Heterogeneous Photoelectrochemistry and
Photocatalysis of TiO2-based Nanomaterials
-Towards photocatalytic CO2 reduction-
DISSERTATION
zur Erlangung des Grades Dr. rer. nat.
der Fakultät für Chemie und Biochemie
der Ruhr-Universität Bochum.
vorgelegt von
Bastian Timo Mei
aus Hagen
Ruhr-Universität Bochum
Lehrstuhl für Technische Chemie
2013
Mei, Bastian Timo - 2013
Diese vorliegende Arbeit wurde im Zeitraum vom August 2009 bis zum Januar 2013 am
Lehrstuhl für Technische Chemie der Ruhr-Universität Bochum angefertigt.
Gutachter:
Dissertation eingereicht am:
Tag der Disputation:
Prof. Dr. Martin Muhler
Prof. Dr. Wolfgang Schuhmann
12.02.2013
19.04.2013
Danksagung
Zunächst möchte ich mich bei Herrn Prof. Dr. Martin Muhler für die Möglichkeit bedanken
diese Arbeit an seinem Lehrstuhl anzufertigen, sowie für wissenschaftliche Leitung. Außerdem
danke ich ihm für die vorhandenen Freiheiten während meiner Arbeit und das mir entgegengebrachte Vertrauen.
Herrn Prof. Dr. Wolfgang Schuhmann danke ich für die Übernahme des Koreferats.
Frau Dr. Jennifer Strunk danke ich für die groartige Betreuung während meiner Arbeit, die
erfolgreiche Zusammenarbeit im Rahmen des ”Photocatalysts for CO2 Reduction” Projekts
und die gute Leitung des Projekts. Allen Mitgliedern danke ich für die zahlreichen fachlichen
Diskussionen und die guten Ratschläge. Im Einzelnen möchte ich mich bei Dr. Freddy Oropeza
für sein stätiges Interesse, bei Ahmet Becerkili für die unzähligen Synthesen und bei Anna
Pougin für Ihre Hilfestellungen und Ideen bei Aufbau des Photoreaktors bedanken.
Frau Dr. Wilma Busser danke ich für die Unterstützung bei photokatalytischen Messungen und
der Präparation einiger Katalysatoren. Außerdem möchte ich mich für die vielen gewinnbringenden Diskussionen bedanken.
Vielen Dank auch an Dr. Thomas Reinecke für die Durchführung der XRD-Messungen und an
Frau Kirsten Keppler für die Durchführung der Elementaranalysen.
Frau Sigrid Kalender danke ich dafür, dass Sie immer ein offenes Ohr hatte und viele Tage mit
Kaffee erträglich machte.
Bei Frau Susanne Buse, Frau Lina Freitag und Frau Noushin Arshadi bedanke ich mich für
BET und IR Messungen.
Allen Mitarbeitern des Lehrstuhls für Technische Chemie danke ich für die freundliche Arbeitsatmosphäre und Hilfsbereitschaft während meiner Zeit am Lehrstuhl.
Dem Graduiertenkolleg des Sonderforschungsbereichs SFB 558 der Deutschen Forschungsgemeinschaft und der Research School der Ruhr-Universität Bochum danke ich für die finanzielle
Unterstützung.
Ein riesiges Dankeschön geht an meine Eltern, Geschwister und Großeltern, die mich jeweils auf ihre eigene Weise während des Studiums und der Promotion unterstützt und motiviert
haben. Danke, dass ich mich immer auf euch verlassen kann.
Besonders danken möchte ich meiner tollen Freundin Marie Holz, die auch in schwierigen
Zeiten zu mir hält und mir dadurch viel Kraft gibt.
Zusammenfassung
Die photokatalytische Umsetzung von Kohlenstoffdioxid mit Wasser zu C1-Basischemikalien
der chemischen Industrie wie z.B. Methanol oder Methan unter Nutzung des Sonnenlichtes bietet eine Möglichkeit, Erdöl oder Erdgas als primäre Energiequellen abzulösen und gleichzeitig
die Kohlenstoffdioxid-Bilanz zu verbessern, indem in einem idealen Prozess CO2 rezykliert
wird. Ziel dieser Arbeit war die Durchführung systematischer Untersuchungen an Titandioxidbasierten Photokatalysatoren für die CO2 -Reduktion und der Aufbau einer geeigneten Testanlage für photokatalytische Aktivitätsmessungen in dieser Gasphasenreaktion.
In der vorliegenden Arbeit wurden, ausgehend von den Arbeiten der Gruppe um Masakazu
Anpo, unterschiedliche Strukturen von auf mesoporösem Siliziumdioxid geträgerten
Titandioxid-Spezies (TiOx -Spezies) synthetisiert.
Die Bestimmung der so erhaltenen
TiOx -Spezies erfolgte durch die Kombination komplementärer spektroskopischer Untersuchungsmethoden. Es konnte gezeigt werden, dass in Abhängigkeit von der Titanbeladung
gezielt isolierte, oligomerisierte oder polymerisierte TiOx -Spezies synthetisiert werden konnten. Somit wurden TiO2-basierte Materialien hergestellt, die als Ausgangsmaterialen für die
Optimierung TiO2-basierter Photokatalysatoren reproduzierbar zur Verfügung stehen. Des
Weiteren eignen sich die synthetisierten Materialien dazu, eine detaillierte Studie des Einflusses der Titandioxid-Struktur durchzuführen und somit Struktur-Aktivitäts-Beziehungen in
der photokatalytischen CO2 -Reduktion zu erhalten.
Die photokatalytische Aktivität isolierter TiOx -Spezies in der CO2 Reduktion konnte im Rahmen dieser Arbeit bestätigt werden. Methan und höhere Alkane konnten als Produkte detektiert werden und Formaldehyd konnte durch spektroskopische Untersuchungen als Intermediat
der photokatalytischen Reduktion von CO2 identifiziert werden. Die Abscheidung von GoldNanopartikeln auf der Oberfläche des Photokatalysators steigerte die Hydrieraktivität des Materials, wodurch höhere Methanausbeuten erzielt wurden und die Anreicherung von stabilen
Formaldehyd-Spezies auf der Oberfläche des Photokatalysators verringert wurde. Des Weiteren konnte mit Hilfe geeigneter Testreaktionen gezeigt werden, dass die photokatalytische
Aktivität der Materialien mit dem Polymerisationsgrad der TiOx-Spezies abfällt.
Die Charakterisierung der spezifischen Wechselwirkungen von CO2 mit isolierten und
polymerisierten TiOx-Spezies mittels temperaturprogrammierter Desorption und InfrarotSpektroskopie zeigten, dass CO2 lediglich bei tiefen Temperaturen mit den Photokatalysatoren
wechselwirkt, wodurch eine Aktivierung des CO2 auf der Katalysatoroberfläche unter realen
Reaktionsbedingungen ausgeschlossen werden kann. Es konnte jedoch gezeigt werden, dass
die Wechselwirkungen von CO2 mit dem Katalysator durch Einbringen kleiner ZinkoxidAgglomerate verbessert wird. Systematische Untersuchungen an Zinkoxid-modifizierten TiOx haltigen Materialien zeigten, dass die Photonenanregung der TiOx -Spezies die photokat-
alytische Aktivität der Materialien bestimmt. Es konnte außerdem gezeigt werden, dass
das Trägermaterial Einfluss auf die photokatalytische Aktivität der TiOx -Spezies hat und
die spezifischen Wechselwirkungen zwischen TiOx - und Zinkoxid-Spezies die Aktivität in
den durchgeführten Testmessungen bestimmen. Abschließend wurde durch hochauflösende
Transmissionselektronenmikroskopie-Messungen gezeigt, dass die Mobilität der TiOx -Spezies
durch die Anwesenheit von Zinkoxid verringert wird, wodurch die Abscheidung von GoldNanopartikeln auf den Materialien beeinflusst wird.
Im zweiten Teil der Arbeit wurde eine neuartige Syntheseroute basierend auf dem
Sprühtrocknungsprozess zur Präparation dotierter TiO2-Materialien entwickelt. Ausgehend
von wässrigen Lösungen von Titan(IV)-oxidsulfat erlaubt die Sprühtrocknungsmethode, homogene Verteilungen eines Dotierelements in TiO2 zu erhalten und durch anschließende
Kalzination dotiertes TiO2 in unterschiedlichen polymorphen Strukturen zu synthetisieren.
niobdotiertes TiO2 wurde mit Hilfe dieser Syntheseroute erfolgreich präpariert und die erhaltenen Materialien konnten erfolgreich in der photoelektrochemischen Wasserspaltung eingesetzt
werden. Der Einbau geringer Konzentrationen von Niob in die Gerüstrstuktur der TiO2 Materialien führte zur einer Erhöhung der gemessenen Photoströme um ca. 30 % im Vergleich
zu undotierten TiO2 Materialien, die über das Sprütrocknungsverfahren hergestellt wurden.
Die verbesserten Eigenschaften der niobmodifizierten TiO2 -Materialien konnte auf einen
verbesserten Elektronentransport in den Materialien zurückgeführt werden. Abschließend
wurde außerdem gezeigt, dass die verbesserten Eigenschaften des Elektronenkollektors auch
nach Modifikation der Oberfläche mit einer Polyheptazin-Hülle beibehalten werden konnten,
wodurch die Materialien unter sichtbarem Licht photokatalytische Aktivität zeigen.
Die Ergebnisse dieser Arbeit zeigen somit eindrucksvoll, dass TiO2 -basierte Materialien gezielt
mit verschiedenen Strukturen oder Dotierungen synthetisiert werden konnten, deren Anwendbarkeit in unterschiedlichen photokatalytischen Prozessen demonstriert wurde. Dabei wurden
neue Strategien verfolgt, um die Aktivität bekannter Systeme zu verbessern und die StrukturWirkungs-Beziehungen der Materialien genauer zu charakterisieren.
Table of Contents
Table of Contents
i
List of Figures
iii
List of Tables
v
Symbols
vii
Abbreviations
ix
1 Introduction
1
2 Basic Principles
2.1 Physical and Chemical Properties of Semiconductors . . .
2.1.1 The Energy Band Model . . . . . . . . . . . . . .
2.1.2 Optical Properties of Semiconductors . . . . . . .
2.2 Heterogeneous Photocatalysis on Semiconductor Particles .
2.2.1 Oxide Photocatalysts - Band gap Engineering . . .
2.2.2 Metal-modified Photocatalysts . . . . . . . . . . .
2.3 TiO2 in Photocatalysis . . . . . . . . . . . . . . . . . . .
2.4 Single-site Photocatalysts . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5
5
5
8
10
12
13
14
16
3 Literature Review - Photocatalytic CO2 Reduction on TiO2 -related Materials
19
4 Development of a Gas-phase Photoreactor
23
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide .
5.1.1 Short Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . .
5.1.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 Photocatalytic CO2 reduction using Au-modified TiOx/SBA-15 materials . .
5.2.1 Short Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . .
5.2.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
30
31
35
52
54
55
56
63
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Table of Contents
ii
5.3 Effects of Au and ZnO on the
TiOx /SBA-15 materials . . . . . .
5.3.1 Short Introduction . . . . .
5.3.2 Results and Discussion . .
5.3.3 Conclusions and Outlook .
structure and
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
photocatalytic activity of
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
64
65
67
80
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
6.1 The synthesis of Nb-doped TiO2 nanoparticles by spray drying . . . . . . . . .
6.1.1 Short Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2 TiO2 -polyheptazine hybrid materials . . . . . . . . . . . . . . . . . . . . . . .
6.2.1 Short Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.3 Nitrogen-modified Nb-doped TiO2 : Towards red TiO2 . . . . . . . . . . . . . .
6.3.1 Short Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.3.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . .
6.3.3 Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . .
81
82
83
86
97
98
99
101
104
105
106
108
112
7 Summary and Prospects
113
Bibliography
I
8 Appendix
8.1 Supporting Figures Chapter 4 . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.2 Supporting Figures Chapter 5 . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.3 Supporting Figures Chapter 6 . . . . . . . . . . . . . . . . . . . . . . . . . . .
XIII
XIII
XIV
XXII
9 Curriculum Vitae
XXVII
List of Figures
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
Energy band diagram . . . . . . . . . . . . . . . . . . .
Characteristics of semiconductors . . . . . . . . . . . .
Optical transitions . . . . . . . . . . . . . . . . . . . . .
Heterogeneous photocatalysis on semiconductor particles
Band gap Engineering . . . . . . . . . . . . . . . . . . .
Metal-modified photocatalysts . . . . . . . . . . . . . .
TiO2 crystal structures . . . . . . . . . . . . . . . . . .
Ti-oxide single-site catalysts . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
6
7
9
11
12
14
15
16
4.1 Flow sheet of the photoreactor . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2 GC operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.3 GC calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
5.10
5.11
5.12
5.13
5.14
5.15
5.16
5.17
5.18
5.19
5.20
XPS and pore size distribution of Ti(x)/Zn(y)/SBA . .
UV-Vis results of Ti(x)/Zn(y)/SBA . . . . . . . . . . .
XPS results of Ti(x)/Zn(y)/SBA . . . . . . . . . . . .
XANES of Ti(x)/SBA-15 . . . . . . . . . . . . . . . .
XANES and EXAFS of ZnOx -containing SBA-15 . . .
CO2 TPD profiles . . . . . . . . . . . . . . . . . . . .
CO2 adsorption on Zn1.0/SBA - UHV-FTIR . . . . . .
CO2 adsorption on Zn0.7/Ti1.0/SBA - UHV-FTIR . . .
CO2 adsorption on Ti1.0/SBA - UHV-FTIR . . . . . .
Photocatalytic CO2 reduction . . . . . . . . . . . . . .
Photocatalytic CO2 reduction in different environments
Difference DRIFTS of Ti/SBA-15 . . . . . . . . . . .
Au photo-deposition on Ti(x)/SBA . . . . . . . . . . .
TA hydroxylation with Au/Ti(x)/Zn(y)/SBA . . . . . .
DR UV-Vis spectra of Au/Ti(x)/Zn(y)/SBA . . . . . .
BF-TEM and EFTEM of Au/Ti1.0/SBA . . . . . . . .
HAADF-STEM of Au/Ti1.0/SBA . . . . . . . . . . .
HAADF-STEM of Au nanoparticles in Au/Ti1.0/SBA .
HRTEM and HAADF-STEM of Au/Zn0.3/Ti1.2/SBA .
Tomographic reconstruction of Au/Zn0.3/Ti1.2/SBA .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
37
38
39
42
43
45
48
50
51
57
60
62
68
70
73
74
75
76
77
78
List of Figures
iv
6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8
6.9
6.10
6.11
TG of different precursors . . . . . . . . . . . . . . .
TG of Nb-doped TiO2 . . . . . . . . . . . . . . . . .
XRD of Nb-doped TiO2 . . . . . . . . . . . . . . . .
Raman and XRD . . . . . . . . . . . . . . . . . . .
Texture and morphology of Nb-doped TiO2 . . . . .
XPS of Nb-doped TiO2 . . . . . . . . . . . . . . . .
UV-Vis of Nb-doped TiO2 . . . . . . . . . . . . . .
Hybrid photoelectrodes . . . . . . . . . . . . . . . .
Photoelectrochemical performance of Nb-doped TiO2
Degradation experiments with Nb-doped TiO2 . . . .
N 1s region of Nb/N co-doped TiO2 . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
87
88
90
91
92
94
95
100
102
108
110
List of Tables
2.1 Wavenumbers of different excitations . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Wavenumber and wavelength TiOx species . . . . . . . . . . . . . . . . . . . . 17
4.1 GC settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.2 Retention times . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.1
5.2
5.3
5.4
BET and EA results of Ti(x)/Zn(y)/SBA .
NH3 and CO2 TPD . . . . . . . . . . . .
Mode assignments of (para-)formaldehyde
ICP-OES results of Au/Ti(x)/Zn(y)/SBA .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
36
46
63
72
6.1 Refined lattice parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.2 Surface areas of Nb-doped TiO2 . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.3 Elemental analysis of co-doped TiO2 . . . . . . . . . . . . . . . . . . . . . . . 109
Symbols
symbol
denotion
α
d
d(BET)
d(XRD)
∆G
∆µ
e−
Ef
∗E
Fn
∗E
Fp
Eg
EOredox
f(E)
F(R∞ )
h+
h
I0
I
λ
NA
nads
N(E)
ν
p
p/p0
pH
R
ρ
ϕ
T
absorption coefficient
thickness of a material
diameter determined by BET
diameter determined by XRD
change in Gibbs energy
energy difference between ∗ EFn and ∗ EF p
electron
Energy of the Fermi level
quasi-Fermi level of electrons
quasi-Fermi level of holes
band gap energy
standard reduction potential
Fermi-Dirac distribution
Kubelka-Munk function
hole in the valence band
Planck constant
incident light
transmitted light
wavelength
Avogadro constant
adsorbed amount of substance
density of states
frequency
pressure
relative pressure
pH-value
gas constant
density
work function
transmittance
unit
[m]
[nm]
[nm]
[kJ/mol]
[eV ]
[eV ]
[Js]
[m]
[mol −1 ]
[mol]
[s−1 ]
[mbar]
[J K −1 mol −1 ]
[kg m−3 ]
[eV ]
Symbols
viii
symbol
denotion
unit
Ti(Oi Pr)4
θ
V
Vpore
Zn(acac)2
titanium tetraisopropoxide
diffraction angle
volume
total pore volume
zinc acetylacetonate
[◦ ]
[ml]
[ml g−1 ]
Abbreviations
abbreviation
denotion
AAS
ANO
BET
BJH
CB
DOE
DOS
DR
DTG
EA
EDX
EPR
EXAFS
FFT
FID
FTIR
FWHM
GC
HAADF
HOMO
ICP-OES
Atomic Absorption Spectroscopy
Ammonium Niobate (V)Oxalate
Brunnauer-Emmet-Teller
Barret-Joyner-Halenda
Conduction Band
Department Of Energy
Density Of States
Diffuse Reflectance
Differential Thermogravimetry
Elemental Analysis
Energy Dispersive X-Ray spectroscopy
Electron Paramagnetic Resonance
Extended X-Ray Absorption Fine Structure
Fast Fourier Transform
Flame Ionization Detector
Fourier Transform Infrared spectroscopy
Full Width at Half Maximum
Gas Chromatograph
High-angle Annular Dark-Field imaging
Highest Occupied Molecular Orbital
Inductively Coupled Plasma Optical Emission
Spectrometry
Intergovernmental Panel on Climate Change
Incident Photon to Current Efficiency
Infrared
Indium Tin Oxide
International Union of Pure and Applied Chemistry
Ligand to Metal Charge Transfer
Lowest Unoccupied Molecular Orbital
Mobil Crystalline Material-48
IPCC
IPCE
IR
ITO
IUPAC
LMCT
LUMO
MCM-48
Abbreviations
x
abbreviation
denotion
MFC
NHE
PVA
PVS
QMS
RHE
SBA-15
SEM
TA
TCD
TEOS
TEM
TG
TGA
TOC
TPD
UHV
VB
XANES
XAS
XPS
XRD
Mass flow controller
Normal Hydrogen Electrode
Polyvinyl Alcohol
Polyvinyl Siloxane
Quadrupole Mass Spectrometry
Reversible Hydrogen Electrode
Santa Barbara-No.15
Scanning Electron Microscopy
Terephthalic Acid
Thermal Conductivity Detector
Tetraethylorthosilicate
Transmission Electron Microscopy
Thermo Gravimetry
Thermal Gravimetric Analysis
Total Organic Carbon
Temperature Programmed Reduction
Ultra High Vaccum
Valence Band
X-ray Absorption Near Edge Structure
X-ray Absorption Spectroscopy
X-ray Photoelectron Spectroscopy
X-ray diffraction
1 Introduction
During the past four decades photochemistry has been a strongly growing field of scientific
interest. Certainly, the pioneering reports by Fujishima and Honda [1] on solar water splitting
into hydrogen and oxygen on a TiO2 photoelectrode in the early 1970’s revealed the practical
potential of photochemical processes for solar energy conversion and storage on semiconductor
materials. Subsequently, the potential of photochemistry was verified for several reactions, e.g.
the removal of organic pollutants from water or air, [2,3] organic synthesis, [4] and for the reduction of carbon dioxide, which was first published in 1979 by Inoue et al. [5] These examples
clearly demonstrated that photocatalysis and photoelectrochemistry exhibit the possibility to
convert thermodynamically stable raw materials, such as water and carbon dioxide, into valuable chemicals with the help of solar energy and a semiconductor catalyst. In a recommended
definition by the International Union of Pure and Applied Chemistry (IUPAC), photocatalysis
is described as the ”Change in the rate of a chemical reaction or its initiation under the action
of ultraviolet, visible, or infrared radiation in the presence of a substance - the photocatalyst that absorbs light and is involved in the chemical transformation of the reaction partners”. [6] In
recent years the developed photocatalytic concepts have gained excessive attention as one of
the most important scientific challenges nowadays is to secure the futures energy supply. [7]
The need for new energy sources can be rationalized considering that today’s energy consumption is mainly covered by fossil fuels (∼ 87 %), [8] which, at reasonable cost, will supply the
world’s energy demand for several years. However, meeting global energy consumption in a
sustainable, carbon-neutral manner is desirable. For instance the USA’s Department of Energy
(DOE) estimated that carbon-free energy will be needed by 2050 [9] and the Intergovernmental
Panel on Climate Change (IPCC) showed that the global warming is mostly due to human activities and the release of green house gases, like carbon dioxide. [10] Biomass conversion will
always compete with the human food supply and wind energy as well as energy from photovoltaic devices is accompanied by severe fluctuations. Therefore, energy storage systems are
required, which of course is a scientific challenge by itself. However, the earth receives 5,000
times more energy from the sun than mankind uses over a year. [11,12] Thus, it is arguable that
utilizing sunlight to convert energy-poor to energy-rich molecules is a promising technology.
Nature already showed us by million years of successful photosynthesis that it is accomplishable. [7] In this sense replacing fossil fuels by solar chemical fuels is a goal of current research
and even though it is arguable to which degree solar irradiance can be efficiently converted into
usable energy-rich molecules, scientists have to explore the possibility of a partly solar energy-
1 Introduction
2
driven society with a reasonably high effort. One of the likely solar fuels would be hydrogen
in case it is efficiently produced from solar water splitting. [8,13] The long-term goal within this
field of research can be defined as the direct hydrogenation of the green house gas carbon dioxide into high-energy carriers using water. [8] The energy required to bridge the thermodynamic
gap between reactants and desired products will be supplied by radiation. Considering the conversion of CO2 and water to methanol, for instance, 698.7 kJ/mol are needed to overcome the
thermodynamic gap of this endergonic reaction [14] :
− CH OH + 3 O
CO2 + 2H2 O →
2
3
hν
2
As this process basically mimics the production of polysaccharide conducted by green plants it
is often referred to as ”artificial photosynthesis”.
Besides the photochemical process, strategies have already been developed to diminish the
atmospheric CO2 content. Numerous efforts have been put into the field of CO2 capture and
separation by high performance adsorbents. But there are also catalytic reactions, where this
highly thermodynamically stable and therefore almost unreactive molecule is used in industrial
relevant processes [15] :
• CO2 is used as a reactant in the urea synthesis which is an important fertilizer,
• in the Kolbe-Schmitt process to form salicylic acid,
• cyclic organic carbonates and epoxides are synthesized from CO2 ,
• and CO2 participates in the methanol synthesis over copper catalysts:
CO2 + 3H2 → CH3 OH + H2 O.
Despite these efforts it was estimated that since the industrial revolution the content of carbon
dioxide in the atmosphere increased by ∼ 35 % to its current level of 383 ppm [16] and recent
considerations claimed that at the current increase of atmospheric CO2 the global temperature
will increase of ∼ 6 ◦ C before the end of the century. [17] Today, still less than 1 % of the overall
emitted CO2 is chemically converted [12] and mostly CO2 is not recycled at all. Instead, CO2 as
a reactant is mainly provided by synthesis gas (CO, CO2 , and H2 ) produced from fossil fuels.
Furthermore, the hardly reactive CO2 molecule has to be activated, e.g. by elevated temperatures facilitating an unfavorable energy balance and raising the carbon footprint. Alternatively,
this work will explore the photochemical fixation of CO2 in useful base chemicals by the help
of photocatalytically active titania materials.
Titania-related materials seem to be the most promising semiconductors for many photocatalytic and photoelectrochemical applications, particularly due to its commercial availability,
non-toxicity, and excellent stability against photocorrosion. [18–20] Thus, it is not remarkable
that titania-based materials are frequently studied for photocatalytic CO2 reduction and other
3
reactions. While the main focus of this dissertation is the study of the photocatalytic CO2 reduction on isolated titania species incorporated in an mesoporous silica matrix, in the second
part bulk TiO2 materials are discussed. The key points and questions addressed in the first part
of this dissertation are:
1. What are the requirements of a gas-phase reactor for photocatalytic CO2 reduction? According to the requirements the gas-phase reactor should be developed allowing for trace
gas analysis and reproducible CO2 reduction experiments.
2. A reproducible synthesis procedure for isolated titania species with high titania loading
and a reliable characterization routine to distinguish between isolated and polymerized
titania should be developed. Synthesis and characterization should be applicable for a
variety of different isolated metal oxide materials.
3. Are isolated titania species really active in the photocatalytic CO2 reduction and what are
the main products of the photocatalytic CO2 reduction reaction on isolated titania sites?
What is the effect of titania polymerization?
4. Is CO2 interacting with isolated titania species?
5. How can materials with beneficial CO2 adsorption properties be synthesized on the basis
of a material containing isolated titania sites? Is there any relation between the CO2
adsorption and the photocatalytic activity of the materials?
Certainly, the use of bulk TiO2 in a photocatalytic CO2 reduction process is preferable due
to its commercial availability and due to its lower band gap, which can be even reduced by
doping. However, in recent studies isolated titania appeared to be more active in comparison
to bulk TiO2 . [21] Nevertheless, the thorough investigation of the CO2 reduction mechanism on
isolated and polymerized titania species may later on provide the knowledge to design highly
active bulk TiO2 materials. In this sense it is desirable to study the properties of bulk TiO2 and
the key points and questions, which were addressed in these studies are:
1. A simple and reliable synthesis method for the preparation of TiO2 should be developed,
which should offer the possibility to synthesize doped TiO2.
2. Do the synthesized TiO2 or doped TiO2 materials possess exceptional properties in a
well-known photocatalytic process?
3. Are there further modification possibilities, e.g. to improve the light-harvesting properties and to increase the photocatalytic activity of TiO2 ?
Finally, the outline of this thesis can be summarized as follows. Chapter 2 will shortly introduce the basic concepts of semiconductor physics and their impact on photocatalysis with
semiconductor materials in general and particularly on TiO2 and isolated titania species. A
short literature review on photocatalytic CO2 reduction on isolated titania species and TiO2 is
given in chapter 3. In chapter 4 the developed gas-phase photoreactor will be described and
4
1 Introduction
the calibration data will be shown. The results obtained in the studies performed with isolated
titania are shown in chapter 5. The chapter is divided in three sections. First, the synthesis and
the characterization results are discussed and the interaction of CO2 is studied. A possibility
to increase the CO2 adsorption by incorporation of zinc oxide is demonstrated. In the second
section the feasible photocatalytic CO2 reduction on isolated titania species is presented and a
possible intermediate of the photocatalytic CO2 reduction on isolated titania is identified. Additionally, differences in the photocatalytic activities of gold-modified materials are discussed.
The effect of zinc oxide species incorporated into the titania materials is investigated by means
of photo-deposition of gold nanoparticles, the hydroxylation of terephthalic acid, and high resolution transmission electron microscopy. Corresponding to chapter 5 chapter 6 is divided into
three sections dealing with a the description of a novel synthesis technique for metal doped
TiO2 materials, which were subsequently highlighted to exhibit unique properties in photoelectrochemical water oxidation and finally first results for an improved visible light harvesting
of non-metal/metal co-doped TiO2 are shown. Finally, the main results obtained within the
scope of this dissertation will be summarized and an outlook will be given in chapter 7.
2 Basic Principles
2.1 Physical and Chemical Properties of Semiconductors
Semiconductors are materials with conductivity between the conductivity of conductors and
insulators. The typical semiconducting properties are caused by atomic interactions, hence by
the nature of the chemical bond and the resulting lattice structure of the material. At absolute
zero temperature semiconductors are not conductive. The intrinsic conductivity increases, to a
certain extent, with increasing temperature. Depending on the temperature the electron mobility
increases and electron excitation into the conduction band is becoming favored. To describe
this behavior the energy band model and the Fermi function are used.
2.1.1 The Energy Band Model
The energy band model, derived as an extension of molecular orbital theory, describes the
interaction of atoms in an ideal crystal and is used to specify the conductivity of metals, semiconductors, and insulators. [22] The model mainly consists of two stacked energy bands (valence
and conduction band) separated by a gap (band gap). In fact the valence band is the highest
occupied band of the lattice structure. Contrary to the valence band the conduction band is the
lowest unoccupied band in the ground state (T = 0 K). The separation or more precisely the size
of the band gap influences the conductivity of the material. Therefore, the energy band model
is a convenient way to describe the electronic and optical properties of a semiconductor.
The atom model developed by Bohr describes a single atom by discrete energy states, which
can be occupied by up to two electrons. In molecular orbital theory the formation of bonding,
antibonding and non-bonding orbitals due to interaction between several atoms is described. [20]
These interactions are possible across several inter-atomic distances and with increasing number of interacting atoms the discrete levels are getting more and more indistinguishable, resulting in broad energy regions or bands (Fig. 2.1). The number of electrons incorporated in one of
these bands is similar to twice the number of atoms involved and the band width increases with
decreasing binding strength between atomic core and electron. The weekly bound electrons
are easily released from the core and hence they can interact with neighboring atoms resulting
in mobile electrons. In case of the valence electrons they are completely merged together and
2 Basic Principles
6
a)
conduction band
1023
E3
E3
valence band
E2
E2
E1
E1
1
2
10 23
3
atoms
x
b)
band gap
-
-
-
-
insulator
- VB
CB
-
band gap
+
-
-
-
- VB
semiconductor
Energy
Energy
Energy
CB
CB
-
-
-
-
-
+
+
+
+
+ VB
conductor
Figure 2.1: a) Schematic description of the interaction of discrete atom energy levels and the formation of energy bands. The figure was adapted from Hoffmann et al. [20] b) Available
electrons in insulator, semiconductor, and metal-type materials. With increasing band
gap the amount of electrons excited into the conduction band decreases.
electrons cannot be assigned to a special atom anymore (valence band). The lowest unoccupied
band consisting of anti-bonding states is associated with the conduction band.
For conduction unoccupied states within the valence band or charge carriers within the conduction band are necessary. In metals the conductivity is generally realized by the overlap of
the valence and the conduction band, or by a half-filled valence band (Fig. 2.1). In any case
free electrons can participate in the conduction process. This is not the case in insulators or
semiconductor materials. The valence bands of these materials are completely filled and a gap
appears between conduction and valence band, which is rather small for semiconductors (Eg <
3 eV). Hence, electrons are rigid in the valence band and have to be excited into the conduction
band to participate in the conduction process (Fig. 2.2). Excitation of electrons is possible
by thermal excitation and/or by the absorption of light and negative, mobile charge carriers
are available within the conduction band. The latter is the primary process in photo-related
mechanisms.
The thermally introduced increase in conductivity is best described by the Fermi function or
Fermi-Dirac distribution f(E) and the density of states (DOS) N(E) concepts (eq. 2.1). The
Fermi function gives the probability that a given energy state is completely filled and the electron electron density of states N(E) describes the available energy states. Thus, the population
2.1 Physical and Chemical Properties of Semiconductors
7
Figure 2.2: Characteristics of a) an intrinsic, b) a n-type, and c) a p-type semiconductor. The figure
was adapted from Principles of electronic materials and devices. [23]
of an energy state is given by the product of the N(E) and f(E).
f (E) =
√
1
, N(E)dE = A · EdE
E − EF
1 + exp
kT
(2.1)
EF is known as the energy of the Fermi level and is associated with the highest electron level
at absolute zero temperature, where no electron will have enough energy to pass through. [24]
For intrinsic semiconductors the Fermi level is situated half-way between the valence and the
conduction band.There are several characteristics of f(E) and DOS explaining the behavior of
intrinsic semiconductor material. Inside the semiconductor’s band gap the DOS is zero, so there
are no electrons inside the gap although the Fermi function has a finite value. Furthermore, the
Fermi function at T = 0 K is zero for the conduction band. Hence, there are no free charge
carriers within the conduction band, even though the DOS has a finite value for the conduction
band. The only possibility to increase the number of charge carriers within the conduction band
is to increase the temperature. With higher temperature both functions have a finite value and
8
2 Basic Principles
a conducting population can be assumed and for certain semiconductors excitation of electrons
at ambient temperature is feasible.
A specific modification of intrinsic semiconductors is possible by introducing impurities into
the lattice structure of the material by impurity doping. Due to the impurity atoms new energy
levels are produced, which are mostly inside the band gap near the valence or the conduction
band. Therefore, the energy difference is getting smaller, electron excitation is getting easier
and the material is more conductive at lower temperature. Doped semiconductors are also called
extrinsic semiconductor. Extrinsic semiconductors are divided into two groups depending on
the nature of chemical impurity. Electron donating impurity atoms provide a single-occupied
energy level close to the conduction band. Less energy is required to excite these electrons to
the conduction band, and therefore, negative charges are the majority carriers of electric currents in these n-type doped materials. In n-type semiconductors the Fermi level discussed above
is shifted towards higher energies and it is located between the donor states and the conduction
band. [24] Analogous to n-type doping positive charges, so-called holes can be introduced by
doping with an acceptor atom (p-type doping). Due to one electron deficiency, an empty energy state is created close to the valence band. These states can easily be occupied by valence
band electrons and new positive charge carriers are the majority carriers of electric currents. In
this case the Fermi level is shifted to lower energies and is situated between the valence band
and the new empty states (Fig. 2.2). Among many semiconducting materials titanium dioxide
(TiO2) is one of the best-known semiconductors exhibiting n-type character which is used for
numerous purposes. [18,19]
2.1.2 Optical Properties of Semiconductors
The ability of a semiconductor to absorb UV or visible light is basically determined by the
separation of the valence (VB) and conduction band (CB). [25,26] Upon illumination of a semiconductor with light possessing energy greater than the difference of energy between the VB
and the CB an electron can be excited from the VB to the CB. Based on the relative position of
the valence band apex and the energy valleys of the CB in the energy/momentum k-space the
excitation of an electron can be classified as direct or indirect excitation (Fig. 2.3). For indirect
excitation in an so-called indirect semiconductor an additional momentum is necessary, e.g. a
phonon. The difference of the mechanisms can be illustrated best by having a closer look at
the real band structure in the momentum k-space. Valence bands as well as conduction bands
are described by different trajectories possessing extreme values. The highest probability for
the excitation of an electron is from the VB maximum to CB minimum. In case of an equal
momentum of VB maximum and CB minimum electrons are directly excited in a one-step process. These materials are semiconductors with direct allowed optical transition. Otherwise,
an extra momentum is necessary for electron excitation. In that case minimum and maximum
appear at two different momentums (Fig. 2.3) and excitation occurs in a two-step process. The
2.1 Physical and Chemical Properties of Semiconductors
9
CB
band gap
CB
Energy
Energy
probability of an electron excitation is typically lower for an indirect transition, which can be
measured in terms of the absorption coefficient α .
band gap
VB
VB
Momentum k
Momentum k
Figure 2.3: Scheme of an allowed direct (left) and an allowed indirect (right) optical transition.
The absorption coefficient α is directly accessible by the absorption spectrum of a material as
a consequence of the Lambert-Beer law (eq. 2.2). Based on the Lambert-Beer law α can be
determined by the logarithm of the intensity of the incident light I0 and the transmitted light I
and the thickness of a material d. [27]
I
(2.2)
ln = lnT = −α d = A
I0
The basic principle of these measurements is to describe the attenuation of transmitted light
intensity. Here, the loss of intensity is due to electron excitation. Since the excitation process
is possible in atoms, molecules, and solids (interband transition) information about bulk and
surfaces properties of a catalyst sample as well as of adsorbed species can be obtained (Tab.
2.1). [27,28]
Table 2.1: Wavenumber ranges of different excitation processes.
Bulk and
surface
Wavenumber range
Adsorbed
species
Wavenumber range
Band gap
UV 30,000 50,000 cm-1
UV 30,000 50,000 cm-1
Transition
metal ions
Molecules
with chromophoric
groups
Vibrational
overtone and
combination
bands
NIR, UV-Vis 5,000
- 50,000 cm-1
UV-Vis 14,000 50,000 cm-1
Defects
Transition
metal ions
NIR, UV-Vis 5,000
- 50,000 cm-1
NIR 5,000 14,000 cm-1
However, the absorption spectrum of a semiconductor powder is not easily accessible and diffuse reflectance data are usually recorded. The interpretation of diffuse reflectance spectra is
2 Basic Principles
10
based on the Kubelka and Munk theory, which relates α and the reflection behavior of a powder
material. [27] The Kubelka-Munk function F(R∞ ) is given by eq. 2.3.
(1 − R∞ )2
F(R∞ ) =
2R∞
(2.3)
Assuming a wavelength-independent scattering F(R∞ ) is proportional to α and the band gap
energy of a powder material is accessible by the Tauc equation (eq. 2.4).
α =A
(hν − Eg )n
hν
(2.4)
Here, A is a constant, hν is the energy of light, Eg is the band gap energy, and n (= 0.5, 1.5, 2
and 3) is a factor, which depends on the nature of the optical transition of the material.
2.2 Heterogeneous Photocatalysis on Semiconductor
Particles
The principles of heterogeneous photocatalysis on small semiconductor particles is explained
best by the schematic representation of a small particle as depicted in Fig. 2.4top.
The key steps of the photocatalytic process are 1) the absorption of light, 2) the migration of
electrons and holes to the particle surface, where they are trapped at reactive surface sites and
the subsequently occurring 3) and 4) simultaneous redox reactions. The photo-exited trapped
electron reduces an acceptor species A to a primary product A · - (process 3) and the hole oxidizes a donor species D · + (process 4). Thus, photocatalytic reactions, e.g. overall water
splitting, can only occur if the redox potential of the photo-excited electron is more negative
than the standard redox potential of protons and the redox potential of the photo-generated hole
is more positive than the standard redox potential of H2 O in the oxygen evolution reaction. [26]
In a simplified picture the bottom of the CB and the top of the VB have to enclose the redox
potentials of the specific redox reactions. In Fig. 2.4middle the position of the band edges in
an aqueous solution at pH 1 are shown in relation to the standard redox potentials of several
redox couples and the energy scales of the normal hydrogen electrode (NHE) and the vacuum
level. [11] However, it should be noted that the Fermi level EF of a semiconductor in thermodynamic equilibrium cannot be used to describe the conditions under illumination. Instead, the
electron and hole densities should be described separately by the quasi-Fermi level of electrons
and holes ∗ EFn and ∗ EF p , respectively, as shown in Fig. 2.4top. Therefore, more precisely the
energy difference ∆µ between ∗ EFn and ∗ EF p has to be larger than the redox potential of e.g.
water for overall water splitting.
The efficiency of a photocatalytic process is mainly determined by the recombination rate
of photo-generated electrons and holes. Recombination either occurs directly after photoexcitation (5) or the trapped electrons and holes recombine in a secondary process (6) (Fig.
2.2 Heterogeneous Photocatalysis on Semiconductor Particles
.
A-
Conduction Band
-
*Efn
energy
Ared
(2)
-
(1)
11
(5)
(3)
A
Ered/ox(H2/H2O)
1.23 eV
Ered/ox(O2/H2O)
Δµ (6)
D
+ (4)
*Efp
+
(2)
Valence Band
a) ΔG > 0
Dox
b) ΔG < 0
CB (LUMO)
CB (LUMO)
energy
.
D+
ΔG < 0
ΔG > 0
VB (HOMO)
VB (HOMO)
Figure 2.4: top: Schematic representation of the main processes, which take place at a semiconductor particle during photocatalysis. The figure was partially adapted from Kaiser et
al. [26] and Park et al. [29] middle: Relationship between band structure of a semiconductor and the redox potentials of water splitting. The figure was adapted from Grätzel
et al. [11] bottom: Changes in Gibbs energy in photocatalytic reactions, a) uphill and b)
downhill reaction. The figure was adapted from Ohtani et al. [25]
2 Basic Principles
12
2.4top). Finally, the efficiency of the overall photo-initiated process is also determined by the
back-reaction of the primary reaction products A · - and D · + . [18,20]
Additionally, an important advantage of light-driven reactions at semiconducting materials is
shown in Fig. 2.4bottom, which is the ability to drive reactions even though the change in
Gibbs energy is positive (∆G > 0). [25,30] For most reactions in photocatalysis, like the oxidative
decomposition of organic matter, ∆G < 0, whereas some energy-storing reactions with ∆G >
0, like photocatalytic overall water splitting and photocatalytic CO2 reduction, can be driven
by photocatalysis if the two redox reactions are spatially or chemically separated. [25]
2.2.1 Oxide Photocatalysts - Band gap Engineering
It is well known that the conduction bands of stable oxide photocatalysts, like TiO2 , are comprised of empty orbitals of metal cations with d 0 or d 10 configuration and that the valence
band of these oxides are usually consisting of O 2p orbitals. [30] Due to the position of the CB
edge with respect to the standard redox potential of desired reactions, like proton reduction, it is
mostly desirable to raise the VB in order to decrease the band gap of a specific oxide. Thus, new
energy levels above the VB have to be created by the incorporation of elements which either
form discrete electron donor levels (DL in Fig. 2.5a) within the band gap or which form a new
VB (Fig. 2.5b). [30] For certain oxides the formation of a solid solution is feasible, in which the
VB and the CB are shifted (Fig. 2.5c). [30] In any case the obtained states or new bands should
offer the thermodynamic potential and the kinetic ability to drive the desired redox reaction.
a) Doping
b) New VB
c) Solid solution
CB
CB
WIDE
DL
VB
NARROW
NEW VB
VB
Figure 2.5: a) Donor levels created in the band gap by foreign element doping. b) Creation of a
new VB within the band gap. c) Formation of a solid solution. The figure was adapted
from Kudo et al. [30]
Typical examples for the formation of new DL within the band gap are metal or non-metal
doping of TiO2 or SrTiO3 , like Rh3+ , Cr3+ , or nitrogen doping. The drawback of this strategy is the discrete character of the DL, which is inconvenient for hole migration and which is
2.2 Heterogeneous Photocatalysis on Semiconductor Particles
13
considered to act as a recombination site. [18,25,31,32] New VBs, which ensure the mobility of
holes, can be formed by orbitals of Pb 6s in Pb2+ , Sn 5s in Sn2+ , Ag 4d in Ag+ , and Bi 6s
in Bi3+ , as in BiVO4 . [30] GaN:ZnO is a well-known example of a solid solution containing a
metal oxide. Furthermore, this system is one of the rare examples of successful implementation
of photocatalytic water splitting on nanoparticular photocatalysts by visible light. [33–36] A different strategy to achieve visible light activity of oxide photocatalysts is by sensitization with
organic dyes, metal complexes, and metal nanoparticles. [11,25,37] Sensitization by organic dyes
or metal complexes is used in dye-sensitized solar cells, in which electrons are excited from the
HOMO to the LUMO of the dye and the electron is subsequently injected into the conduction
band of the oxide photocatalyst. [11] Sensitization of a photocatalyst by metal nanoparticles is
extensively discussed in literature. [29,37–40] Mainly Au nanoparticles deposited on TiO2 were
shown to act as sensitizer. While the mechanism of visible light sensitization is not fully understand yet, it is believed that excitation of small Au nanoparticles on the Au plasmon can lead
to an electron injection into the CB of the semiconductor. [29,37–40] Furthermore, trapping of
electrons on metal co-catalysts due to a metal/photocatalyst junction preventing electron-hole
recombination is discussed in the following section.
2.2.2 Metal-modified Photocatalysts
In general, the interaction between metal nanoparticles and a semiconducting photocatalyst can
be explained by the junctions formed in metal/semiconductors. Two different interactions between metal and semiconductor might be discussed, the Ohmic contact and the Schottky junction. [22] While the formation of an Ohmic contact is discussed [25] it is generally excepted that
Schottky junctions are formed on metal-modified photocatalysts, and thus the ohmic contact
will be neglected in the following considerations. [18,29,41,42] Considering an n-type semiconductor and a metal, which are electrically neutral and isolated from each other the Fermi level
of the semiconductor is situated close to the CB. The semiconductor’s work function ϕS is
therefore determined by the energy difference between the vacuum level and the Fermi level
(Fig. 2.6a). The work function of a metal ϕM is normally larger compared to ϕS of a n-type
semiconductor. Hence, the Fermi level of a metal is situated at a more anodic position, and thus
it is more difficult to promote an electron from the metal to the vacuum level.
