ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2018

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1726

Infrared spectroscopy studies ofadsorption and photochemistry onTiO2 surfaces

From single crystals to nanostructured materials

ANDREAS MATTSSON

ISSN 1651-6214ISBN 978-91-513-0456-4urn:nbn:se:uu:diva-362326

Dissertation presented at Uppsala University to be publicly examined in Häggsalen,Ångströmlaboratoriet, Lägerhyddsvägen 1, Uppsala, Wednesday, 21 November 2018 at09:15 for the degree of Doctor of Philosophy. The examination will be conducted in English.Faculty examiner: Associate Professor Per-Anders Carlsson (Chalmers University ofTechnology).

AbstractMattsson, A. 2018. Infrared spectroscopy studies of adsorption and photochemistry on TiO2

surfaces. From single crystals to nanostructured materials. Digital Comprehensive Summariesof Uppsala Dissertations from the Faculty of Science and Technology 1726. 117 pp. Uppsala:Acta Universitatis Upsaliensis. ISBN 978-91-513-0456-4.

TiO2 based photocatalysis is a green nanotechnology that can be used for removal of pollutantsfrom water and air, as well as making synthetic fuels from water and carbon dioxide.Said photocatalysis has received major research interests during the last decades. Despitethese efforts, many elementary processes that occur on the photocatalyst surface are notfully understood and, therefore, limit our ability to purposefully manufacture more efficientphotocatalytic materials. The objective of this thesis is to provide new understanding at amolecular level of important adsorbate species on the TiO2 surfaces.

Fundamental properties of adsorption and photochemistry of primarily formic acid ondifferent TiO2 surfaces, ranging from single crystals to nanoparticles, have been studied usinginfrared spectroscopy. A method to simulate IR spectra have been developed and, combinedwith experimental data, has been proven to be a powerful tool to identify different adsorbategeometries on the surface. In the presence of oxygen, a thermally activated and irreversiblereaction between formate and oxygen adatoms takes place on the single crystal rutile (110)surface to yield hydrogen bicarbonate surface complexes. For disordered single crystal surfaces,the adsorption geometry of formate changes due to exposure of Ti3+ atoms on the surface,and the adsorption spectra shows resemblances with that observed for formate adsorption onnanocrystalline surfaces.

Illumination with UV light results in small changes of the formate coverage on the disorderedsingle crystal and nanocrystalline rutile surfaces, whereas on the rutile (110) surface onlyminiscule changes in formate coverages are seen. This is due to the lack of oxygen electronacceptors and OH/H2O electron donors in the vacuum environment, which results in a muchlower degradation rate compared to measurements made at ambient conditions. Furthermore, itis shown that the coordination of the formate molecule on various TiO2 surfaces has a profoundeffect on the photocatalytic degradation rate, with bidentate coordinated formate moleculesbeing most resilient towards oxidation.

The results presented here shows that additional insight in the processes on the TiO2

photocatalyst surface can be obtained by combining spectroscopic studies of single crystals andnanocrystalline films and that it is possible to unravel adsorption geometries on surfaces bycombining experimental and simulated IR spectra.

Andreas Mattsson, Department of Engineering Sciences, Solid State Physics, Box 534,Uppsala University, SE-751 21 Uppsala, Sweden.

© Andreas Mattsson 2018

ISSN 1651-6214ISBN 978-91-513-0456-4urn:nbn:se:uu:diva-362326 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-362326)

To my family

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Mattsson, A., Hu, S., Hermansson, K., Österlund, L. (2014) Ad-

sorption of formic acid on rutile TiO2 (110) revisited: An infra-red reflection-absorption spectroscopy and density functional theory study. The Journal of Chemical Physics, 140:034705

II Mattsson, A., Hu, S., Hermansson, K., Österlund, L. (2014) In-frared spectroscopy study of adsorption and photodecomposi-tion of formic acid on reduced and defective rutile TiO2 (110) surfaces. Journal of Vacuum Science and Technology A, 32:061402

III Hu, S., Wang, Z., Mattsson, A., Österlund, L., Hermansson, K. (2015) Simulation of IRRAS spectra for molecules on oxide surfaces: CO on TiO2 (110). The Journal of Physical Chemistry C, 119:5403-5411

IV Mattsson, A., Österlund, L. (2017) Co-adsorption of oxygen and formic acid on rutile TiO2 (110) studied by infrared reflec-tion-absorption spectroscopy. Surface Science, 663:47-55

V Mattsson, A., Österlund, L. (2018) Formic acid adsorption and photodecomposition on TiO2: A comparative study of rutile single crystals, sputtered and sol-gel synthesized anatase, brookite and rutile thin films. In manuscript,

Reprints were made with permission from the respective publishers.

My contribution to the appended papers

I All experimental work except DFT calculations most of the data

analysis and most of the writing II All experimental work and most of the writing III Part of the writing IV All experimental work, data analysis and most of the writing V All experimental work, data analysis and most of the writing

Papers not included in the thesis

I Mattsson, A., Österlund, L. (2011) Adsorption and photoin-duced decomposition of acetone and acetic acid on anatase, brookite and rutile TiO2 nanoparticles. Journal of Physical chemistry C. 114:14121-14132.

II Mattsson, A., Lejon, C., Bakardjieva, S., Štengl, V., Österlund, L. (2013) Phase stability and surface chemical properties of photocatalytic active Zr and Y co-doped anatase TiO2 nano-particles. Journal of Solid State Chemistry, 199:212-223.

III Wäckelgård, E., Bartali, R., Gerosa, R., Laidani, N., Mattsson, A., Micheli, V., Rivolta, B. (2014) New cermet coatings for mid-temperature applications for solar concentrated combine heat and power system. Energy Procedia 48:242-249.

IV Henych, J., Štengl, V., Kormunda, M., Mattsson, A., Öster-lund, L. (2014) Role of bismuth in nano-structured doped TiO2 photocatalyst prepared by environmentally benign soft synthe-sis. Journal of Materials Science, 49:3560-3571.

V Stefanov, B. I., Lebrun, D., Mattsson, A., Granqvist, C. G., Österlund, L. (2015) Demonstrating online monitoring of air pollutant photodegradation in a 3D printed gas phase photoca-talysis reactor. Journal of Chemical Education, 92:678-682.

VI Henych, J., Štengl, V., Mattsson, A., Österlund, L (2015) In situ FTIR spectroscopy study of the photodegradation of acet-aldehyde and azo dye photobleaching on bismuth-modified TiO2. Photochemistry and Photobiology, 91:48-58.

VII Wäckelgård, E., Mattsson, A., Bartali, R., Gerosa, R., Got-tardi, G., Gustavsson, F., Laidani, N., Micheli, V., Primetzho-fer, D., Rivolta, B. (2015) Development of W-SiO2 and Nb-TiO2 solar absorber coatings for combined heat and power sys-tems at intermediate operation temperatures. Solar Energy Materials & Solar Cells, 133:180-193.

VIII Johansson, M. B., Mattsson, A., Lindquist, S-E., Niklasson, G. A., Österlund, L. (2017) The importance of oxygen vacancies in nanocrystalline WO3-x thin films prepared by DC magnetron sputtering for achieving high photoelectrochemical efficiency. Journal of Physical Chemistry C. 121:7412-7420.

IX Österlund, L. Mattsson, A., Brischetto, M., Johansson Byberg, J. Stefanov, B. I., Ji, Y., Niklasson G. A. (2018) Spectral se-lective solar light enhanced photocatalysis: TiO2/TiAlN bi-layer films. Topics in Catalysis, https://doi.org/10.1007/s11244-018-1011-5.

X Henych, J. Stengl, V. Mattsson, A. Tolasz, J. Österlund, L. (2018) Chemical warfare agent simulant DMMP reactive ad-sorption on TiO2/graphene oxide composites prepared via tita-nium peroxo-complex or urea precipitation. Journal of Haz-ardous Materials 359:482-490.

Contents

1  Introduction ......................................................................................... 17 

2  Photocatalysis ...................................................................................... 21 2.1  Principles of semiconductor photocatalysis .................................... 21 2.2  Applications of photocatalysis ........................................................ 26 

3  Titanium dioxide .................................................................................. 29 3.1  Titanium dioxide phases ................................................................. 29 3.2  Rutile TiO2 surface structures ......................................................... 31 

3.2.1  The rutile (110) surface .......................................................... 32 3.2.2  The rutile (011) surface .......................................................... 33 3.2.3  The rutile (100) surface .......................................................... 34 

3.3  Defects in rutile TiO2 ...................................................................... 35 3.3.1  Bulk defects ........................................................................... 35 3.3.2  Surface defects ....................................................................... 35 3.3.3  Surface oxygen vacancies ...................................................... 36 

3.4  Anatase and brookite TiO2 surface structures ................................. 36 3.4.1  The anatase (101) surface ...................................................... 37 3.4.2  The anatase (001) surface ...................................................... 38 

3.5  Adsorption on the rutile (110) surface ............................................ 39 

4  Vibrational spectroscopy ..................................................................... 41 4.1  Principles of infrared spectroscopy ................................................. 41 

4.1.1  Fundamental vibrations .......................................................... 41 4.1.2  Anharmonicity and line broadening ....................................... 43 4.1.3  Vibration modes ..................................................................... 46 

4.2  Surface infrared spectroscopy ......................................................... 47 4.2.1  Polarization of light ............................................................... 47 4.2.2  Three layer model .................................................................. 49 4.2.3  Simulation of infrared-reflection absorption spectrum .......... 51 

4.3  The FT-IR spectrophotometer ......................................................... 54 4.4  Raman spectroscopy ....................................................................... 57 

5  Single crystal sample preparation and characterisation ....................... 59 5.1  Sample preparation ......................................................................... 59 5.2  Low energy electron diffraction ...................................................... 60 5.3  Formic acid dosing .......................................................................... 62 

5.4  Infrared reflection-absorption spectroscopy setup .......................... 62 5.5  UV illumination .............................................................................. 63 

6  Thin film sample preparation and characterisation .............................. 65 6.1  Magnetron sputtering deposition ..................................................... 65 6.2  Raman characterisation ................................................................... 66 6.3  Grazing incidence X-ray diffraction ............................................... 67 6.4  Scanning electron microscopy ........................................................ 68 6.5  In-situ IR transmission measurements ............................................ 69 

7  Results ................................................................................................. 71 7.1  CO adsorption on the rutile (110) surface ....................................... 71 7.2  Formic acid adsorption on the rutile (110) surface ......................... 74 7.3  Interaction between formate and oxygen on the rutile (110) surface ...................................................................................................... 78 7.4  Formate adsorption on disordered rutile single crystal and rutile nanocrystalline surfaces ........................................................................... 85 7.5  Formic acid adsorption on anatase and brookite nanoparticle films at atmospheric conditions ................................................................ 89 7.6  Interaction between formate and light on the rutile TiO2 surface at UHV conditions .................................................................................... 90 7.7  Interaction between formate and light on anatase, brookite and rutile TiO2 nanoparticle films at atmospheric conditions ......................... 93 

8  Concluding remarks and outlook ......................................................... 95 

9  Sammanfattning på svenska .............................................................. 101 

10  Acknowledgements............................................................................ 105 

11  Bibliography ...................................................................................... 107 

Abbreviations

, Real space lattice vectors

*, *, * Primitive reciprocal vectors

AA,X Peak area of adsorbate (A) or surface hydrogen complex (X)

c Speed of light

dhkl Interplanar distance for the {hkl} planes

De Potential well depth

DC Direct current

DFT Density functional theory

e* Effective charge

E Electric field

Eg Bandgap

En,p Normal component of p-polarized electric field

Ep P-polarized electric field

Er Activation energy for chemical reaction

Et Activation energy for dehydrogenation reaction

Et,p Tangential component of p-polarized electric field

Et,s S-polarized electric field

eV Electron volt

FT-IR Fourier Transform-Infrared

FWHM Full width half maximum

Reciprocal lattice vector

{hkl} Miller index for a set of lattice planes

h Plancks constant

ht Adsorbate layer thickness

h+ Hole

H+ Hydrogen ion

HObr Bridging OH group

IR Infrared

IRRAS Infrared reflection-absorption spectroscopy

k Reaction rate

Wave vector

kb Boltzmanns constant

kr Pseudo reaction rate

kt Thermal removal rate

LEED Low energy electron diffraction

LN2 Liquid nitrogen

m* Effective mass

MCT Mercury cadmium tellurium

ML Monolayer

Complex refractive index

Surface normal vector

n Real part of complex refractive index

N Degrees of freedom

Ns Number of adsorbate molecules

np-TiO2 Nanocrystalline TiO2

NHE Normal hydrogen electrode

Oad Oxygen adatom

Obr Bridging oxygen atom

O2c Two-fold coordinated oxygen atom

Ovac Bridging oxygen vacancy

OH OH-group

OHt OH group on Ti5c atom, terminal OH

o-TiO2 Oxidized TiO2

Oxygen partial pressure

Vector coordinate

re Equilibrium distance

Transition quantum number

R0 Reflectivity without adsorbate layer

Ra Reflectivity with adsorbate layer

Rp Reflectivity for p-polarized light

Rs Reflectivity for s-polarized light

r-TiO2 Reduced TiO2

s-TiO2 Stoichiometric TiO2

sp-TiO2 Disordered TiO2 single crystal

SEM Scanning electron microscopy

SNR Signal to noise ratio

STM Scanning tunneling microscope

t Time

Ti5c Five-fold coordinated titanium atom

Ti6c Six-fold coordinated titanium atom

TiO2 Titanium dioxide

TPD Temperature programed desorption

UHV Ultra-high vacuum

UV Ultraviolet

VOC Volatile organic compounds

Å Ångström

a Absorption coefficient

e Electronic part of the polarizability

v Vibrational part of the polarizability

p Polarizability of adsorbate layer or molecule

Line width

Wagging vibration

R Change in reflectivity, R0-Ra

a Dielectric function of adsorbate layer

a,z z-component of the adsorbate dielectric function

s Dielectric function for the substrate

v Dielectric function for vacuum

A Surface coverage of formate type A

i Angle of incidence

O Surface coverage of oxygen adatoms

X Surface coverage of surface hydrogen complex

d Diffraction angle

Imaginary part of complex refractive index

Wavelength

Electric dipole moment operator

µind Induced dipole moment

Stretching vibration

as Asymmetric stretching vibration

s Symmetric stretching vibration

Fundamental vibration frequency

Quantum number

Wavenumber

Azimuthal angle

Wavefunction

e Wavefunction for excited state

i Wavefunction for initial state

Angular frequency

C Catalyst

C* Excited state of catalyst

I Intermediate species in chemical reaction

P Product of chemical reaction

R Reactant in chemical reaction

R* Excited state of reactant

17

1 Introduction

With increasing energy demand in the modern society sustainable solutions are necessary if the aim to limit global warming should be successful. The International Energy Agency (IEA) has projected that the worlds energy consumption, Figure 1.1, will increase with about 30% until the year 2040[1]. The largest increase is expected to occur in the developing countries as a result of economic growth and improved living conditions. In the developed countries the energy increase is estimated to be lower but none the less in-creasing. Thus in order to adhere to the UN climate goals and have a sustain-able economic growth energy saving measures, besides an increased energy production from renewable sources, is necessary. This is especially true for the developed countries where most of the world energy demand, counted per capita, is used today.

Figure 1.1. World energy consumption from 1990 to 2015 and projected towards year 2040. The energy demand is divided between OECD and non-OECD countries. The figure is adapted from the International energy outlook 2017 provided by IEA[1].

IEA have divided the total energy use in the world into three different sec-tors; industrial, transport and buildings. The major consumer is the industrial sector whereas the energy use for the transportation and buildings sectors are similarly large. These later sectors is expected to have a more rapid growth in energy demand than the industrial sector and in 2040 it is forecasted that about 20% of the world’s total energy consumption is related to buildings[1]. In the industrialized countries buildings’ contribution to the energy demand

18

is even higher, and in 2010 it was reported that as much as 40% of the Unit-ed States energy demand was due to buildings[2]. Thus by making more en-ergy efficient buildings, the need for energy can be reduced. By designing buildings to be more air tight, not only does the building become more ener-gy efficient but also the indoor climate becomes more comfortable due to the decreased heating and cooling demands. On the downside an increased ener-gy efficiency of the building may result in a build-up of high concentrations of volatile organic compounds (VOC) in the indoor climate. VOCs are com-pounds that can evaporate at normal indoor conditions and they emanate from a large variety of household products and domestic chemicals, such as plastics and glues used in adhesives, carpets, paint, furniture upholstery as well as detergents and cleaning products[3]–[6]. The most commonly found VOCs are BTEX compounds (benzene, toluene, ethylbenzene and xylene) or chlorinated hydrocarbons (chloroform and tetrachloroethene)[7]. Other com-pounds such as carbonyls (formaldehyde, acetaldehyde and acetone) are also common[4],[6]. These chemicals are regarded as hazardous chemicals and are in some cases also reported to be cancerogenic[8],[9]. This problem with in-door air pollution is commonly manifested as the sick building syndrome (SBS), where the low quality of the indoor air is known to cause diffuse illnesses in people working or living in such building. It has been reported that as many as 30% of all new or refurbished buildings can be affected by the SBS syndrome[10]. The outdoor environment also exposes us to a wide-spread range of different chemicals stemming from emissions into air and water from industrial- and manufacturing processes, contamination of water by leakage from contaminated soils and waste dumps as well as emissions such as hydrocarbons and nitrogen oxides from combustion engines in our vehicles. The International Agency for Research on Cancer (IARC) recently declared that the urban outdoor air is cancerogenic and that more than 200 000 people where estimated to have died from lung cancer caused by out-door air pollution in 2010[11]. More recently WHO also concluded that nine out of ten people breathe polluted air and that worldwide seven million peo-ple die prematurely due to air pollution[12].

