, becoming
green and oily within twenty minutes of deposition. This implied that the vanadium chloride
was not fully oxidised within the reactor, most likely depositing as a partially oxidised vanadium
169
chloride species that upon exposure to air reacted with atmospheric water to give an oily
vanadium hydroxide species. Deposition with no oxidising agent gave no deposition.
Figure 89 shows the mass spectrum of VCl4. Initially, VCl4 was passed through the reactor at
600 ˚C without any oxidising agent and a predictable spectrum, with peaks at m/z 192, 155, 122
and 86 representing the parent ion VCl4 and the subsequent loss of one, two and three chlorine
atoms respectively was observed. Furthermore, the presence of peaks at m/z 227 and 264
indicates that chlorination of the parent ion via the fragmented Cl∙ radicals, as observed for
TiCl4 was also occurring with VCl4.
chemical vapour deposition reactor at a temperature of 600 ˚C
Following the deposition using only VCl4, spectra were collected for depositions with the same
pressure of VCl4 with different pressures of ethyl acetate passing through the reactor. In the
case of this study, the mass flows of VCl4 and ethyl acetate were closely monitored as the final
170
product has been found to relate closely to the stoichiometry of VCl4 and the oxidising species.
The mass flows of all three experiments are shown in Table 15.
By tuning the flow rates, atmospheres were tailored so that VCl4: ethyl acetate ratios of 2.5:1,
1:1 and 1:4 were used (experiments 11, 12 and 13 respectively), in order to observe the effect
of increased levels of oxidising agent on the atmosphere inside the reactor.
Table 15: Approximate vapour pressure and molar flow rates of VCl4 and ethyl acetate during mass
spectrometry APCVD depositions 11-13
VCl4 ethyl acetate
Deposition Bubbler Flow Vapour Mass flow Bubbler Flow Vapour Mass flow
Temp /˚C rate pressure Mol min-1 Temp /˚C rate pressure Mol min-1
L min-1 mmHg L min-1 mmHg
11 70 0.6 60 0.002 30 0.05 120 0.0005

12 70 0.6 60 0.002 30 0.2 120 0.002
13 70 0.6 60 0.002 30 0.6 120 0.005
The results are shown in Figures 90 and 91. As the flow rate of ethyl acetate increases, relative
heights of the ethyl acetate peaks to the vanadium species increases, however the ratios do
not reflect the amount entering the reactor, as in all three cases the highest peaks in the
spectrum are those at m/z 88, 70 and 66, commonly associated with ethyl acetate. This is due
to both the ethyl acetate having fewer breakdown products and a lower boiling point and
reactivity than VCl4 meaning it is far more likely to reach the detector and not deposit or
breakdown on the walls of the sampling tube or mass spec chamber. In all three cases the
171
presence of the ethyl acetate appears to lead to the emergence of peaks that relate to VOCl 3,
as observed in Figure 91.
Figure 90: Mass spectra of APCVD depositions of VCl4 at 600 ˚C with flows of ethyl acetate at 0.2 L min-
1
(top) and 0.05 L min-1 (bottom).
172
Figure 91: Mass spectra of APCVD depositions of VCl4 at 600 ˚C with flows of ethyl acetate at 0.6 L min-
1
with peaks in the range of m/z- 180 to 300 enhanced.
173
Peaks at m/z 172, 138 and 102 with isotope patterns matching those of three, two and one
chlorine ligands can be seen in all three spectrum with the intensities increasing as the mass
flow of ethyl acetate increases.
In the depositions carried out with ethyl acetate, flow rates of 0.2 L min-1 and 0.05 L min-1, only
those peaks and the peaks of the unreacted VCl4 and ethyl acetate were observed (Figure 91).
When the ethyl acetate level is highest, the flow rate of 0.6 L min-1 , however, new peaks are
observed at m/z 281, 244 and 209. These are thought to relate to the VCl4 with one intake ethyl
acetate ligand coordinated to the metal centre via the carbonyl oxygen, along with the same
species having lost one and two chlorine ligands respectively. The very low intensity of these
peaks indicates that this structure is somehow unstable and likely breaks down in the gas phase
to form VOCl3.
5.3.6.2. Analysis of Thin Films
i) X-ray Diffraction
X-ray diffraction patterns of the thin films confirmed that increasing the level of ethyl acetate
and by extension the amount of VOCl3 forming within the reactor increased the oxidation state
of the final deposition product (Figure 92). The film deposited with 0.6 L min-1 ethyl acetate
flow shows a clear pattern for V2O5. The film deposited using a 0.2 L min-1 shows a pattern very
similar to that resulting from the VOCl3 depositions, that of monoclinic VO2. In both cases the
patterns were considerably broadened, suggesting low crystallinity in the films.
The film deposited from the lowest flow of ethyl acetate appears to be largely amorphous, with
small peaks that may represent V2O5 just visible below the baseline. This is in keeping with the
174
incomplete removal of chlorine in the reactor resulting in slow oxidation of an intermediate
oxychloride species on the substrate in air to vanadium (V).
Figure 92: X-ray diffraction pattern of the films deposited from VCl4 and ethyl acetate onto float glass
at 600 ˚C using the APCVD / mass spectrometry apparatus with ethyl acetate flow rates of 0.6 L min-
1
(top), 0.2 L min-1 (middle) and 0.05 L min-1 (bottom) and alongside standards for monoclinic (PDF 72-
0514)188 and tetragonal (PDF 80-0690) VO2.189
ii) X-ray Photoelectron Spectroscopy
X-ray photoelectron spectroscopy (XPS) of the thin films deposited from VCl4 and ethyl acetate
further confirm the pattern of greater oxidation as a result of higher levels of ethyl acetate. The
film deposited with a flow of 0.6 L min-1 of ethyl acetate gave the largest signal for the V(V)
oxidation state over V(IV), with a ratio of 3.2:1. The presence of some V(IV) in the film shows
that the vanadium in the material has not been completely oxidised from the 4+ state (Figure
93a).
175
a) 40000
V2p Scan
V 2p 3/2 (V)
V 2p 3/2 (VI)
V2p Scan (counts per second)

Background
30000
Envelope
20000
10000
0
535 530 525 520 515 510
Binding energy (eV)
b) V2p Scan
V 2p 3/2 (V)
V 2p 3/2 (VI)
V2p Scan (Counts per second)
100000 Background
Envelope
50000
535 530 525 520 515 510

Binding energy (eV)
Figure 93: Detailed XPS scans of: a) V 2p of film deposited from 0.2 Lmin-1 VCl4 and 0.6 L min-1 ethyl
acetate and b) V 2p of film deposited from 0.2 Lmin-1 VCl4 and 0.2 L min-1 ethyl acetate, V(IV) (dotted
line) vs. V(V) (dashed line).
176
The film deposited from 0.2 L min-1 flow rate of ethyl acetate gave a predominantly V(IV) signal,
with the ratio of the 5+ to 4+ oxidation states at 1:2.1. The presence of some V2O5 in any VO2
film is explained by surface oxidation of the film (Figure 93 b). This is exacerbated by the nature
of XPS which is a highly surface sensitive technique.
XPS of the film deposited from 0.05 L min-1 of ethyl acetate gave a film that was roughly 1:1
V(V) to V(IV), which is in keeping with the idea that the material deposited was only partially
reacted, and that a large amount of vanadium chlorine bonds remained, which is evidenced by
the colour change and non adherent nature of the film deposited. The resulting oily film would
likely contain a number of different species, especially following exposure to air. XPS was also
taken of the film deposited from VCl4 without any ethyl acetate, revealing that the glass, which
appeared uncoated, contained a very small amount of V(IV).
iii) Scanning Electron Microscopy
SEM imaging of the films deposited from VCl4 appears to show a considerable difference in the
morphology as a result of increased levels of ethyl acetate present in the gas flow. When the
amount of ethyl acetate was maximised and the mass flow ratio to VCl4 was roughly 5:2 (Figure
94a), the films are at their most porous and disordered, consisting of roughly spherical
particulates of roughly 100 nm in diameter, adhered together to form protruding structures.
When the ethyl acetate flow was set to 0.2 L min-1 and the mass flow ratio to VCl4 was around
1:1, the morphology was observed to consist of vertical columns, roughly cylindrical in shape
and 200 nm in diameter, densely packed together on the substrate surface. When the mass
flow of ethyl acetate was reduced even further, with the flow rate set to 0.05 L min -1 and the
mass
177
Figure 94: SEM images of the film deposited from a 0.6L min-1 flow of VCl4 onto float glass with 0.6 L min-1 (a), 0.2 L min-1 (b) and
0.05 L min-1 (c) flows of ethyl acetate using the APCVD / mass spectrometry apparatus.
178
flow ratio to VCl4 at 1:4 ethyl acetate to VCl4, square plate- like particles of between 200 – 800
nm in width and no more than 50 nm in thickness were seen, arranged in a disorganised fashion
across the surface with some flat to the surface, some perpendicular. Some cube shaped
particles of roughly 500 nm in width were also observed
5.4. Conclusion
This chapter details a novel, sensitive and versatile approach to analysing the contents of an
APCVD reactor using mass spectrometry, during high temperature depositions. The approach
has proved robust enough to sample highly reactive metal chloride species that would be
beyond the scope of most spectrometers. The methodology has also been shown to be capable
of ‘capturing’ species that form in the gas phase from the reaction of the metal oxide precursors
and the oxygen source, giving new insight into the intermediate phases observed in APCVD.
Initial testing using TiCl4 and VOCl3 with ethyl acetate as a source of carbonyl oxygen appear to
reaffirm what was previously postulated with regards to the reactivity of the two species. In
the case of the TiCl4 a species featuring two ethyl acetate species coordinated to the titanium
species was observed with only the loss of a single chlorine atom from the metal centre. An
analogous complex with VOCl3 was not observed. In that case with each ethyl acetate molecule
that was observed to have coordinated to the metal centre, chlorine was lost, as observed with
the diethyl malonate. This is important as it shows that the metal chloride species behave the
same way in the gas phase, as in the solid phase. The ability to observe the intermediate
gaseous species in-situ rather than in solution, (as with the NMR) and solid, (as with the
crystallography) gives new weight to these conclusions.
In addition to this, new light has been shed on numerous other APCVD precursors. Depositions
using butyltin chloride reveal that the reaction with the ethyl acetate can proceed via two
179
pathways, either displacing the butyl group, or the chloro group first, with the latter being the
more favourable of the two.
In the case of titanium isopropoxide, deposited alone, at high temperatures the possible
formation of a dimeric intermediate to deposition of TiO2 was observed. Ti-O-Ti linkages were
observed to form in the gas phase, before contact with the substrate surface. These
intermediates are then likely to nucleate on the substrate surface, resulting in the TiO2 thin film
product.
Finally, with VCl4 when deposited alongside ethyl acetate, the observation of VOCl3 as an
intermediate in the deposition of vanadium oxide thin films were observed, with the height of
the VOCl3 peaks correlating to the quantity of ethyl acetate in the system. Increasing the level
of ethyl acetate increases the vanadium: oxygen ratio in the final film. This could mean that
VOCl3 is the gaseous intermediate to VO2, the vanadium oxygen bonds forming in the gas phase
before depositing.
The range of applications of this apparatus is broad and has by no means been fully explored.
In addition to work with APCVD, the mass spec CVD linkage has been utilised in the study of
aerosol assisted chemical vapour deposition reactions, gas sensor arrays and low pressure
chemical vapour deposition applications. Further work would likely involve increasing the range
of precursors studied with iron oxide precursors being a likely candidate, as well as investigation
of dual source co-depositions, although the resulting spectra would likely be extremely
complex.
180
Chapter 6
Aerosol Assisted CVD of Photocatalytically Active TiO2-