Upon contacting the metal and the semiconductor metal and semiconductor can be regarded
as being short-circuited with each other. [25] The junction is formed as electrons are transferred
from the semiconductor to the metal and the Fermi level of the metal is shifted upwards (cathodically) until the Fermi levels are aligned (Fig. 2.6b). The flow of electrons from the semiconductor to the metal will cause an accumulation of positive charges at the semiconductor
interface and an accumulation of negative charges on the metal. The semiconductor’s VB is
bent upwards in a Schottky junction. [18]
The effect on the photocatalytic performance of different metal co-catalysts can be thus ex-
2 Basic Principles
14
a)
vacuum
level
φs
b)
vacuum
level
φs
φm
CB
CB
EF
+
-
EF
VB
VB
n-type SC
+
+ +
energy
energy
EF
φm
metal
n-type SC
metal
Figure 2.6: a) Position of the work functions ϕM ϕS for separated metals and n-type semiconductors. b) Formation of a Schottky junction (depletion layer) for metal/n-type semiconductor junctions. The figure was adapted from Yates et al. [43]
plained by the differences in the metal/semiconductor work functions. The larger ϕM the higher
is the Schottky junction. Photo-generated electrons are expected to be more efficiently trapped
on metals with higher work function. Experimental results of Au-, Ag-, and Pt-modified TiO2
showed that H2 evolution due to photocatalytic reforming of alcohols increased in the order Ag
< Au < Pt in agreement with the increase of the work functions of the metals. [41]
2.3 TiO2 in Photocatalysis
Titanium dioxide is the most important titanium compound for technical applications. This
compound is nontoxic, stable against acids and bases and with a melting point of 1850 ◦ C resistant against severe temperature. Usually TiO2 is a white powder, which is used as a pigment
and in cosmetics. [19,44] Catalysis is another important area of application of TiO2 where it is
used either as promoter, support material, additive, or as the catalytically active compound.
The most stable modification is the tetragonal rutile phase. Other possible modifications are the
tetragonal anatase and the orthorhombic brookit structures, which convert into the rutile structure upon heating to temperatures above 580 ◦ C. [46] Of these different modifications the anatase
phase is the most widely used TiO2 polymorph for photo-related processes. All modifications
consist of titanium ions coordinated by six oxygen ions arranged in an octahedral structure.
Therefore, all oxygen ions are involved in three coordination spheres. The different thermody-
2.3 TiO2 in Photocatalysis
15
Figure 2.7: Crystal structure of (a) the rutile and (b) the anatase TiO2 polymorph. The figure was
adapted from Glassford et al. [45]
namic stabilities, different optical and electronic properties, and differences in density of the
modifications can be explained by TiO6 -octahedron distortion, e.g. in anatase phase Ti-Ti distances are bigger and Ti-O distances are smaller (Fig. 2.7). Well established characterization
methods to differentiate between the phases are XRD, Raman, and UV-Vis spectroscopy. [47–49]
Titanium dioxide is either referred to be an insulator or a wide-band gap semiconductor material with an indirect allowed optical transition of around 3.0 eV in the rutile and around 3.2 eV
in the anatase phase. [19] The valence bands of TiO2 are composed of O 2p states hybridized by
Ti 3d, resulting in a band width of around 6 eV. [45,50] The conduction bands are dominated by
unoccupied Ti 3d, 4s, and 4p states. [18] However, these statements and values are only valid for
stoichiometric bulk material. The surface structure mostly differs from the bulk structure. This
is due to the formation of defect sites created by different atmospheres and/or oxygen partial
pressures during synthesis. Common defects are oxygen vacancies and titanium interstitials.
A consequence of oxygen vacancies are electron-donating Ti 3d states (Ti3+ ) states appearing
at 1.18 eV below the conduction band, similar to states introduced by n-type doping. Consequently, the conductivity of TiO2 crystals is mainly determined by defects and increases with
increasing defect density. [19] The number of oxygen vacancies can be increased by reversible
reduction treatments, however, in ambient atmosphere oxygen vacancies are generally compensated by physisorption or dissociative chemisorption of water. [51] High defect densities,
especially at the surface, can also change the optical properties of the TiO2 and depending on
the reduction conditions and the resulting electron states within the band gap colored titanium
suboxides can be obtained. The color varies from green to black for highly reduced samples.
The photocatalytic properties of TiO2 have been reviewed by several articles. [18–20,37] Usually,
the white, defect-free TiO2 is used in photocatalysis and it was shown to be applicable in the
photocatalytic oxidation of organic compounds and hydrogen evolution due to photocatalytic
reforming of alcohols. [18,37] Furthermore, TiO2 is used as electron collector in dye-sensitized
solar cells and inorganic/organic hybrids for photoelectrochemical water splitting. [52–54] The
main drawback of TiO2 is the inherent large band gap. Thus, only small fractions of the solar
light spectrum can be utilized by TiO2 . Therefore, metal and non-metal doping as well as
16
2 Basic Principles
sensitization by metal nanoparticles are currently discussed to improve the light absorption
properties of TiO2 . [18,37] Within these efforts non-metal doping, especially nitrogen doping,
was proven to enhance the visible light activity. However, the beneficial activity under visible
light operation is mostly associated with a decrease of the overall activity. [55] More recently,
blue or black TiO2 was shown to exhibit significant activity in visible light-driven reactions
although defects are generally expected to decrease the catalytic activity. [56,57]
2.4 Single-site Photocatalysts
So far photocatalysis has exclusively been related to semiconductor materials and the redox
reactions performed at the surfaces of the semiconductor particles. One of the first examples
of the photocatalytic activity of isolated metal oxide species was reported by Anpo et al. [58]
They showed that the photoreduction of CO2 on highly dispersed or even isolated metal oxide
species, in this case TiO4 species, was feasible. [58] Since then isolated metal oxides received
some interest in heterogeneous catalysis, as isolated metal oxides offer a generally applicable
strategy for the design of new heterogeneous catalysts. [59,60] More recently, the interest in
single-site photocatalysts is increasing and successful examples of molybdenum, copper, and
chromium single-site photocatalysts were reported. [61,62]
Figure 2.8: From TiO2 semiconductors to Ti-oxide single-site catalysts. Illustration of: size quantization effect, small particle effect, and change in coordination number. The figure was
adapted from Anpo et al. [61]
2.4 Single-site Photocatalysts
17
While for semiconductor particles the excitation of electrons occurs from the VB to the CB the
creation of excited electronic states proceeds differently in the case of isolated surface species.
As shown in Fig. 2.8 and explained previously for the energy band model electronic states
in single atoms are characterized by discrete levels. Therefore, with decreasing TiOx domain
sizes the overlap between bonding, antibonding, and non-bonding states are reduced and they
appear to be more discrete.
The excitation process in an isolated metal oxide species is a Ligand to Metal Charge Transfer
(LMCT) and corresponds to an electron being excited from the VB to the CB leaving a hole
(h+ ) and an electron in the conduction band (e- ) in a semiconductor particle (Fig. 2.8). Ultimately, the excitation of an electron, whether through the formation of excitons or an LMCT, is
the result of the absorption of radiation. The required energy is in the case of isolated species
the distance between Highest Occupied Molecular Orbital (HOMO) and Lowest Unoccupied
Molecular Orbital (LUMO). Accompanied with the mentioned changes an increase in the absorption can be observed due to a lowering of the VB edge and the rise of the CB edge, which is
also known as the quantum size effect. Therefore, UV-vis spectroscopy is frequently employed
to determine if a metal oxide is present as an isolated species. [63]
Table 2.2: Correlation of UV-Vis absorption and degree of isolation of TiOx species.
Ti [wt%]
LMCT band
wavenumber [cm-1 ]
LMCT band wavelength [nm]
isolated
dimeric/one
dimensional
polymeric/two
dimensional
1.05
47,600
6.6
40,600
15
39,000
210
246
256
Different supports can be used to synthesize isolated metal oxides, however, mesoporous silica is commonly used due to the high specific surface area. For isolated, dimeric, and twodimensional polymerized TiOx species on SiO2 the absorption maxima determined by UV-Vis
spectroscopy are summarized in Table 2.2. For highly isolated TiOx species the LMCT enabling photocatalytic reactions is believed to be due to a charge transfer in the tetrahedral
Ti(OH)(OSi)3 species as described in eq. 2.5.
− Ti3+ + O−
Ti4+ + O2− →
hν
(2.5)
The TiOx loadings listed in Table 2.2 are standard values and might change with the support or
the preparation technique. [64,65]
3 Literature Review - Photocatalytic
CO2 Reduction on TiO2-related
Materials
Several molecular approaches to photocatalytic reduction CO2 were recently reviewed by Morris et al. [66] and more specifically the photocatalytic CO2 reduction on TiO2 -related materials
was reviewed by Dhakshinamoorthy et al. [14] Most of the reviewed work within these two articles performed photocatalytic CO2 reduction in the liquid phase, which cannot easily be linked
to gas-phase reactions. Therefore, this chapter will mainly focus on the studies performed with
mesoporous silicates containing isolated/polymerized titania sites usually performed in the gasphase, and the recent results on the mechanistic aspects of photocatalytic CO2 reduction will
be discussed, which are normally discussed for bulk TiO2 materials.
Even though photocatalytic CO2 reduction was first published in 1979 by Inoue et al. [5] , the
first report of a successful reduction of CO2 at isolated or highly dispersed titania under UV irradiation was published by Anpo and coworkers in 1992. [58] In contrast to bulk-like TiO2 where
electrons are excited form the valence band to the conduction band of the material, the excitation in isolated titania can be described by a ligand-to-metal charge transfer (LMCT) from the
oxygen anion to the titanium cation. In this pioneering work isolated or highly dispersed titania
was obtained by the reaction of TiCl4 with the OH-terminated surface of a porous glass yielding an active anchored Ti-photocatalyst for the formation of mainly methane in the presence
of gas-phase CO2 and H2 O. Furthermore, minor contributions of methanol, ethane, ethylene,
carbon monoxide, and dioxygen were detected. [58] The product distribution was confirmed to
strongly depend on the H2 O/CO2 ratio in the feed gas resulting in the highest product yields
for a H2 O/CO2 ratio of 5. [58] Continuing their research Anpo et al. [67] were able to confirm
the presence of isolated titania in the active anchored Ti-photocatalyst by means of X-ray absorption spectroscopy (XAS). They proposed a mechanism, in which the formation of methane
occurred by the photoreduction of CO2 into bare carbon radicals, which subsequently react with
hydrogen atoms produced by the reduction of protons supplied from adsorbed water. [58,67] The
radical mechanism was mainly based on in-situ electron paramagnetic resonance spectroscopy
(EPR) studies evidencing the formation of the radicals upon illumination of the photocatalsyt.
Thorough investigations of mesoporous materials containing a variety of different titania structures, from isolated tetrahedrally coordinated Ti-sites to octahedral coordinated Ti-sites with
20
3 Literature Review - Photocatalytic CO2 Reduction on TiO2 -related Materials
bulk TiO2 characteristics, revealed that the tetrahedrally coordinated Ti-sites show a higher selectivity to methanol compared with the octahedral Ti-sites. However, methane was always
the main product independent of the Ti-coordination. [68] The comparison of the obtained total rates of CH4 and CH3 OH formation indicated that the formation of methanol depends on
the Ti-coordination in the mesoporous photocatalyst and a high selectivity towards methanol
formation was obtained for the catalysts containing mostly tetrahedrally coordinated Ti-sites.
Based on these results Anpo et al. [68] assumed that the spatial separation of photo-excited electrons and holes is of major importance for the product distribution. In consideration of their
proposed radical mechanism the formation of methanol on isolated tetrahedrally coordinated
Ti-sites was explained by the following reactions:
• CO2 photoreduction to carbon radicals occurs at the isolated Ti-site,
• H2 O photoreduction to hydrogen atoms occurs at the isolated Ti-site,
• simultaneously OH anions can react with trapped holes to form OH radicals at the isolated Ti-site.
Thus, they proposed that the OH radicals can easily react with the carbon radicals formed
by photoreduction. Additionally, hydrogen atoms and carbon radicals can recombine to form
methane. On the other hand, the reaction between OH and carbon radicals can be neglected
for samples with bulk TiO2 -like structure as in the samples containing octahedral Ti-sites. The
photo-excited electrons and holes are rapidly separated in the bulk-like TiO2 samples and photooxidation of the OH anion to an OH radicals occurs at a different active site, and therefore
according to Anpo et al. [68] , the spatial separation of the active sites determines the product
distribution. It was also considered beneficial that the lifetime of photo-excited LMCT states,
as in isolated titania, is significantly higher and the electron and hole transfer to CO2 and H2 O
is favored.
CO2 + 2H+ + 2e−
CB → HCOOH
+
Eredox
/NHE = −0.61V
O
HCOOH + 2H
→ HCOH + 2H2 O Eredox
/NHE = −0.48V
O
redox
HCOH + 2H + 2e−
/NHE = −0.38V
CB → CH3 OH EO
redox
CH3 OH + 2H+ + 2e−
/NHE = −0.24V
CB → CH4 + H2 O EO
+
Eredox
/NHE = 2.32V
H2 O + h +
O
CB → OH + H
+
2H2 O + 2h+
Eredox
/NHE = 1.35V
O
CB → H2 O2 + 2H
+
2H2 O + 4h+
Eredox
/NHE = 0.82V
O
CB → O2 + 4H
+ 2e−
CB
+
(3.1)
(3.2)
(3.3)
(3.4)
(3.5)
(3.6)
(3.7)
The potentials of all reactions were obtained at pH 7.
Here, it should be noted that Inoue et al. [5] already proposed a relation between the conduction band edge position of a semiconductor and the methanol yield. The yields of methanol
significantly increase as the reduction potential becomes more negative, which is in agreement
with the redox potentials shown below in eq. 3.1-3.7. [69] Certainly, the reduction potential of
21
photo-excited electrons is more negative for isolated titania species, and therefore the increase
in methanol yield might also be rationalized with the higher reduction potential of electrons.
More recently, Anpo’s group reported that higher methane yields can be obtained with hydrophillic samples and that more open pore structures are beneficial for the photocatalytic CO2
reduction. [21,70] Additionally, the selectivity towards the formation of methane was reported to
be promoted upon incorporation of noble metal co-catalysts, e.g. platinum or palladium. [21,68]
Inspired by the pioneering work of Anpo et al. [58,67] mesoporous materials containing isolated
titania sites were also studied by Frei et al. [71,72] and Mul et al. [73] In-situ infrared spectroscopy
was performed by Frei et al. [71] using either H2 or methanol as sacrificial electron donor. Only
in the presence of sacrificial methanol formic acid formation associated with the formation of
CO was observed, whereas the only product for the reduction of CO2 was CO. [71] In presence of water no hydrogen atom able to reduce activated CO2 is formed, and instead two OH
radicals recombine to hydrogen peroxide and subsequently oxygen is released. Mul et al. [73]
followed the mechanism proposed by Frei et al. [71] However, as mainly methane and longer
chain hydrocarbons were observed by Mul et al. [73] the reaction mechanism was extended and
formaldehyde was identified as a possible reaction intermediate, which can easily be converted
to methane. Similar to the studies by Anpo et al. [58,67] and Frei et al. [71,72] all reactions proceed at one Ti-center. Recently, Stair et al. [74] doubted that CO2 sufficiently interacts with
isolated titania. They unambiguously showed that isolated titania is interacting with water by
means of UV-Raman spectroscopy, but the interaction with CO2 was not confirmed. Anpo et
el. [21,68] tried to confirm the interaction by photoluminescence spectroscopy. A quenching of
the characteristic photoluminescence band was observed, which was attributed to a transfer
of the photo-excited electron from the Ti-center to CO2 . [21,68] Theoretical studies by Kanai et
al. [75] indicated that there is a significantly reduced energy barrier for the first reduction of CO2
at isolated titania, which might be attributed to a physical or chemical interaction. However,
the the photoluminescence studies performed by Anpo et al. [21,68] are no direct evidence for
the interaction of isolated titania and CO2 as the quenching of a photoluminescence band might
be the result of different optical properties of the materials in presence of CO2 . Furthermore,
the studies were performed at 77 K and it is reasonable to assume that CO2 is in the solid state.
Thus, until know, the interaction of CO2 with isolated titania is still not fully understood, and
it is reasonable to assume that an improved interaction will lead to strong improvements in the
efficiencies. [73]
Even though there is a discrepancy in the observed product distributions, the CO2 interaction,
and the suggested reaction mechanisms, a detailed knowledge of the different steps involved
in the overall photocatalytic CO2 processes is of great interest. Mechanistic studies of TiO2
surfaces might provide further insights. The interaction of CO2 and TiO2 is well known and
the photocatalytic properties of TiO2 are better understood. [18–20,37] It certainly should be determined whether or not the reaction proceeding at isolated titania is different from the reaction
mechanism on bulk-like TiO2 structures [14] as Anpo et al. [21] showed that the reaction on iso-
22
3 Literature Review - Photocatalytic CO2 Reduction on TiO2 -related Materials
lated titania is drastically enhanced and lower energy barriers were observed by Kanai et al. [75]
facilitating the formation of CO. However, the discussed mechanisms for CO2 reduction on isolated titania should be compared to in-situ spectroscopic studies and theoretical calculations on
TiO2 surfaces, which may not provide complementary information to the mechanistic studies
on isolated titania. [14]
The recent experimental studies and theoretical calculations by Zapol et al. [69,76] and Rajah
et al. [77] should be mentioned: The role of water and carbonates in the formation of methane
was investigated by Zapol et al. [69] The EPR results showed the formation of hydrogen atoms,
hydroxyl radicals, and CO−
3 and it was specified that water is involved in the formation of
OH radicals, competes with CO2 as electron acceptor and efficiently stabilizes charges diminishing electron-hole recombination. [69] The influence of possible reaction intermediates like
methanol, formaldehyde, and formic acid in the multistep CO2 reduction process to methane
(eq. 3.1-3.7) on TiO2 was studied by Rajah et al. [77] also by means of EPR. They concluded that
formic acid can be a reaction intermediate, which is subsequently reduced to methane, whereas
methanol and formaldehyde are mainly acting as hole scavengers and tend to photooxidize. [77]
Within this study the simplest reaction scheme for CO2 reduction to methane was considered
consisting of four consecutive two-electron proton-coupled reactions (eq. 3.1-3.4). All reaction
intermediates in this sequence are one carbon atom molecules. Recently, Zapol et al. [76] considered a reaction mechanism in which reaction intermediates with two carbon atom molecules
are involved. Interestingly, the molecules are formed spontaneously upon photooxidation of
methanol and formaldehyde and are therefore complementary to the studies by Rajah et al. [77]
It should be emphasized that mainly EPR was used, which is exclusively able to detect radical
species. This may not give a complete picture of all the reaction pathways and intermediates
involved. [14]
Finally, it should be noted that performing photocatalytic CO2 reduction is rather challenging
and thorough blind experiments have to be performed as carbon contaminations on the catalyst
surface significantly influence the product yields [73] and photolysis of CO2 to methane can take
place at 185 nm irradiation regardless of the presence of any solid. [78]
4 Development of a Gas-phase
Photoreactor
CO2
V3
N2
V5
He
TC
Sat
V2
cooling
GC
P detection
V4
V7
V6
V1
QMS
Figure 4.1: Flow scheme of the photoreactor set-up including the gas supply and the analysis devices.
A gas-phase photoreactor on the basis of a set-up design recently presented by Grimes et
al. [17,79] was developed. A schematic drawing of the photoreactor is shown in Fig. 4.1. Pictures
of the set-up are shown in Fig A1. The set-up mainly consists of a gas supply, a saturator, a
reactor, analysis devices (Gas Chromatograph and Quadrupol Mass Spectrometer), and a pump
system. Thus, the fully metal sealed home-made set-up has a total volume of 27 ml, which can
be pumped down to a pressure of 10-8 mbar. While the temperature of the reactor and the saturator are maintained by water cooling, the gas lines can be heated to avoid water condensation.
Reactions can be performed either in continuous or in batch mode. For this purpose product
gases can be analyzed by a Quadrupol Mass Spectrometer (QMS, Balzer, Omnistar) in continuous mode and a Gas Chromatograph (GC, Shimadzu 14B) in batch mode. Trace gas analysis
is performed with the GC, which is equipped with Thermal Conductivity Detector (TCD) and
Flame Ionization Detector (FID) detectors. Two Porapak N columns and one molecular sieve
column are used for separation. The photocatalyst is spread out on the bottom of the reactor to
obtain a homogeneous light illumination. A 200 W Hg/Xe (Newport Oriel) lamp is used to irra-
4 Development of a Gas-phase Photoreactor
24
diate the samples through a quartz window allowing for high transmissions of high energy UV
light. The lamp is equipped with a water filter to avoid heating of the system by infrared (IR) irradiation. Thus, the light irradiance at the sample surface was measured to be 200 mW/cm2 and
110 mW/cm2 using a 320 nm cut-off filter. The gas supply consists of three different gas lines
connected to helium, nitrogen, and carbon dioxide, which can be used as feed gases. Helium
is mainly used to vary the carbon dioxide concentration in the feed gas. However, in case of
water splitting experiments the helium feed is saturated with water. Nitrogen instead is mainly
used as an internal standard during batch mode operation to adjust for the pressure drop during
sampling.
b) Event 92 Molecular sieve online c) Event -92 Molecular sieve bypass
Porapak
column
Porapak
column
Molecular sieve
Porapak
column
Porapak
column
Molecular sieve
Figure 4.2: a) GC program containing the column- and the temperature program, the separation of
gas molecules according to time is indicated. Flow sheet of the column arrangement in
their switching events, to either set the molecular sieve column online (b) (event 92) or
bypass (c) (event -92).
While the QMS, as mentioned, is mainly used for gas-phase water splitting experiments in a
continuous mode, the GC was adjusted allowing for trace gas analysis of short-chain hydrocarbons. Trace gas analysis was achieved by the GC equipped with two Porapak N columns
25
Table 4.1: Standard GC settings. Temperature and current of the TCD are chosen to avoid water
condensation.
detector
temperature
current
FID
TCD
200 ◦ C
150 ◦ C
50 mA
carrier gas flow behind the columns
adjusted carrier gas pressure
adjusted reference gas pressure
19.2 ml/min
500 kPa
165 kPa
and a molecular sieve column. Furthermore, a Thermal Conductivity Detector (TCD) and a
Flame Ionization Detector (FID) connected in series at the column outlet are used. The standard settings of the GC are summarized in Table 4.1. With this arrangement of columns and
detectors the analysis of the different compounds in the reaction product stream is feasible. The
product gas stream normally containing nitrogen (N2 ), helium (He), carbon dioxide (CO2 ), and
water (H2 O), as well as the desired products, such as C1 to C4 hydrocarbons and alcohols can
be separated using a temperature program and column switching procedure as shown in Fig.
4.2. In a first step of the analysis programm the gases, which are not interacting with the Porapak N column, are flushed straight to the molecular sieve column. Afterwards, the molecular
sieve column is set bypass and the molecules are stored there, while mainly H2 O, CO2 , H2 ,
and C2 -C4 are separated by the Porapak N column. Reasonable retention times of the gases
are achieved by heating the Porapak N column. After cooling the system the molecular sieve
column is switched online and the retaining molecules are analyzed. While small or incombustible gases like CO2 , N2 , H2 O, H2 , O2 , and CO are detected by the TCD, all gases being
easily flammable (C1 -C5 and small alcohols) are detected by an FID.
Table 4.2: Retention times of different molecules using the specified sampling settings. Assignment of the molecules to the separation column and the appropriate detector.
molecule
retention time
column
detector
H2 O
N2
CO2
CH4
C2 H4
C2 H6
C3 H8
C4 H10
C4 H12
CH3 OH
10
30.2 min
2.8 min
31.9 min
3.4 min
3.7 min
7.8 min
11.9 min
17.4 min
11.6 min
PP
MS
PP
MS
PP
PP
PP
PP
PP
PP
TCD
TCD
TCD
FID
FID
FID
FID
FID
FID
FID
4 Development of a Gas-phase Photoreactor
26
160
160
a)
140
C1
120
peak area [a.u.]
peak area [a.u.]
140
C2
C3
100
C4
C5
80
60
40
20
0
0
10
b)
20
30
40
50
60
70
80
90
8 ppm
18 ppm
120
18 ppm (2)
22 ppm
100
29 ppm
43 ppm
80
81 ppm
60
40
20
0
theoretical fraction of alkane
C1
C2
C3
C4
C5
number of carbon atoms
Figure 4.3: a) Plot of peak area derived by the integration of the peak area of the respective FID
signal vs. the theoretical fraction of the C1-C5 alkanes in the feed gas. b) GC calibration with C1-C5 hydrocarbons, peak area vs. the number of carbon atoms in the
respected hydrocarbon.
Calibration of the GC for quantitative product analysis
To study the detection of small substance concentrations with the gas chromatograph, a dilution
series was applied. Therefore, a gas mixture containing 1000 ppm of C1 -C5 hydrocarbons
and 10 % N2 balanced with He was used and mixed with high purity He (6.0). Thus, C1 -C5
hydrocarbon concentrations of 9 ppm were achieved.
The integrated peak areas plotted against the calculated fraction in ppm of different hydrocarbons are shown in Fig. 4.3a. Additionally, the peak areas plotted against the number of carbon
atoms are shown in Fig. 4.3b. It is evident from these plots that the signal area increases for
each compound with increasing number of carbon atoms in a linear manner, demonstrating the
high quality of the calibration procedure. Furthermore, it can be seen that the peak area also
increases linearly with the overall concentration of the detected compound. The deviations
from the linear graph shape is caused by an overall increasing gas flow in the dilution series
necessary to achieve low C1 -C5 hydrocarbon concentrations, and thus an increase in the reactor
pressure. However, as mainly low C1 -C5 hydrocarbon concentrations are expected a linear fit
was chosen in agreement with the low concentration region (Fig. 4.3a). Besides C1 -C5 hydrocarbons, methanol was calibrated using a 3000 ppm MeOH/He gas mixture. Concentrations of
25 ppm MeOH were achieved using the dilution procedure with He, which was easily detected
by the GC. By repeating the calibration procedure for different hydrocarbon concentrations an
uncertainty of the calibration of 5% was estimated. The retention times of the most frequently
detected molecules using the explained sampling programm are summarized in Table 4.2.
Photocatalytic CO2 reduction experiments
Even though the reactor was mainly developed for gas-phase photocatalytic CO2 reduction it
is also possible to perform gas-phase photocatalytic water splitting experiments in continuous
27
mode using the QMS as mentioned above. Photocatalytic CO2 reduction experiments are normally performed in batch mode due to the low product yields. Due to the low product formation
yields a pretreatment of the photocatalyst by irradiating the material in humid He is necessary
to confirm that possible carbon contaminations on the catalyst surface are not involved in the
product formation. [73] The actual photocatalytic CO2 reaction is conducted by introducing a
water-saturated reactant feed containing x% CO2 /1% N2 balanced by He into the photoreactor. The water content in the reactant feed is mainly determined by the temperature of the two
stainless steel saturators, which can be easily adjusted between 0 ◦ C and 25 ◦ C using a cryostat.
Thus, the water content can be varied between 6042 ppm and 3.14 %. Prior to the actual CO2
reduction experiment the reactor is pumped down to 5 x 10-3 mbar by means of a rotary pump
or a turbomolecular pump if lower pressures are required. In this way a base pressure of 5 x 10-8
mbar can be achieved. Afterwards, the reactant feed is introduced into the reactor until atmospheric pressure is obtained. A fully water-saturated gas-phase can be achieved by flowing the
reactant feed through the reactor for a certain period of time or by repeating pumping/dosing
cycles. Afterwards, the reactor pressure is adjusted to a desired value and the reactor is closed.
Thus, the reaction is performed in batch mode by illumination the sample for 7 h. Sampling can
be performed at given time intervals, which is mainly limited by the sampling time of the GC
programm. With the described settings sampling takes 50 min in total. Each sampling causes
a pressure drop in the reactor of 100 mbar. A quantitative analysis of the products is achieved
using 1% N2 as internal standard.
5 Photocatalytic CO2 Reduction Single-site TiOx Materials
In this chapter Ti-single site catalysts for the photocatalytic reduction of CO2 are presented. In
the first section synthesis and characterization of the single-site photocatalysts are described
and the incorporation of additional CO2 by means of isolated ZnOx species is introduced. The
second section is dealing with the photocatalytic CO2 reduction on isolated TiOx species and
the effect of photo-deposited Au nanoparticles is discussed. Finally, the photocatalytic activity
of the as-synthesized and Au-modified materials with respect to TiOx loading and the state of
ZnOx species is discussed.
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
30
5.1 Modification of titanate-loaded mesoporous silica by
grafting of zinc oxide
Abstract1
Mesoporous silica (SBA-15) loaded with TiOx species was synthesized by anhydrous grafting
of titanium isopropoxide, and a novel procedure for the preparation of ZnOx /SBA-15 materials
by grafting of Zn(acac)2 was explored. The TiOx /SBA-15 and ZnOx /SBA-15 materials as well
as subsequently prepared bifunctional ZnOx - and TiOx -containing SBA-15 materials were characterized in depth by combining N2 physisorption measurements, UV-Vis, X-ray photoelectron
and X-ray absorption spectroscopy, and CO2 and NH3 temperature-programmed desorption
experiments. The characterization results confirmed a close proximity of ZnOx and TiOx in
the subsequently grafted materials. Due to strong interactions between the Zn precursor and
the SiO2 surface the order of the ZnOx and TiOx grafting steps affected the amount of Ti-O-Zn
bonds formed in the materials. When ZnOx is present in SBA-15, subsequently grafted TiOx
is higher coordinated and more Ti-O-Zn bonds are formed compared to SBA-15 in which TiOx
was introduced first indicating strong interactions between the Ti precursor and ZnOx . While
all TiOx and ZnOx -containing samples exhibit a large amount of acidic sites, ZnOx present as
isolated species or small clusters in SBA-15 significantly improves the CO2 adsorption capacity
by introducing basic sites. In the subsequently grafted samples the amount of acidic and basic
sites is found to be unaffected by the order in which the two transition metals are introduced.
1
The main content of this section was published as ”Tuning the Acid/Base and Structural Properties of TitanateLoaded Mesoporous Silica by Grafting of Zinc Oxide”, B. Mei, A. Becerikli, A. Pougin, D. Heeskens, I. Sinev,
W. Grünert, M. Muhler, and J. Strunk J. Phys. Chem. C, 2012, 116, 1431814327, DOI: 10.1021/jp301908c.
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
31
5.1.1 Short Introduction
In recent years single-site catalysts have been widely studied, and especially TiO2 dispersed in
mesoporous SiO2 has attracted much attention as catalyst or catalyst support in various fields
such as photocatalysis or acid catalysis. [65,80,81] Due to its different coordination, the reactivity
and selectivity of TiOx species in mesoporous materials considerably differs from that of bulk
TiO2 . Differences in the activity and selectivity were observed in liquid-phase and vapor-phase
epoxidation reactions, [82–84] or in the gas-phase oxidation of methanol. [80,85] In these studies
different preparation routes have been developed to achieve a maximum amount of isolated
TiOx sites. While the incorporation of the tetrahedrally coordinated TiO4 species into the silica
walls has the highest probability of yielding exclusively isolated sites, the maximum amount
that can be incorporated is very low, [73] and a considerable fraction of the sites may not even be
accessible within the wall structure. Impregnation or chemical vapor deposition, on the other
hand, make it possible to achieve a much higher loading of the transition metal on the silica
surface, but the probability of the Ti species being isolated decreases significantly. [86,87] In this
respect, anhydrous grafting is a powerful method: the selective reaction of the precursor with
the silanol groups makes it possible to achieve rather high loadings of relatively isolated species
on the silica walls. [60,65,88] Furthermore, it was shown that anhydrous grafting is an effective
method to control both the amount and the degree of agglomeration of TiOx species. [65]
Different degrees of agglomeration of the titania component on the silica surface influence
their behavior. For example, previous work on TiOx grafted onto MCM-48 has shown that
the reducibility of the titania component is a function of its degree of agglomeration. [89] It
is reasonable to assume that the degree of agglomeration, the mobility/delocalization of the
electrons or the reactivity of the lattice oxygen may influence the product spectrum in many
(photo)catalytic reactions.
In their pioneering work, Anpo et al. [58,67,68,90] demonstrated that isolated tetrahedral TiOx
species on silica are active photocatalysts for the formation of methane and methanol from
CO2 . It was suggested that both CO2 and H2 O adsorb on the tetrahedral TiO4 species, and the
mechanism was postulated to involve hydrogen and methyl radicals, which were both detected
by electron paramagnetic resonance (EPR) spectroscopy. [67]
There is currently a dispute about the adsorption of the reactants CO2 and H2 O at the tetrahedral
titania site. Anpo’s group observed that the addition of either of the two reactants decreased the
photoluminescence emission originating from the electron-hole recombination at the tetrahedral Ti site. [67] It was inferred that the recombination does not take place, because the electron
is transferred to the adsorbate. A study by Danon et al. [74] monitored the UV resonance Raman
vibrational band of the tetrahedral TiO4 species upon adsorption of CO2 or H2 O confirming the
adsorption of H2 O at this site, but there was no indication for the adsorption of CO2 . Neither the
studies by Anpo et al. [67] nor that by Danon et al. [74] proved the adsorption of CO2 on isolated
TiOx species by means of a direct or quantitative technique such as temperature-programmed
32
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
desorption (TPD), thermogravimetry, or microcalorimetry. The adsorption of CO2 usually requires the presence of basic sites, and CO2 is a frequently used probe molecule for the detection
of basic sites. [91] The acidic properties of grafted TiO2 /SiO2 materials have been characterized
by Iengo et al. [92] by means of TPD of pyridine. In a study of amine-functionalized TiOx/SBA15 Srivastava et al. [93] detected a small amount of CO2 adsorbed on amine-free TiOx /SBA-15.
However, the amount did not differ from the bare SBA-15 and no activated CO2 was found by
infrared spectroscopy. Only the amine-functionalized TiOx/SBA-15 materials were shown to
effectively activate CO2 in chloropropene and styrene carbonate synthesis. [93]
Magnesium- and alumina-modified SBA-15, as well as amine-functionalized TiOx /SBA-15,
have been reported to possess acidic and basic sites, however, no transition metal-modified
TiOx/SBA-15 material providing both acidic and basic sites has been reported yet. [93–95] Zinc
oxide is known for the amphoteric nature of its surface, adsorbing CO2 on a variety of basic
sites. [91,96,97] Apart from the weakly adsorbed linear species, a variety of carbonates and bicarbonates can be observed. [98] ZnO is a photocatalyst with a similar band gap as TiO2. Moreover, it is a common active ingredient of methanol synthesis catalysts such as ZnO/Cr2 O3 and
Cu/ZnO/Al2O3 . [99] With respect to the photocatalytic CO2 reduction it appears to be favorable
to combine TiOx with ZnO or ZnOx species in a close interfacial contact in order to make use of
the water splitting properties of titania and the CO2 sorption properties of ZnO. Similar to the
thermal catalytic processes, the adsorption of the reactants can be one of the rate-limiting steps
in photocatalytic reactions. It has also been suggested that due to the generally fast electronhole recombination kinetics, an interfacial electron transfer seems only kinetically competitive,
when the reactant is already preadsorbed. [74,100] Therefore, it is desirable not only to characterize, but also to increase the amount of surface-bound CO2 and H2 O. [93] While the synthesis of
ZnO inside micro- or mesoporous silica has previously been reported, the acid/base properties
of the obtained species are barely characterized. [101–104] Furthermore, a reproducible grafting
technique for small, well-defined ZnOx domains in silica using Zn(acac)2 has not been reported
yet.
In this contribution, titania and zinc oxide species in SBA-15 are synthesized on the silica surface by means of anhydrous grafting of chloride-free precursors. A thorough characterization
with different spectroscopic techniques revealed that high loadings of isolated or slightly agglomerated TiOx species can be obtained. With respect to ZnOx , it is shown that grafting of Zn
acetylacetonate results in the formation of isolated ZnOx species. When both ZnOx and TiOx
are present on SBA-15, there is an interaction between these species. The characterization results indicate that the structure of titania depends on the order in which the transition metals
are introduced into SBA-15.
Furthermore, it is shown that both TiOx and ZnOx in SBA-15 act as acidic sites. In addition,
the presence of ZnOx enhances the sorption capacities of CO2 , as it introduces basic sites. In
spite of the differences that were observed in the coordination of titania, the order in which
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
33
the transition metals are grafted onto SBA-15 does not influence the total amount of acidic and
basic sites.
Experimental
Sample synthesis
The SBA-15 support was synthesized according to well-established procedures. [105] 4 g of
Pluronic P123 (Sigma Aldrich) were dissolved in 30 ml of deionized water and stirred for 4 h
to ensure good mixing. After adding 120 ml of 2 M HCl, stirring was continued for 1 h at room
temperature and for 1 h at 313 K. At this temperature, 9 g (8.4 ml) of tetraethylorthosilicate
(TEOS) were added dropwise under vigorous stirring. The mixture was kept at 313 K for 24
h and afterwards at 373 K for 48 h. The reaction mixture was cooled to room temperature,
filtered, washed with deionized water and dried for 2 h. Calcination was performed in synthetic air ramping the temperature at 1 K min-1 from room temperature to 623 K, holding the
temperature for 1 h, then ramping to 823 K at 1 K min-1 with a holding step of 6 h.
The grafting procedure has been reported previously for the deposition of TiOx on MCM48. [89,106] It is completely performed under inert atmosphere to avoid the presence of H2 O. 2 g
of the support material (here SBA-15) were dried over night at 393 K. The precursor Ti(Oi Pr)4
(99.999 %, Sigma Aldrich, stored in a glove box) was dissolved in about 50 ml of dry toluene.
The amount was chosen according to the desired surface coverage, but it was never higher
than an equivalent of 1.5 Ti nm-2 . The precursor solution was brought in contact with the dry
support at room temperature and stirred for at least 4 h. The support was separated from the
solution either by sedimentation or centrifugation (6000 min-1 for 3 to 5 min). The support was
washed three times with dry toluene (about 30 ml), separating sample and solution each time
as mentioned above. It was then dried under dynamic vacuum.
Calcination was performed by heating the sample with 2 K min-1 in a flow of 100 ml min-1 N2
to 573 K holding the temperature for 1 h. During the time at 573 K, the gas flow was switched
to synthetic air, before the temperature was ramped at 2 K min-1 to 773 K and kept for 4 h. The
samples were stored in a desiccator prior to use.
The grafting of ZnOx species was performed using zinc acetylacetonate (Zn(acac)2 ) as precursor (Alfa Aesar). Due to the undefined purity and high water content of the precursor, it was
purified by means of sublimation. All steps of the grafting were similar except for the solvent,
which was changed from toluene to dry tetrahydrofurane (THF). The temperature during the final calcination was lowered, so that the sample was heated in N2 to 523 K only, and in synthetic
air only up to 673 K.
Characterization
The materials were investigated by means of elemental analysis, nitrogen physisorption, UVVis spectroscopy, UHV-Fourier-Transformed Infrared spectroscopy(UHV-FTIR), X-ray photoelectron spectroscopy (XPS), X-ray absorption spectroscopy (XAS), and temperature pro-
34
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
grammed desorption (TPD). Elemental analysis was performed by inductively coupled plasma
optical emission spectroscopy (ICP-OES) to determine the amount of titanium in the samples,
while atomic absorption spectroscopy (AAS) was used to determine the Zn content. ICP-OES
measurements were performed with a PU701 instrument supplied by UNICAM. A SpectrAA
220 instrument (Varian) was used for AAS measurements. Static N2 physisorption experiments
were performed at the boiling point of liquid N2 subsequent to out-gassing at 200 ◦ C for 2 h
in a Bellsorp max instrument from Belcat, Inc. The surface areas of all samples were estimated according to the method by Brunauer, Emmett and Teller (BET), whereas the pore size
distribution was obtained applying the method by Barrett, Joyner and Halenda (BJH). [107,108]
Micropore volumina were calculated using the t-method. [109] UV-Vis diffuse reflectance spectra (DRS) were recorded in a Perkin Elmer Lambda 650 UV-Vis spectrometer equipped with a
Praying-Mantis mirror construction using MgO as the 100 % reflection reference. The samples
were dehydrated in flowing syn. air in an in-situ UV-Vis cell (Harrick) at 400 ◦ C for 1 h prior
to the measurement. As TiOx /SiO2 is not a semiconductor, nor a structurally homogeneous
material, the Tauc-Plot cannot be used to determine the absorption on sets. Instead, absorption
onsets were estimated from a linear fit to the incline of the Kubelka-Munk function.
XPS was performed in a UHV set-up equipped with a Gammadata-Scienta SES 2002 analyzer.
The base pressure in the measurement chamber was 5 x 10−10 mbar. Monochromatic Al Kα
(1486.6 eV; 13.5 kV; 37 mA) was used as incident radiation, and a pass energy of 200 eV was
chosen resulting in an effective instrument resolution higher than 0.6 eV. Charging effects were
compensated using a flood gun, and binding energies were calibrated based on positioning the
main C 1s peak at 285 eV, which originates from carbon contaminations. Prior to the measurements the samples were dehydrated at 200 ◦ C under ultra-high vacuum (UHV) conditions.
Measured data were fitted using Shirley-type backgrounds and a combination of GaussianLorentzian functions with the CasaXPS software. The Ti/Si atomic concentration ratios were
obtained by determining the integral area of the Gaussian-Lorentzian functions and correcting
the values by the specific atomic sensitivity factors proposed by Wagner. [110]
UHV-FTIR experiments with the materials were carried out using a UHV apparatus, which
combines a state-of-the-art vacuum IR spectrometer (Bruker, VERTEX 80v) with a novel UHV
system (Prevac). (For details, please see ref. [111] ) The powder samples were first pressed into
a gold covered stainless steel grid and then mounted on a sample holder that was particularly
designed for the FTIR transmission measurements under UHV conditions. The base pressure
in the measurement chamber was 2 x 10-10 mbar. The optical path inside the IR spectrometer
and the space between the spectrometer and UHV chamber was evacuated to avoid atmospheric
moisture adsorption, thus resulting in a high sensitivity and long-term stability.
XAS around the Ti K-edge (4966 eV) and the Zn K-edge (9659 eV) was measured in transmission mode at Hasylab, beamline C (Hamburg, Germany). Both X-ray absorption near edge
spectroscopy (XANES) and extended X-ray absorption fine structure (EXAFS) were measured.
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
35
The double Si(111) crystal monochromator was detuned to 50 % of its maximum intensity to
reduce the higher harmonics. Prior to the measurements, the samples were re-calcined at 773 K
for 1 h in synthetic air, purged with nitrogen and sealed in Kapton tape inside a glove box. The
spectra were then recorded under vacuum at liquid nitrogen temperature. A foil of Ti or Zn,
respectively, inserted between the second and third ionization chamber, was measured simultaneously and used as a reference for energy calibration. Data treatment was carried out with
the software package VIPER. [112] In the spectra of the absorption coefficient µ , a Victoreen
polynomial was fitted to the pre-edge region for background subtraction. A smooth atomic
background µ 0 was evaluated using a smoothing cubic spline. The Fourier analysis of the
km2 -weighted experimental function µ = (µ - µ 0 )/µ 0 was performed with a Kaiser window.
TPD measurements were performed in a stainless-steel flow set-up equipped with a calibrated
on-line mass spectrometer (Balzers GAM400). In a typical experimental sequence 50 mg of
catalyst were placed in a quartz-lined stainless-steel U-tube reactor. A thermocouple was placed
into the catalyst bed to measure the temperature during the desorption experiments. As a pretreatment the sample was heated to 400 ◦ C with a heating ramp of 10 K min-1 in 50 Nml min-1
1 % O2 /He and then kept for 1 h. After cooling to room temperature the reactor was purged
with He and subsequently CO2 adsorption was performed with a flow rate of 50 Nml min-1 4.1
% CO2 /He for 15 min. After purging with He for 30 min the temperature was increased to 400
◦ C with a heating ramp of 10 K min-1 and held at this temperature for 1 h. During desorption
the concentrations of CO2 and He were measured continuously. Afterwards, NH3 was adsorbed
at 100 ◦ C with a flow rate of 50 Nml min-1 0.8 % NH3 /He for 15 min and subsequently, after
cooling to 60 ◦ C in He, the NH3 TPD was conducted similarly.