How to efficiently and cost-effectively deal with these threats to our daily life is an open question. Of course, the best solution would be to eliminate the sources of hazardous chemical compounds. Although this must be the prime goal, it is probably not feasible even in a long-term perspective due to the diverse and widespread nature of these sources, which sometime are not well-characterized. Moreover, human activity influences the chemistry in the indoor environment, which is both unavoidable and unpredictable[13],[14]. The use of functional and specialized materials can be one part of the solution to the problem. Titanium dioxide, TiO2, is a material that has found its way into many different uses due to its material properties. It is used as a white pig-ment in paint and as a physical UV blocker in sun creams[15] and is investi-gated in medical applications as a way to treat cancer[16],[17]. TiO2 is also

19

being used in renewable energy production in mesoporous solar cells where it acts as a support for the light harvesters and as a material for charge transport[18]. This new type of photovoltaics is expected to find its ways into areas such as domestic device charging and architectural facades, where the traditional crystalline silicon and thin film photovoltaic devices are less suit-able[19]. TiO2 is also investigated as a material used for water splitting, and is thus intended to be used in hydrogen production, using sun light as an ener-gy source[20]–[22]. A main trend in the photocatalysis research today is the production of synthetic chemicals and in particular fuels by photo-reduction of CO2

[23],[24]. Historically the main application of titanium dioxide photoca-talysis, and also the topic of this thesis, is however to use it as a photocata-lytic material, for air and water cleaning. In this process sun light is harvest-ed in the material to create chemical reactions that decompose and mineral-ize pollutants[25]–[30]. By employing photocatalytic air filtering technologies, reduced ventilation capacity and down-sizing of ventilation rates are possi-ble, and energy savings up to 30% have been reported with down-sizing of the buildings heat, ventilation and air condition equipment by 25% and with no additional initial investments costs[31].

With the increasing research interest in photocatalysis both from a fun-damental and an application perspective our knowledge of the processes that occurs on the photocatalyst surface has increased tremendously[32]–[35]. For instance it is now known that TiO2 has a high selectivity for small carbox-ylates, which is manifested in the formation of a formate monolayer on the surface at ambient conditions even though the formate concentration is much lower than for other organic species in the ambient air[36]. Despite this, even the most simple chemical reactions on a model catalyst surface, such as the rutile TiO2 (110) surface, is still not fully understood. Therefore additional studies on the fundamental processes are needed in order to increase our knowledge and thereby enable us to produced photocatalytic materials with improved properties. The results from these fundamental studies on proto-typical surfaces also need to be translated into application scenarios with polycrystalline anatase and rutile materials. In the present thesis fundamental investigations of how small molecules such as formic acid and carbon mon-oxide adsorb on and interact with the rutile TiO2 photocatalyst surface, both the model single crystal rutile (110) surface and polycrystalline surfaces have been investigated.

21

2 Photocatalysis

In short one can say that a photocatalyst is a material that uses light to facili-tate chemical (reduction and oxidation) processes on its surface without be-ing consumed. It may sound simple but the actual process is complicated and involves several steps with limitations and pitfalls as will be discussed in the following section.

2.1 Principles of semiconductor photocatalysis Up till now the word photocatalysis has been used without an explanation of the actual reaction mechanisms and how it is different to thermal catalytic and photochemical reactions. For a thermal reaction, Figure 2.1a, the reac-tant, R, needs thermal energy to overcome the reaction barrier to form an intermediate product, I, before the end product, P, is formed. If the reac-tions take place on a catalyst surface, C, different reaction intermediates, I’ is possible and as a result the energy barrier can be lowered. In a photochem-ical reaction, Figure 2.1b, photons are used to put the reactant in an excited state, R*, which is the starting state of the chemical reaction. For a photo-catalyzed reaction, Figure 2.1c, the catalyst is active only in the excited state, C*, which is obtained by absorption of light in the catalyst. The reac-tion to the end product then proceeds through intermediates of the reactant and the catalyst surface, C’. This can be compared to the thermal reaction where the reactions occur on the ground state surfaces of the catalyst.

The reaction mechanisms that occur inside and on the surface of a semi-conductor photocatalyst particle is given in Figure 2.2, and as it can be seen, it involves several different elementary steps.

22

Figure 2.1. Reaction paths for a reaction from reactant R to end product P via in-termediate species, I, for a) thermal reaction, where the dotted and solid line illus-trates the reaction without and on a catalyst, C, surface, b) photochemical reaction where the reactant is placed in an excited state, R*, due to photon adsorption and c) photocatalyzed reaction where the catalyst surface, C*, is excited by photon adsorp-tion.

Figure 2.2. Schematic figure of the photocatalytic process on a nanoparticle starting with (1) creation of electron-hole (e- and h+) pairs by absorption of photons followed by (2) diffusion of the charge carriers to the surface where (3) charge transfers occur between photogenerated charge carriers and adsorbed molecules.

The first step in the photocatalytic process is the creation of charge carriers by absorption of light, which facilitates inter-band transitions in the material. There are a wide range of different materials that can be used for photocatal-

23

ysis ranging from single metal oxide semiconductors to doped and mixed metal oxides layered (perovskites) structures, as well as nanostructured composites and core-shell structures[17],[25],[29],[37],[38]. Most of the research in photocatalysis has been done on wide band gap metal oxide semiconductors, and TiO2 in particular[17],[29],[30],[37]. In these types of photocatalysts the charge carriers are created by excitation of an electron from the valence band to the conduction band and thereby creating an electron-hole pair. This is done by absorption of a photon whose energy is larger than the optical band gap, Eg. One of the drawbacks with TiO2 as a photocatalyst is the large band gap. The energy of the band gap changes with the TiO2 crystal structure, and for the three common TiO2 phases it is Eg(anatase) = 3.2 eV, Eg(rutile) = 3.0 and Eg(brookite) = 3.3 eV[39]. For small nanoparticles, less than 5 nm in size, a blue shift of the band gap due to quantum confinement have been reported to occur[33]. Much of the research into TiO2 photocatalysis has been devoted to find a way to dope the photocatalyst in order to lower the band gap without introducing negative side effects, such as recombination centres that lowers the performance[17],[29],[40].

The second step is migration of the charge carriers to the surface. The lifetime of an electron-hole pair is not infinite, after a certain time the elec-tron and the hole will recombine unless they are separated from each other. After the creation of the electron-hole pairs the charge carriers will, if they not recombine immediately, perform a random walk (if no bias is applied and since band bending usually is negligible) until they either reach the sur-face and are trapped there, or are trapped in the bulk and eventually recom-bine there. Trapping of electrons and holes can be beneficial for the lifetime of the charge carriers because the electrons and holes need to be in the vicin-ity of each other to recombine. If the trapping sites for electrons and holes are separated the trapped charge carriers are hindered to get close to each other and thus prevent recombination. The trapping process is relatively quick, in the order of nanoseconds, compared to the lifetimes of the trapped charge carriers[28],[29],[32]. In nano-porous TiO2 charge carrier lifetimes in the microsecond range have been measured[32], and even hours have been in-ferred for deep traps[41],[42]. In a photocatalyst trapping of electrons and holes normally occurs in band gap states and this is the main competitor of the charge carriers’ random walk to the photocatalyst surface. Thus the mobility of the electron and holes needs to be high in order to obtain a high rate of charge carriers reaching the photocatalyst surface.

Charge carriers trapped at the surface can either recombine or participate in the next step of the photocatalytic process, which is the charge transfer from the photocatalyst to molecules adsorbed at the photocatalyst surface. Since oxygen and water vapour is always present in air, molecular oxygen and water molecules and its dissociated fragments, hydroxyl groups and protons will be absorbed on the surface. Other types of molecules present in air, like VOCs, will also adsorb on the surface. The charge carriers at the

24

surface can then interact with the adsorbed molecules. Conduction band electrons will be interacting with so called acceptor molecules such as oxy-gen, whereby an electron is transferred from the photocatalyst to the lowest unoccupied energy level of the molecule. The reason for the name acceptor molecules hence comes from the fact that these molecules accept an extra electron from the photocatalyst. Likewise the holes will be transferred to donor molecules like water and hydroxyls, or in a physically more correct picture, an electron is transferred from the molecule to the photocatalyst. Thus oxygen and hydroxyl radicals are formed on the surface of the photo-catalyst. These OH• and O2

•- radicals are long-lived, lifetimes in the order of 10 minutes have been observed[43], reactive and readily interact with other molecules in reduction-oxidation processes, or with each other to form other radicals and reactive species. Thus reduction and oxidation processes of air and water pollutants can be obtained. The charge transfer between the pho-tocatalyst and the adsorbed molecules are affected by bandgap states, caused by defects and oxygen vacancies in the photocatalyst. The likelihood of the charge carriers to be trapped increases with an increasing number of band gap states. This in turn reduces the charge transfer rate between the photo-catalyst and the electron acceptors and donors on the surface. The creation of oxygen and hydroxyl radicals is summarized by the following reaction schemes, eq. (2.1) - (2.4).

(2.1)

• (2.2)

• (2.3)

• (2.4)

Following this scheme one can be led to believe that any semiconductor will be suitable for photocatalysis and the smaller the band gap the better it is because less energetic photons are needed to create the electron-hole pairs. Unfortunately it is not as simple as that. The band gap of the photocatalyst influences the photocatalytic activity in several ways and it has been shown that for metal oxides an increased band gap energy results in an increased activity for photooxidation of oxygen and photoreduction of methane[44]. Furthermore for every reduction-oxidation process there is a certain energy needed. For these chemical reactions to occur it is important that the conduc-tion band and valence band edges in the photocatalyst are positioned in such a way that it is energetically favourable for the charge carriers to create radi-cals. For a reduction process to occur the conduction band edge must be

25

more positive (negative) than the reduction potential of the chemical species relative to the vacuum level (normal hydrogen electrode, NHE). Likewise for an oxidation process to occur the valence band edge should be more neg-ative (positive) than the oxidation potential of the chemical species. In Fig-ure 2.3 below, the valence and conduction band edges for different semicon-ducting metal oxides are shown together with the energy potentials, versus the absolute vacuum energy scale and the normal hydrogen electrode energy scale, for different reduction and oxidation processes. The absolute value of the NHE at 25°C is 4.44 eV with respect to the vacuum level. Thus the ener-gy levels versus the NHE can easily be converted to absolute values. From Figure 2.3 it can be seen that the conduction band edge for TiO2 is suitably positioned for reduction of O2 atoms into oxygen radicals O2

•- and that the valence band edge is much more positive (relative the NHE level) than the energy needed for oxidation of OH- into hydroxyl radicals, OH•. For other semiconductors, such as iron oxide (hematite), the band edges are not posi-tioned equally favourable for the formation of these radicals, although some degree of band gap engineering can be made by novel nanostructuring of the photocatalyst material. Another aspect of photocatalysis is that the photo-catalyst needs to be resilient to the created charge carriers. Cadmium sul-phide, which from Figure 2.3 seems attractive for photocatalytic applications due to the lower band gap and favourably positioned valence and conduction band edges suffers from problems with photo-induced corrosion, i.e. degra-dation of the photocatalyst by reduction-oxidation processes created from the photogenerated charge carriers[25].

26

Figure 2.3. Valence band and conduction band edges for different metal oxide semi-conductors at pH 1 versus the normal hydrogen electrode (NHE) as well as the abso-lute potential. The energy for common reduction-oxidation processes is also shown. The figure is adapted from Grätzel[45].

2.2 Applications of photocatalysis Even though all elementary reactions in a photocatalytic process is not fully understood applications and devices with photocatalytic materials are nowa-days commercially available since more than 20 years[46]. One of the first widely spread commercially available photocatalytic products was the self-cleaning window where a thin layer of TiO2 is coated on window glass[47]. These TiO2 coatings uses two photocatalytic processes: (i) dissociation of hydrogen bonded water clusters, and (ii) decomposition of hydrocarbons adsorbed on the TiO2 surface[48],[49]. The effect is that the surface will be hydroxylated and becomes superwetting (or super-hydrophilic as it is also called in the literature) when it is irradiated with UV-light. A thin water film is thus formed on the window in wet and moist conditions. This water film washes away the dirt and leaves no streaks on the window, Figure 2.4. The exact processes that occur during the transition from the super-wetting state back to the hydrophobic dark state sans UV illumination is not fully under-

27

stood. It has recently been suggested that it is caused by the spontaneous formation of a well-ordered layer of hydrophobic carboxyl molecules from the ambient environment on the surface[36].

Figure 2.4. A photograph of a self-cleaning window (right) and an ordinary window (left) after a rainfall. On the self-cleaning window no water streaks and droplets are seen due to the hydrophilic properties of the TiO2 layer. The figure is adapted from Pilkington[50].

TiO2 has also been incorporated into building materials, such as ceramic tiles, cements and plastics, to obtain a self-cleaning surface, which will keep the aesthetic appearance of the building over time[47]. It has been estimated that a conventional building should be washed at least every five years to keep a good appearance, whereas a building using photocatalytic self-cleaning materials should preserve the good appearance for at least twenty years[32]. This self-cleaning effect can clearly be seen in Figure 2.5, where a tent, placed outdoors for an extended period of time becomes black, whereas the part coated with TiO2 still remains white. Another benefit of having a photocatalytic material incorporated in the urban environment, either on the façade of the buildings or in the roads and pavements, is that it helps to re-duce pollutants such as NOx emitted from automotive exhausts. Field trials at different locations in Europe (Figure 2.6) and Japan have shown substantial reductions of the NOx levels, depending on the actual conditions at the test site[32],[47],[51],[52].

28

Figure 2.5. Outdoor exposure test of a PVC tent material where the left half of the tent is coated with TiO2. (a) Picture is taken at the start of the outdoor test and (b) three years later. The figure is adapted from Fujishima et. al.[32].

Figure 2.6. Parking lanes in Leien, Antwerpen with photocatalytic pavement blocks. The figure is adapted from Boonen et. al.[51]

29

3 Titanium dioxide

Although TiO2 exists in different phases and crystals, which have a variety of surface facets that are exposed to the surrounding environment, most of the fundamental surface science studies to date is performed on the rutile TiO2 (110) surface, simply because it is by far the most stable surface (i.e. it has the lowest surface energy), and can thus readily be prepared as large single crystals.

3.1 Titanium dioxide phases TiO2 can be found in several different phases and the most common ones are anatase, brookite and rutile[53]. These three can all be found naturally in dif-ferent minerals whereas others, like baddeleyite- and -PbO2-types of TiO2, are exotic high pressure phases, or metastable systems that can only be made synthetically[54]–[56], and are therefore of limited interest for photocatalytic applications. Anatase, brookite and rutile all have the same basic building block: a titanium atom surrounded by six oxygen atoms in an octahedral geometry, Figure 3.1. The arrangement of these octahedral is different for the anatase, brookite and rutile phases, creating a tetragonal crystal structure for anatase and rutile and an orthorhombic structure for brookite, Figure 3.2.

Figure 3.1. Arrangement of a TiO2 octahedral with a titanium atom (red) six-fold coordinated to oxygen atoms (blue) in an octahedral structure.

30

Rutile is the most stable phase at all temperatures and at pressures up to 60 kbar according to thermodynamic calculations[56],[57]. The differences in Gibbs free energy between the three phases is however small, so brookite and especially anatase can readily be synthesized[29],[37]. For nanoparticles size effects has to be accounted for, and then anatase and brookite are calcu-lated to be thermodynamically stable for particle sizes less than 11 nm and between 11-35 nm, respectively[29],[58],[59]. If the particles are larger than ap-proximately 35 nm, the rutile phase becomes the most thermodynamically stable phase. Anatase and brookite can readily be converted to rutile by heat-ing at ambient pressures. The exact phase transformation temperature de-pends on a variety of different factors such as crystallite size, morphology, impurity concentration, and is thus not exact and difficult to control[53],[58],[60]. Nonetheless a general consensus is that temperatures above 600°C is needed for transformation of anatase to rutile[39],[53]. This is one of the reasons why single crystals of rutile TiO2 with large area are readily available, whereas anatase TiO2 single crystals can only be obtained at small crystal sizes main-ly from naturally occurring minerals[61]–[63].

Figure 3.2. Crystal structures for a) anatase, b) rutile and c) brookite showing the stacking of the octahedral TiO6 building blocks using crystal data from the Ameri-can mineralogist[64],[65].

The photocatalytic properties of the three different TiO2 polymorphs differ[39] and it is commonly perceived that anatase is a better photocatalyst than rutile[35] whereas studies of brookite is more scarce but those that have been made report photocatalytic properties similar to that of ana-tase[33],[39],[66]. As nearly all comparison between the different TiO2 poly-morphs are made using powder samples where the size, shape and quality of the sample can greatly affect the photocatalytic activity a comparison be-tween the phases is not straightforward. It has been shown that the crystallin-ity of the nanoparticles have a large impact of the photocatalytic activity as a

31

sample with poor crystallinity will have more low-coordinated atoms and lattice distortions that will act as recombination centres and trap the charge carriers[33]. On the other hand, doping - either due to oxygen vacancies or other elements - moves the Fermi level towards the conduction band edge and results in higher carrier mobility, which is beneficial for charge separa-tion. Thus it is likely that there is an optimal balance between doping con-centration and crystallinity (to avoid charge carrier trapping), although this has hitherto not been verified either experimentally or theoretically. Fur-thermore differences in shape and grain size, where the rutile particles gen-erally are larger and thus have a lower specific surface area also affects the photocatalytic yield. Synthesis of small rutile nanoparticles, with sizes com-parable to those of anatase has showed comparable photocatalytic activity between the anatase and rutile phases[33]. It have also been shown that differ-ent facets of the TiO2 crystal have different photocatalytic activity[67],[68] making comparison between nanoparticles of different TiO2 polymorphs even more difficult. Photochemical single crystal studies of anatase are scarce[69]. Xu and co-workers have studied photo-oxidation of CO on anatase (101) and rutile (110) single crystal at 100 K and saw a higher photo-activity for the anatase single crystal, which was accompanied by a longer life time of the charge carriers in anatase[70]. On the other hand Ohsawa and col-leagues measured similar photodecomposition rates of trimethyl acetate on epitaxially grown anatase(001) and rutile(110) surfaces[71]. The effect of surface reconstructions complicates however the comparisons of the differ-ent polymorphs even more. The anatase (001) surface is not stable and re-constructs from a (1x1) to a (1x4) structure when heated to elevated tem-peratures, which is not the case for the anatase (101) surface.

There are also synergetic effects between different facets on the TiO2 crystal, where the oxidation and reduction sites in the photocatalytic process occurs on different facets[72],[73]. Similarly there is also a synergetic effect between anatase and rutile, where a mixture of the two has a higher photo-catalytic activity than phase pure TiO2. This is attributed to an improved charge separation due to the interfacial contact between the two different crystalline phases[33].