SiO2 Microfractal Thin Films
6.1. Introduction
The work presented in this chapter is based around a continuation of the coordination
chemistry of carbonyl ligands with TiCl4. In addition to the malonate and triglyceride examined
in chapter 4, several carbonyl species featuring silane groups were reacted with TiCl4. Whilst
initially linked to the discovery of organosilanes within the industrial samples for TiCl4, the
emergence of one compound in particular was found to be highly effective as an aerosol
assisted CVD precursor for the deposition of TiO2-SiO2 composite thin films, which show
increased photocatalytic activity over TiO2 deposited from TiCl4 using the same deposition
technique. The results of the reaction of four silyl-keto species, methyltrimethylsilyl acetate,
trimethylsilyl acetate, silicon tetraacetate and bis-trimethylsilyl malonate with TiCl4 to produce
complexes 10 [Ti{OC(OMeCH2SiMe3}Cl3(µ-Cl)]2, 11, 12 [Ti3O2Cl3(CH3CO2)5 and 13 are presented.
Additionally the resulting TiO2-SiO2 composite thin films fabricated from the deposition of 10
onto float glass using AACVD are characterised and tested from photocatalytic efficiency and
hydrophilicity.
6.1.1. Aerosol Assisted CVD
Aerosol Assisted CVD differs from conventional chemical vapour deposition in that
transportation of the precursor is primarily a liquid phase process. The precursor is dissolved
or suspended in a solvent, from which an aerosol is generated using an ultrasonic humidifier.202
This creates a mist of aerosol droplets containing solvated precursor molecules that are then
181
transported into the reactor via a carrier gas (Figure 95). In this case nitrogen was used. Upon
entering the reactor chamber, the aerosol droplets evaporate, leaving the precursor, which
deposits onto the substrate.203
Figure 95: Schematic diagram of the AACVD apparatus.
AACVD has a number of advantages over conventional methods of CVD, most notably that it
opens up a new range of precursors as the primary prerequisite for an AAVCD precursor is its
solubility rather than its volatility, as is the case with other methods. 204 The process is also far
simpler to set up, requiring no heated tubing to carry the hot precursor, simply an aerosol
generator and a reactor. AACVD has been used to deposit a large number of metal oxide
coatings. These include titanium dioxide,205,206 thermochromic vanadium dioxide,207–209 tin
oxide ,210–212 zinc oxide213 and tungsten oxide,214,215 as well as composites216,217 and doped
films.218–220
The use of a solution based medium allows for greater control over the composition of doped/
composite films and also the deposition of mixed metal oxides by allowing the pre-mixing of
the precursors in solution.221 These one pot depositions allow the correct stoichometery of the
two components to be achieved before the deposition of the material. 222 The wider range of
precursors available also means that there is a much more expansive combination of precursors
that can be combined in order to give the desired properties.223,224
182
AACVD is a highly tunable technique, with many variants.225 In general a piezoelectric mister is
used to generate aerosol droplets of a solution of the precursor in a solvent. The deposition
can be altered by the frequency of the ultrasound used to generate the mist, with higher
frequency producing more numerous, finer droplets.162 The use of a solution based technique
allows for control over deposition speed and/or film thickness by altering the concentration of
the precursors within the solution.226
AACVD has also been shown to produce different morphologies compared to films deposited
by traditional CVD methods.227 AACVD has been used to produce wide a range of
nanostructures from the same precursor, for instance tungsten oxide from W(OEt) 6 which has
been deposited as particles, wires, hollow tubes, rods and even ‘nanoflowers’.228,229 Further
control of morphology can be demonstrated by use of surfactant additives to the precursor
solution, such as substituted ammonia salts. These have been found to favour certain growth
patterns.230
Figure 96: Schematic representation of various deposition pathways possible in AACVD.
Depositions within an AACVD process can occur via a number of pathways, relating to the
substrate temperature, concentration of precursor, solvent used and other factors (Figure 96).
In each case the precursor arrives at the reactor solvated in droplets of the chosen solvent. The
solvent either evaporates before reaching the substrate (a) or afterwards (e).231 In the case of
183
(a) the precursor species commonly decomposes in the gas phase, forming a reactive
intermediate vapour or particles of the desired product ((b) + (c) respectively), similar to
atmospheric pressure chemical vapour deposition.225 Following this, in both cases the products
are absorbed onto the substrate to form the film. The gaseous reagents can also adsorb onto
the substrate intact before decomposing and coordinating onto the surface of the substrate
(d).231 In the case of (e) the solution deposits onto the heated substrate and spreads out before
rapidly evaporating, leaving the precursor species to react as seen in (d).232
Problems with this method are mainly based around the time taken for the deposition to
complete, taking anywhere from ten minutes to over an hour depending on the ease with which
an aerosol of the solvent is generated. There are also issues with industrial scale up 233 and
reproducibility.234
6.1.2. TiO2-SiO2 Composite Films
The photocatalytic properties of TiO2 are widely recognized235 and have received much
literature attention.236 Irradiation of TiO2 with UV light causes the excitation of electrons to the
conduction band, leaving a positive hole that can migrate to the surface, oxidising water to
form highly reactive hydroxyl radical species capable of breaking down surface organics. 237
Efforts to improve the photocatalytic properties of TiO2 have focussed on the use of dopants
such as nitrogen,238 fluorine,239 and tungsten,240 as well as composite materials such as TiO2-
SiO2, which have been produced in the form of both thin films 241 and nanoparticles,242 both of
which are frequently doped to further improve catalytic activity.243
TiO2-SiO2 thin films have themselves been widely studied for a number of uses. These films
have been shown to have exceedingly high wettability.244,245 Work by Hourmard et al. showed
that films of 20:80 and 60:40 composites of TiO2:SiO2, prepared via sol gel of
tetraethylorthosilicate and isopropyl orthotitanate show increased superhydrophilicity
184
compared to TiO2 prepared in the same manner.246 This was attributed to increased surface
roughness of the composites, as well as the presence of charged TiOx- and SiOx+ at the
interfaces.247
SiO2 / TiO2 composite films have also been shown to have the potential for excellent
photocatalytic behaviour. Films with a TiO2:SiO2 ratio of 1:1, deposited onto ethylene
terepthalare via dip coating using a peroxotitanium acid/(aminopropyl) triethoxysilane sol,
were shown to photocatalytically degrade Rhodamine B under UV light at a rate roughly twice
that of pure TiO2 deposited in the same way. This was attributed to the generation of surface
acid sites, as well as smaller grain size.248 Likewise, a study by Yu et al. found that a 5%
composite film of SiO2 in TiO2 was more effective at degrading methyl orange under UV/Vis
conditions than TiO2 prepared in the same way, however at concentrations greater than 5% the
performance became poorer.249
Chemical vapour deposition has been widely used in the deposition of TiO2 thin films, using a
variety of methods such as atmospheric pressure (APCVD),250 aerosol assisted (AACVD),251 and
plasma enhanced (PECVD).252 Examples of TiO2-SiO2 thin films grown via CVD are limited.
Klobukowski et al. succeeded in depositing a 1:1 TiO2-SiO2 film onto float glass via APCVD using
a [Ti(OiPr)2{OSi(OtBu)3}2] precursor.253 Additionally, Yoon et al. fabricated a TiO2-SiO2 composite
film from the co-deposition of tetraethylorthosilicate (TEOS) with titanium isopropoxide at 400
˚C using AACVD.254
This work describes the reactions of TiCl4 with methyl trimethylsilyl acetate (mtmsa) to form a
novel single source precursor for the deposition of photocatalytically active TiO2-SiO2 thin films
onto glass via AACVD over a range of temperatures. AACVD has a number of advantages over
conventional methods of CVD, the most notable being that it opens up a new range of
precursors as their effectiveness is dependent on solubility rather than volatility as is the case
185
with other methods. The process is also far simpler to set up, requiring no heated tubing to
carry the precursor vapour, simply an ultrasonic humidifier and a reactor.225
Figure 97: Silicon containing ligands reacted with TiCl4 used in the synthesis of 10 (top left), 11 (top
right) 12 (bottom left) and 13 (bottom right)
In this section TiCl4 was reacted with four silicon containing carbonyl species, methyl
trimethylsilyl acetate, trimethylsilyl acetate, bis-trimethylsilyl acetate and silicon tetraacetate
(Figure 97) to produce compounds 10 [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)]2, 11, 12 [Ti3O2Cl3(O2CCH3)5
and 13 respectively. The resulting species were characterised by NMR spectroscopy and single
crystal X-ray diffraction analysis where possible. Following this, 10 was used as a single source
precursor for the deposition of TiO2-SiO2 films for use as self cleaning coatings as evidenced by
their enhanced propensity for the breakdown of stearic acid when compared with TiO2 films
deposited from TiCl4 in the same manner. This is the first deposition of a TiO2-SiO2 composite
film using a single source precursor via AACVD. Depositing in this manner lead to films with
novel fractal like morphologies.
186
6.2. Experimental
All starting materials were purchased from Sigma Aldrich and used without further purification.
The solvents were dried over activated alumina by the Grubbs method using anhydrous
engineering equipment, such that the water concentration was 5 – 10 ppm.83 All products were
synthesised under an atmosphere of nitrogen obtained from BOC in anhydrous solvents using
standard Schlenk techniques.
6.2.1. Synthesis of [Ti{OC(OMeCH2SiMe3}Cl3(µ-Cl)]2 [10]
Methyl trimethylsilyl acetate (1 g, 0.006 mol) was added to an excess of titanium(IV) chloride
(1 ml, ~ 10 mmol), dissolved in anhydrous hexane (30 ml) in a Schlenk flask under nitrogen. The
resulting mixture was stirred for two hours. Following this the solvent was removed via
filtration, and the product washed three times with anhydrous hexane and dried under vacuum
in order to ensure the complete removal of all remaining TiCl4, leaving a light yellow, free-
flowing powder product [10] (1.931 g, 92.1%). Single crystals of 10 were grown from a 5 ml
saturated solution of the product in anhydrous toluene, layered with 10 ml of anhydrous
hexane and left under nitrogen for one week. 1H NMR (CDCl3): δ 4.04 (s, 2H, CH2Si), 2.54 (s, 3H,
OCH3), 0.44 (s, 9H, SiMe3) 13C {1H} NMR (CDCl3): δ 184.8 (C=O), 56.5 (CH2), 28.8 (CH3), 2.4
(SiMe3). MS: m/z = 671.46 [M + H+].
6.2.2. Reaction of TiCl4 with trimethylsilyl acetate to produce [11]
Trimethylsilyl acetate (1 g, 0.006 mol) was added to an excess of titanium(IV) chloride (1 ml, ~
0.01 mol), dissolved in anhydrous hexane (30 ml) in a schlenk flask under nitrogen. The resulting
mixture was stirred for two hours. Following this the solvent was removed via filtration, and
the product washed three times with anhydrous hexane and dried under vaccum in order to
187
ensure the complete removal of all remaining TiCl4, leaving a bright yellow powder product (11)
(1.344 g, 55.3%). Attempts were made to grow single crystals of 11 by layering a 5 ml saturated
solution of 11 in CHCl2 with hexane, an leaving for one week, and by freezing a 5 ml saturated
solution over one month, however in both cases the product was observed to break down into
a white solid over time, without the formation of crystals. 1H NMR (CDCl3): δ 2.63 (s, 3H, CH3),
0.44 (s, 9H, SiMe3), 0.6 (s, 3H), 2.1 (s, 0.5), 2.3 (s, 0.5), 2.4 (s, 0.8). 13C {1H} NMR (CDCl
3): δ 173.3
(C=O), 22.76 (CH3), 3.6 (SiMe3), (Appendix 11). The NMR spectra of 11 shows a number of minor
peaks adjacent to the product (0.6, 0.8, 2.1, 2.3) peaks that are likely the result of unreacted
starting material remaining in solution.
6.2.3. Reaction of TiCl4 with silicon tetraacetate to produce [Ti3O2Cl3(O2CCH3)5] [12]
Silicon tetraacetate (1 g, 0.004 mol) was added to an excess of titanium (IV) chloride (1 ml, ~
0.01 mol), dissolved in anhydrous hexane (30 ml) in a schlenk flask under nitrogen. The resulting
mixture was stirred for two hours. Following this the solvent was removed via filtration, and
the product washed three times with anhydrous hexane and dried under vaccum in order to
ensure the complete removal of all remaining TiCl4, leaving a white powder product (10) (1.701
g, 93.87%). Single crystals of 10 were grown from a 5 ml saturated solution of the product in
anhydrous toluene, layered with 10 ml of anhydrous hexane and left under nitrogen for one
week. 1H NMR (CDCl3): δ 2.32 (COCH3) (s, 3H). 13C {1H} NMR (CDCl3): δ 184.8 (C=O), 23.1 (CH3).
6.2.4. Reaction of TiCl4 with bis-trimethylsilylmalonate to produce [13]
Bis-trimethylsilylmalonate (1 g, 0.004 mol) was added to an excess of titanium (IV) chloride (1
ml, ~ 0.01 mol), dissolved in anhydrous hexane (30 ml) in a Schlenk flask under nitrogen. The
resulting mixture was stirred for two hours. Following this the solvent was removed via
filtration and the product washed three times with anhydrous hexane and dried under vacuum
in order to ensure the complete removal of all remaining TiCl4, leaving an orange, powder
188
product [11] (1.068 g, 79.2%). The product was found to be insoluble in CHCl2 and CDCl3 with
NMR showing only signals for the trimethylsilyl chloride formed as a side product (0.5 ppm),
hexane, toluene and vacuum grease.
6.2.5. Deposition of Thin Films
Following characterisation of 10 – 13 as well as some test runs it was determined that only 10
was a suitable precursor for the AACVD of TiO2-SiO2 composite thin films, as the other three
were found to be unsuitable precursors. Compounds 11 and 12 were found to decay rapidly
upon exposure to a dynamic flow of nitrogen. In the case of 13, despite being air stable, the
compound did not dissolve in any common AACVD solvents and did not contain any silicon in
the crystallised product.
For all three depositions using 10, 0.5 g (0.00071 mol) was dissolved in 20 ml of anhydrous
toluene and transferred to a glass bubbler under nitrogen using a cannula. The resulting
solution was atomized for 5 minutes to ensure complete solvation of the precursor. An aerosol
was generated using a Piezo Liquifog® ultrasonic atomiser. The deposition was carried out using
nitrogen as the carrier gas at a flow rate of 0.5 L min −1. The glass substrate was standard float
glass of 4 mm thickness. A top plate was suspended level, 0.5 cm above the substrate to ensure
a laminar flow.
A stream of air was introduced into the baffle via a second tube, from the house line fitted with
a filter. Air is frequently used in AACVD rather than nitrogen as a carrier gas in order to ensure
that the product is fully oxidised in the reactor chamber. 255 In this case however, due to the air
sensitive nature of the precursor, using air as the main carrier gas would result in the pre
oxidation of the precursor, forming TiO2 before the materials could enter the chamber and
deposit. By using the split baffle and two streams, one of nitrogen through the bubbler and one
of air directly into the chamber the benefits of both carrier gasses can be achieved without loss
189
of product to oxidation before entering the chamber. The depositions were carried out at 450,
500 and 550 °C, each taking approximately 30 minutes. A final deposition using a solution of
TiCl4 (0.5 ml, 0.005 mol) in 20 ml of anhydrous toluene was carried out at 500 °C using the same
conditions as the above reactions in order obtain a control film of TiO2.
6.2.6. Analytical Methods
X-ray crystallography, both obtaining and solving the spectra was carried out by Dr. Caroline
Knapp. Single crystal diffraction patterns were collected on a SuperNova, Dual, Cu,
Atlas diffractometer. The crystal was kept at 150 K during data collection. Data was processed
with Olex2,87 the structure was solved with the olex2.solve88 structure solution program using
Charge Flipping and refined with the ShelXL89 refinement package using Least Squares
minimisation. X-ray diffraction (XRD) patterns of the deposited films were obtained using a
Bruker-Axs D8 (GADDS) diffractometer equipped with a monochromated (Kα1 and K𝛼2) Cu X-
ray source and a 2D area X-ray detector with a resolution of 0.01°. The films were analysed with
a glancing incident angle of 5°. The diffraction patterns obtained were compared with database
standards. X-ray photoelectron spectroscopy (XPS) was performed using a Thermo ScientificTM
K-alphaTM spectrometer, with monochromated Al Kα radiation, a dual beam charge
compensation system and consta pass energy of 50 eV. Survey scans were collected in the
energy range of 0 – 1200 eV.
1H and 13C{1H} NMR spectroscopy was carried out on a Bruker A-600 Mz spectrometer,
operating at 295 K and 600.13 MHz (1H). Signals are reported relative to SiMe4 (δ = 0.00 ppm)
and the following abbreviations are used s (singlet), d (doublet), t (triplet), q (quartet), m
(multiplet), b (broad). Deuterated CDCl3 was obtained from GOSS Scientific and was degassed
and dried over 3 Å molecular sieves. Liquid Chromatography Mass spectroscopy of 10 was
recorded using a Thermo Finnigan LTQ two-dimensional ion-trap mass spectrometer operating
190
in ESI mode. High-resolution peaks were used for the principal peaks Ti (2p), O (1s), C (1s), Si
(2p) and Cl (2p) with a spot size of 400 μm. Elemental composition was inferred from the
integral of the peak corresponding to the signal of that particular element. A Renishaw 1000
spectrometer equipped with a 514 nm laser was used to obtain Raman spectra of the samples.
The Raman system was calibrated using a silicon reference. Transmission and reflectance
UV/Vis spectroscopy were performed using a double beam; double monochromated Perkin
Elmer Lambda 950 UV/vis/NIR Spectrophotometer. The absorption spectra were recorded
directly on the films as deposited on quartz slides, clamped against an integrating sphere in
perpendicular position to the beam path. A Labsphere reflectance standard was used as
reference in the UV/vis measurements. SEM analysis was carried out using secondary electron
image taken using a JEOL 6301 field-emission instrument (5 kV).
6.2.7. Photocatalytic Testing
The photocatalytic activity of the films was measured by their ability to break down stearic acid,
a layer of which was applied to the films via dip coating from a stock solution of 0.005 M stearic
acid in dichloromethane. The degradation of the stearic acid was measured by Fourier
transform infrared (FTIR) spectroscopy in the range 2700–3000 cm−1, using a Perkin Elmer RX-I
instrument. The area beneath the peaks relating to stretching vibrations of C-H bonds in the
stearic acid (2820-2970 cm-1) was integrated giving an estimation of the number of molecules
of stearic acid degraded using a conversion factor reported in the literature (1 cm−1 ≡ 9.7 × 1015
mol)].256 The photoactivity rates were estimated from line regression of the initial 30–40%
degradation steps (zero-order kinetics). These rates are expressed as formal quantum efficiency
(FQE) values, which are defined as the number of acid molecules degraded per incident photon.
This degradation of the stearic acid is shown in equation 1.
CH3(CH2)16CO2H + 26O2TiO2 (hν≥ Ebg) 18CO2 + 18H2O (1)
191
A blacklight-bulb UVA lamp (Vilber-Lourmat) 2 × 8 W, was used to irradiate samples during the
photocatalytic tests. The irradiance of the UV lamp (4.1 mWcm−2) was measured using a UVX
meter (UVP). The films were cleaned under wet air conditions under a 365 nm UV lamp for 24
hours and kept in the dark for another 24 hours prior to the photocatalysis testing.
TiCl4 was reacted with methyltrimethylsilyl acetate, trimethylsilyl acetate, silicon tetraacetate
and bis-trimethylsilyl malonate complexes in hexane at room temperature under nitrogen to
form 10 [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)]2, 11, 12 [Ti3O2Cl3(CH3CO2)5 and 13 in good yield.
Following this, product 10 was deposited onto plain glass using aerosol assisted CVD in toluene
over a range of temperatures. The films all had excellent coverage, evenly coating the entire
substrate, and showed good adherence, being resistant to scratching and the ‘Scotch tape test’.
6.3.1. Reaction of TiCl4 with methyl trimethylsilylacetate to give
[Ti{OC(OMeCH2SiMe3}Cl3(µ-Cl)]2 [10]
Methyl trimethylsilyl acetate (Sigma Aldrich, 97%) was added to a solution of TiCl4 and stirred
under nitrogen to give [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)] [10], a bright yellow powder in good yield.
Yellow crystals were grown by layering a solution of 10 in toluene with dichloromethane. NMR
analysis shows a downfield shifting of all of the proton signals of the uncoordinated ligand, with
the magnitude of the shift decreasing with distance from the carbonyl group (Figure 98). This
is indicative of the dative coordination of the ligand to the TiCl4 centre observed for the
malonate species reacted with TiCl4 in Chapter 3.
This hypothesis was confirmed by the X-ray diffraction analysis of the crystal grown. The crystal
structure of the precursor of [Ti{OC(OMeCH2SiMe3}Cl3(µ-Cl)]2 (10) comprises of a dimeric
species, bridged by two chloride ligands (Figure 99). Each titanium atom adopts a distorted
octahedral geometry, having retained all four chlorine atoms, with the silyl acetate molecule
192
coordinating via the carbonyl oxygen. This structure is very common with titanium chloride
species, seen frequently for complexes of TiCl4 with carbonyl ligands, such as 2-propanone257
and acetate.258
Figure 98: ¹H NMR spectra of methyl trimethylsilylacetate and the product of its reaction with TiCl4 [10], all peaks
exhibiting a downfield shift indicating coordination to the titanium metal centre.
193
Figure 99: ORTEP representation of crystal structure of 10 with thermal ellipsoids at the 50%
probability level. Hydrogen omitted for clarity.
Table 16: Crystallographic Data and Selected bond lengths angles for the crystal structure of 1a the
product of reaction TiCl4 and methyltrimethylsilyl acetate [10]
Selected Bond
Structural Data Selected Bond Angles /˚
Lengths /Å
crystal system monoclinic Cl(21)-Ti(1)-Cl(1) 166.816(18) Ti(1)-Cl(1) 2.2327(4)
space group C2/c Cl(2)-Ti(1)-Cl(1) 88.766(16) Ti(1)-Cl(2) 2.5157(4)
a 16.98585(13) Cl(4)-Ti(1)-Cl(1) 99.810(18) Ti(1)-Cl(21) 2.4615(4)
b 15.68339(10) Cl(4)-Ti(1)-Cl(21) 90.860(16) Ti(1)-Cl(4) 2.2085(4)
c 24.15174(15) Cl(4)-Ti(1)-Cl(2) 169.677(18) Ti(1)-Cl(3) 2.2349(4)
α 90 Cl(3)-Ti(1)-Cl(1) 96.526(16) Ti(1)-O(1) 2.0238(10)
β 95.1852(6) Cl(3)-Ti(1)-Cl(21) 89.995(15)
γ 90 Cl(3)-Ti(1)-Cl(2) 88.729(14)
R1 0.0289 Cl(3) -Ti(1)-Cl(4) 95.930(17)
wR2 0.0756 O(1) -Ti(1)-Cl(1) 87.55(3)
O(1) -Ti(1)-Cl(21) 84.11(3)
O(1) -Ti(1)-Cl(2) 81.33(3)
O(1) -Ti(1)-Cl(4) 93.23(3)
194
The Ti-O bond lengths present (Table 16) are 2.0238(10) and 2.0238(10) Å, indicating covalent
coordination to the titanium centre by the acetate species, as observed with the similar
structure of di-chloro-bis[(allylacetate-O)trichlorotitanium(IV)]. The Ti-Cl bond lengths to the
µ-2 bridging chlorides were 2.5157(4) and 2.4615(4) Å, longer than that of the terminal chloride
ligands, all of which are between 2.20 and 2.25 Å. This is expected due to the electron density
being shared over two metal centres, which in turn reduces the attractive forces felt between
the individual titanium centres and the electron cloud around the chlorine.
The decreased electron density also results in the bridging chlorides lying closer to the other
ligands. The bond angle between the terminal Cl(3) and the bridging Cl(2) is 88.729(14)˚,
whereas the angle between two mutually adjacent terminal chlorines Cl(1) and Cl(3) is
96.526(16)˚. This is due to the reduced electron density around the bridging chlorines resulting
in a decrease in electronegativity of the ligand and therefore a decreased repulsion between
the ligands.
6.3.2. Reaction of TiCl4 with trimethylsilylacetate to give [11]
Trimethyl silylacetate (TSA) (Sigma Aldrich, 97%) was added to a solution of TiCl4 in hexane
under an atmosphere of nitrogen and stirred for two hours. The resulting off-white powder was
washed with hexane and dried under vacuum and a 1H NMR spectrum of 11 was obtained.
Attempts were made to grow crystals by layering saturated solutions of 11 in CHCl2 and toluene
with hexane, as well as storing saturated solutions of 11 in CHCl2 and toluene in a freezer for
one month however in all cases, a white, powdery, non-crystalline precipitate was observed.
The NMR spectrum of 11 (Figure 100) shows a range of products with the signal at 0.2 ppm in
the starting material shifting from 0.2 to 0.5 ppm, and the emergence of a range of new peaks
at 0.55, 2.25, 2.40, 2.48, 2.51, 2.57 and 2.68 ppm. The integrals of these peaks do not align,
suggesting that a range of products are present. The large peak at 0.5 ppm, is likely to be that
195
of trimethylsilylchloride, the trimethylsilyl (TMS) group having undergone ligand exchange with
the TiCl4. TSA is a well used acylating agent and has been shown in the literature to readily
form trimethylsilyl chloride.259 The peaks in the 2.2 – 2.7 ppm range suggest a number of ethyl
species could have been formed, perhaps coordinating to the titanium centre in different
numbers or forming clusters, such as the one shown in Figure 102 (vide infra). This explains the
difficulty in isolating a crystalline product. It also suggests that this species would be
inappropriate for use as a TiO2 / SiO2 thin film precursor.
Figure 100: ¹H NMR spectra of trimethylsilylacetate and the product of its reaction with TiCl4 [11],
showing a series of new peaks suggesting that multiple products form.
6.3.3. Reaction of TiCl4 with silicon tetraacetate to give [Ti3O2Cl3(O2CCH3)5 [12]
Silicon tetraacetate (STA) (Sigma Aldrich, 96%) was added to a solution of TiCl4 in hexane and
stirring for two hours under nitrogen, leading to the precipitation of a pale yellow product
196
[Ti3O2Cl3(O2CCH3)5 [12]. Crystals were obtained by layering a 5 ml saturated solution of 12 in
anhydrous toluene with 10 ml of anhydrous hexane. Over a period of one week, pale yellow
crystals formed.
As with the previous experiment, the NMR strongly suggests displacement of the acetate
groups on the silicon by the TiCl4 (Figure 101). The NMR of the starting material shows a single
major peak at 2.15 ppm with satellites suggesting that the silicon tetraacetate is prone to loss
of the acetate groups in solution, as well as suggesting strong presence of impurities. The
product shows a slight shift in the acetate CH3 peak to 2.3 ppm, as well as a broadening that
could hide the formulation of multiple species. The new peak at 0.1 ppm is very likely to be
from vacuum grease as there are no methyl silyl species present in the starting material and
the formation of a methyl silane species in solution is highly unlikely.
Figure 101: ¹H NMR spectra of trimethylsilylacetate and the product of its reaction with TiCl4 [12],
showing a series of new peaks suggesting that multiple products form.
Unlike 11, crystals of 12 could be grown via solvent layering, suggesting that although the
reaction does not follow a simple coordination pathway as per 10, a product formed in great
197
enough abundance to nucleate as crystals. These crystals were surprisingly facile to produce,
meaning it is likely that 12 favours ordered crystalline growth. Analysis of the crystal revealed
that rather than the silicon tetraacetate coordinating to the titanium chloride, the titanium
displaces the silicon, coordinating to the acetate ligand. This leads to the formation of a trimeric
cluster with two central oxygen atoms, one bridging all three titanium centres, the other
bridging only two (Figure 102). The bonds between the titanium and the central oxygen atoms
are all in the region of 1.87 Å, similar to other Ti-O bond lengths calculated for TiOx cluster
molecules. 260,261
One titanium atom is coordinated to four acetate groups, with the remaining two coordinated
to three, each titanium centre retaining one chlorine atom. The two oxygen atoms in the
carboxylate groups of the acetate ligands coordinate to different metal centres. This leads to
the formation of highly distorted, six membered rings featuring the two titanium centres linked
by the central oxygen atom and the carboxylate group of the acetate.
The Ti-O-C bond angles are all between 132° and 138°, rather than 120° as is conventional for
a six membered ring, in order to accommodate coordination to both metal centres. This relates
to the highly differing bond lengths, Ti(1)-O(3) is 1.9697(11) Å, fitting with a dative Ti-O bond
compared with the considerably shorter C(1)=O(3) bond in the acetate ligand at 1.2806(19) Å.
198
Figure 102: ORTEP representation of crystal structure of 12 with thermal ellipsoids at the 50% probability level.
Hydrogen omitted for clarity.
Table 17: Crystallographic Data and Selected bond lengths angles for the crystal structure of 1a the product of
reaction TiCl4 and silicon tetraacetate [12]
Structural Data Selected Bond Angles /˚ Selected Bond Lengths /Å

crystal system monoclinic O(1)-Ti(1)-Cl(1) 175.03(4) Ti(1)-Cl(1) 2.2820(5)
space group P21/c O(1)-Ti(1)-O(3) 87.04(5) Ti(1)-O(1) 1.8733(11)
a 8.4469(3) O(1)-Ti(1)-O(7) 90.44(5) Ti(1)-O(3) 1.9697(11)
b 13.6092(4) O(1)-Ti(1)-O(9) 86.97(5) Ti(1)-O(5) 1.9929(11)
c 12.4255(4) O(3)-Ti(1)-Cl(1) 89.84(4) Ti(1)-O(7) 1.9889(11)
α 90.00(5) O(3)-Ti(1)-O(5) 87.72(5) Ti(1)-O(9) 1.9603(11)
β 98.424(3) O(3)-Ti(1)-O(7) 171.57(5) Ti(2)-O(1) 1.9868(10)
γ 90.00(5) O(5)-Ti(1)-Cl(1) 92.90(4) Ti(3)-O(1) 1.9885(10)
R1 2.70 O(7)-Ti(1)-Cl(1) 93.19(4)
wR2 7.40 O(9)-Ti(1)-Cl(1) 89.69(4)
O(9)-Ti(1)-O(3) 98.96(5)
O(9)-Ti(1)-O(5) 172.85(5)
O(9)-Ti(1)-O(7) 88.93(5)
O(7)-C(5)-O(8) 123.06(14)
Ti(1)-O(1)-Ti(2) 130.53(5)
Ti(1)-O(1)-Ti(3) 130.92(5)
199
The reason for the formation of 12, as well as the likely failure of the other reactions to produce an
acceptable precursor is the propensity of TiCl4 to simply displace the silicon species, rather than
coordinating directly to the acetate species carbonyl group. The reaction likely begins with a ligand
exchange step, the titanium coordinating to the acetate, which in turn leads to the displacement of
Cl- from the metal centre as unlike the coordination to TiCl4 seen in previous structures, the oxygen
atom would be negatively charged. This chlorine would be attracted to the newly positive silyl acetate
species.
The methyl trimethylsilylacetate ligand used to form 10 features a trimethylsilyl (TMS) group bound
to carbon, not oxygen. The carbon oxygen bond of the methyl trimethyl silyl acetate is more stable
than the silicon oxygen of bond in the trimethylsilylacetate.262 Furthermore there is a far greater
bond polarity between the carbon-oxygen bond (χC = 3.4 vs χSi = 1.9)263 than the carbon - silicon bond
(χC = 2.5 vs χSi = 1.9)264. The greater electronegativity of the oxygen facilitates the coordination of the
titanium leading to the ligand exchange.
6.3.4. Reaction of TiCl4 with bis-trimethylsilylmalonate to give [13]
Bistrimethylsilyl malonate (BTMSM) was added to a solution of TiCl4 in hexane under an atmosphere
of nitrogen and stirred for two hours. The resulting orange product was washed with hexane and
dried under vacuum. This material was found to be highly unstable and sensitive to atmospheric
oxygen and water, breaking down to a white, insoluble powder (presumable TiO2) over time even
when stored under nitrogen and almost instantaneously if the flask was opened. This made obtaining
an NMR spectrum impossible as the white precipitate would crash out of the CDCl3 as soon as the
solution was made up. Attempts were made to grow crystals by layering saturated solutions of 13 in
CHCl2 and toluene with hexane, as well as well as storing saturated solutions of 13 in CHCl2 and
toluene in a freezer for one month however in all cases, a white, powdery, non crystalline precipitate
was observed.
200
6.3.5. Characterisation Thin Films
A range of thin films were deposited from the single source precursor [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)]2
[10] in toluene at temperatures 450, 500 and 550 °C. The films were observed to display increasing
opacity as deposition temperature increased, with those deposited at 450 °C being fully transparent,
at 500 °C also transparent but with a slight fogginess in places and at 550 °C having a cloudy, almost
opaque appearance. The film deposited from TiCl4 was largely transparent but somewhat hazy, with
fogginess around the edges of the substrate. All four films were left undamaged by the “Scotch Tape”
test.
The morphology of the films was found to vary considerably with deposition temperature (Figure
103). The most notable of the films were those deposited at 500 °C. These films consisted of a base
of cubic grains of approximately 100 nm in diameter as is commonly seen in the deposition of TiO2
films.265 However on top of this film additional growth in the form of snowflake-like fractal structures
was observed. These growths also consisted of more homogenised, less defined grains that were
more reflecting than those upon which they were grown. The micro-fractals ranged in size from 5-50
µm and appear to grow outward from a central point, from which nucleation occurs. This two tiered
morphology would result in an increased surface area compared to that of a simple nanostructured
thin film (Figure 103).
201
Figure 103: SEM images of the microstructure of film deposited from [Ti{SiMe3CH2CO2Me}Cl3(µ-
Cl)]2 (10) at 500 ˚C at magnifications of x200 (a), 1,200 (b), x15,000 (c) and 2,600 x (d) with addition
of ratios of titanium to silicon at different points in the film, as measured by EDX spectroscopy.
x15,000 magnified images of films deposited at 550˚C and 450˚C (e) + (f) respectively.
202
Films deposited at 450 °C were by comparison extremely flat and featureless, with protruding
spherical grains of diameter 200 – 500 nm appearing sporadically across the surface.
Depositions at 550 °C have a more textured morphology consisting of plate-like crystals of
lengths between 0.5 and 1 µm. Only films deposited at 500 °C showed any sort of fractal
growth, the others appearing homogenous across the surface. EDX spectroscopy carried out on
the film deposited at 500 °C showed that the brighter fractal areas (Figure 103) at the top left
and bottom left contain more titanium that other parts of the film with Ti: Si ratios of 1.67:1
and 1.50:1 compared with around 1.3:1 for the other parts of the film. This suggests that
although the surface of the film consisted of a mixture of SiO2 and TiO2 particles, the area of
fractal growth on top of the films base surface morphology are more titanium rich, potentially
pure TiO2.
Side-on SEM images show that film thickness increases with temperature, with the depositions
at 450 °C giving a film of approximately 600 nm thickness and deposition at 500 °C leading to
films of around 1 µm. The depositions at 550 °C showed the thickest and also most varied film
thickness, varying from approximately 1 - 3 µm across the substrate.
Depth Profile XPS of the thin films showed that the vast majority of the silicon was situated at
the surface, with films across the temperature range showing between 15-18 at.% silicon at
surface level (Figure 106). The opposite was true of titanium, with initial concentrations of
around 20 at.% more than doubling in all three films below surface level to 40 at.%, 48 at.%, 52
at.% for films deposited at 450, 500 and 550 °C respectively (Figure 104). This is in keeping with
the EDX data, where the ratio of Ti:Si at the surface of the film was around 1.5:1, as EDX is a
more penetrative technique than XPS, which would result in it picking up on some of the higher
concentrations of titanium in the bulk. The bulk ratio of TiO2:SiO2 in films deposited at 500 °C
is far lower than other films, at 12:1, compared with 21:1 (450 °C) and 24:1 (550 °C). The ratio
203
of TiO2: SiO2 at the surface of the films was highest in the 550 °C deposited film, at 1.68:1,
whereas with the two lower temperature deposited films this ratio was 1:19 and 0.96 for 500
°C and 450 °C respectively.
XPS shows that the percentage of silicon making up the bulk of the film was considerably lower
than at the surface, with titanium being 12-24 times more abundant. This composition is more
consistent silicon doped TiO2,266,267 suggesting that the bulk phase of the film is largely Si doped
TiO2 with a composite of TiO2 and pure phase SiO2 at surface level.
The XRD patterns of all three films are shown in Figure 104. All were dominated by the peak
pattern of anatase TiO2, the phase commonly formed at this temperature range. XRD analysis
of the films showed that the depositions using 10 at 500 and 550 °C lead to the formation of a
mixed system of anatase and brookite TiO2 polymorphs. Of the three common structures of
TiO2, brookite is the least studied due to its lower refractive index of approximately 2.5 and
reddish brown colour being less desirable for use in pigments.268
700
TiO₂
TiO₂/SiO₂ 450 ˚C
B(101)
600
A (101) TiO₂/SiO₂ 500 ˚C