5.1.2 Results and Discussion
Structural characterization
The synthesis of the SBA-15 support was repeated several times, and surface areas from 787
m2 g-1 to 830 m2 g-1 were obtained reproducibly, resulting in an average surface area of 800
m2 g-1 and an average pore radius of 3.5 nm.
The transition metal loading of the Ti(x)/SBA-15 materials was obtained by means of ICPOES, where x is specified in Ti atoms nm-2. The Ti loadings and the corresponding band edges
are summarized in Table 5.1. The weight percentages of Ti determined by ICP-OES were
converted to Ti nm-2 using the BET surface of SBA-15 of 800 m2 g-1 , according to Ti/nm2 =
(m(Ti)/M(Ti) x NA) / (800 m2 /g x (1-m(Ti)-m(Zn)) x 1018 nm2/m2 ) in which m(Ti) denotes the
grams of Ti per gram of sample, M(Ti) is the molar mass of titanium, and NA is the Avogadro
number. For Zn/nm2 calculations were done accordingly. The calculated values show that
the desired loading of Ti is obtained for the Ti0.3/SBA sample. A loading of ∼ 1 Ti nm-2 is
obtained reproducibly, when an equivalent amount of 1.5 Ti nm-2 as in case of Ti1.0/SBA is
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
36
Table 5.1: Ti and Zn contents, edge energies, and BET surface areas of the grafted samples.
Samplea
Ti0.3/SBA
Ti1.0/SBA
Ti2.0/SBA
Ti2.7/SBA
Zn1.0/SBA
Ti0.9/Zn0.9/SBA
Zn0.7/Ti0.9/SBA
Zn0.3/SBA
Zn0.3/Ti1.2/SBA
Ti1.2/Zn0.3/SBA
Ti loading (ICP-OES)
Zn loading (AAS)
Edge energy
(UV-Vis)
BET surface
area
[wt%Ti]
[Ti/nm2 ]
[wt%Zn]
[Zn/nm2 ] [eV]
[m2 /g]
2.1
6.3
11.3
14.8
4.5
4.9
6.7
6.9
0.3
1.0
2.0
2.7
0.9
0.9
1.2
1.2
7.5
6.7
4.9
2.3
2.6
2.3
1.0
0.9
0.7
0.3
0.3
0.3
749
653
575
460
760
515
508
760
552
n. d.
a
4.34
4.11
3.87
3.77
4.21
4.08
4.13
4.15
Ti(x)/Zn(y)/SBA: x and y denote atoms nm-2
used in the grafting procedure. Furthermore, subsequent grafting of two or even three times
using a solution equivalent to 1.5 Ti nm-2 resulted in ∼ 2 Ti nm-2 and ∼ 2.7 Ti nm-2 in case of
Ti2.0/SBA and Ti2.7/SBA. These results indicate that a maximum of 1 Ti nm-2 can be deposited
per grafting step. The same observation has been made for the grafting of Ti(OiPr)4 onto
MCM-48, [89] and a maximum achievable loading of roughly 12 to 13 wt % TiO2 (not Ti metal)
has previously been reported in a study of different grafting procedures on SBA-15. [80,88] The
increase in coverage by about 1 Ti nm-2 after each grafting step also suggests that the OH
groups as anchoring sites for the precursor are recovered after the calcination step. [113]
In order to check for the possible segregation of TiOx to the external surface of the SBA-15
particles, the surface Ti/Si ratios were determined by XPS measurements. The results of the
XPS measurements are shown in Fig. 5.1. Surface segregation of Ti can be excluded as the
Ti/Si ratios determined by ICP-OES match those derived from XPS. Obviously, the diffusion
of the precursor through the pores of the SBA-15 support material is fast and the TiOx species
are evenly distributed within the pores of the SBA-15.
For the synthesis of Zn-containing samples the as-received Ti1.0/SBA sample was subsequently grafted with the equivalent amount of 1.5 Zn/nm2 resulting in the sample labeled
Zn0.7/Ti0.9/SBA. Furthermore, in sample Ti0.9/Zn0.9/SBA the order of grafting steps was
changed, so that Ti was grafted on a sample that already contained Zn. Finally, Zn1.0/SBA was
prepared without an additional grafting of Ti to investigate the effect of ZnOx species incorporated in a SBA-15 matrix. The Zn and the Ti loadings are summarized in Table 5.1. Small
differences in the calculated surface loadings (in metal nm-2) might be attributed to a slightly
different surface area or the error limit in the results obtained by AAS or ICP-OES. In agree-
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
3
14
12
10
8
6
4
2
0
0
2
4
6
8
10
12
Ti/Si bulk ratio [%]
(a) XPS
14
16
1500
1350
1.8
A)
desorption Dv(d) [cc/A/g]
Volume adsorbed [cm /g ]
Ti/Si surface ratio [%]
16
1200
1050
900
750
600
450
300
150
0
37
e)
d)
c)
b)
a)
0.0
0.2
0.4
p/p
0
0.6
0.8
[a.u.]
1.0
1.6
B)
1.4
1.2
1.0
0.8
0.6
0.4
e)
d)
0.2
c)
b)
0.0
a)
2
4
6
8
10
pore diameter [nm]
(b) Pore size distribution
Figure 5.1: (a) Ti/Si surface ratios derived from XPS as a function of the Ti/Si bulk ratios;
Ti(x)/SBA (black squares), Zn-containing samples (red circles). (b) A) N2 isotherms
obtained with a) SBA-15, b) Ti1.0/SBA, c) Zn0.7/Ti0.9/SBA, d) Ti0.9/Zn0.9/SBA, and
e) Zn1.0/SBA. B) Pore size distribution of these samples derived by applying the BJH
equation to the desorption branch of the isotherms.
ment with the results obtained for Ti(x)/SBA-15 materials, the maximum loading of Zn which
can be obtained in one grafting step is 1 Zn/nm2 .
The N2 isotherms and the pore size distribution of the synthesized materials shown in Fig.
5.1 are nearly unchanged for all materials independent of the grafting order or the species deposited onto SBA-15 with a narrow intense signal with a pore size maximum at a radius of
3.5 nm. The overall pore volume decreases after the first grafting step for Ti1.0/SBA accompanied by a decrease in BET surface area of roughly 150 m2 /g. This is not observed when
only a small amount of Ti (0.3 Ti/nm2) is deposited (Table 5.1). The BET surface area and
the pore volume of Zn1.0/SBA are almost unchanged compared to the bare SBA-15 support.
SBA-15 displays a certain degree of microporosity, [114,115] and the decrease in surface area in
Ti(x)/SBA-15 materials is attributed to a clogging of the micropores by TiOx species, whereas
the micropores were unchanged after grafting of Zn. [116] Using the t-method we confirmed this
suggestion by calculating the micropore volume of the SBA-15, Ti1.0/SBA, and the Zn1.0/SBA
material. A micropore volume of 0.054 cm3 /g was calculated for the SBA-15. The micropore
volume was not changed upon grafting of Zn as in Zn1.0/SBA (0.06 cm3 /g), whereas the micropore volume drops to 0.02 cm3 /g in case of Ti1.0/SBA. For higher loaded samples there is
no further decrease in the overall pore volume, while the specific BET surface area decreases
further in Ti2.0/SBA, Ti2.7/SBA, and in the subsequently grafted samples Zn0.7/Ti0.9/SBA
and Ti0.9/Zn0.9/SBA. This can be attributed to a higher content of heavier elements that do not
contribute to the surface area. Interestingly, the decrease in surface in the subsequently grafted
samples Ti0.9/Zn0.9/SBA and Zn0.7/Ti0.9/SBA is more pronounced than in Ti2.0/SBA.
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
38
B)
F(R) [a.u.]
A)
6)
4)
5)
7)
3)
2)
increasing
Ti loading
8)
3)
1)
250 300 350 400 450
250 300 350 400 450
Wavelength [nm]
Figure 5.2: A) Diffuse reflectance UV-Vis spectra of dehydrated Ti(x)/SBA-15 materials with varying Ti loading and B) DR UV-Vis spectra of dehydrated samples containing ZnOx
species. 1) SBA-15, 2) Ti0.3/SBA, 3) Ti1.0/SBA, 4) Ti2.0/SBA, 5) Ti2.7/SBA, 6)
Zn0.7/Ti0.9/SBA, 7) Ti0.9/Zn0.9/SBA, and 8) Zn1.0/SBA materials.
UV-Vis diffuse reflectance spectroscopy, XPS, and XAS were used in order to investigate the
chemical state and the relative dispersion of Ti and Zn within the SBA-15 matrix.
From UV-Vis diffuse reflectance measurements, the light absorption of the Ti(x)/SBA-15 materials was estimated by using the Kubelka-Munk function. The edge energies for all samples are
summarized in Table 5.1. The absorption onsets of the four different dehydrated Ti(x)/SBA15 materials are shown in Fig. 5.2A. All samples exhibit an absorption onset in the range of
270 - 350 nm, which is clearly blueshifted compared to bulk anatase. While SBA-15 shows
no clear absorption feature in the UV-Vis spectrum, Ti0.3/SBA absorbs below 270 nm. Thus,
the absorption feature observed in the UV-Vis measurements is solely influenced by the presence and the local electronic structure of TiOx. The absorption edge of Ti0.3/SBA was derived
to be 4.34 eV. Compared to Ti0.3/SBA the higher loaded sample Ti1.0/SBA is redshifted by
∼ 0.2 eV. Ti2.0/SBA and Ti2.7/SBA are further redshifted, and the corresponding absorption
edge energies were calculated to be 3.87 and 3.77 eV, respectively. UV-Vis measurements performed by Wachs et al. [65] showed that the absorption edge clearly depends on the degree of
agglomeration of the TiOx domains. They obtained an absorption edge energy of 4.3 eV for
isolated tetrahedrally coordinated TiOx , which is in good agreement with the absorption edge
energy determined for Ti0.3/SBA. Furthermore, they pointed out that absorption edge energies
close to 3.4 eV can be observed for polymeric TiOx chains. Similar trends with increasing Ti
content were reported by Brub et al. [84,117] and Yamashita et al. [118] . The assignment of the
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
b) O 1s
a) Ti 2p
c) Zn 2p
39
d) Si 2p
x 0.5
CPS [a.u.]
7)
7)
6)
6)
6)
5)
5)
7)
4)
5)
4)
3)
4)
3)
2)
5)
3)
1)
TiO
2
TiO
469
462
455
2
536 532 528
4)
1050 1035 1020
1)
108
104
100
Binding energy [eV]
Figure 5.3: XP spectra in the a) Ti 2p region, b) O 1s region, c) Zn 2p region, and d) Si 2p region:
1) SBA-15, 2) Ti0.3/SBA, 3) Ti1.0/SBA, 4) Zn0.7/Ti0.9/SBA, 5) Ti0.9/Zn0.9/SBA, 6)
Ti2.7/SBA, and 7) Zn1.0/SBA. Bulk TiO2 is included as a reference.
intermediate species with absorption edge energies in between those values is still not fully
clear.
In Fig. 5.2B the UV-Vis spectra of the different Zn-containing SBA-15 materials and the
Ti1.0/SBA sample are shown. Zn1.0/SBA (Fig. 5.2B, trace 8)) absorbs some deep UV light
comparable to the bare SBA-15 support (Fig. 5.2B, trace 1)). No indication for the presence
of bulk ZnO is observed. Similar UV-Vis spectra were reported by Chen et al. [101,119] for subnanometric ZnOx clusters in zeolites and Mihai et al. [103] for amorphous ZnOx supported on
SBA-15. Subsequently grafted samples containing TiOx and ZnOx species show absorption
features, which are clearly related to TiOx grafted onto the SBA-15 supports. While the absorption onsets of Zn0.7/Ti0.9/SBA, Zn0.3/Ti1.2/SBA, and Ti1.2/Zn0.3/SBA match the onset
of Ti1.0/SBA, there is a blueshift of the absorption onset in case of Ti0.9/Zn0.9/SBA (Fig.
5.2B, trace 7)). This suggests that the TiOx species has a lower coordination in this sample
probably due to the slightly lower Ti loading (Table 5.1).
Characteristic XP spectra of all samples in the Ti 2p, O 1s and Zn 2p regions of different materials are shown in Fig. 5.3. The Ti 2p3/2 peak of Ti0.3/SBA is centered at 460.4 eV and shifts
to 459.9 eV in Ti1.0/SBA. No further shift to lower binding energies was observed when either
the Ti loading was increased or when Zn was added to the SBA-15 materials. These values are
significantly higher than those of Ti 2p3/2 for bulk TiO2 at a binding energy of 459.0 eV (Fig.
40
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
5.3). Xia et al. [87] showed that already small quantities of bulk TiO2 can be resolved with XPS,
and therefore the formation of bulk TiO2 can be excluded in the Ti(x)/SBA-15 materials. Generally, it has been reported that binding energies of isomorphously substituted Ti in silicates
are shifted to higher values as compared to the oxides. [120,121] A binding energy of the Ti 2p3/2
peak of 460.3 - 460.7 eV for tetrahedrally coordinated Ti was observed by Brube et al. [84] for
samples with Ti/Si ratios smaller than 5 %. With increasing Ti/Si ratio they observed a continuous shift to lower binding energies. Stakheev et al. [122] also observed a decrease in binding
energy with increasing Ti content. However, they pointed out that the binding energy remains
relatively stable up to 10 wt% TiO2 . In these studies XP spectra were measured for TiO2/SiO2
materials with no specific pretreatment. Gao et al. [80] performed XPS measurements after dehydration of the TiO2 /SiO2 powders. They observed a shift of the Ti 2p3/2 binding energy with
increasing TiO2 content from 1 wt% to 5 wt%, but they did not observe a further shift with
increasing TiO2 content up to 15 wt%. Therefore, titanium species in the Ti0.3/SBA material
can be attributed to tetrahedrally coordinated TiO4. Furthermore, we observed a broader Ti 2p
signal and a larger Ti 2p splitting in the Ti(x)/SBA-15 materials with a FWHM of 2.5 eV and
Ti 2p splitting of 5.8 eV for the Ti0.3/SBA sample compared to a FWHM of 1.0 eV and a Ti
2p splitting value of 5.7 eV for the TiO2 bulk reference material. With increasing Ti loading a
decrease in the Ti 2p peak width and a decrease in the Ti 2p splitting values for the Ti(x)/SBA15 were observed. The decrease in the Ti 2p signal width might be explained by an increase in
the amount of higher coordinated Ti atoms in the Ti(x)/SBA-15 materials at higher Ti loadings
as shown by UV-Vis spectroscopy. This trend is in agreement with the observed shifts of the Ti
2p signals to lower binding energies which can also be explained with a higher coordination of
the Ti atom. [84] The larger Ti 2p splitting might be rationalized by a relaxation occurring in the
Ti atom due to its observed lower coordination.
The addition of Zn did not significantly alter the Ti XP spectra. The peak positions and the
line shapes of the Ti 2p3/2 peaks of the Ti0.9/Zn0.9/SBA and the Zn0.7/Ti0.9/SBA samples
are in good agreement with those of Ti1.0/SBA-15. It becomes more evident that Zn did not
significantly change the XPS data by analyzing the O 1s XP spectra, which are shown in Fig.
5.3b. The O 1s signal of pure SBA-15 consists of one intense line centered at 534 eV. With
increasing Ti content the peak position of the O 1s signal shifts to 533.5 eV for the Ti0.3/SBA
sample and is further shifted to 533.2 eV for the Ti1.0/SBA signal (Fig. 5.3). A gradual
decrease is also observed for the Si 2p signal. These shifts of the O 1s and the Si 2p signals
are well known for silicon oxides containing small amounts of Ti or other transition metals
and can be attributed to Si-O-Ti bonds. [84] In addition to the shifts to lower binding energies
a new feature at 531.5 eV can be resolved, which increases in intensity with increasing Ti
content. The comparison with the O 1s signal observed for bulk TiO2 indicates that this feature
cannot be associated with oxygen species in bulk TiO2 , because their XPS peak is centered
at a binding energy of 530.2 eV. Gao et al. [80] observed a similar feature at ∼ 531.2 eV for
samples containing at least 10 wt% TiO2 . They assigned this peak to Si-O-Ti bonds, which
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
41
were not resolved at lower loadings. The shift of the O 1s peak to higher binding energies
compared to TiO2, i.e. oxygen in a Ti-O-Ti bond, was explained with more electronegative and
less polarizable Si atoms being bound to oxygen in a Si-O-Ti bond. The energy difference ∆(Ti
2p3/2 -O 1s) was calculated to be 73.1 eV for the Ti0.3/SBA, while 71.3 eV were calculated for
the reference sample. However, the energy difference between ∆(Si 2p-O 1s) was comparable
for the SBA-15 and the Ti(x)/SBA-15 materials. Considering these energy differences it is
evident that the Ti-O bond is more affected than the Si-O bond. Hence, the shift and the low
binding energy feature observed in this study are attributed to Si-O-Ti bonds.
In contrast to these observations, Zn1.0/SBA exhibits essentially the same peak position and
the same line shape in the O 1s region as the bare SBA-15 support. The lack of a peak shift
can be attributed to a smaller difference in electronegativity between Zn and Si compared to Ti
and Si. Similarly, the O 1s region of Zn0.7/Ti0.9/SBA is comparable to Ti1.0/SBA indicating
that the amount of Ti-O-Si bonds is similar. In contrast, the O 1s peak of Ti0.9/Zn0.9/SBA
is shifted to higher binding energy compared to Ti1.0/SBA indicating that less Ti-O-Si bonds
are formed. For this sample differences in coordination have also been suggested by UV-Vis
spectroscopy. The XP Zn 2p spectra are shown in Fig. 5.3c. Zn 2p3/2 peaks with binding
energies of 1022.9 - 1023.9 eV can be detected, which are about 0.7 - 1.7 eV higher than that
of bulk ZnO. [109] Similar shifts of the Zn 2p3/2 binding energy were reported by Tkachenko
et al. [102] This shift is indicative for reasonably well isolated ZnOx species with Zn in the
divalent oxidation state in all samples as explained above for isolated TiOx. For the Zn1.0/SBA
sample the highest binding energy of the Zn 2p3/2 signal is observed. Depending on the order
of the grafting steps, the Zn 2p3/2 peaks shift to lower binding energy with the lowest binding
energy observed for Zn0.7/Ti0.9/SBA. The shift of the peak position in the subsequently grafted
samples Zn0.7/Ti0.9/SBA and Ti0.9/Zn0.9/SBA can be attributed to an effective blocking of
surface sites by TiOx species, and therefore ZnOx species seem to be less isolated compared to
Zn1.0/SBA.
The Ti and Zn K-edge XANES spectra were measured in order to gain further information
about the coordination of dehydrated Ti(x)/SBA-15 and Ti(x)/Zn(x)/SBA-15 materials. The
XANES spectra of bulk TiO2 in the anatase and rutile phase and Ti0.3/SBA are shown in Fig.
5.4a. XANES measurements revealed that the pre-edge features of Ti(x)/SBA-15, showing one
distinct pre-edge peak maximum, clearly differ from the pre-edge feature of octahedrally coordinated bulk TiO2 exhibiting three weak pre-edge peaks. [67] In agreement with the UV-Vis and
XPS results the formation of crystalline bulk-like structures can be excluded from the XANES
measurements for each Ti(x)/SBA-15 sample. The intensities of the pre-edge feature determined by the normalized XANES data are plotted against the Ti loading in Ti/nm2 in Fig. 5.4b.
For Ti(x)/SBA-15 materials a linear relationship between the Ti loading and the pre-edge intensity was found, that is, with increasing Ti loading the pre-edge height decreases linearly. Farges
et al. [123] found that reliable information about the Ti coordination can be obtained by the preedge intensity. They also pointed out that the pre-edge position of the Ti K-edge XANES
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
2.0
b)
a)
Pre-edge peak intensity
Normalised absorption [a.u.]
42
1.5
1.0
0.5
3)
2)
1)
1)
0.7
0.6
0.5
4)
2)
5)
0.4
3)
0.3
0.0
4965
4980
4995
5010
Photon energy [eV]
0
1
2
3
2
Ti/nm
Figure 5.4: a) Comparison of the XANES spectra of 1) TiO2 rutile modification, 2) TiO2 anatase
modification, and 3) isolated TiOx in Ti0.3/SBA. b) Pre-edge intensities obtained
from normalized Ti K-edge XANES of Ti(x)/SBA-15 materials plotted against the
Ti loading in Ti/nm2 obtained from ICP-OES. The red circles are obtained by plotting the Ti pre-edge intensity of Zn0.7/Ti0.9/SBA and Ti0.9/Zn0.9/SBA against the
Ti loading, whereas blue stars are obtained for Ti pre-edge intensities plotted against
the overall transition metal content. 1) Ti0.3/SBA, 2) Ti2.0/SBA, 3) Ti2.7/SBA, 4)
Zn0.7/Ti0.9/SBA, 5) Ti0.9/Zn0.9/SBA.
spectra should be taken into account. Based on these variables three different well-defined
groups were identified, namely four-, five-, and sixfold coordinated titania. These observations
were successfully adopted to determine the coordination of Ti in different TiO2 /SiO2 materials
by Gao et al. [80] In the present study the pre-edge peak maximum was centered at photon energies ranging from 4669.9 eV for Ti0.3/SBA to 4670.1 eV for Ti2.7/SBA. According to Farges’
results these pre-edge energies identify fourfold coordinated Ti species. [123] For Ti0.3/SBA the
pre-edge intensity perfectly fits into the predicted range. However, the pre-edge intensity is far
too low in case of Ti2.0/SBA and Ti2.7/SBA to be exclusively fourfold coordinated indicating the presence of mixtures of different Ti coordination spheres. As pointed out by Vining
et al. [106] , a distortion of the tetrahedral geometry of TiOx causes a less intense pre-edge peak
attributed to lower Ti 3d-4p orbital hybridization and less overlap with the O 2p orbitals. Correspondingly, for Ti2.0/SBA it is reasonable to assume a mixture of predominantly four- and
fivefold coordinated titania with minor contributions of sixfold coordinated titania and slightly
distorted tetrahedral TiOx structures. In case of Ti2.7/SBA the amount of sixfold coordinated
Ti is clearly increased.
The Ti pre-edge peak intensities of the subsequently grafted samples Zn0.7/Ti0.9/SBA and
Ti0.9/Zn0.9/SBA plotted against the loading of Ti (Ti/nm2, red circle) and plotted against the
43
2.0
b)
a)
5)
(k)|
1.5
0.5
2
1.0
4)
|FT(k
Normalised absorption [a.u.]
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
5)
3)
4)
2)
3)
2)
0.0
1)
1)
9645
9660
9675
9690
Photon energy [eV]
0
2
4
6
8
R [A]
Figure 5.5: a) XANES spectra and b) Fourier-transformed EXAFS spectra (Zn K-edge) of Zncontaining SBA-15 materials. 1) Zn foil, 2) ZnO, 3) Zn1.0/SBA, 4) Zn0.7/Ti0.9/SBA,
and 5) Ti0.9/Zn0.9/SBA.
total transition metal loading ((Zn+Ti)/nm2, blue stars) are also shown in Fig. 5.4b. Compared to the pure Ti(x)/SBA-15 materials, titania in Zn0.7/Ti0.9/SBA and Ti0.9/Zn0.9/SBA
appears to be higher coordinated, especially in the Ti0.9/Zn0.9/SBA sample. While the Ti coordination in Zn0.7/Ti0.9/SBA is quite similar to the Ti2.0/SBA sample, the Ti coordination in
the Ti0.9/Zn0.9/SBA sample resembles that of Ti2.7/SBA. A similar trend in the Ti pre-edge
peak intensity was reported by Vining et al. [106] for subsequently prepared VOx /TiOx/SiO2
material with different V/Ti ratios, which was attributed to V-Ti-interactions. Therefore, the
results of the present study indicate a Zn-O-Ti interaction after subsequent grafting in the
Ti(x)/Zn(x)/SBA-15 materials. Less Zn-O-Ti interactions are formed in Zn0.7/Ti0.9/SBA compared to the Ti0.9/Zn0.9/SBA material. On the contrary, more Zn-O-Ti interactions are present
in Ti0.9/Zn0.9/SBA. The small differences in total transition metal loading cannot explain the
differences in pre-edge intensities. Instead, they may be an indication for a preferred interaction of the Ti precursor with surface ZnOx species during the sequential grafting. This is in
agreement with the presence of less Ti-O-Si bridges as indicated by the XPS results and the
blueshift of the absorption edge as observed by UV-Vis spectroscopy. On the other hand, the
Zn precursor does not seem to coordinate to TiOx species preferably, indicating that the order
of grafting steps significantly affects the Ti coordination.
Comparing the Zn absorption edge position measured by XANES of the Zn1.0/SBA and the
Ti(x)/Zn(x)/SBA materials indicates that Zn in all samples is exclusively present in the 2+ state
and not in the metallic state (Fig. 5.5a). The FT-EXAFS spectra of the Zn region of the three
44
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
Zn-containing samples Zn1.0/SBA, Zn0.7/Ti0.9/SBA, and Ti0.9/Zn0.9/SBA are shown in Fig.
5.5b. ZnO and Zn foil references are also included.
While the first coordination shell is significantly lower compared to the ZnO reference, the
second and further coordination shells can be barely seen in FT-EXAFS plots of the SBA-15
materials. Similar FT-EXAFS results were reported previously. [102,104] Tkachenko et al. [102]
pointed out that the best agreement between experimental and model spectra is obtained with
Si as second neighbor. Based on the FT-EXAFS data it can be concluded that the ZnOx species
are present as isolated species in all Zn-containing materials, while the presence of very small
islands cannot be excluded. It has to be noticed that a thorough inspection of the region between
2 and 4 Å in the Fourier-transformed EXAFS spectra revealed the second coordination shell to
be slightly more pronounced in the Zn0.7/Ti0.9/SBA sample. This might be another indication
that during subsequent grafting of Zn onto TiOx /SBA-15, the Zn precursor does not coordinate
to TiOx, but rather ZnOx species are formed in closer proximity on the remaining silica surface,
which is not blocked by TiOx, due to a strong interaction between Zn and silica as already
reported by Tkachenko et al. [102] Combining the results obtained by the UV-Vis, XPS, and XAS
studies the structure of TiOx in dehydrated SBA-15 can be described as isolated, tetrahedrally
coordinated TiO4 at low Ti loadings, while with increasing Ti content agglomerates are formed.
These findings are in good agreement with the results obtained by Wachs et al. [65,80] ZnOx in
all samples was found to be isolated or agglomerated in very small clusters. The tendency
to form higher agglomerated ZnOx seems to be increased when TiOx is already present on
the surface of the material due to a less favorable interaction between the Zn precursor and
TiOx species and a strong interaction between Zn and the silica support. [102] In contrast, the
Ti precursor seems to interact with ZnOx preferably. Correspondingly, in the two subsequently
grafted samples, TiOx and ZnOx are in close proximity, but more Ti-O-Si bonds of isolated
TiOx species are present in Zn0.7/Ti0.9/SBA.
CO2 /NH3 adsorption capacities determined by TPD
Having established the structures and interactions of surface TiOx and ZnOx species, the
question needs to be addressed as to whether they are accessible for adsorption and whether
they act as acidic or basic sites. The CO2 adsorption capacities of the Ti(x)/SBA and the
Ti(x)/Zn(x)/SBA samples were evaluated by TPD experiments. The obtained desorption profiles are shown in Fig. 5.6, and the calculated amounts of desorbed CO2 are summarized in
Table 5.2.
ZnO is known to favor the adsorption of CO2 , [96,124] and indeed the TPD results indicate that all
SBA-15 materials loaded with ZnOx species adsorb CO2 . Zn1.0/SBA shows the most intense
CO2 desorption signal in the low-temperature region and a small increase at higher temperatures above 650 K. The high-temperature desorption is related to strongly chemisorbed CO2 ,
most likely in the form of carbonates. The heating ramp was not extended beyond the pretreatment temperature in order to quantify the amount of the high-temperature carbonate, as it
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
5.0
x 10
-3
f)
e)
3.0
CO
2
[%]
4.0
45
2.0
d)
c)
1.0
0.0
b)
a)
250
300
350
400
450
500
550
600
650
700
750
Temperature [K]
Figure 5.6: CO2 TPD profiles obtained with a) SBA-15, b) Ti1.0/SBA, c) Zn0.3/SBA, d)
Zn1.0/SBA, e) Ti0.9/Zn0.9/SBA, and f) Zn0.7/Ti0.9/SBA.
cannot be excluded that the desorbing species have been present already prior to CO2 adsorption. Furthermore, species desorbing at such high temperature are likely irrelevant in catalysis
or for the use as reversible adsorbent. The high-temperature desorption feature was not included in the integration of the adsorbed amount of CO2 for all samples. Zn0.7/Ti0.9/SBA and
Ti0.9/Zn0.9/SBA also exhibit a CO2 desorption signal in the low-temperature region.
With decreasing Zn content the signal intensity clearly decreases in the low-temperature region and the high-temperature desorption feature more or less disappears. The total CO2
adsorption capacity decreases in the order Zn1.0/SBA > Zn0.3/SBA > Ti0.9/Zn0.9/SBA
> Zn0.7/Ti0.9/SBA. Calculations of the amount of CO2 desorbed normalized by the Zn content as determined by AAS, i.e. the number of adsorbed CO2 molecules per ZnOx site present
in the materials, revealed that only a small fraction of the ZnOx species adsorb CO2 . It is
possible that only those ZnOx species that have a certain geometric arrangement are suitable
for CO2 adsorption. The amount of CO2 per ZnOx site is similar for Ti0.9/Zn0.9/SBA and
Zn0.7/Ti0.9/SBA, but lower than for titanium-free samples. This observation implies that the
presence of TiOx lowers the amount of CO2 adsorbed, but this effect is not influenced by the
order of the grafting steps. This indicates that the grafting of TiOx onto ZnOx /SBA-15, during
which the Ti-precursor seems to coordinate preferably to ZnOx , does not lead to an additional
blocking of adsorption sites for CO2 .
Furthermore, the CO2 TPD measurements revealed that SBA-15 materials, which do not contain ZnOx species, do not exhibit any significant CO2 desorption signals. As the adsorption
of CO2 is a general indicator for basic sites on surfaces, it can be concluded that the samples
contain hardly any basic sites in the absence of ZnOx . [125] Srivastava et al. [93,126] and Srinivas
et al. [127] observed a small amount of CO2 during the TPD experiments. However, a similar
amount of CO2 was adsorbed on the bare SBA-15 support. In the present study bare SBA-15
46
Table 5.2: Amount of acidic and basic sites determined by NH3 and CO2 TPD measurements for the Ti(x)/Zn(y)/SBA samples.
sample
SBA-15
CO2 [mol/g] 0.0
NH3 [mol/g] 0.0
Zn [mol/g]
a
metal
[mol/g]
n(CO2 )/n(Zn) n(NH3 )/n(metal)a
Ti0.3/SBA Ti1.0/SBA Ti2.0/SBA Ti2.7/SBA Zn0.3/SBA Zn1.0/SBA Zn0.7/Ti0.9/SBA Ti0.9/Zn0.9/SBA
0.0
92.1
438
0.2
204.0
1315
0.4
360.0
2359
0.5
465.3
3090
2.3
151.3
520
520
4.4
272.8
1147
1147
1.4
389.4
750
1772
2.0
422.7
1025
1964
0.21
0.16
0.15
0.15
4.4x10-3
0.30
3.8x10-3
0.24
1.9x10-3
0.22
1.9x10-3
0.22
total transition metal content (Ti+Zn) from elemental analysis
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
Amount of
substance
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
47
did not adsorb any CO2 . Differences in the observations may be explained by different calcinations and the pretreatment temperatures. Srivastava et al. [126] pointed out that probably surface
OH groups are the sites were CO2 adsorption or activation of the CO2 molecule takes place.
As the pre-treatment temperature in this study was higher, these surface sites may have been
removed. In a recent study Danon et al. [74] studied the interaction of TiOx /SBA-15 with CO2
and H2 O by means of Raman spectroscopy. In agreement with the results of our study, they
observed that there is hardly any interaction between CO2 and TiOx /SBA-15.
On the contrary, all TiOx -grafted and subsequently grafted ZnOx /TiOx samples contain acidic
sites as shown by adsorption-desorption experiments using NH3 . The amount of acidic sites is
about a factor of 102 larger than the amount of basic sites in all materials. The NH3 adsorption
capacity increases linearly with increasing Ti loading except for the Ti0.3/SBA sample. This
observation points to an improved interaction between fully isolated TiOx and NH3 . The presence of acidic sites in similar systems was already reported by Iengo et al. [92] Using pyridine as
basic probe molecule for TPD experiments they reported a nearly linear relationship between
the TiOx content and the surface acidic sites. [64] In a more recent study Srivastava et al. [126]
pointed out that Ti-SBA-15 possesses only weak Lewis centers, whereas SBA-15 did not contain any acidic sites. These observations are in good agreement with our observations that all
Ti atoms in Ti(x)/SBA-15 materials seem to be accessible for NH3 adsorption. In case of the
two subsequently grafted samples Zn0.7/Ti0.9/SBA and Ti0.9/Zn0.9/SBA, NH3 adsorption is
slightly more favorable in the Ti0.9/Zn0.9/SBA material. This is in good agreement with the Zn
and Ti loadings determined by ICP-OES measurements revealing that a higher Zn content was
achieved in Ti0.9/Zn0.9/SBA. Furthermore, these results illustrate that the overall amount of
NH3 adsorbed on Ti0.9/Zn0.9/SBA is approximately equal to the sum of the amount adsorbed
on Ti1.0/SBA and the amount adsorbed on Zn1.0/SBA. Consequently, subsequent grafting of
Zn and Ti did not block a significant number of acidic adsorption sites, indicating that the
species do not block each other.
In summary, it was shown that the grafting of either TiOx or ZnOx introduces acidic sites, and
that the amount scales almost linearly with the total amount of transition metals. Basic sites
can be generated by the grafting of ZnOx , but their number is a factor of 102 smaller than that
of the acidic sites. Further studies are required concerning the nature of the adsorbed species
formed from CO2 on TiOx/ZnOx /SBA-15.
CO2 adsorption studied by UHV-FTIR
The interaction of CO2 was studied using Fourier-Transformed Infrared (FTIR) spectroscopy.
The FTIR spectrometer is coupled to a UHV system allowing for high sensitivities and low
contamination levels during the measurements. Prior to CO2 adsorption the samples were pretreated at 700 K in vacuum. The adsorption studies were performed at 90 K. The stability
of adsorbed species was evaluated by heating the samples up to 300 K after CO2 adsorption.
Additional CO2 adsorption experiments were carried at 300 K.
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
48
a)
b)
H
G
Absorbance
2350
2343
0.01
F
2278
clean surface
0.001
1400
1442
G
1349
1667
E
2
1610
1570
1582
1616
0.2
+ CO
c)
1375
D
H
F
G
E
F
D
E
C
D
C
C
B
B
B
A
A
A clean surface
2400 2350 2300 2250
-1
Wavenumber [cm ]
1800 1650 1500 1350
-1
Wavenumber [cm ]
clean surface
1650 1500 1350
-1
Wavenumber [cm ]
Figure 5.7: UHV-FTIR spectra obtained after exposing the clean Zn1.0/SBA-15 sample to CO2 at
90 K: (A) clean sample, (B) 1x10-5 mbar CO2 , (C) 5x10-5 mbar CO2 , (D) 1x10-4 mbar
CO2 , and heating to: (E) 150 K, (F) 200 K, (G) 250 K, and (H) 300 K. a) CO2 range, b)
carbonate range. c) CO2 dosing at 300 K: (A) clean surface, (B) 1x10-6 mbar CO2 , (C)
5x10-6 mbar CO2 , (D) 1x10-5 mbar CO2 , (E) 5x10-5 mbar CO2 , and (F) 1x10-4 mbar
CO2 . Afterwards the sample was evacuated for (G) 10 min.
The UHV-FTIR spectra obtained upon CO2 adsorption at 90 K at different CO2 pressures on
Zn1.0/SBA and upon subsequent heating of the Zn1.0/SBA sample are shown in Fig. 5.7a and
Fig. 5.7b. Obviously, three bands appear at relatively low pressures of 1 x 10-6 mbar CO2 .
These bands are situated at 2343, 2278, and 1375 cm−1 . With increasing CO2 pressures a
broad band at ∼ 1660 cm−1 appears. Additionally, the band at 2343 cm−1 is getting broader.
ZnO nanoparticles were recently studied under similar conditions by Noei et al. [98] . Based on
their assignments the intense band at 2343 cm−1 can be related to linearly physisorbed CO2 .
As Noei et al. [98] already pointed out that these values are close to those wavenumber observed
for gas-phase CO2 , and interestingly they also observed a broadening or even a splitting of
this band upon larger CO2 doses. Finally, the assignment of this broad band at 2343 cm−1
to linearly physisorbed CO2 was verified by the fact that it disappeared during heating to 200
K. [98] During heating of the sample the weak band at 2278 cm−1 and the sharp band at 1375
cm−1 decreased, too. During heating a so far unresolved band at 1442 cm−1 appeared due to
the disappearance of the sharp band at 1375 cm−1 . Furthermore, the heating procedure revealed
that the band at ∼ 1660 cm−1 remained unchanged up to 150 K. However, the band maxima
was shifted to lower wavenumbers (1616 cm−1 ) and a shoulder at 1582 cm−1 appeared upon
further heating to 300 K. A similar shift of certain bands upon heating was also observed by
Noei et al. [98] Usually, the bands observed at wavenumbers between 1300 and 1700 cm−1 are
assigned to carbonate or carboxylate species at different ZnO surfaces. [98,124,128] Especially
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
49
bicarbonates that are reversibly adsorbed at room temperature were assigned to bands at 1635
cm−1 and 1424 cm−1 . [124] However, in the case of Zn1.0/SBA the assignment of the bands in
this region to a certain ZnO surface or a certain carbonate/carboxylate species is not feasible.
The above presented characterization results (UV-Vis, EXAFS, and XPS) clearly indicate that
there is no defined bulk-like ZnO structure in case of the Zn/SBA material. The ZnOx species
are rather isolated, and therefore comparison to ZnO nanoparticles is not possible. Esken et
al. [128] studied ZnO nanoparticles with a diameter of ∼ 2 nm, which were incorporated into a
host system. Given that the CO2 adsorption features are similar to the results reported by Noei
et al. [98] the ZnOx species in Zn1.0/SBA are much smaller. Nevertheless, the UHV-FTIR data
unambiguously provide evidence for the interaction of CO2 and Zn1.0/SBA and the formation
of carbonate-like species at the surface of the Zn1.0/SBA material upon exposure to CO2 at 90
K might be possible. Furthermore, it was shown that certain species remained on the surface
even after heating to 300 K.
The observation of the formation of carbonate-like species was strengthened by the CO2 adsorption experiments performed at 300 K. The results of these experiments are shown in Fig.
5.7c. During adsorption at different CO2 pressures two broad bands appear at 1400 cm−1 and
1610 cm−1 . As mentioned above similar bands at 1635 cm−1 and 1424 cm−1 were assigned
to reversibly adsorbed bicarbonates. [124] These bands remained stable during evacuation of the
system, which is a clear evidence of a strong absorption at the surface of the Zn1.0/SBA sample. The bands are less intense demonstrating that the interaction of CO2 with the surface of
Zn/SBA is less likely at elevated temperatures. By combining the UHV-FTIR results and the
above presented TPD results it was possible to show that CO2 adsorption occurs at the surface
of Zn1.0/SBA, and that the TPD signal at low temperatures is possibly due to weakly bound
carbonate species, thus desorbing at low temperatures.
The adsorption of CO2 was also studied for Zn0.7/Ti0.9/SBA and Ti1.0/SBA by means of UHVFTIR measurements. The results obtained for low-temperature CO2 at different CO2 pressures
as well as representative data recorded during heating of the Zn0.7/Ti0.9/SBA and Ti1.0/SBA
samples are shown in Fig. 5.8, Fig. 5.9, and Fig. A2, respectively. Similar to the Zn1.0/SBA a
band was observed for the Zn0.7/Ti0.9/SBA sample in the CO2 region of the FTIR spectra at ∼
2345 cm−1 , which can be assigned to linearly physisorbed CO2 . However, the intensity of this
band is weaker and the band is significantly broader compared to the Zn1.0/SBA. Thus, an additional species situated at lower wavenumber (∼ 2336 cm−1 ) is present for the Zn0.7/Ti0.9/SBA
sample upon CO2 . Even though the signal gets narrower and the overall intensity of the signal
decreases upon heating the linearly physisorbed CO2 appeared to be more stable at the surface
of Zn0.7/Ti0.9/SBA, because at 220 K the band ∼ 2345 cm−1 can be still observed. Finally, the
band at 2278 cm−1 was also observed for the Zn0.7/Ti0.9/SBA sample. It is striking that the
intensity ratio between the two signals at ∼ 2345 cm−1 and ∼ 2278 cm−1 was much smaller
for the Zn0.7/Ti0.9/SBA.
In agreement with the UHV-FTIR CO2 adsorption results obtained for Zn1.0/SBA carbonate-
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
50
a)
2345
2336
Absorbance
2355
0.1
F
1378
b)
0.02
1598
1614
c)
1409
1440
0.001
F
E
D
1404
1578
D
1549
C
2278
1620
C
B
D
B
C
clean surface
clean surface
2400
2300
B
2200
-1
Wavenumber [cm ]
A
clean surface
1600
1400
-1
Wavenumber [cm ]
1650 1500 1350
-1
Wavenumber [cm ]
Figure 5.8: UHV-FTIR spectra obtained after exposing the clean Zn0.7/Ti0.9/SBA sample to CO2
at 90 K: (A) clean surface, (B) 1x10-4 mbar CO2 , and heating to: (C) 150 K, (D) 175 K,
(E) 200 K, and (F) 220 K. a) CO2 range, b) carbonate range. c) CO2 dosing at 300 K:
(A) clean surface, (B) 1x10-6 mbar CO2 , (C) 1x10-5 mbar CO2 , and (D) 1x10-4 mbar
CO2 .
related bands between 1300 and 1700 cm−1 were detected for the Zn0.7/Ti0.9/SBA sample
upon adsorption of CO2 at 90 K, but the bands are less intense compared to the Zn1.0/SBA
material. While a sharp signal at 1378 cm−1 was the most intense feature observed in the
carbonate region, which is in good agreement with the results of the Zn/SBA sample, a small
signal at 1549 cm−1 , which was absent or barely detectable for the Zn1.0/SBA sample, was
resolved for the Zn0.7/Ti0.9/SBA sample. Furthermore, the broad signal at ∼ 1660 cm−1 observed for the Zn1.0/SBA could not be detected for the Zn0.7/Ti0.9/SBA sample upon CO2
adsorption. The results of CO2 adsorption experiments performed with the Zn1.0/SBA and the
Zn0.7/Ti0.9/SBA indicate that the broadening of the band detected at ∼ 2345 cm−1 , the band at
∼ 2278 cm−1 and the sharp feature observed in the carbonate region at ∼ 1378 cm−1 are likely
to be related with each other. During heating the sharp band at 1378 cm−1 and the smaller signal at 1549 cm−1 disappeared and new signals at 1409 cm−1 , 1440 cm−1 , and 1616 cm−1 were
resolved. The enhanced stability observed for the bands in the CO2 region corresponds to the
enhanced stability of the band at 1378 cm−1 further indicating the relationship of the two signals. Besides this enhancement in stability the species observed after heating of the Zn/Ti/SBA
are similar to the species observed for Zn/SBA sample, and it is resonable to conclude that the
influence of TiOx species on the CO2 adsorption is small when CO2 is adsorbed at 90 K and
the sample is heated afterwards. An influence of TiOx can be only assumed at 90 K because of
the absence of the broad band at ∼ 1660 cm−1 observed for Zn1.0/SBA.
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
a) 2343
b)
E
51
0.001
D
Absorbance
D
C
C
0.01
B
A
clean surface
B
clean surface
2400
2300
A
2200
-1
Wavenumber [cm ]
1650 1500 1350
-1
Wavenumber [cm ]
Figure 5.9: a) UHV-FTIR spectra obtained after exposing the clean Ti1.0/SBA sample to CO2 at 90
K: (A) clean surface, (B) 1x10-4 mbar CO2 , and annealing to: (C) 120 K, (D) 150 K, and
(E) 170 K. b) Spectra obtained (carbonate region) after exposing the clean TiOx/SBA15 sample to CO2 at 90 K in the CO2 range: (A) clean surface, (B) 1x10-6 mbar CO2 ,
(C) 1x10-5 mbar CO2 , and (D) 1x10-4 mbar CO2 .