3.2 Rutile TiO2 surface structures Detailed understanding of the elementary processes occurring at the surface of the photocatalyst is of paramount importance. The reason for its im-portance lies in the charge transfer between the photogenerated electrons and holes and the adsorbed molecules, which occurs at the surface. Another as-pect of its importance is that the surface structure can also affect how mole-cules adsorb, i.e. the binding geometries of the molecules. This is of im-portance because the molecules ability to be oxidized or reduced is greatly

32

affected by the binding strength and adsorption geometry, since it directly affects the molecular energy levels and charge population of intramolecular bonds. For example, it has been reported that the surface structure and expo-sure of different facets affects the bonding of formic acid on anatase nano-particles and thereby influence the photocatalytic reactivity[67].

From an application point of view a good oxidation photocatalyst should have five surface properties, (i) a large surface area allowing for a large number of sites for molecules to adsorb on; (ii) exposure of facets that pro-motes adsorbate structures beneficial for photodecomposition; (iii) appropri-ate band gap energy for inter-band absorption; (iv) high mobility and low trapping rate of the charge carriers and (v) appropriate surface electronic properties of exposed facets, i.e. position of CB and VB energy levels for efficient interfacial charge transfer with respect to adsorbed O2 and OH, or adsorbed molecules[74],[75]. It should be noted that fulfilment of all five crite-ria may not be simultaneously possible on a single crystal facet, since charge transfer and adsorbate bonding may compete[75], and that instead an optimum balance should be found. Alternatively, as described above, communication between surfaces with different surface electronic properties can be consid-ered for optimum performance for a purposeful design of photocatalytic materials.

From a surface science point of view the pre-requisites are different and a surface that is stable and uniform is preferred as it makes the sample prepa-ration easier and interpretation of the results more straightforward. Therefore single crystals are often used as they always expose the same facet unlike nanoparticles or nanoporous films where a multitude of facets can be ex-posed simultaneously, and importantly, where low-coordinated surface at-oms may completely determine the observed reactivity. Of the different ru-tile TiO2 facets the rutile (110) surface is the most stable and easy to prepare as large crystals, and therefore is the most investigated surface[34], although theoretical and experimental investigations of the other low index surface orientations have also been performed[34],[76]–[79]. The three most common rutile (110), (011) and (100) surfaces are described in more detail below. Other low index facets of rutile TiO2 include the (001) and the (111) facets whose properties is only scarcely investigated and thus there is an ambiguity of the exact structure of these facets[34],[76].

3.2.1 The rutile (110) surface The rutile (110) surface, depicted as a ball model in Figure 3.3, has the low-est formation energy of all the different TiO2 surfaces. It is being built up of rows with six-fold coordinated Ti atoms (Ti6c) that are surrounded by oxygen atoms as depicted in Figure 3.1. In every second row along the [001] direc-tion the top-most oxygen is removed and thus exposing the titanium atoms. These Ti atoms are then five-fold coordinated, Ti5c. As a result the surface

33

consists of two different types of titanium atoms. The Ti5c atoms just de-scribed above and the Ti6c atoms below the topmost oxygen rows. These oxygen atoms are two-fold coordinated and are also called bridging oxygen, Obr, atoms. The bridging oxygen atoms are oriented in rows along the [001] direction. Finally there is also tree-fold coordinated oxygen atoms, O3c, on the surface and these oxygen atoms are found around the Ti5c atoms in the trenches.

Figure 3.3. Ball model of the rutile TiO2 (110) surface.

3.2.2 The rutile (011) surface The (011) surface have gained attention during the last decade after reports of this surface facet exhibiting enhanced photocatalytic properties[72],[80],[81], although an ultra-high vacuum, UHV, investigation show less favourable activity of the (011) surface compared to the (110) surface[82]. The (011) surface is not similar to the (110) surface due to the tetragonal unit cell of the rutile TiO2 and the (011) facet is built up of a zig-zag rows of two-fold coor-dinated oxygen atoms, Figure 3.4.

34

Figure 3.4. Ball and stick model of the bulk-terminated rutile TiO2 showing the rutile (011) surface with oxygen atoms (blue) in a zig-zag rows between rows of titanium atoms (red). The figure is adapted from Torrelles et. al.[83].

3.2.3 The rutile (100) surface The (100) surface is analogous to the (110) surface in that respect that it is created by breaking the same number of Ti→O as O→Ti bond in a bulk crystal. This result in a grooved surface, Figure 3.5 consisting of two-fold coordinated oxygen atoms in bridging rows and five-fold coordinated titani-um atoms and three-fold coordinated oxygen atoms in between.

Figure 3.5. Ball and stick model of the rutile TiO2 displaying the (100) surface, consisting of oxygen atoms (white) with two-fold coordination in bridging rows and titanium atoms (grey) with five-fold coordination and three-fold coordinated oxygen atoms in between the bridging oxygen rows. The figure is adapted from Diebold[34].

35

3.3 Defects in rutile TiO2 What has been described so far is a perfect rutile surface. Although the sur-face preparation techniques and methods to characterize them on an atomic level have advanced tremendously, allowing us to manipulate and investi-gate surfaces down to the atom level using tools such as scanning tunnelling microscopes, a completely perfect surface with a relatively large area, in the order of cm2 is still not achievable.

3.3.1 Bulk defects The standard preparation method of the TiO2 single crystal is argon sputter-ing and vacuum annealing. This creates a reduced crystal due to the prefer-ential sputtering of oxygen. The reduction of the TiO2 crystal is accompanied by a change in colour from white to bluish, Figure 3.6. The more reduced the crystal is the more blue it becomes. This substoichiometry creates colour centres associated with bulk defects such as doubly charged oxygen vacan-cies and Ti3+ and Ti4+ interstitials[34].

Figure 3.6. Stoichiometric (left) and reduced substoichiometric (right) rutile (110) single crystal.

3.3.2 Surface defects Sputtering of a TiO2 single crystal with argon ions creates a rough surface with about 2 nm protrusions that contain a significant amount of Ti3+ cations as a result of the preferential sputtering of oxygen atoms[84]. During anneal-ing thermal energy is transferred to the atoms so that they can diffuse in the crystal and throughout the subsequent cooling the surface is reconstructed in such a way that the cohesive energy of the surface atoms is minimized. As a result the protrusions evens out and creates smooth terraces that grows larger with increasing annealing temperature[34],[84]. At the step edges under-coordinated atoms are exposed, typically Ti4c atoms[34]. These under coordi-

36

nated atoms generally have larger chemical activity and thereby the sizes of the terraces can be of importance.

3.3.3 Surface oxygen vacancies On the TiO2 (110) surface the most common defect is an oxygen vacancy in the bridging oxygen rows (Ovac). Depending on the surface preparation methods the number of Ovac sites can be somewhat controlled. Typically the bridging oxygen vacancies are in the range of 5-10% of a monolayer for a reduced surface but on heavily reduced surfaces the number of vacancies can be as much as 15% of a monolayer[34],[76],[85]–[87]. In order to differentiate a reduced surface, with an excess of oxygen vacancies, from a more stoichio-metric surface, the term r-TiO2 is used for the reduced TiO2 surface. Like-wise the term s-TiO2 will here be used for a stoichiometric TiO2 surface, i.e. a surface prepared to obtain a minimum of surface defects. A general con-sensus when working with the TiO2 (110) surface is that cleaning the surface with Argon-ion sputtering and subsequent annealing in UHV will create a reduced surface[34],[76],[85]. To replenish the oxygen vacancies annealing can be done in the presence of a low partial pressure of O2. This will reduced the number of Ovac and result in large defect free domains, of the order of several thousand nm2 [34],[88].

The study of the oxygen vacancies on the rutile (110) surface requires however some precautions. It is widely known that Ovac sites on the TiO2 (110) surface dissociate water molecules and thereby becomes hydroxylated with an OH group filling the bridging oxygen vacancy position (OHbr)

[34],[76]. This process is reported to be swift unless the surface is studied at extreme UHV conditions and special care is taken to reduce the water partial pressure in the vacuum chamber[76],[87],[89].

3.4 Anatase and brookite TiO2 surface structures As mentioned above, single crystal studies of anatase TiO2 is scarce due the limited availability of commercially available single crystals. The studies on anatase made so far have mainly used naturally occurring minerals, with low but not zero amount of contaminants, or surfaces that are grown in house by UHV epitaxy. These studies are limited to the (001) and the (101) anatase surface where the latter is the more stable of two. On nanoparticle sized ana-tase, depending on the particle size and synthesis method, other facets than these two also occur, Figure 3.7.

Experimental surface science studies on brookite single crystals are non-existing and only a few theoretical calculations have been done for the dif-ferent brookite surfaces[90],[91]. These calculations have shown that the for-

37

mation energy for the brookite surfaces is larger than that of the most stable anatase and rutile surfaces.

Figure 3.7. Compilation of different nanoparticle shapes and the most probable crystal facets that are exposed on these nanoparticles. Different anatase crystals (AI, AII and AIII) are shown to the left and rutile (RI and RII) crystals are shown to the right. The figure is adapted from Österlund[67].

3.4.1 The anatase (101) surface The anatase (101) surface is the anatase surface configuration with the low-est formation energy. Calculations have shown that there is a rather big for-mation energy difference between the different anatase surface structures and also that the formation energy is somewhat larger than that for the sta-bile rutile (110) surface[92]. The anatase (101) surface is made up five-fold and six-fold coordinated titanium atoms and two- and three-fold coordinated oxygen atoms, Figure 3.8. From a cross section view the surface appears somewhat corrugated with the O2c atoms positioned at the topmost position creating the ridges on the surface. These O2c atoms are bonded to a Ti6c atom found in the “lowest” valley of the cross section profile and a Ti5c atom. The Ti5c and Ti6c atoms are separated from each other by O3c atoms.

This surface exhibits some similarities to that of the rutile (110) surface. The density of Ti5c atoms is similar on both surfaces and oxygen vacancies on the anatase (101) surface can be created by removing the O2c atoms just as on the rutile (110) surface[92]. This will create Ti3+ ions that are four- and five-fold coordinated on the surface compared to the rutile (110) surface where removal of the O2c atoms creates Ti5c ions. Measurements using pho-toemission techniques have shown that the anatase (101) crystal has a higher oxygen vacancy concentration compared to the rutile (110) surface[93], but virtually no point defects on the surface have been seen by STM investiga-tions of the surface[94]. This conundrum is due to diffusion of surface oxygen vacancies from the surface into the subsurface layers at temperatures above 200K[95]. Thereby the anatase (101) surface displays in principle no oxygen vacancies in the topmost layer, while the majority of the oxygen vacancies

38

are found in the subsurface layers where they still will affect the properties of the surface adsorbates[96].

Figure 3.8. Ball and stick model of the anatase (101) surface with different coordi-nations of the titanium and oxygen atoms. The figure is adapted from Bourikas et. al.[97].

3.4.2 The anatase (001) surface The anatase (001) surface easily reconstructs when heated and only on as-grown samples have the unreconstructed (001) surface been observed[92]. This surface consists of Ti5c atoms bonded to O2c and O3c atoms, Figure 3.9. When heated the (001) surface reconstructs into a 1x4 structure, where peri-odically a TiO2 species is added with one of the O2c surface atoms[92],[98],[99]. This surface construction is also reportedly to have half the formation energy compared to the unreconstructed anatase (001) surface making it the energet-ically preferred surface structure[97].

Figure 3.9. Ball and stick model of the anatase (001) surface with different co-ordinations of the titanium and oxygen atoms. The figure is adapted from Bourikas et. al.[97].

39

3.5 Adsorption on the rutile (110) surface Molecular oxygen does not adsorb on the stoichiometric rutile TiO2 (110) surface at room temperature, only at cryogenic temperatures up to 1.5 ML of O2 can be physisorbed on the surface. For a non-stoichiometric rutile (110), i.e. a surface with Ovac, OHbr or Ti interstitials and thus charged, O2 can chemisorb on the surface. At temperatures above 150 K the chemisorbed oxygen molecules will interact with the Ovac on the surface, heal them and create oxygen adatoms[76],[85]. Thus exposure of the s-TiO2 surface to oxygen at room temperature are expected to dissociate the oxygen molecules and heal the remaining Ovac sites present on the surface, as well as to create oxy-gen adatoms, Oad, on the surface. This oxidized surface is by analogy called o-TiO2 to differentiate it from the other surface preparations[87],[100],[101].

It has been shown that with increasing OH coverage on the surface, the adsorption energy for O2 chemisorption increases[102]–[104] and for a hydrox-ylated surface different behaviours of the OHbr and Ovac have been observed depending on the magnitude of the O2 dose. For low O2 doses, the oxygen molecule interacts with the Ovac and OHbr species and creates Oad atoms via surface intermediates, eq (3.1) - (3.3).

→ (3.1)

OH O → (3.2)

→ (3.3)

If the oxygen dose is large enough to heal all the Ovac the Oad atoms will interact with the OHbr and create terminal OH (OHt) i.e. a hydroxyl on top of a Ti5c atom. These OHt will then in turn react with each other and desorb leaving an Oad atom behind as is illustrated in eq (3.4)-(3.5) [105]–[107]. Investi-gations utilizing large O2 doses, much larger than what is enough to consume all OH on the surface according to the previous reaction scheme, eq (3.4)-(3.5), Oad is still present on the surface. Thus a reaction route involving only one OH for the creation of one Oad atom is suggested[108].

→ (3.4)

2 → (3.5)

These Oad atoms are reactive and stable. Scanning tunnelling microscopy (STM) images have shown Oad atoms on the surface at 300 K and tempera-

40

ture programmed desorption (TPD) measurements suggest that they exist on the surface at temperatures up to 600 K[109],[110].

On the rutile (110) surface water also dissociates on Ovac sites and form two OHbr species. When the surface is heated to a temperature of about 500 K the OHbr groups on the surface starts to recombine and form water and create and Ovac site. From isotope labelled TPD studies it was shown that during this process the hydrogen atom can diffuse to different Obr atoms. There is also a reaction pathway for water adsorption and dissociation on Ti5c atoms. This reaction route requires the presence of an Oad atom on a nearby Ti5c atom. This water dissociation pathway creates two OHt species that eventually recombines into water, desorbs and leaving the Oad atom on the surface[85],[111].

41

4 Vibrational spectroscopy

Raman and particular infrared (IR) spectroscopy is a widely used tool for chemical analysis. There are different specialized IR spectroscopy tech-niques for different applications, but they are all based on the same principle, absorption of light to excite molecular vibrations. In this thesis infrared re-flection-absorption spectroscopy (IRRAS) has been used to study adsorbate geometry on rutile TiO2 and the interactions between the adsorbate and oxy-gen and light.

4.1 Principles of infrared spectroscopy

4.1.1 Fundamental vibrations A molecule consists of atoms, ranging from only a few to several thousands, depending on the size and complexity of the molecule. A single isolated atom has three degrees of freedom since its position needs to be defined in a coordinate system, for instance the Cartesian system with an x-, y- and z-axis. A molecule with N atoms thus has 3N degrees of freedom. Three of these refer to the translational movement of the molecule as a whole along the x-, y- and z-axis. Three additional degrees of freedom are attributed to the rotation of the whole molecule along any of these axes. The remaining 3N-6 degrees of freedom for the molecule then represent the vibrations of the atoms in respect to each other. For a linear molecule there is symmetry around the internuclear axis, which means that there is no moment of inertia in this direction. Hence there is no degree of freedom for rotation in this direction and as a result the number of vibration modes in such a molecule is 3N-5. Furthermore the symmetry in a molecule will cause some of the vibra-tional modes to be degenerate and thus have the same vibrational frequency[112].

The atoms in the molecule are bonded to each other by the outermost (va-lence) electrons in the atom. One separates between ionic, physical (van der Waals) and chemical (covalent) bonds depending on the characteristics of the bond. van der Waals bonds are the weakest of the three and are caused by fluctuations in the electron cloud around the atoms. In ionic bonds, which are the strongest binding type of the three, electrons are transferred to the

42

atom(s) that has the highest electronegativity. Thereby creating a positive and a negative ion that attracts each other by means of electrostatic interac-tions, i.e. there is a permanent dipole in the molecule. For covalent bonds the difference in electronegativity between the atoms is small and it can be viewed as the electrons are shared between the atoms. However as long as there are a difference in electronegativity between the atoms there will be a small displacement of the electrons towards one of the atoms and hence a small permanent dipole can also be formed in covalent bonds. Alternatively, an induced dipole is created as the atoms in the molecule vibrate. The pres-ence of the dipole is important in IR spectroscopy since it is the electromag-netic field in the IR light that excites the movement (or vibration) of the at-oms in the molecule. This can only happen if there is a dipole moment in the molecule. Hence linear diatomic molecules such as N2 are inactive in IR spectroscopy whereas carbon monoxide and non-linear molecules, which has a dipole moment, is easily seen with IR spectroscopy. Similarly, vibrational modes that are asymmetric, and hence will have an induced dipole moment are IR active, while symmetric modes are IR inactive. An example is the linear CO2 molecule, whose symmetrical stretch of CO2 is inactive, while the bending modes (scissoring modes) are IR active.

The quantum mechanical explanation for this is briefly described below. It is referred to as the dipole selection rule and it is based on how light (elec-tromagnetic waves) interacts with molecules. From the selection rule it is possible to derive which vibrations that are allowed and hence IR active[112],[113]. If the vibration of the atoms is modelled as a harmonic oscilla-tor, which is a good approximation for small displacements of the atoms in diatomic molecules, the dipole selection rule can be expressed as

∗ (4.1)

where R

is the vibrational transition moment, i and e is the wave func-tions of the initial and excited state of the molecule,

is the electric dipole

moment operator, and “*” represents the complex conjugate of the wave function. If the integral is zero the transition to the exited state is for-bidden and thus IR inactive whereas it is allowed when the integral is non zero. As the wave functions for the two states of the molecule are eigenfunc-tions of the same Hamiltonian they are therefore orthogonal and thus the integral over them becomes zero

∗ 0 (4.2)

43

If the dipole moment operator is expanded around the equilibrium position, r=0,

(4.3)

it can be seen that for eq. (4.1) to be non-zero two criterions must be ful-filled. First, there must be a change in dipole moment during the vibration, i.e. / 0. Secondly, in the harmonic oscillator model the wave func-tion for the ground state has even parity and the dipole moment operator in turn has odd parity, the wave function for the excited state must also have odd parity for the integral to be non-zero. As a consequence of this the change in quantum number, , between the two states must be ±1. Thus only transition to an energy level just above or below the present level in the har-monic oscillator is allowed. From the dipole selection rule it also follows that the intensity of the vibration is proportional to the square of the vibra-tional transition moment.