TiO₂/SiO₂ 550 ˚C
500
brookite
Intensity / a.u.
B(121) anatase
400
A(112) A(200)
A(211) A(213
300
)
200
100
10 20 30 40 50 60
2θ / °
Figure 104: XRD patterns of films deposited from AACVD of [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)]2 (10) at
various temperatures compared with that of anatase TiO2 (bottom).
204
Figure. 105: Raman Spectra of films deposited from AACVD of [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)]2 (10) at
various temperatures compared with that of anatase TiO2 (bottom)
Despite this it does possess a narrower band gap than anatase (2.9 eV), and therefore its
inclusion could be beneficial to photocatalytic applications. This is confirmed by Raman
spectroscopy, once again showing that whilst the main TiO2 polymorph deposited was anatase
TiO2, brookite signals were visible in films deposited from 10 at 500 and 550 °C (Figure 105).
The film deposited at 450 °C gave a weak Raman spectrum with high background noise.
This is most probably due to the relatively low surface concentration of TiO2. Transmission and
reflectance UV/Vis spectroscopy was carried out on all of the films (Figure 108). The data was
then used to determine the absorption coefficient in order to construct Tauc plots 269 and in
turn approximate the band gap of the films. The absorbance characteristics of the TiO2-SiO2
films are strongly affected by the deposition temperature.
205
Figure. 106: Overlay of the depth profile XPS peaks for Si (2p) with darkening colour representing
deeper penetration below the films surface a), Atomic percentages of titanium and silicon in the
Composite films as measured using depth profile XPS b).
Figure 107: UV/Vis Absorption spectra of films deposited from AACVD of [Ti{SiMe3CH2CO2Me}Cl3(µ-
Cl)]2 (10) at various temperatures compared with That of anatase TiO2.
206
Figure. 108: Tauc plots of films deposited from AACVD of [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)]2 (10) at various
temperatures used to estimate the band gap of these films.
The film deposited at 550 °C saw the optical band gap reduced from 3.27 to 3.20 eV (Figure
108). This is a considerable difference in band gap, mirrored in the results of the photocatalysis
testing (vide infra), especially as the ratio of TiO2:SiO2 in the bulk in the two films is similar.
Films deposited from 10 at 500 °C also appear to undergo a reduction in the band gap, although
it is less pronounced than the film deposited at 550 °C, with the band gap estimated to be 3.23
eV. The deposition at 450 °C was found to increase the size of the band gap to 3.31 eV. This is
in agreement with previous studies on TiO2-SiO2 composite thin films. 270
The presence of SiO2 as a separate phase is unlikely to lead to a reduction in band gap, due to
SiO2 having a substantially larger band gap at 9.3 eV.271
The reduction in band gap from the film deposited from TiCl4 was found to correlate well with
the results of the photocatalytic testing. The reduction in bandgap for films grown at 550 °C is
explained by the incorporation of more silicon into the TiO2 lattice at higher temperatures as
demonstrated in the shifting of the anatase (101) peak of the XRD pattern for the film deposited
207
at 550 °C. The presence of silicon within the TiO2 lattice leads to a narrowing of the optical
band gap due to the introduction of strain into the TiO 2 lattice, which has previously been
observed in both theoretical and experimental studies. 272
6.3.6. Water Contact Angle Measurements
Water contact angles for the TiO2 film and the three TiO2-SiO2 composites were taken by casting
a 2.5 µL droplet of water onto the surface and imaging them with a high speed camera. The
drop shape and contact angle were obtained using FTA3.2 software. The results are shown in
Figure 109, as well as Table 18.
The contact angle for the TiO2 film behaves as expected based on previous studies of anatase
phase TiO2 wettability,206 initially displaying somewhat hydrophilic behaviour, which is greatly
enhanced by exposure to UV irradiation, the contact angle falling from 73.9° to 11.2°. This
thought to be due to the presence of surface- OH sites being deprotonated by the UV
irradiation, as well as the UV catalysed breakdown of organic on the surface that would prevent
contact with the TiO2 surface.273 This behaviour is not shared with the TiO2 /SiO2 films. All three
films show no hydrophilicity in their non-irradiated state, with contact angles between 90 and
110°. Upon irradiation, a minor increase in the hydrophilicity of the films deposited at 500 and
550 °C was observed, the contact angles decreasing by 6.1° and 6.9° respectively. No discernible
change was observed in the contact angle of the film deposited at 450 °C. This corresponds
with the findings from the XPS studies, where the film deposited at 450 °C was observed to
have the lowest surface concentration of TiO2 as well as the highest SiO2:TiO2 ratio.
Furthermore the SEM images of the films shows that the films deposited at higher
temperatures have a protruding microstructure giving increased surface roughness, whereas
the film deposited at 450 °C has a flat, featureless microstructure that is not conducive to
wetting.14,274 the lack of UV induced hydrophilicy is supported by the XPS data which shows a
208
considerable amount of SiO2 at the surface of the films, which is considerably reduced just
below. This would greatly reduce the presence of these acid sites to which the water is able to
weakly adhere to
Table 18: Contact angles of before and after UV irradiation
Contact angle before Contact angle after

Film irradiation (°) irradiation (°)
TiO₂ 72.1 ± 1.5 10.4 ±1.4
TiO₂ /SiO₂ 450 °C 109.1 ± 1.2 109.9 ± 2.4
TiO₂ /SiO₂ 500 °C 94.8 ± 1.0 87.4 ± 1.6
TiO₂ /SiO₂ 550 °C 102.0 ± 0.7 96.1 ± 0.4
Figure 109: Comparison of the behaviour of water droplets on thin films of TiO2 and the TiO2-SiO2
composite deposited at 450 °C, 500 °C and 550 °C before UV irradiation (a, c, e and g respectively) and
after UV irradiation. (b, d, f, h respectively).
209
6.3.7. Photocatalytic Testing
Photocatalytic degradation of a coat of stearic acid applied to the films under λ = 365 nm UV
irradiation was carried out on the films over the course of three hours (Figure 110). It was
observed that of the three depositions, two yielded films that displayed a greater propensity to
break down the organic material than the TiO2 deposited in an equivalent manner, via AACVD
of TiCl4. The formal quantum efficiency (FQE) of TiO2 deposited from TiCl4 was 9.04 x10-5 which
is slightly lower than values found in the literature for undoped TiO2238 and can be explained by
the patchier nature of the films. Figure 110 shows that the TiO2-SiO2 films deposited from 10
improved as photocatalysts with deposition temperature. At 450 °C the film deposited gave an
FQE of 4.4532 x10-5, roughly half that of the plain TiO2. Films deposited at 500 °C give an FQE
of 1.6764x10-4 81% superior to that of TiO2 from TiCl4.
Film FQE (molecules photon -1) error (molecules photon -1)

450 °C 4.4532E-5 4.85783E-6
500 °C 1.6764E-4 1.32888E-5
4.0x10-4 550 °C 3.06624E-4 2.78684E-5
TiO2 9.04305E-5 7.3104E-6
FQE (molecules photon-1)
3.0x10-4
2.0x10-4
1.0x10-4
0.0
450 °C 500 °C 550 °C TiO2
Film
Figure 110: Bar graph showing the Formal Quantum Efficiency (FQE) of films deposited from AACVD
of [Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)] (1) at various temperatures compared with That of anatase TiO2 and
plain glass, as measured by the UV catalysed decomposition of the stearic acid on the films surface.
210
The optically opaque films deposited at 550 °C performed best with an FQE of 3.0662x10-4, an
improvement of over 250% above the film deposited from TiCl4 with air. UV/Vis analysis
suggests that in the case of the 500 °C deposited film the improved photocatalytic activity was
not the result of tuning the optical properties of the film, as the band gap was effectively the
same as the undoped TiO2. It is more likely that the improved performance is in fact the result
of the surface area of the film being greater due to the small grain size and the fractal structures
growing on top providing a greater capacity for catalysis.
Depositing at 550 °C saw a narrowing of the band gap to 2.93 eV, compared with the 3.15 eV
observed for pure TiO2. This is reflected in its improved catalytic activity. Reducing the
difference in energy between the highest occupied state and lowest unoccupied state means
that more electrons are capable of overcoming the energy barrier to excitation, resulting in the
formation of a greater number of positive holes in the valence band migrating to the film
surface. This in turn leads to an increase in the rate at which surface organics are decomposed.
The lowering of the band gap is likely a combination of the incorporation of the silicon which
distorts the matrix to lower the band gap, and the presence of small amounts of brookite in the
film. This is further corroborated by the shifting of the anatase peaks observed into the XRD of
the film deposited at 550 °C, not seen in any of the other films.
211
1.3 5
1.3 3
1.3 1
Intensity /a.u.
1.2 9
1.2 7
1.2 5
1.2 3
1.2 1
1.1 9
1.1 7
1.1 5
2820 2870 2920 2970

wavenumber cm¯¹
Figure 111: Overlay of the FT-IR spectra of the stearic acid on the surface of the 500 ˚C film over a two
hour period of UV irradiation with spectra taken every 20 minutes
6.4. Conclusion
A novel precursor for the deposition of TiO2 / SiO2 composite films has been synthesised by
reacting TiCl4 with methyl trimethylsilyl acetate to give a chloro-bridged dimeric species
[Ti{SiMe3CH2CO2Me}Cl3(µ-Cl)]2 [10] . The structure of the precursor was confirmed using single
crystal X-ray crystallography. Using the aerosol assisted chemical vapour deposition technique,
films were successfully deposited at 450, 500 and 550 °C from a solution of 10 using nitrogen
as a carrier gas and a flow of air entering the reactor in order to oxidise the precursor.
The resulting films were found to have ratios of TiO2: SiO2 of between 2:1 and 1:1 at the surface
and 12:1 and 24:1 in the bulk. In films deposited at higher temperatures the films were found
212
to contain both Anatase and Brookite phases of TiO2. SEM revealed a novel morphology of
‘microfractals’ of material nucleating out from central points across the films surface, leading
to improved surface area.
Studying the UV catalysed decomposition of stearic acid on the films surface; it was revealed
that films deposited at 500 °C and 550 °C showed formal quantum efficiencies 73% and 150%
higher than TiO2 films deposited via the same method using TiCl4 as a precursor. This work
shows that SiO2 has potential as an additive for improving photocatalytic efficiency of TiO2 thin
films.
Chapter 7
213
Derivatives of VOCl3 and Their Uses in Materials
Synthesis
7. Introduction
7.1.1. VO2 Thin Films
Monoclinic vanadium dioxide thin films and nanoparticles for energy saving, infrared radiation
reflecting windows has come under considerable attention in the literature, due to its
considerable potential to help curtail energy intensive means of heating and cooling homes,
thereby reducing pressure on the world’s non-renewable energy resources.275–278 By
developing windows that can promote and prevent heat loss at different temperatures the
need to use such resources can be lessened, reducing both household bills and global energy
usage.
The monoclinic phase of vanadium(IV) oxide (VO2 (M)) undergoes a structural transition to the
rutile vanadium(IV) oxide (VO2 (R)) at 68˚C. The temperature at which this change occurs is
known as the metal to semiconductor transition temperature (MST). This phenomenon, in
which the permittivity of electromagnetic radiation is dependent on temperature, is known as
thermochromism.203,279 VO2 (M) is of particular interest as its MST is relatively close to room
temperature, unlike other thermochromic VxOy species, such as V 2O3.280 Below the MST, VO2
(M) is a semiconductor that transmits infrared and ultraviolet radiation consistently, whereas
above the MST, VO2 (R) exhibits metallic conduction, reflecting in the same region.281 This
property is crucial in the design of smart window coatings, that are transmissive of sunlight at
low temperatures, but reflective at high temperatures, thus alleviating the use of expensive
and energy-intensive air conditioning units in hot countries (Figure 112).282

214
Figure 112: Diagram of the behaviour of visible and near infrared radiation upon contact with a
VO2(M) coated window, at below (left) and above (right) the MST275
Although close to room temperature, the MST must still be reduced in order to make the
application applicable to everyday life. The most common means of achieving this reduction is
the incorporation of dopant atoms into the VO2 matrix. Tungsten is by far the most widespread
and effective dopant in lowering the MST. Furthermore, incorporation of tungsten lends the
windows a blue green tint as a result of a narrowing of the optical band gap due to lattice strain,
altering the absorption of energy in d-d transitions that give the material its colour. This masks
the less aesthetically acceptable brown/ yellow of VO2 (M).146,282–284 Other dopants such as:
niobium,285 fluorine,286 magnesium287 and molybdenum288 have also been shown to reduce the
MST. The incorporation of higher valence atoms with large atomic radii, such as tungsten, is
thought to reduce the MST by introducing distortions into the VO2 lattice around the dopant
atom centre,289 as well as introducing areas of greater electron density into the system. 290
215
The production of thin films of VO2 (M) is synthetically challenging due to the large number of
stable oxide polymorphs of vanadium that can form under CVD conditons.280 This has led to a
wide variety of methods for the synthesis of VO2 (M) thin films and nanoparticles.291–293 Thin
films of VO2 have been very well studied, with highly effective thermochromic VO2 (M) and
doped VO2 (M) films deposited on a variety of substrates.129,291,294 The scarcity of low molecular
weight, volatile precursors containing vanadium(IV), suitable for APCVD has led to the use of
air-sensitive VCl4 295,296 and stable [VO(acac)2]297 precursors. VOCl3 is also a well used VO2 (M)
thin film precursor in APCVD, even though the vanadium present is in the +5 oxidation
state.295,298
Thin films of VO2 (M) have also been deposited via aerosol assisted chemical vapour deposition
(AACVD) methods.220,299,300 The advantage of AACVD is that a wider range of precursors can be
employed, giving a greater scope to tailor the vanadium precursor, potentially leading to the
formation of more effective thin films.144
Synthetic methods for nanoparticles formation vary greatly.292,293,301 Vanadium oxide
nanoparticles have been synthesised using a variety of methods, including pyrolysis, 302 sol-
gel,303 hydrothermal methods304 and ball milling305 for use in coatings. Hydrothermal (and sol-
gel methods) have been shown to produce highly crystalline VO2 (M) nanostructures, from
vanadium precursors in the presence of reducing agents.301,191,306 Manipulating the
environment inside the high temperature synthesis vessel by tailoring the amount of oxygen
present preventing the over/under oxidation of vanadium(IV) species is crucial in nanoparticle
synthesis, a technique that is employed here as well, in the AACVD of the VO2 thin films. 307–309
The majority of VO2 (M) synthesis focuses on a vanadium precursor that is reacted with an
oxygen-containing species in situ. This work however, shows the fabrication of doped and
216
undoped VO2 (M) thin films using AACVD, and surfactant controlled anisotropic vanadium oxide
nanomaterials from the previously synthesised single source precursor dichloro(oxo)
Vanadium(V) diethyl malonate [{VOCl2(CH2(COOEt2)}4] (6). The films were doped with tungsten
by adding small amounts of tungsten(VI) hexaphenoxide [W(OPh)6] to the precusor solution,
and a lowering of the MST was observed. Vanadium oxide nanostructures were synthesised by
the thermal decomposition of 6 in the presence of a high boiling point solvent (1-octadecene)
and structure directing surfactants (oleic acid and oleylamine). The structure of 6 was
determined by X-ray crystallography, elemental analysis and 1H NMR spectroscopy in chapter
4. Thin films and nanostructures were characterised using X-ray diffraction (XRD), scanning
electron microscopy (SEM), transmission electron microscopy (TEM), X-ray photoelectron
spectroscopy (XPS) and Energy-dispersive X-ray spectroscopy (EDS). Thermochromic
measurements were measured using UV-Vis spectroscopy with a variable temperature stage.
7.1.2. Synthesis of Vanadium Carbide and Vanadium Nitride
Vanadium nitride (VN) and vanadium carbide (VC) are the subject of investigation by materials
scientists due to their exceptional hardness, high melting points, high thermal conductivities
and solid lubricating properties.310 The Vickers hardness of VC311,312 has been measured at 2600
- 3200 kg mm-2 considerably harder than tungsten carbide313 and similar to titanium carbide
(2400 kg mm-2 & 1900 - 3200 kg mm-2 respectively)314 as well as a measured Young’s modulus
comparable with that of tungsten carbide.311,315,316 VN exhibits similar properties, with a Vickers
hardness of 1500 kg mm-2 and a melting point (2619 K) comparable to VC (3103 K).310,317
VN is widely used to harden steel, with the surface of the steel treated in such a way that a thin
layer of VN is present by annealing at high temperatures under a flow of nitrogen. This increases
217
wear resistance for use in high performance steels.318,319 VN is also a strong coupled
superconductor, with potential use in supercapacitors.320
VN is often synthesised by controlled direct nitridation of vanadium containing species using
nitrogen or ammonia gas at varying temperatures. Using this technique, foams of VN were
formed through the nitridation of vanadium oxides with ammonia gas with high control over
morphology.321 Other methods that have proved successful include microwave synthesis,322,323
hydrothermal synthesis324 and high temperature plasma routes.325
VN thin films have been synthesised using molecular precursors with pre formed V-N bonds,
for instance V(NEt2)5, with a carrier gas composed of 10% NH3 in He using AACVD326. VN has
also been produced from a variety of molecular precursors including the direct reaction of
vanadium tetrachloride with sodium amide,327 ammonia328 and hexamethyldisilazane329
vanadium-urea complexes,330 as well as chloroimidovanadium compounds and metal oxide
nanoparticles with cyanamide and urea.331
VC has been shown to be a highly effective additive to tungsten carbide in improving the
hardness and thermal conductivity of highly durable ceramic-metal “cermet” composites,332
and an ideal material for improving the wear resistance of tools.333 Furthermore, precipitation
of vanadium carbide nanoparticles into ferrite-martensite dual phase steel has been shown to
lead to a consistent improvement to Vickers hardness over a range of synthesis conditions. 334–
336 It has been reported that depositing a layer of vanadium carbide onto the surface of high
carbon steel via a salt bath improves its surface hardness by six times.337 Similar results have
been reported for VC coatings deposited onto a die steel substrate via high temperature
reactive diffusion using NH4Cl/ferro-vanadium/naphthalene precursor with surface hardness
improved approximately fivefold.338