For CO2 adsorption on Zn0.7/Ti0.9/SBA performed at 300 K weak bands at 1620 cm−1 and
1404 cm−1 at high CO2 pressures were detected (Fig. 5.8c). Based on the CO2 adsorption
results obtained at 90 K and 300 K it can be concluded that in both cases reversibly adsorbed
bicarbonates are present at 300 K. [124] At lower temperatures additional absorption modes were
observed, which disappeared or shifted during heating. Additionally, it can be concluded that
the structure of the ZnOx species in Zn1.0/SBA and Zn0.7/Ti0.9/SBA is similar. Given that the
intensity of all bands is smaller in case of the Zn0.7/Ti0.9/SBA compared to the Zn1.0/SBA
sample less CO2 can be adsorbed at the Zn0.7/Ti0.9/SBA sample. This observation is in good
agreement with the smaller CO2 adsorption capacity determined by TPD experiments.
In contrast to the results obtained for Zn1.0/SBA and Zn0.7/Ti0.9/SBA a different CO2 adsorption behavior was observed for the Ti1.0/SBA sample. The CO2 region upon CO2 adsorption at
90 K of the Ti1.0/SBA is shown in Fig. 5.9a. The corresponding carbonate region during CO2
dosing is shown in Fig. 5.9b. Again, linearly physisorbed CO2 at 2343 cm−1 was detected.
The additional peak usually observed at 2278 cm−1 is absent in case of CO2 adsorption on
Ti1.0/SBA. Furthermore, the peak at 2343 cm−1 appeared to be completely symmetric even at
high CO2 pressures and in contrast to Zn1.0/SBA and Zn0.7/Ti0.9/SBA samples a slight broad-
52
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
ening of the signal at 2343 cm−1 is only observed during heating. In the carbonate region only
a broad band appeared during the CO2 dosing (Fig. 5.9b), which disappeared during heating.
Based on these observations it can be concluded that the typical features at 2278 cm−1 and
1378 cm−1 are due to the presence of ZnOx species in the Zn1.0/SBA and Zn0.7/Ti0.9/SBA
samples. Accordant to the results of low temperature CO2 adsorption on Ti1.0/SBA, adsorption
of CO2 at 300 K revealed that there is no specific interaction between the Ti1.0/SBA sample
and CO2 at elevated temperatures (Fig. A3). This is in agreement with the presented CO2 TPD
data evidencing that CO2 adsorption at RT is unlikely at isolated TiOx species as present in the
Ti1.0/SBA sample.
In summary, UHV-FTIR measurements of CO2 adsorption at Ti1.0/SBA, Zn1.0/SBA, and
Zn0.7/Ti0.9/SBA performed at 90 K and at 300 K provide insight into the interaction of the different materials with CO2 . It was shown that during adsorption at 300 K carbonates are formed
on Zn1.0/SBA and Zn0.7/Ti0.9/SBA. On Ti1.0/SBA CO2 was mainly adsorbed linearly at 90 K,
and the formation of carbonates upon CO2 dosing at 90 K and 300 K can be excluded. Therefore, the CO2 adsorption capacities as determined by TPD measurements of Zn0.7/Ti0.9/SBA
can be clearly attributed to the presence of ZnOx species rather than the presence of TiOx.
However, the UHV-FTIR measurements show that TiOx species influence the modes of CO2
adsorption in the Zn0.7/Ti0.9/SBA sample during CO2 dosing at 90 K. In general, the obtained
results are in good agreement with the structural characterization of the Zn0.7/Ti0.9/SBA sample indicating that the structure of ZnOx species is similar to the structure of ZnOx species in
Zn1.0/SBA.
5.1.3 Conclusions
Using an efficient grafting procedure, TiOx - and ZnOx -containing SBA-15 materials with different transition metal loadings were synthesized. The extensive characterization of the prepared materials showed that depending on the Ti loading isolated and polymerized TiOx species
were obtained. ZnOx species were found to be present as isolated species or very small islands.
XPS and X-ray absorption spectroscopy measurements indicated strong interactions between
ZnOx and TiOx species. When titania is grafted first, the sample contains considerable amounts
of isolated titania species. In the subsequent grafting of ZnOx onto TiOx /SBA-15 Zn(acac)2 interacts preferably with the free silica surface. Therefore, there is a tendency to form higher
agglomerated ZnOx species, while the decrease in the total number of isolated titania species
is less pronounced. In contrast, when TiOx is grafted on ZnOx /SBA-15 a preferred interaction
between the Ti precursor and surface ZnOx species was observed. Due to this interaction more
Ti-O-Zn and less Ti-O-Si bonds are formed in the material, and titania appears to be higher
coordinated. Using NH3 TPD a large amount of acidic sites was found in all materials. A significant CO2 adsorption capacity was observed only for the Zn-containing materials. The total
amount of acidic and basic sites was not influenced by the order of the grafting steps, indicating
5.1 Modification of titanate-loaded mesoporous silica by grafting of zinc oxide
53
that the transition metal oxide species do not block each other. UHV-FTIR studies confirm that
CO2 is interacting with ZnOx species incorporated into SBA-15. No specific interaction at 300
K was observed for TiOx species.
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
54
5.2 Photocatalytic CO2 reduction using Au-modified
TiOx/SBA-15 materials
Abstract2
Photo-deposition of Au nanoparticles enhances the hydrogenation rate in photocatalytic CO2
reduction to methane on titanate in SBA-15. Product formation from contaminants is ruled
out experimentally by a thorough pre-cleaning of the samples and the design of the vacuumtight gas-phase photoreactor. Without Au, an active carbon pool accumulates on the catalyst
and higher hydrocarbons are formed. By infrared spectroscopy it was found that the carbon
pool consists of formaldehyde/paraformaldehyde. Furthermore, the obtained results indicate
that adsorbed water in the mesopores of the catalyst is sufficient to achieve CO2 reduction and
convert CO2 mainly to methane. The results contribute to the understanding and systematic
improvement of photocatalysts.
2
The content of this section was submitted for publication as ”Influence of photodeposited gold nanoparticles
on the photocatalytic activity of titanate species in the reduction of CO2 to hydrocarbons” to Energy and
Environmental Science.
5.2 Photocatalytic CO2 reduction using Au-modified TiOx/SBA-15 materials
55
5.2.1 Short Introduction
Chemical fixation of CO2 into value-added materials or fuels by photocatalytic reduction at
semiconductor surfaces is currently attracting much attention as this kind of artificial photosynthesis is a promising technology for the future energy supply. [129] TiO2 -related materials in
various shapes and with different kinds of metal or non-metal doping are the most frequently
studied materials for this application. [14] It was already shown by Anpo et al. [58,67] in their
pioneering work that isolated TiOx -species incorporated into different matrices are active materials for the photocatalytic reduction of CO2 . More recently, it was shown by Anpo et al. [58,67]
that these materials exhibit better performance than bulk TiO2 when compared on a per gram
Ti basis. [130] However, there is still the need to further understand the processes involved in
photocatalytic reduction. It was suggested that CO is a likely intermediate by Frei et al. [71]
This was further supported by Mul et al. [73] , who proposed that formaldehyde is another likely
intermediate generated by CO and water. Furthermore, there is currently a dispute about CO2
adsorption on isolated Ti-species. While Anpo et al. [67] used the quenching of the characteristic TiOx photoluminescence feature in presence of CO2 as an indication for the interaction
between CO2 and the titania site, no specific interaction was observed by Stair et al. [74] by
means of UV Raman spectroscopy.
It was already shown by Mori et al. [131,132] that the deposition of noble metals like Au, Ni, or
Pd by a photo-assisted procedure is feasible for Ti single-site catalysts. Even though noble
metals are widely accepted to enhance photocatalytic activities due to an enhanced charge
separation, there is, to the best of our knowledge, no report about the effect of noble metals
or a comparative study between noble metal-modified and unmodified Ti single-site catalysts.
Therefore, we report for the first time on the effect of photo-deposited Au on the photocatalytic
CO2 reduction activities of Ti/SBA-15 materials.
Experimental
Sample synthesis
Photo-deposition of Au nanoparticles was performed in a continuous flow stirred-tank reactor equipped with a 700 W Hg immersion lamp cooled by water circulating in a double wall
jacket. [133] During photo-deposition the concentration of CO2 , O2 , and H2 were analyzed with
a non-dispersive IR photometer, a paramagnetic and a thermal conductivity detector (XStream,
Emerson Process Management), respectively. 350 mg of catalyst were dispersed in a solution
of 550 ml dest. H2 O and 50 ml MeOH. The reactor was deaerated for 60 min using pure
N2 prior to irradiation. The activity for photocatalytic methanol reforming of the Ti/SBA was
evaluated by irradiation for 60 min (in the absence of the Au precursor). 3.5 ml of a 1.5 x 10-3
molar auric acid solution (HAuCl4 99.9 %, Sigma Aldrich) were added to the suspension and
the reactor was again deaerated for 60 min using pure N2 . Photo-deposition was performed at
56
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
30 ◦ C with 50 % irradiation power (350 W) for 2.5 h. After successful photo-deposition the
materials were filtered and freeze-dried over night.
Characterization
DRIFT spectra of the samples diluted with diamond powder in a 1:1 ratio (by weight) were
recorded at different temperatures between RT and 400 ◦ C to remove surface-adsorbed water.
At 120 ◦ C the temperature was kept constant while the spectra were measured. The obtained
spectra were first converted to log(1/R) and then normalized to the absorption of the harmonic
vibration of SiO2 (∼2088 cm-1 - ∼1718 cm-1). Difference spectra were obtained as follows:
For spectrum a) the spectrum of Ti/SBA-15 (as-prepared) was subtracted from Ti/SBA-15 after
photocatalytic CO2 reduction. Both spectra were measured after drying at 120 ◦ C. Spectrum
b) was obtained by subtracting the spectrum of Ti/SBA-15 after photocatalytic CO2 reduction from the spectrum of Au/Ti/SBA-15 after photocatalytic CO2 reduction. Both spectra
were measured after drying at 120 ◦ C. Spectrum c) was obtained by subtracting the spectra
of Ti/SBA-15 after photocatalytic CO2 reduction and the spectra Au/Ti/SBA-15 after photocatalytic CO2 measured after drying at 400 ◦ C. XPS and UV-Vis spectra were obtained as
described previously.
5.2.2 Results and Discussion
The preparation and in-depth characterization of the Ti/SBA-15 material was described in detail
in the previous section. [134] In short, it was proven by UV-Vis, X-ray photoelectron and X-ray
absorption spectroscopy that tetrahedrally coordinated TiO4-species were obtained within the
mesoporous SBA-15 matrix. The Ti content was confirmed to be 2 wt% resulting in 0.3 Ti/nm2.
Deposition of Au nanoparticles onto the Ti/SBA-15 support was achieved by a photo-assisted
deposition procedure, [133] in which Au is deposited from auric acid in an aqueous methanol
solution. The evolution of H2 was constantly measured during the photo-deposition procedure
revealing that hydrogen evolution took place in the presence of auric acid (Fig. A4). This observation is a first indication of the successful deposition of Au nanoparticles onto Ti/SBA-15.
However, to further proof the existence of Au nanoparticles diffuse reflectance UV-Vis spectroscopy was employed showing a characteristic plasmon absorption peak of metallic Au at
∼ 550 nm. [135,136] Elemental analysis revealed that the obtained Au/Ti/SBA-15 sample contains 0.25 wt% of Au corresponding to the amount supplied during photo-assisted deposition.
Therefore, all the supplied Au ions were successfully deposited onto the support material.
The two different materials Ti/SBA-15 and Au/Ti/SBA-15 were thoroughly cleaned by illumination in humid He under static conditions in a fully metal-sealed gas-phase photoreactor
(Chapter 4). Recently, Mul et al. [73,137] showed that this pre-illumination procedure in the
presence of water vapor is mandatory to achieve reliable results in the photocatalytic activity
in gas-phase CO2 reduction. During the cleaning procedure in humid He mainly CH4 , C2 H4 ,
5.2 Photocatalytic CO2 reduction using Au-modified TiOx/SBA-15 materials
60
a)
C H
4
10
C H
3
40
8
C H
2
40
35
evacuation
20
15
10
5
0
1h
5h
7h
0h
time
1h
5h
7h
Au/Ti/SBA-15
10
25
Au/Ti/SBA-15
20
30
Ti/SBA-15
30
0h
6
CH
4
yield [ppm]
[ppm]
b)
Ti/SBA-15
4
45
Au/Ti/SBA-15
50
yield CH
Ti/SBA-15
57
0
1. run
1. run
2. run
2. run
Figure 5.10: a) Time course of the methane evolution observed for Ti/SBA-15 and Au/Ti/SBA-15
during two subsequent CO2 reduction experiments with 7h of irradiation, uncorrected
values. For Au/Ti/SBA-15, no GC sampling of products was performed after 7 h in the
first run. b) Evolution of different hydrocarbon species after 5 h of irradiation in two
subsequent CO2 reduction experiments. The yield of the four different hydrocarbon
species is corrected for possible carbon contaminations left on the catalyst.
C2 H6 , and C3 H8 were detected (Fig. A5), which is in good agreement with the carbon contaminations observed by Mul et al. [73] In addition to these hydrocarbons the evolution of MeOH was
observed for Au/Ti/SBA-15, which is most likely a result of the photo-deposition procedure,
in which methanol was used as sacrificial electron donor. MeOH was presumably converted
to CO2 , which was detected throughout prolonged irradiation in the Au photo-deposition procedure. It is remarkable that the overall amount of carbon contaminations was higher than the
contamination previously observed by Mul et al. [73] Therefore, subsequent cleaning steps were
necessary to obtain a clean catalyst surface. In case of Ti/SBA-15 four subsequent cleaning
steps were conducted, whereas Au/Ti/SBA-15 was cleaned seven times for 7 h each. Additionally, Au/Ti/SBA-15-C was prepared by calcination at 250 ◦ C in synthetic air with the aim
to reduce the number of necessary cleaning steps. This calcination temperature has previously
been chosen for the calcination of PVA-protected gold colloids on metal oxides. [138] However,
four additional cleaning steps in humid He were still necessary to sufficiently remove the carbon contaminations. The decrease of the concentration of the three main contaminants CH4 ,
C2 H4 , and C3 H8 of each cleaning step can be represented with high accuracy as an exponential
decay curve in case of the Au/Ti/SBA-15. For Ti/SBA-15 and calcined Au/Ti/SBA-15-C, on
the other hand, the decrease of the contamination concentration was linear.
The actual CO2 reduction experiments were performed under static conditions in an atmosphere
of 5 % CO2 and 6000 ppm water, which was balanced by He. These conditions were achieved
by saturating a CO2 /He stream with water at 273 K. A CO2 excess was chosen because of
the unfavorable interaction of CO2 with the isolated titanate species. The initial pressure was
adjusted to 1500 mbar. The hydrocarbon formation under illumination with a 200 W Hg/Xe
58
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
lamp was monitored during a period of 7 h. For this purpose samples were taken for GC
analysis after 1, 5, and 7 h. The decrease in pressure during sampling was taken into account
by normalization of the obtained signals to N2 used as an internal standard. After reaction the
reactor was pumped down to 5 x 10-2 mbar and flushed with high purity He. Blank experiments
revealed that no hydrocarbons are formed in the absence of a photocatalyst or in absence of
high-energy UV light, which was excluded using a 435 nm cut-off filter. The hydrocarbon
formation within 7 h of illumination for Ti/SBA-15 and Au/Ti/SBA-15 is shown in Fig. 5.10.
Obviously, in the first reaction run the yield of CH4 is significantly higher for the Au/Ti/SBA15 sample. However, a subsequent reaction period without further pretreatment revealed that a
threefold increase in the CH4 yield is observed for the Ti/SBA-15 sample, whereas the overall
yield for the Au/Ti/SBA-15 material is comparable to the first reaction run. Furthermore, it is
evident that the yield of CH4 after 1 h of irradiation for Ti/SBA-15 resembles the value obtained
after 7 h of irradiation in the first reaction run.
The comparison between the Au/Ti/SBA-15 and the calcined Au/Ti/SBA-15-C sample shows
that the CH4 yield is significantly lower for the calcined Au/Ti/SBA-15-C sample (Fig. A6).
After an additional surface cleaning in humid He, the performances of the different materials
were reproduced. The CH4 yields were similar for two subsequently performed reactions on
Au/Ti/SBA-15 and Au/Ti/SBA-15-C. An increase in the CH4 yield was again observed in the
2nd reaction run for Ti/SBA-15. However, it has to be noted that the overall yield was lower
indicating that both catalysts deactivated during prolonged irradiation (Fig. A6).
In addition to CH4 longer-chain hydrocarbons were produced in different quantities with both
materials. While C2 H6 , C3 H8 , and C4 H10 were produced in the two subsequent reactions using the Au/Ti/SBA-15 catalyst, negligible amounts of longer chain hydrocarbons were detected
with Ti/SBA-15 in the first run (Fig. 5.10b). However, after deactivation of the Ti/SBA-15 material longer chain hydrocarbon formation was also observed during the first reaction run (Fig.
A6). In addition to the threefold increase in the yield of CH4 , a threefold increase in ethane
formation was observed in the subsequent reaction on Ti/SBA-15 (Fig. 5.10b). Again, these
observations were reproducible after additional surface cleaning in humid He. We therefore
suppose that under the present reaction conditions a carbon pool is reproducibly built up during
the first reaction run on the Ti/SBA-15 material, which cannot be removed by evacuation. This
carbon pool can then participate in the product formation at the beginning of the second run.
As a consequence, no time for the accumulation of intermediates is required in the second run,
product formation starts earlier and the total yield is higher. It is assumed that saturation of
the catalyst surface with these less reactive carbon species is required to achieve high hydrocarbon yields. On the other hand, in the presence of Au, this carbon accumulation does not
seem to occur and is either removed by evacuation or these carbon species are not reactive in
subsequent reaction runs. Consequently, the yields observed in the second run are similar to
the first ones. It is reasonable to assume that a higher hydrogenation rate in the presence of
Au leads to a carbon-free surface after the first reaction run. In general, noble metals usually
5.2 Photocatalytic CO2 reduction using Au-modified TiOx/SBA-15 materials
59
show high activity in hydrogenation reactions. [139,140] In photocatalytic reactions it is accepted
that H2 formation occurs at these electron trapping sites, and more hydrogen is available in
close proximity to gold facilitating hydrogenation. [37,38] It should be noted that this is in agreement with the higher yield of H2 that has been observed during photo-deposition experiments.
Visible light-assisted photoreactions in the presence of Au can be excluded as no product formation was observed when the samples were irradiated with Vis light. The 30 ppm of CH4
produced within 5 h of irradiation in the presence of Au/Ti/SBA-15 is equivalent to 0.047 mol
resulting in 0.188 mol/gcat/h or 9.4 mol/gTi/h. For Ti/SBA-15 the yield in the second run can
be calculated to amount to 12.1 mol/gTi/h. The comparison illustrates that on the basis of one
gram of catalyst the Au/Ti/SBA-15 is roughly 7 times more active than the catalysts studied
by Mul et al. [73] However, comparing the yields on a basis of g-1 Ti Mul et al. [73] observed
five times higher methane yields as only 0.05 wt% Ti were incorporated into the SBA-15. The
applied reaction conditions differ considerably rendering the comparison difficult. Reactions
with stoichiometric CO2 /H2 O mixtures are in progress.
To get further insight in the carbon pool formation, two consecutive reaction cycles were performed followed by irradiation of the photocatalysts in humid He. Surprisingly, on both catalysts evolution of hydrocarbons was observed under irradiation in humid He (Fig. 5.11a). A
steep increase in methane and longer-chain hydrocarbon formation was observed after 1 h of
irradiation for Ti/SBA-15. This is in line with the proposed formation of a carbon pool. For
the Au/Ti/SBA-15 sample a slower formation of Cx Hy is characteristic. The formation of hydrocarbons during irradiation in humid He, however, is a clear indication that a carbon pool is
also created on Au/Ti/SBA-15, but its effects on hydrocarbon formation are less pronounced.
Possibly, the nature or location of these carbon species is different.
Finally, the influence of the water content on the photocatalytic CO2 reduction reaction was
investigated using a reaction gas mixture containing 5 % CO2 /He. Consequently, only preadsorbed water in the pores of the catalysts was available as reducing agent. The results of these
reaction conditions with only surface-bound H2 O are shown in Fig. 5.11b. Prior and subsequent
to these experiments CO2 reduction was performed using the reaction gas mixture containing
gas-phase water. These results are also included in Fig. 5.11b. Unexpectedly, less longer-chain
hydrocarbons are formed in the absence of gas-phase water on both catalysts systems. The
CH4 yield is significantly higher for both catalysts. It should be noted that the carbon pool
was already formed on the Ti/SBA-15 surface during a previously performed experiment with
gas-phase water. Considering the overall produced carbon content (CH4 , C2 H6 , C3 H8 , C4 H10 )
the carbon yield (C*) increased by about 20 % in the absence of water, which was mainly due
to the higher CH4 yields compensating the decrease in longer-chain hydrocarbon yields. The
overall hydrocarbon yield was even increased in the consecutive experiment performed with the
CO2 /H2 O mixture. For Ti/SBA-15 the CH4 yield decreased, but the C2 H6 yield was drastically
higher (Fig. 5.11b). A higher ethane yield and a higher methane yield were observed for the
Au/Ti/SBA-15 sample, which might indicate that a carbon pool was formed on the surface of
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
60
70
a)
CO
2
60
45
Ti/SBA-15
+ H O
2
40
Au/Ti/SBA-15
H O
2
25
1h
5h
7h
Au/Ti/SBA-15
0h
time
~43 ppm C*
7h
Ti/SBA-15
5h
~38 ppm C*
1h
Au/Ti/SBA-15
10
Ti/SBA-15
15
~34 ppm C*
20
~36 ppm C*
yield CH
30
5
0h
6
4
Au/Ti/SBA-15
10
2
CH
Ti/SBA-15
20
10
8
C H
~39 ppm C*
30
0
4
3
~29 ppm C*
yield [ppm]
40
C H
C H
35
50
4
[ppm]
evacuation
b)
0
CO
2
+ H O
2
CO
2
CO
2
+ H O
2
Figure 5.11: a) CH4 yield during irradiation of the photocatalysts in the reaction gas mixture and
CH4 yield during irradiation in humid He subsequently performed. Evolution of CH4
during irradiation in humid He is due to remaining carbon species on the surface.
b) Hydrocarbon yields in photocatalytic CO2 reduction experiments obtained with
varying gas-phase water contents in the reactant feed after 5 h for the Ti/SBA-15
and Au/Ti/SBA-15 samples. The yield of the four different hydrocarbon species is
corrected for possible carbon contaminations left on the catalyst.
the Au/Ti/SBA-15 sample. Furthermore, these experiments show that the water content on the
sample surface is sufficient to perform photocatalytic CO2 reduction for at least five hours of
irradiation with both photocatalysts. In contrast to these observations Anpo et al. [58] observed
the formation of larger quantities of CH4 for high H2 O/CO2 ratios, and only small quantities
of CH4 were produced in the absence of gas-phase water. However, it is well known that
water strongly binds to isolated Ti-species and high temperatures are required to remove water
effectively. [80] Therefore, the surface of the photocatalysts should be saturated with H2 O and,
as a result, CO2 reduction is feasible. In addition, high methane yields in water-poor conditions
are in agreement with a C2 H6 /CH4 ratio exceeding 1 for stoichiometric CO2 /H2 O mixtures
observed by Mul et al. [73] , and increasing ethane yield for higher water content reaction gasmixtures as observed by Anpo et al. [58] Further studies are clearly required to fully explain
these observations. However, a hypothesis which should be considered is the inhibition of a
subsequent photo-reforming reaction of the produced CH4 occurring in the presence of gasphase H2 O. CO and H2 may be produced due to this process, which then may be involved
in the formation of higher hydrocarbons. In terms of irradiation in CO2 , this will diminish
the formation of higher hydrocarbons. Furthermore, an influence of the carbon pool in higher
hydrocarbon production was observed. Thus, it should be considered that high water contents
are required to hydrogenate these species. Furthermore, the water content may influence the
mobility of intermediates on the catalyst surface.
After reaction all samples were characterized by DRIFTS, UV-Vis, and XPS and compared to
the unreacted materials to get further insight in possible intermediates, the deactivation of the
5.2 Photocatalytic CO2 reduction using Au-modified TiOx/SBA-15 materials
61
catalysts, and the carbon pool, which was built up during CO2 reduction. The carbon content
as determined by XPS analysis was similar for all the samples. Structural changes of the Tispecies within the course of gas-phase photocatalytic CO2 reduction causing deactivation of
the materials can be excluded comparing Ti/SBA-15 before and after reaction or Au/Ti/SBA15 before and after reaction (XPS results, Fig. A7). UV-Vis indicates a narrowing of the Au
plasmon for Au/Ti/SBA-15 after reaction, which may be caused by different Ti-Au interactions
(Fig. A8). One has to keep in mind that TiOx -species have the potential to alter the local
environment of the Au particles. [135,136] However, there is no clear evidence for a loss of TiOx
or the polymerization of the TiOx species, which could explain the deactivation of the Ti/SBA15 and the Au/Ti/SBA-15 during prolonged irradiation.
On the other hand, structural changes of the Ti-species caused by the photo-deposition of Au
were detected by UV-Vis and XPS (Fig. A8 and Fig. A9). The Ti 2p signal of the Au/Ti/SBA15 is shifted to lower binding energies compared to the Ti/SBA-15 sample, and two different
Ti-species were needed to obtain a reasonable fit (Fig. A9). This additional signal was in good
agreement with an additional shoulder observed in the O 1s region of the Au/Ti/SBA-15. From
UV-Vis spectroscopy the absorption onsets of Ti/SBA-15 and Au/Ti/SBA-15 were similar (Fig.
A8). However, heating of the sample revealed that there is no increase of the absorption edge
energy for Au/Ti/SBA-15 due to the loss of water molecules bound to the Ti-species, which
is usually observed after a drying of the samples. [134] These observations clearly point to a
change in the coordination of the Ti-species. Most likely the changes of the Ti-species occur
in close proximity to the Au nanoparticles. It can be assumed that the combination of Ti and
Au in close proximity, rather than the presence of Au alone, might be necessary to obtain
the higher hydrogenation activities observed during CO2 reduction. However, based on the
presented results a distinction of the different species is not feasible and further studies are
required to fully describe the changes of the Ti-species upon Au photo-deposition. A HRTEM
study combined with photocatalytic test reactions was performed, which is presented in the
next section. Ti/SBA-15 materials containing different TiOx loadings are used to investigate
structural changes and to relate them with the activities.
Difference IR spectra of the reacted and the fresh Ti/SBA-15 revealed that formaldehyde and
paraformaldehyde were accumulated at the surface of the Ti/SBA-15 sample after CO2 reduction reaction (Fig. 5.12). The bands observed at 1685 and 2853 cm-1 are in good agreement
with values of the ν (C=O) and ν (CH2 ) stretching modes of formaldehyde adsorbed at different surfaces. [141–144] The intense band at 2972 cm-1 is representative for the asymmetric
stretching (ν a (CH2 )) of paraformaldehyde, which is known to be obtained by the polymerization of formaldehyde at low temperatures (Table 5.3). [141,143,144] The assignment of the bands
is clearly verified by the thermal stability of paraformaldehyde. Interestingly, less formaldehyde was detected for the Au/Ti/SBA-15 sample and the signals assigned to paraformaldehyde
are shifted to higher wavenumbers (Fig. A10). Assuming that paraformaldehyde is the intermediate of higher hydrocarbons it is likely that it is immediately available at the surface of
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
62
2972
2929
2910
2853
1877
1685
1602
log (1/R)
400°C treated
b)
a)
120°C treated
3050 3000 2950 2900 2850
2000 1900 1800 1700 1600 1500
-1
wavenumbers [cm ]
Figure 5.12: Difference DRIFT spectra of the Ti/SBA-15 before and after photocatalytic CO2 reduction obtained after heating to 120 ◦ C and 400 ◦ C. Difference spectra were obtained
as follows: For spectrum a) the spectrum of Ti/SBA-15 (as-prepared) was subtracted
from Ti/SBA-15 after photocatalytic CO2 reduction. Both spectra were measured after drying at 120 ◦ C. Spectrum b) was derived by subtracting the spectra obtained after
drying at 400 ◦ C.
the Au/Ti/SBA-15 sample and higher hydrocarbon formation can be produced, whereas at the
Ti/SBA-15 sample paraformaldehyde is formed upon accumulation of formaldehyde at the surface. This might explain the inhibition of ethane production on Ti/SBA-15, as observed during
the 1st reaction run.
In the literature different reaction mechanisms are proposed for photocatalytic CO2 reduction.
Anpo et al. [67] propose a mechanism in which CO2 and H2 O decompose competitively into C
and OH radicals at the excited Ti-O centers, presumably with CO as an intermediate species.
Similar observations were recently reported for crystalline TiO2 by Zapol et al. [69] Mul et al. [73]
followed the proposed route by Frei et al. [71] who suggested a mechanism involving CO in the
initial state and formaldehyde, which is converted to CH4 and longer-chain hydrocarbons in
the presence of water. A Ti-hydroperoxid species is assumed to be essential in this process.
Due to the results obtained within this study the mechanism proposed by Mul et al. [73] and Frei
et al. [71] appears to be more likely as 1) formaldehyde and paraformaldehyde were observed
by diffuse reflectance IR measurements, 2) higher CH4 yields are observed for water-poor
reaction conditions, and 3) CO2 is weakly interacting with isolated Ti-species, [134] whereas
water is strongly bound to the surface.
5.2 Photocatalytic CO2 reduction using Au-modified TiOx/SBA-15 materials
63
Table 5.3: Wavenumbers (cm-1 ) and mode assignments of formaldehyde and paraformaldehyde on
different substrates reported in literature.
Mode Assignment
H2 CO [144]
(H2 CO)n [144] H2 CO [141]
(H2 CO)n [141] H2 CO [143]
(H2 CO)n [143]
ν a (CH2 )
ν s (CH2 )
ν (C=O)
2884
2820
1727
2982
2917
-
2928
2928
-
2980
2928
2928
1694
2894
2830
1717
-
5.2.3 Conclusions
In summary, Ti/SBA-15 and Au nanoparticles embedded into Ti/SBA-15 were studied in photocatalytic CO2 reduction. Both materials were active in photocatalytic CO2 reduction forming CH4 . However, an increase in the hydrogenation ability of the Au-containing catalyst
was found, whereas an active carbon pool formation was observed on Ti/SBA-15. Unlike the
less reactive carbon pool formed on Au/Ti/SBA-15 the carbon pool on Ti/SBA-15 can participate in subsequent reactions. Experiments with varying water content indicate that the
carbon chain growth is significantly affected by the gas-phase water content, and formaldehyde/paraformaldehyde were identified as reaction intermediates.
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
64
5.3 Effects of Au and ZnO on the structure and
photocatalytic activity of TiOx/SBA-15 materials
Abstract3
Gold nanoparticles can be efficiently photo-deposited on titanate-loaded SBA-15, and the formation of hydrogen is observed during the photo-deposition process. Additionally, the presence
of the gold nanoparticles greatly enhances the performance of Ti(x)/SBA-15 in the photocatalytic hydroxylation of terephthalic acid. HRTEM studies show that larger Au nanoparticles
are photo-deposited at the outer surface of Ti(x)/SBA-15 materials and that TiOx tends to form
agglomerates in close proximity to the Au nanoparticles. The activity of Au/Ti(x)/SBA-15 materials was found to be dependent on the TiOx loading. Photocatalytic tests indicated that the
amount of initially isolated TiO4 tetrahedra determines the activity of the Au/Ti(x)/SBA-15,
and Au/Ti(x)/SBA-15 catalysts in which the TiOx species are already clustered showed lower
photocatalytic activities. When isolated zinc oxide (ZnOx ) species are present on TiOx/SBA15, the photo-deposited gold nanoparticles are smaller, well dispersed within the pore structure
of SBA-15, and agglomeration of TiOx-species can be neglected. The dispersion of Au in the
SBA-15 matrix obtained due to the photo-deposition procedure seems to depend on the mobility
of the TiOx -species, which is mainly determined by the agglomeration of TiOx . Effective hydrogen evolution requires Au/TiOx assemblies as in Au/Ti(x)/SBA-15, whereas hydroxylation
of terephthalic acid can also be performed with Au/ZnOx /TiOx/SBA-15 materials. However,
isolated TiOx species have to be grafted onto the support prior to the zinc oxide species. In
conclusion, the results are strong evidence for the necessary presence Ti-O-Si bridges for high
photocatalytic activity in terephthalic acid hydroxylation.
3
The results discussed in this section were obtained in a cooperative project with the Chair of Inorganic Chemistry II, Ruhr-University Bochum. For TEM measurements Christian Wiktor and Stuart Turner are gratefully
acknowledged. Furthermore, the support from the European Union under the Framework 7 program under a
contract from an Integrated Infrastructure Initiative (Reference 262348 ESMI) is acknowledged. To be published as ”Effects of Au and ZnO on the structure and photocatalytic activity of TiOx /SBA-15 materials”.
5.3 Effects of Au and ZnO on the structure and photocatalytic activity of TiOx /SBA-15
materials
65
5.3.1 Short Introduction
In the last decades highly dispersed Au nanoparticles on metal oxides have become of major
interest for several reactions such as CO oxidation, selective oxidation of alcohols or methanol
synthesis. [139,145–147] Recently, the potential of Au nanoparticles in photocatalysis has received
considerable attention. [37,38] Within several studies improved photocatalytic performance due
to visible light absorption of the Au plasmon and due to good electron storage capacities
are reported. [148,149] One of the most studied composites in this regard are Au/TiO2 materials, which exhibit catalytic activity in organic pollutant degradation and hydrogen evolution
due to photocatalytic reforming of alcohols. Usually the utilized composites are obtained by
deposition-precipitation, colloidal synthesis, or by photo-deposition of Au nanoparticles. [37,138]
Even though bulk TiO2 and Au/TiO2 are versatile materials, single-site titania catalysts are
widely studied particularly in photocatalytic CO2 reduction, which was already shown by Anpo
et al. [58] Mori et al. [131,132] showed that the deposition of Au nanoparticles using photo-excited
Ti-containing zeolites is feasible. We demonstrated that Au/Ti/SBA composite materials prepared by a photo-deposition procedure of Au were highly active in photocatalytic CO2 reduction, presumably due to a higher hydrogenation activity of the composite material in presence
of Au nanoparticles. However, the UV-Vis spectroscopic results and the XPS characterization showed that the Ti coordination of the isolated TiO4 tetrahedra changed upon Au photodeposition. Two different TiOx species were observed by XPS and UV-Vis spectroscopy results
inidcated that less water interacts with the Au/Ti/SBA material. However, the interaction between the TiOx species and photo-deposited Au is still unclear and the origin of the effect of
Au on the photocatalytic activity of the Ti/SBA material is not yet solved.
Recently, we have shown that using a reproducible grafting method of Ti(Oi Pr)4 high loadings of isolated or slightly agglomerated TiOx species can be obtained. [134] Furthermore, we
were able to introduce ZnOx species into the materials by grafting of Zn(acac)2 . [134] Crucial
properties of the photocatalysts, such as light absorption, CO2 adsorption, basic and acidic
sites on the surface of the different materials were studied. [134] It was shown that ZnOx species
significantly enhance the CO2 adsorption properties, whereas the light absorption is mainly
depending on the HOMO-LUMO excitation of the TiOx species. XPS and EXAFS showed
that preferably Ti-O-Zn bonds are formed when ZnOx is grafted first. [134] In this contribution in-depth characterization of photo-deposited Au nanoparticles was performed depending
on the TiOx loading in Ti(x)/SBA samples. The activity of the as-received and Au-modified
materials was determined by hydrogen evolution reaction due to photo-reforming of methanol
and the hydroxylation of terephthalic acid. Thus, the electron and the hole availability upon
photo-excitation were assessed. Furthermore, the effect of ZnOx species was determined.
It is shown that the photo-deposition of Au nanoparticles enhances the photocatalytic properties
of the materials regarding photocatalytic hydrogen production and hydroxylation of organic
compounds. Agglomeration of TiOx species occurring during the diffusion-controlled photo-
66
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
deposition of Au is independent of the initial TiOx loading, but the agglomeration of TiOx in
the presence of ZnOx is clearly inhibited and Au is photo-deposited homogeneously throughout
the whole material. Furthermore, our results provide evidence that Ti-O-Si bonds are essential
for photocatalytic activities in the terephthalic acid hydroxylation reaction.
Experimental
Sample synthesis
The synthesis of SBA-15 and the grafting procedure for obtaining isolated and slightly agglomerated TiOx and ZnOx species on the SBA-15 surface have been described in detail in the
previous section. [134] In brief, Ti(Oi Pr)4 (99.999 %, Sigma-Aldrich) or Zn(acac)2 (98 %, Alfa
Aesar, purified by means of sublimation) were dissolved in dry toluene and contacted with the
dried SBA-15 support at room temperature. After 4 h of stirring, the solution was removed by
sedimentation or centrifugation. Samples were washed three times with dry toluene, dried in
vacuum and calcined. Coverages up to 1 Ti nm−2 or 1 Zn nm−2 can be obtained in one grafting
step, while for higher coverages the grafting procedure had to be repeated.
Characterization
Characterization of the Au-modified materials was performed by elemental analysis, UV-Vis
spectroscopy, and Transmission Electron Microscopy (TEM). In-depth characterization of the
different as-received TiOx- and ZnOx -grafted SBA-15 supports by means of elemental analysis,
nitrogen physisorption, UV-Vis spectroscopy, X-ray photoelectron spectroscopy (XPS), and Xray absorption spectroscopy (XAS) is described in the previous section. [134]
UV-Vis diffuse reflectance spectra (DRS) were recorded in a Perkin Elmer Lambda 650 UVVis spectrometer equipped with a Praying-Mantis mirror construction using MgO as the 100 %
reflection reference.
XPS was performed in a UHV set-up equipped with a Gammadata-Scienta SES 2002 analyzer.
The base pressure in the measurement chamber was 5 x 10−10 mbar. Monochromatic Al Kα
(1486.6 eV; 14.5 kV; 30 mA) was used as incident radiation, and a pass energy of 200 eV was
chosen resulting in an effective instrument resolution higher than 0.6 eV. Charging effects were
compensated using a flood gun, and binding energies were calibrated based on positioning the
main C 1s peak at 285 eV, which originates from carbon contaminations. Prior to the measurements the samples were dehydrated at 200 ◦ C under ultra-high vacuum (UHV) conditions.
Measured data were fitted using Shirley-type backgrounds and a combination of GaussianLorentzian functions with the CasaXPS software. The Ti/Si atomic concentration ratios were
obtained by determining the integral area of the Gaussian-Lorentzian functions and correcting
the values by the specific atomic sensitivity factors proposed by Wagner. [110]
Bright field TEM, high resolution TEM (HRTEM), electron energy-loss spectroscopy (EELS),
and energy-filtered TEM (EFTEM) were conducted on a Philips CM30 equipped with a Schottky field emission gun and a post-column GIF 200 energy filter, operated at 300 kV. High-angle
5.3 Effects of Au and ZnO on the structure and photocatalytic activity of TiOx /SBA-15
materials
67
annular dark-field scanning transmission electron microscopy (HAADF-STEM), spatially resolved energy-dispersive X-ray spectroscopy (STEM-EDX), and the electron tomography series were acquired on a FEI Tecnai G2 equipped with a Schottky field emission gun operated
at 200 kV and FEI Titan ”cubed” microscope equipped with a probe corrector, operated at 300
kV. EDX spectra were recorded on Philips CM20 microscope equipped with a LaB6 filament
operated at 200 kV.
Photocatalytic measurements
Photo-deposition of Au was performed in a continuous flow stirred-tank reactor equipped with
a 700 W Hg immersion lamp cooled by water circulating in a double wall jacket. [133] During
photo-deposition the concentration of CO2 , O2 , and H2 were analyzed with a non-dispersive
IR photometer, a paramagnetic and a thermal conductivity detector (XStream, Emerson Process Management), respectively. 350 mg of catalyst were dispersed in a solution of 550 ml
dest. H2 O, 50 ml MeOH, and 3.5 ml of a 1.5 x 10−3 molar auric acid solution (HAuCl4 99.9
%, Sigma Aldrich). The reactor was deaerated for 60 min using pure N2 prior to irradiation.
Photo-deposition was performed at 30 ◦ C with 50 % irradiation power (350 W) for 2.5 h. After
successful photo-deposition the materials were filtered and freeze-dried over night. Terephthalic acid (TA) hydroxylation was performed to determine the amount of OH radicals generated during irradiation. 150 mg of catalyst was ultra-sonicated for 5 min in 500 ml of a 0.01
M NaOH solution containing 3 mM TA. [150] The suspension was stirred for additional 30 min
in the dark before illumination was started by means of a 150 W Xe immersion lamp cooled
by water circulating in a double wall jacket. Samples were taken every 15 min, filtered with a
Filtropur S and the fluorescence spectra were measured with a double monochromatic fluorescence spectrometer (Fluorolog FL3-22, HORIBAJobin Scientific). The emission wavelength
was set to 320 nm and the characteristic fluorescence at 426 nm which is directly correlated
with the OH radicals generated was measured.
5.3.2 Results and Discussion
Sample structure of the as-received materials
Characterization of the Ti(x)/SBA and TiOx- and ZnOx -grafted SBA-15 support materials is
described in the previous section. [134] In short, UV-Vis, X-ray photoelectron (XPS), and X-ray
absorption spectroscopy (XAS) proved that in dry samples TiO4 species in a tetrahedral coordination sphere were obtained within the SBA-15 material at low Ti loadings of 2.1 wt%.
With increasing titania loading the species start to polymerize, however, for Ti loadings up to
7 wt%, corresponding to about 1 Ti nm-2 , most of the TiOx species are still isolated. For all
materials a homogeneous titania distribution was obtained as evidenced by XPS and EFTEM.
A wavelength of less than 300 nm is required to excite electrons across the HOMO-LUMO gap
in case of fully isolated TiOx species. For agglomerated TiOx at a loading of 2.7 Ti nm-2, the re-
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
5000
a)
CO
H
4000
0.45wt%
4000
0.55wt%
0.35wt%
2
Concentration [ppm]
Concentration [ppm]
68
2
O
2
0.25wt%
0.05wt% Au
3000
0.15wt%
2000
1000
0
0
100
200
300
400
500
600
700
b)
Ti0.3/SBA
Ti1.0/SBA
Ti2.0/SBA
3000
Ti2.7/SBA
2000
1000
800
Time [min]
0
0
30
60
90
120
150
180
Time [min]
Figure 5.13: a) The H2 evolution during the stepwise photo-deposition of Au nanoparticles on
Ti2.0/SBA is shown and the overall amount of deposited Au is indicated. b) H2 evolution measured during the photo-deposition process of Au (0.25 wt%) on Ti(x)/SBA
materials with increasing Ti loading.
quired wavelength shifts to 320 nm. For the TiOx - and ZnOx -grafted SBA-15 samples, namely
Ti1.2/Zn0.3/SBA and Zn0.3/Ti1.2/SBA, strong interactions between TiOx - and ZnOx were detected. The TiOx species were shown to be modified by the order of the grafting steps, and more
isolated TiOx was obtained when ZnOx was grafted subsequent to TiOx as in Zn0.3/Ti1.2/SBA.
Photocatalytic hydrogen evolution in presence of Au
It is well known that Au can act as a co-catalyst in photocatalysis. [37,38] In general metal cocatalysts are assumed to act as an electron or hole sink enhancing the electron-hole lifetime
in photocatalysis. [37,38] Furthermore, the photo-deposition procedure is reported to be a suitable technique to deposit Au or other metal co-catalysts on the surface of a photoactive material. [131,132] Thus, Au was photo-deposited using methanol as sacrificial agent. [148] During
photo-deposition, the hydrogen evolved due to sacrificial water splitting was measured online. [133] For all measurements it was confirmed that H2 evolution occurs only in a dispersion
of Ti(x)/SBA-15 samples in auric acid solution under irradiation, whereas no H2 was detected
in absence of any of the components. Bowker et al. [148] as well as other groups reported that
there is a maximum amount of co-catalyst which is beneficial for photocatalytic reactions. In
these studies they pointed out that too high loadings might block active sites which would, at a
certain loading, overwhelm the otherwise positive effect of the co-catalyst. Therefore, the maximum amount of deposited gold was evaluated using a stepwise photo-deposition procedure,
following the amount of H2 evolved during photo-deposition as previously reported by Busser
et al. [133]
For low gold loadings of 0.05 wt% deposited on Ti2.0/SBA the evolution of hydrogen slowly
increases within 60 min of irradiation time (Fig. 5.13a). Afterwards, the Au concentration was
increased stepwise and during each deposition interval irradiation was performed for ∼ 60 min.