4.1.2 Anharmonicity and line broadening So far we have only considered the linear expansion of the dipole moment operator and in this case the dipole moment operator is said to vary harmoni-cally with . If the expression is expanded to include higher order terms, eq (4.4),

12

16

(4.4)

the relationship will become anharmonic. One result from this anharmonici-ty, due to the higher order terms, is that the vibrational selection rule changes to also allow transitions where the vibrational quantum wavenumber changes with ±2, ±3 …±n. The contribution from these higher power terms are usual-ly small and is manifested in the spectrum as vibrational overtones at fre-quencies corresponding to 20, 30 and so on where 0 refers to the vibra-tional frequency of the fundamental vibration, i.e. the vibration when =±1.

The potential well of the harmonic oscillator model predicts that the forc-es felt by the molecule are the same regardless whether the atoms are moved towards each other or away from each other. In reality the forces felt by the molecules should raise steeper when the atoms are moved towards each oth-er as the electric repulsion grows very quick with decreasing distance be-tween the atoms. On the other hand when the atoms are moved away from each other the forces experienced by the molecule starts to trail off and even-

44

tually disappear when the atoms are separated far enough to dissociate the molecule. An empirical determination of such a potential function have been made by Morse[114] and it is described by the following relationship,

1 (4.5)

where De is the depth of the potential well, is a measure of the curvature at the bottom of the well and re is the equilibrium distance between the atoms.

Figure 4.1. The Morse potential (purple line) and the harmonic oscillator potential (blue) and the corresponding energy levels.

In the Figure 4.1 above, the Morse potential and the harmonic potential curve is shown. There the steeper rise of the potential for small separations between the atoms is seen as well as the flattening out of the potential when the separation is large. One of the bigger differences between the Morse potential and the harmonic oscillator potential is that in the harmonic oscilla-tor model the energy separation between the different states are constant, h0 between two adjacent energy states whereas the energy separation decreases with increasing vibration quantum number, in the Morse potential. As a consequence of this the energy for the transition from ground state to the first excited state → is larger than that from the first excited state to the second excited state, → and so on. This will result in new vibration bands and the intensity of these bands depends on the population of the higher levels, which is temperature dependent. If the IR spectroscopic measurements are performed at room temperature one can expect that most of the vibrations are in their ground state and that the transition between

45

excited states are less likely but with increasing temperature transitions from excited states becomes more probable and will thereby influence the position and appearance of the IR vibration band. Different variants of the Morse potential have been developed during the years to better match actual condi-tions. Another aspect to consider is that two different vibrations can be ex-cited by the same photon, ab→ ab which will result in a combination band, whose energy is the sum of the two fundamental vibra-tions.

Line broadening of all vibrational peaks occurs due to the heterogeneity of the molecules. For each of the molecules, no matter whether they are in gas phase, in a solution or adsorbed on a surface, the vibrational transition will not be identical, since every molecule experiences a different environ-ment due to statistical distribution of thermal energies. As a result of this a slight shift of the vibrational frequency for the different molecules occurs and results in line broadening of the vibration bands in the measured spec-trum. These individual contributions from each molecule cannot be resolved and as a result a broader peak with an intensity distribution is measured, Figure 4.2.

Figure 4.2. Line broadening of the infrared vibration band (blue curve) due the indi-vidual shift of the energy of each vibration (red lines) caused by collisions with the surrounding molecules.

46

4.1.3 Vibration modes To discriminate between different vibrational modes, which depend on how the atoms move in respect to each other, terms like stretching and wagging is used. These different vibrations are distinguished in the text by the use of different Greek letters. As formic acid (HCOOH) or formate (HCOO) being the molecule of interest here it will be used as an example, Figure 4.3. Stretching, , is when the atoms move in the plane of the bond and the dis-tance between them is altered. Whether the atoms movement is in phase our out of phase with each other it is called a symmetric (s) or asymmetric (as) stretch, respectively. Wagging () is a change in angle between the plane of a group of atoms, such as the C-H bond and the C2v symmetry axis of the HCOOH molecule. Additional modes, not pertinent for HCOOH, are rock-ing, twisting, in-plane bend, and out-of-plane bending vibrations.

The different vibrational modes are distinguished from each other by the energy needed to excite them. Stronger bonds between the atoms and larger masses of the atoms results in a higher resonance frequency. As every type of molecule is unique they all have a characteristic IR spectrum that can be used for chemical identification. From the IR spectrum it is thus possible to identify a molecule and how it binds to the surface.

Figure 4.3. Different vibration modes of the formate molecule. Oxygen atoms are depicted as blue spheres, carbon as purple and hydrogen as a red sphere. a) symmet-ric a(OCO) stretch, b) asymmetric as(OCO) stretch, and c) in-plane (C-H) wag-ging.

For gaseous species the molecules can also rotate giving rise to rotation bands. The energy difference between these rotational bands is small and hence far infrared radiation is used to probe only the rotational motion. However, the gaseous molecule also rotates when the molecules is in an excited vibrational state. Thus the rotational bands can be superimposed on the vibrational transition, Figure 4.4. For molecules adsorbed on a surface the rotational freedom is lost and therefore these rotational-vibrational bands are not seen. Likewise for molecules in a solution, where collisions between molecules creates a line broadening that makes the rotational fine structure also to be lost.

47

As water and CO2 is always present in the surrounding atmosphere and both of these molecules have a large dipole transition moment, i.e. they are easily seen in the infrared spectra, the presence of these rotational-vibrational bands can pose a problem if not special precautions are consid-ered, for instance by purging the spectrometer with dry air or nitrogen or even better, evacuate the spectrometer to remove all water vapour and car-bon monoxide.

Figure 4.4. Rotational bands superimposed on the water vapour vibrational band centred at ~1600 cm-1.

4.2 Surface infrared spectroscopy

4.2.1 Polarization of light Like electrons, light can either be described as a particle, a phonon, or as an oscillating electromagnetic field, i.e. an electromagnetic wave. The set of equations that describes the relations between electromagnetic fields was formulated by James Maxwell in the 19th century[115]. From these equations it is possible to derive that light is composed of a transverse electromagnetic wave. This means that the electromagnetic fields are oriented in directions that are perpendicular to the propagating direction of the wave, Figure 4.5.

48

Figure 4.5. A transverse electromagnetic wave with the electric (E) and magnetic (B) field oriented perpendicular to the propagation direction of the wave.

Depending on how the orientation of the electric field (E-field) vectors change with time the light is said to have different polarizations. These are called linear, circular and elliptical polarizations. In linearly polarized light the E-field vectors oscillate in the same direction with time, whereas in cir-cular or elliptical polarization the direction of the electric field vectors ro-tates around the azimuth of the propagation direction of the light, like a cork screw. In materials optics, when talking about polarized light, one mainly speaks about two special cases where the E-field of the linearly polarized light is oriented in certain directions with respect to the interface between two media (here vacuum/TiO2 or air/TiO2). These two cases of polarized light are denoted s-polarization (from the German word senkrecht, which means perpendicular) and p-polarization (parallel). Light incident on a sur-face is said to have a plane of incidence defined by the propagating direction of the light and the surface normal. For p-polarized light the E-field is in this plane of incidence and for s-polarized light the E-field is perpendicular to the plane of incidence.

If the surface is oriented in the x-y plane and the light is propagating in the x-z plane as indicated in, Figure 4.6, the plane of incidence is defined to be the x-z plane. For s-polarized light the E-field is then in the y-direction perpendicular to the plane of incidence and thereby tangential to the surface, Et,s. For p-polarized light the electric field, Ep, is thus conversely oriented in the plane of incidence. With the orientation of the sample and light used in Figure 4.6 the E-field for p-polarized light will have two vector components, one in the x-direction tangential to the surface, Et,,p, and one in the z-

49

direction normal to the surface, En,p. Their relative magnitude depends on the angle of incidence, i, of the light. Only at the extreme cases with i equal to 0 or 90° the p-polarized light will only have a component in either the x or z direction, respectively. If there is an orientation of the adsorbed molecules on the surface it is possible to use polarized light to probe certain vibrational modes in the molecule. This stems from the fact that if the dipole moment for the vibrational mode does not have a component in the same direction as the E-field of the IR light, there will be no coupling between the vibration in the molecule and the electromagnetic field and hence the vibration cannot be excited.

Figure 4.6. Directions of the electric field for s- and p-polarized light incident on a surface.

4.2.2 Three layer model Using Maxwell’s equations as a starting point the interaction of light with matter can be calculated[116]. This can be a tedious task depending on the complexity of the system. By making some assumptions about the system the complexity can be reduced. For spectroscopic studies of thin films on a substrate a three-layer model can be employed. The model consists of three different layers; the incident medium the thin surface film layer, and the substrate, denoted with subscripts 1, 2 and 3 respectively. Each of these lay-ers is characterized by their own complex refractive index and is separated by sharp boundaries[117]–[120]. By picturing the adsorbate layer as a homoge-nous film with an effective refractive index, n2, and specific thickness, ht, the

50

reflectivity for s- and p-polarized light (Rs and Rp respectively) can then be described by the following equations[120]

1 4 (4.6)

1 4

(4.7)

where

(4.8)

with being the complex refractive index with n as the real part and as the imaginary part and the subscript index j (j = 1, 2, or 3) denotes the differ-ent layers, a is the absorption coefficient for the adsorbate layer, expressed as 4 ⁄ with as the wavelength of the incident light. These simpli-fied expressions allow one to easily see how different parameters such as the angle of incidence, refractive index of the substrate and adsorbate layer af-fect the reflectivity of the system. Thus the changes of reflectivity, R, when a thin adsorbate layer adsorbs on a surface can be calculated[121]. Here R is defined as R0-Ra, where R0 and Ra is the reflectivity without and with the adsorbate layer, respectively. In Figure 4.7 the calculated R for a thin ad-sorbate layer (h=3 Å) with refractive index, 1.4 0.4 is shown using typical values for the refractive index of stoichiometric (n3=2.4)[122] and re-duced (n3=1.9)[123] rutile TiO2, at a wavelength of 1500 cm-1 using eq. (4.6)-(4.8). From these results one can clearly see that the largest |Δ | occurs at different angles of incidence for s- and p-polarized light. The optimum angle also depends on the refractive indexes of the system, Figure 4.7. The change in reflectivity for p-polarized light can either be positive, negative or in some cases also zero, whereas for s-polarized light the change in reflectivity is always negative. These basic considerations are crucial when interpreting an infrared reflection-absorption spectroscopy (IRRAS) spectra, in particular if the refractive index of the substrate is expected to change, e.g. as a function of sample preparation; sputtering and vacuum annealing results in a reduced

51

surface, while oxygen or ozone pre-treatments render it more oxidized, which will affect the refractive index.

Figure 4.7. Difference in reflectivity (R = R0 - Ra) for a surface without (R0) and with (Ra) adsorbate layer as a function of angle of incidence (i) for s- and p-polarized light (red and blue lines, respectively). In the plot two different values of refractive index for the substrate, n3, is shown. Solid line is for n3=2.4 and dashed is for n3 = 1.9. The adsorbate layer thickness, ht, is set to 3 Å with a refractive index

1.4 0.4 . The results are calculated using a wavenumber, = 1500 cm-1 of the IR light.

4.2.3 Simulation of infrared-reflection absorption spectrum The description of the adsorbate layer as a homogenous layer with a refrac-tive index is in some cases, like studies of adsorbates on a surface, not satis-factory. Thus the model can be rewritten so that the orientation of the ad-sorbate layer is accounted for by the use of the polarizability of the adsorbate layer[124],[125]. Thereby the change in reflectivity with a thin layer can be ex-pressed as

∆ 4

Im (4.9)

where is the frequency given by 2 with as the wavenumber, and c the speed of light. F(i) is a function that depends on the incidence angle, the polarization of the light and the dielectric response of the surface, NS is the number of adsorbate molecules per surface unit cell and Im(p is the imaginary part of the polarizability of the adsorbate layer. For thin layers of adsorbed molecules the molecules are going to be oriented on the surface. As

52

a consequence of this the polarizability of the adsorbate is anisotropic but has its principal axis along the x-, y- and z-axis. The dielectric tensors of the thin layer is therefore diagonal[124]. For a semiconducting substrate the ex-pression for F(i) is different depending on the incident polarization of the light. With light incident in the x-y plane along the x-direction the s-polarized light will be oriented in the y-direction and the normal and tangen-tial component of the p-polarized will be in the z- and x direction respective-ly, see Figure 4.6. Thus R/R0, in eq. (4.9) becomes;

Δ

,

24 cos

4 Im , (4.10)

Δ

,

24 cos 1

14 Im ,

(4.11)

Δ

,

24 cos

1

4 Im ,

,

(4.12)

Here Et,s is the electric field for s-polarized light, Et,p the tangential and En,p the normal component of the E-field for p-polarized light. The relative die-lectric function for the incident vacuum medium and the substrate are denot-ed v and s respectively, a,z is the z-component of the adsorbate dielectric function and p(i,j) is the polarizability of the adsorbate layer in the different directions, denoted by either (t,s), (t,p) or (n,p) for s-polarized light in the tangential direction, p-polarized light in the tangential direction or p-polarized light in the normal direction. The complex dielectric function ̂ is related to the complex refractive index according to ̂ . The polarizabil-ity and the dielectric function are material properties that governs how the material, or in this case the molecule, is affected by external electromagnetic fields.

Modelling the adsorbate molecule as a Lorentz oscillator allows one to al-so determine the polarizability and the dielectric function of the adsorbate layer, a

[116],[124]. The polarizability for a Lorentz oscillator can be described as a summation of the electronic (e) and vibrational (v) part of the polar-izability for the different modes

53

, ,

∗

∗

Γ (4.13)

With e* and m* being the effective charge and mass, 0 the resonant fre-quency and the line width. The dielectric function is related to the polar-izability and is given by

14

14

(4.14)

Here d is given by d=4NS/CNS3/2 with C as a constant whose value depends

on the arrangement of the dipoles on the surface[124]. Since the imaginary part of eq. (4.13) always will be positive, R/R0 for s-

polarized light, eq. (4.10), will always be negative, and as a result the vibra-tional bands probed with s-polarized light will show up as emission peaks, i.e. increased reflectivity compared to the reference surface without adsorb-ate layer, similar to the results presented in Figure 4.7. Similarly the tangen-tial and normal component of R/R0 , eq. (4.11) and eq. (4.12), for p-polarized light will be negative and positive, respectively, for grazing inci-dence spectroscopy. Thus the same vibrational mode can either appear as an emission or absorption peak, i.e. increased or decreased reflectivity com-pared to the reference surface without adsorbate layer, depending on the polarization of the light and how the molecule is oriented on the surface. A small frequency shift between the tangential and normal component, equa-tion (4.11) and (4.12) respectively, for the same vibrational mode can also occur. This is due to the frequency dependence of the a,z term in eq. (4.12).

By the use of density functional theory (DFT) it is possible to calculate the polarizability and dielectric function of the adsorbate layer[126]–[129]. It is thus possible to simulate IRRAS spectrum of a thin adsorbate layer and compare with experimental data[130], as shown in Figure 4.8 since the orien-tation of the adsorbate molecules are known in the DFT calculations infor-mation about adsorbate geometry on the surface can therefore be obtained.

54

Figure 4.8. Comparison of experimental (left) and simulated (right) IRRAS spectra for formate on TiO2 with p-polarized light. Adapted from Mattsson el. al.[130].

4.3 The FT-IR spectrophotometer That IR light existed was unknown until the year 1800 when it was discov-ered by Sir William Herschel. Herschel had constructed a rudimentary mon-ochromator in order to investigate the heating power of the different colours in the sunlight. For this he used a thermometer with a blackened bulb as a sensor. The thermometer was placed in the different colour regions of the light with two other thermometers placed in the dark as controls. When he out of curiosity placed the thermometer outside of the red light region he could still observed a temperature rise and could thus conclude that the sun-light also contained light that was not visible to the human eye. During the following years Herschel performed experiments on IR light and discovered that it was reflected and transmitted in the same way as visible light[131]. Af-ter this discovery it would take almost a century before the infrared light was utilized for spectroscopic measurements. In 1881 the first near infrared spec-trum was recorded by Abney and Festing with the use of photographic plates as an IR detector[132]. About two decades later William Coblentz further de-veloped the IR spectrometer. Up till then a handful of IR spectra had been published with wavelength ranges up to 5 µm. Coblentz improved the design of the spectrometer and was able to obtain spectra using wavelengths up to 15 µm. He also published the first collection of IR spectra, consisting of more than 100 pure compounds[133]. During these early years of IR spectros-copy, the instruments used were dispersive spectrometers with either prisms or gratings to disperse the light. Even though the advantages of using a Mi-chelson interferometer instead of a prism or grating was known the technical limitations for implementing such a solution were at this time to large. With the rapid development of electronics these hurdles would eventually be

55

overcame and the first commercially available spectrometer based on the Michelson interferometer became available during the 1960s. After this it would not take many years before the Fourier Transform IR (FT-IR) spec-trometer design outperformed the dispersive IR spectrometers and nowadays in principle all IR spectrophotometers that are sold are FT-IR spectrometers. Compared to the traditional dispersive instruments, FT-IR instruments have three main advantages commonly named after their inventors

Connes wavelength accuracy advantage Fellgets multiplex advantage Jacquinots throughput advantage

and they will be discussed more in detail below. Another key component of the IR spectrometer is the IR detector. In the early days when dispersive instruments were used thermocouples were used as an IR detector. With the discovery of the ternary HgCdTe (MCT) compound in 1959 a new type of detector was born[134]. In the HgCdTe alloy the bandgap can be varied by composition changes and thus allow for a wide variety of detector configura-tions. This type of detector also has a shorter response time compared to the thermocouple detectors which made it more suitable to use in IR spectrome-ters equipped with a Michelson interferometer. For today’s spectrometer the vital parts, a Michelson interferometer for light modulation, a MCT light detector and a SiC globar for light generation are the same as those in the 1960, albeit the performance of the instruments have improved significantly. The most recent development in infrared spectroscopy is related to methods to improve the lateral resolution. With the use of microscopes connected to an IR spectrometer, a lateral resolution in the µm range can be obtained. More recently incorporation of IR spectroscopy with AFM instruments has provided lateral resolutions down to 10 nm. There are two different tech-niques to facilitate this, either by using the near field of the light i.e. scan-ning near field optical microscopy, SNOM, where the broadband light is focused on the tip of a metal coated AFM probe. This will create an optical near field between the probe and the sample and the elastically scattered light then carries information about the sample. The other method is based on absorption resonance of the IR light where an AFM tip is used to detect the thermal expansion of the material when it is illuminated with pulsed IR light. The frequency of the incoming IR light is then scanned in order to obtain spectra[135]. Both these techniques use tuneable IR lasers as light sources instead of the broadband radiation source emitted from a heated silicon carbide globar lamp found in conventional IR spectrometers.