218
There are numerous methods for the synthesis of vanadium carbide nanoparticles.
Preparations can involve vanadium(V) oxide (V2O5) with various carbonaceous species at high
temperatures in a reducing gaseous environment.332,339,340 Refluxing V2O5 powder in n-
dodecane has been shown to produce VC nanoparticles.341 Furthermore V2O5 can be
decomposed with magnesium filings and acetone in an autoclave.336,342–344 Nanostructured thin
films of vanadium carbide have been deposited using the CVD of single source precursors such
as vanadocene345 and cyclopentadienyl vanadium tetracarbonyl.346
In this study, the products of the experiments involving VOCl3 and the bidentate ligands 2,4-
pentadione, diethyl malonate, and diethyl succinate (5, 6 and 7) were used as precursors for
vanadium nitride and carbide formation. Each of the precursors was converted to vanadium
nitride and carbide via heating in a furnace under an inert atmosphere at 1200 °C, similar
temperatures to those used in the formation of austenitic steel. 347,348 Conversion to the carbide
or nitride was dependent on the carrier gas used, with argon giving a carbide and nitrogen
predictably producing a nitride species. The conversion of the precursor molecules to vanadium
carbide/ nitride was determined by X-ray diffraction (XRD) (carried out by Dr. Michael Powell),
X-ray photoelectron spectroscopy (XPS), energy dispersive X-ray spectroscopy (EDS) and
transmission electron microscopy (TEM) (carried out by Dr. Joseph Bear).
7.2. Formation of VO2 (M) thin films and VOx Nanoparticles Using the Single Source
Precursor [{VOCl2(CH2(COOEt2)}4] (6)
Solutions of [{VOCl2(CH2(COOEt2)}4] [6] in toluene were used to deposit films across a range of
temperatures, resulting in the deposition of thermochromic VO2(M). Preliminary attempts
were made to deposit in the same way with dichloro(oxo)(diethyl succinate) vanadium(IV) [7]
and dichloro(oxo)(2,4-pentanedione) vanadium(V) [5], however the films deposited by, these
219
precursors were found to be thinner and more uneven than those from 6. This is likely due to
the higher stability of 5 and the lower solubility of 7 in toluene.
7.2.1. Experimental
[VOCl2(CH2(COOEt2)}]4 [6] was synthesised as described in Chapter 3. Oleic acid (techn. grade,
90%), oleylamine (techn. grade, 70%), 1-octadecene (techn. grade, 90%) (dried over sodium)
and 1,2-tetradecanediol (techn. grade, 90%) were purchased from Sigma Aldrich Ltd. and used
as received. Laboratory solvents were purchased from Sigma Aldrich Ltd. and were of analytical
grade. The solvents were dried over activated alumina via the Grubbs method using anhydrous
engineering equipment, ensuring that the concentration of water in the solvents was below 5
– 10 ppm.83 Deuterated chloroform (CDCl3) was obtained from GOSS Scientific and was
degassed and dried over 3 Å molecular sieves.
7.2.2. Deposition of Thin Films
All films were deposited using AACVD. Compound 6 (0.30 g, 0.25 mmol) was dissolved in
anhydrous toluene (20 ml) and transferred to a glass bubbler under nitrogen using a cannula.
The resulting solution was atomized for 5 minutes to ensure complete solvation of the
precursor. An aerosol was generated using a piezoelectric humidifier (Ultrasonic Liquids
Atomizer LIQUIFOG®). The depositions were carried out using a carrier gas of 2% O2 in N2 at a
flow rate of 3 L min-1. A separate 1 L min-1 flow of air was introduce via a second inlet in the
baffle, from the house line fitted with a filter. The glass substrate was standard float glass of 4
220
mm thickness. A top plate of the same glass was suspended 0.5 cm above the substrate to
ensure a laminar flow of the precursor/solvent vapour. Depositions were carried out at 540 °C,
550 °C and 560 °C, each taking approximately 30 minutes.
7.2.3. Tungsten Doping
Tungsten doping was affected by the addition of tungsten(VI) ethoxide [W(OEt) 6] or
tungsten(VI) hexaphenoxide [W(OPh)6] into the precursor solution prior to an AACVD
deposition. The amounts used in each depsotion, as well as the atomic percentage of W present
in the precursor solution, are detailed in Table 19. [W(OEt)6] was purchaseed from Sigma
Aldrich and added unmodifed to the precursor solution. [W(OPh)6] was synthesised using a
literature procedure first published by Cross et al.349
7.2.4. Synthesis of Nanoparticles
Nanoparticle samples were prepared using the thermal decomposition of 6 in the presence of
high boiling point solvents and alkyl surfactants. Compound 6 (0.5 g, 0.41 mmol) was weighed
out into a nitrogen-purged three-necked 250 ml flask with a condensor. 1-octadecene (dried
over sodium) (20 ml), oleylamine, oleic acid and 1,4-tetradecanediol were added in varying
amounts as described in Table 20. The flask was put under vacuum and then back filled with
nitrogen 3x to purge any residual water from the system. Following this the flask was heated
to the required synthesis temperature at a heating rate of 3.3 °C min-1 under a flow of nitrogen.
Once reached, the reaction temperature was maintained for 1 hour before allowing the mixture
to cool to room temperature. The reaction remained under nitrogen until completely cooled.
The nanoparticles were precipitated with ethanol (ca. 100 ml) and centrifuged at 3000 × g. This
was repeated three times in order to ensure maximum purity before the products were stored
221
in a vacuum dessicator overnight to remove any residual moisture. Due to the presence of the
alkyl surfactants, the nanomaterials exhibited a high level of dispersibility in organic solvents
such as chloroform and n-hexane.
7.2.5. Annealing to produce VC + VN powders
Samples were annealed in a tube furnace under a constant argon flow of 0.6 L min -1 for 10
hours at 550 °C at a heating rate of 10 °C min-1 and allowed to cool to room temperature
naturally under argon flow before analysis.
7.2.7. Film deposition
All films were deposited using AACVD, in which the precursor is dissolved in a suitable solvent,
from which an aerosol was generated using a piezoelectric mister. The precursor-containing
aerosol droplets were transported into the reaction vessel via a flow of carrier gas, at which
point the high temperature inside the reactor causes the solvent to evaporate bringing the
precursor in contact with the substrate. The precursor was then free to decompose, nucleate
and grow crystals to form a thin film. In all the depositions, toluene was used as a solvent. The
precursor for both the VO2 film and the tungsten dopant were both dissolved in the same
solution and transported to the reactor using a carrier gas consisting of 98% nitrogen and 2%
oxygen. It was found that in order to encourage film deposition and to attain VO2 (M), a
separate oxygen source had to be introduced in the form of a separate flow of air from the
house line, in order to counteract the reducing nature of an almost total azotic atmosphere.
222
Key depositions using solutions of 6 in toluene are denoted in Table 19. The best results were
found to be at 550 ˚C using a carrier gas of 2% O2 in N2 at a flow rate of 3 L min-1. (experiments
3 + 4). Depositing using the same conditions but with the temperature lowered to 500 ˚C
yielded patchy films whilst at 600 ˚C the temperature was too high encouraging the full
oxidation of the product, giving V2VO5thin films (experiments 1 and 7 respectively). The films
deposited at 550 ˚C were of good coverage and pale yellow in colour, with some darker,
metallic patches formed at higher concentrations of 6. The optimum concentration of 6 was
found to be 0.3 g in 20 ml of toluene, or a concentration of 0.533 mol dm-3 (experiment 4).
Following the optimising of the precursor concentration and successful formation of
thermochromic VO2 (M), depositions were carried out at 540 and 560 ˚C in order to ensure the
process was fully optimised. Thermochromic testing with UV/Vis spectroscopy found neither
film performed as well as the one deposited at 550 ˚C. Films deposited under these conditions,
over the temperature range of 540 to 560 ˚C displayed good optical transparency as well as
durability, passing the Scotch tape test. They were however fairly easy to remove by scratching
with a coin. Films deposited outside this temperature range on the other hand showed poor
adherence and were more powdery.
223
Table 19: AACVD depositions carried out using [VOCl₂(DEM)]₄ [6] with and without the presence of tungsten dopants.
No. Amount Amount Amount of V:W Flow Rate of Temp Observations

W(OPh)₆ proportion carrier (2%O₂ in
of [VOCl₂(DEM)]₄ of W(OEt)₆ (g) N₂) (L min¯¹ ) ˚C
[6] (g) (g)
1 0.5 0 0 0.00 2 500 Patchy + uneven mixed phase film
2 0.5 0 0 0.00 2 550 Thermochromic VO₂ covers half substrate
3 0.5 0 0 0.00 3 550 Thermochromic VO₂ full coverage
4 0.3 0 0 0.00 3 550 thermochromic VO₂ full coverage + film thinner and more
transparent than 3
5 0.3 0 0 0.00 3 540 thermochromic VO₂ full coverage, less thermochromic than 4
6 0.3 0 0 0.00 3 560 thermochromic VO₂ full coverage, less thermochromic than 4
7 0.3 0 0 0.00 3 600 Full coverage of black V₂O₅
8 0.3 0.032 0 0.066 3 550 Blue film, not thermochromic
9 0.3 0.022 0 0.045 3 550 Yellow VO2 film. thermochromism not observed
10 0.3 0.028 0 0.058 3 550 Weakly thermochromic film Tc = 60˚C
11 0.5 0 0.025 0.019 3 550 Termochromic film Tc = 60˚C
12 0.5 0 0.018 0.014 3 550 Yellow VO2 film. thermochromism not observed
13 0.5 0 0.03 0.023 3 550 Thermochromic film Tc = 40˚C
14 0.5 0 0.05 0.038 3 550 Thermochromic film Tc = 30˚C
224
Tungsten was doped into the films over a range of concentrations (Table 19) by adding small
amounts of tungsten(VI) ethoxide [W(OEt)6] or tungsten(IV) hexaphenoxide [W(OPh)6] into
the precursor solution. The concentration of [W(OEt)6] as a percentage of 6 present in the
precursor solution ranged from 4.5 to 6.6%, whereas the concentration of [W(OPh) 6] in the
solutions ranged from 1.8 to 5.0%. The W doped films each had greater optical transparency
than the undoped films deposited at the same temperature, with colours ranging from pale
yellow to azure, becoming more blue with increased concentration. All W doped films were
fairly adherent, resisting the “Scotch Tape” test.250,350 [W(OPh)6] was observed to be the
superior tungsten dopant precursor of the two, as the MST was found to be far less sensitive
to the presence of the precursor, giving a ‘larger window’ with which to alter the
concentration.
The thermochromic transition (MST) was observed to decrease linearly from 60 ˚C to 30 ˚C
as the amount of [W(OPh)6] added to the solution increased from 0.025 to 0.05 g. The use of
[W(OEt)5] resulted in not enough dopant present at 0.022 to having far too much at 0.032 g,
which made finding the ideal amount highly challenging. This is partially due to the far greater
molecular weight of [W(OPh)6] than [W(OEt)6] (746 g mol-1 and 454 g mol-1 respectively),
meaning that the amount of actual tungsten per gram of precursor is lower allowing for more
control. [W(OPh)6] is also more soluble in toluene, allowing for better incorporation into the
VO2 film and therefore resulting in more even coverage as well as being a more reliable
precursor with greater reproducibility.
SEM images were obtained for films of VO2 deposited at 550˚C, both with and without the
incorporation of tungsten into VO2 film. The W-free film consisted of highly disordered rod-
like particles of width approximately 50 nm and length 300 - 500m (Figure 113a).
225
Figure 113: SEM images of the VO2 film deposited from [6] in toluene at 550 °C using a 2% O2 in N2
carrier gas with no dopant (a), 2.8% [W(OEt)6] (b) and 5.0% [W(OPh)6 ] (c).
226
Figure 114: XRD patterns of the VO2 film deposited from [6] in toluene at 550 °C using a 2% O2 in N2
carrier gas, both with and without 5.0% [W(OPh)6].
Figure 115: XPS scans of: a) V 2p3/2 of a VO2 (M) thin film from the AACVD of [{VOCl2(CH2(COOEt2)}4] [6]
at 550 ˚C (showing V(IV) at lower binding energy (thin solid line) vs. V(V) (dotted line)), b) V 2p3/2 of a W-
doped VO2 (M) thin film (film 12) at 550 ˚C and c) corresponding W4f scan.
227
Incorporation of tungsten into the film appears to greatly alter the morphology, with the film
depositing densely packed, rounded, vertical protrusions with a width of around 100 nm (Figure
113b). This is in keeping with the more smooth and even appearance of the doped films.
XRD patterns were referenced against those of VO2 (M) (ICSD: 34033), VO2 (A) (ICSD: 51213),
VO2 (B) (ICSD: 199) V2O5(ICSD: 15798) and V2O3 (ICSD: 1473). Films deposited at 550°C were
found to be exclusively VO2 (M) as seen in Figure 114. Tungsten incorporation induces a slight
shift in the XRD pattern (27.8° for undoped VO2 versus 27.9° for W-doped) and a loss of
crystallinity. This, coupled with the lowering of the MST is highly indicative of successful doping.
Typical XPS data from VO2 (M) and W-doped VO2 (M) thin films are shown in Figure 115. The V
2p spectra are split into two regions which were then fitted with two separate environments
as it was clear that more than one oxidation state was present. The visible shoulders of the
main V 2p3/2 peak indicate the presence of V(IV) in the films. XPS is highly surface sensitive, and
the VO2 on the surface of the films had fully oxided, creating a thin layer as V 2O5, as a higher
proportion of V2O5 was present at the surface than in the film, hence a large V(V) peak. The
fitted peak positions of 516.5 eV and 516.7 eV for V 2p 3/2 in the undoped and doped films
respectively are indicative of V(IV) in VO2.44 The second feature at higher binding energy in both
spectra (518.1 eV and 518.2 eV respectively) can be attributed to V2O5. 44
The W 4f environment showed the presence of only one oxidation state. This gave a value of
35.28 eV (W 4f7/2) and 37.38 eV (W 4f5/2) which is consistent with W(VI). This has been
previously seen for W-doped VO2 in both thin film and nanoparticle synthesis.275, 283, ,351
Variable temperature transmission UV/Vis spectroscopy was carried out on the films deposited
from 6 in order to determine whether the films underwent the structural transition associated
228
with VO2 (M) (Figure 6). A spectrum was taken of the film at room temperature. Following this
the films were rapidly heated to 90 ˚C and the measurement repeated. The films were then
cooled to room temperature and the initial measurement repeated to ensure that the sample
has remained in the same position throughout the heating as heating the ample can cause the
sample to slip downward in the heater cell, which would affect the UV/Vis data.
Figure 116: Variable temperature UV/Vis/NIR spectra showing the change in optical properties of thin
films of undoped and W-doped VO2 thin films deposited by AACVD of synthesised vanadium precursor
solutions. Solar modulation values (ΔTsol) for each film have been included in the Figure. a) VO2 (M)
film deposited at 540 ˚C, b) 550 ˚C, c) 560 ˚C and d) W-doped VO2 (M) thin film deposited at 550 ˚C.
Films deposited at 550 ˚C showed by far the highest MST, with the transmittance at 2500 nm
falling from 60% to around 2% when the sample was heated from room temperature to 90 ˚C.
Films deposited at 560 ˚C demonstrated similar characteristics at room temperature to those
deposited at 550 ˚C. However heating the samples reveals that the monoclinic phase of the VO2
229
is less pure, roughly halving the transmission to 30%. The VO2 film deposited at 540 ˚C was the
least effective by a considerable margin, seeing a reduction in transmission from 50% to 40%
at 2500 nm. These measurements reveal that the optimum temperature for the formation of
monoclinic VO2 from 6 was 550 ˚C.
Solar modulation calculations (ΔTsol) were performed as described by Taylor et al.352 Solar
modulation values take into account the effect that gases, such as CO 2 and H2O, have on the
intensity of solar radiation at ground level. This gives a value which can be used to give a better
representation of the absolute change in the optical properties of the material. The majority of
the energy in the solar spectrum, at ground level, is accounted for by the UV/Visible region
(380-780 nm); this leaves a maximum change of ca. 20% for the near IR region (780-2500 nm).
When comparing the solar modulation values for the VO2 films, it can be seen that the sample
deposited at 550 °C shows a solar modulation of 15.9% which is close to the maximum allowed
solar modulation and is comparable to the best results seen for VO2 thin films by other research
groups.353
7.2.8. VOx Nanoparticle Synthesis and Characterisation
[{VOCl2(CH2(COOEt2)}4] 6 was additionally used to produce nanoparticles, via thermal
decomposition in a high boiling point solvent with structure-directing surfactants under an
atmosphere of nitrogen. This follows the standard procedure for many nanoparticle synthesis
in which a long chain unsaturated alcohol is coordinated to a metal species, which is then
decomposed at high temperatures to form surfactant capped nanoparticles that are easily
dispersed in organic solvents.354–356 Paik et al. has used a similar methodology to synthesise VOx
nanocrystals from VOCl3 in the presence of oleylamine and octadecanol, resulting in mixed-
phase VOx nanocrystals.357 In that case the synthesised particles were flash-annealed at 500 ˚C
230
for 5 minutes in air at 1 mTorr in order to convert to thermochromic VO2 (M). The reacton is
facilitated by the reaction of VOCl3 with the 1-octadecanol, allowing for the formation of
vanadium-oxygen bonds.358,359 In this synthesis, alcohol is absent, the V-O bonds instead were
provided by the diester species.
The decomposition of 6 in the presence of surfactants in a high boiling point solvent was carried
out repeatedly, differing the ratios of the two surfactants, as well as the temperature and the
inclusion of a fatty alcohol, 1,4-tetradecanediol (Table 20). All reactions were undertaken in 20
ml of dry 1-octadecene (dried over sodium) and after 3 vacuum/re-fill purge cycles with
nitrogen. The reaction was heated from room temperature to the target temperature at a rate
of 3.3 °C min-1 under dynamic nitrogen, during which the reaction mixture was observed to
darken in colour gradually from green to black. The addition of oleylamine also induced a colour
change of the precursor to blue, presumably due to complexation with the vanadium centre,
as a probable result of a d→d transition due to the formation of an octahedral complex.
Following subsequent cooling to room temperature, the particles were precipitated with
ethanol and separated by centrifugation. All samples gave black precipitates that were readily
dispersable in organic solvents, with the exception of sample G.
TEM analysis clearly demonstrates the extent to which the nanoparticle morpholgy is effected
by the ratios of surfacant used. Oleylamine was observed to have the most pronounced effect,
promoting the growth of spine-like structures as well as nanoparticles which were not seen in
its absence (Figure 7a and 7b). Oleic acid purely promoted particle growth only (Figure 7g and
7h). Mixtures of oleic acid and oleylamine were found to be the most effective at generating
readily dispersible products, with the optimum blend of equimolar amounts of oleic acid and
oleylamine a commonly used in iron oxide nanoparticle syntesis.360
231
Table 20: Compiled reaction conditions for the formation of VOx
nanoparticles using [6] (*No reaction, non-colloidal blue solution formed.
ǂAddition of 1,4-tetradecanediol).
Synthesis Oleylamine Oleic acid
Sample temperature °C mmol mmol
A 320 6 6
B 320 9 3
C 320 3 9
D 320 0 12
E 320 12 0
F 280 6 6
G* 240 6 6
Hǂ 320 6 6
In samples A, F, G and H, the “6:6“ blend produced a mixture of mainly spine-like structures of
50-100 nm in length with some round particles also present. These features were preserved on
annealing in nitrogen for 8 hours, but were not as pronounced in sample F, in which the
synthesis temperature is lower (280 °C), suggesting a degree of temperature dependance on
the formation of elongated spines as well as surfactant. However, the features re-emerged with
annealing treatment at 550 °C under nitrogen for 10 hours (Figure 117d) and e)). Sample G, in
which the synthesis temperature was reduced to 240 °C was evidently lower than the
nucleation temperature for this system as no nanoparticles were produced.
It proved demanding to gather detailed TEM analysis of the as-synthesised VOx nanostructures,
in particular the lattice plane imaging; due to the presence of high carbon contamination,
caused by surfactant layers and decomposition products from the precusor itself. The latter
was in evidence through the presence of chlorine by EDS (Figure 117, c) and f).
232
Sample A and F were both annealed in order to induce the formation of VO2 (M). However,
analysis of lattice planes in both of the the annealed samples (Figure 117e) gave a d-spacing of
0.326 nm corresponding to the <111> plane of vanadium oxide (V4O9, ICSD 15041), confirming
that heat treactment had an identical effect on both samples despite intial differences with
synthesis temperatures. It also suggests that the samples were slightly over-oxidised in the
annealing process when trying to synthesise VO2 (M).
The addition of an oxygen source, 1,4-tetradecanediol in sample H with a „6,6“ blend of oleic
acid and oleylamine did not induce a phase transformation to VO2 (M). Instead , the addition
of 1,4-tetradecanediol prevented the structure-directing effects of oleylamine from taking
effect, giving almost exclusively particulate material (Figure 117i).
EDS spectra showed the persistance of chlorine in the system as a consequence of the
precursor. It also showed the reduction in carbon from the heavily carbon contaminated pre-
annealed samples versus post annealed samples. Atomic percentages of vanadium and oxygen
in the samples are shown in able 3. The at% ratios of V:O for sample A and F pre-anneal was
26.20 : 73.80 and 1.91 : 98.09 respectively, with heat treatment dramatically effecting sample
F with the ratio changing to 26.22 : 73.78. The ratio for sample A however, was relatively
unchanged (26.40 : 73.60). The very high ratio of oxygen to vanadium in sample G can be
attributed to the high levels of carbon contamination obtained from the presence of the
surfactants.
233
Figure 117: TEM images and EDS spectra of sample A, pre-annealing (a and b) and post annealing (d
and e). C and G are the EDS spectra of the pre-annealed and post annealed sample 1 respectively.
Also shown are TEM images of samples B, C and H (g, h and i repectively).
The lack of phase transition to VO2 (M) with thermal treatment at elevated temperatures was
confirmed by XRD analysis. The XRD pattern did not match the standard of VO2 forming a
different species. Once again this is likely the result of the high level of carbon contamination
from the remnants of the oleic acid/oleylamine ligand system.
234
Table 21: EDS at% ratios of vanadium to oxygen in nanomaterial samples.
Bracketed quantities are those in the annealed spectra.
Sample number V / at% O / at%
A 26.2 (26.4) 73.8 (73.6)