The slow increase in hydrogen evolution was observed for Au loadings of up to 0.35 wt%, at
5.3 Effects of Au and ZnO on the structure and photocatalytic activity of TiOx /SBA-15
materials
69
which the hydrogen evolution stabilizes, and the maximum hydrogen evolution was observed
for a gold content between 0.45 and 0.55 wt% Au for the Ti2.0/SBA material (Fig. 5.13a). The
same trend in hydrogen evolution with increasing Au loading was observed for Ti1.0/SBA.
Therefore, the maximum Au loading for the Ti(x)/SBA materials was estimated to be 0.45 wt%
for the two materials.
In a second set of experiments the amount of Au in the solution was fixed to 0.25 wt% of
Au with respect to the amount of sample dispersed in the solution. These experiments were
performed in order to study the effect of the different Ti loadings as well as the effect of Zn.
The amount of Au was chosen to be lower than the maximum beneficial Au loading obtained
during the stepwise deposition procedure to avoid blocking of active sites for samples grafted
with lower TiOx loadings. The evolution of hydrogen as a function of irradiation time during the
in-situ photo-deposition experiments is shown in Fig. 5.13b for different Ti(x)/SBA materials.
From these results it can be clearly seen that the measurable amount of H2 in the gas-phase
significantly increases with the Ti loading up to 7 wt% as in Ti1.0/SBA, while it decreases
afterwards with higher Ti loadings as in case of Ti2.0/SBA and Ti2.7/SBA. This trend is a first
indication for different activities of the isolated and polymerized TiOx species. The higher
loaded Ti1.0/SBA is more active compared to the Ti0.3/SBA, however, in both cases Ti is
mostly present as isolated sites that have a tetrahedral coordination sphere in the dry state.
Those species should be entirely surrounded by Ti-O-Si bonds and are able to coordinate two
water molecules, upon which they change into an octahedral configuration. [80] For the higher
loaded Ti1.0/SBA more of these isolated species can contribute to the hydrogen generation.
Increasing the TiOx loadings the evolved hydrogen drops off due to polymerization of the TiOx
species. Therefore, less isolated TiO4 is present on the SBA-15 support that can contribute to
the hydrogen generation. The beneficial effect of Au supported on different semiconductors in
photocatalytic H2 production or photocatalytic dye degradation has already been demonstrated
by several research groups. [37,38,149] It is supposed that Au-containing materials show enhanced
dye degradation rates and a significant increase in H2 productivity due to efficient electron
storage on the Au nanoparticles resulting in a decrease in recombination rates and due to the
ability of the Au nanoparticles to act as an electron transfer site. [38,149] An influence of the Au
loading on the H2 production rate has been reported recently. [38,148] In the present study the
influence of Au loading on the hydrogen evolution rate was confirmed for the Ti(x)/SBA-15
materials, and it is reasonable to assume that H2 evolution measured during photo-deposition of
gold onto Ti(x)/SBA materials was due to water splitting in presence of methanol as sacrificial
agent. Furthermore, at a fixed Au loading the obtained results indicate an effect of the TiOx
coordination and agglomeration on the hydrogen evolution ability.
For the Zn0.3/Ti1.2/SBA and Ti1.2/Zn0.3/SBA catalysts no hydrogen evolution was observed
during deposition of 0.25 wt% Au. However, the Zn0.3/Ti1.2/SBA and the Ti1.2/Zn0.3/SBA
material became bluish after the experiments, which was also observed for the Ti(x)/SBA samples. Furthermore, it should be mentioned that within the detection limit also no hydrogen
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
70
2.5
2.5
5
Rel. intensity x 10
5
Rel. intensity x 10
f)
B)
A)
f)
2.0
1.5
e)
1.0
d)
c)
0.5
b)
0
20
40
60
Time [min]
80
1.5
1.0
d)
c)
0.5
b)
a)
a)
0.0
e)
2.0
0.0
100
0
20
40
60
80
100
Time [min]
Figure 5.14: A) Time-dependent terephthalic acid hydroxylation reaction for a) Ti0.3SBA,
b) Au/SBA-15, c) Au/Ti2.7/SBA, d) Au/Ti0.3/SBA, e) Au/Ti2.0/SBA, and f)
Au/Ti1.0/SBA. B) Time-dependent terephthalic acid hydroxylation reaction for a)
Zn0.3/Ti1.2/SBA, b) Au/SBA-15, c) Au/Zn0.3/SBA, d) Au/Ti1.2/Zn0.3/SBA, e)
Au/Ti1.0/SBA, and f) Au/Zn0.3/Ti1.2/SBA.
evolution was observed for the Zn0.3/SBA sample or the bare SBA-15 support and the samples
remain white after the photo-deposition experiments.
Measurements of the photocatalytic H2 evolution mainly probed the availability of photogenerated electrons upon light irradiation. To probe the availability of photo-generated holes
for photocatalytic reactions the terephthalic acid (TA) hydroxylation was used as a simple test
reaction. It is known that the OH radical, which is formed upon reaction of surface adsorbed
water or hydroxyl groups, can react with terephthalic acid (TA) to generate 2-hydroxy terephthalic acid (TAOH), which, in its terephthalate form, emits unique fluorescence at around 425
nm. [151,152] Firstly, the as-prepared Ti(x)/SBA and TiOx - and ZnOx -grafted SBA-15 materials
were tested. The TA test revealed that the as-made materials, regardless of the coverage of
titanate species or the presence of ZnOx , do not show significant photocatalytic activity under
the experimental conditions for the hydroxylation of TA, which result in a strong increase in
the characteristic TAOH fluorescence band. This was expected as a Xe lamp was used, which
does not emit intense UV radiation to excite the isolated or polymerized TiOx species in the
Ti(x)/SBA and TiOx- and ZnOx -grafted SBA-15 materials. The two possible reasons for the
low activity may be the low intensity of deep UV light and a fast electron-hole recombination.
The influence of the latter should be diminished using the Au-modified Ti(x)/SBA and TiOxand ZnOx -grafted SBA-15 samples obtained after photo-deposition experiments.
The TA hydroxylation test reaction was repeated with the Au/Ti(x)/SBA, Au/Zn0.3/Ti1.2/SBA,
and Au/Ti1.2/Zn0.3/SBA samples. The obtained results are shown in Fig. 5.14a and Fig. 5.14b,
respectively. The TA test results indicate that the Au/Ti(x)/SBA-15 materials are able to generate OH radicals, which can easily react with terephthalic acid to form TAOH, whereas Au/SBA15 is nearly inactive. Furthermore, the almost linear relationship between irradiation time and
5.3 Effects of Au and ZnO on the structure and photocatalytic activity of TiOx /SBA-15
materials
71
fluorescence signal intensity indicates that Au/Ti(x)/SBA-15 materials are stable under these
reaction conditions. A small formation of TAOH in the case of Au/SBA-15 may be related to
surface hydroxyl groups of the SBA-15 support participating in the reaction. It has to be noted
that in a blank experiment without the addition of a catalyst or SBA-15 substrate material no
reaction was observed. However, all Au/Ti(x)/SBA catalysts exhibit higher fluorescence signals of TAOH compared to Au/SBA-15. Within the Au/Ti(x)/SBA materials, Au/Ti1.0/SBA
shows the highest production rate of TAOH. The higher loaded samples Au/Ti2.0/SBA and
Au/Ti2.7/SBA and the Au/Ti0.3/SBA sample with the low Ti loading are clearly less active.
These samples contain many isolated TiO4 species as shown by XAS and XPS. [134] Higher
loaded samples contain mainly polymerized TiOx chains, which appear to be less active. It
is striking that the activity of Au/Ti0.3/SBA is as low as the activity of Au/Ti2.7/SBA, even
though it was shown that only isolated TiOx species are present. This observation may be attributed to the larger absorption edge energy in case of Ti0.3/SBA or the lower amount of active
TiOx centers. However, the observed trend in the TA test reaction is in good agreement with the
hydrogen evolution rates obtained during Au photo-deposition. Thus, the results strongly suggest that isolated TiOx species are required to achieve high activities in the TA hydroxylation
reaction and the photocatalytic reforming of methanol.
The time-dependent evolution of hydroxylated TA for all Zn-containing SBA-15 samples is
shown in Fig. 5.14b. Au/Ti1.0/SBA is shown for comparison. While Au/Zn0.3/Ti1.2/SBA
is at least as active as Au/Ti1.0/SBA, the other two Zn-containing samples, especially
Au/Zn0.3/SBA, are significantly less active in the TA hydroxylation test reaction. These results indicate that TiOx species rather than ZnOx species bound to the silica surface are the
photocatalytically active species. Furthermore, these results demonstrate that the order of the
grafting steps clearly influences the photocatalytic activity: Au/Ti1.2/Zn0.3/SBA is clearly less
active than Au/Zn0.3/Ti1.2/SBA. In the case of Au/Ti1.2/Zn0.3/SBA TiOx is grafted onto a
Zn0.3/SBA, thus it is likely that Ti-O-Zn bonds are formed rather than Ti-O-Si bonds. Indeed, the characterization results based on X-ray photoelectron spectroscopy and X-ray absorption spectroscopy indicate an intense interaction between Zn and Ti in samples prepared
with a similar composition as the Au/Ti1.2/Zn0.3/SBA material. [134] In the opposite case as
in Au/Zn0.3/Ti1.2/SBA, where ZnOx is grafted after TiOx grafting is completed, the photocatalytic activity in TA hydroxylation is still high. For this material it can be assumed that Ti-O-Si
bonds are formed during the first grafting step and that the formation of Zn-O-Ti bonds due
to the post grafting of ZnOx is negligible. In summary, a structure-activity correlation was
observed for the ZnOx -containing materials, as the order of the grafting steps significantly influences the photocatalytic activity especially in the TA hydroxylation test reaction. Ti-O-Si
bonds rather than Ti-O-Zn bonds are required to run the catalytic reaction. Additionally, it was
shown that ZnOx species are not able to perform the photocatalytic TA hydroxylation reaction,
which is in agreement with the poor absorption properties. [134]
By comparing the photocatalytic activity towards hydroxylation of TA of all the as-prepared
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
72
materials and all the Au-modified samples it can be clearly seen that Au is able to enhance the
photocatalytic activity of the Ti(x)/SBA materials and the ZnOx /TiOx/SBA samples. Presumably, Au is acting as a co-catalyst in these cases. Based on these results it is not clear whether
Au is acting as an electron sink or additionally enables the materials to be activated by visible light due to the Au plasmon absorption. Recent results for photocatalytic CO2 reduction
indicate that visible light-driven CO2 reduction by excitation of the Au plasmon was not feasible for the Au/Ti03/SBA catalyst, and thus Au is mainly acting as an electron sink in these
samples (Chapter 5.2). However, it is striking that even in the presence of Au photocatalytic
reforming of methanol was obviously not possible for Au/Zn0.3/Ti1.2/SBA as no hydrogen
was evolved. On the other hand, Au/Zn0.3/Ti1.2/SBA exhibited similar photocatalytic activity
towards TA hydroxylation as the most active Au/Ti1.0/SBA sample. In order to elucidate this
phenomena on the different materials were characterized regarding the nature of the Au particles. Characterization was performed by means of ICP-OES, UV-Vis spectroscopy, XPS, and
TEM analysis.
Characterization of Au particles
Table 5.4: Au content after stepwise and photo-deposition as determined by ICP-OES measurements for Au/Ti(x)/SBA-15, Au/Zn0.3/Ti1.2/SBA, and Au/Ti1.2/Zn0.3/SBA materials.
Sample
Au/Ti0.3/SBA Au/Ti1.0/SBA Au/Ti2.0/SBA
Au/Ti2.7/SBA
Au [wt%]
0.25
0.29/0.54
0.28
Sample
Au/SBA-15
Au/Zn0.3/SBA Au/Ti1.2/Zn0.3/SBA Au/Zn0.3/Ti1.2/SBA
Au [wt%]
0.02
0.05
0.27/0.55
0.1
0.14
ICP-OES results indicate that the amount of deposited Au is in good agreement with the loading
of 0.25 wt% and the 0.55 wt% supplied during the different photo-deposition procedures on
the Au/Ti(x)/SBA materials (Table 5.4). As expected, Au can hardly be deposited on Au/SBA15, most likely due to a lack of photoactive sites. The residual Au found on Au/SBA-15 was
probably due to a simple decomposition or photobleaching process of auric acid on the support,
which is not necessarily photo-induced. For the Zn-containing materials Au/Zn0.3/Ti1.2/SBA
and Au/Ti1.2/Zn0.3/SBA it can be clearly seen that less Au was deposited, however, the results
show that the presence of TiOx species is required for the photo-deposition of Au driven by a
photocatalytic reduction of auric acid. In case of Au/Zn0.3/SBA, where no TiOx species are
present in the material, clearly less Au can be detected by ICP-OES. The deposition of Au in
this case may be related to a photo-bleaching process of auric acid solution as observed for
the bare SBA-15 material rather than to the photocatalytic reduction, and indeed the measured
values of Au are quite similar for Au/SBA-15 and Au/Zn0.3/SBA materials. Additionally, ICPOES results confirmed that leaching of Ti or Zn species during the photo-deposition process
can be neglected (results not shown).
5.3 Effects of Au and ZnO on the structure and photocatalytic activity of TiOx /SBA-15
materials
A)
73
B)
553
541
556
F(R) [a.u.]
F(R) [a.u.]
g)
f)
e)
d)
d)
c)
c)
b)
b)
a)
a)
250
300
350
400
450
500
550
600
Wavelength [nm]
650
700
750
250
300
350
400
450
500
550
600
650
700
750
Wavelength [nm]
Figure 5.15: A) Diffuse reflectance UV-Vis spectra of Au-modified materials with varying Ti
loading a) Au/SBA-15, b) Au/Ti0.3/SBA, c) Au/Ti1.0/SBA, d) Au/Ti2.0/SBA, e)
Au/Ti2.7/SBA, f) Au/Ti1.0/SBA (0.55 wt% Au), g) Au/Ti2.0/SBA (0.55 wt%
Au). B) Diffuse reflectance UV-Vis spectra of the Au-modified materials containing ZnOx in the SBA-15 matrix a) Au/Zn0.3/SBA, b) Au/Zn0.3/Ti1.2/SBA, c)
Au/Ti1.2/Zn0.3/SBA, and d) Au/Ti1.0/SBA.
The structure and position of the Au plasmon peaks were analyzed by UV-Vis spectroscopy.
The results obtained for the Au/Ti(x)/SBA samples are shown in Fig. 5.15a. Except for
Au/SBA-15, all materials showed a characteristic Au plasmon absorption at ∼ 550 nm. There
are obvious differences in the position of the plasmon absorption depending on the Au and the
Ti loading. A plasmon centered at a wavelength higher than 550 nm is observed after photodeposition of 0.25 wt% Au. The plasmon bands of the stepwise prepared 0.55 wt% Au samples
are clearly shifted to shorter wavelength. Additionally, the Au plasmon is shifted to slightly
lower wavelength with higher Ti loading. This phenomenon can be related to the Au particle
size, the reflective index of the supporting materials, or to different TiOx structures. [135,136] For
the subsequently grafted samples Au/Zn0.3/Ti1.2/SBA and Au/Ti1.2/Zn0.3/SBA the plasmon
peak is also centered at ∼ 550 nm (Fig. 5.15b). However, there are certain differences in the
nature of photo-deposited Au nanoparticles of the Zn-containing samples compared with the
Au/Ti(x)/SBA materials. Thus, it can be assumed that the differences clearly depend on the
presence of ZnOx -species incorporated in the SBA-15 matrix. A more intense Au plasmon was
observed in case of Au/Ti1.2/Zn0.3/SBA compared to the Au plasmon of Au/Zn0.3/Ti1.2/SBA.
The overall intensities of these plasmon peaks are lower for the Au/Zn0.3/Ti1.2/SBA and
Au/Ti1.2/Zn0.3/SBA materials compared to the Au/Ti(x)/SBA samples, which is in good
agreement with the ICP-OES results indicating that less Au was photo-deposited on the Zncontaining samples. However, the Au/Ti1.2/Zn0.3/SBA sample with a lower loading of Au as
measured by ICP-OES exhibits the higher plasmon peak, possibly due to different scattering
properties of the sample. For the Au/Zn0.3/Ti1.2/SBA and the Au/Ti1.2/Zn0.3/SBA samples
it can be assumed that metallic Au is present, as the plasmon of Au in the metallic state is
74
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
Figure 5.16: a) Typical BF-TEM of Au/Ti1.0/SBA. NPs are imaged as dark contrast, and are polydisperse. The SBA-15 channels are visibly intact. b) A FFT analysis of the white
frame in a) shows the pore spacing to be 104 Å. c) EFTEM Ti map of the tip of
the crystal visible in a). Some of the particles visible in a) give rise to strong Ti signals. The dark area in the middle of the crystal is due to thickness effects. d) EELS
spectrum of Au/Ti1.0/SBA showing the Ti-L2,3 and the O-K edge. e) Intensity profile
following the stripe in c) from top left to down right. Besides the particle, a clear Ti
background is present throughout the material.
accepted to be in the range of 500 - 600 nm. [135,136] No Au plasmon peak was observed for the
Au/Zn0.3/SBA sample most likely due to the low Au loading.
In-depth high-resolution TEM analysis was performed for the three samples with similar TiOx
loading, namely Au/Ti1.0/SBA, Au/Zn0.3/Ti1.2/SBA, and Au/Ti1.2/Zn0.3/SBA samples. BFTEM images of Au/Ti1.0/SBA showed the presence of intact channels as sxpected for SBA-15
mesoporous materials (Fig. 5.16a). The channels are highly ordered and exhibit a channel
spacing of 104 Å, determined by FFT analysis (Fig. 5.16b) of the area in the white frame in Fig.
5.16a. The channel spacing does, however, slightly change from crystal to crystal. The average
distance found in both characterized samples is approximately 100 Å. The nanoparticles that
are visible in the material shown in Fig. 5.16a are quite polydisperse with diameters ranging
from 10 to 90 nm.
The presence of Ti and O in the Au/Ti1.0/SBA material were proven by EELS. The EELS
spectrum acquired from the material (Fig. 5.16d) shows the presence of both the Ti-L2,3 and
the O-K edges. The Ti-L2,3 signal was used to acquire an Ti elemental map using EFTEM (Fig.
5.16c). This elemental map shows a homogeneous Ti distribution in the tip of the SBA-15
crystal as shown in Fig. 5.16c. The dark contrast at the center of the Au/Ti1.0/SBA crystal
5.3 Effects of Au and ZnO on the structure and photocatalytic activity of TiOx /SBA-15
materials
75
Figure 5.17: a) HAADF-STEM images of Au/Ti1.0/SBA. Numbered white frames label the areas
where STEM-EDX spectra were recorded. b) and c) show magnified HAADF-STEM
images of the areas where spectrum 2 and 3 were recorded. d) EDX spectra of the
material in the areas 1 (black), 2 (red) and 3 (blue), only 3 shows an Au signal.
is an imaging artifact (thickness effect). Fig. 5.16e shows the intensity profile over the line
indicated in the Ti map in Fig. 5.16c, from top left to bottom right. Some of the particles
visible in BF-TEM in Fig. 5.16a give rise to strong signals in the Ti map (Fig. 5.16c) and
resulted in peaks in the Ti intensity profile (Fig. 5.16e). Thus, these particles are then clearly
enriched in Ti. STEM-EDX (Fig. 5.17) and HR HAADF-STEM (Fig. 5.18) was used to further
explore the nature of these particles.
Most of the particles of Au/Ti1.0/SBA, which are visible in BF-TEM (Fig. 5.16c) and HAADFSTEM proved to be Ti-rich. STEM-EDX spectra, however, reveal that there are two types
of nanoparticles in this material; most particles exhibit only a small contrast difference with
respect to the framework material in the mass-thickness sensitive HAADF-STEM images (Fig.
5.17a and 5.17b area 2) are Ti-rich nanoparticles as proven by STEM-EDX spectra (Fig. 5.17d
spectra 1 and 2). Au nanoparticles detected based on the stronger image contrast with respect
to the framework material and the according STEM-EDX spectrum (Fig. 5.17d spectrum 3) are
also present. Often these Au nanoparticles are embedded in shells that exhibit a very similar
composition as the framework material.
76
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
Figure 5.18: a) HAADF-STEM image of Au/Ti1.0/SBA showing the presence of several Au
nanoparticles (bright white contrast). b) High-resolution HAADF-STEM image of
a single Au nanoparticle A, covered by a low-contrast, amorphous shell. c) The FFT
analysis of (b) confirms the particle is crystalline Au. d) Au nanoparticle B imaged
along the [111] zone axis orientation, as evidenced by FFT analysis shown in e) (scale
bar 2 nm).
The Au particles of Au/Ti1.0/SBA were further investigated by a FFT analysis of HAADFSTEM images in atomic resolution. Particle A in Fig. 5.18a exhibits the presence of multiple
twinning defects, typical for small Au nanoparticles. The FFT (Fig. 5.18c) of the HAADFSTEM image depicting it in atomic resolution shows only reflections fitting to the FCC Au
crystal structure. Fig. 5.18b shows a typical low-contrast amorphous shell surrounding the gold
nanoparticle. Particle B (Fig. 5.18d) is smaller than A and is a non-defected Au particle imaged along its [111] zone axis as proven by the FFT analysis (Fig. 5.18e). Even though analysis
was not performed in similar detail, HRTEM images of the Au/Ti0.3/SBA and Au/Ti2.0/SBA
revealed that polydisperse nanoparticles are formed, as was observed for Au/Ti1.0/SBA and
have been proven to be Ti-rich in nature. Furthermore, it can be assumed that the Au particles detected for the Au/Ti0.3/SBA and Au/Ti2.0/SBA are embedded in similar shells (Fig.
5.18b). It is likely that the formation of these core-shell structures and the Ti-enrichment in
the Au/Ti1.0/SBA sample observed at several positions occurs during photo-deposition of Au,
because the presence of Ti-rich areas was excluded by HRTEM measurements for a Ti1.0/SBA
support sample.
Similarly to the intact channel structure of the Au/Ti1.0/SBA sample the HRTEM image
5.3 Effects of Au and ZnO on the structure and photocatalytic activity of TiOx /SBA-15
materials
77
Figure 5.19: a) HRTEM image of a typical Au/Zn0.3/Ti1.2/SBA crystal. The channels are visibly intact. The FFT of a) (displayed in b)) reveals an average distance of approximately 100 Å. c) HAADF-STEM image of a typical Au/Zn0.3/Ti1.2/SBA crystal. d)
HAADF-STEM image signal intensity and Ti-K/Zn-L EDX intensity profiles over the
grey line in a) from the top left to the bottom right (Zn-L, black; Ti-K, red; image
intensity, green).
(Fig. 5.19a) of the Au/Zn0.3/Ti1.2/SBA material shows intact channels with an average distance of ∼ 100 Å (Fig. 5.19b). From BF imaging the presence of nanoparticles as observed for Au/Ti1.0/SBA was excluded and Au particles incorporated by photo-deposition
into the Au/Zn0.3/Ti1.2/SBA sample were hardly visible. Only vague shadows were imaged (Fig. 5.19a). Again, the HAADF-STEM mode was used to further characterize the
Au/Zn0.3/Ti1.2/SBA material. Using HAADF-STEM it was again verified that the channel
structure of SBA-15 support material is intact and the Au particles incorporated in the material
due to the photo-deposition procedure became visible (Fig. 5.19c). Most likely the large particles that can be seen at the outer surface are agglomerates of Au particles. These are found
very rarely and most of the Au particles within these agglomerates as well as in the channels of
the Au/Zn0.3/Ti1.2/SBA sample are monodisperse, with an average size of 8 ± 2 nm (see size
distribution in Fig. A11). A 2D line scan was performed in STEM-EDX to measure the Ti and
Zn distribution. Both elements are evenly distributed as it can be seen in the according element
profiles in Fig. 5.19d. The presence of Ti, Zn, and Au in Au/Zn0.3/Ti1.2/SBA was proven by
EDX (Fig. A12) and the Ti distribution was shown to be homogeneous without Ti-rich areas
(Fig. A13).
The position of the Au nanoparticles within the Au/Zn0.3/Ti1.2/SBA material was investigated
by electron tomography. Using the electron tomography technique, a three-dimensional reconstruction of the investigated material was retrieved. Fig. 5.20 shows a slice through the
three-dimensionally reconstructed volume of the investigated Au/Zn0.3/Ti1.2/SBA material.
The intact SBA-15 channels are resolved in the reconstruction, and the Au particles are visibly
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
78
y
x
z
xz
xy
yz
Figure 5.20: Tomographic reconstruction of a typical Au/Zn0.3/Ti1.2/SBA crystal, reconstruction
size = 622x738x817 nm (boundaries indicated by the orange box). In the 3D representation (left top) the Au nanoparticles are displayed in yellow, the SBA framework
in soft-red. The arrangement of the channels is clearly visible as well as the Au particles. The orthoslices through the reconstruction show that the SBA framework pores
are resolved. The Au nanoparticles are found within the SBA-15 crystal as well as at
the surface.
distributed throughout the whole SBA-15 crystal, and at the surface of the Au/Zn0.3/Ti1.2/SBA
material. The artifacts in the reconstruction are a result of the missing wedge and the tendency
of the material to charge under the beam which can result in movement of the SBA-15 crystal. Using HRTEM measurements it was unambiguously shown that larger Au particles and
Ti-rich areas, observed for the Au/Ti1.0/SBA, were not existing in the Au/Zn0.3/Ti1.2/SBA
sample. The HRTEM measurements are supported by UV-Vis measurements, which reveal
that the intensity of the characteristic Au plasmon of the Au/Zn0.3/Ti1.2/SBA sample is less
intense. It can be assumed that monodisperse homogeneously distributed Au particles within
a mesoporous matrix certainly will possess a less intense Au plasmon compared with larger
particles mainly situated at the outer surface of a material. Taking this into account small Au
particles mainly in the pores of the material are expected for the Au/Ti1.2/Zn0.3/SBA sample
and indeed HRTEM images confirmed the presence of small Au particles (Fig. A14). However,
a slight enrichment of Ti in certain areas, possibly associated with larger Au particles was observed in the Ti EFTEM images for the Au/Ti1.2/Zn0.3/SBA. The photocatalytic activity of the
5.3 Effects of Au and ZnO on the structure and photocatalytic activity of TiOx /SBA-15
materials
79
three different materials clearly decreases in the order Au/Ti1.0/SBA > Au/Zn0.3/Ti1.2/SBA
> Au/Ti1.2/Zn0.3/SBA as Au/Ti1.0/SBA was active for the evolution of H2 and the hydroxylation of TA. Regarding the presented HRTEM characterization results these differences in
photocatalytic activity can be explained by the enrichment in TiOx and the differences in the
particle sizes of the incorporated Au nanoparticles.
The above presented results provide evidence that highly active isolated or slightly polymerized TiOx were obtained by the anhydrous grafting of Ti(Oi Pr)4 on SBA-15 support. The isolated species appear to be mobile, at least during photo-deposition of nanoparticular Au, and
tend to agglomerate during the irradiation process. Due to the high photocatalytic activity of
the initially isolated and presumably then agglomerated TiOx species photo-deposition of Au
nanoparticles is mainly taking place at the outer surface: larger Au particles are photo-deposited
and the probability of Au precursor diffusion into the channels of the Au/Ti(x)/SBA samples
is low. The agglomeration and photo-deposition of larger Au particles was observed for all
Au/Ti(x)/SBA samples and a relation between the hydrogen evolution activity and these insitu generated Au/TiOx assemblies is likely. Even though it was not possible to be definitively
confirmed by means of HRTEM, the observed trends of photocatalytic hydrogen evolution due
to methanol steam reforming for the Au/Ti(x)/SBA samples might be explained as follows:
Less hydrogen was evolved during Au photo-deposition in presence of Ti0.3/SBA compared to
Ti1.0/SBA due to the lower TiOx loading. With increasing TiOx loading as in Ti2.0/SBA or in
Ti2.7/SBA again less hydrogen was evolved during photo-deposition. These samples exhibited
less isolated TiOx species. Instead, TiOx species are present as oligomers and even polymerized
TiOx at the surface of the SBA-15 was confirmed. [134] Presumably, these interconnected TiOx
species are less mobile at the surface and only remaining isolated TiOx sites are able to move.
Thus, less Au/TiOx assemblies can be formed in the photo-deposition resulting in less active
sites and lower hydrogen evolution. It is also possible to assume that Ti-O-Si bonds still need
to be present in the agglomerates formed around the gold nanoparticles as the composition of
the shell was confirmed to be similar to the framework of the material.
In case of the Au/Zn0.3/Ti1.2/SBA and the Au/Ti1.2/Zn0.3/SBA samples monodisperse, evenly
distributed particles are photo-deposited in the channels and at the outer surface of the material. Furthermore, less Au is deposited on both samples compared to Au/Ti(x)/SBA materials
as evidenced by ICP-OES measurements. The mobility of the TiOx species after subsequent
deposition of ZnOx appears to be lower or is even inhibited in analogy to the higher loaded
Ti2.0/SBA and Ti2.7/SBA samples, and thus the agglomeration of TiOx is avoided. The TiOx
species remain mainly isolated and the photo-deposition of Au is slow for the Zn0.3/Ti1.2/SBA
sample resulting in a lower Au loading after 90 min of irradiation and the hydrogen evolution
due to photoreforming of methanol, if occurring at all, is below the detection limit.
The Au/Zn0.3/Ti1.2/SBA and the Au/Ti1.0/SAB samples exhibit similar photocatalytic activities towards TA hydroxylation. Two different sites seem to be responsible for the hydrogen evolution reaction and the TA hydroxylation, e.g. Au/TiOx assemblies arranged in a core-shell like
80
5 Photocatalytic CO2 Reduction - Single-site TiOx Materials
structure as observed for Au/Ti(x)/SBA samples may be responsible for the hydrogen evolution
and Au in close contact to isolated TiOx species facilitates the hydroxylation of TA. Although
the Au distribution of the Au/Ti1.2/Zn0.3/SBA sample is similar to the Au/Zn0.3/Ti1.2/SBA
sample it is less active in the TA test reaction, and therefore the lower activity is exclusively
assigned to the formation of Ti-O-Zn bonds instead of Ti-O-Si bonds as already explained.
Finally, it should be mentioned that the observed behavior of Ti(x)/SBA-15 materials during
Au photo-deposition resembles the SMSI effect observed for metal/TiO2 materials. It is believed that reduction of TiO2 gives rise to TiOx suboxide species (x < 2), which is believed
to migrate onto metal particles. The state of titania present in the Ti(x)/SBA-15 materials can
be considered as a TiOx species, which can migrate onto the metal particles. Furthermore, the
desired reduction of titania can be achieved by the UV-light excitation of the isolated Ti-sites.
The SMSI effect is in good agreement with the considerations of an enhanced mobility of the
titania in species Ti(x)/SBA-15.
5.3.3 Conclusions and Outlook
The presented results show that Au/Ti(x)/SBA-15 samples are photocatalytically active in the
hydrogen production by the photoreforming reaction of methanol and in the hydroxylation
of terephthalic acid. Therefore, the availability of electrons and holes upon photo-excitation
was proven. Furthermore, a correlation of titania loading and photocatalytic activity was observed: the highest activity was observed for materials with the majority of isolated titania
sites. In contrast to the behavior observed for Au/Ti(x)/SBA-15, the photocatalytic activity
is drastically changed upon incorporation of zinc oxide species as in Au/Zn0.3/Ti1.2/SBA
and Au/Ti1.2/Zn0.3/SBA. A photocatalytic activity resembling the activity of the most active Au/Ti(x)/SBA-15 was only observed for the Au/Zn0.3/Ti1.2/SBA sample. The differences
in activity were attributed to titania species in a Zn-O-Ti environment, which appeared to be
unfavorable for the photocatalytic activity. Moreover, HRTEM studies indicated a diffusioncontrolled photo-deposition of Au for Ti(x)/SBA-15, and larger Au particles at the outer surface of the Au/Ti(x)/SBA-15 were detected. These particles were mostly associated with a
shell of similar composition as the framework material, and it seems that Au-Ti agglomerates are formed. These agglomerates were completely absent in Au/Zn0.3/Ti1.2/SBA and
Au/Ti1.2/Zn0.3/SBA, and only small Au particles in the pores of the SBA-15 were observed.
The presence of ZnOx seems to decrease the mobility of titania species required to form the
Au-Ti agglomerates.
6 Heterogeneous
Photoelectrochemistry and
Photocatalysis of TiO2
Nanomaterials
In this chapter Nb-doped TiO2 bulk materials are discussed. In the first section a novel synthesis
procedure is described. In-depth characterization of the synthesized materials is used to proof
the suitability of the procedure. The improvement in photo-to-current in hybrid photoelectrodes
prepared with the Nb-doped TiO2 samples as electron collector material are reported in the
second chapter and a strategy to improve non-metal doping is discussed in the third section of
this chapter.
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
82
6.1 The synthesis of Nb-doped TiO2 nanoparticles by spray
drying
Abstract1
Nb-doped TiO2 nanoparticles were prepared by a continuous spray drying process using ammonium niobate (V) oxalate and titaniumoxysulfate as water-soluble precursors. The structural
and electronic properties were investigated using thermogravimetric analysis, X-ray diffraction,
X-ray photoelectron spectroscopy, and Raman spectroscopy. Nb was found to be mainly incorporated as Nb5+ into the TiO2 lattice resulting in a charge compensation by Ti vacancies. The
characterization results indicate that Nb was homogeneously distributed within the titania lattice, and that the surface segregation of Nb, which is commonly observed for Nb-doped TiO2,
was significantly less pronounced. The high homogeneity and the lower extent of surface segregation originate from the efficient atomization of homogeneous precursor solutions and the
fast evaporation of the solvent in the spray drying process. As a result, the ion mobility is
diminished and spheres of well-mixed precursor materials are formed. Using the continuous
spray drying process followed by a controlled heat treatment, the phase composition, the crystal
size, and the surface area of the Nb-doped TiO2 nanoparticles are easily adjustable.
1 The
content of this section is published as ”The synthesis of Nb-doped TiO2 nanoparticles by spray drying: an
efficient and scalable method”, B. Mei, M.D. Sanchez, T. Reinecke, S. Kaluza, W. Xia, M. Muhler, J. Mater.
Chem., 2011, 21, 11781-11790, DOI:10.1039/C1JM11431J.
6.1 The synthesis of Nb-doped TiO2 nanoparticles by spray drying
83
6.1.1 Short Introduction
Transition metal oxides such as titanium dioxide (TiO2 ) have been extensively studied in the last
decades for various applications. TiO2 -based materials of different particle size and morphology are used in scientifically and technologically interesting areas such as the photocatalytic
splitting of water, [153] environmental cleanup, [154] gas sensing, [155,156] and in fuel cells. [157]
Furthermore, TiO2 is applied as support material in heterogeneous catalysis due to strong
metal-support interactions, high stability in oxidizing environments, low cost, and commercial
availability. [18] Due to the wide band gaps of the commonly used TiO2 modifications anatase
(3.2 eV) and rutile (3.0 eV), it is often necessary to modify their band structure. [18,158,159] For
photocatalytic applications the modification of the band structure is intended to lead to visible
light absorption, whereas in electrochemical applications the conductivity of TiO2 needs to be
enhanced by tuning its band gap.
The applicability of wide band gap oxides can be improved by different methods including
doping with anions or cations. [160,161] Typically, anions such as N, C, F, or S can be used to substitute lattice oxygen anions, whereas transition metal cations substitute Ti cations. [19,162–164]
Depending on the application it is crucial to consider the splitting of the d orbitals of transition
metals and the effect of their different electron configurations. In addition, the formation of
solid solutions of two or more oxides is commonly used to change the band structures of wide
band gap oxides.
In recent years Nb was employed to dope TiO2 particles or films for different applications. [156,157,165,166] These investigations showed that the doped materials can be applied as
conductive support for cathode catalysts in polymer electrolyte membrane fuel cells (PEMFCs), [157,167] as counter electrode in dye-sensitized solar cells (DSSCs), [168–170] and as transparent conductive oxides (TCOs). [171]
The incorporation of Nb into the titania lattice leads to changes in the electronic structure,
which are extensively discussed in literature. The charge compensation of Nb5+ substituting
Ti4+ depends on the oxygen partial pressure, the temperature applied during the synthesis, and
the Nb concentration. It can be achieved by two mechanisms: one possibility is the reduction
of one Ti4+ to Ti3+ for every Nb5+ incorporated (eq. 6.1), the other possibility is the formation
of one Ti vacancy per four Nb5+ introduced (eq. 6.2).
′
1/2Nb2O5 + TixTi → Nb.Ti + TiTi + 5/4O2
′′′
1/2Nb2O5 + TixTi → Nb.Ti + 1/4VTi + TiO2 + 1/4O2
(6.1)
(6.2)
Reaction (6.1) mainly occurs at low oxygen partial pressures and in the synthesis of samples
with high Nb content. In this case one excess electron occupies the Ti 3d orbital leading to
an increase of the electron density. Using high oxygen partial pressures favors the formation
of titanium vacancies (eq. 6.2). In addition, the substitution of Ti4+ by Nb4+ can be obtained
84
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
by a high temperature treatment in reducing atmosphere. The preparation of doped powder
materials can be achieved by different methods, such as hydrothermal synthesis, [157,169] pyrolysis, [172] [173] high temperature ceramic routes, [174–176] and sol-gel methods. [162,177–179] These
methods usually involve complicated processes, and the resulting properties of the product
strongly depend on the various process parameters. Furthermore, the scale-up of these methods is rather difficult. For instance, Kubacka et al. [162,163] prepared doped TiO2 materials by
a microemulsion method. This sophisticated preparation comprises stirring for 24 h followed
by several additional steps, such as centrifugation, drying, and washing, which is rather timeconsuming. The synthesis of Nb-doped TiO2 by a hydrothermal method described by Park and
Seol [157] is another example for a rather time-consuming process. Furthermore, both preparation techniques require very hydrophilic materials, which have to be handled skillfully. The
high-temperature ceramic route used for the synthesis of doped TiO2 is an energy-intensive
process. For example, Morris et al. [174] used ceramics for valence band studies, where the
starting materials were treated at temperatures as high as 1000 ◦ C for several days.
Recently, spray drying has been successfully employed as a fast and scalable method for the
preparation of oxide materials. Kaluza et al. [180–182] used a coupled precipitation-spray drying
approach to prepare high surface area ZnO nanoparticles and ZnO-Al2 O3 nanocomposites as
supports for copper catalysts applied in methanol synthesis. By using this spray drying technique, it was possible to investigate the aging process of the polycrystalline precipitates as well
as the role of the post-treatment during the catalyst preparation. Hence, important preparation
parameters can be easily controlled using this approach, whereas in conventional batch processes the investigation of the aging of the precipitate appeared to be rather difficult. Spray
drying has also been used for the synthesis of TiO2 materials. Sizgek et al. [183] prepared titanates by a coupled sol-gel and spray drying approach resulting in well-defined microspheres.
This method revealed that the particle porosity and morphology strongly depend on the dispersion of the used sol. Porous TiO2 nanowires and macroporous particles for photocatalytic
applications were prepared by Zhang et al. [184] and Iskandar et al. [185] Zhang et al. [184] used
suspensions of powdered nanowires and a surfactant to synthesize hierarchically porous TiO2
microspheres consisting of nanowires, whereas Iskandar et al. [185] focused on the preparation
of macroporous brookite nanoparticles using suspensions of brookite nanoparticles together
with a template for spray drying. The obtained macroporous particles showed higher photocatalytic activity in waste water treatment compared to dense particles due to higher surface
area. [185] Thus, applying the spray drying approach combines several major advantages, such
as homogeneous particle size distribution, fast drying under mild conditions, and good parameter control. However, finding a soluble TiO2 precursor suitable for spray drying is challenging,
and, to the best of our knowledge, the synthesis of cation-doped TiO2 by spray drying has not
yet been reported.
In this contribution, we present a continuous method for the synthesis of Nb-doped TiO2
nanoparticles by spray drying. A peristaltic pump transports an acidic solution containing Nb
6.1 The synthesis of Nb-doped TiO2 nanoparticles by spray drying
85
and Ti ions into the two-fluid nozzle of the spray dryer. After rapid evaporation of the liquid
and separation of the gas, the solid can be collected as colorless powder. The segregation of the
ions and particle aggregation are avoided due to the rapid drying process. Thus, the Nb and Ti
cations are well mixed within the powder. The spray drying process leads to a narrow particle
size distribution, because the ion content of the feed solution is homogeneously dispersed and
the atomization by the nozzle is well defined. Furthermore, the scale-up of this synthesis route
to an industrial level is easily feasible.
Experimental
Sample Preparation
Titaniumoxysulfate sulfuric acid hydrate (TiOSO4 x H2 SO4 x H2 O, synthesis grade) and ammonium niobate (V) oxalate hydrate ((NH4 )NbO(C2 O4 )2 x H2 O, 99.9 %, denoted as ANO)
were supplied by Aldrich and used as received. The Ti complex was dissolved in nitric acid
solution (c = 1 mol L-1 ) by stirring for at least 4 h. Subsequently, a predetermined amount of
the ANO complex was added to the colorless solution to achieve Nb concentrations between
0.2 - 50 at % relative to Ti. The obtained solution was stirred for 1 h and used for the synthesis
of Nb-doped TiO2 precursor materials through a continuous spray drying method. A benchtop spray dryer (B-290, Büchi) with a two-fluid nozzle was used for fast evaporation of the
fluid. [180–182] The colorless powder samples were collected from the dryer and calcined under
flowing synthetic air (20.5 % O2 in He, 100 sccm) at 600 ◦ C, 700 ◦ C or 800 ◦ C for 1 h. A
heating rate of 1 K min-1 was applied leaving sufficient time for flushing out volatile species
evolving during the decomposition of the precursors. After calcination the samples were thoroughly washed with distilled water and dried in static air for 24 h at 110 ◦ C. The prepared
samples are labeled based on the composition and the calcination temperature. For instance,
Ti0.998 Nb0.002 -700 indicates a theoretical composition of 99.8 at% Ti ions and 0.2 at% Nb ions
and a calcination temperature of 700 ◦ C. For comparison, monometallic Ti or Nb oxides were
prepared using single Ti or Nb sources, where ANO was also dissolved in acidic solution.
Characterization Methods
The materials were investigated by means of thermogravimetry coupled with mass spectrometry (TG-MS), X-ray diffraction (XRD), Raman spectroscopy, UV-Vis spectroscopy, nitrogen
physisorption, and X-ray photoelectron spectroscopy (XPS). TG-MS analysis was carried out
in a Cahn TG2131 thermobalance coupled with a quadrupole mass spectrometer (Balzer, Omnistar). Samples were heated to 800 ◦ C with a heating rate of 2 K min-1 in synthetic air with
at total flow rate of 30 sccm. XRD studies were performed with a PANalytical MPD diffractometer with Cu Kα radiation in a 2θ range from 10 to 70◦ . Powder Diffraction Files (PDF2)
from the International Centre of Diffraction Data (ICDD) combined with the XPert Line software (Panalytical, Almelo) were used for qualitative phase analysis. Raman measurements
were performed with a Nexus FT-NIR spectrometer equipped with a nitrogen-cooled germanium detector and a 1064 nm laser (Nd:YAG) provided by Thermo Fisher Scientific. UV-Vis
86
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
spectra were recorded in the diffuse reflectance mode in a Perkin Elmer Lambda 650 UV-Vis
spectrometer equipped with a Praying-Mantis mirror construction. The obtained spectra were
converted by the Kubelka-Munk function F(R) into absorption spectra using BaSO4 as white
standard. Static nitrogen physisorption experiments were performed at the boiling point of liquid N2 subsequent to out-gassing at 150 ◦ C in a Quantachrome Autosorb-1-MP setup. Data
were analyzed according to the BET equation assuming that the area covered by one nitrogen
molecule is 0.162 nm2 . The mean diameter d(BET) was calculated based on the assumption of
non-porous spherical particles. The density of anatase (ρ anatase = 3.85 x 103 kg m-3 ) [186] was
used in all cases. It has to be noted that this is only a rough approximation for materials with
imperfect structures. The pore size distribution was obtained applying the BJH equation. XPS
measurements were carried out in an UHV set-up equipped with a Gammadata-Scienta SES
2002 analyzer. The base pressure in the measurement chamber was 5 x 10-10 mbar. Monochromatic Al Kα (1486.6 eV; 13.5 kV; 37 mA) was used as incident radiation and a pass energy of
200 eV was chosen resulting in an energy resolution better than 0.6 eV. Charging effects were
compensated using a flood gun, and binding energies were calibrated based on positioning the
main C 1s peak at 284.5 eV, which originates from carbon contaminations.