The heart of the modern FT-IR spectrometers is the Michelson interfer-ometer, which as mentioned previously is used to modulate the light. The principle of the Michelson interferometer is that the incoming light from a broadband light source is divided into a reflected and transmitted beam using

56

a beam splitter, Figure 4.9. The two beams are reflected back towards the beam splitter by a fixed and moving mirror, respectively. For every position of the moving mirror, interference, either constructive or destructive, occurs for a specific wavelength. By sampling the signal in the detector at different positions of the moving mirror an interferogram is constructed. This inter-ferogram can then be Fourier transformed into an infrared spectrum[136]. As the spectral resolution, , is inversely proportional to the distance the mir-ror moves, x, it is important to keep track of the mirror movements because any error in mirror position when sampling is done will affect the spectral resolution. For this a He-Ne laser is used. If the laser light is also modulated by the interferometer destructive interference will occur for certain positions of the moving mirror. By sampling the detector signal exactly when the sig-nal from the laser is zero it is possible to accurately determine the distance the mirror moves and thus create an interferogram with well-defined sam-pling intervals. As the wavelength of the He-Ne is precisely known and sta-ble the FT-IR spectrometers is said to have a built in wavenumber calibra-tion of high accuracy, in the order of 0.01 cm-1. This benefit of the FT-IR spectrometers is generally referred to as the Connes advantage[137].

The use of the interferometer in the FT-IR instrument also allows for simultaneous measurement of all wavelengths. This has two implications, which both will improve the signal to noise ratio (SNR). As all wavelengths is measured simultaneously there will be higher energies impinging on the detector (higher photon flux) that will enhance the SNR as the readout noise will consist of a smaller part of the signal, compared to a dispersive instru-ment where a lower photon flux is recorded due to the use of narrow slits before and after the monochromator. This is commonly called the Jacquinot advantage[137]. Furthermore the SNR will also be improved as the square root of the number of sampling points compared to a dispersive instrument, when the signal noise is dominated by the noise from the detector. This comes from the fact that for FT-IR instruments all wavelengths are measured simul-taneously and thus the detector noise averages out with increasing sampling points. The number of sampling points is given by (1-2)/, where and 2 is the start and end wavenumber respectively and the resolution. This benefit of the FT-IR instrument is normally referred to as the Fellgett ad-vantage[137]. Another benefit with the FT-IR spectrometer is that the resolu-tion is constant throughout the measurements as the aperture is fixed during the scan, whereas in a dispersive instrument the slit width changes with wavelength to maintain a similar input power on the detector. Finally the measurement time for FT-IR instrument is governed by the time it takes to move the mirror a distance, x, corresponding to the desired resolution. This can be done quite fast in modern instruments and thus numerous repetitive scans can be made and averaged within a reasonable timeframe. This will further improve the signal quality as the noise is reduced by the square root of the number of scans. Everything is however not positive with the use of a

57

FT-IR instrument compared to a dispersive instrument. The main drawback of the FT-IR is that it is a single beam instrument. Thus the signal is not referred to a reference during the measurements and therefore the stability of the instrument and the measurement environment is crucial for highly sensi-tive measurements or measurements running for a long time. This can to most extent be dealt with by purging the instrument with N2 or even better evacuate the instrument and performed the most sensitive measurements in a climate controlled environment.

Figure 4.9. Schematic picture of a Michelson interferometer.

4.4 Raman spectroscopy Another type of vibrational spectroscopy is Raman spectroscopy. Here the polarizability of the molecule is used instead of the dipole moment as in IR spectroscopy. If the molecule is placed in an electric field it will have an induced dipole moment, µind, since the electrons and nuclei in the molecules are attracted to the positive respectively the negative pole of the electric field. The magnitude of the induced dipole is proportional to the strength of the electric field, with a proportional constant, p that describes how easily the field is induced in the molecule and pis therefore called the polarizabil-ity of the molecule. All atoms and molecules will have non-zero polarizabil-ity. The polarizability is not constant but varies as the molecule oscillates. Just as previously done for infrared spectroscopy, eq (4.1) - (4.3), the Raman vibrational transition moment can be calculated. This gives, just as for the infrared vibrational transitional case, that the vibrational quantum number must change by ±1, if the polarizability is expanded as a Taylor series to the linear term in order for the transition to be allowed.

58

In Raman spectroscopy the oscillating electric field is created by illumi-nating the sample using light with a fixed wavelength. This will trigger the molecule to a virtual quantum state before it returns back to its original state and the incoming photon is elastically scattered. On some occasions the pho-ton is inelastically scattered, or Raman scattered, which means that when the molecule relaxes from its virtual state it does not go back to its original state, and thus the energy of the scattered photon has changed. Which vibrational transition states the molecule can end up in is governed by selection rules described earlier. If the initial state is the ground state of the molecule and the final vibrational state is the first excited state, i→ f i.e. the inelastically scattered photon have lost energy and the process is denoted Stokes Raman scattering. For the opposite case where the final state has lower energy than the initial state, i→ fi.e. the scattered photon has gained energy and the process is called anti-Stokes Raman scat-tering.

Macroscopically the creation of the induced dipole moment can be seen as an interaction between the incident light with frequency whose electric field, varies as costAs the polarizability varies when the mol-ecule oscillates the polarizability can be described by a permanent and oscil-lating part, eq. (4.15).

, Δ cos 2 (4.15)

where p,0 is the equilibrium polarizability and 0 is the vibration frequency of the molecule. Since the induced dipole is given by µind=p one gets

μ , ∆ cos 2 cos 2 + (4.16)

Using trigonometrical relations equation (4.16) can be rewritten so that

μ , cos 2

12∆ cos 2 cos 2

(4.17)

From this, eq (4.17), one can see that the induced dipole will emit radiation with three different frequencies , +0 and -0. This corresponds to elastically Rayleigh scattered light, inelastically scattered light (anti-Stokes) and (Stokes) scattered light respectively as described above.

59

5 Single crystal sample preparation and characterisation

To fully benefit from the use of a single crystal it is necessary that the sur-face is well defined. Therefore the surface preparation and measurements are carried out in an UHV chamber in order to prevent adsorption of pollutants on the surface.

5.1 Sample preparation In this thesis, sample preparation is made by argon ion sputter cleaning of the 20 mm by 20 mm big rutile TiO2 (110) single crystal (PI-KEM Ltd, Tamsworth, UK). A differentially pumped sputter gun (OCI Vacuum Micro-engineering, London, Canada) mounted on one flange of the UHV chamber was used for surface cleaning. The working principle of the sputter gun is that electrons are emitted from a thermally heated tungsten-rhenium fila-ment. The emitted electrons are accelerated in an electro-magnetic field and by impactation the argon atoms in the sputter gun will be ionized. These argon ions are subsequently accelerated in the gun with a user selected volt-age, and finally emitted through an aperture into the main UHV chamber where they impinge on the grounded sample. Argon ions with an energy of 1 keV and a sputter current of 1 µA on the TiO2 single crystal was employed for 10 minutes at 323 K with the argon ion beam having an angle of inci-dence of 30° with respect to the sample surface. Depending on the desired surface properties (r-TiO2, s-TiO2 or o-TiO2) different preparation schemes were used. For the reduced surface, annealing at 973 K in vacuum for 20 minutes was employed. For the stoichiometric surface the annealing was done for 10 minutes at 1×10-6 mbar O2 (99.998% purity), and subsequently 10 min in UHV. The sputter-annealing cycles were repeated four times (ex-cept for a new crystal where several more cycles were used). For the oxi-dized surface the same pre-treatment as for the s-TiO2 was used except that after the sputtering-annealing cycles the sample was cooled down to 140 K and 2 Langmuir (L) of O2 was dosed via backfilling of the chamber. Heating of the sample was done by the on-board resistive heater and cooling was done with copper braids connected to the sample holder and a liquid nitrogen

60

(LN2) reservoir. To minimise contaminations of the sample the experiments were made in an UHV chamber with a base pressure of 2×10-10 mbar.

5.2 Low energy electron diffraction Low energy electron diffraction, LEED, is a technique that allows for studies of the surface structure of a well-defined system, such as single crystals and adsorbates orientated on the surface[138]. The fact that the electron is a charged particle means that it interacts strongly with matter. The low energy of the incoming electrons, up to a few hundred eV, implies that the electron cannot penetrate deep into the sample only in the orders of a few Ångström. As a result the structure of the topmost atoms, i.e. the surface is probed. As the name of the technique tells, LEED is based on the diffraction of elec-trons. The wave-particle duality of the electron gives it properties of both a particle and a wave, meaning it has the possibility to create interference ef-fects. When probing the surface the incoming electrons can either be inelas-tically or elastically scattered in the forward and backward direction. For LEED measurements we are only interested in the elastically backscattered electrons, which compromise about a few percent of the electrons in the incoming beam[138]. In order for diffraction to occur the path difference for the waves must be a multiple of the wavelength. By using the notation of wave-vectors for the incoming, , and diffracted, , electrons the require-ment for diffraction to occur can then be expressed as

(5.1)

∗ ∗ ∗ (5.2)

(5.3)

where is the reciprocal lattice vector. The magnitude of the incoming and diffracted wave is the same, eq. (5.3), since we are working with elas-tically scattered electrons. Given that the electron only interacts with the topmost atomic layers the Laue condition for diffraction expressed in eq. (5.1) is reduced to a two-dimensional case where represents a reciprocal lattice vector in the surface plane. By choosing a set of primitive vectors so that ∗ and ∗ are the primitive reciprocal lattice vectors in the surface plane and ∗ is perpendicular to the surface, i.e. the surface normal, , one can relate the reciprocal lattice vectors to the real space lattice vectors;

61

∗ 2∙

(5.4)

∗ 2⋅

(5.5)

Thus the diffraction spots in the LEED pattern can thereby be related to the surface structure in real space, Figure 5.1.

Figure 5.1. Lattice surface structure in real space (left) and the corresponding recip-rocal structure (right), which also corresponds to the observed LEED diffraction pattern.

A standard LEED equipment (BDL800, OCI Vacuum Microengineering, London, Canada) was used to check the results of the surface preparation before each measurement. An ordered rutile TiO2 surface with little defects results in sharp spots in the diffraction pattern organized in a 1x1 structure. A typically obtained LEED pattern is shown in Figure 5.2.

62

Figure 5.2. Typically observed LEED pattern with a 100 eV primary electron beam.

5.3 Formic acid dosing Exposure of the sample to formic acid was made with a directed gas doser. The gas doser consisted of a 1 µm pinhole mounted on a linear translator arm. Before dosing the arm is positioned in front of the sample and then made into contact with a reservoir of liquid formic acid. Prior to dosing the formic acid is subjected to several freeze thaw cycles using liquid nitrogen to remove impurities. After dosing the linear translator arm is evacuated and retracted. By controlling the dosage time, 60 s, saturation coverage of for-mate can be obtained on the rutile (110) surface.

5.4 Infrared reflection-absorption spectroscopy setup A vacuum pumped FT-IR spectrometer (Bruker IFS 66v/S) was used for the IRRAS measurements. The spectrometer is connected to the UHV chamber via flexible metal tube, which in turn is connected to a KBr window mount-ed on the UHV chamber. Thereby the entire beam path, from source to the UHV chamber, is evacuated down to approximately 2 mbar using the spec-trometer pump. As a result of the spectrometer being evacuated contributions from water vapour and carbon dioxide is non-existing in the obtained spec-tra. A parabolic mirror (focal length 250 mm) is used to focus the IR beam through the KBr window and on the sample. A wire grid polarizer (KRS-5, Specac) is used to polarize the light before entering the UHV chamber. The orientation of the wire grid polarizer can be adjusted manually to obtain the

63

desired polarization. The sample surface is probed at grazing incidence angle (i=85°) and the beam is specular reflected towards an elliptical mirror (focal lengths 203 and 34 mm), which directs the light on a LN2 cooled MCT de-tector. The mirror and detector is mounted in a compartment placed outside the UHV chamber. The compartment is separated from the UHV chamber with a KBr window and it is evacuated down to 8 mbar. A schematic picture of the beam path through the UHV chamber can be seen in Figure 5.3.

Figure 5.3. The IR beam from the interferometer is directed on the sample, through a polarizer with a parabolic mirror. An elliptical mirror is used to focus the reflected IR light on the detector after the sample.

5.5 UV illumination UV irradiation of the samples was done using the 365 nm line from a 200 W Hg(Xe) lamp (Oriel Instruments) in combination with a narrow band pass filter (Oriel Instruments, FWHM = 10 nm). A 54 mm long water filter was used to remove the IR part of the radiation to avoid excess heating of the sample. The light was directed on the sample via a plano-convex quartz lens (f = 200 mm) through a quartz viewport mounted on one flange of the UHV chamber and thus reaching the sample at an incident angle of 40°. The dis-tance between the lens and the sample was 450 mm with a distance of 175 mm between the quartz window and the sample. The incident light flux, was measured in the focal plane of the lens with a thermophile detector (Ophir). From the measured photon power an upper limit of the illumination power incident on the single crystal was estimated to be between 0.2 – 2 mW cm-2.

65

6 Thin film sample preparation and characterisation

6.1 Magnetron sputtering deposition Physical vapour deposition is a common name for deposition methods where a solid material, the target, is transformed into the gas phase before it is de-posited on the surface. The phase transformation can be achieved by among other techniques, resistive heating, laser pulsed illumination or bombardment of energetic particles, also known as sputtering, of the target. In sputtering these energetic particles are created by applying a large electric field be-tween a cathode and an anode in the presence of an inert gas, typically ar-gon. This will create argon plasma and the argon ions are then accelerated towards the target, which act as the cathode. The target is surrounded by a metallic ring that acts both as an anode and a shield to prevent sputtering of the target holder. A set of permanent magnets, behind the target are used to create a magnetic field to increase the confinement of the plasma in the vi-cinity of the target and thereby also increase the sputter rate, Figure 6.1. For standard direct current magnetron sputtering, the target needs to be conduc-tive, and is therefore often made of metal, either pure or a compound of sev-eral different metals. Introduction of reactive gases, such as oxygen or nitro-gen, together with the argon gas is done in order to obtain oxides or nitrides of the sputtered metals. Sputtering in general performs rather compact films but the morphology of the films can be somewhat tweaked by altering the process parameters. Higher sputter rates and deposition pressures results in more porous films, whereas denser films, on the other hand, are obtained by employing a lower pressure in the sputtering chamber. Heating of the sub-strate during deposition will allow for fabrication of different crystalline phases of the material.

In this work the equipment used consists of a deposition chamber with four different magnetrons, placed at 13 cm distance from the substrate at an angle of 50° from the surface normal, Figure 6.1. Deposition of TiO2 films, with a sputter rate of 5 nm/min, were made with a constant sputter current of 0.75A from a titanium target at a working pressure of 4×10-2 mbar with a flow rate of 60 ml min-1 of argon (99.9997% purity) and 2.5 ml min-1 of ox-ygen (99.998% purity). This will result in amorphous TiO2 films that are

66

transformed into crystalline rutile films by heating in air at 650°C for one hour.

Figure 6.1. Schematic set-up of the sputter deposition chamber and the magnetron with the generation of the argon plasma.

6.2 Raman characterisation As previously described Raman spectroscopy is a vibrational spectroscopy technique that utilizes the polarizability of the molecules to induce Raman scattering of the incident light. The probability is low for a photon to be ine-lastically scattered, about 1 out of 106 photons are Raman scattered, and hence lasers are used to illuminate the sample as they provide a large amount of photons with the same wavelength. After illumination of the sample the elastically scattered light is filtered out before it is passed through a mono-chromator and finally detected, Figure 6.2. It is important to employ a filter to remove the elastically scattered light, as it otherwise will saturate the de-tector and make it impossible to detect the much weaker signal from the inelastically scattered light. By either moving the monochromator or dis-perse the inelastically scattered light on a matrix detector a Raman spectrum with detected intensity versus energy shift of the light is obtained. The peaks in the spectra are related to different molecular vibrations and can thus be used for identification of crystalline phase and composition of the sample.

Here a confocal Renishaw inVia Raman microscope was used with lasers emitting light with a wavelength of 514 nm and 534 nm was used. The light was focused on the sample using a x100 objective and neutral density filters was employed to prevent sample damage. The Raman scatted light was de-tected with a matrix CCD sensor.

67

Figure 6.2. Schematic set-up of a Raman microscope. The figure is adapted from Begum et. al.[139].

6.3 Grazing incidence X-ray diffraction X-ray diffraction is a common tool for determining the structure and compo-sition of samples and is based on Bragg reflection from different crystal planes. Parallel beams of X-rays incident on a sample are elastically scat-tered by the atoms in the sample. As the wavelength of the incoming light have the same magnitude as the distance between the lattice planes in the crystal, the scattered X-rays will constructively interfere and thus form a so called diffracted beam. For X-ray diffraction of thin films a grazing inci-dence of the light is used. Typical angles close to the critical angle for total reflection is employed. This gives some benefits compared to normal X-ray diffraction, namely that by employing grazing incidence the X-ray beam is evanescent and only penetrates the topmost part of the film, ~100 Å, making it a surface sensitive technique and thus decreases contributions from the substrate. The grazing incidence also provides an enhancement of the inten-sities at the surface[140]. The conditions for diffraction is given by Braggs law,

2 (6.1)

where j is an integer, dhkl is the interplanar distance for the {hkl} plane, 2d. the angle between the scattered beam and the surface tangent when grazing incidence of the X-rays are used.

By recording the scattered intensity as a function of 2d , a so called dif-fractogram is recorded. Different materials and crystal structures have

68

unique diffraction patterns and thus identification of the phase and composi-tion of the material can be done by comparing with a database of measured and calculated reference diffractograms.