B 9.66 90.34
C 1.82 98.18
D 0.18 99.82
E 28.95 71.05
F 1.91 (26.22) 98.09 (73.78)
H 10.21 89.79
Figure 118: PXRD diffraction patterns for as synthesised powders, a) and b) and following annealing
treatment, c) and d). λ= 0.7093 Å.
Powder X-ray Diffraction (PXRD) was performed on the as-synthesised and the post-annealed
nanoparticle samples (Figure 118). The as-synthesised nanoparticles, Figure 118 a) and b), have
only a few diffraction patterns and it was not possible to confirm the phase of the material from
235
these. The broad reflections between 5-12° 2θ also suggest that there was a large amorphous
component to the samples.
The post-annealed samples, Figure 118c and d), have a greater number of more intense
diffraction peaks in the data, with less broadening suggesting that the annealing has resulted
in improved crystallinity. Phase analysis of these samples suggested a mix of Magneli phases,
with V3O7 – 22.2° (0,2,0), V4O9 – 11.1° (1,0,2), 16.5° (1,1,3) and V5O9 – 5.2° (0,0,1), 24.1° (2,1,1)
and 28.8° (2,0,3) lattice planes all being identified in the data. The formation of multiple
oxidation states in annealed vanadium samples is relatively commonplace.360,361
Figure 119: XPS scans of: a) V 2p of Sample 1, VO2 nanomaterials before annealing and b) after
annealing showing V(IV) at lower binding energy (thin solid line) vs. V(V) (dotted line).
X-ray photoelectron spectrcopy was also used to compare the samples before and after
annealing. The XPS spectra of the V 2p region in Sample 1 pre (Figure 119a) and post (Figure
119b) anneal show similar profiles to the thin films in Figure 115, with V(IV) peaks reported at
516.5 eV and 516.3 eV for non-annealed and annealed samples respectively (V 2p 3/2) and
assigned as VO2. The V(V) peaks, at 517.7 eV and 517.8 eV respectively were again assigned as
V2O5.362
236
7.3. Using adducts of VOCl3 and 2,4 pentadione, diethyl malonate and diethyl
succinate as precursors for Vanadium Nitride and Carbide
7.3.1. Experimental
All precursors were synthesised as described in Chapter 4. All other starting materials were
purchased from Sigma Aldrich and used without further purification. The solvents were dried
over activated alumina by the Grubbs method using anhydrous engineering equipment, such
that the water concentration was 5 – 10 ppm.83 All products were synthesised under an
atmosphere of nitrogen obtained from BOC in anhydrous solvents using standard Schlenk
techniques.
7.3.1.2. Synthesis of vanadium nitride and carbide powders
Vanadium nitride and carbide powders were synthesised by weighing out ~ 0.2 g of compounds
5, 6 and 7 in a nitrogen filled glove box onto ceramic bricks. The samples were sealed with a
layer of laboratory film before removal from the glovebox. The samples were then transferred
immediately to a tube furnace. The samples were purged with a flow of nitrogen (for vanadium
nitride, BOC, 99.9%) or argon (for vanadium carbide, BOC “pureshield” 99.9%) at room
temperature for 30 minutes (flow rate 40 cm min-1) before heating to 1200 °C (heating rate 20
°C min-1) for 24 hours. The samples were then allowed to cool to room temperature under the
flow of inert gas.
237
Compounds 5, 6 and 7, as synthesised in chapter 4 were heated to temperatures akin to those
used in steel tempering under oxygen free flows of inert gases nitrogen and argon in order to
obtain VN and VC powders respectively. These powders were analysed by XPS, XRD and TEM in
order to deduce their composition and particulate size.
7.3.2.1. Synthesis And Characterisation Of Vanadium Nitride Powders
Vanadium nitride powders were synthesised by annealing compounds 5, 6 and 7 under nitrogen
gas flow at 1200 °C for 24 hours. It is necessary to treat vanadium species at such high
temperatures to illicit the formation of vanadium nitride due to the intrinsic stability of various
vanadium oxides such as V2O5.
It was initially the intention of the experiment to produce VC powder from the decomposition
of the precursors, with the ligands acting as the carbon sources, with probable inclusion of
some carbon species from the lab film. However compounds 5, 6 and 7 showed near complete
transformation to vanadium nitride at 1200 °C as evidenced by the XRD patterns in Figure 120a.
The XRD patterns were compared to a VN standard (ICSD 22321), all samples were shown to
be phase pure with only diffraction peaks for VN visible. Annealing at the lower temperature of
1000 °C for 24 hours yielded mixed phases with some VN so it was necessary to perform the
reaction at 1200 °C in order for the reaction to reach completion. Yu et al.363 determined the
reaction mechanism for the reaction of vanadium(III) oxide (V2O3) and carbon in the presence
of nitrogen gas at elevated temperatures (up to 1180 °C). Nitridation was simultaneous with
carbothermal reduction, from V2O3 to V8C7 to VN, and was completed at lower temperatures
than just V2O3 in nitrogen. Compounds 5, 6 and 7 were able to decompose in a similar way to
238
this under a nitrogen gas atmosphere due to the presence of oxygen and carbon in the
precursor compounds.
Figure 120: Composite Figure for the VN sample derived from complex 6. a) shows XRD patterns of VN
derived from complex 6 compared to a VN ICSD standard (22321). b)shows an EDS spectrum
demonstrating the presence of vanadium and nitrogen with little carbon present. Copper emanated
from the copper mesh TEM grid. b) Fitted N1s XPS spectrum showing the regions assigned as VN and
VC. ) d) is an HRTEM image of a VN crystal with the <111> plane of VN highlighted.
239
Ortega et al.364 followed the progress of the reaction of by measuring the partial pressures of
evolved carbon monoxide, supporting the V2O5–V2O4-VO2–V2O3–VO0.9-VN pathway, despite the
difficulty in the detection of V0.9.
XPS analysis, Figure 120b) showed the presence of vanadium nitride in both the N 1s and
V 2p3/2 at 397.2 and 513.6 eV respectively. These match with literature values.365 The XPS also
showed evidence for vanadium carbide and oxidised vanadium(V) The carbide was likely due
to excess carbon presence in the precursor and the V(V) signal is typical for vanadium, as the
vanadium at the surface will readily oxidise when exposed to air.366
TEM analysis of all VN samples showed the formation of large, polydisperse crystallites of VN
(42.9 ± 17.0 nm (sample derived from 5), 88.4 ± 47.7 nm (sample derived from 6) and 72.4 ±
71.4 nm (sample derived from 7). Small amounts of crystalline carbon was also present around
the edges of the VN crystallites, resulting from the film used to seal the samples burning within
the furnace. HRTEM analysis of all samples demonstrated the presence of VN, via analysis of
the d-spacings. Several d-spacings were indexed (Figure 120 d) with the <111> plane of VN
present and identifiable in all samples, with d-spacings from the samples derived from
compounds 5,6 and 7: 0.24 nm, 0.2 nm and 0.23 nm respectively.
EDS analysis indicated the presence of nitrogen and vanadium with low amounts of carbon
(Figure 120 a) and in all samples analysed. Quantitative EDS analysis in all samples showed a
ca. 50:50 at% ratio, indicative of the formation of VN. The presence of residual carbon species
presented difficulties in integrating the nitrogen peak as their EDS fingerprints appear in the
same region of the spectrum. The relatively low amount of carbon in the sample and the large
VN crystallites allowed for accurate confirmation of a 50:50 at% ratio of V:N, and therefore
conformation of the VN empirical formula.