6.1.2 Results and Discussion
Common Ti compounds such as TiCl4 are typically air- and water-sensitive rendering them
unsuitable as Ti source for spray drying. Titaniumoxysulfate dissolves slowly in HNO3 under
stirring and is stable in acidic solutions. As the spray dryer was equipped with glass reactors
and Teflon tubing, it was possible to use the Ti- and Nb- containing acidic solution directly for
the spray drying synthesis resulting in colorless powders collected in the product vessel.
TG-MS applied to the precursor materials
The decomposition of the powders obtained from spray drying, i.e. the precursor for mixed
TiNb oxides, was first studied by TG-MS in flowing air. The TG-MS results of pure Ti1 , pure
Nb1 , and Ti0.5Nb0.5 mixed precursors are shown in Fig. 6.1 It can be seen that the weight loss
of the pure Ti precursor occurred in two steps: at around 78 ◦ C accompanied by the release of
H2 O (m/e = 18), and at about 580 ◦ C accompanied by the release of a small amount of CO2
(m/e = 44) (Fig. 6.1a). Surprisingly, SO2 (m/e = 64) and SO3 (m/e = 80) were not detected
by MS in the whole temperature range. Obviously, the Ti precursor decomposed completely
to TiO2 at about 600 ◦ C. Further heating to 800 ◦ C caused a negligible weight loss. The
transformation temperature to TiO2 of 540 - 580 ◦ C is in good agreement with the results
reported by Johnson et al. [187] They observed that the weight loss in the temperature region
below 540 ◦ C is only related to the release of water. According to the obtained TG data of the
spray-dried TiOSO4 sample, water mainly desorbs below 200 ◦ C, and no further weight loss
is detected up to 540 ◦ C. Hence, water is bound less strongly to the spray-dried sample. The
major weight loss above 540 ◦ C can be assigned to the decomposition of the precursor and the
6.1 The synthesis of Nb-doped TiO2 nanoparticles by spray drying
78°C
210°C
580°C
a)
87
142°C
b)
c)
235°C
CO
2
630°C
510°C
142°C
H O
2
540°C
290°C
585°C
DTG [a.u.]
DTG [a.u.]
DTG [a.u.]
558°C
CO
CO
90°C
2
2
CO
CO
H O
H O
2
0
200
400
600
temperature [°C]
800
0
200
400
2
600
temperature [°C]
800
0
200
400
600
800
temperature [°C]
Figure 6.1: DTG curves and MS profiles of the spray-dried materials Ti1 , Nb1 , and Nb0.5 Ti0.5 using
a heating rate of 2 K min-1 a) DTG curve of the spray-dried Ti1 precursor and the MS
profile of H2 O and CO2 during decomposition, b) DTG curve of Nb1 precursor (ANO)
and the MS profiles of H2 O, CO, and CO2 , and c) DTG curve of spray-dried Nb0.5 Ti0.5
precursor and the MS profile of H2 O, CO, and CO2 .
release of different sulfur species as suggested by Johnson et al. [187] Due to the lack of strong
contributions from SO2 and SO3 , the sulfur content in the precursor is presumably released as
H2 SO4 in the applied temperature region. Thus, the decomposition process can be described
by the following equation:
TiOSO4 · H2 SO4 · xH2 O(s) → TiO2(s) + 2H2 SO4 (g) + (x − 1)H2O(g)
(6.3)
The decomposition profile of the spray-dried ANO precursor shown in Fig. 6.1b can also be
divided into two main regions. The first weight loss region from 50 ◦ C to 250 ◦ C is associated
with the release of CO (m/e = 28), CO2 (m/e = 44), and water (m/e = 18). The second decomposition region is indicated by a sharp increase of the CO2 signal at 510 ◦ C. Furthermore,
sharp CO and CO2 decomposition peaks are detected at 540 and 560 ◦ C, and no further weight
loss is observed above 600 ◦ C. The second decomposition can be assigned to the decomposition of strongly bound oxalate or, more likely, to oxalate encapsulated in Nb2 O5 , whereas a
weight loss below 200 ◦ C is mainly due to the loss of water and ammonia. [188] The loosely
bound oxalate is decomposed between 250 and 350 ◦ C. [188] In case of the spray-dried ANO
sample the decomposition of the oxalate already starts at 140 ◦ C and is terminated at 250 ◦ C.
Thus, for the spray-dried ANO precursor the conversion into Nb2 O5 is favored. It is interesting
to note that both the low and high temperature regions consist of three different CO and CO2
peaks. Therefore, the carbon species in the spray-dried ANO powder could exist in three different environments. Hence, the overall equation of the decomposition of the spray-dried ANO
88
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
137°C
Ti
582°C
Nb
0.995
Ti
0.005
Nb
0.99
Ti
0.01
Nb
0.95
Ti
CO
0.05
Nb
0.9
0.1
DTG [a.u.]
2
535°C
0
100
200
300
400
500
600
700
800
temperature [°C]
Figure 6.2: DTG curves of the spray-dried precursor powders (bottom) containing 0.5, 1, 5, and 10
at% Nb, and the corresponding CO2 MS profiles (top).
precursor can be described as:
2[(NH)4NbO(C2 O4 )2 ] · xH2 O(s) →
Nb2 O5 (s) + 2NH3(g) + 6CO(g) + 2CO2(g) + (x + 1)H2O(g)
(6.4)
In Fig. 6.1c the decomposition profile of the spray-dried Ti0.5 Nb0.5 precursor is shown. The
profile obtained with this sample is more related to the decomposition of a mixed phase rather
than the decomposition of pure TiOSO4 (Ti1 ) or the pure ANO (Nb1 ) precursor. The first strong
weight loss signal is located at 140 ◦ C. This weight loss is clearly related to the evolution of
CO and CO2 and can be assigned to the decomposition of the oxalate species from the ANO
precursor. A second major weight loss is detected at 290 ◦ C, which is also related to the
evolution of CO and CO2 and is not observed in any of the reference samples. Compared to the
ANO (Nb1 ) sample the decomposition accompanied by the release of CO2 is shifted towards
higher temperatures. Therefore, this weight loss can be assigned to the decomposition of a
mixed phase containing Nb and Ti. The third strong weight loss occurs in the region of 540 to
640 ◦ C with a maximum located at 630 ◦ C. The peak obtained by DTG analysis is rather broad
and the MS signals show two decomposition steps. The first step centered at 585 ◦ C may be
related to the decomposition of a pure Ti phase, whereas at higher temperatures mixed Ti-Nb
phases may be fully converted into oxides. Additionally, it can be seen from Fig. 6.1 that the
water signals vary for different precursors. The Ti1 precursor shows mainly one peak with a
maximum at around 80 ◦ C and a change in the background of the m/e = 18 signal at 580 ◦ C,
which may be related with the release of another species. The ANO precursor exhibits two
steps mainly below 500 ◦ C and in case of 1:1 mixtures of Nb and Ti (Ti0.5 Nb0.5 ) two broad
water peaks of low intensity are detected at 150 and 585 ◦ C.
The TG-MS profiles of the precursors containing a small amount of Nb ions between 0.05 -
6.1 The synthesis of Nb-doped TiO2 nanoparticles by spray drying
89
10 at% are presented in Fig. 6.2. The profiles obtained are quite similar to the decomposition
profile of the spray-dried TiOSO4 precursor. With increasing Nb content two new features
appear: first, a decomposition peak at low temperatures is observed, which is associated with
the evolution of CO2 at 137 ◦ C. In addition, a second weight loss signal in the high temperature
region at 582 ◦ C appears, which becomes broader with increasing Nb amount. Thus, the second
decomposition process already starts at lower temperature. Due to the low amount of Nb ions
the conversion into TiO2 is not significantly affected as indicated by the DTG curve. Even with
higher Nb doping the measured TG-MS profiles are similar suggesting that even with a loading
of 10 at% the Nb ions are well dispersed within the TiO2 particles without the formation of
NbOx islands. The formation of Nb2 O5 or the formation of solid solutions of a NbTix Oy type
should result in TG profiles similar to those presented in Figs. 6.1b and 6.1c, respectively.
None of the characteristic peaks in the DTG curves is observed in case of the samples with
low Nb content. In comparison with the TG-MS data presented in Figs. 6.1b and 6.1c the
decomposition of the loosely bound oxalate occurs at lower temperature, and the release of
strongly bound carbon and sulfur species is also shifted to lower temperatures. These shifts
can be rationalized by assuming that small amounts of Nb ions have a positive influence on
the decomposition of sulfate, and Ti ions are positively influencing the oxidation of carbon
species. It is worth mentioning that the oxidation of CO to CO2 is no longer favored for the
Ti0.9 Nb0.1 sample and that the MS signal detected at mass 18 becomes broader in this case.
Both observations indicate that the local environment of the Nb ions corresponds more and
more to the environment of Nb ions in the Ti0.5Nb0.5 sample forming a solid NbTixOy solution.
Structural characterization by XRD and Raman spectroscopy
Based on the TG-MS results the precursors with different Ti to Nb ratios were calcined at 600,
700, or 800 ◦ C. The structures of the calcined samples were characterized by XRD and Raman
spectroscopy. The obtained diffraction patterns are shown in Fig. 6.3. The calcination of Ti1
and Ti0.95 Nb0.05 at 600 ◦ C leads to the full decomposition of the precursor as shown by the TG
analysis. However, conversion into the tetragonal anatase phase is not fully completed, as minor diffraction lines of the monoclinic β -TiO2 phase are additionally detected (Fig. 6.3a). The
β -phase is structurally related to anatase, with which it may be semicoherently intergrown, [189]
and to which it irreversibly reacts upon heating. [190] In case of the Ti0.95 Nb0.05 -600 sample the
diffraction peaks are broader and less intense than those of Ti1-600. Thus, the incorporation of
Nb is accompanied by the formation of smaller crystallites (see below). In the 700 ◦ C calcination series, the decomposition of the precursor and the transformation to the anatase structure
is complete for samples with low Nb-loading up to 1 at% (Fig. 6.3b). For Ti0.95 Nb0.05 -700
and Ti0.9Nb0.1 -700 small diffraction peaks of β -TiO2 are detected in addition to anatase, but
no characteristic peaks of the rutile phase are observed in the diffraction patterns. Furthermore, the diffraction maxima of the Nb-doped anatase specimens (corrected with the NIST
SRM 660a LaB6 standard for instrumental aberrations in 2θ ) slightly shifts to lower diffraction
angles indicating the incorporation of Nb ions into Ti lattice sites. [191] The lattice expansion of
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
90
Anatase
1
2) Ti
1000 counts
0.995
3) Ti
0.005
*
Nb
0.99
4) Ti
0.95
5) Ti
Rutile
Nb
ß-TiO
b) 700°C
1000 counts
a) 600°C
1) Ti
2
0.01
5)
Nb
0.05
Nb
0.9
0.1
4)
3)
* *
*
* *
*
4)
2)
1)
1)
25
30
35
40
45
50
55
60
2
25
30
35
40
45
50
55
60
[°]
1000 counts
c) 800°C
5)
4)
2)
1)
20
25
30
35
40
2
45
50
55
60
[°]
Figure 6.3: X-ray diffraction patterns of the spray-dried precursors calcined at different temperatures. Spray-dried powder without Nb and low Nb doping levels calcined at a) 600 ◦ C,
b) 700 ◦ C, and c) 800 ◦ C. The diffraction pattern of anatase TiO2 (21-1272) and rutile
TiO2 (88-1175) are included for comparison.
anatase due to the incorporation of Nb is quantified with the refined lattice parameters (Table
6.1). While the lattice parameters of the Ti1-700 anatase are a = 3.784(1) Å and c = 9.505(5)
Å, which is in good agreement with values reported in the literature, [192] the cell parameters
increase up to a = 3.791(3) Å and c = 9.510(12) Å in the Ti0.9 Nb0.1 -700 sample. This increase of
the cell parameters with increasing Nb content confirms that Nb is incorporated in the anatase
lattice, because the ionic radius of the Nb5+ ions into octahedral coordination is larger than that
of Ti4+ . [193] In addition, the effect of Nb on the size and morphology of anatase crystallites
was investigated by means of the integral breadth method. [194] The apparent size of all anatase
lattice planes (hkl) (measured by pattern decomposition and corrected for instrumental line
broadening) is concordant indicating the absence of crystallite shape anisotropy in all samples.
Ignoring the potential minor effects of lattice imperfections (e.g. intercalation of β -TiO2 lamellae with anatase) and strain on line broadening, a volume-weighted mean d(XRD) is calculated
from the arithmetic mean of the apparent size values in each sample (Table 6.1). These data
clearly suggest a decrease in crystallite size with increasing Nb content.
6.1 The synthesis of Nb-doped TiO2 nanoparticles by spray drying
91
1) Ti
Nb
-700
2) Ti
Nb
-800
0.5
*
Nb O
*
2
0.5
5
2
counts [a.u.]
1)
450 400 350 300 250 200 150
*
**
0.5
0.5
3) Nb -700
TiNb O
4)
3)
2)
10
Raman intensity [a.u.]
5)
* *
1
7
#
Anatase
*
Rutile
4) Nb -800
1
#
4)
*
3)
*
5)
*
*
*
*
#
# #
*
#
4)
2)
3)
1)
2)
1)
1200
1000
800
600
400
-1
200
0
20
25
30
35
40
45
50
55
60
2
Raman shift [cm ]
(a) Raman Spectroscopy
(b) X-Ray Diffraction
Figure 6.4: a) Raman spectra of spray-dried Nb-doped TiO2 precursor powders calcined at 700 ◦ C.
1) Ti1 , 2) Ti0.995 Nb0.005 , 3) Ti0.99 Nb0.01 , 4) Ti0.95 Nb0.05 , 5) Ti0.9 Nb0.1 . Bands assigned
to β -phase TiO2 are labeled with an asterisk. b) Spray-dried ANO precursor (Nb1 ) and
spray-dried Ti0.5 Nb0.5 precursor calcined at 700 and 800 ◦ C. The diffraction pattern of
anatase TiO2 (21-1272), rutile TiO2 (88-1175), Nb2 O5 (30-0873), and TiNb2 O7 (720116) are included for comparison.
Table 6.1: Refined lattice parameters and volume-averaged crystallite sizes of anatase derived from
XRD measurements.
pretreatment
sample
Ti1
Ti0.995 Nb0.005
Ti0.99 Nb0.01
Ti0.95 Nb0.05
Ti0.9 Nb0.1
700 ◦ C, 1 h
lattice parameter a/c [Å] d(XRD) [nm]
3.784(1)/9.505(5)
3.784(1)/9.505(6)
3.788(6)/9.510(5)
3.791(2)/9.512(8)
3.791(3)/9.510(12)
20
18
19
14
13
800 ◦ C, 1 h
lattice parameter a/c [Å] d(XRD) [nm]
3.7828(8)/9.511(3)
3.7839(9)/9.516(4)
/
3.791(1)/9.515(5)
3.785(1)/9.513(6)
41
35
/
22
22
The qualitative phase composition of the 700 ◦ C calcination series obtained from the XRD
studies is clearly confirmed by Raman spectroscopy (Fig. 6.4a). It is known that the most
characteristic and strongest band of anatase appears at 151 cm-1 , and that weaker bands can
be detected at 639, 515, 400, and 204 cm-1 , whereas the Raman bands of rutile phase can be
observed at 612, 448, 240, and 147 cm-1 . [46] In all spectra the detected bands can be clearly
assigned to the anatase phase. Three weak bands in the 200 - 400 cm-1 region of the Ti0.95 Nb0.05
and Ti0.9 Nb0.1 samples are likely due to the presence of β -TiO2 . Bands of niobia phases are
not observed indicating that there is no phase separation or formation of NbOx islands on the
surface of TiO2 particles. Similar changes in crystallite size and cell parameters with bulk
composition are observed for the spray-dried precursors calcined at 800 ◦ C. While a large
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
Desorption Dv(d) x10
Ti
0.5
Ti
Ti
0.4
Ti
1.2
1
Nb
0.995
Nb
0.95
Nb
0.9
Nb
0.5
0.3
0.2
0.1
0.0
10
100
1000
0.005
0.05
0.1
0.5
[cc/A/g]
Ti
a)
-2
-2
[cc/A/g]
0.6
Desorption Dv(d) x10
92
Ti
b)
Ti
1.0
Ti
Ti
0.8
Ti
1.00
Nb
0.995
Nb
0.99
Nb
0.95
Nb
0,90
0.005
0.01
0.05
0,10
0.6
0.4
0.2
0.0
10
100
1000
Pore diameter [A]
(a) Pore size distribution
(b) SEM micrographs
Figure 6.5: (a) Pore size distribution of Nb-doped TiO2 samples with different Nb content. All
samples were treated at a) 800 ◦ C and b) 700 ◦ C. Pore size distribution derived from
the N2 physisorption measurements applying the BJH method. (b) SEM micrograph of
a) Ti1 -700, b) Ti0.95 Nb0.05 -700, and Ti0.9 Nb0.1 -700.
quantity of Ti1-800 is converted into the rutile phase, the degree of conversion decreases with
increasing Nb content.
Figure 6.4b shows the XRD results of the Ti0.5 Nb0.5 and Nb1 reference samples. The Nb1
sample is fully converted into Nb2 O5 at both calcination temperatures. The diffraction patterns
show sharp diffraction peaks suggesting that these are highly crystalline samples. Detailed
analysis of the XRD pattern of the Ti0.5 Nb0.5 sample reveals the presence of anatase and rutile,
of TiNb2O7 oxide, and of a minor amount of Nb2 O5 . At lower calcination temperature the
diffraction peaks of the mixed oxide TiNb2 O7 are rather weak, and sharp well-defined peaks
of the anatase and rutile phases can be detected. When increasing the calcination temperature
to 800 ◦ C, a highly crystalline TiNb2O7 oxide phase is observed. The poor crystallinity of
the TiNb2O7 oxide in the Ti0.5 Nb0.5 -700 sample is in good agreement with the results and the
interpretation of the TG-MS data demonstrating the decomposition of the pure Ti-precursor
at lower temperatures and the decomposition of a mixed precursor at elevated temperatures.
Arbiol et al. [172] explained the phase separation at higher contents of Nb ions by the reduction
of stress by removing Nb5+ ions from the lattice. However, the high rutile-to-anatase ratio
cannot be explained only by removing Nb5+ ions from the lattice, as based on the data obtained
with Ti1 -700 sample, the anatase phase was expected to form exclusively. Possibly, the number
of oxygen vacancies, which act as transformation centers is strongly increased compared to Ti1700 after removing the Nb from the lattice, and therefore, the anatase-to-rutile transformation
is favored. [172]
Characterization of the texture and the morphology
The obtained BET surface areas are summarized in Table 6.2. The surface area of the parti-
6.1 The synthesis of Nb-doped TiO2 nanoparticles by spray drying
93
Table 6.2: Specific BET surfaces areas (BET SA) and particle diameters derived from the N2 physisorption measurements for the differently treated samples with varying Nb content
and the respective treatment of the sample.
pretreatment
sample
600 ◦ C, 1h
BET SA [m2 /g]
Ti1
Ti0.995 Nb0.005
Ti0.99 Nb0.01
Ti0.95 Nb0.05
Ti0.9 Nb0.1
105
115
129
135
/
700 ◦ C, 1h
BET SA [m2 /g] d(BET) [nm]
56
51
60
69
90
28
30
26
22
17
800 ◦ C, 1h
BET SA [m2 /g] d(BET) [nm]
17
26
/
46
48
91
59
/
33
32
cles clearly increases with increasing Nb content, while it decreases with increasing calcination
temperature. Thus, by varying the calcination temperature or the Nb content the surface area
of the particles can be tuned from 40 to 140 m2 /g. A mean particle diameter of the different
powders can be calculated from the measured surface area values by assuming spherical particles (Tab. 6.2). Based on this simple calculation, mean particle diameters of 15 - 60 nm were
derived, which are in rather good agreement with the crystallite diameters obtained by XRD.
It should be mentioned that the density of the anatase particles was used to calculate the mean
particle size of the powders. The evolution of the pore size distribution of the samples calcined
at 700 and 800 ◦ C is presented in Fig. 6.5 The data are derived from the N2 physisorption
experiments applying the BJH equation. The sample Ti1-700 shows two broad signals: smaller
pores with a pore size maximum at 100 Å and a less intense signal in the range of 400 - 1000
Å can be observed. After treatment at 800 ◦ C the peak detected at 100 Å disappears and the
intensity of the other peak is significantly decreased. Obviously, even a small amount of Nb
changes the pore size distribution and increases the number of smaller pores with a diameter
of smaller than 200 Å. Further incorporation of Nb up to 10 at % led to even smaller pores.
As shown in Fig. 6.5 the intensity of the maxima is changed by changing the calcination temperature. The presence of these small pores can be correlated with the formation of smaller
particles. A change in the evolution of the pore size can be seen for the Nb0.5 Ti0.5 -800 sample,
for which the pore diameter increases, while the overall number of pores decreases reflecting
the trend of the N2 physisorption measurements. Characterization of the materials by means
of SEM (Fig. 6.5) revealed that the calcined materials consist of collapsed spherical particles. It is known that spray drying led to spherical particles due to fast evaporation of solvent,
which collapsed during calcination due to volume contraction resulting from decomposition
of the precursors. [180] Furthermore, Nb concentration does not show significant influence on
the morphology of the obtained composites, and all the samples show similar morphologies as
indicated by SEM studies.
Probing the electronic structure by XPS and UV-Vis
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
a)
16
b)
Nb 3d
Ti
Nb
-700
Ti
Nb
-700
0.99
c)
14
458.4 eV
0.05
Nb
0.9
206.8 eV
3/2
0.01
0.95
Ti
Ti 2p
5/2
-700
0.1
Ti
Nb
-700
Ti
Nb
-700
0.99
0.01
0.95
CPS [a.u.]
Ti
0.05
Nb
0.9
-700
0.1
XPS-derived Nb:Ti ratio
94
12
10
8
6
4
2
214
212
210
208
206
204
binding energy [eV]
202 470 468 466 464 462 460 458 456 454
binding energy [eV]
0
2
4
6
8
10
12
theoretical Nb doping [at.%]
Figure 6.6: Results of the XPS analysis of samples with Nb content from 1 to 10 at% calcined at
700◦ C. a) Nb 3d XP spectra, b) Ti 2p XP spectra, and c) the measured (solid line) and
the ideal Nb/Ti ratio (dashed line) based on the XPS analysis vs. the theoretical Nb
doping.
The analysis of the XPS survey spectra (not shown) demonstrated that the sulfate ions had been
removed completely from the sample surface by calcination at 700 ◦ C, as no S 2p signal was
detected. The XP spectra of the undoped and Nb-doped TiO2 powder samples prepared with
different Nb contents (0, 0.01, 0.05, and 0.1), which were annealed at 700 ◦ C, are shown in
Fig. 6.6. Based on the peak positions, the differences of the Ti 2p3/2 and the Nb 3d5/2 binding
energies and the line shapes, the presence of Ti4+ species with a Ti 2p3/2 binding energy of
458.3 - 458.5 eV and the presence of Nb5+ ions with a Nb 3d5/2 binding energy of 206.8 eV is
clearly detected. [174] Within the XPS information depth there is no evidence for Ti3+ species,
which are usually observed at 1.8 eV lower binding energy than that of Ti4+ . [195] Furthermore,
there is no evidence for the presence of Nb4+ species, which are normally observed at binding
energies below 206.5 eV. [165,196] By deriving the Nb:Ti atomic ratio from the Ti 2p3/2 and the
Nb 3d5/2 peaks a slight enrichment of the Nb ions at the sample surface can be detected as
shown in Fig. 6.6c.
It is well known that the doping of heteroatoms can change the band gap of TiO2 . In Fig. 6.7
the results of the UV-Vis measurements of the undoped, the Ti0.995 Nb0.005 , and the Ti0.95 Nb0.05
samples calcined at 600, 700, and 800 ◦ C are shown. From these spectra the band gap values
can be derived, and unoccupied electronic states can be detected. The measured band gap of the
Ti1 samples amounts to 3.2 eV, which is in good agreement with reported values of the anatase
modification. It shifts to 3.0 eV corresponding to the rutile modification when calcining at 800
◦ C. The band gap of the doped TiO sample is slightly shifted to lower values with increasing
2
calcination temperature. The changes in the magnitude of the band gap can be explained by
the different anatase-to-rutile ratios of the samples: the undoped sample was converted into
6.1 The synthesis of Nb-doped TiO2 nanoparticles by spray drying
Ti
Ti
Ti
95
1
Nb
0.995
Nb
0.95
0.005
0.05
F(R) [a.u.]
600°C
700°C
800°C
300
350
400
450
500
550
600
wavelength [nm]
Figure 6.7: UV-Vis results of the Ti1 , the Ti0.995 Nb0.005 , and the Ti0.95 Nb0.05 precursor powders
calcined at 600, 700, and 800 ◦ C.
the more stable rutile modification, whereas the degree of phase transformation of the doped
sample is low due to the hindered phase transition. Therefore, the shift in the band gap is less
pronounced. In case of the samples calcined at 700 ◦ C a slight blueshift of the adsorption edge
of the low doped Ti0.995 Nb0.005 -700 sample and a slight redshift of the Ti0.95 Nb0.05 -700 sample
compared with the adsorption edge of the undoped titania Ti1-700 is observed. It is important
to mention that no additional electronic states giving rise to electronic transitions were found
at higher wavelength for all samples.
Discussion
The characterization of the structural properties and the texture analysis of the synthesized
powder materials are in good agreement. The XRD results show that the crystallite size of the
materials as well as the particle size derived from BET measurements decrease with increasing
Nb content. The differences in the XRD- and BET-derived crystallite sizes can be explained by
agglomeration effects influencing the physisorption results. However, the trend in both cases
is clearly maintained. The inhibition of grain growth and the hindered anatase to rutile phase
transformation are in good agreement with other studies using sol-gel methods [167,173] or laserinduced pyrolysis [172] for the preparation of Nb-doped TiO2. The inhibition of grain growth is
mostly explained by stress introduced into the lattice by the slightly larger Nb5+ ions, [172,178]
and the different phase transformation behavior is explained by the lower oxygen vacancy concentration acting as transformation centers. [19,178] On the other hand, Pittmann and Bell, [197]
who studied titania-supported niobia using Raman spectroscopy, pointed out that the formation of strong Nb-O-Ti bonds at the surface inhibits the Ti movement and hinders the phase
transformation from anatase to rutile, because the Ti mobility is required to initiate the phase
transformation. [198] These Nb-O-Ti bonds are formed, when the surface of titania is covered
by NbOx islands. [197] However, in the present study it is not likely that amorphous NbOx is-
96
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
lands are present on the surface of samples with low Nb content. This is clearly supported by
the increase of the refined cell parameter with increasing Nb content and the absence of any
feature in the XRD and Raman spectra, which can be found in case of high Nb content. The
increase of the cell parameters verifying Vegard’s law is a strong hint for the effectiveness of
doping as pointed out by Hirano et al. [191] Combining the results of the XRD, Raman, and BET
analysis, it is highly likely that Nb is incorporated in the lattice of TiO2. Thus, using the simple
spray drying approach, Ti ions are substituted by Nb ions at low doping levels, and solid solutions are formed. The chemical oxidation state of Nb derived by the XPS measurements is 5+.
This is in good agreement with other studies using sol-gel methods [178,198] or other preparation
procedures. [162,172] Furthermore, the XPS analysis did not reveal any evidence of Ti3+ , and
the UV-Vis measurements did not show any additional absorption. It is therefore reasonable
to assume that charge compensation is achieved by the formation of Ti vacancies according to
equation 6.2.
The blueshift of the absorption edge of the Ti0.995 Nb0.005 -700 is in good agreement with previous studies on Nb-doped TiO2. Density functional theory calculations of the electronic structure of the pure and the Nb-doped anatase by Liu et al. [199] showed that the difference between
the top of the valence band and the Fermi level is larger in doped samples. As a result, the
electron transition from the valence band to the conduction band requires more energy, which
is clearly consistent with the observed blueshift in the UV-Vis spectra. Furthermore, Chandiran
et al. [192] recently observed a similar blueshift of the absorption edge in low content Nb-doped
TiO2. They ascribed this blueshift to the Burstein-Moss effect, and the higher visible light
transparency and enhanced charge collection efficiency resulted in an increase in power conversion in dye-sensitized solar cells. [192] A redshift in higher doped samples was measured
by Lü et al. [169] They used Nb-doped TiO2 with loadings above 2.5 mol% Nb to enhance the
performance of the photoanode in a dye-sensitized solar cell and observed a small redshift of
the adsorption edge upon Nb doping, which is in agreement with our measurements. However,
the redshift in the higher doped Ti0.95 Nb0.05 -700 may also be ascribed to the presence of the
beta-TiO2 phase, which was detected by XRD and Raman measurements.
Finally, the advantages of the spray drying process compared to the conventional methods have
to be addressed. Ruiz et al. [198] studied solid solutions of Nbx Ti1-x O2 prepared by a sol-gel
method with a Nb content of 0 < x < 0.1 and found that the solubility limit of Nb in the
anatase phase is higher than 0.1. However, the XPS analysis revealed that there is a large discrepancy between the concentration of Nb at the surface and the theoretical loading. Even at
low calcination temperature the measured surface concentration was two times higher than the
theoretical loading. Similar discrepancies between theoretical loading and surface concentration were found by Chandiran et al. [192] , who used a rather complicated hydrothermal process
and subsequent annealing treatment to synthesize Nb-doped TiO2 materials. Even though there
is also a minor discrepancy between the theoretical Nb content and the actual amount of Nb in
the near-surface region detected by XPS, it should be pointed out that the surface segregation is
6.1 The synthesis of Nb-doped TiO2 nanoparticles by spray drying
97
much less pronounced, and at calcination temperatures of 700 ◦ C and a theoretical Nb content
of 10 at% the detectable amount of Nb at the surface is less than 15 %. Thus, more Nb should
remain in the bulk of the material effectively doping the titania particles. The differences in
segregation behavior are assumed to be due to the rapid evaporation in the spray drying process
compared to the slow gelation process in the sol-gel approach. Furthermore, XRD and Raman
results provide evidence that there is no additional phase, and it is not likely that NbOx islands
are present on the surface of the material. Hence, Nb is well distributed in the bulk of the
spray-dried materials. In summary, spray drying is considered an efficient and advantageous
synthetic method leading to high levels of Nb doping in TiO2 .
6.1.3 Conclusions
Spray drying was applied as efficient, scalable, and reliable method for the synthesis of doped
metal oxides using Nb-doped TiO2 as a test system. The suppression of the ion mobility by the
fast evaporation of water during the spray drying process is assumed to favor the homogeneous
distribution of the Nb ions. The subsequent calcination in the temperature range from at 600 to
800 ◦ C results in crystalline materials with narrow pore size distributions. Subsequent to calcination, the fully oxidized Nb ions were found to be homogenously distributed in the crystalline
TiO2 powder materials, and Nb surface segregation had occurred only to a minor extent in spite
of the high-temperature treatment. By using different levels of Nb doping, the particle size and
the anatase-to-rutile ratio can be easily adjusted: with increasing Nb content up to Ti0.9Nb0.11
the particle size decreases continuously, and the transformation of anatase into rutile at high
temperatures is more kinetically hindered. The spray drying approach combined with a controlled high-temperature calcination is considered a promising technique for the homogeneous
doping of TiO2 and other oxides with various metal ions. Future work should focus on the
development of more suitable precursor molecules.
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
98
6.2 TiO2-polyheptazine hybrid materials
Abstract2
Nb-doped TiO2 prepared by spray drying exhibits enhanced photoelectrochemical performance
both in its bare form (under UV irradiation) and when used as an electron collector in hybrid
TiO2-polyheptazine photoanodes for water photooxidation under visible (λ > 420 nm) light.
2 To
be published as ”Beneficial effect of Nb doping on the photoelectrochemical properties of TiO2 and TiO2 polyheptazine hybrids”, B. Mei, H. Byford, M. Bledowski, J. Strunk, M. Muhler, R. Beranek.
6.2 TiO2-polyheptazine hybrid materials
99
6.2.1 Short Introduction
Direct conversion of solar energy into chemical energy by sunlight-driven water splitting into
hydrogen and oxygen in a photoelectrochemical cell is one of promising strategies to secure
the future supply of clean and sustainable energy. [200] Although very high solar-to-hydrogen
efficiency of up to 18 % have been already achieved, such cells were typically based on multijunction photoelectrodes using expensive materials. [201,202] An economically viable solution,
therefore, requires the development of new, low-cost materials. Notably, due to the complex
chemistry involved in four-electron oxidation of water to dioxygen, [203] the major challenge in
photoelectrochemical water splitting is the development of cheap, efficient, and stable photoanodes. The research efforts have mostly focused on low-band gap transition metal oxides, like
WO3 (2.6 eV, ∼ 480 nm) [204,205] or α -Fe2 O3 (2.1 eV, ∼ 590 nm), [206–208] which combine light
absorption in the visible with high stability. However, their rather positive conduction band
edge normally translates into the need to apply significant external electric bias or use tandem
cell configurations in order to achieve proton reduction at the counter electrode.
An attractive alternative is to use a nanocrystalline layer of a stable wide-band gap oxide (e.g.
TiO2 ) sensitized by a visible light absorbing dye coupled to a co-catalyst for water oxidation. [209–211] In this approach the oxide is used essentially as a collector for electrons injected
from the dye, whereby the positive charges (holes) are channeled to a colloidal co-catalyst for
water oxidation (IrO2 nanocrystals). Interestingly, such hybrid photochemical assemblies not
only resemble the Photosystem II of green plants by exhibiting kinetic charge separation, [212]
but they also take advantage of the relatively negative potential of the conduction band edge of
the wide-band gap metal oxide (e.g. -0.15 V vs. RHE for TiO2) which promises significantly
reduced need for external bias to drive complete water splitting. Obviously, the stability issue
plays a crucial role in this type of devices since many dyes are not stable enough to survive the
harsh conditions during water oxidation. [209,211,213]
Recently, we have been developing photoanodes (Figure 6.8) based on a novel class of visible
light photoactive inorganic/organic hybrid materials - TiO2 modified at the surface with polyheptazine (also known as ”graphitic carbon nitride”). [53,54,214,215] The most attractive feature
of these inorganic/organic hybrids is the high thermal (up to 550 ◦ C in air) and chemical stability of polyheptazine-type compounds as compared to conventional organic dyes. Notably, we
have shown that the optical absorption edge of the TiO2 -polyheptazine hybrids is red-shifted
into the visible (2.3 eV; ∼540 nm) as compared to the band gap of both of the single components - TiO2 (3.2 eV; ∼390 nm) and polyheptazine (band gap of 2.9 eV; ∼428 nm), which is
due to the formation of an interfacial charge-transfer complex between polyheptazine (donor)
and TiO2 (acceptor). In other words, the direct optical charge transfer leads to generation of
electrons with a relatively negative potential in the conduction band of TiO2, while the holes
photogenerated in the surface polyheptazine layer can drive water photooxidation, as evidenced
100
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
Figure 6.8: a) Hybrid photoelectrode consists of a layer of nanocrystalline TiO2 powder pressed
onto an ITO-glass, sintered, and subsequently modified at the surface with polyheptazine; b) Potential scheme (at pH 7) illustrating the visible light photoactivity of TiO2 polyheptazine hybrids based on direct optical charge-transfer excitation of an electron
from the polyheptazine HOMO (valence band) into the conduction band of TiO2 .
by visible light-driven evolution of dioxygen on hybrid electrodes modified with IrOx or CoOx
nanoparticles acting as oxygen evolution co-catalysts. [53,54,214,215]
An attractive strategy, so far untried, for improving such hybrid photoelectrodes consists in tuning the electronic properties of the electron-collecting component, i.e. of TiO2 in our case. In
this context, particularly doping of TiO2 with niobium represents an interesting option since
it is known that the replacement of tetravalent Ti with pentavalent Nb can shift the Fermi
level closer to the conduction band edge of titania and thus increase significantly its conductivity. [166,198,216,217] The enhanced mobility of photogenerated electrons would then lead to
improved charge collection and diminished recombination. Indeed, some promising results
along these lines have been recently reported for Nb-doped TiO2 electrodes used either for
water photooxidation [218,219] or as photoanodes in dye-sensitized solar cells. [169,192,192] However, the beneficial effect of Nb doping has not been observed always, [166] and it seems clear
that the performance of Nb-doped TiO2 is highly dependent on Nb content and its preparation
conditions. [169,192,192,218,219]
We have recently developed a continuous spray drying approach to synthesize of Nb-doped
TiO2 materials from aqueous solution (Chapter 6.1). [220] This method is not only efficient and
scalable but allows also for highly homogeneous distribution of Nb in crystalline TiO2, [220]
which is important in order to avoid formation of undesired defects. [217] Herein, we show that
Nb-doped TiO2 prepared by spray drying exhibits enhanced photoelectrochemical performance
both in its UV-only-active bare form and also when used as an electron collector in hybrid
photoanodes for water photooxidation under visible light.
Experimental
Nb-doped TiO2 powders (0-10 at%) were prepared by spray drying followed by calcination at
6.2 TiO2-polyheptazine hybrid materials
101
700 ◦ C. [220] Working electrodes of catalyst materials were prepared on cleaned ITO glass substrates by doctor blading of powder pastes as described previously. [53] The powder pastes were
prepared by ultrasonication of a mixture of 0.37 ml ethanol, 0.13 ml titaniumtetraisopropoxid
and 100 mg of sample. Modification of the working electrodes with polyheptazine was performed by a procedure described in detail elsewhere. [53] In short, the electrode was statically
heated to 450 ◦ C in a two compartment glass flask containing the electrode and 1 g of urea. The
electrode was further modified with IrO2 by hydrolysis of Na2 IrCl6 . Characterization of the
as-received materials was performed using Raman spectroscopy using a Nexus FT-NIR spectrometer equipped with a nitrogen-cooled germanium detector and a 1064 nm laser (Nd:YAG)
provided by Thermo Fisher Scientific. UV-vis diffuse reflectance spectra (DRS) were recorded
in a Perkin-Elmer Lambda 650 UV-vis spectrometer equipped with a Praying-Mantis mirror
construction using MgO as the 100 % reflection reference. Photocurrent measurements were
performed with a SP-300 BioLogic potentiostat and a three-electrode cell using a Pt counter
and a Ag/AgCl (3M KCl) reference electrode in a phosphate buffer solution (pH 7). A tuneable
monochromatic light source with a 150 W Xenon lamp was used for sample irradiation. IPCE
data for each wavelength were obtained according to the equation IPCE (%)=(iph hc)/(λ Pq)100,
where iph is the photocurrent density, h is Planck’s constant, c the velocity of light, P the light
power density, λ the irradiation wavelength, and q the elementary charge. The measurement
of quasi-Fermi level of electrons by the suspension method of Roy was carried out in a two
electrode setup with a 150 W Xenon lamp equipped with heat absorbing IR filter. The pH dependence of the potential was recorded in the presence of the pH independent electron acceptor
methyl viologen as previously reported. [221] Oxygen evolution was measured in a phosphate
buffer (pH 7) by an OxySense 325i oxygen analyzer in a two-compartment cell. [53]
6.2.2 Results and Discussion
All investigated photoelectrodes consisted of ∼ 6 µ m thick porous crystalline powder layers
pressed onto an ITO-glass substrate (Fig. A15). A Raman spectroscopy analysis of Nb-doped
samples prepared by spray drying and calcined at 700 ◦ C confirmed that as expected exclusively
the anatase structure was obtained for a large concentration range of Nb (0.1-10 at% Nb with
respect to Ti). [220] The optical absorption properties of undoped and Nb-doped materials were
nearly the same, exhibiting a band gap of 3.2 eV (Fig. A16). The photocurrent response of Nbdoped TiO2 electrodes was strongly dependent on the doping concentration as shown by the
wavelength-dependent photocurrent measurements (Figure 6.9a; raw data in Fig. A17). The
Nb-doped TiO2 with 0.1 at% Nb content (”Nb01”) has shown the best UV-light photoresponse
exhibiting the monochromatic incident photon-to-current efficiency (IPCE) values higher by
ca. 25 % as compared to a pristine TiO2 reference, whereby a maximum is observed at 350
nm. The drop of the photocurrent at low wavelengths is due to enhanced recombination which
is a typical consequence of photogeneration of charges in a thin surface layer upon irradiation
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
102
12
b)
TiO
a)
2
TiO -PH (440nm)
2
Nb01
10
Nb01-PH (440nm)
Nb02
Nb10
I [mA]
IPCE [%]
Nb5
8
6
0.06 mA
4
2
0
300
350
400
450
500
-0.4
0.50
c)
TiO
10
6
4
2
Nb01 + IrO
]
-3
2
Nb01-PH
Oxygen [mg dm
IPCE [%]
0.2
Nb01-PH + IrO
0.45
TiO -PH
0
300
d)
2
Nb01
8
0.0
0.4
0.6
0.8
potential [V vs. Ag/AgCl (3M)]
wavelength [nm]
12
-0.2
2
light off
>420nm
2
0.40
0.35
0.30
light on
>420nm
0.25
0.20
0.15
350
400
450
500
wavelength [nm]
550
600
0.10
0
15
30
45
60
75
90
105 120 135 150 165
t [min]
Figure 6.9: a) Photoaction spectra measured in a phosphate buffer solution (pH 7) at 0.5 V vs.
Ag/AgCl for a TiO2 reference and Nb-doped TiO2 materials with different Nb concentrations (0.1-10 at%). b) Potential-dependent and c) wavelength-dependent photocurrents measured at different excitation wavelength for the polyheptazine-modified TiO2
reference and Nb-doped TiO2 materials. d) Oxygen evolution measured at 0.5 V vs.
Ag/AgCl in the liquid phase upon visible light irradiation for the two materials. IrO2
was used as an oxygen evolution co-catalyst.
with short wavelength light. When increasing the Nb concentration to 0.2 at% Nb a decrease
in photocurrent was observed, which is further pronounced with increasing Nb concentration.
This can be explained by an increase in the number of defects within the TiO2 lattice which
will increase the recombination rate.
A similar trend has been recently observed by Schmuki et al. [218,219] However, it should be
noted that Nb-doped TiO2 nanotubes investigated by Schmuki et al. [218,219] contained significant portion of rutile, whereas our samples contain only the anatase phase and the effect of
Nb incorporation into the TiO2 lattice can be evaluated without any interference of structural
changes. We can therefore conclude that in our case the observed photocurrent enhancement
is chiefly due to Nb incorporation into the TiO2 lattice, which increases the mobility of photogenerated electrons and diminishes the recombination. Interestingly, under potentiodynamic
6.2 TiO2-polyheptazine hybrid materials
103
conditions only a small improvement of the photocurrents can be observed at high bias potentials, and first at more negative potentials the improvement becomes obvious at the 0.1 at% Nb
electrode (Fig. A18a). This is in line with our assumption that the enhancement is due to improved electron transport properties, which can be expected to make itself apparent particularly
at low bias potentials. Furthermore, we exclude the possibility that the enhancement is simply
due to the shift of the band edges in the Nb-doped TiO2 . The photocurrent onset potential was
determined to be ∼ -0.5 V vs. Ag/AgCl (at pH 7) for both the pristine TiO2 and Nb-doped TiO2
(Fig. A18b). Similarly, the quasi-Fermi level of electrons (i.e. the electrochemical potential of
photogenerated electrons), as determined by the suspension method of Roy, [221] shows only a
very small difference (-0.56 V vs. NHE for pristine TiO2 and -0.59 V vs. NHE for Nb-doped
TiO2 at pH 7; Fig. A19). Though this difference is well within the experimental error of this
measurement, it cannot be completely excluded that the difference is precisely due to the presence of new donor levels in Nb-doped samples which leads to an increase of conductivity and
shifts thereby also the quasi-Fermi level closer to the conduction band edge. [198,216,217]
Modification with a thin (1-3 nm) layer of polyheptazine using a chemical vapor deposition
method [53,54,214,215] did not significantly change the morphology of the electrodes as evidenced
by SEM analysis (not shown). The optical absorption edge has been shifted to 2.3 eV upon
modification (Fig. A20). As compared to polyheptazine-free photoanodes, the photocurrent
onset, measured under monochromatic visible (λ = 440 nm) light irradiation, has been shifted
anodically by 0.2 V to -0.3 V vs. Ag/AgCl for hybrid photoelectrodes based on both pristine
and Nb-doped TiO2 (Figure 6.9b). This shift is in agreement with our previous observations
and can be ascribed to enhanced recombination rate due to slower kinetics of holes and their
accumulation in the polyheptazine layer. [214] However, more importantly, the Nb-doped samples show much better performance, particularly at low-bias potentials. This can clearly be
attributed to the enhanced mobility of electrons in the Nb-doped TiO2 which renders the electron transport into the ITO substrate more efficient and the charge recombination less likely.