6.4 Scanning electron microscopy Information about morphology and topography on the nanometre level is routinely obtained and visualized using scanning electron microscopy, SEM. In the SEM electrons that imping on the surface, with energies in the range of a few keV, are used as the probe. There they interact with the sample and generate a variety of signals that carries different information, Figure 6.3. For topographic information, secondary electrons that are generated by ine-lastic scattering of the electrons in the primary beam are collected. By col-lecting only low energy secondary electrons, information about the surface topography is obtained as these low energy secondary electrons are generat-ed within a few nm from the surface. The amount of secondary electrons that is detected depends on the incident angle of the primary beam, with edges and steps generating higher recorded intensities. Sweeping the primary elec-tron beam over an area, a two-dimensional topographic picture can thus be obtained. Other information from the sample can be obtained by collecting other types of signals emanating from the sample. Compositional contrast can be provided by backscattered electrons and composition mapping of the sample can be obtained from characteristic X-rays that are emitted due to the interaction between the sample and the primary electron beam.

Figure 6.3. Schematic drawing of the volume of interaction between the sample and the primary electron beam. The different signals emanating from the sample and the typical depth for their origin are also shown.

69

Figure 6.4. Scanning electron microscopy picture, showing the topography of a rutile TiO2 film made by sputtering and subsequent annealing.

6.5 In-situ IR transmission measurements Formic acid adsorption and photocatalytic decomposition experiments at atmospheric pressures were made using an in-situ reaction transmission cell, Figure 6.5, mounted in a vacuum pumped FTIR spectrometer[39],[141],[142]. The reaction cell allows for simultaneous IR measurements and UV illumination in a controlled environment where both the sample temperature and gas composition could be controlled. A set of mass flow controllers was used to regulate the synthetic air flow (20% O2 with 99.998% purity and 80% N2 with 99.98% purity) through the sample cell. Typically a flow rate of 100 ml min-1 was used. Formic acid dosing was done by letting the synthetic air pass over a reservoir with liquid formic acid (98% purity) held at a constant tem-perature of 303 K. This resulted in a steady state formic acid concentration of 420 ppm in the synthetic air flow. A series of valves was used to either direct the synthetic air flow over the formic acid reservoir or to bypass it. In a typical experiment 15 minutes of formic acid dosing was employed, fol-lowed by 10 minutes dwell time, in order to remove any gaseous traces of formic acid from the sample cell, before illumination commenced. Sample illumination, which was provided by a Xenon lamp equipped with a water filter and a set of AM1.5 filter, both from Oriel, to mimic the solar radiation. A fused silica optical fibre bundle was used to direct the light on the sample

70

and the illumination power was 165 mW cm-2 as measured with a thermopile detector (Ophir). IR spectra was obtained every minute and consisted of a collection of 135 scans, obtained during a 30 second long acquisition time. Prior to the measurements the sample was cleaned at 400°C in the sample cell under a 100 ml min-1 flow of synthetic air for 15 minutes.

Figure 6.5. Picture of the in-situ IR transmission cell.

71

7 Results

In this chapter selected results from paper I-V, which are included in this thesis, is presented and discussed. First CO adsorption on the rutile (110) surface is examined. In the two following sections the interaction between formic acid, oxygen and the rutile single crystal TiO2 surfaces is investigat-ed. This is followed by two sections on adsorption of formic acid on disor-dered single crystals and nanocrystalline surfaces. The chapter is then con-cluded with two sections about the interaction of formate adsorbed on differ-ent TiO2 surfaces and light.

7.1 CO adsorption on the rutile (110) surface CO is a widely used probe molecule in surface science as it provides vital information about the gas-surface interactions and the adsorption sites on the surface. In heterogeneous catalysis oxidation of CO on the rutile TiO2 (110) surface have become a benchmark system used to study the oxidation pro-cesses of molecules at the surface. In order to understand the CO oxidation reaction it is necessary to have knowledge about the adsorption of CO on the surface. This has previously been studied both experimentally and theoreti-cally[143]–[147]. There is a general consensus that the CO molecule weakly adsorbs with its carbon atom to a fivefold coordinated titanium atom on the stoichiometric surface at coverages below one monolayer (ML). The adsorp-tion on the surface cause an upshift of the CO stretching vibrational frequen-cy compared to the gas phase. For coverages at one 1 ML and above there are different proposals as to whether the first monolayer adsorbs with the CO molecular axis oriented parallel to or tilted away from the surface normal in either the [110] or [110] direction in a zig-zag pattern[144]–[146]. Simulation of the IRRAS spectra for different CO adsorption geometries was done using a three layer model and DFT calculations as described previous in section 4.2.3. These results was then compared to experimentally obtained IRRAS spectra different CO coverages by Petrik and Kimmel[145], in order to unravel the adsorption geometry of CO on the rutile (110) surface.

72

Figure 7.1. Influence of the CO adsorption geometry on the simulated IRRAS spec-tra for 1ML coverage. The left column (a) shows the case for a tilted absorption geometry of the CO molecules in a side view and a top view and the right column (b) shows an upright adsorption geometry. Simulated IRRAS spectra for s- and p-polarized light incident in the [110] and [001] direction for (c) the tilted CO adsorp-tion geometry and (d) the upright adsorption geometry is also shown. The figure is adapted from paper III[148].

At precisely 1 ML coverage the calculations show that the adsorption energy is slightly more favourable, ~20meV, for a tilted adsorption geometry com-pared to an upright orientation of the adsorbed CO molecules. The simulated IRRAS spectra display small differences between the two cases. With p-polarized light there is in principle no difference in the simulated IRRAS spectra between the two absorption geometries, only a shift of the peaks with 1 cm-1, Figure 7.1. For both the tilted and the upright adsorption geometry the normal component of the electric field of the incoming p-polarized light couples to the molecular vibration. This is not the case for s-polarized where the electric field of the incoming light is parallel to the surface. For the up-right adsorption geometry the electric field of the incident light is perpendic-

73

ular to the molecular axis regardless of incident direction and does therefore not couple to the dipole moment of the CO vibrational stretch. As result no vibration bands appear in the simulated IRRAS spectra. When the adsorbate is tilted the dipole moment will have a component that is parallel to the sur-face in the [110] direction and therefore it will couple to s-polarized light incident in the [001] direction. Consequently there is a small absorption band at 2149 cm-1 in the simulated spectra. The signal intensities for the s-polarized light are small and can therefore be difficult to experimentally detect but examination of the IRRAS spectra measured by Petrik and Kim-mel[145] with 1ML coverage shows a weak absorption band for s-polarized light incident along the [001] direction, suggesting a tilted adsorption ge-ometry for CO.

Figure 7.2. Simulated IRRAS spectra for s- and p-polarized light incident in the [110] and [001] direction for 1.5 ML of CO. (a) Side and top view of the adsorption geometry. (b) Simulated IRRAS spectra for s- and p-polarized light incident in the [110] and [001] direction. (c) Decomposition of the spectra to show the contribu-tions from the first and second ML as well as the normal (n,p) and tangential (t,p) components of the p-polarized light. The figure is adapted from paper III[148].

With an increased coverage of CO to 1.5 ML the calculations show that the configuration with the lowest energy is that of a tilted monolayer, as de-scribed previously, below half a monolayer of CO molecules that is situated above the bridging oxygen atoms. In the second layer the CO molecules are oriented such that the molecular axis is oriented in the 110 direction with an angle of 11° to the surface plane, Figure 7.2. This gives rise to a new set of vibration bands in the simulated spectra. From the simulation, deconvolu-tion of the contributions from the first monolayer and the topmost layer to the total spectra can be can calculated, and is also shown in Figure 7.2

Comparison between the simulated spectra and the experimentally ob-tained spectra[145] is shown in Figure 7.3. A good agreement between the

74

experimentally measured and simulated spectra is shown. The main differ-ence is that the simulations propose a tilted structure for the first monolayer and hence a small vibration band next to the large band due to the topmost layer of adsorbed CO molecules for the s-polarized IRRAS spectra in the [001] direction. In the experimental spectra this band is not marked out but can be discerned.

Figure 7.3. (a) Simulated IRRAS spectra for 1.5ML CO adsorption on TiO2 (110) and the suggested adsorption structure. The contributions from the different mole-cules to the IRRAS spectra are showed. (b) Experimental IRRAS spectra by Petrik and Kimmel[145] and their proposed adsorption structure, with the blue atoms indicat-ing adsorption of CO for coverages larger than 1 ML. The figure is adapted from paper III[148].

7.2 Formic acid adsorption on the rutile (110) surface Exposure of formic acid to a rutile TiO2 (110) surface leads to dissociation of the acid on the rutile surface into a formate ion (HCOO-) and a proton (H+). The proton binds to a bridging oxygen, Obr, and forms a bridging hy-droxyl OHbr species whereas the formate ion can bind to the surface in dif-

75

ferent configurations, Figure 7.4. It has been established that a majority of the formate species bind to the rutile (110) surface in a bridge bonded fash-ion via two Ti5c atoms. Thus the formate molecule is oriented in parallel to the bridging oxygen rows in the [001] direction[34],[76],[149]. This orientation for formate is commonly denoted as species A [130],[150]–[152]. On the perfect rutile (110) surface saturation coverage of formate in type A configuration corresponds to a coverage of A = 0.5 where the formate is ordered in a 2×1 configuration. The coverage is calculated as the ratio between the number of formate molecules and the number of Ti5c sites.

Figure 7.4. The rutile (110) surface with four different proposed formate adsorption geometries (denoted A-D). Also shown is the orientation of the incoming IR light. The figure is adapted from paper I[130].

Different minority species configurations have also been observed on the surface. A formate species bonded via a bridging oxygen vacancy, Ovac, and a Ti5c atom is denoted as species B and have been reported in the literature[151],[152]. In this configuration the formate molecular axis is oriented in the 110 direction. Using scanning tunnel microscopy another formate species where one of the formate oxygen atoms is at the Ovac site whereas the other oxygen atom forms a hydrogen bond with a OHbr have also been ob-served[152]. This species is denoted species C in the literature and is oriented

76

along the bridging oxygen rows in the [001] direction. An additional config-uration that is oriented along the 110 direction where the formate is bond-ed to a Ti5c atom and a OHbr species have been proposed in Paper I and is called species D[130]. DFT+U calculations were performed on these four dif-ferent configurations to determine their binding energy. These calculations showed that the binding energy was very similar for species A, B and C (-1.84, -1.85 and -1.87 eV respectively), whereas the binding energy for spe-cies D was lower (-1.31 eV) and it is thus the least stable species. A method to simulate the IRRAS spectra was developed using a three layer model (see section 4.2.3) and the results from the DFT+U calculations. These simulated spectra were combined with experimentally measured IRRAS spectra on a TiO2 rutile (110) single crystal in order to investigate the adsorption geome-try of formate. The single crystal was prepared according to different proto-cols to achieve either a stoichiometric, reduced or oxylated surface in order to obtain surfaces with a varying concentration of oxygen vacancies and hydroxyl groups. This was done by repeated cycles of Argon sputtering and annealing as described in section 5.1. After surface preparation the LEED pattern was checked and showed bright spots in a 1×1 orientation typical for the rutile (110) surface.

To be able to distinguish between different vibration modes of the mole-cule both s- and p-polarized light was employed incident into two different directions on the crystal, either the [001] or the 110 direction. Spectra for the different polarizations and incident direction of the light on r-TiO2 are shown in Figure 7.5 and Figure 7.6. Differences can be seen in the different IRRAS spectra for p-polarized (Figure 7.5) and s-polarized light (Figure 7.6). The differences seen are due to both the coupling of the polarized IR light to different vibrational modes in formate, but also due to the different orientations of the majority species and the minority species. From geomet-rical reasons it can be concluded that p-polarized light couples to the sym-metric s(OCO) mode in species A regardless of incident direction, whereas the asymmetric mode only couples to p-polarized light incident in the [001] direction (= 90°). Likewise only s-polarized light incident in the 110 direction (= 0°) couples to the asymmetric as(OCO) mode whereas the symmetric s(OCO) mode do not couple to s-polarized light regardless of incident direction. This agrees well with the measured spectra where the symmetric and asymmetric (OCO) band is seen at 1360 and 1534 cm-1 for the majority species. Furthermore, for s-polarized light an emission band at 1374 cm-1 is seen only for light incident in the [001] direction. This band is therefore attributed to the in-plane wagging, (C-H), for the majority spe-cies. Moreover this also implies that the formate (C-H) wagging occurs along the molecular axis of the molecule.

77

Figure 7.5. IRRAS spectra measured with p-polarized light for formate adsorbed on a r-TiO2 surface. The IR light is incident in the 110 (=0°) and [001] (=90°) direction. The figure is adapted from paper I[130].

Figure 7.6. IRRAS spectra measured with s-polarized light for formate adsorbed on a r-TiO2 surface. The IR light is incident in the 110 (=0°) and [001] (=90°) di-rection. The figure is adapted from paper I[130].

The band at 1562 cm-1 is attributed to the asymmetric as(OCO) stretch for the minority species present on the surface. In order to resolve which of the minority species (species B, C or D in Figure 7.4) it is, comparisons with simulated IRRAS spectra were made. From this, species C (formate ad-sorbed in a monodentate configuration to an Ovac site) can be ruled out as the vibrational frequency for this species does not match the measured spectrum. As the reduced surface contains a larger amount of oxygen vacancies than the stoichiometric surface, a difference in intensities for the vibrational

78

modes for a minority species bonded to an Ovac site (specie B) is expected to occur with Ovac concentration. This is however not the case (Figure 7.7) and it is attributed to the rapid hydroxylation of oxygen vacancies on rutile TiO2 creating a hydroxylated surface[76],[85],[87]. This can be seen by the presence of hydrogen bonded formate ions at 1582 cm-1 [153]. Thus the band at 1564 cm-1 is attributed to the asymmetric as(OCO) mode for species D (formate bridge bonded to a Ti5c atom and a protonated oxygen atom in the bridging oxygen row), despite this being the least stable species, as this specie does not re-quire oxygen vacancies on the surface. Therefore, due to the rapid hydrox-ylation of the bridging Ovac sites formate bonded to a hydroxylated oxygen vacancy (or a protonated oxygen atom) and a Ti5c atom is expected to appear also at ambient conditions. Surface hydroxylation therefore plays an im-portant role for the reactivity of the rutile (110) surface and should thus be taken into account when modelling reactions on nanoparticles at ambient conditions in addition to bridging oxygen vacancies.

Figure 7.7. IRRAS spectra of formate adsorbed on differently prepared rutile TiO2 single crystals. The measurement is performed with p-polarized light incident along the 110 direction. The figure is adapted from paper I[130].

7.3 Interaction between formate and oxygen on the rutile (110) surface

Exposure of formate adsorbed on the rutile (110) surface to oxygen mole-cules results in the formation of a new vibration band at 1516 cm-1 in the IRRAS spectra. Concomitant to the formation of the band at 1516 cm-1 a decrease in the asymmetric as(OCO) stretch for formate at 1532 cm-1 is seen, Figure 7.8. The rate of the changes in the IRRAS spectra depends on

79

the partial pressure of oxygen in the vacuum chamber. With a O2 partial pressure of 1×10-6 mbar the new vibration band is clearly seen after 30 minutes of O2 exposure and after 90 minutes it is the dominant peak in the spectra, whereas for lower pressures the build-up is slower and for O2 pres-sures below 1×10-7 mbar no sign of this new band is seen even after 3 hours exposure. Removal of the O2 molecules by evacuating the vacuum chamber leaves the IRRAS spectra unchanged, i.e. it does not revert back to its initial form. Thus the addition of O2 to the gas phase induce an irreversible reaction that converts the bridge bonded HCOO into a new surface species. This new species is thermally stable as exposure of the new surface species to the UHV environment for an extended time, about 3 hours, shows no additional changes to the spectra The same changes in the IRRAS spectra is also seen if the order of exposure is reversed. Exposure of the pristine TiO2 surface to 100-1000 Langmuir of oxygen molecules also creates the vibration band at 1516 cm-1. As previously a larger O2 exposure generates a larger contribu-tion from the new species.

It can be speculated that the origin of these changes stem from the pres-ence of formate minority species adsorbed on the surface or the presence of oxygen vacancies. Based on previous simulations and measurements (paper I) it can be concluded that changes seen in the IRRAS spectra is not due to conversion from minority species B and D, which are not seen with p-polarized light incident in the [001] direction into species C, which couples to the infrared light incident into the [001] direction. Firstly measurements with s-polarized light, also incident in the [001] direction show no presence of these minority species on the surface. As the changes are large especially with high O2 partial pressures these species would be detected if present. Secondly from the simulations in paper I, the vibration bands for species C is shifted to higher wavenumbers for species C compared to species A. This shift is about 100 cm-1 whereas the new vibration band is shifted with about 15 cm-1 to lower wavenumbers compared to the as(OCO) stretch for species A. As previously mentioned, on a typical rutile (110) surface prepared by sputtering and annealing one can expect to have an Ovac concentration that is about 5-10% depending on the exact surface preparation procedure.[34],[86],[87] As oxygen vacancies have been reported to be healed by dissociation of ox-ygen molecules resulting in an oxygen adatom, Oad, bonded to a Ti5c site on the (110) surface[105],[106] any vacancies present after the sample preparation procedure is expected to be healed when the surface is exposed to oxygen molecules. As the conversion into the new species occurs both when oxygen is supplied before and after formic acid exposure it can therefore be ruled out the changes in the IRRAS spectra is caused by reactions of formate with Ovac sites on the surface. This is also corroborated with the measurements done without O2 exposure. If Ovac would be responsible for the chemical transfor-mation into the new species, these changes would also occur without expo-sure of the adsorbed formate to oxygen molecules.

80

Figure 7.8. IRRAS spectra of formate (HCOO=0.5) adsorbed on TiO2 (110) after different O2 exposure times and different pressures. (a) PO2=1×10-6 mbar, (b) PO2=5×10-7 mbar, and (c) PO2=1×10-7 mbar. The spectra were recorded at 273K with p-polarized light incident in the [001] direction. Figure is adapted from paper IV[154].

81

This is not the case as measurements without O2 exposure exhibits no ap-pearance of the new vibrational band at 1516 cm-1 but only a slow decay of the formate asymmetric band due to thermal desorption of formate. It can also be concluded that oxygen vacancies have a high affinity to water and are rapidly healed by the residual water present in the UHV chamber. STM studies have shown that Ovac sites are hydroxylated into OHbr species within a couple of minutes even at UHV pressures in the 10-10 mbar range.[76],[85],[155] Thus most of the Ovac created during the sample preparation procedure is expected to be healed and hydroxylated prior to oxygen exposure experi-ments and therefore the Ovac concentration is expected to be very low.