240
7.3.3. Synthesis and characterisation of vanadium carbide powders
Vanadium carbide powders were synthesised by annealing compounds 5, 6 and 7 under an
argon flow at 1200 °C for 24 hours, yielding a black powder. The VC derived from compound
showed excellent conversion as evidenced by XRD analysis in Figure 121a). The samples
produced were phase pure, with only diffraction peaks for VC present in the patterns compared
to a VC ICSD standard (159870). Powders derived from 5 and 7 show evidence of a partial
transformation to VC, however phase purity could not be attained. Figure 121 d shows an
HRTEM image of a VC crystal with the <111> plane of VC highlighted.
XPS analysis (Figure 121c) showed the presence of vanadium carbide in both the C1s and V 2p3/2
at 282.8 and 513.7 eV respectively. These are in agreement with literature values.367 The C1s
showed several other environments, which were matched to C-C, C-O and C-OR- with these
environments being more intense than the V-C signal. This was also reflected in the V 2p signal,
which was significantly weaker than the C1s, this was attributed to the formation of a carbon
‘shell’ around the VC particles during annealing.
TEM analysis of all VC samples showed the formation of large, polydisperse crystallites of VC
(42.9 ± 17.0 nm , 88.4 ± 47.7 nm and 72.4 ± 71.4 nm from samples derived from 5, 6 and 7
respectively. These are similar distributions to those obtained for VN. A large amount of
crystalline carbon was detected in these particles, moreso than in the VN equivalents.
241
Figure 121: Composite Figure for the VC sample derived from 6. a) shows XRD patterns of VC derived
from 6 b) shows an EDS spectrum demonstrating the presence of vanadium and high levels of carbon.
copper emanated from the copper mesh TEM grid. c) shows the fitted V 2p XPS spectrum showing the
regions assigned as VC and VO.d) is an HRTEM image of a VC crystal with the <111> plane of VC
highlighted.
242
HRTEM analysis of all samples demonstrated the presence of VN, via analysis of the d-spacings.
Several d-spacings were indexed (Figure 121 e) and Figure S2 b) and d)), with the <111> plane
of VC identified in all samples, with d-spacings from the samples derived from compounds 5, 6
and 7 : 0.22 nm, 0.22 nm and 0.23 nm respectively.
EDS analysis indicated the presence of carbon and vanadium with high amounts of carbon
(Figure 121b) in all samples analysed. Quantitative EDS analysis showed a small difference in
each of the samples. The sample derived from 5 showed a ca. 95:5 at% ratio of C:V, the sample
derived from 6 a ca. 90:10 at% ratio, and the sample derived from 7 a ca. 85:15 at% ratio. The
carbon film on the TEM grid skews the ratio in favour of carbon, but the carbon-rich nature of
the structures produced was confirmed by quantitative XPS giving ratios of 94.3 carbon : 5.7
vanadium for the sample derived from 5.
7.4. Conclusions
[{VOCl2(CH2(COOEt2)}4] (6) has shown synthetic verstility as a precursor for the AACVD of
vanadium oxide thin films and nanocrystals. Firstly, 6 was shown to be a highly effective
precursor for the formation of thermochromic VO2 (M) thin films onto glass by AACVD. VO2 (M)
films produced at 550 ˚C were shown to have superior solar modulation when compared to
films produced at 540 ˚C and 560 ˚C and compare favourably with literature values (15.9%).
This suggests that although a very effective precursor for themochromic VO2(M), the window
for such characteristics is very narrow, with the purity of the monoclinic phase being greatly
reduced by varying temperature by only 10 ˚C in either direction. The thermochromic switching
temperature was reduced through the addition of W(OEt)6 or W(OPh)6 to the precursor
solution, reducing the MST to ~30 ˚C. This demonstrates the viability of the AACVD method
243
when using a designed molecular precursor to form thermochromic materials with tunable
properties.
[{VOCl2(CH2(COOEt2)}4] (6) was also decomposed in the presence of alkyl surfactants at
elevated temperatures in an inert atmosphere. It was shown that different surfactants affected
the shape of the final product, with oleylamine promoting "spine-like" growth, and oleic acid
particulate material. Although the heat-induced phase tansformation to VO2 (M) was ultimately
unsuccessful, control over shape and size of the resultant nanomaterials via manipulation of
surfactant type and concentration in the decomposition of [6] was deonstrated.
Complexes 5, 6 and 7 were also evaluated for their propensity to form VN and VC on annealing
at 1200 °C under nitrogen and argon gas respectively. All samples converted to VN at 1200 °C
under nitrogen, but complex 6 was the only molecule shown to be a viable precursor for both
VC and VN.
244
Chapter 8
Conclusions
This thesis tackles the industrial problem of metal chlorides separation in a number of ways
using laboratory scale chemical techniques, some of which were developed purely for the
study, some of which were well established. The reactivity of TiCl4 and VOCl3 with ester groups
was analysed in the solid, liquid and gas phases separately. The results of these studies,
corroborate on several points, although the techniques involved vary greatly.
Firstly the solid state phase, TiCl4 and VOCl3 were reacted with diesters and triesters. The
resulting compounds were characterised using NMR spectroscopy and X-ray crystallography.
This lead to the discovery that coordination to a diester species results in the loss of chlorine
from the vanadium centre but not the TiCl4 all cases. It was inferred fron this that although the
TiCl4 may coordinate in the gas phase to the triglyceride group of the oil, the reaction was so
reversible in the gas phase that it would readily able to break away from the molecule, in a way
that the VOCl3 could not, explaining the preferential reactivity.
The flow cell experiment was a liquid and gas phase study, focussing not on the separation
itself, but on potential side products. The gaseous TiCl4, VOCl3 or mixture was pushed through
the chamber with the agitated oil, the gaseous species interacting with the droplets kicked up
before distilling into the top of the apparatus. Analysis of the resulting product lead to the
discovery that the TiCl4 does react with the glyceride group, chlorinating the species by
displacing the two side ester groups, leaving the central intact. This is corroborated by the
glycerol tribenzoate TiCl4 complex [4], which shows the two side groups coordinated to the
same titanium centre, with a second titanium atom coordinating to the central ester groups of
245
a different molecule. This reaction would be unfavourable in the gas phase and so it is likely
that the reaction simply reverses, however in some cases a titanium centre is able to undergo
exchange with the glycerol, coordinating resulting in a species that has a low enough boiling
point to distil over. This explains the issues that occur later in the plant with furnace fowling, as
well as why studies by Huntsman Pigments and Additives found that changing the oil did not
alter the problem, as all of the oils tested were triglyceride based.
The final study involved in situ gas spectroscopy measurements. The finding of this work
appeared to correlate well with the initial experiments. Although the ester used had only one
carbonyl group present it was observed that it took coordination of two to a single titanium
centre to result in chlorine loss, where as with VOCl3 the relationship between ethyl acetate
coordination and chlorine loss is 1:1. It should be noted that the high temperatures and the in-
situ nature of the study means that the species formed should be considered intermediates.
This work has therefore shown that the gas phase reactions of TiCl4 and VOCl3 in the Chloride
Process can be modelled using simpler structures, giving the same conclusions in the solid and
gas phases.
Future work in this field would likely involve broadening the scope of the mass spectrometer
CVD apparatus to investigate other common inorganic reactions. The most effective precursors
in the field would be those with as few atoms present as possible, and also with only one type
of ligand, however the results with titanium isopropoxide and butyltin trichloride would
suggest that more complex precursors can be studied successfully. Species such as aluminium
oxychloride, trimethylindium, trimethylgallium and ferrocene would all be promising
candidates.
246
Another possible study could involve changing the oxidation source. Ethyl acetate worked well
in this study due to its similarity to part of the oils used in the Chloride Process but simpler
oxygen sources, such as water and oxygen itself, could also be investigated if the amounts
entering the system could be monitored and controlled closely. A separate study on the effect
of moisture on SiO2 film growth carried out using the same instrument and reactor proved
highly successful.
247
References
1. Mcnulty, G. S. Production Of Titanium Dioxide. in Proc. Conf. Seville, Spain 1–16 (2007).
2. Enghag, P. Encyclopedia of the Elements Technical Data - History - Processing. 501 (Wiley,2008).
3. Winkler, J. Titanium Dioxide: Production, Properties and Effective Usage, Vincentz Network
GmbH & Co. 11 (2013).
4. Moellmann, J., Ehrlich, S., Tonner, R. & Grimme, S. A DFT-D Study Of Structural And Energetic
Properties Of TiO2 Modifications. J. Phys. Condens. Matter 24, 424206 (2012).
5. Winkler, J. Titianium Dioxide, 15 (Vincentz Network GmbH & Co, 2003).
6. Daoud, W. A . Self-Cleaning Materials and Surfaces: A Nanotechnology Approach, 45, (John
Wiley & Sons, 2013).
7. Ye, L. The Total Amount Of Global TiO2 Market Value Will Reach USD 31.19 Billion In Year 2022.
(Pulse, 2016).
8. Hornbostel, C. Construction Materials: Types, Uses and Applications, 883, (Wiley, 1991).
9. Benvenuto, M. A. Industrial Inorganic Chemistry, 883, (Walter de Gruyter GmbH & Co,2015).
10. Merchant Research And Consulting ltd. Titanium Dioxide: 2016 World Market Outlook and
Forecast up to 2020. (2016).
11. Wicks, Z. W. J., Jones, F. N., Pappas, S. P. & Wicks, D. A. Organic Coatings - Science And
Technology, Vol. 5. Progress in Organic Coatings 13, 38 (2007).
12. Mills, A., Elliott, N., Parkin, I. P., Neill, S. A. O. & Clark, R. J. Novel TiO2 CVD Films For
Semiconductor Photocatalysis. J. Photochem. Photobiol. A, 151, 171–179 (2002).
13. Kumar, C. S. S. R. Nanocomposites, 11, (John Wiley & Sons, 2010).
14. Parkin, I. P. & Palgrave, R. G. Self-cleaning coatings. J. Mater. Chem. 15, 1689 (2005).
15. Cai, Y. Titanium Dioxide Photocatalysis in Biomaterials Applications. (2013).
16. Dunnill, C. W. H. et al. Enhanced photocatalytic activity under visible light in N-doped TiO2 thin
films produced by APCVD preparations using t-butylamine as a nitrogen source and their
potential for antibacterial film, J. Photochem. Photobiol. A, 207, 244 (2009).
17. O’Regan, B. & Grätzel, M. A Low-Cost, High-Efficiency Solar Cell Based On Dye-Sensitized
Colloidal TiO2 Films. Nature 343, 737–740 (1991).
18. Eric R. Force, Geology of Titanium-Mineral Deposits. GSA Special Papers, 259, (1991).
19. Breen, G. T Mineral Sands Update Price Revisions - A Turning Point & Time To Gain Exposure To
The Sector The Dark Clouds Begin To Lift. (2013).
20. Sahu, K., Thomas, A., Devabrat, M. & Archana, A. An Overview On The Production Of Pigment
Grade Titania From Titania-Rich Slag. Wate Manag. Res 24, 74–79 (2006).
21. Craver, C. & Carrahe, C. Applied Polymer Science: 21st Century. 485 (Elsevier, 2000).
22. Kawaja, S. TiO2 -2016 Outlook - Are Prices At Rock Bottom? (Tize, 2015).
23. Saklatwalla, B., Dunn, H. & Marshall, A., Sulfate Proces Of Making Titianium Dioxide. (1932).
24. Shreve, R. N. & Austin, G. T. Shreve’s Chemical Process Industries. 434 (McGraw-Hill, 1984).
25. Ti-Cons Ldt., Sufate Process Overview, (2009).
26. Cody, C., Reichert, W., Kemnetz, S. & Magauran, D. Process For Preparing High Quality Titnaium
Dioxide, US4505886 A (1966)
27. Goldschmidt, A. & Streitberger, H.J. BASF Handbook on Basics of Coating Technology. (William
Andrew, 2003).
28. Nosaka, Y. & Nosaka, A. Introduction to Photocatalysis - From Basic Science to Applications.
10–12 (RSC Publishing, 2016).
29. Parker, R. Greatham Works Process Manual ICON 1 Chlorination Section Contents, (Huntsman
Tioxide, 2001)
30. Boyd, C. . Purification Of Crude Titanium Tetrachloride With White Mineral Oil, US3253885 A
(1966).
31. Cronin, J. , Elkins, T., Helberg, L. & Merkle, J. Purification Of Titanium Tetrachloride. US2600881
248
A (2004).
32. Glaeser, H. H. Removal of AlCl3 from TiCl4 by addition of H2O and NaCl US4125586 A (1978).
33. Sutcliffe, S. Best Operating Practice Guide Titanium Tetrachloride Purification. 1–23 (Huntsman
Tioxide, 2008).
34. Cristal Performance Chemicals. TiOCl2.(H2O)x.(HCl)y MSDS. (2016).
35. Coates, W. Improvements in or relating to the Purification Of Titainum Tetrachloride by
Distillation, US3156630 A (1954).
36. Haddow, A. Oxidation Of Titanium Tetrachloride to Form Titanium Dioxide. US5599519 A
(1995).
37. Tolinski, M. Additives for Polyolefins Getting the Most out of Polypropylene. 141 (Elsevier,
2009).
38. Image Of Rutile Sand as sold by Tianjin Wangda Chemical Co. Ltd.
http://www.ecvv.com/product/3831604.html accessed 30/01/2017
39. Image of Purified TiO2 as sold by SAAQIN ltd. https://www.amazon.com/Titanium-Dioxide-
Powder-Lb-SAAQIN/dp/B004W0EQB8 accessed 30/01/2017
40. Gunstone, F. Fatty Acid and Lipid Chemistry. (Springer, 1996).
41. Reck, E. & Richards, M. Titanium dioxide — Manufacture, environment and life cycle analysis:
The Tioxide experience. Surf. Coatings Int. 80, 568–572 (1997).
42. Müller, U. Inorganic Structural Chemistry. (John Wiley & Sons, 1994).
43. Hooper, A. B. & Lipscomb, J. D. Resolution of multiple heme centers of hydroxylamine
oxidoreductase from Nitrosomonas. 2. Mössbauer spectroscopy. J. Am. Chem. Soc. 42, 3973–
3976 (2002).
44. Sutton, L. Tables of Interatomic Distances and Configuration in Molecules & Ions (supplement).
(The Chemical Society, 1956).
45. Webb, S. P. & Gordon, M. S. Intermolecular Self-Interactions of the Titanium Tetrahalides TiX4 (
X = F , Cl , Br ). J. Am. Chem. Soc. 4, 2552–2560 (1999).
46. Thomas, T., Pugh, D., Parkin, I. P. & Carmalt, C. J. Syntheses, X-ray structures and CVD of
titanium(IV) arsine complexes. Dalt. Trans. 39, 5325–31 (2010).
47. Thomas, T., Pugh, D., Parkin, I. P. & Carmalt, C. J. Novel ion pairs obtained from the reaction of
titanium(IV) halides with simple arsane ligands. Acta Crystallogr. Sect. C Cryst. Struct. Commun.
67, 96–99 (2011).
48. Kim, C. S., Nakaso, K., Xia, B., Okuyama, K. & Shimada, M. A new observation on the phase
transformation of TiO2 nanoparticles produced by a CVD method. Aerosol Sci. Technol. 39, 104–
112 (2005).
49. Hyett, G., Green, M. A & Parkin, I. P. The use of combinatorial chemical vapor deposition in the
synthesis of Ti(3-δ)O4N with 0.06 < δ < 0.25: a titanium oxynitride phase isostructural to
anosovite. J. Am. Chem. Soc 129, 15541–8 (2007).
50. Smith, M. B. Organic Synthesis. (Academic Press, 2011).
51. Chadwick, J. C., Duchateau, R., Freixa, Z. & Leeuwen, P. W. N. M. van. Homogeneous Catalysts.
(John Wiley & Sons, 2011).
52. Malpass, D. B. Introduction to Industrial Polyethylene: Properties, Catalysts, and Processes.
(John Wiley & Sons, 2010).
53. Natta, G. Stereospecific polymerizations by means of coordinated anionic catalysis. J. Inorg.
Nucl. Chem. 8, 589–611 (1958).
54. Cossee, P. Ziegler-Natta catalysis I. Mechanism of polymerization of alpha;-olefins with Ziegler-
Natta catalysts. J. Catal. 3, 80–88 (1964).
55. Anslyn, V. A. & Dougherty, D. A. Modern Physical Organic Chemistry. (University Science Books,
2006).
56. Prakash, S., Tuli, G., Baku, S. & Banda, R. Advanced Inorganic Chemistry, Volume 2. 352 (S.
Chand, 2000).
57. Pelletier, E. & Harrod, J. F. Cyclosiloxanes as Frameworks for Multimetallic Compounds. 3. ‘H
NMR Spectra of Some Substituted Methylcyclosiloxanes’. Organometallics 3, 1070–1075
249
(1984).
58. Gottlieb, H. E., Kotlyar, V. & Nudelman, A. NMR Chemical Shifts of Common Laboratory
Solvents as Trace Impurities In the course of the routine use of NMR as an aid for organic
chemistry , a day-to-day problem is the identifica- tion of signals deriving from common
contaminants literature , but the. J. Org. Chem. 62, 7512–7515 (1997).
59. Health & Safety & Environmental Deparment. Hexamethylcyclotrisiloxane - Material Safety
Data Sheet. (2009).
60. Harrod, J. F., Shaver, A. & Tucka, A. Cyclosiloxanes as Frameworks for Multimetallic
Compounds. 3. 1H NMR Spectra of Some Substituted Methylcyclosiloxanes. J. Chem. Inf. Model.
53, 1689–1699 (2013).
61. Sigma Aldrich. Datasheet for 2,4,6,8-Tetramethylcyclotetrasiloxane (CAS Number 2370-88-9).
(2016).
62. Sigma Aldrich. Datasheet for Hexamethylcyclotrisiloxane (CAS Number 541-05-9). (2015).
63. Sigma Aldrich. Datasheet forHexamethylcyclotrisiloxane (CAS Number 541-05-9 ). (2016).
64. Sigma Aldrich Datasheet for Hexamethyldisiloxane (CAS Number 107-46-0). (2016).
65. Sigma Aldrich. Datasheet for (Dichloromethyl)trimethylsilane (CAS no. CAS Number 5926-38-5).
66. Sigma Aldrich. Datasheet for 1-chloro-pentamethyl-disilane (CAS No. 1560-28-7). (2016).
67. Sigma Aldrich. Datasheet for Chloromethyl(dimethyl)silane (CAS Number 3144-74-9 ). (2016).
68. Sigma Aldrich. Datasheet for Dichloro(chloromethyl)methylsilane (CAS Number 1558-33-4 ).
(2015).
69. B.S. Middleditch. Artifacts: GC, MS, HPLC, TLC and PC. (Elsevier, 1989).
70. M. Alger. Polymer Science Dictionary. (Springer Science & Business Media, 1996).
71. Wypych, G. Handbook of Antiblocking, Release, and Slip Additives. (ChemTec Publishing, 2005).
72. Brandolini, A. & Hills, D. NMR spectra of polymers and polymer additives. 31 (Mobil Chemical
Company, 2000).
73. Mishra, B. Review of Extraction, Processing, Properties, and Applications of Reactive Metals:
1999 TMS Annual Meeting, San Diego, CA, February 28 - March 15, 1999. (John Wiley & Sons,
2013).
74. Yates, J.G., Lettier, P., Fluidized-Bed Reactors: Processes and Operating Conditions, 69,
(Springer, 2016).
75. Buxbaum, G. Industrial Inorganic Pigments. 55 (Wiley-VCH Verlag GmbH, 1998).
76. Krchma, I. J., Distillation of Titanium Tetrachloride. US2463396 A (1946).
77. Xiong, S., Yuan, Z., Yin, Z. & Yan, W. Removal Of The Vanadium Impurities From Crude TiCl4 with
High Content Of Vanadium Using A Mixture Of Al Power and White Mineral Oil.
Hydrometallurgy 119–120, 16–22 (2012).
78. Cossee, P. Ziegler-Natta Catalysis I. Mechanism of polymerization of α-olefins with Ziegler-
Natta catalysts. J. Catal. 3, 80–88 (1964).
79. Maier, G., Ulrich, S. & Boese, R. Isolierung und Kristallstrvkturanalyse Eines Titantetrachlorid-
Komplexes Eines 1,3-Dieetons. Tetrahedron Lett. 28, 4515–4516 (1987).
80. Shao, M. Reactions of TiCl4 or Ti(OiPr)Cl3 with 2,4-Pentanedione . The Molecular Structures of
TiCl3(acac)(THF) and Ti(OiPr)Cl2(acac)(THF). J. Chin. Chem. Soc 3, 901–906 (2000).
81. Sobota, P., Szafert, S. & Lis, T. Reactions of TiCl4 with diesters. Crystal structures of
[CH2(CO2Et)2Cl4Ti] and [C2H4(CO2CH2CH2OPh)2Cl4Ti]. J. Organomet. Chem. 443, 85–91 (1992).
82. Kakkonen, H. J. & Pursiainen, J. TiCl4 diester complexes : relationships between the crystal
structures and properties of Ziegler-Natta catalysts. J. Organomet. Chem 453, 175–184 (1993).
83. Pangborn, A. B., Giardello, M. A., Grubbs, R. H., Rosen, R. K. & Timmers, F. J. Safe and
Convenient Procedure for Solvent Purification. Organometallics 15, 1518–1520 (1996).
84. Japan, R. C. T. CrystalClear-SM Expert 3..1 b18. (2013).
85. Farrugia, L. J. WinGX and ORTEP for Windows: an update. J. Appl. Cryst 54, 849–854 (2012).
86. Allen, F. H., Johnson, O., Shields, G. P. & Smith, B. R. CIF applications. XV. enCIFer: a program
for viewing, editing and visualizing CIFs. J. Appl. Cryst 37, (2004).
87. Dolomanov, O.V., Bourhis, L.J., Gildea, R.J, Howard, J.A.K. & Puschmann, H. OLEX2: a complete
250
structure solution, refinement and analysis program. J. Appl. Cryst 42, 339–341 (2009).
88. Palatinus, L. & Chapuis, G. SUPERFLIP - a computer program for the solution of crystal
structures by charge flipping in arbitrary dimensions. J. Appl. Cryst 40, 786–790 (2007).
89. Sheldrick, G. M. A short history of SHELX. Acta Cryst.`` A64, 112–122 (2008).
90. Bondi, A. van der Waals Volumes and Radii. J. Phys. Chem 3, 441–451 (27643BC).
91. Pennington, D. A., Hughes, D. L., Bochmann, M. & Lancaster, S. J. The synthesis, structure and
catalytic activity of mono(salicylaldiminato) titanium complexes. Dalt. Trans. 2, 3480 (2003).
92. Sobota, P. Reaction between [MgCl4] 2- anion and titanium chlorides. The crystal structure of
[NBu4][trans-TiCl4(THF)2]⋅THF. Polyhedron 8, 2013–2016 (2013).
93. Gallucci, J. C., Kozmina, N. & Paquette, L. a. A [{TiCl2(isodiCp)}2(μ-O)] Dimer and a Twinned
Cyclic [TiCl(isodiCp)(μ-O)]4 Tetramer. Acta Crystallogr. Sect. C Cryst. Struct. Commun. 54, 1609–
1612 (1998).
94. Huang, Y. & Stephan, D. W. Synthesis and structure of titanium diolate complexes derived from
CpTiCI ,. Can. J. Chem 962, 956–962 (1995).
95. Schormann, M. [mu]-Oxo-bis[trichlorobis(dimethylsulfane)titanium(IV)]. Acta Cryst. E59,
M674–M675. (2003).
96. Wu, Y., Ho, Y., Lin, C. & Gau, H. Stepwise Reactions of TiCl4 and Ti(OiPr)Cl3 with 2-Propanol.
Variable-Temperature NMR Studies and Crystal Structures of [TiCl2(OiPr)(HOiPr)(µ-Cl)]2 and
[TiCl2(OiPr)(HOiPr)(µ-OiPr)]2. Inorg.Chem. 2, 5948–5952 (1996).
97. Sobota, P., Szafert, S. & Lis, T. Reactions of TiCl4 with diesters . Crystal structures of
[CH2(CO2Et)2Cl2Ti]. 443, 85–91 (1993).
98. Sobota, P. et al. Synthesis and characterization of new di- and tetra-meric vanadium
intermediates of olefin polymerization catalysts. Crystal structures of [V2(µ-Cl)2Cl4(MeCo2Et)4]
and [{VOCl2[CH2(CO2Et)2]}4]·2CH2Cl2. J. Chem. Soc. Dalt. Trans. 424, 1727–1732 (1995).
99. Crans, D. C., Felty, R. A., Anderson, O. P. & Miller, M. M. Structure and solution properties of a
dimeric tetrahedral vanadium(V) chloride alkoxide complex. Inorg. Chem. 32, 247–248 (1993).
100. Rosenthal, E. C. E. From vanadium(V) to vanadium(IV) - and backwards. Pure Appl. Chem. 81,
1197–1204 (2009).
101. Zhang, S. et al. Mixed-metal-organic frameworks (M′MOFs) from 1D to 3D based on the
‘organic’ connectivity and the inorganic connectivity: Syntheses, structures and magnetic
properties. CrystEngComm 17, 3312–3324 (2015).
102. Phan, A., Czaja, A. U., G??ndara, F., Knobler, C. B. & Yaghi, O. M. Metal-organic frameworks of
vanadium as catalysts for conversion of methane to acetic acid. Inorg. Chem. 50, 7388–7390
(2011).
103. Sun, Q. et al. Ideal Spintronics in Molecule-Based Novel Organometallic Nanowires. Sci. Rep. 5,
12772 (2015).
104. Lv, B., Shi, X., Ma, X., Zhang, Z. & Wang, K. Controllable fabrication of multifunctional 1D Ag-
based coordination polymer@PVP nanowires. New J. Chem. 39, 349–354 (2015).
105. Zhang, X.-M., Tong, M.-L., Lee, H. K. & Chen, X.-M. The First Noncluster Vanadium(IV)
Coordination Polymers: Solvothermal Syntheses, Crystal Structure, and Ion Exchange. J. Solid
State Chem. 160, 118–122 (2001).
106. Hosseini-Monfared, H., Asghari-Lalami, N., Pazio, A., Wozniak, K. & Janiak, C. Dinuclear
vanadium, copper, manganese and titanium complexes containing O,O,N-dichelating ligands:
Synthesis, crystal structure and catalytic activity. Inorganica Chim. Acta 406, 241–250 (2013).
107. Clegg, J. K., Kim, Y. & Lindoy, L. F. Discrete and polymeric supramolecular architectures derived
from dinuclear oxovanadium(IV) complexes of aryl-linked bis-diketonato ligands and nitrogen
donor co-ligands. J. Incl. Phenom. Macrocycl. Chem. 82, 247–257 (2015).
108. Roy, S. et al. Synthesis, characterization and magnetic property of a succinate bridged 1D
coordination polymer of cobalt(II) containing benzidine as end-capping ligand. J. Mol. Struct.
1061, 54–60 (2014).
109. Luthria, D. Oil Extraction and Analysis: Critical Issues and Competitive Studies. 104 (2004).
110. Banfi, D.; Patiny, L. www.nmrdb.org: Resurrecting and processing NMR spectra on-line. Chimia
251
(Aarau). 62(4), 280–281 (2008).
111. Castillo, A. M., Patiny, L. & Wist., J. Fast and Accurate Algorithm for the Simulation of NMR
spectra of Large Spin Systems. J. Magn. Reson. 209, 123–130 (2011).
112. Aires-de-Sousa, M. & Hemmer, J. G. Prediction of -1H NMR Chemical Shifts Using Neural
Networks. Anal. Chem. 74(1), 80–90. (2002).
113. Darrell Ebbing & Gammon, S. D. General Chemistry. (Cengage Learning, 2016).
114. Barrow, H., Brown, D. a., Alcock, N. W., Clase, H. J. & Wallbridge, M. G. H. Reactions of titanium
tetrachloride with carboxylic acids. Crystal and molecular structure of the dinuclear titanium
oxo compound [{TiCl2(O2CBut)(ButCO2H)}2O]. J. Chem. Soc. Dalt. Trans. 1, 195 (1994).
115. Dijkgraaf, C. Charge-Transfer Complexes of Titanium Tetrachloride, Titanium Tetrabromide, and
Vanadium Oxytrichloride with Aromatic Hydrocarbons. J. Phys. Chem. 69, 660–661 (1965).
116. Jones, Anthony C.; Hitchman, M. L. Chemical Vapour Deposition - Precursors, Processes and
Applications. 40 (Royal Society of Chemistry, 2009).
117. Ponraj, J. S., Attolini, G. & Bosi, M. Review on Atomic Layer Deposition and Applications of
Oxide Thin Films. Crit. Rev. Solid State Mater. Sci. 38, 203–233 (2013).
118. Moravej, M. & Hicks, R. F. Atmospheric plasma deposition of coatings using a capacitive
discharge source. Chem. Vap. Depos. 11, 469–476 (2005).
119. Gordon, R. G., Barry, S., Barton, J. T. & Broomhall-Dillard, R. N. R. Atmospheric pressure
chemical vapor deposition of electrochromic tungsten oxide films. Thin Solid Films 392, 231–
235 (2001).
120. Ha, H. Y., Nam, S. W., Lim, T. H., Oh, I. H. & Hong, S. A. Properties of the TiO2 membranes
prepared by CVD of titanium tetraisopropoxide. J. Memb. Sci. 111, 81–92 (1996).
121. Park, H. S., Waezsada, S. D., Cowley, A. H. & Roesky, H. W. Growth of GaN layer from the single-
source precursor (Et2GaNH2)(3). Chem. Mater. 10, 2251–2257 (1998).
122. Jones, A. C. & Hitchman, M. L. Overview of Chemical Vapour Deposition. Chem. Vap. Depos.
Precursors, Process. Appl. 1–36 (2009).
123. Barron, A. in Chemistry for Electronic Materials 204–206 (Connexions, 2011).
124. Chua, C. S. et al. Effect of annealing temperature on microstructure and UV light photocatalytic
activity of TiO2 films grown by atmospheric pressure CVD. Chem. Vap. Depos. 20, 44–50 (2014).
125. Patil, D. R., Patil, L. S., Bange, J. P. & Gautam, D. K. Deposition of oriented nanocrystalline TiO 2
thin films. J. Optoelectron. Adv. Mater. 10, 3251–3256 (2008).
126. Hocine, D. et al. Characterization of TiO2 antireflection coatings elaborated by APCVD for
monocrystalline silicon solar cells. Phys. Status Solidi 12, 323–326 (2015).
127. Vernardou, D., Louloudakis, D., Spanakis, E., Katsarakis, N. & Koudoumas, E. Thermochromic
amorphous VO2 coatings grown by APCVD using a single-precursor. Sol. Energy Mater. Sol.
Cells 128, 36–40 (2014).
128. Kiria, P., Hyett, G. & Binionsa, R. Solid state thermochromic materials. Adv. Mater. Lett. 1, 86–
105 (2010).
129. Manning, T., Parkin, I.P., Blackman, C. & Qureshi, U. APCVD of thermochromic vanadium
dioxide thin films - solid solutions V2-xMxO2 (M = Mo, Nb) or composites VO2 : SnO2. J. Mater.
Chem. 15, 4560–4566 (2005).
130. Hatem, D. & Belkaid, M. S. SnO2 Thin Films Prepared by APCVD for Organic Solar Cells
Application. Adv. Mater. Res. 685, 166–173 (2013).
131. Maleki, M. & Rozati, S. M. Investigation of the Effect of the Substrate Position Relative to the
Source on the Optoelectrical and Structural Properties of Pure Nanostructured Tin Oxide by
APCVD. Chem. Vap. Depos. 20, 352–355 (2014).
132. Morales, C. et al. Low temperature SnO2 films deposited by APCVD. Microelectronics J. 39, 586–
588 (2008).
133. Chai, C. C., Peng, J. & Yan, B. P. Preparation and Gas-Sensing Properties of a-Fe203 Thin Films.
24, (1995).
134. Ashraf, S., Binions, R., Blackman, C. S. & Parkin, I. P. The APCVD of tungsten oxide thin films
from reaction of WCl6 with ethanol and results on their gas-sensing properties. Polyhedron 26,
252
1493–1498 (2007).
135. Riaz, U. Low-temperature atmospheric-pressure chemical vapor deposition of tungsten oxide
thin films. Thin Solid Films 235, 15–16 (1993).
136. Pacio, M. et al. Study of (100) orientated ZnO films by APCVD system. Mater. Sci. Eng. B 174,
38–41 (2010).
137. Yoo, Y. Z. et al. Development of tilted hexagonal platelet ZnO using atmospheric pressure
chemical vapor deposition and investigation of its growth mechanism. Appl. Phys. Lett. 100, 1–
4 (2012).
138. Maleki, M. & Rozati, S. M. Dependence of ZnO Nanostructured Thin Films Properties on Growth
Temperature by APCVD Method. Acta Phys. Pol. A 128, 367–373 (2015).
139. Ruppi, S. Deposition, Microstructure and properties of texture-controlled CVD α-Al2O3 coatings.
Int. J. Refract. Met. Hard Mater. 23, 306–316 (2005).
140. Blittersdorf, S., Bahlawane, N., Kohse-Hoinghaus, K., Atakan, B. & Muller, J. CVD of Al2O3 Thin
Films Using Aluminum Tri-isopropoxide. Chem. Vap. Depos. 9, 194–198 (2003).
141. Morosanu, C. E. & Siddall, G. Thin Films by Chemical Vapour Deposition. 138–140 (2013).
142. Bodurov, G., Ivanova, T., Aleksandrova, M. & Gesheva, K. A. Optical characterization of WO3-
VOx thin films for application in electrochromic devices—‘smart windows’. J. Phys. Conf. Ser.
356, 12016 (2012).
143. Powell, M. J. et al. Intelligent Multifunctional VO2/SiO2/TiO2 Coatings for Self-Cleaning, Energy-
Saving Window Panels. Chem. Mater. 28 (5), 1369–1376 (2016).
144. Tiwari, A., Mishra, A. K., Kobayashi, H. & Turner, A. P. F. Intelligent Nanomaterials. (John Wiley
& Sons, 2012).
145. Sotelo-Vazquez, C., Quesada-Cabrera, R., Darr, J. A. & Parkin, I. P. Single-step synthesis of
doped TiO2 stratified thin-films by atmospheric-pressure chemical vapour deposition. J. Mater.
Chem. A 2, 7082 (2014).
146. Manning, T. D., Parkin, I. P., Pemble, M. E., Sheel, D. & Vernardou, D. Intelligent Window
Coatings: Atmospheric Pressure Chemical Vapor Deposition of Tungsten-Doped Vanadium
Dioxide. Chem. Mater. 16, 744–749 (2004).
147. Ullmann, F., Yamamoto, Y. S., Campbell, F. T., Pfefferkorn, R. & Rounsaville, J. F. Ullmann’s
Encyclopedia of Industrial Chemistry. (VCH, 1991).
148. Shriver, D. A. & Drezden, M. A. The manipulation of air-sensitive compounds. 2nd Ed., 95
(Wiley-Interscience, 1986).
149. Processes, L. D. Applications of Metal Amidinate Precursors in CVD / ALD. 1–9 (Azo Nano, 2013).
150. Moinpour, M., Lubic, K., Nguyen, B. & Moghadam, F. Factors Affecting The Temperature
Stability Of APCVD Systems In A Rapidly Changing Semiconductor Manufacturing Environment.
(1993).
151. O’Neill, S. A., Clark, R. J. H., Parkin, I. P., Elliott, N. & Mills, A. Anatase Thin Films on Glass from
the Chemical Vapor Deposition of Titanium(IV) Chloride and Ethyl Acetate. Chem. Mater. 15,
46–50 (2003).
152. Cross, A. J., Dunnill, C. W. & Parkin, I. P. Production of Predominantly Anatase Thin Films on
Various Grades of Steel and Other Metallic Substrates From TiCl4 and Ethyl Acetate by
Atmospheric Pressure CVD. Chem. Vap. Depos. 18, 133–139 (2012).
153. Hyett, G., Darr, J. A., Mills, A. & Parkin, I. P. An Investigation into the Optimum Thickness of
Titanium Dioxide Thin Films Synthesized by Using Atmospheric Pressure Chemical Vapour
Deposition for Use in Photocatalytic Water Oxidation. Chem. A Eur. J. 16, 10546–10552 (2010).
154. Wilkinson, M. et al. Combinatorial atmospheric pressure chemical vapor deposition of graded
TiO2-VO2 mixed-phase composites and their dual functional property as self-cleaning and
photochromic window coatings. ACS Comb. Sci. 15, 309–319 (2013).
155. Kotz, J., Treichel, P. & Townsend, J. Chemistry and Chemical Reactivity 575 (Brooks Cole, 2008).
156. Smith, J. D. The Chemistry of Metal CVD. Journal of Organometallic Chemistry 522, 161–162
(1996).
157. Drosos, C. & Vernardou, D. Perspectives of energy materials grown by APCVD. Sol. Energy
253
Mater. Sol. Cells 140, 1–8 (2015).
158. Meyers, L. & P, W. Atmospheric Pressure Chemical Vapor Deposition of CdTe for High Efficiency
Thin Film PV Devices Atmospheric Pressure Chemical Vapor Deposition of CdTe for High
Efficiency Thin Film PV Devices Annual Report. (1999).
159. Downie, N. A. Industrial Gases. 394 (Springer, 2007).
160. Hornyak, G. L., Tibbals, H. F., Dutta, J. & Moore, J. J. Introduction to Nanoscience and
Nanotechnology: A Workbook. 1112 (Notre Dame, 2008).
161. Choy, K. Chemical vapour deposition of coatings. Prog. Mater. Sci. 48, 57–170 (2003).
162. Jones, A. C. & Hitchman, M. L. Chemical Vapour Deposition: Precursors, Processes and
Applications. (Royal Society of Chemistry, 2009).
163. Turgambaeva, A. E., Prud’homme, N., Krisyuk, V. V. & Vahlas, C. Mass spectrometric monitoring
of the gas phase during the CVD of copper from copper cyclopentadienyl triethylphosphine.
Chemical Vapor Deposition 18, 209–214 (2012).
164. Schropp, R. E. I. Hot Wire Chemical Vapor Deposition: Recent Progress, Present State of the Art
and. ECS Trans. 25, 3–14 (2009).
165. Li, X. M., Eustergerling, B. D. & Shi, Y. J. Mass spectrometric study of gas-phase chemistry in a
hot-wire chemical vapor deposition reactor with tetramethylsilane. Int. J. Mass Spectrom. 263,
233–242 (2007).
166. Rego, C. A. et al. In-situ mass spectrometric study of the gas-phase species involved in CVD of
diamond as a function of filament temperature. Diamond and Related Materials 4, 770–774
(1995).
167. Tsang, R. S., Rego, C. A., May, P. W., Ashfold, M. N. R. & Rosser, K. N. Examination of the effects
of nitrogen on the CVD diamond growth mechanism using in situ molecular beam mass