In accord with the behavior of polyheptazine-free electrodes, also in case of hybrids the best
performing photoanodes were those with 0.1 at% Nb. The enhanced performance of Nb-doped
samples is also clearly apparent from the IPCE photoaction spectra (Fig. 6.9c). The beneficial
effect of Nb doping is more pronounced at lower-energy light (longer wavelengths) where the
influence of electron mobility in Nb-doped TiO2 can be expected to exert major influence on
recombination processes. As compared to irradiation by short wavelength light where nearly
all charges are photogenerated in a small volume near the ITO glass (electrodes are irradiated from the backside - through the ITO glass), the longer-wavelength light penetrates deeper
into the porous electrode and renders the transport of electrons more important. Accordingly,
at wavelengths longer than 450 nm the IPCE values are practically doubled as compared to
hybrid photoanodes based on pristine TiO2.3 After depositing IrO2 nanoparticles as a water
3 It has to be noted that the relatively smaller enhancement of IPCE values as compared to photocurrents measured
under potentiodynamic conditions (Fig. 6.9a) is due to the fact that the IPCE values were measured under
potentiostatic conditions where the photocurrent transients show a strong spike-like behavior. [214]
104
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
oxidation co-catalyst, polychromatic visible light (λ > 420 nm) induced dioxygen evolution
on Nb-doped TiO2 -polyheptazine hybrids (Fig. 6.9c), while the unmodified 0.1 wt% Nb-doped
TiO2 did not show any oxygen evolution. The corresponding photocurrent transient (Fig. A21)
reveals insufficient stability, which is mainly related to poor coupling to the IrO2 co-catalyst
that is far from optimum in this case. Our current work is mainly focused on solving this issue.
6.2.3 Conclusions
In summary, Nb-doped TiO2 prepared by spray drying exhibits enhanced photoelectrochemical
performance both in its bare form and when used as an electron collector in visible light-active
TiO2-polyheptazine hybrid photoanodes for water photooxidation. The optimum Nb content
was 0.1 at%, and the photoelectrochemical analysis indicates that the enhancement is chiefly
due to enhanced mobility of electrons in the Nb-doped samples. Finally, our findings highlight
the importance of modulation of electronic properties of electron-collecting substrates in hybrid
photoelectrochemical devices.
6.3 Nitrogen-modified Nb-doped TiO2: Towards red TiO2
105
6.3 Nitrogen-modified Nb-doped TiO2: Towards red TiO2
Abstract4
Nb-doped TiO2 prepared by spray drying were subsequently treated by ammonia to obtain
Nb/N co-doped TiO2 . The as-received Nb-doped TiO2 and ammonia-modified Nb/N co-doped
TiO2 were tested for the photocatalytic activity for the degradation of organic dyes. Surface
adsorption of the dye was found to significantly affect the photocatalytic activity of the photocatalysts and incorporation of up to 5 at% of Nb into TiO2 exhibit beneficial degradation properties for the degradation of methylene blue (MB). The photocatalytic activities of the Nb/N
co-doped TiO2 catalysts were small. However, an optimum in the photocatalytic activity of the
Nb/N co-doped TiO2 materials was determined for the ammonia-modified Nb-doped TiO2 with
10 at% of Nb. By means of elemental analysis and X-ray photoelectron spectroscopy a correlation between the Nb and nitrogen content incorporated into the TiO2 materials was observed
indicating that nitrogen incorporation is improved in the Nb-doped TiO2 materials. Based on
this preliminary results possible strategies are discussed to further improve the nitrogen incorporation into TiO2-based materials and enhance the photocatalytic activity under visible light
irradiation.
4 To
be published as ”Nb/N co-doped TiO2 : Improvement of the visible light activity by effective charge compensation”.
106
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
6.3.1 Short Introduction
Recently, co-doping of TiO2 by non-metal/non-metal or metal/non-metal co-incorporation into
the anatase TiO2 structure attracted some interest. [32,36,222–227] Several examples discussed in
literature demonstrate that the visible light photocatalytic activity is mainly found to be lower
than that of pure TiO2 under UV irradiation, although visible light absorption can be achieved
by doping. This can be ascribed to different problems occurring for doped TiO2 : [32,55]
• High formation energies and differences in ionic radius prevent high doping levels. [32]
• The recombination of charge carriers is increased due to localized states in the band gap
acting as recombination centers. [228]
• The migration of charge carriers is limited and the oxidation power of the localized states
is lower. [32]
Synergetic effects between two complementary doping species are believed to reduce the defects introduced during doping by charge compensation, and thus partially overcome the problems of single-doped TiO2 . [226] An interesting example of successful co-doping of TiO2 was
recently presented by Liu et al. [224] They synthesized a TiO2 material with boron-enriched
surface, which was subsequently treated in ammonia atmosphere yielding a red anatase. Characterization of the red anatase revealed that no Ti3+ -related defects were created at the surface
of the sample during the synthesis, and boron was shown to be present in BN-like environment.
Finally, the B/N co-doped TiO2 was shown to exhibit enhanced visible light performances in
a photoelectrochemical cell. [224] Another study by Ikeda et al. [36] with Cr/Sb co-doped TiO2
showed that the ratio of Cr and Sb is of a high relevance to obtain a visible light active material.
While the material was black in color and photocatalytically inactive for high Cr to Sb-ratios,
the color changed to orange and visible light activity in the oxygen evolution reaction in presence of a sacrificial agent was observed. [36]
A more recent theoretical study based on density functional theory suggest that certain combinations of metal/non-metal co-incorporation into anatase TiO2, or more generally spoken
alloying donor-acceptor combinations with anatase TiO2 , are promising materials for photoanodes in a photoelectrochemical cell. [229] According to the calculations they should fulfil the
main requirements of a photoanode such as an appropriate optical band gap and suitable valence and conduction band-edge positions with respect to the redox potential for hydrogen and
oxygen evolution. [229] Within this study two different concentration regimes and several donoracceptor combinations were calculated. Finally, four different combinations were suggested.
For the low concentration alloys, where 3.1 or 6.2 % of oxygen were replaced by electron acceptors depending on the donor atom, a combination of Mo and N or W and N appeared to be
good choices. In the high alloy concentration regime Nb/N and Ta/N combinations passed the
selection criteria.
6.3 Nitrogen-modified Nb-doped TiO2: Towards red TiO2
107
So far, only one experimental study was performed aiming at an improvement of the visible
light absorption by co-alloying TiO2 with Nb and N or Ta and N. [222,230] Indeed, it was shown
that Nb/N co-alloying enhanced the photocatalytic activity towards methylene blue decolorization. [222] A reddish Nb/N co-alloyed TiO2 sample was obtained after ammonia treatment of a
material with a molar Nb/Ti ratio of 1:3. Within this study the Nb/Ti ratio was fixed, and only
the effect of slight variations in the sample pretreatment were presented. [222] Obviously, the
B/N and the Nb/N co-doped TiO2 materials synthesized by Liu et al. [224] and Breault et al. [222]
exhibit similar absorption properties. Niobium is well known to segregate to the surface. Therefore, a Nb gradient within the anatase particle is feasible, which will result in a similar structure
than the B/N co-doped TiO2. [172,198] Ta/N co-doped TiO2 materials with undefined Ta and N
concentrations were used for photoelectrochemical water oxidation. [230] By means of TEM investigations Hoang et al. [230] observed a suppression of the formation of an amorphous layer
during the nitridation process and noted that the enhanced photoelectrochemical performance
of the Ta/N co-doped TiO2 is due to less recombination centers. Therefore, it is reasonable to
assume that co-doping is a suitable strategy to enhance the photocatalytic activity compared to
single-doped materials. Due to the above mentioned reasons further investigations with Nb/N
co-doped TiO2 are necessary to understand the effect of co-doping and to evaluate the photocatalytic performance of co-doped materials. Here, a larger Nb concentration range is explored
in order to obtain a better match of Nb and N. Again spray drying was used to synthesize Nbdoped TiO2 . Based on the experimental results and theoretical calculations samples with Nb/Ti
ratios of up to 1:3 were prepared. [222,229] . For a comparison with the experimental results presented by Breault et al. [222] the photocatalytic activity with respect to decolorization of organic
dyes was used as a test reaction to evaluate the photocatalytic properties of the as-received
Nb-doped TiO2 and the subsequently NH3 -modified Nb-doped TiO2 samples.
Experimental
Nb-doped TiO2 was prepared by spray drying as explained in this chapter. All samples were
calcined at 700 ◦ C to obtain the anatase structure exclusively using the described heating procedure. [220] The ammonia treatment was performed at different temperatures in a vertical furnace
using pure ammonia with a flow rate of 150 ml/min. The design of the furnace allows for a homogeneous distribution of the NH3 , and thus a homogeneous nitrogen doping can be achieved.
Photocatalytic degradation of organic dyes was performed by suspending 50 mg of catalyst
in 85 ml of a 10 ppm methylene blue (MB) (20 ppm methyl orange (MO)) solution. The
absorption-desorption equilibrium was obtained by stirring for 60 min in the dark. Irradiation
of the suspension was performed with a 150 W Xe lamp equipped with an IR filter and a 320
nm cutoff filter. The solution was permanently saturated by bubbling synthetic air through the
suspension. The degradation of the dyes was monitored by taking 2 ml of the solution at constant time intervals. The catalyst was removed by filtering, and the characteristic absorbance of
the dye was recorded by means of UV-Vis spectroscopy.
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
108
6.3.2 Results and Discussion
0min
1.0
1.0
6)
TiO
10min
1.0
2
20min
7)
Absorbance
0
C/C
0.6
0.4
2)
1)
0.2
0.8
50min
30min
50min
Absorbance
0.8
Nb5%
10min
20min
30min
0.8
0min
80min
0.6
0.4
80min
0.6
0.4
0.2
0.2
5)
3)
0
20
40
60
0.0
0.0
4)
0.0
400
80
500
600
700
0.6
4)
3)
0.4
2)
0.2
1)
40
60
80
100
120
Irradiation time [min]
140
160
TiO
20min
800
40min
60min
0.8
100min
0.6
0.4
0.2
Nb5%
20min
60min
160min
100min
0.6
0.4
0.2
0.0
300
700
0min
1.0
2
40min
Absorbance
C/C
0
8)
Absorbance
0.8
0.8
600
Wavelength [nm]
0min
1.0
1.0
20
500
Wavelength [nm]
Irradiation time [min]
0
400
800
0.0
400
500
Wavelength [nm]
600
700
300
400
500
600
700
Wavelength [nm]
Figure 6.10: First row: Results obtained for the MB degradation experiments (10 ppm MB). Second row: Degradation of MO (20 ppm solution) with different Nb-doped TiO2 samples. 1) TiO2 , 2) Nb 1 at%, 3) Nb 2 at%, 4) Nb 5 at%, 5) Nb 10 at%, 6) TiO2 , 400 nm
filter, 7) Nb 5 at%, 400 nm filter , 8) Nb 15 at%.
Initially the as-received Nb-doped TiO2 materials with different Nb concentrations were characterized regarding their structure and absorption characteristics and afterwards tested as photocatalysts. In short, characterization by means of Raman and UV-Vis spectroscopy revealed
that the desired anatase structure was obtained exclusively and that all samples posses an band
edge of 3.2 eV, which is usually observed for the anatase structure. [220] The activities towards
methylene blue (MB) degradation are shown in Fig. 6.10. Obviously, there is an increase in the
activity towards MB decolorization upon Nb doping of up to 5 at%. For the Nb-doped TiO2
with 10 at% the activity is still higher than the pristine TiO2 material. However, it is necessary to take the MB adsorption at the materials surface into account. Nearly no adsorption was
observed at the pristine TiO2, but with increasing Nb doping concentration the surface absorption of MB is getting more pronounced. This is evident by comparing the UV-Vis absorption
spectra measured during the degradation reactions of pristine TiO2 and Nb 5 at%. It is obvious
that ∼ 40 % of MB is pre-adsorbed at the Nb 5 at%, as in Ti0.95 Nb0.05 , surface at the beginning
of the reaction (Fig. 6.10). Pre-adsorbed MB might change the decolorization mechanism.
An indication of changes upon adsorption was obtained by studying MB decolorization under
visible light irradiation using a 400 nm cutoff filter. While no decrease in the characteristic
MB absorption spectra was observed for pristine TiO2 , the intensity of the MB peak maximum
at 664 nm is slightly decreasing upon irradiation (Fig. 6.10). The UV-Vis absorption spectra
show that the band gap of TiO2 samples is unchanged upon Nb doping (Fig. 6.7). Only UV
light can be absorbed, and electron excitation from the valence band to the conduction band to
6.3 Nitrogen-modified Nb-doped TiO2: Towards red TiO2
109
Table 6.3: Chemical composition of the spray-dried samples with different Nb content after NH3
treatment at 500 ◦ C in 150 ml/min for 4 h.
sample
TiO2
Ti0.99 Nb0.01
Ti0.98 Nb0.02
Ti0.95 Nb0.05
Ti0.9 Nb0.1
Ti0.85 Nb0.15
Ti0.8 Nb0.2
Ti0.75 Nb0.25
Ti content [wt%]
Nb content [wt%]
N content [wt%]
XPS(N/Ti)-ratio)
57.52
56.63
55.82
55.11
51.05
47.36
47.20
44.77
0.81
1.42
3.07
6.06
8.85
12.16
15.38
0.1
0.2
0.37
0.45
0.7
1.02
1.32
1.83
0.05
0.11
0.13
0.27
-
create electron-hole pairs is not feasible. Therefore, the decolorization of MB may be due to
visible light absorption of MB and charge injection into the conduction band of the material.
This sensitization seems to be favored upon MB adsorption.
To further study the influence of surface adsorption of the dye onto the photocatalytic activity of the samples the degradation experiments were performed with a different dye. Methyl
orange (MO) was chosen instead of MB (cationic dye) due to the anionic nature of MO. The
results of the degradation experiments are also included in Fig. 6.10. Upon Nb doping of TiO2
the activity towards the decolorization of MO is steadily decreasing with increasing Nb concentration. Additionally, a different trend in MO adsorption was observed. As shown by the
UV-Vis absorption spectra obtained at different reaction times for pristine TiO2 and Nb 5 at%
(Ti0.95 Nb0.05 ) (Fig. 6.10) MO is preferably adsorbed at the pristine TiO2 . The two presented
sets of photocatalytic dye degradation demonstrate that dye adsorption is indeed an important
factor considerably changing the activities of a material. The importance of adsorption can
be rationalized by the different types of degradation mechanisms. [231,232] While it is widely
believed that dye degradation is occurring by the attack of generated OH radicals, the direct
degradation by holes has to be taken into account. [232] OH radicals are mainly produced by the
reaction of surface adsorbed water with holes. The presented results show that direct degradation by holes is responsible for the degradation of MB at TiO2 -like samples. However, it
is evident that surface adsorption of dyes has to be considered when organic dye decolarization/degradation is used as a test reaction to measure photocatalytic activities. The indirect
oxidation of the dye by activated oxygen is neglected in this considerations due to the lower
oxidation potential. [232] To obtain further insight into the photocatalytic activities of the different Nb-doped TiO2 samples the determination of the total organic carbon content (TOC) during
photocatalytic degradation of the different dyes would be necessary.
Subsequently to calcination in synthetic air at 700 ◦ C the Nb-doped TiO2 materials with different Nb contents were treated in pure NH3 at different temperatures to obtain co-doped Nb/N-
110
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
TiO2. The materials described in the following section were obtained after ammonia treatment
at 500 ◦ C for 4 h. Elemental analysis of the samples revealed that with increasing Nb concentration the amount of nitrogen incorporated into the materials is increasing from 0.1 wt%
for the pristine TiO2 to ∼ 1.8 wt% for the Ti0.75 Nb0.25 (Table 6.3). According to the overall increase in nitrogen content the XPS analysis showed that the N/Ti-ratio at the surface of
the materials steadily increased (Table 6.3). Furthermore, the XPS analysis indicates that two
different species can be detected at binding energies of ∼ 396 eV and ∼ 400 eV (Fig. 6.11).
The latter signal is usually assigned to interstitial nitrogen (Ni ) and the low binding energy
signal is assigned to substitutional nitrogen (Ns ). [233] The intensity of both signals increases
and the ratio of Ns /Ni changes with increasing Nb content in the Nb/N co-doped TiO2 samples.
While Ns could be almost neglected for N-doped TiO2 the ratio of Ns /Ni is already almost 1
for the Nb/N co-doped Ti0.9Nb0.1 sample. The substitutional Ns compound was determined to
be the main nitrogen compound in the Nb/N co-doped Ti0.8 Nb0.2 sample. Elemental analysis
additionally confirmed that the Nb content in the different materials is still in good agreement
with the theoretical Nb concentrations (Table 6.3). After ammonia treatment with increasing
Nb content the color of the Nb/N co-doped samples is gradually changing from yellowish for
the ammonia-treated pristine TiO2 to dark green for the ammonia-treated Ti0.75 Nb0.25 . The
visually observed change in color is supported by an increase in the visible light absorption
at higher wavelength (600 - 800 nm) with increasing Nb content in the Nb/N co-doped TiO2
materials as determined by UV-Vis spectroscopy. Additionally, from the XRD measurements
it is evident that structural changes due to the ammonia treatment at 500 ◦ C can be neglected
as shown in Fig. A22. For all samples, regardless of the Nb- or N-content incorporated into the
TiO2 lattice, the anatase structure was obtained exclusively.
4) N-Ti
Nb
3) N-Ti
Nb
0.8
0.9
2) N-Ti
0.95
0.2
0.1
Nb
0.05
1) N-TiO
2
404
402
400
398
396
394
392
binding energy [eV]
Figure 6.11: XPS N 1s region scans of Nb-doped TiO2 materials with different Nb concentrations
treated at 500 ◦ C in 150 ml/min NH3 for 4 h. The theoretical Nb content in the
different materials was 1) 0 at%, 2) 5 at%, 3) 10 at%, and 4) 20 at%.
6.3 Nitrogen-modified Nb-doped TiO2: Towards red TiO2
111
The photocatalytic activity of the materials was tested by the degradation of MB and MO.
Surprisingly, none of the ammonia-modified materials exhibit any activity for the degradation
of MO and only a slight degradation of MB was detected. In the MB degradation tests the Nb/N
co-doped TiO2 sample theoretically containing 10 at% Nb (Ti0.9 Nb0.1 ) showed the highest
degradation rates (in this comparison differences in MB adsorption on the Nb/N co-doped
TiO2 samples were neglected). However, the photocatalytic activity towards degradation of
MB was still significantly lower for the ammonia-treated Ti0.9Nb0.1 sample compared with
photocatalytic activities of the pristine TiO2 and the Ti0.95 Nb0.05 samples.
For the Ti0.9 Nb0.1 sample showing the best photocatalytic activities in the applied test reactions
different ammonia treatment temperatures up to 700 ◦ C were tested. The photocatalytic activity
towards degradation of MB decreased with increasing temperature. X-ray diffraction (Fig.
A22) and XPS (Fig. 6.11) measurements revealed that the Ti0.9Nb0.1 sample treated at 700
◦ C was mainly converted into a the corresponding nitride, which certainly explains the loss in
activity. While the XRD results point to the complete transformation of the anatase phase into
the nitride (Fig. A22), the XPS results reveal that Ti in a TiO2-like and Nb in a Nb2 O5 -like
state are also present in the near surface region of the material. Interestingly, the Nb content as
determined by XPS was lower for the ammonia-modified Ti0.9Nb0.1 sample. The interdiffusion
of Ti in TiN/NbN single crystals is well-known, and it can be assumed that the decrease in the
surface concentration of Nb is due to the diffusion of Ti to the surface or Nb into the bulk of
the material. [234] For the B/N co-doped TiO2 samples prepared by Liu et al. [224] diffusion of B
to the surface of TiO2 was observed, thus reducing the defect formation during nitridation. The
opposite trend as observed for the Nb/N co-doped TiO2 might facilitate homogeneous doping
of the materials increasing the visible light absorption.
Strategies for improved nitrogen incorporation
The above presented results clearly show that there is a preferential incorporation of nitrogen
into the TiO2 lattice depending on the concentration of Nb in the Nb-doped TiO2 materials. The
improved nitrogen incorporation is considered to be necessary reduce the defect formation in
TiO2 during ammonia treatment by charge compensation, to increase the nitrogen content, and
to enhance the visible light response of N-doped TiO2 . [226] Thus, the results are in agreement
with the theoretical calculations by Yin et al. [229] and the experimental results by Breault et
al. [222] . Even though there is no clear indication of charge compensation by Nb in the Nb/N
co-doping TiO2 samples resulting in superior photocatalytic properties of the materials with
respect to visible light-driven photocatalytic reactions the performance of the Nb/N co-doped
Ti0.9 Nb0.1 sample exceeded the performance of N-doped TiO2 sample in the MB degradation
test reaction. In the experimental work by Breault et al. [222] the obtained material with high
photocatalytic activity was red. A similar color was obtained for the B/N co-doped TiO2 prepared by Liu et al. [224] The materials prepared here are yellowish and tend to get black with
increasing Nb content. This is a clear indication for defects at the surface of the material, which
are usually considered to act as recombination sites. Therefore, a preparation resulting in a less
112
6 Heterogeneous Photoelectrochemistry and Photocatalysis of TiO2 Nanomaterials
defective surface is needed, which can be obtained by ammonia treatments at lower temperatures with lower ammonia concentrations for nitrogen incorporation or post-calcination of the
ammonia-treated samples. Preliminary results show that defects can be significantly reduced
by post-annealing. Post-treatment at 400 ◦ C for 30 min in static air of the 500 ◦ C ammoniamodified Ti0.9 Nb0.1 and the Ti0.8Nb0.2 samples showed that the color changes to a bright yellow,
which is indicative for a less defective surface. First XPS results indicate that mainly the surface concentration of the substitutional Ns species decreased, whereas the interstitial Ni was
nearly unchanged after post-treatment (Fig. A23). Similarly, the color of the 700 ◦ C ammoniamodified Ti0.9Nb0.1 sample changed to brown/red upon post-treatment. The nitrogen content
determined by XPS decreased significantly. The substitutional nitrogen Ns was still the main
nitrogen species and a new nitrogen species at ∼ 402 eV was detected, which is usually attributed to nitrogen adsorbed on surface oxygen sites, instead of interstitial Ni with a binding
energy of ∼ 400 eV . [235,236] Furthermore, region scans of the Ti 2p and the Nb 3d region
showed that only Ti4+ as in TiO2 and Nb5+ as in Nb2 O5 was detected indicating a lower surface
defect concentration. XRD measurements showed that a composite of the nitride and the oxide
was obtained after calcination, and it is reasonable to assume that the surface mainly consists
of the oxide and nitride is still present in the bulk of the material. Thus, post-treatment appears
to be a suitable method to enhance the photocatalytic activity of Nb/N co-doped TiO2. Again,
it should be emphasized that the Nb to N-ratio is of major importance.
6.3.3 Conclusions and Outlook
First studies of the photocatalytic properties of Nb-doped and Nb/N co-doped TiO2 were performed. While the optimum Nb concentration for effective MB degradation was determined to
be 5 at%, it was shown using the anionic MO dye that adsorption of the cationic MB on the
surface is crucial for the beneficial degradation rates. Therefore, it is emphasized that the total
organic carbon content (TOC) should always be determined to verify the activity data obtained
by simple dye degradation test reactions. Although the photocatalytic activities determined for
the Nb/N co-doped TiO2 materials were small compared to the untreated Nb-doped TiO2 samples a correlation between the Nb and the N content incorporated into TiO2 was observed. The
results show that there is a certain potential of metal/anion co-doping of TiO2 , and particularly
for Nb/N co-doping, to increase the visible light activity of TiO2-based materials.
7 Summary and Prospects
It is well known that the activity of a photocatalyst is determined by a variety of properties,
and it is assumed that a photocatalyst has to be designed for each specific reaction. TiO2 was
shown to be one of the most versatile photocatalysts suffering mainly from its large band gap.
A variety of TiO2 materials with different structures and numerous modifications were already
tested, but till now the processes involved in photocatalytic reactions are not fully understood
and a knowledge-based approach to design a photocatalytically active material is still missing. This is particularly true for rather challenging photocatalytic reactions such as the overall
water splitting or the photocatalytic CO2 reduction with water as reducing agent involving the
thermodynamic stable molecules CO2 and H2 O. In both reactions energy is absorbed by the reactants and energy-rich molecules, like hydrogen and methane, are obtained. A successful implementation of this technology is of major importance for the future energy supply. However,
a photocatalyst has to be developed working on an industrial scale. So far the number of active
photocatalysts, especially for photocatalytic CO2 reduction, is rather limited. However, titaniabased materials were proven to exhibit a reasonable activity for the photocatalytic reduction
of CO2 . Among the different titania materials tested isolated tetrahedrally coordinated Ti-sites
were reported to offer high activities and significant selectivity towards methanol, whereas
methane production is favored at the octahedral coordinated Ti-sites mainly present in bulk
TiO2 materials. The specific interaction of the reactants with the photocatalyst, particularly the
interaction of CO2 with isolated titania, is not fully understood and the intermediates involved
in the photocatalytic CO2 reduction mechanism, either on tetrahedral or on octahedral Ti-sites,
are still not known. Therefore, several species are discussed to be possible intermediates in
CO2 reduction and responsible for the product distributions. Regardless of the coordination of
the Ti-site, all materials have in common that their photocatalytic activities are still not sufficient. Among others this is normally attributed to the inherent poor light absorption properties
of titania. In case of photocatalysts with isolated titania the low achievable loadings of isolated
titania are discussed to be responsible for the low activities and additionally it is speculated that
the adsorption of CO2 in close proximity to the active titania site will enhance the activity of
the photocatalyst.
Several questions are raised by the results presented in literature, namely:
1. What is the active site in titania-based photocatalysts in the photocatalytic reduction of
CO2 ?
114
7 Summary and Prospects
2. How to correlate titania coordination and activity?
3. Are the reactants interacting with the photocatalyst, e.g. is there any specific interaction
between isolated Ti-sites and CO2 ?
4. How to increase the photocatalytic activity for CO2 reduction of titania-based photocatalysts?
Addressing these questions was the main scope of this work. In chapter 5 the received results
are presented, which can be summarized as follows: In the first part of chapter 5 the photocatalyst was synthesized. A synthesis procedure was chosen facilitating the selective synthesis
of differently coordinated titania in a host material. SBA-15, a mesoporous silica, was used
as support material, since SBA-15 can be easily synthesized and high surfaces areas can be
obtained reproducibly. The incorporation of titania was achieved by the anhydrous grafting of
Ti(Oi Pr)4 . The suitability of the applied synthesis procedure with respect to the titania loading and the titania coordination was verified by in-depth characterization of the materials by
UV-Vis spectroscopy and X-ray spectroscopic methods (XAS, XPS). It was unambiguously
confirmed that isolated titania can be obtained with reasonably high titania loadings and that
the titania coordination can be easily controlled by the applied synthesis procedure. With increasing titania loading the coordination gradually changed from isolated tetrahedrally to octahedrally coordinated titania species. Therefore, titania incorporated in a mesoporous matrix
(Ti/SBA-15) is a suitable model catalyst to study the photocatalytic CO2 reduction, and a valuable pool of catalysts with varying titania coordination was readily available. Simultaneously,
the interaction of the materials with CO2 was studied by means of temperature-programmed
desorption experiments and infrared spectroscopy. It was proven by means of the direct, quantitative temperature-programmed desorption technique that isolated titania is not able to adsorb
CO2 at standard conditions usually applied in CO2 reduction. These results were clearly confirmed by IR spectroscopy. Even though the adsorption of CO2 was found to be negligible,
the Ti/SBA-15 materials were shown to exhibit reasonable photocatalytic activities in different
test reactions, and in case of solely isolated titania species the activity for the photocatalytic
CO2 reduction was demonstrated (Chapter 5.2). Within the photocatalytic CO2 reduction tests
formaldehyde/paraformaldehyde were found to be potential reaction intermediates of the CO2
reduction. Furthermore, structure-activity relationships were confirmed for the Ti/SBA-15 materials with varying titania coordination (Chapter 5.3). The photocatalytic activity data presented in Chapter 5.3 obtained for the photoreforming of methanol and the photocatalytic hydroxylation of terephthalic acid confirmed that there is a correlation of the titania coordination
and the activity of the photocatalyst (at least in presence of Au co-catalysts).
Two strategies were subsequently developed to increase the activity of Ti/SBA-15: the modification of Ti/SBA-15 with a material known to exhibit an amphoteric nature of the surface
to increase the CO2 adsorption properties, and the incorporation of a noble metal co-catalyst.
Zinc oxide was chosen, which is known to facilitate the adsorption of different carbonates and
115
bicarbonates. Moreover, ZnO was chosen as it is a common active ingredient of the methanol
synthesis catalyst and it is a photocatalyst by itself. Similarly, to the synthesis of titania species
the incorporation of ZnO was also achieved by anhydrous grafting. The characterization routine developed to study the titania coordination was also employed to study the zinc oxide
species incorporated into the mesoporous support. Zinc oxide was concluded to be present as
an isolated species or as small clusters in SBA-15. The beneficial CO2 adsorption properties of
the prepared Zn/SBA-15 materials were demonstrated by TPD and IR spectroscopy. The CO2
adsorption capacity can be easily correlated to the ZnO loading (Chapter 5.1). Combinations of
zinc oxide and titania-modified SBA-15 were synthesized by subsequent anhydrous grafting of
the two compounds. While the CO2 adsorption properties of the obtained materials were shown
to be nearly unchanged by the order of the grafting steps it was found that the titania precursor
preferably interacts with zinc oxide species, whereas the zinc precursor favors the interaction
with the SBA-15 support (Chapter 5.1). As a result of the preferential interactions of the two
precursors it can be concluded that the titania environment is different for the subsequently
grafted samples resulting in different photocatalytic activities of the materials especially for
the photocatalytic hydroxylation of terephthalic acid, which is discussed in detail in Chapter
5.3. Activities similar to Ti/SAB-15 were observed when titania was grafted first, whereas the
interaction of the titania precursor with the zinc oxide species was found to be unfavorable.
Another attempt to increase the photocatalytic activity was based on the incorporation of gold
nanoparticles. Gold nanoparticles were incorporated by photo-deposition. It was confirmed
that the photocatalytic activity of solely isolated titania species towards the reduction of CO2
is enhanced in presence of Au nanoparticles. The higher activity is mainly attributed to an
increase in the hydrogenation rate of a CO2 reduction intermediate. It was shown that less
formaldehyde/paraformaldehyde species are present on the Au-modified material, and therefore it is reasonable to assume that the hydrogenation of formaldehyde/paraformaldehyde is
favored. Additionally, it was confirmed that Au nanoparticles indeed enhance the photocatalytic activity, e.g. for the hydrogen production due to photoreforming of methanol or the
hydroxylation of terephthalic acid of the Ti/SBA-15 samples independent of the titania coordination. However, the photocatalysis data obtained for these test reactions showed that upon
zinc oxide modification of the materials the photocatalytic activity towards the photoreforming
of methanol was negligible, even though the activity towards terephthalic acid hydroxylation
resembles the activity of the unmodified titania samples. Yet, this phenomena is not fully understood. Transmission electron microscopy showed that larger Au particles are deposited on
the outer surface of the Ti/SBA-15 materials, whereas small particles are deposited in the pores
of the materials containing zinc oxide. Furthermore, titania species in close proximity to the Au
particles appear to be agglomerated pointing to a high mobility of titania species in Ti/SBA-15.
In contrast to these results the distribution of titania is homogeneously and the mobility of the
titania species in presence of zinc oxide appears to be lower. So far it can only be speculated
that larger agglomerates of titania and Au are required to produce hydrogen due to the pho-
116
7 Summary and Prospects
toreforming of methanol. Small Au particles close to titania sites are sufficient to drive the
hydroxylation reaction.
In summary, the results presented in Chapter 5 provide clear evidence that the employed strategy, e.g. using Ti/SBA-15 materials obtained by anhydrous grafting as model catalyst, is suitable to study the photocatalytic activity of titania-based materials for the reduction of CO2 .
Furthermore, it was shown that the titania coordination can be selectively modified, and thus
the Ti/SBA-15 materials are also suitable to study the effect of the titania coordination in great
detail. Reasonable attempts were already made to increase the inherent low photocatalytic
activities of titania-based photocatalysts, e.g. by the incorporation of gold nanoparticles as cocatalyst and zinc oxide species. Latter act as a CO2 adsorption site, which might be beneficial
for the activity in the photocatalytic CO2 reduction. While several key questions were answered
within the framework of this dissertation further studies are needed to unambiguously confirm
the structure-activity relationships observed for the Ti/SBA-15 materials in the photoreforming of methanol and the hydroxylation of terephthalic acid as well as for the CO2 reduction.
Additionally, the presence of CO2 adsorption sites, e.g. zinc oxide, was not yet proven to be
beneficial for the CO2 reduction.
In addition to these two complementary studies there are interesting approaches to increase
the activities of titania-related photocatalysts for CO2 reduction, which should be considered
in further research. First, it should be mentioned that in further studies other transition metals
can be used to modify the Ti/SBA-15 materials. It was already shown that isolated vanadate,
copper, molybdate, and chromate species can be obtained by anhydrous grafting. Most of them
exhibit beneficial light absorption properties compared with the titania and zinc oxide species
used in this study. Zinc oxide was mainly used due to the amphoteric character of zinc oxide
surfaces. Copper surfaces are also known to adsorb CO2 and additionally it is known to be
active in the electrochemical conversion of CO2 . Furthermore, isolated copper and molybdate
species were already shown to possess photocatalytic activities in test reactions like CO oxidation or decomposition of NO. [61] Second, the support for isolated titania sites can be modified.
The support used within this study to synthesize isolated titania species is known to be inert in
photocatalysis. Therefore, photo-excitation of an electron can only be achieved in the titania
species. For isolated titania the number of photo-excited electrons is additionally limited by the
number of isolated Ti-sites as only one electron can be excited for each isolated titania site. The
combination of a photocatalytically active support, e.g. ZnO or TiO2, modified with isolated
titania sites may be beneficial for the photocatalytic CO2 reduction properties. For ZnO and
TiO2 it is reported that oxygen evolution from water splitting is feasible increasing the number
of protons available for the CO2 reduction proceeding at the isolated titania site. In this sense
the material might work as a bifunctional catalyst. Even though it was already shown that
tetrahedrally coordinated titania can be obtained at the surface of TiO2 by iron doping, [237] it is
reasonable to assume that the modification of TiO2 with isolated Ti-sites by post-treatment will
be difficult as an octahedral titania coordination should be favored. A possible disadvantage
117
of ZnO as support material was already demonstrated by the results obtained within this study.
It was shown that isolated Ti-sites in a Zn-O-Ti environment exhibit only negligible photocatalytic activities compared to Ti-sites in Ti-O-Si environments. These problems of ZnO or TiO2
supports might be addressed by passivation of the surfaces by a thin silica coating, which will
favor the formation of isolated Ti-sites as observed for the Ti/SBA-15 materials. Moreover, it
was recently calculated that SiO2 passivation may be beneficial for the photocatalytic properties
of TiO2 materials. [238]
In addition to the studies directly related to the photocatalytic CO2 reduction a novel synthesis
method for the preparation of pristine and doped TiO2 was developed. A homogeneous distribution of a substitutional atom within the lattice structure of a host is of high relevance and a
successful doping can only be achieved when partial enrichment can be avoided. The spraydrying technique used to dope TiO2 with Nb was shown to be a suitable technique to achieve
a homogeneous doping of the TiO2 lattice. This was confirmed by different characterization
techniques in Chapter 6.1. Later on it was additionally shown that Nb-doped TiO2 can be effectively used in a photoelectrochemical cell (Chapter 6.2). It was concluded that an enhanced
mobility of electrons in the Nb-doped samples allows for higher efficiencies in the devices.
The properties of the Nb-doped TiO2 were maintained even upon post-treatment with urea and
higher efficiencies compared to the pristine urea-modified TiO2 were achieved. Finally, first attempts have been made to co-dope the Nb-doped TiO2 materials by ammonia, and interestingly
a preferential incorporation of nitrogen in the TiO2 lattice with increasing Nb content was confirmed. However, the photocatalytic activity of these materials was negligible compared with
the pristine and the Nb-doped TiO2. Therefore, the obtained results prove that the nitrogen
incorporation into the Nb-doped TiO2 has to be optimized. Most likely, the ammonia treatment
temperature, the ammonia treatment time, or the pretreatment have to be modified to achieve
photocatalytically active Nb/N co-doped TiO2 materials. Additionally, calcination of the spray
dried precursor powders in pure nitrogen might be useful to obtain photocatalytically active
Nb/N co-doped TiO2 materials, which then may also be suitable substrates for the modification
with isolated titania sites as discussed above.
Bibliography
[1] Fujishima, A.; Honda, K. Nature 1972, 238, 37–38.
[2] Frank, S. N.; Bard, A. J. J. Phys. Chem. 1977, 81, 1484–1488.
[3] Fujihira, M.; Satoh, Y.; Osa, T. Nature 1981, 293, 206–208.
[4] Balzani, V. Electron Transfer in Chemistry; Wiley-VCH Verlag GmbH: Weinheim, Germany, 2001.
[5] Inoue, T.; Fujishima, A.; Konishi, S.; Honda, K. Nature 1979, 277, 637–638.
[6] Braslavsky, S. E. Pure Appl. Chem. 2007, 79, 293–465.
[7] Gray, H. B. Nature Chemistry 2009, 1, 7.
[8] Dahl, S.; Chorkendorff, I. Nature Materials 2012, 11, 100–101.
[9] Kruger, P. Int. J. Hydrogen Energy 2001, 26, 1137–1147.
[10] Kočı́, K.; Obalová, L.; Lacný, Z. Chemical Papers 2008, 62, 1–9.
[11] Grätzel, M. Nature 2001, 414, 338–344.
[12] Xu, X.; Moulijn, J. A. Energy & Fuels 1996, 10, 305–325.
[13] Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. 2006, 103, 15729–15735.
[14] Dhakshinamoorthy, A.; Navalon, S.; Corma, A.; Garcia, H. Energy Environ. Sci. 2012,
5, 9217–9233.
[15] Freund, H.-J.; Roberts, M. Surf. Sci. Rep. 1996, 25, 225–273.
[16] Choi, S.; Drese, J. H.; Jones, C. W. ChemSusChem 2009, 2, 796–854.
[17] Roy, S. C.; Varghese, O. K.; Paulose, M.; Grimes, C. A. ACS Nano 2010, 4, 1259–1278.
[18] Linsebigler, A. L.; Lu, G.; Yates, J. T. Chem. Rev. 1995, 95, 735–758.
[19] Nowotny, M. K.; Sheppard, L. R.; Bak, T.; Nowotny, J. J. Phys. Chem. C 2008, 112,
5275–5300.
[20] Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W. Chem. Rev. 1995, 95,
69–96.
II
BIBLIOGRAPHY
[21] Mori, K.; Yamashita, H.; Anpo, M. RSC Adv. 2012, 2, 3165.
[22] Sze, S. M. Physics of semiconductor devices, 2nd ed.; Wiley: New York, 1981.
[23] Kasap, S. O. Principles of electronic materials and devices, 3rd ed.; McGraw-Hill:
Boston, 2006.
[24] Thomas, J. M.; Thomas, W. J. Principles and practice of heterogeneous catalysis; VCH:
Weinheim and , New York, 1997.
[25] Ohtani, B. J. Photochem. Photobiol., C 2010, 11, 157–178.
[26] B. Kaiser, W. J. S. F.; Lewerenz, H.-J. Bunsen-Magazin 2011, 13, 104–111.
[27] Gates, B. C.; Knözinger, H.; Jentoft, F. Advances in catalysis; Academic: London, 2009.
[28] Delannay, F. Characterization of heterogeneous catalysts; M. Dekker: New York, 1984.
[29] Park, H.; Park, Y.; Kim, W.; Choi, W. J. Photochem. Photobiol., C 2012.
[30] Kudo, A.; Miseki, Y. Chem. Soc. Rev. 2009, 38, 253–278.
[31] Osterloh, F. E. Chem. Mater. 2008, 20, 35–54.
[32] Dozzi, M. V.; Selli, E. J. Photochem. Photobiol., C 2012.
[33] Maeda, K.; Domen, K. Chem. Mater. 2010, 22, 612–623.
[34] Maeda, K.; Higashi, M.; Lu, D.; Abe, R.; Domen, K. J. Am. Chem. Soc. 2010, 132,
5858–5868.
[35] Maeda, K.; Takata, T.; Hara, M.; Saito, N.; Inoue, Y.; Kobayashi, H.; Domen, K. J. Am.
Chem. Soc. 2005, 127, 8286–8287.
[36] Maeda, K. J. Photochem. Photobiol., C 2011, 12, 237–268.
[37] Primo, A.; Corma, A.; Garcı́a, H. Phys. Chem. Chem. Phys. 2010, 13, 886.
[38] Murdoch, M.; Waterhouse, G. I. N.; Nadeem, M. A.; Metson, J. B.; Keane, M. A.; Howe,
R. F.; Llorca, J.; Idriss, H. Nature Chemistry 2011, 489–492.
[39] Du, L.; Furube, A.; Hara, K.; Katoh, R.; Tachiya, M. J. Photochem. Photobiol., C 2012.
[40] Wang, P.; Huang, B.; Dai, Y.; Whangbo, M.-H. Phys. Chem. Chem. Phys. 2012, 14,
9813.
[41] Chiarello, G. L.; Aguirre, M. H.; Selli, E. J. Catal. 2010, 273, 182–190.
[42] Kamat, P. V. J. Phys. Chem. B 2002, 106, 7729–7744.
[43] Zhang, Z.; Yates, J. T. Chem. Rev. 2012, 112, 5520–5551.
BIBLIOGRAPHY
III
[44] Wiberg, E.; Wiberg, N. Lehrbuch der anorganischen Chemie, 101st ed.; Walter de
Gruyter: Berlin, 1995.
[45] Glassford, K.; Chelikowsky, J. Phys. Rev. B 1992, 46, 1284–1298.
[46] Chang, H.; Huang, P. J. J. Raman Spectrosc. 1998, 29, 97–102.
[47] Ohsaka, T.; Izumi, F.; Fujiki, Y. J. Raman Spectrosc. 1978, 7, 321–324.
[48] Kelly, S.; Pollak, F. H.; Tomkiewicz, M. J. Phys. Chem. B 1997, 101, 2730–2734.
[49] Zhang, J.; Li, M.; Feng, Z.; Chen, J.; Li, C. J. Phys. Chem. B 2006, 110, 927–935.
[50] Zhang, Z.; Jeng, S.-P.; Henrich, V. Phys. Rev. B 1991, 43, 12004–12011.
[51] Müller, U. Anorganische Strukturchemie, 5th ed.; Teubner: Wiesbaden, 2006.
[52] Grätzel, M. J. Photochem. Photobiol., C 2003, 4, 145–153.
[53] Bledowski, M.; Wang, L.; Ramakrishnan, A.; Khavryuchenko, O. V.; Khavryuchenko,
V. D.; Ricci, P. C.; Strunk, J.; Cremer, T.; Kolbeck, C.; Beranek, R. Phys. Chem. Chem.
Phys. 2011, 13, 21511.
[54] Bledowski, M.; Wang, L.; Ramakrishnan, A.; Bétard, A.; Khavryuchenko, O. V.; Beranek, R. ChemPhysChem 2012, 13, 3018.
[55] D’Arienzo, M.; Siedl, N.; Sternig, A.; Scotti, R.; Morazzoni, F.; Bernardi, J.; Diwald,
O. J. Phys. Chem. C 2010, 114, 18067–18072.
[56] Chen, X.; Liu, L.; Yu, P. Y.; Mao, S. S. Science 2011, 331, 746–750.
[57] Hamdy, M. S.; Amrollahi, R.; Mul, G. ACS Catal. 2012, 2641–2647.
[58] Anpo, M.; Chiba, K. J. Mol. Catal. 1992, 74, 207–212.
[59] Thomas, J. M.; Raja, R.; Lewis, D. W. Angew. Chem. Int. Ed. 2005, 44, 6456–6482.
[60] Maschmeyer, T.; Rey, F.; Sankar, G.; Thomas, J. M. Nature 1995, 378, 159–162.
[61] Anpo, M.; Thomas, J. M. Chem. Comm. 2006, 3273.
[62] Amano, F.; Yamaguchi, T.; Tanaka, T. J. Phys. Chem. B 2006, 110, 281–288.
[63] Lee, E.; Wachs, I. J. Phys. Chem. C 2007, 111, 14410–14425.
[64] Gao, X.; Wachs, I. E. Catal. Today 1999, 51, 233–254.