The new species causing the absorption band at 1516 cm-1 in the IRRAS spectra is instead attributed to a hydrogen carbonate surface complex, whose prerequisite to be formed is the presence of an Oad or hydroxyl group on the surface. With the use of STM Oad atoms have been observed at 300K and TPD data suggests that they can exist on the surface up to 600K.[109],[110] This suggests that the Oad atom is stable and long-lived on the rutile (110) surface, which is corroborated with our experiments on O2 dosing prior to formic acid exposure where the Oad must be present on the surface in the order of a minute or two before the formic acid exposure is commenced. With the pres-ence of water, even in very small amounts, Oad can react with a water mole-cule and form a terminal OH group bonded to a Ti5c site, OHt. Thus it can be expected that with our experimental conditions O2 exposure will lead to the creation of Oad and OHt on the surface. These will in turn react with formate and form a surface hydrogen carbonate species, O···O-OCH, which sche-matically is depicted in Figure 7.9.

Measurements of the O2 formate interactions at different temperatures ranging from 273K to 343K shows that with increased temperature the inten-sity of the surface hydrogen carbonate band increases, Figure 7.10. At the highest temperatures the formate as(OCO) peak has almost completely dis-appeared after 90 minutes of exposure whereas it is clearly seen as a shoul-der at lower temperatures. Performing a peak deconvolution of the spectra shows that the peak due to the formate as(OCO) stretch decreases exponen-tially whereas the band due to the hydrogen carbonate species simultaneous-ly grows exponentially. With the initial condition that the total coverage of the two different species equals saturation coverage, = 0.5, it is possible to calculate the coverage versus time by using the measured absorbance for type A formate, AA , and O···O-OCH, AX. Figure 7.11. Here it is assumed that the infrared cross section of the as(OCO) stretch in O···O-OCH and in formate is the same, which is reasonable given the similarity of the species and their orientation.

82

Figure 7.9. Schematic picture of the surface hydrogen carbonate species, O···O-OCH on the rutile (110) surface. Here the bridging oxygen atoms are red, bulk oxy-gen atoms green, titanium atoms are grey and the surface Oad atom is orange. The formate oxygen atoms are yellow, the carbon atom is blue and the formate hydrogen atom is white.

Figure 7.10. IRRAS spectra recorded with an O2 partial pressure of =1×10-6 mbar at different temperatures (a) 273K, (b) 303K, (c) 323K and (d) 343K. The spectra is recorded with IR light incident in the [001] direction. The figure is adapted from paper IV[154].

83

The kinetics for the transformation from formate into a surface hydrogen carbonate complex according to the reaction

⋅⋅⋅ (7.1)

can be modelled using a first order reaction kinetics, which is determined by the change in coverage of the two species according to

k (7.2)

k (7.3)

Here A is the coverage of type A formate, O is the oxygen adatom coverage and X is the coverage of the surface hydrogen carbonate complex associated with the 1516 cm-1 band. k is the reaction rate and kt is the thermal removal of formate into the gas phase by a dehydrogenation reaction[156]. As there is a constant supply of oxygen from the gas phase it can be assumed that the supply of oxygen is not limiting the reaction. The changes in absorbance for the two species is then given by a pseudo first order reaction, where the pseudo reaction rate, kr, is defined as kr

= kO. Using the same initial condi-tion as described above one then gets

, (7.4)

, 1 , (7.5)

by utilizing the assumption that at t = 0, the sum of AA,0 and AX,0 equals satu-ration coverage (AA,0 + AX,0 = 0.5). Thus the reaction rates at different tem-peratures is determined and presented in Table 1. The activation energy for the transformation of formate into a hydrogen carbonate surface complex and dehydrogenation of formate can be determined using Arrhenius analysis.

,

,

(7.6)

84

where Er is the activation energy for the reaction to form O···O-OCH and Et

is the activation energy for the dehydrogenation of formate, kb is Boltzmanns constant and T is the temperature. An activation energy of Er = 0.25 eV is found for the reaction into O···O-OCH and Et = 0.19 eV for the dehydro-genation of formate. The hydrogenation activation energy determined here is close to that previously reported, 0.16 eV, by Onishi et. al.[156].

Table 1. Reaction rates determined using a pseudo first order reaction kinetics with an oxygen partial pressure of 1×10-6 mbar.

Temperature kr (min-1) kt (min-1)

273 K 0.2×10-2 0.9×10-3

303 K 0.4×10-2 1.0×10-3

323 K 0.9×10-2 3.2×10-3

343 K 1.9×10-2 3.9×10-3

Figure 7.11. Formate (A, red circles) and surface hydrogen carbonate (x, black squares) coverage deduced from the 1516 cm-1 and the 1532 cm-1 absorption bands as a function of time at (a) 273 K (b) 303K, (c) 323K and (d) 343K. The absorbance bands were obtained by peak deconvolution of the IRRAS spectra in Figure 7.10. The solid lines are least squares fits to experimental data. The figure is adapted from paper (IV)[154].

85

7.4 Formate adsorption on disordered rutile single crystal and rutile nanocrystalline surfaces

The importance of the surface structure for formate adsorption has been in-vestigated by performing measurements either on a reduced and ordered, or a reduced and disordered TiO2 single crystal surface (paper II) and a nano-crystalline TiO2 sample (paper V). The reduced surface is created by repeat-ed argon ion and annealing cycles like those described earlier in section 5.1. The disordered surface was obtained by argon sputtering cycles without an-nealing-cooling cycles and is further on denoted sp-TiO2. LEED measure-ments on the disordered sp-TiO2 surface showed no pattern at all, which suggest a disordered surface structure within the coherence length of the LEED electrons.

From the IRRAS spectra, obtained with p-polarized light incident in ei-ther the 110 (Figure 7.12) or the [001] (Figure 7.13) direction, it can be seen that regardless of incident direction of the IR light the formate spectra looks the same on the sp-TiO2 surface, which is not the case for the r-TiO2 surface. The position of the peaks on the sp-TiO2 surface is also shifted compared to the reduced TiO2. The change in appearance of the formate adsorption spectra is attributed to the lack of order on the sp-TiO2 surface. As a result of this formate cannot adsorb in the same structured way as on the r-TiO2 sample, which manifests itself as the IRRAS spectrum being simi-lar in both incident directions of the IR light. Another aspect of the argon ion sputtering is that the surface consist of a larger amount of defects and also, due to the preferential sputtering of oxygen, under-coordinated Ti atoms. These defects and Ti3+ atoms will change the way the formate molecule is adsorbed on the surface. Instead of the bridge bonding fashion seen for or-dered TiO2 surfaces the formate molecule is expected to bond asymmetrical-ly to the Ti3+ atoms[84],[157]. This change in adsorption geometry will affect the position of the vibrational bands resulting in the appearance of the 1391 cm-1 band and the shift of the as(OCO) band. There are some resemblances in the observed peaks between the sp-TiO2 and the r-TiO2 surfaces, which suggests the existence of some domains with (110) planes at the surface.

86

Figure 7.12. IRRAS spectra on an ordered (r-TiO2), a disordered (sp-TiO2) rutile (110) surface obtained with p-polarized light incident along the 110 direction. The IRRAS spectra, obtained with p-polarized light, for a nanocrystalline (np-TiO2) sample is also shown as well as the absorbance spectra for a film with nanocrystal-line rutile particles (R9x26) obtained at atmospheric conditions.

Figure 7.13. IRRAS spectra on an ordered (r-TiO2) and a disordered (sp-TiO2) rutile (110) obtained with p-polarized light incident along the [001] direction. The IRRAS spectra, obtained with p-polarized light, for a nanocrystalline (np-TiO2) sample is also shown as well as the absorbance spectra for a film with nanocrystalline rutile particles (R9x26) obtained at atmospheric conditions.

87

The non-transparent nature of the metallic substrate for the thin film nano-crystalline sample, np-TiO2, also changes the appearance of the spectra to always exhibit absorption peaks compared to the single crystal measure-ments where both emission and absorption peaks can be seen. Looking in more detail it can be seen that for the nanocrystalline TiO2 sample the as(OCO) band is shifted to higher wavenumbers and broadened compared to the single crystal surfaces. For this sample no separation between the s(OCO) band and the (C-H) band can be seen as was the case for the sin-gle crystal surfaces, instead a broad band at 1380 cm-1 is seen. As the np-TiO2 sample exhibits several different facets for formic acid to adsorb on, and the atomic distances is not the same for different facets, a variance in the vibrational energy for formate is expected and this is manifested as a broad-ening of the peaks in the IR spectra. For instance on the rutile (111) surface the distance between two neighbouring titanium atoms is 5.46Å compared to 2.95Å for the rutile (110) surface[34],[158]. The longer distance between the titanium atoms on the (111) surface, makes bridge bonding of formate less favourable as this distance is much longer than the oxygen-oxygen distance in formate making monodentate and bidentate bonding of formate to the surface more favourable, and as a consequence of this the position of the vibration bands changes[158]. Furthermore the nanocrystalline sample, due to the small size of the particles (Figure 7.14) exhibits more edges and defects that also influence the adsorption geometry of formate and thus affect the peak position of the vibrational bands.

Figure 7.14. SEM image of the nanocrystalline TiO2 rutile sample (np-TiO2), made by DC reactive magnetron sputtering, showing the semi-porous structure of the np-TiO2 sample.

88

Figure 7.15. X-ray diffractogram of the nanocrystalline sample showing the contri-butions from the different crystal planes.

Compared to the rutile (110) single crystal surface, no interaction between the adsorbed formate molecule and oxygen molecules are seen when the surface is exposed to an oxygen rich environment. Even though there is a contribution from the (110) crystal plane in the X-ray diffractogram, Figure 7.15, it does not necessarily mean that all these planes are exposed at the surface and thus the (110) surface contribution to the overall spectrum might be low. This also suggests that the spontaneous reaction between formate and Oad and OHt to form hydrogen carbonate surface complexes does not occur on all TiO2 facets, which is a consequence of the different adsorption geometries of formate on these surfaces.

For adsorption of formic acid on rutile nanoparticles (R9x26) at atmos-pheric conditions the adsorption of formic acid is also shown in Figure 7.12 and Figure 7.13. These rutile nanoparticles have a slightly elongated shaped with a width of about 9 nm and a length of 26 nm. The absorption spectrum exhibits broader peaks compared to that of the sputtered film and single crystal spectra. In the region of the asymmetric as(OCO) vibration two main peaks are seen at 1557 cm-1 and 1588 cm-1. The former of these two peaks is attributed to bridge-bonded formate while the later peak is associated with formate ions on the surface. A small shoulder at 1650 cm-1 is also seen that is due to the (C=O) bond found in monodentate (-coordinated) formate. At even higher wavenumbers, 1715 cm-1, a small peak is seen, which is associ-ated with the (C=O) bond found in formic acid hydrogen bonded to the surface. These different formate coordinations also give rise symmetric vi-bration bands and C-H wagging bands mainly in the region between 1350-1390 cm-1. These bands are often close to each other and are therefore more difficult to identify than the as(OCO) and (C=O) bands. Most notably are

89

the C-O vibration band for monodentate formate at ~1260 cm-1 and for hy-drogen bonded formic acid it is shifted to even lower wavenumbers, ~1210 cm-1. These two adsorbates also have their (C-H) vibration band shifted to lower wavenumbers, 1335 and 1310 cm-1 respectively, compared to bridge bonded formate[159].

7.5 Formic acid adsorption on anatase and brookite nanoparticle films at atmospheric conditions

On the anatase and brookite surface formic acid adsorbs differently com-pared to the rutile surface, Figure 7.16. On the anatase surface the amount of formate ions present on the surface is relatively less compared to the rutile surface, as seen by the lack of a peak around 1588 cm-1. Furthermore the amount of monodentate formate is relatively large. Similar trends are also seen on the brookite sample, with a relatively large amount of monodentate formate and also a large contribution from hydrogen bonded formic acid. From TEM and XRD measurements of these films it has been concluded that the anatase film (A25) consists of particles with a size of around 23 nm ex-posing mainly <101> and <112> facets. The other anatase film (A40) con-sists of larger anatase particles with a size about 40 nm, which exposes, to a large extent the same facets. Whereas the brookite film (B11) on the other hand consists of particles with a diameter of 11 nm and a doughnut shape on some particles that suggest coalescence of two or more particles[39].

90

Figure 7.16. Adsorption spectra of formic acid on TiO2 films consisting of nanopar-ticles of anatase with a size of 25 nm (A25) and 40 nm (A40) as well as for brookite nanoparticles with a size of 11 nm (B11).

7.6 Interaction between formate and light on the rutile TiO2 surface at UHV conditions

Illumination with UV light of formate adsorbed on differently prepared rutile TiO2 surfaces has been done. On the reduced TiO2 surface a miniscule change in peak intensity was seen with illumination whereas on the sputtered disordered rutile (110) single crystal surface a larger, albeit still small, change in intensity of the formate vibration bands is seen, Figure 7.17. On the nanocrystalline samples larger absolute changes in the adsorbed formate is seen with illumination, Figure 7.18. The small changes seen, especially on the single crystal surface in UHV conditions are not surprising as measure-ments on large rutile nanoparticles in atmospheric conditions exhibit a very low photocatalytic activity[67]. The kinetics of the photocatalytic degradation of the adsorbed formate can be modelled with a first order kinetics reaction similar to that described in section 7.3 where the intensity or area for a for-mate vibrational band is assumed to decrease exponentially with time,

where A(t) is the peak intensity or peak area at time t and kr is the reaction rate for the photocatalytic degradation. For the sputtered single crystal sur-face a degradation rate of 9×10-4 min-1 is obtained after 100 minutes of illu-

91

mination, whereas the changes on the reduced single crystal surface was too small compared to the long term drift of the system to quantify a reaction rate. On the nanocrystalline sample a similar reaction rate, 8×10-4 min-1, as that observed for the sp-TiO2 surface is measured. As a result of the in-creased number of formate adsorption sites on this surface, due to the porosi-ty of the TiO2 film, the changes in formate coverage can clearly be followed with time when the film is illuminated with UV light, Figure 7.18.

Figure 7.17. IRRAS spectra before, after 50 and 100 minutes of UV irradiation of formic acid adsorbed on a reduced rutile (110) surface (r-TiO2) and sputtered rutile (110) surface (sp-TiO2). The spectra is obtained with p-polarized light incident in the [110] direction (a and d) and [001] direction (b and e). For comparison measure-ments without UV illumination (c and f) incident in the (001) direction is also shown. The figure is adapted from paper II[101].

For measurements performed in an UHV environment the supply of oxygen radicals (O2

●-) and hydroxyl radicals (OH●) will be limited due to the low availability of oxygen and water. Thus the photocatalytic activity will be low since these radicals, as previously described in section 2.1, is important for the photocatalytic degradation process where formate is decomposed into carbon dioxide and water. At oxygen deficit conditions it have previously been shown using a powder mixture of anatase and rutile that surface lattice oxygen participates in the photocatalytic decomposition leaving an oxygen deficit surface that is replenished by diffusion of oxygen from the bulk[160],[161]. Thus for long illumination times the degradation rate becomes limited by the oxygen diffusion rate from the bulk of the TiO2

[142],[160]. As these studies was done on anatase or anatase and rutile powder mixtures, the mechanism is not necessarily the same on a pure rutile sample, since STM studies of anatase have shown that the oxygen vacancies diffuses into the

92

subsurface layers to create a vacancy free surface[94]–[96]. This mechanism is not seen or proceeds much more slowly on the rutile surface where oxygen vacancies are readily observed[85],[111].

Figure 7.18. IRRAS spectra before and after UV irradiation of formic acid adsorbed on rutile thin film nanocrystalline sample (np-TiO2) made by DC reactive magnetron sputtering. Spectra are obtained without (a) and with (b) O2 exposure at a partial pressure of 1×10-6 mbar during UV illumination. For comparison measurements (c) with exposure to O2 without UV illuminations and (d) no UV or O2 exposure is also showed.

The similar degradation rate observed for both the nanocrystalline surface and the sputtered single crystal surface can be attributed to different reasons. Either the two samples expose similar facets and defects yielding a similar degradation rate. This seems less likely as the XRD for the np-TiO2 sample shows contributions from a variety of facets, whereas the sp-TiO2 surface is a disordered and atomically rough single crystal rutile (110) surface. Another explanation for the similar degradation rate for the two surfaces is that the process is limited by the scarce availability of oxygen and hydroxyl radicals. As the measurements are performed at the same UHV conditions, this then result in a similar degradation rate due to the same amount of oxygen and hydroxyl present in the UHV chamber. Thus if oxygen is supplied during the process an increase in the degradation rate would be expected. Measure-ments with an increased oxygen partial pressure, 1×10-6 mbar, compared to the UHV condition where is less than 1×10-8 mbar show a somewhat increased activity, Figure 7.19, with a degradation rate of 9×10-4 min-1. That

93

the difference between the two different oxygen partial pressures is not larg-er is attributed to the fact that hydroxyls or Ovac are needed on the TiO2 sur-face to form Oad and these are not expected to change much with O2 dosing. Also the low incoming photon flux can be a limiting factor on the reaction rate. Thus suggesting that any difference in the photocatalytic activity be-tween the different oxygen concentrations cannot fully be revealed at UHV conditions with the photon flux used here. Noteworthy is also the very low activity for the r-TiO2 surface compared to the sp-TiO2 and np-TiO2.

Figure 7.19. Fit of a first order reaction kinetics model for the photodegradation of formate on the np-TiO2 sample with UV illumination in UHV conditions (dots) and with an increased oxygen partial pressure, = 1×10-6 mbar (squares).

7.7 Interaction between formate and light on anatase, brookite and rutile TiO2 nanoparticle films at atmospheric conditions

Compared to the measurements done at UHV conditions the photocatalytic degradation of the adsorbed formic acid proceeds much quicker at atmos-pheric conditions. This is mainly due to the increased presence of oxygen and water on the surface. Another contributing factor is the larger photon flux used in these experiments. Comparing the photocatalytic degradation on the different crystal structures, Figure 7.20, large differences can be seen. On the rutile R9x26 and anatase A25 sample the degradation proceeds rather slowly with monodentate and hydrogen bonded molecules primarily disap-pearing whereas the bridge bonded formate molecules are more resilient to photocatalytic degradation. On the anatase A25 film it can also be seen that monodentate and hydrogen bonded molecules are converted into bridge bonded formate, as the asymmetric vibration band at 1560 cm-1 associated to bridge bonded formate increases during the first 30 minutes of illumination.