spectrometry. Diamond and Related Materials 6, 247–254 (1997).
168. Petherbridge, J. R., May, P. W., Pearce, S. R. J., Rosser, K. N. & Ashfold, M. N. R. Low
temperature diamond growth using CO2/CH4 plasmas: Molecular beam mass spectrometry and
computer simulation investigations. J. Appl. Phys. 89, 1484–1492 (2001).
169. Tsang, R. S., May, P. W., Ashfold, M. N. R. & Rosser, K. N. Influence of phosphine on the
diamond growth mechanism: a molecular beam mass spectrometric investigation. Diam. Relat.
Mater. 7, 1651–1656 (1998).
170. Rahtu, A., Alaranta, T. & Ritala, M. In situ quartz crystal microbalance and quadrupole mass
spectrometry studies of atomic layer deposition of aluminum oxide from trimethylaluminum
and water. Langmuir 17, 6506–6509 (2001).
171. Matero, R., Rahtu, A. & Ritala, M. In Situ Quadrupole Mass Spectrometry and Quartz Crystal
Microbalance Studies on the Atomic Layer Deposition of Titanium Dioxide from Titanium
Tetrachloride and Water. Chem. Mater 13, 4506–4511 (2001).
172. Rahtu, A. & Ritala, M. Integration of a Quadrupole Mass Spectrometer and a Quartz Crytal
Microblance For In-Situ Characterization of Atomic Layr Deposition Processes in Flow Rate
Typre Reactors. in CVD XV, proceedings of the fifteenth International Symposium on Chemical
Vapor Deposition 105–110 (2000).
173. Johnstone, R. A. W. Mass Spectrometry for Organic Chemists. 21 (Science, 1972).
174. Colombini, M. P. & Modugno, F. Organic Mass Spectrometry in Art and Archaeology, 56, (John
175. Ashcroft, A. E. Ionization Methods in Organic Mass Spectrometry. 13 (Royal Society of
Chemistry, 1997)
176. de Hoffmann Edmond, S. V. Mass Spectrometry: principles and applications, 3rd Edition . 1–4
(John Wiley and Sons, 2007).
177. Watson, J. T. & O. David Sparkman. Introduction to Mass Spectrometry - Instrumentation,
Applications, and Strategies for Data Interpretation. University of Arizona, 444, (John Wiley &
Sons, 2013).
178. Williams, D. H. Mass Spectrometry, 89 , (Royal Society of Chemistry, 1973).
179. Walker Johnstone, R. A, Rose, M. E, Mass Spectrometry For Chemists and Biochemists, 36,
254
(Cambridge University Press, 1996).
180. Agnieszka Kraj, Dominic M. Desiderio, N. M. N. Mass Spectrometry: Instrumentation,
Interpretation, and Applications. 46 (Wiley, 2008).
181. Herbert, C. G. & Johnstone, R. A. W. Mass spectrometry Basics, 189 (CRC Press, 2002).
182. Ramanathan, R., Mass Spectrometry in Drug Metabolism and Pharmacokinetics, 62 (John Wiley
& Sons, 2008).
183. Harrison, A. G. Chemical Ionization Mass Spectrometry, Second Edition. 59 (CRC Press, 1992).
184. Hiden Analytical. ltd Mass Spectrometers for Residual Gas Analysis - RGA For Applications in a
Vacuum Environment Quadrupole Mass Spectrometers (2016).
185. Kool, J. & Niessen, W. M. A. Analyzing Biomolecular Interactions by Mass Spectrometry. (John
186. Stein, S. . Titanium Tetrachloride. NIST Webook (2016).
187. Howard, C. J., Sabine, T. M. & Dickson, F. Structural and thermal parameters for rutile and
anatase. Acta Crystallographica Section B 47, 462–468 (1991).
188. Gurvitch, M. et al. VO2 films with strong semiconductor to metal phase transition prepared by
the precursor oxidation process. J. Appl. Phys. 102, 1–13 (2007).
189. Gao, Y., Kang, L., Chen, Z. & Luo, H. Solution Processing of Nanoceramic VO2 Thin Films for
Application to Smart Windows. Nanofabrication (2011).
190. Qureshi, U., Manning, T. & Parkin, I. Atmospheric pressure chemical vapour deposition of VO2
and VO2/TiO2 films from the reaction of VOCl3, TiCl4 and water. J. Mater. 14, 1190–1194 (2004).
191. Kam, K. C. & Cheetham, A. K. Thermochromic VO2 nanorods and other vanadium oxides
nanostructures. Mater. Res. Bull. 41, 1015–1021 (2006).
192. Kam, K. C. Investigations of Inorganic and Hybrid Inorganic-organic Nanostructures. (University
of California, Santa Barbara, 2007).
193. Cheng, C. et al. Self-Assembly and Horizontal Orientation Growth of VO2 Nanowires. Sci. Rep. 4,
5456 (2014).
194. Shirahata, N., Masuda, Y., Yonezawa, T. & Koumoto, K. Control over film thickness of SnO2
ultrathin film selectively deposited on a patterned self-assembled monolayer. Langmuir 18,
10379–10385 (2002).
195. Wyckoff, R. W. G. Crystal Structures Vol. 1 (2nd Ed.). (New York : Interscience Publishers, 1964).
196. Chew, C. K. T., Salcianu, C., Bishop, P., Carmalt, C. J. & Parkin, I. P. Functional thin film coatings
incorporating gold nanoparticles in a transparent conducting fluorine doped tin oxide matrix. J.
Mater. Chem. C 3, 1118–1125 (2015).
197. Wang, J. T. et al. Influence of preferred orientation on the electrical conductivity of fluorine-
doped tin oxide films. Sci. Rep. 4, 3679 (2014).
198. Evans, P., Pemble, M. E. & Sheel, D. W. Precursor-directed control of crystalline type in
atmospheric pressure CVD growth of TiO2 on stainless steel. Chem. Mater. 18, 5750–5755
(2006).
199. Lee, M. S. Mass Spectrometry Handbook. 302, (John Wiley & Sons, 2012).
200. Montaudo, G. & Montaudo, M. S. Mass Spectrometry of Polymers: Polypropylene Glycol. Z.
Naturforsch 35a, 1090–1097 (1980).
201. Niederberger, M. & Pinna, N. Metal Oxide Nanoparticles in Organic Solvents: Synthesis,
Formation. (Springer Science & Business Media, 2009).
202. Knapp, C. E., Kafizas, A., Parkin, I. P. & Carmalt, C. J. The use of combinatorial aerosol-assisted
chemical vapour deposition for the formation of gallium-indium-oxide thin films. J. Mater.
Chem. 21, 12644 (2011).
203. Marchand, P., Hassan, I. A, Parkin, I. P. & Carmalt, C. J. Aerosol-assisted delivery of precursors
for chemical vapour deposition: expanding the scope of CVD for materials fabrication. Dalton
Trans. 42, 9406–22 (2013).
204. Special Issue on Aerosol-Assisted Chemical Vapor Deposition - Choy - 2006 - Chemical Vapor
Deposition - Wiley Online Library.
205. Lim, S. P., Huang, N. M., Lim, H. N. & Mazhar, M. Aerosol assisted chemical vapour deposited
255
(AACVD) of TiO2 thin film as compact layer for dye-sensitised solar cell. Ceram. Int. 40, 8045–
8052 (2014).
206. Edusi, C., Hyett, G., Sankar, G. & Parkin, I. P. Aerosol-Assisted CVD of Titanium Dioxide Thin
Films from Methanolic Solutions of Titanium Tetraisopropoxide; Substrate and Aerosol-
Selective Deposition of Rutile or Anatase. Chem. Vap. Depos. 17, 30–36 (2011).
207. Saeli, M., Piccirillo, C., Warwick, M. E. A. & Binions, R. in Materials and processes for energy:
communicating current research and technological developments 736–746 (Formatex, 2013).
208. Naik, A. J. T., Bowman, C., Panjwani, N., Warwick, M. E. a. & Binions, R. Electric field assisted
aerosol assisted chemical vapour deposition of nanostructured metal oxide thin films. Thin
Solid Films 544, 452–456 (2013).
209. Piccirillo, C., Binions, R. & Parkin, I. Synthesis and functional properties of vanadium oxides:
V2O3, VO2, and V2O5 deposited on glass by aerosol-assisted CVD. Chemical vapor deposition 13,
145–151 (2007).
210. Stoycheva, T. T. et al. Aerosol-assisted CVD of SnO2 thin films for gas-sensor applications. Chem.
Vap. Depos. 17, 247–252 (2011).
211. Naeem, R. et al. Electric-Field Aerosol-Assisted CVD: Synthesis, Characterization, and Properties
of Tin Oxide Microballs Prepared from a Single Source Precursor. Chem. Vap. Depos. 360–368
(2015).
212. Pan, J. Chemical Vapor Deposition of One Dimensional Tin Oxide Nanostructures: Structural
Studies, Surface Modifications and Device Applications. Chem. Vap. Depos. 231 (2010).
213. Chen, S., McLachlan, M., Sapelkin, A. & Binions, R. Aerosol assisted chemical vapour deposition
of transparent conductive ZnO thin films with hexagonal microplate surfaces and ultrahigh
haze values. J. Mater. Chem. A 3, 22311–22315 (2015).
214. Palgrave, R. & Parkin, I. P. Aerosol assisted chemical vapour deposition of photochromic
tungsten oxide and doped tungsten oxide thin films. J. Mater. Chem., 14, 2864–2867 (2004).
215. Qadri, M. U. et al. WO3 nano-needles by Aerosol Assisted CVD for optical sensing. Procedia Eng.
25, 761–764 (2011).
216. Chew, C., Bishop, P., Salcianu, C., Carmalt, C. J. & Parkin, I. P. Aerosol-assisted deposition of
gold nanoparticle-tin dioxide composite films. RSC Adv. 4, 13182 (2014).
217. Chadwick, N. P. et al. The use of time resolved aerosol assisted chemical vapour deposition in
mapping metal oxide thin film growth and fine tuning functional properties. J. Mater. Chem. A
3, 4811–4819 (2015).
218. Chadwick, N. P. et al. Photo-activity and low resistivity in N/Nb Co-doped TiO 2 thin films by
combinatorial AACVD. J. Mater. Chem. A 407–415 (2016).
219. Noor, N. & Parkin, I. P. Halide doping effects on transparent conducting oxides formed by
aerosol assisted chemical vapour deposition. Thin Solid Films 532, 26–30 (2013).
220. Piccirillo, C., Binions, R. & Parkin, I. P. Synthesis and characterisation of W-doped VO2 by
Aerosol Assisted Chemical Vapour Deposition. Thin Solid Films 516, 1992–1997 (2008).
221. Bawaked, S. M. et al. Aerosol assisted chemical vapor deposition of conductive and
photocatalytically active tantalum doped titanium dioxide films. J. Mater. Chem. A 2, 12849–
12856 (2014).
222. Bachu, D., Sankar, G. & Parkin, I. P. Aerosol Assisted Chemical Vapour Deposition of
Transparent Conductive Zinc Oxide Films. Chem. Mater. 24 (24), 4704-4710, (2012).
223. Ehsan, M. A. et al. Cobalt titanate-cobalt oxide composite thin films deposited from
heterobimetallic precursor. Appl. Organomet. Chem. 26, 493–498 (2012).
224. Binions, R. & Parkin, Ivan, P. in Advances in Nanocomposites - Synthesis, Characterization and
Industrial Applications (ed. Boreddy, R.) (2007).
225. Hou, X. & Choy, K.-L. Processing and Applications of Aerosol-Assisted Chemical Vapor
Deposition. Chem. Vap. Depos. 12, 583–596 (2006).
226. Jaaniso, R. & Tan, O. K. Semiconductor Gas Sensors. 439 (Elsevier, 2013).
227. Knapp, C. E. & Carmalt, C. J. Solution based CVD of main group materials. Chem. Soc. Rev. 45,
1036–1064 (2016).
256
228. Vallejos, S., Di Maggio, F., Shujah, T. & Blackman, C. Chemical Vapour Deposition of Gas
Sensitive Metal Oxides. Chemosensors 4, 4 (2016).
229. Vallejos, S., Umek, P. & Blackman, C. Aerosol Assisted Chemical Vapour Deposition Control
Parameters for Selective Deposition of Tungsten Oxide Nanostructures. J. Nanosci.
Nanotechnol. 11, 8214–8220 (2011).
230. Kaye, K., Turner, D., McKenna, D., Cackett, A. & Hyett, G. The Use of Additives to Control the
Morphology of Thin Films Synthesized Using Aerosol Assisted Chemical Vapour Deposition.
Phys. Procedia 46, 21–26 (2013).
231. Tahir, A. A. et al. Photoelectrochemical and photoresponsive properties of Bi2S 3 nanotube and
nanoparticle thin films. Chem. Mater. 22, 5084–5092 (2010).
232. Sakka, S. Handbook of sol-gel science and technology. (Springer Science & Business Media,
2005).
233. Allendorf, M. D., Bernard, C., Division, E. S. H. T. M., Science, E. S. D. & Division, T.
Requirements of Precursors for conventional and Aerosol Assisited CVD. in Chemical Vapor
Deposition: Proceedings of the Fourteenth International Conference and EUROCVD-11 324–330
(The Electrochemical Society, 1997).
234. Lewis, D. J. & O’Brien, P. Ambient pressure aerosol-assisted chemical vapour deposition of
(CH3NH3)PbBr3, an inorganic-organic perovskite important in photovoltaics. Chem. Commun.
50, 6319–21 (2014).
235. Fujishima, A. & Honda, K. Electrochemical Photolysis of Water at a Semiconductor Electrode.
Nature 238, 37–38 (1972).
236. Schneider, J. et al. Understanding TiO2 Photocatalysis: Mechanisms and Materials. Chem. Rev.
114, 9919–9986 (2014).
237. Linsebigler, A. L., Lu, G. & Yates, J. T. Photocatalysis on TiO2 Surfaces: Principles, Mechanisms,
and Selected Results. Chem. Rev. 95, 735–758 (1995).
238. Quesada-Cabrera, R., Sotelo-Vazquez, C., Darr, J. a. & Parkin, I. P. Critical influence of surface
nitrogen species on the activity of N-doped TiO2 thin-films during photodegradation of stearic
acid under UV light irradiation. Appl. Catal. B Environ. 160–161, 582–588 (2014).
239. Park, J. S. & Choi, W. Enhanced remote photocatalytic oxidation on surface-fluorinated TiO2.
Langmuir 20, 11523–7 (2004).
240. Rampaul, A. et al. Titania and tungsten doped titania thin films on glass; active photocatalysts.
Polyhedron 22, 35–44 (2003).
241. Lee, D., Omolade, D., Cohen, R. E. & Rubner, M. F. pH-Dependent Structure and Properties of
TiO2 /SiO2 Nanoparticle Multilayer Thin Films. Chem. Mater. 19, 1427–1433 (2007).
242. Li, X. & He, J. Synthesis of Raspberry-Like SiO2–TiO2 Nanoparticles toward Antireflective and
Self-Cleaning Coatings. ACS Appl. Mater. Interfaces 5, 5282–5290 (2013).
243. Kaewtip, C. et al. Photocatalytic of TiO2-SiO2 Thin Films Co-Doped with Fe3+ and Thio-Urea in
the Degradation of Formaldehyde by Indoor and Outdoor Visible Lights. Adv. Mater. Phys.
Chem. 2, 40–44 (2012).
244. Machida, M., Norimoto, K. & Watanabe, T. The effect of SiO2 addition in super-hydrophilic
property of TiO 2 photocatalyst. J. Mater. Sci. 34, 2569–2574 (1999).
245. Huang, T., Huang, W., Zhou, C., Situ, Y. & Huang, H. Superhydrophilicity of TiO2/SiO2 thin films:
Synergistic effect of SiO2 and phase-separation-induced porous structure. Surf. Coatings
Technol. 213, 126–132 (2012).
246. Houmard, M. et al. Morphology and natural wettability properties of sol–gel derived TiO2–SiO2
composite thin films. Appl. Surf. Sci. 254, 1405–1414 (2007).
247. Permpoon, S. et al. Natural and persistent superhydrophilicity of SiO2/TiO2 and TiO2/SiO2 bi-
layer films. Thin Solid Films 516, 957–966 (2008).
248. Zhou, L., Yan, S., Tian, B., Zhang, J. & Anpo, M. Preparation of TiO2–SiO2 film with high
photocatalytic activity on PET substrate. Mater. Lett. 60, 396–399 (2006).
249. Yu, J., Jimmy, C. & Zhao, X. The effect of SiO2 addition on the grain size and photocatalytic
activity of TiO2 thin films. J. Sol-Gel Sci. Technol. 24, 95–103 (2002).
257
250. Mills, A., Elliott, N., Parkin, I. P., O’Neill, S. a. & Clark, R. J. Novel TiO2 CVD films for
semiconductor photocatalysis. J. Photochem. Photobiol. A Chem. 151, 171–179 (2002).
251. Panjawi, N., Naik, A., Warwick, M. E. A., Hyett, G. & Binions, R. The Preparation of Titanium
Dioxide Gas Sensors by the Electric Field Assisted Aerosol CVD Reaction of Titanium
Isopropoxide in Toluene. Chem. Vap. Depos. 18, 102–106 (2012).
252. Yamauchi, S. Plasma-Assisted Chemical Vapor Deposition of TiO2 Thin Films for Highly
Hydrophilic Performance. Cryst. Struct. Theory Appl. 2, 1–7 (2013).
253. Klobukowski, E. R., Tenhaeff, W. E., McCamy, J. W., Harris, C. S. & Narula, C. K. Atmospheric
pressure chemical vapor deposition of high silica SiO2–TiO2 antireflective thin films for glass
based solar panels. J. Mater. Chem. C 1, 6188 (2013).
254. Yoon, J., Oh, H. K. & Kwag, Y. J. Structural and Optical Properties of TiO2 -SiO2 Composite Films
Prepared by Aerosol-Assisted Chemical-Vapor Deposition. J. Korean Phys. Soc. 33, 699–704
(1998).
255. Noor, N. & Parkin, I. P. Enhanced transparent-conducting fluorine-doped tin oxide films formed
by Aerosol-Assisted Chemical Vapour Deposition. J. Mater. Chem. C 1, 984–996 (2013).
256. Mills, A. & Wang, J. Simultaneous monitoring of the destruction of stearic acid and generation
of carbon dioxide by self-cleaning semiconductor photocatalytic films. J. Photochem. Photobiol.
A Chem. 182, 181–186 (2006).
257. Cozzi, P. G., Solari, E., Chiesi-Villa, A. & Rizzoli, C. Interaction Modes of Titanium Tetrachloride
with the Carbonyl Functionality, Euro. J. Inorg. Chem. 129 (11) 1361–1368 (1996).
258. Johnson, B. F. . Inorganic Chemistry of the Transition Elements, Volume 6. (Royal Society of
Chemistry, 1978).
259. Shtelman, A. V. & Becker, J. Y. A convenient method for the synthesis of α-silylacetic acids.
Tetrahedron Lett. 49, 3101–3103 (2008).
260. Gaba, R., Bhandari, M. & Kakkar, R. Adsorption studies of acetaldehyde on TiO2 nanosurface. 4,
769–778 (2013).
261. Liu, L., Liu, Q., Xiao, W., Pan, C. & Wang, Z. O2 adsorption and dissociation on an anatase (101)
surface with a subsurface Ti interstitial. Phys. Chem. Chem. Phys. 18, 1–8 (2016).
262. Hooper, A. B. & Lipscomb, J. D. Resolution of multiple heme centers of hydroxylamine
oxidoreductase from Nitrosomonas. 2. Mössbauer spectroscopy. J. Am. Chem. Soc. 42, 3973–
3976 (2002).
263. Clayden, J., Greeves, N. & Warren, S. Organic Chemistry. (OUP Oxford, 2012).
264. Harding, C., Janes, R. & Johnson, D. Elements of the P Block. (Royal Society of Chemistry, 2002).
265. Diesen, V. et al. Visible Light Photocatalytic Activity in AACVD-Prepared N-modified TiO2 Thin
Films. Chem. Vap. Depos. 20, 91–97 (2014).
266. Sun, M. et al. Thermal formation of silicon-doped TiO2 thin films with enhanced visible light
photoelectrochemical response. Electrochem. commun. 16, 26–29 (2012).
267. Jin, R., Wu, Z., Liu, Y., Jiang, B. & Wang, H. Photocatalytic reduction of NO with NH3 using Si-
doped TiO2 prepared by hydrothermal method. J. Hazard. Mater. 161, 42–8 (2009).
268. Di Paola, A., Bellardita, M. & Palmisano, L. Brookite, the Least Known TiO2 Photocatalyst.
Catalysts 3, 36–73 (2013).
269. Tauc, J. Optical properties and electronic structure of amorphous Ge and Si. Mater. Res. Bull. 3,
37–46 (1968).
270. Noctun, M., Slawomir, K. & Zontek, J. Optical properties of SiO2/TiO2 thin layers prepared by
sol-gel method. Opt. Appl. XLI, 979–987 (2011).
271. Sokrates, T. P. The Physics of SiO2 and Its Interfaces Proceedings of the International
Proceedings of the International Topical Conference on the Physics of SiO2 and its Interfaces.
(Elsevier, 2013).
272. Long, R., Dai, Y., Meng, G. & Huang, B. Energetic and electronic properties of X- (Si, Ge, Sn, Pb)
doped TiO2 from first-principles. Phys. Chem. Chem. Phys. 11, 8165–72 (2009).
273. Niu, F., Zhai, J., Jiang, L. & Song, W.G. Light induced activity switch in interfacial hydrogen-bond
catalysis with photo sensitive metal oxides. Chem. Commun. (Camb). 4738–40 (2009).
258
274. Kim, H., Seo, S. B., Kim, D. Y., Bae, K. & Sohn, S. Y. Enhanced Hydrophilic Property of TiO 2 Thin
Film Deposited on Glass Etched with O2 Plasma. Trans. Electr. Electron. Mater. 14, 152–155
(2013).
275. Warwick, M. E. A. & Binions, R. Advances in thermochromic vanadium dioxide films. J. Mater.
Chem. A 2, 3275–3292 (2014).
276. Granqvist, C. G. Oxide-based chromogenic coatings and devices for energy efficient
fenestration: Brief survey and update on thermochromics and electrochromics. J. Vac. Sci.
Technol. B, Nanotechnol. Microelectron. Mater. Process. Meas. Phenom. 32, 60801 (2014).
277. Kamalisarvestani, M., Saidur, R., Mekhilef, S. & Javadi, F. S. Performance, materials and coating
technologies of thermochromic thin films on smart windows. Renew. Sustain. Energy Rev. 26,
353–364 (2013).
278. Granqvist, C. G. Transparent conductors as solar energy materials: A panoramic review. Sol.
Energy Mater. Sol. Cells 91, 1529–1598 (2007).
279. Babulanam, S. M., Eriksson, T. S., Niklasson, G. A. & Granqvist, C. G. Thermochromic VO2 Films
For Energy-Efficient Windows. in (eds. Holly, S. & Lampert, C. M.) 16, 8–18 (1987).
280. Goodenough, J. B. Anomalous Properties of the Vanadium Oxides. Ann. Rev. Mater. Sci. 1, 101–
138 (1971).
281. Saeli, M., Piccirillo, C., Parkin, I. P., Binions, R. & Ridley, I. Energy modelling studies of
thermochromic glazing. Energy Build. 42, 1666–1673 (2010).
282. Batista, C., Ribeiro, R. M. & Teixeira, V. Synthesis and characterization of VO2-based
thermochromic thin films for energy-efficient windows. Nanoscale Res. Lett. 6, 301 (2011).
283. Blackman, C. S., Piccirillo, C., Binions, R. & Parkin, I. P. Atmospheric pressure chemical vapour
deposition of thermochromic tungsten doped vanadium dioxide thin films for use in
architectural glazing. Thin Solid Films 517, 4565–4570 (2009).
284. Louloudakis, D. et al. Atmospheric pressure chemical vapor deposition of amorphous tungsten
doped vanadium dioxide for smart window applications. 7, 10–14 (2016).
285. Jorgenson, G. V. & Lee, J. C. Doped vanadium oxide for optical switching films. Sol. Energy
Mater. 14, 205–214 (1986).
286. Burkhardt, W. et al. Tungsten and fluorine co-doping of VO2 films. Thin Solid Films 402, 226–
231 (2002).
287. Hu, S. et al. Optical properties of Mg-doped VO2: Absorption measurements and hybrid
functional calculations. Appl. Phys. Lett. 101, 10–14 (2012).
288. Hanlon, T. J., Coath, J. A. & Richardson, M. A. Molybdenum-doped vanadium dioxide coatings
on glass produced by the aqueous sol-gel method. Thin Solid Films 436, 269–272 (2003).
289. Tan, X. et al. Unraveling metal-insulator transition mechanism of VO₂ triggered by tungsten
doping. Sci. Rep. 2, 466 (2012).
290. Long, R. et al. Identifying structural distortion in doped VO2 with IR spectroscopy. Phys. Chem.
Chem. Phys. 14, 7225–8 (2012).
291. Manning, T. D. & Parkin, I. P. Vanadium(IV) oxide thin films on glass and silicon from the
atmospheric pressure chemical vapour deposition reaction of VOCl3 and water. Polyhedron 23,
3087–3095 (2004).
292. Powell, M. J. et al. Direct and continuous synthesis of VO2 nanoparticles. Nanoscale 7, 18686–
18693 (2015).
293. Son, J. H., Wei, J., Cobden, D., Cao, G. & Xia, Y. Hydrothermal synthesis of monoclinic VO2
micro- and nanocrystals in one step and their use in fabricating inverse opals. Chem. Mater. 22,
3043–3050 (2010).
294. Kim, D. H. & Kwok, H. S. Pulsed laser deposition of VO2 thin films. Appl. Phys. Lett. 65, 3188–
3190 (1994).
295. Qureshi, U., Manning, T. D., Parkin, I. P. & Oaj, W. C. H. Atmospheric pressure chemical vapour
deposition of VO2 and VO2/TiO2 films from the reaction of VOCl3 , TiCl4 and water. 1190–1194
(2004).
296. Fang, B. et al. Optical properties of vanadium dioxide thin film in nanoparticle structure. Opt.
259
Mater. (Amst). 47, 225–230 (2015).
297. Sahana, M. B., Dharmaprakash, M. S. & Shivashankar, S. a. Microstructure and properties of
VO2 thin films deposited by MOCVD from vanadyl acetylacetonate. J. Mater. Chem. 12, 333–
338 (2001).
298. Manning, T. D. & Parkin, I. P. Atmospheric pressure chemical vapour deposition of tungsten
doped vanadium(IV) oxide from VOCl3, water and WCl6. J. Mater. Chem. 14, 2554–2559
(2004).
299. Piccirillo, C., Binions, R. & Parkin, I. P. Nb-Doped VO2 Thin Films Prepared by Aerosol-Assisted
Chemical Vapour Deposition. Eur. J. Inorg. Chem. 2007, 4050–4055 (2007).
300. Warwick, M., Smith, R., Furlan, N., Crane, J. & Binions, R. Electric Fields and Chemical Vapor
Deposition. in ECS Transactions 28, 1–13 (2010).
301. Nguyen, T. D. & Do, T. O. Solvo-hydrothermal approach for the shape-selective synthesis of
vanadium oxide nanocrystals and their characterization. Langmuir 25, 5322–5332 (2009).
302. Zhang, H. et al. A cost-effective method to fabricate VO2 (M) nanoparticles and films with
excellent thermochromic properties. J. Alloys Compd. 636, 106–112 (2015).
303. Nag, J. & Haglund Jr, R. F. Synthesis of vanadium dioxide thin films and nanoparticles. J. Phys.
Condens. Matter 20, 264016 (2008).
304. Zhong, L. et al. Star-shaped VO2 (M) nanoparticle films with high thermochromic performance.
CrystEngComm 17, 5614–5619 (2015).
305. Billik, P. et al. Synthesis and Transport Properties of Nanostructured VO2 by Mechanochemical
Processing. Meas. Sci. Rev. 11, 29–33 (2011).
306. Zhang, Y. et al. Influence of different additives on the synthesis of VO2 polymorphs. Ceram. Int.
39, 8363–8376 (2013).
307. Wang, H., Yi, X., Chen, S. & Fu, X. Fabrication of vanadium oxide micro-optical switches. Sensors
Actuators, A Phys. 122, 108–112 (2005).
308. Pauli, S. A. et al. X-ray diffraction studies of the growth of vanadium dioxide nanoparticles. J.
Appl. Phys. 102, (2007).
309. Chae, B.-G. et al. Highly Oriented VO2 Thin Films Prepared by Sol-Gel Deposition. Electrochem.
Solid-State Lett. 9, C12–C14 (2006).
310. Krawiec, P.; Kaskel, S. in Nanoscale Materials in Chemistry (eds. Klabunde, K. J. & Richards, R.
M.) 111–125 (John Wiley & Sons, 2009).
311. Wu, L. et al. Understanding the mechanical properties of vanadium carbides: Nano-indentation
measurement and first-principles calculations. J. Alloys Compd. 548, 60–64 (2013).
312. Didziulis, S. V. & Butcher, K. D. A perspective on the properties and surface reactivities of
carbides and nitrides of titanium and vanadium. Coord. Chem. Rev. 257, 93–109 (2013).
313. Stanciu, V. I., Vitry, V. & Delaunois, F. Tungsten carbide powder obtained by direct
carburization of tungsten trioxide using mechanical alloying method. J. Alloys Compd. 659,
302–308 (2016).
314. Hussainova, I. Effect of microstructure on the erosive wear of titanium carbide-based cermets.
Wear 255, 121–128 (2003).
315. Liu, K., Li, X. P., Rahman, M. & Liu, X. D. CBN tool wear in ductile cutting of tungsten carbide.
Wear 255, 1344–1351 (2003).
316. Hannink, R. H. J. & Murray, M. J. Elastic moduli measurements of some cubic transition metal
carbides and alloyed carbides. Journal of Materials Science 9, 223–228 (1974).
317. Kral, C., Lengauer, W., Rafaja, D. & Ettmayer, P. Critical review on the elastic properties of
transition metal carbides, nitrides and carbonitrides. J. Alloys Compd. 265, 215–233 (1998).
318. Matrosov, Y. I. & Anashenko, V. N. Vanadium nitride in low-alloy low-carbon steels. Met. Sci.
Heat Treat. 13, 893–896 (1971).
319. Muije, P., Horton, R. M. & Duran, S. A. Kinetics of solid vanadium-nitrogen reactions at high
temperatures. Metall. Trans. 5, 97–104 (1974).
320. Choi, D., Blomgren, G. E. & Kumta, P. N. Fast and reversible surface redox reaction in
nanocrystalline vanadium nitride supercapacitors. Advanced Materials 18, 1178–1182 (2006).
260
321. Oyama, S. T., Kapoor, R., Oyama, H. T., Hofmann, D. J. & Matijevic, E. Topotactic synthesis of
vanadium nitride solid foams. 5814, 2–6 (1993).
322. Vaidhyanathan, B., Agrawal, D. K. & Roy, R. Novel synthesis of nitride powders of microwave-
assisted combustion. Journal of Materials Research 15, 974–981 (2000).
323. Huang, J. W., Peng, H. & Xia, G. B. Microwave synthesis of vanadium nitride for industrial
applications. Ironmak. Steelmak. 36, 110–114 (2009).
324. Huang, T. et al. Hydrothermal synthesis of vanadium nitride and modulation of its catalytic
performance for oxygen reduction reaction. Nanoscale 6, 9608–13 (2014).
325. Hong, Y. C., Shin, D. H. & Uhm, H. S. Production of vanadium nitride nanopowders from gas-
phase VOCl3 by making use of microwave plasma torch. Mater. Chem. Phys. 101, 35–40 (2007).
326. Fix, R., Gordon, R. & Hoffman, D. Chemical Vapor Deposition of Vanadium, Niobium, and
Tantalum Nitride Thin Films. Chem. Mater. 5, 614–619 (1993).
327. Chen, L. et al. A room-temperature synthesis of nanocrystalline vanadium nitride. Solid State
Commun. 132, 343–346 (2004).
328. Parkin, I. P. & Elwin, G. S. Atmospheric pressure chemical vapour deposition of vanadium
nitride and oxynitride films on glass from reaction of VCl4 with NH3. J. Mater. Chem. 11, 3120–
3124 (2001).
329. Newport, A., Carmalt, C. J., Parkin, I. P. & O’Neill, S. A. Formation of VN front VCl4 and
NH(SiMe3)2 by APCVD - A potential solar control coating. Eur. J. Inorg. Chem. 4286–4290
(2004).
330. Giordano, C., Erpen, C., Yao, W., Milke, B. & Antonietti, M. Metal nitride and metal carbide
nanoparticles by a soft urea pathway. Chem. Mater. 21, 5136–5144 (2009).
331. Buha, J., Djerdj, I., Antonietti, M. & Niederberger, M. Thermal transformation of metal oxide
nanoparticles into nanocrystalline metal nitrides using cyanamide and urea as nitrogen source.
Chem. Mater. 19, 3499–3505 (2007).
332. Ma, J. et al. Low temperature synthesis of vanadium carbide (VC). Mater. Lett. 63, 905–907
(2009).
333. Aguzzoli, C., Figueroa, C. A., de Souza, F. S., Spinelli, A. & Baumvol, I. J. R. Corrosion and
nanomechanical properties of vanadium carbide thin film coatings of tool steel. Surf. Coatings
Technol. 206, 2725–2731 (2012).
334. Yang, G., Sun, X., Li, Z., Li, X. & Yong, Q. Effects of vanadium on the microstructure and
mechanical properties of a high strength low alloy martensite steel. Mater. Des. 50, 102–107
(2013).
335. Ollilainen, V., Kasprzak, W. & Holappa, L. The effect of silicon, vanadium and nitrogen on the
microstructure and hardness of air cooled medium carbon low alloy steels. J. Mater. Process.
Technol. 134, 405–412 (2003).
336. S., G. & Cochrane, R. C. Influence of cooling rate and tempering on precipitation and hardness
of vanadium microalloyed steel. Mater. Des. 26, 486–492 (2005).
337. Chicco, B., Borbidge, W. E. & Summerville, E. Experimental study of vanadium carbide and
carbonitride coatings. Mater. Sci. Eng. A 266, 62–72 (1999).
338. Aghaie-Khafri, M. & Fazlalipour, F. Kinetics of V(N,C) coating produced by a duplex surface
treatment. Surf. Coatings Technol. 202, 4107–4113 (2008).
339. Schmuecker, S. M. & Leonard, B. M. Formation mechanism of nanostructured metal carbides
via salt-flux synthesis. Inorg. Chem. 54, 3889–3895 (2015).
340. Claridge, J. B., York, A. P. E., Brungs, A. J. & Green, M. L. H. Study of the temperature-
programmed reaction synthesis of early transition metal carbide and nitride catalyst materials
from oxide precursors. Chem. Mater. 12, 132–142 (2000).
341. Liu, F. et al. Synthesis and characterization of vanadium carbide nanoparticles by thermal
refluxing-derived precursors. J. Mater. Sci. 46, 3693–3697 (2011).
342. Chen, Y., Zhang, H., Ye, H. & Ma, J. A simple and novel route to synthesize nano-vanadium
carbide using magnesium powders, vanadium pentoxide and different carbon source. Int. J.
Refract. Met. Hard Mater. 29, 528–531 (2011).
261
343. Mahajan, M., Singh, K. & Pandey, O. P. Single step synthesis of nano vanadium carbide - V8C7
phase. Int. J. Refract. Met. Hard Mater. 36, 106–110 (2013).
344. Mahajan, M., Singh, K. & Pandey, O. P. Synthesis of vanadium carbide nanoparticles by thermal
decomposition of the precursor. AIP Conf. Proc. 1536, 271–272 (2013).
345. Poirier, L. et al. Vanadocene used as a precursor for the chemical vapor deposition of vanadium
carbide at atmospheric pressure. Thin Solid Films 249, 62–69 (1994).
346. Brown, G. M. & Maya, L. Cyclopentadienylvanadium carbonyl derivatives as precursors to
vanadium carbide. Inorg. Chem. 28, 2007–2010 (1989).
347. Lo, K. H., Shek, C. H. & Lai, J. K. L. Recent developments in stainless steels. Mater. Sci. Eng. R
Reports 65, 39–104 (2009).
348. Sourmail, T. Precipitation in creep resistant austenitic stainless steels. Mater. Sci. Technol. 17,
1–14 (2001).
349. Cross, W. B. et al. Tungsten oxide coatings from the aerosol-assisted chemical vapor deposition
of W(OAr)6 (Ar = C6H5, C6H4F-4, C6H3F2-3,4); photocatalytically active ??-WO3 films. Chem.
Mater. 15, 2786–2796 (2003).
350. Buijnsters, J. G. et al. CVD diamond deposition on steel using arc-plated chromium nitride
interlayers. Diam. Relat. Mater. 11, 536–544 (2002).
351. Halada, G. P. & Clayton, C. R. Comparison of Mo-N and W-N synergism during passivation of
stainless steel through x-ray photoelectron spectroscopy and electrochemical analysis. J. Vac.
Sci. Technol. A 11, 2342–2347 (1993).
352. Taylor, A. et al. A bioinspired solution for spectrally selective thermochromic VO2 coated
intelligent glazing. Optics express 21, A750--A764 (2013).
353. Liu, C., Wang, N. & Long, Y. Multifunctional overcoats on vanadium dioxide thermochromic thin
films with enhanced luminous transmission and solar modulation, hydrophobicity and anti-
oxidation. Appl. Surf. Sci. 283, 222–226 (2013).
354. Soon, G. K. & Hyeon, T. Colloidal chemical synthesis and formation kinetics of uniformly sized
nanocrystals of metals, oxides, and chalcogenides. Acc. Chem. Res. 41, 1696–1709 (2008).
355. Kwon, S. G.; Hyeon, T. Formation mechanisms of uniform nanocrystals via hot-injection and
heat-up methods. Acc Chem Res 41 (12), 1696–1709.
356. Bronstein, L. M. et al. Nanoparticles by decomposition of long chain iron carboxylates: From
spheres to stars and cubes. Langmuir 27, 3044–3050 (2011).
357. Paik, T. et al. Solution-Processed Phase-Change Vanadium Oxide ( VO x ) Nanocrystals. ACS
Nano 797–806 (2013).
358. Niederberger, M., Bartl, M. H. & Stucky, G. D. Benzyl alcohol and transition metal chlorides as a
versatile reaction system for the nonaqueous and low-temperature synthesis of crystalline
nano-objects with controlled dimensionality. J. Am. Chem. Soc. 124, 13642–13643 (2002).
359. Niederberger, M. Nonaqueous sol-gel routes to metal oxide nanoparticles. Acc. Chem. Res. 40,
793–800 (2007).
360. Sun, S. & Zeng, H. Size-Controlled Synthesis of Magnetite Nanoparticles Size-Controlled
Synthesis of Magnetite Nanoparticles. 124, 8204–8205 (2002).
361. Demeter, M., Neumann, M. & Reichelt, W. Mixed-valence vanadium oxides studied by XPS.
Surf. Sci. 454, 41–44 (2000).
362. Alexander V. Naumkin, Anna Kraut-Vass, Stephen W. Gaarenstroom, and C. J. P. NIST X-ray
Photoelectron Spectroscopy Database. NIST Standard Reference Database 20, Version 4.1
(2012).
363. Yu, S., Fu, N., Gao, F. & Sui, Z. Synthesis of Vanadium Nitride by a One Step Method. J. Mater.
Sci. Technol 23, 43–46 (2007).
364. Ortega, A., Roldan, M. A. & Real, C. Carbothermal synthesis of vanadium nitride: Kinetics and
mechanism. Int. J. Chem. Kinet. 38, 369–375 (2006).
365. Liu, C.; Wang, N.; Long, Y. Carbon Nitride Deposited Using Energetic Species: A Two-Phase
System. Phys. Rev. Lett. 73 (1), 118–121 (1994).
366. Azhan, N. H. et al. Large modification in insulator-metal transition of VO2 films grown on Al2O3
262
(001) by high energy ion irradiation in biased reactive sputtering. J. Appl. Phys. 119, 55308
(2016).
367. Choi, J. G. Ammonia decomposition over vanadium carbide catalysts. J. Catal. 182, 104–116
(1999).
368. Hiden Analytical. Hiden HPR60 Molecular Beam Mass Spectrometer System vacuum analysis
surface science plasma diagnostics gas analysis HPR60 MBMS Overview. (2015).
263