[65] Ross-Medgaarden, E. I.; Wachs, I. E.; Knowles, W. V.; Burrows, A.; Kiely, C. J.; Wong,
M. S. J. Am. Chem. Soc. 2009, 131, 680–687.
[66] Morris, A. J.; Meyer, G. J.; Fujita, E. Acc. Chem. Res. 2009, 42, 1983–1994.
IV
BIBLIOGRAPHY
[67] Anpo, M.; Yamashita, H.; Ichihashi, Y.; Ehara, S. J. Electroanal. Chem. 1995, 396,
21–26.
[68] Anpo, M.; Yamashita, H.; Ichihashi, Y.; Fujii, Y.; Honda, M. J. Phys. Chem. B 1997,
101, 2632–2636.
[69] Dimitrijevic, N. M.; Vijayan, B. K.; Poluektov, O. G.; Rajh, T.; Gray, K. A.; He, H.;
Zapol, P. J. Am. Chem. Soc. 2011, 133, 3964–3971.
[70] Ikeue, K.; Yamashita, H.; Anpo, M.; Takewaki, T. J. Phys. Chem. B 2001, 105, 8350–
8355.
[71] Lin, W.; Han, H.; Frei, H. J. Phys. Chem. B 2004, 108, 18269–18273.
[72] Ulagappan, N.; Frei, H. J. Phys. Chem. A 2000, 104, 7834–7839.
[73] Yang, C.-C.; Vernimmen, J.; Meynen, V.; Cool, P.; Mul, G. J. Catal. 2011, 284, 1–8.
[74] Danon, A.; Stair, P. C.; Weitz, E. Catal. Lett. 2011, 141, 1057–1066.
[75] Lee, D.; Kanai, Y. J. Am. Chem. Soc. 2012, 134, 20266–20269.
[76] Shkrob, I. A.; Marin, T. W.; He, H.; Zapol, P. J. Phys. Chem. C 2012, 116, 9450–9460.
[77] Dimitrijevic, N. M.; Shkrob, I. A.; Gosztola, D. J.; Rajh, T. J. Phys. Chem. C 2012, 116,
878–885.
[78] Sastre, F.; Corma, A.; Garcı́a, H. J. Am. Chem. Soc. 2012, 134, 14137–14141.
[79] Varghese, O. K.; Paulose, M.; LaTempa, T. J.; Grimes, C. A. Nano Lett. 2009, 9, 731–
737.
[80] Gao, X.; Bare, S. R.; Fierro, J. L. G.; Banares, M. A.; Wachs, I. E. J. Phys. Chem. B
1998, 102, 5653–5666.
[81] Ricchiardi, G.; Damin, A.; Bordiga, S.; Lamberti, C.; Spanò, G.; Rivetti, F.; Zecchina,
A. J. Am. Chem. Soc. 2001, 123, 11409–11419.
[82] Sheldon, R. J. Mol. Catal. 1980, 7, 107–126.
[83] Sinha, A. K.; Seelan, S.; Tsubota, S.; Haruta, M. Angew. Chem. Int. Ed. 2004, 43, 1546–
1548.
[84] Brub, F.; Nohair, B.; Kleitz, F.; Kaliaguine, S. Chem. Mater. 2010, 22, 1988–2000.
[85] Bronkema, J.; Leo, D.; Bell, A. J. Phys. Chem. C 2007, 111, 14530–14540.
[86] Yang, H.-C.; Lin, H.-Y.; Chien, Y.-S.; Wu, J. C.-S.; Wu, H.-H. Catal. Lett. 2009, 131,
381–387.
[87] Xia, W.; Mei, B.; Sánchez, M. D.; Strunk, J.; Muhler, M. J. Nanosci. Nanotechnol. 2011,
11, 8152–8157.
BIBLIOGRAPHY
V
[88] Perathoner, S.; Lanzafame, P.; Passalacqua, R.; Centi, G.; Schlögl, R.; Su, D. Micropor.
Mesopor. Mater. 2006, 90, 347–361.
[89] Strunk, J.; Vining, W. C.; Bell, A. T. J. Phys. Chem. C 2010, 114, 16937–16945.
[90] Yamashita, H.; Honda, M.; Harada, M.; Ichihashi, Y.; Anpo, M.; Hirao, T.; Itoh, N.;
Iwamoto, N. J. Phys. Chem. B 1998, 102, 10707–10711.
[91] Auroux, A.; Gervasini, A. J. Phys. Chem. 1990, 94, 6371–6379.
[92] Iengo, P.; Aprile, G.; Di Serio, M.; Gazzoli, D.; Santacesaria, E. Appl. Catal. A: General
1999, 178, 97–109.
[93] Srivastava, R.; Srinivas, D.; Ratnasamy, P. Micropor. Mesopor. Mater. 2006, 90, 314–
326.
[94] Shi, L. Y.; Wang, Y. M.; Ji, A.; Gao, L.; Wang, Y. J. Mater. Chem. 2005, 15, 1392.
[95] Young, P. D.; Notestein, J. M. ChemSusChem 2011, 4, 1671–1678.
[96] Lavalley, J.; Saussey, J.; Raı̈s, T. J. Mol. Catal. 1982, 17, 289–298.
[97] Xia, X.; Strunk, J.; Busser, W.; Naumann d’Alnoncourt, R.; Muhler, M. J. Phys. Chem.
C 2008, 112, 10938–10942.
[98] Noei, H.; Wöll, C.; Muhler, M.; Wang, Y. J. Phys. Chem. C 2011, 115, 908–914.
[99] Kurtz, M.; Strunk, J.; Hinrichsen, O.; Muhler, M.; Fink, K.; Meyer, B.; Wöll, C. Angew.
Chem. Int. Ed. 2005, 44, 2790–2794.
[100] Fox, M. A.; Dulay, M. T. Chem. Rev. 1993, 93, 341–357.
[101] Chen, J.; Feng, Z.; Ying, P.; Li, C. J. Phys. Chem. B 2004, 108, 12669–12676.
[102] Tkachenko, O. P.; Klementiev, K. V.; Löffler, E.; Ritzkopf, I.; Schüth, F.; Bandyopadhyay, M.; Grabowski, S.; Gies, H.; Hagen, V.; Muhler, M.; Lu, L.; Fischer, R. A.;
Grünert, W. Phys. Chem. Chem. Phy. 2003, 5, 4325.
[103] Mihai, G. D.; Meynen, V.; Mertens, M.; Bilba, N.; Cool, P.; Vansant, E. F. J. Mater. Sci.
2010, 45, 5786–5794.
[104] Almutairi, S. M. T.; Mezari, B.; Magusin, P. C. M. M.; Pidko, E. A.; Hensen, E. J. M.
ACS Catal. 2011, 71–83.
[105] Zhao, D.; Huo, Q.; Feng, J.; Chmelka, B. F.; Stucky, G. D. J. Amer. Chem. Soc. 1998,
120, 6024–6036.
[106] Vining, W. C.; Goodrow, A.; Strunk, J.; Bell, A. T. J. Catal. 2010, 270, 163–171.
[107] Brunauer, S.; Emmett, P. H.; Teller, E. J. Am. Chem. Soc. 1938, 60, 309–319.
[108] Barrett, E. P.; Joyner, L. G.; Halenda, P. P. J. Am. Chem. Soc. 1951, 73, 373–380.
VI
BIBLIOGRAPHY
[109] Lippens, B. C.; Boer, J. H. d. J. Catal. 1965, 4, 319–323.
[110] Moulder, J. F.; Chastain, J. Handbook of X-ray photoelectron spectroscopy: A reference
book of standard spectra for identification and interpretation of XPS data, update. ed.;
Perkin-Elmer Corporation: Eden Prairie, Minn, 1992.
[111] Wang, Y.; Glenz, A.; Muhler, M.; Wöll, C. Rev. Sci. Instrum. 2009, 80, 113108.
[112] Klementev, K. V. J. Phys. D: Appl. Phys. 2001, 34, 209–217.
[113] Bouh, A. O.; Rice, G. L.; Scott, S. L. J. Am. Chem. Soc. 1999, 121, 7201–7210.
[114] Joo, S. H.; Ryoo, R.; Kruk, M.; Jaroniec, M. J. Phys. Chem. B 2002, 106, 4640–4646.
[115] Ryoo, R.; Ko, C. H.; Kruk, M.; Antochshuk, V.; Jaroniec, M. J. Phys. Chem. B 2000,
104, 11465–11471.
[116] Strunk, J.; Vining, W. C.; Bell, A. T. J. Phys. Chem. C 2011, 115, 4114–4126.
[117] Bérubé, F.; Kleitz, F.; Kaliaguine, S. J. Mater. Sci. 2009, 44, 6727–6735.
[118] Yamashita, H.; Kawasaki, S.; Ichihashi, Y.; Harada, M.; Takeuchi, M.; Anpo, M.; Stewart, G.; Fox, M. A.; Louis, C.; Che, M. J. Phys. Chem. B 1998, 102, 5870–5875.
[119] Chen, J.; Feng, Z.; Ying, P.; Li, M.; Han, B.; Li, C. Phys. Chem. Chem. Phys. 2004, 6,
4473.
[120] Grünert, W.; Schlögel, R. Mol. Sieves, 2004, 467.
[121] Vayssilov, G. N. Catal. Rev. 1997, 39, 209–251.
[122] Stakheev, A. Y.; Shpiro, E. S.; Apijok, J. J. Phys. Chem. 1993, 97, 5668–5672.
[123] Farges, F.; Brown, G.; Rehr, J. Phys. Rev. B 1997, 56, 1809–1819.
[124] Saussey, J.; Lavalley, J.-C.; Bovet, C. J. Chem. Soc., Faraday Trans. 1982, 78, 1457.
[125] Danon, A.; Stair, P. C.; Weitz, E. J. Phys. Chem. C 2011, 115, 11540–11549.
[126] Srivastava, R.; Srinivas, D.; Ratnasamy, P. J. Catal. 2005, 233, 1–15.
[127] Srinivas, D.; Ratnasamy, P. Micropor. Mesopor. Mater. 2007, 105, 170–180.
[128] Esken, D.; Noei, H.; Wang, Y.; Wiktor, C.; Turner, S.; van Tendeloo, G.; Fischer, R. A.
J. Mater. Chem. 2011, 21, 5907.
[129] Hoffmann, M. R.; Moss, J. A.; Baum, M. M. Dalton Trans. 2011, 40, 5151.
[130] Hwang, J.-S.; Chang, J.-S.; Park, S.-E.; Ikeue, K.; Anpo, M. Top. Catal. 2005, 35, 311–
319.
[131] Fuku, K.; Sakano, T.; Kamegawa, T.; Mori, K.; Yamashita, H. J. Mater. Chem. 2012.
BIBLIOGRAPHY
VII
[132] Mori, K.; Miura, Y.; Shironita, S.; Yamashita, H. Langmuir 2009, 25, 11180–11187.
[133] Busser, G. W.; Mei, B.; Muhler, M. ChemSusChem 2012, 5, 2200–2206.
[134] Mei, B.; Becerikli, A.; Pougin, A.; Heeskens, D.; Sinev, I.; Grünert, W.; Muhler, M.;
Strunk, J. J. Phys. Chem. C 2012, 116, 14318–14327.
[135] Buso, D.; Pacifico, J.; Martucci, A.; Mulvaney, P. Adv. Funct. Mater. 2007, 17, 347–354.
[136] Escamilla-Perea, L.; Nava, R.; Pawelec, B.; Rosmaninho, M.; Peza-Ledesma, C.; Fierro,
J. Appl. Catal. A: General 2010, 381, 42–53.
[137] Yang, C.-C.; Yu, Y.-H.; van der Linden, B.; Wu, J. C. S.; Mul, G. J. Amer. Chem. Soc.
2010, 132, 8398–8406.
[138] Comotti, M.; Li, W.-C.; Spliethoff, B.; Schüth, F. J. Am. Chem. Soc. 2006, 128, 917–924.
[139] Strunk, J.; Kähler, K.; Xia, X.; Comotti, M.; Schüth, F.; Reinecke, T.; Muhler, M. Appl.
Catal. A: General 2009, 359, 121–128.
[140] Ryndin, Y.; Hicks, R.; Bell, A.; Yermakov, Y. J. Catal. 1981, 70, 287–297.
[141] Busca, G.; Lamotte, J.; Lavalley, J. C.; Lorenzelli, V. J. Am. Chem. Soc. 1987, 109,
5197–5202.
[142] McGill, P. R.; Söhnel, T. Phys. Chem. Chem. Phys. 2011, 14, 858.
[143] Qiu, H.; Idriss, H.; Wang, Y.; Wöll, C. J. Phys. Chem. C 2008, 112, 9828–9834.
[144] Li, Z.; Zhang, Z.; Kay, B. D.; Dohnlek, Z. J. Phys. Chem. C 2011, 115, 9692–9700.
[145] Kähler, K.; Holz, M. C.; Rohe, M.; Strunk, J.; Muhler, M. ChemPhysChem 2010, 11,
2521–2529.
[146] Bond, G. C.; Louis, C.; Thompson, D. T. Catalysis by gold; Imperial College Press:
London, 2006.
[147] Widmann, D.; Behm, R. J. Angew. Chem. Int. Ed. 2011, 50, 10241–10245.
[148] Bowker, M.; Millard, L.; Greaves, J.; James, D.; Soares, J. Gold Bulletin 2004, 37,
170–173.
[149] Du, L.; Furube, A.; Yamamoto, K.; Hara, K.; Katoh, R.; Tachiya, M. J. Phys. Chem. C
2009, 113, 6454–6462.
[150] Hirakawa, T.; Nosaka, Y. Langmuir 2002, 18, 3247–3254.
[151] Fang, X.; Mark, G.; Sonntag, C. Ultrason. Sonochem. 1996, 3, 57–63.
[152] Liu, G.; Wang, L.; Sun, C.; Yan, X.; Wang, X.; Chen, Z.; Smith, S. C.; Cheng, H.-M.;
Lu, G. Q. Chem. Mater. 2009, 21, 1266–1274.
VIII
BIBLIOGRAPHY
[153] Matsuoka, M.; Kitano, M.; Takeuchi, M.; Tsujimaru, K.; Anpo, M.; Thomas, J. M.
Catal. Today 2007, 122, 51–61.
[154] Ang, T. P.; Toh, C. S.; Han, Y.-F. J. Phys. Chem. C 2009, 113, 10560–10567.
[155] Atashbar, M. Z.; Sun, H. T.; Gong, B.; Wlodarski, W.; Lamb, R. Thin Solid Films 1998,
326, 238–244.
[156] Yamada, N.; Hitosugi, T.; Hoang, N. L. H.; Furubayashi, Y.; Hirose, Y.; Konuma, S.;
Shimada, T.; Hasegawa, T. Thin Solid Films 2008, 516, 5754–5757.
[157] Park, K.-W.; Seol, K.-S. Electrochem. Commun. 2007, 9, 2256–2260.
[158] Liu, G.; Wang, L.; Yang, H. G.; Cheng, H. M.; Lu, G. Q. J. Mater. Chem. 2010, 20,
831–843.
[159] Choi, W.; Termin, A.; Hoffmann, M. R. J. Phys. Chem. 1994, 98, 13669–13679.
[160] Kazumoto, N.; Takaaki, K.; Naoya, M.; Toshiki, T.; Teruhisa, O. Int. J. Photoenergy
2007, 2008.
[161] Anpo, M.; Takeuchi, M. J. Catal. 2003, 216, 505–516.
[162] Kubacka, A.; Fernández-Garcı́a, M.; Colón, G. J. Catal. 2008, 254, 272–284.
[163] Kubacka, A.; Colón, G.; Fernandez-Garcı́a, M. Catal. Today 2009, 143, 286–292.
[164] Castro, A. L.; Nunes, M. R.; Carvalho, M. D.; Ferreira, L. P.; Jumas, J. C.; Costa, F. M.;
Florêncio, M. H. J. Solid State Chem. 2009, 182, 1838–1845.
[165] Mattsson, A.; Leideborg, M.; Larsson, K.; Westin, G.; Asterlund, L. J. Phys. Chem. B
2006, 110, 1210–1220.
[166] Emeline, A. V.; Furubayashi, Y.; Zhang, X.; Jin, M.; Murakami, T.; Fujishima, A. J.
Phys. Chem. B 2005, 109, 24441–24444.
[167] Koninck, M.; Manseau, P.; Marsan, B. J. Electroanal. Chem. 2007, 611, 67–79.
[168] Hasin, P.; Alpuche-Aviles, M. A.; Li, Y.; Wu, Y. J. Phys. Chem. C 2009, 113, 7456–
7460.
[169] Lü, X.; Mou, X.; Wu, J.; Zhang, D.; Zhang, L.; Huang, F.; Xu, F.; Huang, S. Adv. Funct.
Mater. 2010, 20, 509–515.
[170] Yum, J.-H.; Chen, P.; Grätzel, M.; Nazeeruddin, M. K. ChemSusChem 2008, 1, 699–707.
[171] Pore, V.; Ritala, M.; Leskela, M.; Saukkonen, T.; Jarn, M. Cryst. Growth Des. 2009, 9,
2974–2978.
[172] Arbiol, J.; Cerda, J.; Dezanneau, G.; Cirera, A.; Peiro, F.; Cornet, A.; Morante J. Appl.
Phys. 2002, 92, 853.
BIBLIOGRAPHY
IX
[173] Almquist, C. B.; Biswas, P. J. Catal. 2002, 212, 145–156.
[174] Morris, D.; Dou, Y.; Rebane, J.; Mitchell, C. E. J.; Egdell, R. G.; Law, D. S. L.; Vittadini,
A.; Casarin, M. Phys. Rev. B 2000, 61, 13445.
[175] Valigi, M.; Cordischi, D.; Minelli, G.; Natale, P.; Porta, P.; Keijzers, C. P. J. Solid State
Chem. 1988, 77, 255–263.
[176] Guoying, C.; Bare, S. R.; Mallouk, T. E. J. Electrochem. Soc. 2002, 149, A1092–A1099.
[177] Hidalgo, M. C.; Bahnemann, D. Appl. Catal. B: Environ. 2005, 61, 259–266.
[178] Pan, H.; Chen, N.; Shen, S.; Huang, J. J. Sol-Gel Sci. Technol. 2005, 34, 63–69.
[179] Tasbihi, M.; Štangar, U. L.; Černigoj, U.; Kogej, K. Photochem. Photobiol. Sci. 2009, 8,
719–725.
[180] Kaluza, S.; Schröter, M. K.; Naumann d’Alnoncourt, R.; Reinecke, T.; Muhler, M. Adv.
Funct. Mater. 2008, 18, 3670–3677.
[181] Kaluza, S.; Muhler, M. J. Mater. Chem. 2009, 19, 3914–3922.
[182] Kaluza, S.; Muhler, M. Catal. Lett. 2009, 129, 287–292.
[183] Sizgek, E.; Bartlett, J. R.; Brungs, M. P. J. Sol-Gel Sci. Technol. 1998, 13, 1011–1016.
[184] Zhang, X.; Pan, J. H.; Du, A. J.; Ng, J.; Sun, D. D.; Leckie, J. O. Mater. Res. Bull. 2009,
44, 1070–1076.
[185] F. Iskandar; A. B. D. Nandiyanto; K. M. Yun; C. J. Hogan Jr.; K. Okuyama; P. Biswas
Adv. Mater. 2007, 19, 1408–1412.
[186] Lide, D. R. CRC handbook of chemistry and physics, internet version 2005. ed.; CRC
Press: Boca Raton, Fla, 2004.
[187] Johnsson, M.; Pettersson, P.; Nygren, M. Thermochim. Acta 1997, 298, 47–54.
[188] Brnicevic, N.; Despotovic, Z. Proceedings Third ICTA Davos, HR Oswald, E. Dubler
(Hrsg.) 1971, 705.
[189] J. F. Banfield, D. R. V.; Smith, D. J. Am. Mineral., 1991, 343–353.
[190] Brohan, L.; Verbaere, A.; Tournoux, M.; Demazeau, G. Mater. Res. Bull. 1982, 17, 355–
361.
[191] Hirano, M.; Matsushima, K. J. Am. Ceram. Soc. 2006, 89, 110–117.
[192] Chandiran, A. K.; Sauvage, F.; Casas-Cabanas, M.; Comte, P.; Zakeeruddin, S. M.;
Grätzel, M. J. Phys. Chem. C 2010, 114, 15849–15856.
[193] Shannon, R. D.; Prewitt, C. T. Acta Crystall. B: Stru. 1970, 26, 1046–1048.
X
BIBLIOGRAPHY
[194] J. I. Langford NIST Spec. Publ., 1992, 110–126.
[195] Gao, Y.; Liang, Y.; Chambers, S. A. Surf. Sci. 1996, 348, 17–27.
[196] Bahl, M. J. Phys. Chem. Solids 1975, 36, 485–491.
[197] Pittman, R. M.; Bell, A. T. J. Phys. Chem. C 1993, 97, 12178–12185.
[198] Ruiz, A. M.; Dezanneau, G.; Arbiol, J.; Cornet, A.; Morante, J. R. Chem. Mater. 2004,
16, 862–871.
[199] Liu, X. D.; Jiang, E. Y.; Li, Z. Q.; Song, Q. G. Appl. Phys. Lett. 2008, 92, 252104.
[200] Bard, A. J. Science 1980, 207, 139–144.
[201] Khaselev, O. Science 1998, 280, 425–427.
[202] Licht, S. Int. J. Hydrogen Energy 2001, 26, 653–659.
[203] Dau, H.; Limberg, C.; Reier, T.; Risch, M.; Roggan, S.; Strasser, P. ChemCatChem 2010,
2, 724–761.
[204] Solarska, R.; Królikowska, A.; Augustyński, J. Angew. Chem. Int. Ed. 2010, 49, 7980–
7983.
[205] Vidyarthi, V. S.; Hofmann, M.; Savan, A.; Sliozberg, K.; König, D.; Beranek, R.; Schuhmann, W.; Ludwig, A. Int. J. Hydrogen Energy 2011, 36, 4724–4731.
[206] Jorand Sartoretti, C.; Alexander, B. D.; Solarska, R.; Rutkowska, I. A.; Augustynski, J.;
Cerny, R. J. Phys. Chem. B 2005, 109, 13685–13692.
[207] Kay, A.; Cesar, I.; Grätzel, M. J. Am. Chem. Soc. 2006, 128, 15714–15721.
[208] Upul Wijayantha, K.; Saremi-Yarahmadi, S.; Peter, L. M. Phys. Chem. Chem. Phy. 2011,
13, 5264.
[209] Youngblood, W. J.; Lee, S. H. A.; Maeda, K.; Mallouk, T. E. Acc. Chem. Res. 2009, 42,
1966–1973.
[210] Moore, G. F.; Blakemore, J. D.; Milot, R. L.; Hull, J. F.; Song, H.; Cai, L.; Schmuttenmaer, C. A.; Crabtree, R. H.; Brudvig, G. W. Energy Environ. Sci. 2011, 4, 2389.
[211] Swierk, J. R.; Mallouk, T. E. Chem. Soc. Rev. 2013.
[212] Tributsch, H. J. Solid State Electrochem. 2009, 13, 1127–1140.
[213] Youngblood, W. J.; Lee, S.-H. A.; Kobayashi, Y.; Hernandez-Pagan, E. A.; Hoertz, P. G.;
Moore, T. A.; Moore, A. L.; Gust, D.; Mallouk, T. E. J. Amer. Chem. Soc. 2009, 131,
926–927.
[214] Wang, L.; Bledowski, M.; Ramakrishnan, A.; Konig, D.; Ludwig, A.; Beranek, R. J.
Electrochem. Soc. 2012, 159, H616–H622.
BIBLIOGRAPHY
XI
[215] Bledowski, M.; Wang, L.; Ramakrishnan, A.; Beranek, R. J. Mater. Res. 2012, 1–7.
[216] Furubayashi, Y.; Hitosugi, T.; Yamamoto, Y.; Inaba, K.; Kinoda, G.; Hirose, Y.; Shimada, T.; Hasegawa, T. Appl. Phys. Lett. 2005, 86, 252101.
[217] Liu, Y.; Szeifert, J. M.; Feckl, J. M.; Mandlmeier, B.; Rathousky, J.; Hayden, O.;
Fattakhova-Rohlfing, D.; Bein, T. ACS Nano 2010, 4, 5373–5381.
[218] Yang, M.; Jha, H.; Liu, N.; Schmuki, P. J. Mater. Chem. 2011, 21, 15205.
[219] Das, C.; Roy, P.; Yang, M.; Jha, H.; Schmuki, P. Nanoscale 2011, 3, 3094.
[220] Mei, B.; Sánchez, M. D.; Reinecke, T.; Kaluza, S.; Xia, W.; Muhler, M. J. Mater. Chem.
2011, 21, 11781.
[221] Beranek, R. Adv. Chem. Phys. 2011, 2011, 1–20.
[222] Breault, T. M.; Bartlett, B. M. J. Phys. Chem. C 2012, 116, 5986–5994.
[223] Kubacka, A.; Bachiller-Baeza, B.; Colon, G.; FernaÌndez-Garcia, M. J. Phys. Chem. C
2009, 113, 8553–8555.
[224] Liu, G.; Yin, L.-C.; Wang, J.; Niu, P.; Zhen, C.; Xie, Y.; Cheng, H.-M. Energy Environ.
Sci. 2012, 5, 9603.
[225] Márquez, A. M.; Plata, J. J.; Ortega, Y.; Sanz, J. F.; Colón, G.; Kubacka, A.; FernándezGarcı́a, M. J. Phys. Chem. C 2012, 116, 18759–18767.
[226] Di Valentin, C.; Finazzi, E.; Pacchioni, G.; Selloni, A.; Livraghi, S.; Czoska, A. M.;
Paganini, M. C.; Giamello, E. Chem. Mater. 2008, 20, 3706–3714.
[227] Neville, E. M.; Mattle, M. J.; Loughrey, D.; Rajesh, B.; Rahman, M.; MacElroy, J. D.;
Sullivan, J. A.; Thampi, K. R. J. Phys. Chem. C 2012, 116, 16511–16521.
[228] Katoh, R.; Furube, A.; Yamanaka, K.; Morikawa, T. J. Phys. Chem. Lett. 2010, 1, 3261–
3265.
[229] Yin, W.-J.; Tang, H.; Wei, S.-H.; Al-Jassim, M.; Turner, J.; Yan, Y. Phys. Rev. B 2010,
82.
[230] Hoang, S.; Guo, S.; Mullins, C. B. J. Phys. Chem. C 2012, 116, 23283–23290.
[231] Kohtani, S.; Yoshioka, E.; Saito, K.; Kudo, A.; Miyabe, H. J. Phys. Chem. C 2012, 116,
17705–17713.
[232] Tada, H.; Jin, Q.; Kobayashi, H. ChemPhysChem 2012, 13, 3457–3461.
[233] Oropeza, F. E.; Harmer, J.; Egdell, R. G.; Palgrave, R. G. Phys. Chem. Chem. Phys.
2010, 12, 960–969.
XII
BIBLIOGRAPHY
[234] Engström, C.; Birch, J.; Hultman, L.; Lavoie, C.; Cabral, C.; Jordan-Sweet, J. L.; Carlsson, J. R. A. J. Vac. Sci. Technol., A 1999, 17, 2920.
[235] Chen, X.; Wang, X.; Hou, Y.; Huang, J.; Wu, L.; Fu, X. J. Catal. 2008, 255, 59–67.
[236] Asahi, R. Science 2001, 293, 269–271.
[237] Wu, Q.; Zheng, Q.; van de Krol, R. J. Phys. Chem. C 2012, 116, 7219–7226.
[238] Seriani, N.; Pinilla, C.; Cereda, S.; Vita, A. d.; Scandolo, S. J. Phys. Chem. C 2012, 116,
11062–11067.
8 Appendix
Figures
8.1 Supporting Figures Chapter 4
a)
b)
Figure A1: a) Picture of complete photoreactor set-up and b) closeup view of the fully metal sealed
home-made photoreactor.
8 Appendix
XIV
8.2 Supporting Figures Chapter 5
1378
0.02
2343
1553
2278
+ CO
+ CO
2
G
Absorbance
Absorbance
H
2
F
E
D
C
B
A
clean surface
clean surface
2500 2400 2300 2200
1600
-1
Wavenumber [cm ]
1400
-1
Wavenumber [cm ]
Figure A2: UHV-FTIR spectra obtained after exposing the clean Zn0.7/Ti0.9/SBA-15 sample to
CO2 at 90 K in an UHV chamber. (A) clean surface, (B) 1x10-6 mbar CO2 , (C) 5x10-6
mbar CO2 , (D) 1x10-5 mbar CO2 , (E) 5x10-5 mbar CO2 , (F) 1x10-4 mbar CO2 , (G)
1x10-4 mbar CO2 , and (H) 1x10-4 mbar CO2 . left) CO2 range, right) carbonate range.
0.002
2
(highest dosing)
+ CO
Absorbance
Absorbance
+ CO
clean surface
2500
0.002
2400
2300
2200
-1
Wavenumber [cm ]
2
(highest dosing)
clean surface
1600
1400
-1
Wavenumber [cm ]
Figure A3: UHV-FTIR spectra obtained after exposing the clean Ti1.0/SBA-15 sample to 1x10-4
mbar CO2 at 300 K in an UHV chamber. left: CO2 range, right: carbonate range.
Concentration [ppm]
8.2 Supporting Figures Chapter 5
XV
2500
light off
H
2
O
2000
light on
2
add HAuCl
4
light off
1500
MeOH/H O
2
light on
1000
500
0
0
60
120
180
240
300
Time [min]
Figure A4: Hydrogen evolution observed during Au photo-deposition onto Ti/SBA-15.
concentration [ppm])
400
CH
300
80
40
0
80
60
40
20
0
C H
2
20
16
12
8
4
0
6
C H
3
0
2
4
6
8
4
8
10
cleaning step
Figure A5: Evolution of different Cx Hy species observed after 5 h of irradiation for the subsequent
pretreatment steps in humid He in presence of Au/Ti/SBA-15. The decay of the different compounds was fitted using exponential curves.
8 Appendix
XVI
C H
45
4
10
C H
40
3
8
2
15
deact.
deact.
deact.
5
deact.
10
Au/Ti/SBA-15
Au/Ti/SBA-15-C
20
Ti/SBA-15
25
Au/Ti/SBA-15
yield [ppm]
6
CH
4
30
Ti/SBA-15
C H
35
0
1. run
1. run
1. run
2. run
2. run
CPS [a.u.]
Figure A6: Methane yield after 5 h of irradiation obtained for the deactivated Ti/SBA-15, the deactivated Au/Ti/SBA-15, and the calcined Au/Ti/SBA-15-C samples.
110 108 106 104 102 100 470
465
460
455 540
535
530
525
binding energy [eV]
Figure A7: Si 2p, Ti 2p, and O 1s XPS region scans of the fresh (black line) and reacted (red line)
Au/Ti/SBA-15 samples.
8.2 Supporting Figures Chapter 5
XVII
Ti/SBA
Ti/SBA(400)
Ti/SBA-react(400)
Au/Ti/SBA
Au/Ti/SBA(400)
F(R) [a.u.]
Au/Ti/SBA-react
Au/Ti/SBA-react(400)
300
400
500
600
700
800
wavelength [nm]
CPS [a.u.]
Figure A8: UV-Vis spectra obtained with the reacted and fresh Ti/SBA-15 and Au/Ti/SBA-15 samples. UV-Vis spectra of each sample were recorded before and after heating the sample
to 400 ◦ C. Furthermore, a red shift of the characteristic Au plasmon of the Au/Ti/SBA15-C sample was observed by UV-Vis spectroscopy possibly causing the decrease in
activity (not shown).
Au/Ti/SBA-15
Ti/SBA-15
470
468
466
464
462
460
458
456
454
binding energy [eV]
Figure A9: Ti 2p signals of the Ti/SBA-15 and Au/Ti/SBA-15 obtained by XPS
8 Appendix
XVIII
Au/Ti/SBA-15
400°C
c)
Au/Ti/SBA-15
log (1/R)
b)
120°C
a)
Ti/SBA-15
120°C
3050 3000 2950 2900 2850
2000 1900 1800 1700 1600 1500
-1
wavenumbers [cm ]
Figure A10: Difference DRIFT spectra obtained for the Ti/SBA-15 sample at 120 ◦ C and for the
Au/Ti/SBA-15 at 120 ◦ C and 400 ◦ C, respectively.
Normalized Frequency
1,0
0,8
0,6
0,4
0,2
0,0
2
3
4
5
6
7 8 9 10 11 12 13 14 15 16
Diameter [nm]
Figure A11: Size distribution of Au particles in the Au/Zn0.3/Ti1.2/SBA material. The size of the
particles was determined from HAADF-STEM images. The diameter was calculated
by assuming a spherical shape for the particles.
8.2 Supporting Figures Chapter 5
XIX
Figure A12: EDX spectrum of a typical Au/Zn0.3/Ti1.2/SBA crystal. The inset table shows the
composition of the crystal in wt%, its error, and the composition in at%.
XX
Figure A13: a) HRTEM image of a typical Au/Zn0.3/Ti1.2/SBA crystal. The channels are visibly intact. The FFT of a) (displayed in b)) reveals an average
distance of approximately 100 Å. c) EFTEM Ti map of the crystal visible in a). The Ti distribution is homogeneous. There are no visible Ti
enrichments. The dark area on the left side is a thickness effect, and is not due to the absence of Ti. d) EELS spectrum of Au/Zn0.3/Ti1.2/SBA
showing the Ti-L2,3 and the O-K edge. e) Ti-K EDX line profile over c) (from top to bottom).
8 Appendix
8.2 Supporting Figures Chapter 5
XXI
Top Row
Middle Row
Bottom Row
Figure A14: Top row, left: BF-TEM image of Au/Ti1.2/Zn0.3/SBA. The crystal in the center of
the image has been imaged along its channel direction. FFT analysis of the area
highlighted by the white frame (inset) shows a spacing of approximately 95 Å of the
hexagonally packed pores for the imaged crystal. Right: HR-TEM of a gold particle
in Au/Ti1.2/Zn0.3/SBA. The particle is imaged along its [110] zone axis orientation
as it can be seen in the FFT analysis (inset). Middle row, left: HAADF-STEM image of a Au/Ti1.2/Zn0.3/SBA crystal along its pore direction. Right: FFT analysis
of the area highlighted by the white frame (inset) shows a spacing of 101 Å for the
imaged crystal. The imaged gold particles are quite monodisperse. Bottom row, left:
BF-TEM, magnification the Au/Ti1.2/Zn0.3/SBA material and Ti EFTEM map of the
Au/Ti1.2/Zn0.3/SBA crystal shown in the top right. The titanium distribution is homogeneous except for small enrichments (intensity is proportional to the Ti concentration). Right: EELS spectrum of Au/Ti1.2/Zn0.3/SBA showing the Ti-L2,3 and the
O-K edge.
8 Appendix
XXII
8.3 Supporting Figures Chapter 6
180
a)
TiO
2
Nb01
Raman Intensity [a.u.]
160
Nb02
140
Nb5
Nb10
120
100
80
60
40
20
0
100
200
300
400
500
600
700
800
Raman Shift [nm]
Figure A15: a) Structural characterization of Nb-doped TiO2 samples with different Nb concentrations, b) SEM picture of the cross-section of a typical anode.
a)
TiO
b)
2
Nb01
TiO
2
Nb05
Nb01
300
0.5
F(R) [a.u.]
Nb1
Nb10
400
500
600
wavelength [nm]
700
Nb05
(F(R) x hv)
Nb5
800
Nb1
Nb5
Nb10
2.00
2.25
2.50
2.75
3.00
3.25
3.50
3.75
energy [eV]
Figure A16: a) UV-Vis absorption measurements and b) Tauc plots of Nb-doped TiO2 samples.
4.00
8.3 Supporting Figures Chapter 6
XXIII
0.22
TiO
increasing wavelength
0.20
2
Nb01
0.18
Nb05
0.16
Nb1
Nb5
I [mA]
0.14
Nb10
0.12
0.10
0.08
0.06
0.04
300 nm
400 nm
0.02
0.00
0
25
50
75 100 125 150 175 200 225 250 275 300 325 350
time [s]
Figure A17: Wavelength-dependent photocurrent measurements of different Nb-doped TiO2 samples.
0.30
0.20
a)
0.25
b)
TiO
2
TiO
2
0.15
Nb01
Nb01
0.20
0.10
I [mA]
I [mA]
0.15
0.10
0.05
0.05
0.00
0.00
-0.05
-0.05
-0.10
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
0.2
0.4
0.6
potential [V vs Ag/AgCl]
0.8
1.0
-0.10
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
potential [V vs Ag/AgCl]
Figure A18: a) Potential-dependent photocurrent response for the TiO2 -reference and Nb (0.1 at%)
materials irradiated at 350 nm in phosphate buffer (pH 7). b) Magnification of the low
potential range of the potential-dependent photocurrent measurements.
8 Appendix
potential [mV vs Ag/AgCl]
XXIV
600
b)
TiO
2
Nb02
Nb05
400
200
0
-200
2
4
6
8
10
12
pH
Figure A19: a) Quasi-Fermi level of electrons (∗EFn ) estimation by the Method of Roy using pH
independent methyl viologen as indicator. This figure was adapted from Ref. [221] .
b) Change of the potential of a Pt electrode immersed in an irradiated suspension of
TiO2 reference sample and different Nb-doped TiO2 materials with increasing pH.
Assuming Nernstian shift of band edges (0.059 V/pH unit) the values of ∗ EFn at pH =
7 can be then obtained from equation: ∗ EFn = EMV2+/+• + 0.059 (pH0 - 7), where pH0
is the inflection point on the titration curve. This procedure gives ∗ EFn = -0.56 V vs.
NHE for pristine TiO2 and ∗ EFn = -0.59 V vs. NHE for Nb-doped TiO2 .
a)
b)
TiO2-PH
TiO -PH
Nb01-PH
2
(F(R) x hv)
F(R) [a.u.]
0.5
Nb01-PH
300
400
500
600
wavelength [nm]
700
800
1.0
1.5
2.0
2.5
3.0
3.5
4.0
energy [eV]
Figure A20: a) UV-Vis absorption measurements and b) Tauc plots of polyheptazine-modified Nbdoped TiO2 samples.
8.3 Supporting Figures Chapter 6
XXV
150
Nb01-PH + IrO
J [ A cm
-2
]
125
Nb01 + IrO
2
2
100
75
50
light on
light off
>420nm
25
>420nm
0
0
15
30
45
60
75
90 105 120 135 150 165
t [min]
counts [a.u.]
1000 counts
Figure A21: Photocurrent response during irradiation with visible light (λ > 420 nm).
7)
6)
5)
4)
3)
2)
1)
20
30
40
2
50
60
70
Figure A22: X-ray diffraction pattern of Nb-doped TiO2 materials with different Nb concentrations
treated at 500 ◦ C in 150 ml/min NH3 for 4 h. The theoretical Nb content in the
different materials was 1) 1 at%, 2) 2 at%, 3) 5 at%, 4) 10 at%, 5) 15 at%, 6) 20 at%,
and 7) 25 at%.
8 Appendix
XXVI
a)
b)
404
402
400
398
396
394
392
binding energy [eV]
Figure A23: XPS N 1s region scans of Nb-doped TiO2 with 10 at% Nb a) treated at 700 ◦ C in 150
ml/min NH3 and b) post-treated in syn. air.
9 Curriculum Vitae
Bastian Mei
Ruhr-University Bochum
Department of Chemistry and Biochemistry
Bochum, 44780
Phone: 0049-(0)234-32-22341
Fax: 0049-(0)-32-15114
Email: Bastian.Mei@rub.de
Homepage: www.techem.rub.de/
Education
Ph.D. Chemistry, Ruhr-University Bochum, 2009 - present.
Supervisor: Prof. Dr. Martin Muhler
Title: ”Heterogeneous photoelectrochemistry and photocatalysis of TiO2 -based nanomaterials: Towards photocatalytic CO2 reduction.”
M.Sc. Chemistry, Ruhr-University Bochum, July 2009.
Supervisor: Prof. Dr. Martin Muhler
Title: ”Spectroscopic characterization of niobium-doped titania for fuel cell applications.”
B.Sc. Chemistry, Ruhr-University Bochum, March 2008.
Research Experience
Ruhr-University Bochum, Department of Chemistry and Biochemistry, Bochum
PhD studies, August 2009 - present.
University of Twente, Netherlands
Visiting Student, May 2012.
Dalian Institute of Chemical Physics, China
DAAD Exchange Programm, September - December 2010.
Cardiff University, Cardiff Catalysis Institute,Wales, GB
Erasmus Internship, October 2007 - January 2008.
9 Curriculum Vitae
Research Interests
Solar fuels; photoelectrochemical hydrogen production, photocatalytic CO2 reduction.
TiO2 -based single-site and bulk photocatalysts
Characterization techniques: UV-Vis spectroscopy, Raman spectroscopy, XRD and ultrahigh vacuum technologies including UPS and XPS.
Research
Peer-Reviewed Journal Articles
B. Mei, H. Byford, M. Bledowski, L. Wang, J. Strunk, M. Muhler, R. Beranek, Solar Energy
Materials and Solar Cells, 2013, accepted.
F. E. Oropeza, B. Mei, I. Sinev, M. Muhler, J. Strunk, Applied Catalysis B: Environmental,
2013, accepted.
J. Schartner, J. Güldenhaupt, B. Mei, M. Rögner, M. Muhler, K. Gerwert, C. Kötting, Journal American Chemical Society, 2013, 135, 4079.
W. Busser, B. Mei, M. Muhler, ChemSusChem, 2012, 5, 2200.
A. Ramakrishnan, S. Neubert, B. Mei, J. Strunk, L. Wang, M. Bledowski, M. Muhler, R.
Beranek, Chemical Communication, 2012, 48, 8556.
B. Mei, A. Becerikli, A. Pougin, D. Heeskens, I. Sinev, W. Grünert, M. Muhler, J. Strunk,
Journal of Physical Chemistry C, 2012, 116, 14318.
B. Mei, M. D. Sanchez, T. Reinecke, S. Kaluza, W. Xia, M. Muhler., Journal of Materials
Chemistry, 2011, 21, 11781.
D. Schäfer, C. Madare, A. Savan, M. D. Sanchez, B. Mei, W. Xia, M. Muhler, A. Ludwig,
W. Schuhmann, Analytical Chemistry, 2011, 83, 1916.
W. Xia, B. Mei, M. D. Sanchez, J. Strunk, M. Muhler, Journal of Nanoscience and Nanotechnology, 2011, 11, 8152.
C. Jin, M. Holz, W. Xia, B. Mei, S. Kundu, M. Muhler, Electrochemical Society Transactions, 2009, 25, 763.
Papers Under Review
B. Mei, A. Pougin, M. Muhler, J. Strunk, submitted to Journal of Catalysis.
E. Ventosa, B. Mei, W. Xia, M. Muhler, W. Schuhmann, submitted to ChemSusChem.
S. Neubert, A. Ramakrishnan, B. Mei, et al., submitted to ACS Catalysis.
Work in Progress
XXVIII
Bastian Mei
B. Mei, et al., ”Effects of Au and ZnO on the structure and photocatalytic activity of
TiOx/SBA-15 materials”.
E. Ventosa, W. Xia, B. Mei, et al., ”Ammonia annealed TiO2 as negative material in Li-ion
batteries: N-doping or oxygen deficiency?”.
Conference Presentations
”Improvement of TiOx/SBA-15 for photocatalytic applications by the addition of ZnO and
Au.” B. Mei, et al. International Conference on Catalysis in Munich, July 1-6, 2012. Poster
Presentation
”Effect of Nb doping on TiO2 and TiO2 -polyheptazine hybrid materials in wavelength and
potential-dependent photocurrent measurements.” B. Mei, et al. 243rd Spring Meeting of
the American Chemical Society in San Diego, CA, March 25-29, 2012. Oral Presentation
”CO2 adsorption of potential photocatalysts for CO2 reduction.” B. Mei, et al. EuropaCat
X, Glasgow, August 2011. Poster Presentation
”The synthesis and characterization of NbOx /TiO2 nano composites by spray drying: an
efficient and scalable method.” B. Mei, et al. Annual Meeting of the German Catalysis
Society, Weimar, March 16-18, 2011. Poster Presentation
Activities
Member of the GDCh (German society of chemistry), Member of the Ruhr-University Research
School (June 2010 present), Organization of the JCF Spring Symposium 2009 & EYCN Satellite Event, Speaker of the Young Chemists Bochum (2010 - 2011).
Interests include running marathon, swimming, reading, soccer and traveling.
Last updated: May 10, 2013
XXIX