Study of Gas-Phase Reactions in Tio Production and Functionalised Thin Films

Uploaded by

Copyright:

Available Formats

Study of Gas-Phase Reactions in Tio Production and Functionalised Thin Films

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Study of Gas-Phase Reactions in Tio Production and Functionalised Thin Films

Uploaded by

Copyright:

Available Formats

Study of Gas-Phase Reactions in TiO2

Production and Functionalised Thin Films

1.1. Project Overview 1

2. Analysis of Samples from Industrial TiO₂ Production 19

3. Reaction of Titanium Chloride and Vanadium Oxytrichloride 47

3.3.3.1. Dichloro(oxo) (2,4-Pentanedione) vanadium(V) [5] 74

3.3.3.2. Dichloro(oxo) (diethyl malonate) vanadium(V) [6] 78

3.3.3.3. Dichloro(oxo)(diethyl succinate) vanadium (V) [7] 83

3.3.3.4. Reaction of VOCl3 with Other Species 88

4.3.1.2. VOCl3 + Soya Oil 107

4.3.1.3. TiCl4 + 1000 ppm VOCl3 + Soya Oil 108

4.3.1.4. TiCl4 + 40,000 ppm VOCl3 + Soya Oil 109

5. Using Mass Spectrometry to Study Gas Phase Reactions in 120

5.1. Introduction 120

6. Aerosol Assisted CVD of Photocatalytically Active TiO₂/SiO₂ 181

6.1. Introduction 181

6.3.4. Reaction of TiCl4 with bis-trimethylsilylmalonate to give [13] 200

7. Derivatives of VOCl3 and Their Uses in Materials Synthesis 214

7.1 Introduction 214

7.4. Conclusions 244

Figure 105: Raman Spectra of films deposited from AACVD of

Figure 107: UV/Vis Absorption spectra of films deposited from AACVD of

Scheme 1: Reaction of TiCl4 with bis-isopropyl malonate to give tetrachloro(bis-

AACVD aerosol assisted chemical vapour deposition

1.1. Project Overview

reactivity in the Chloride Process.

deposition of thin films.

grown and were analysed.

1.2. Titanium Dioxide

1.2.1. Structure and Phases

when heated to 915 ˚C and 730 ˚C, respectively. 3

polymorphs it is brown in colour.

1.2.2. Uses of TiO2

large number of products, with a market worth $15.17 billion in 2014. 7

TiO₂ Usage 2003 TiO₂ Usage 2013

Plastic Paint Other

Figure 2: Breakdown of TiO2 use in 20031 compared with 2013.10

material exhibits super-hydrophilicity as well as photocatalysing the breakdown of dirt.12 This

responsible titanium dioxide’s effectiveness in killing bacteria. This is believed to be due to a

bacteria in light conditions typical of hospitals.16

used as an anode material.

cost of titanium containing mineral has fallen dramatically (Figure 3).

solid iron compounds damaging waste disposal relative to rutile.21

1.3.1. The Sulfate Process

Sulfate process has been in a state of decline.3

1.3.2. The Chloride Process

stages as described below and is shown in Figure 6.29

the following chlorination reaction:

TiO2(s) + 3/2 C(s) + 2 Cl2(g) → TiCl4(g) + CO(g) + ½ CO2(g)

the ore, most prominently:

2 VO2(s) + 3 Cl2(g) → 2 VOCl3(g) + O2

2 Fe2O3(s) + 6 Cl2(g) → 4 FeCl3(g) + 3 O2

2 Al2O3(s) + 3 Cl2(g) → 2 AlCl3(g) + 3 O2

SiO2(s) + 2 Cl2(g) → SiCl4(g) + 2 O2

quench tower with the TiCl4 and the carbon dioxide.

the top of the cyclone, still in their gaseous form.

TiCl4(g)(excess) + H2O(g) → TiOCl2.2HCl(l)

TiOCl2.2HCl(l) + AlCl3(l) → TiCl4(g) + AlOCl(s) + HCl(g)

are less soluble in the TiCl4 to crash out as solid.