94

Figure 7.20. Photocatalytic degradation of formic acid adsorbed on a) anatase A25, b) anatase A40, c) brookite B11 and d) rutile R9x26. Measurements are done at atmospheric conditions with simulated solar light illumination.

This transformation into bridge bonded formate does not occur on the larger anatase particles (A40). Instead a rather rapid decrease of also bridge-bonded formate is seen on this anatase sample. After 30 minutes of illumination only a small fraction of the initial coverage of bridge bonded formate is left on the surface, whereas the monodentate and hydrogen bonded formate has all dis-appeared. On the brookite sample (B11) the disappearance of the adsorbed format and formic acid also proceeds quickly. The relatively larger amount of bridge-bonded formate on this sample compared to the A40 sample results in a somewhat slower degradation rate. After 60 minutes of illumination the vibration band due to bridge bonded formate is still clearly seen in the spec-tra, although the area has decreased substantially. Using a first order kinetics reaction, as described above, section 7.6, to quantify the degradation rate using the integrated peak area for the different formate coordinations found in the wavenumber region from 1500 cm-1 to 1800 cm-1 one can see that the degradation of formate proceeds almost equally slow on the rutile R9x26 sample (0.012 min-1) as on the anatase A25 sample (0.011 min-1). Both of these samples also favours bridge-bonded formate, whereas the larger ana-tase particles, which has a relatively low amount of bridge-bonded formate, exhibits a much quicker degradation rate (0.061 min-1). This is also true for the brookite sample, which displays similar photocatalytic performance (0.054 min-1).

95

8 Concluding remarks and outlook

When investigating fundamental processes on materials that is used in dif-ferent applications there often exists a discrepancy between the experimental conditions and the environment of the application and for photocatalysis one often talk about a materials and pressure gap. In order to have precise control of the photocatalyst surface and the experimental conditions, which often are needed for fundamental investigations, these studies are often done in UHV and on single crystal surfaces. Most commonly the rutile (110) single crystal surface is used as this surface is readily available in large area substrates since it is the thermodynamically most stable TiO2 surface[34]. This is in con-trast to most applications that takes place at ambient conditions using na-nosized particles of anatase TiO2 either as porous films, powders or compo-site materials[29],[30],[46]. This results in a materials and pressure gap. With the development of infrared spectrometers, resulting in an increased sensitivity, it is now possible to perform IR spectroscopic measurements in UHV of sub-monolayer coverage of adsorbate molecules on small area metal oxide single crystals using state of the art apparatus[69],[162],[163]. This makes it possible to provide fundamental information about the adsorbate structure on different TiO2 structures, such as anatase, for which large area single crystals are not available and thereby providing a method to start to bridge the materials gap[69],[164].

Here another approach is used in other to bridge the materials and pres-sure gap in photocatalysis where adsorption and photochemistry properties are studied in a step-by-step approach. As a starting point rutile single crys-tals with different surface pre-treatments are investigated in UHV environ-ment. Nanocrystalline surfaces and nanoparticle films with different crystal-line phases is subsequently studied both in UHV and at atmospheric condi-tions. Formic acid is mainly used as a probe molecule for different reasons. This molecule is the simplest of the carboxylates and with only a few atoms it gives rise to IR spectra that is relatively easy to interpret compared to more complex molecules. Additionally photocatalytic decomposition of more complex organic molecules occurs by splitting bonds and creating gas phase CO2 and H2O molecules. Thereby creating simpler and simpler molecules on the surface, where the last decomposition step often include decomposition of formate on the photocatalyst surface[33],[35],[141]. Furthermore it has been shown that TiO2 has a high affinity for formic acid and monolayers of for-mate builds up on the surface during exposure to ambient air[36].

96

The adsorption of formic acid on different preparations of the rutile (110) surface have been investigated in order to examine how the amount of va-cancies and hydroxyl groups on the surface affect the formate adsorption behaviour. A method to simulate the IRRAS spectra using a three layer opti-cal model and the polarizabilities of the adsorbed molecules, obtained from DFT calculations, was developed. A combination of simulation and experi-mental data made it possible to differentiate the contribution from different adsorbate configurations in the measured IRRAS spectrum. These results point to the presence of a minority formate species on the surface that is bonded to a proton found on the O2c bridging oxygen, created either by hy-droxylation of an oxygen vacancy or by the proton dissociated from formic acid when it adsorbs on the rutile (110) surface[130].

The same simulation method was also used to investigate the adsorption geometry of CO on the rutile surface. This highlighted different adsorption geometries of carbon monoxide for different adsorbate coverages. It was found that for coverages below 1 ML, CO adsorbs in an upright tilted struc-ture whereas the CO molecules in the second adsorbate layer, i.e. the mole-cules that creates coverage larger than 1 ML, is adsorbed in a tilted horizon-tal geometry[148].

Exposure of formic acid adsorbed on the rutile (110) surface to an oxygen rich UHV environment induces a thermally activated reaction between the adsorbate and oxygen. This will result in the formation of a surface hydrogen carbonate species, O···O-OCH. The transformation into the new surface species can clearly be seen in the IRRAS spectra where a decrease of the asymmetric (OCO) band is accompanied with a concomitant increase of a new vibration band. This conversion of formate also occurs simultaneously with a dehydrogenation of formate into the gas phase. Measurements at dif-ferent temperatures have shown that these processes are thermally activated[154].

In order to better mimic nanocrystalline samples the rutile (110) surface was prepared in a way that resulted in a disordered surface. Adsorption of formic acid on this surface showed a relatively large shift in peak position and broadening of the asymmetric as(OCO) vibration band compared to the rutile (110) surface. The lost order on the surface give rise to different ad-sorption configurations that in turn has different adsorption energies and thereby results in different IR vibration bands. This disordering of the sur-face is clearly seen in the IRRAS spectra as a lack of difference between measurements made with the IR light incident along the [001] and the 110 direction. On an ordered surface these different orientations of the crystal results in different IRRAS spectra due to the orientation of the formate on the surface. For a nanocrystalline thin rutile TiO2 film, made by reactive sputtering and subsequent annealing, an even larger shift and broadening of the formate asymmetric (OCO) band occurs. Compared to the adsorption spectra of formic acid on nanoparticle samples obtained in atmospheric con-

97

ditions both the nanocrystalline thin film sample and the rutile (110) surface shows similarities and discrepancies. For the large rutile nanoparticles shown in Figure 8.1 the main contribution comes from bridge bonded for-mate just as on the rutile (110) surface as these nanoparticles dominantly expose <110> facets. Whereas smaller rutile nanocrystals shows larger simi-larities to the thin sputtered TiO2 film, Figure 7.13. As both these samples expose different facets, and thereby allows for different adsorption configu-rations, which in turn will affect the position of the formate vibration band compared to that of bridge bonded formate on the rutile (110) surface[101].

Figure 8.1. Adsorption of formic acid on a reduced rutile (110) single crystal surface (r-TiO2), a disordered rutile (110) single crystal surface (sp-TiO2) and large (6 x 80 nm) rutile nanoparticles exhibiting a large fraction of <110> facets. The figure is adapted from paper II[101].

Illumination of formic acid with UV light results in decomposition of the adsorbed formate species. Depending on the surface the photodecomposition proceeds at different rates. On the ordered rutile single crystal surface a de-crease of the formate coverage is seen with illumination but the changes are too small to be quantified with any certainty. Whereas on the disordered surface and the nanocrystalline thin film samples a larger decrease in for-mate coverage with illumination is seen. On these two types of surfaces the decomposition rate is similar and no large changes of the formate degrada-tion rate is seen in the presence of an oxygen rich UHV environment. This can be compared to the nanoparticle samples illuminated at atmospheric conditions where the degradation rate is much larger. The surface structure of the nanoparticles is important as it affects the degradation rate. Nanoparti-cle samples exhibiting a large amount of rutile (110) surfaces has a lower degradation rate compared to that of rutile particles with less fraction of (110) surfaces. This agrees well with measurements at UHV conditions where the degradation proceeds much quicker on the disordered single crys-tal surface compared to the rutile (110) surface, due to the different format

98

coordinations on the surface. For anatase and brookite nanoparticles, mono-dentate adsorption of formate and hydrogen bonding of formic acid is rela-tively more common compared to rutile nanoparticles, and thus yields a higher photocatalytic degradation rate.

If one allows oneself to look into the near-term future several different in-teresting research questions lies ahead. The most obvious is the continuation of the present work, adsorption and photochemistry of formic acid, on ana-tase single crystals and also other rutile surface orientations. As mentioned earlier this poses some challenges due to the small area of the available ana-tase substrates. These substrates usually have an area of around 25 mm2, which can be compared to the 400 mm2 area of the rutile (110) single crystal used in these studies. Thus the signal from the adsorbate on the anatase sin-gle crystal substrates will be very weak requiring good control of the exper-imental conditions. As have been shown in Figure 4.7 the change in reflec-tance for a surface with and without a thin layer of adsorbate is dependent, among other factors, on the polarization of the incoming light and the angle of incidence. For p-polarized light the maximum change in reflectance oc-curs at around 85° whereas for s-polarized light the maximum change in reflectance takes place at a lower angle of incidence, around 70°. Thus by changing the angle of the incoming IR light it should be possible to achieve a higher sensitivity, especially for s-polarized light. The sensitivity can also be increased by moving the IR detector inside the UHV chamber which will both result in a better vacuum around the detector reducing gas phase contri-butions and also reducing intensity loss from the optical windows in the light path.

To fully determine whether the small differences seen in photocatalytic activity for degradation of formate on the np-TiO2 sample with and without the presence of an oxygen rich environment stems from the low illumination power or is due to the low oxygen adatom formation rate on the surface, new experiments with high intensity UV-light either from diodes or a laser could be done. The use of monochromatic radiation is preferred, since the use of broadband light could cause heating of the sample due to the low cooling efficiency for large thermal loads in the UHV chamber.

Another aspect that could be interesting to investigate is the adsorption of molecules on single crystals at high pressure. Due to the low number of ad-sorption sites on a single crystal grazing incidence is needed in order to re-solve the adsorbates. At near ambient pressures the small signal from the adsorbate will then disappear in larger contribution from the gas phase mole-cules. This problem has on metallic substrates been circumvented by the use of polarization modulated IRRAS spectroscopy (PM-IRRAS). On metallic substrates s-polarized light does virtually not interact with the substrate and thus only carries information from the gaseous molecules in the light path. P-polarized light on the other hand interacts both with the adsorbates on the metal surface and the gaseous environment. By rapidly modulating the polar-

99

ization of the IR light it is thus possibly to almost simultaneously probe both the gaseous environment with s-polarized light and the adsorbates on the surface as well as the gaseous environment with p-polarized light. The con-tribution from the adsorbates on the metallic surface can then be extracted allowing for studies of thin adsorbates layers in near ambient pressures[165]–

[167]. For TiO2 single crystals substrates PM-IRRAS cannot be applied since s-polarized light interacts with surface adsorbates on non-metallic samples. However, deposition of an ultrathin single crystal layer of TiO2 on a metallic substrate will allow the system to be treated as a metallic substrate. Thus the interaction between the surface adsorbates and s-polarized light can be ne-glected and allow for discrimination between the contributions from the ad-sorbates and the gaseous environment. This method has previously been used to investigate self-assembled monolayers on thin SiO2 films deposited on gold substrates[168].

101

9 Sammanfattning på svenska

För att kunna uppnå FNs klimatmål måste både energiframställningen från förnybara källor öka samtidigt som energibesparande åtgärder krävs. I väst-världen kan energiförbrukningen i byggnader utgöra upp till 40 % av det totala energibehovet och genom att göra byggnaderna mera energieffektiva är det möjligt att göra stora energibesparingar. Tätare och energieffektivare hus kan dock resultera i ökade halter av flyktiga organiska ämnen i inomhus-luften. Dessa ämnen är hälsoskadliga och kommer från bland annat rengö-ringsmedel, plaster och lim i mattor och möbler. Aktiv nedbrytning av dessa hälsoskadliga ämnen med hjälp av fotokatalys gör det möjligt att skapa ett hälsosamt inomhusklimat även i byggnader med reducerad ventilation.

I en fotokatalytisk process används ett material som absorberar ljus, vilket generar elektron-hål par i materialet. Dessa laddningsbärare kan i sin tur skapa syre- och hydroxylradikaler på fotokatalysatorns yta genom reaktioner med syre och vatten, Figur 9.1. Vatten och syre finns alltid på fotokatalysa-torns yta eftersom dessa molekyler alltid finns i den omgivande luften. Det vanligaste förekommande fotokatalytiska materialet är TiO2 som existerar i tre olika kristallstrukturer, anatas, brookit och rutil. Av dessa tre är rutil den kristallstruktur som är termodynamiskt mest stabil och de andra två kristall-strukturerna kan omvandlas till rutil genom värmning av TiO2 till temperatu-rer över 600°C. Detta har gjort att enkristaller av rutil, och då främst rutil (110) ytan till en typyta för fundamentala undersökningar av adsorption och fotokemiska reaktioner hos TiO2. Dessa fundamentala undersökningar ge-nomförs i ultrahögt vakuum (UHV) för att få en kontrollerbar miljö där re-aktioner på ytan kan noggrant studeras.

IR spektroskopi är en metod för att undersöka bindningarna mellan olika atomer i molekyler. När en molekyl adsorberas på en yta kommer vibration-erna mellan atomerna att ändras vilket gör det möjligt att dels särskilja huruvida molekylerna finns inbundna till en yta eller befinner sig i gasfas i den omgivande miljön. IR spektroskopi gör det också möjligt att se hur mo-lekylen binder in till ytan eftersom detta också kommer att påverka vibrat-ionerna mellan atomerna i molekylen. Genom att använda olika polarisation-er på det inkommande ljuset och en flack infallsvinkel är det möjligt att un-dersöka adsorption av molekyler med en täckning som är mindre än ett mo-nolager på en enkristallin TiO2 yta. I denna avhandling har, av flera skäl, främst myrsyra, HCOOH, använts som testmolekyl för att undersöka ab-sorption och fotodegradationen på olika TiO2 ytor. Myrsyra har bland annat

102

den fördelen att det är en enkel molekyl vilket gör det uppmätta IR absorpt-ionsspektrumet mindre komplicerat och därmed enklare att tolka. När myr-syra adsorberar på TiO2 bryts bindningen till en av väteatomerna och format, HCOO, binds in på ytan. För många mera komplicerade organiska molekyler är format det sista steget i en total nedbrytning av ursprungsmolekylen. Detta sista nedbrytningssteg är vanligtvis också det som tar längst tid, vilket bidrar till att göra myrsyra till en intressant testmolekyl.

Figur 9.1. Principskiss för fotokatalys i en nanopartikel. (1) Absorption av fotoner som skapar elektron-hål par. (2) Dessa diffunderar sedan ut till ytan av partikeln där de (3) reagerar med vatten och hydroxyler som adsorberats på ytan och skapar radi-kaler. Dessa radikaler reagerar i sin tur med de molekyler som finns adsorberade på ytan av partikeln.

Beroende på hur den enkristallina ytan preparerats, ändras hur format adsor-beras på ytan. För en enkristallin yta med mycket hydroxyler binder en del av formatmolekylerna in via dessa hydroxylgrupper istället för att binda in direkt till titanatomerna som finns på ytan. Under det här arbetet har också en metod för att simulera det uppmätta IR spektrumet tagits fram. Denna simuleringsmetod gör det möjligt att särskilja hur olika bindningar av mole-kylerna bidrar till det uppmätta IR spektrumet och därmed identifiera vilka molekyler som finns på ytan. Denna simuleringsmetod har även använts för att studera hur kolmonoxid binder in på en rutil (110) yta och resultaten har jämförts med tidigare uppmätta IR spektra.

Om format som har adsorberats på rutil (110) ytan utsätts för en syrerik miljö startar en reaktion mellan formatmolekylerna och syreatomerna vilket resulterar i bildandet av ett syre-format komplex på ytan. Ombildningen från format till denna nya molekyl kan tydligt följas i IR spektrumet där en tydlig minskning av det asymmetriska (OCO) vibrationsbandet tillhörande format ses samtidigt som ett nytt vibrationsband tillhörande syre-formatkomplexet

103

växer upp. Denna omvandling är irreversibel, termiskt aktiverad och sker därmed snabbare med ökad substrattemperatur.

För att bättre kunna efterlikna nanokristallina material som användes i olika tillämpningar har den enkristallina ytan även förberetts på ett sett som skapar en oordnad ytstruktur. Detta gav upphov till ett skifte till högre vågtal och en breddning av det asymmetriska vibrationsbandet. Detta beror på om-struktureringen av ytan. På den oordnande ytan (benämnd sp-TiO2) finns flera olika möjligheter för format att binda in. Detta kommer att ändra posit-ionen på vibrationsbandet då detta är känsligt för hur molekylerna absorberar på ytan. Mätningar som gjorts med myrsyra på en annan rutil ytstruktur (111) har visat att på denna yta finns olika adsorptionskonfigurationer av format som ger upphov till olika vibrationsband. Undersökningen av en tunn polykristallin film av rutil TiO2 (np-TiO2) påvisar också ett skifte och bredd-ning av den asymmetriska (OCO) vibrationsbandet. Jämfört med adsorption-en av myrsyra på nanopartiklar finns det likheter i adsorptionsspektrumen för både den ordnade rutila (110) ytan (benämnd r-TiO2) och den oordnade rutila (110) ytan, beroende på hur nanopartiklarna ser ut. För stora nanopar-tiklar som har en större andel av rutil (110) facetter finns det större likheter mellan den ordnade enkristallina ytan (Figur 9.2) medan mindre nanopartik-lar av rutil visar större likheter med den oordnade enkristallina ytan (Figur 9.3).

När TiO2 belyses med UV ljus sker en fotokatalytisk nedbrytning av for-matmolekylerna. Hur fort denna nedbrytning går beror förutom på fotonflö-det även på ytans struktur och den omgivande miljön. För mätningar i UHV finns det inget syre eller vattenånga runt provet varför nedbrytningen går långsammare då den styrs av hur fort det syre som förbrukats på ytan kan återfyllas genom diffusion av syreatomer från bulken av materialet eller via omgivningen. Dessa mätningar visar att på den ordnade enkristallina ytan går nedbrytningen så långsamt att det är svårt att tillförlitligt kvantifiera skillnaderna. På den oordnande ytan och den tunna nanokristallina filmen är nedbrytningen snabbare. Detta beror att på dessa ytor kan format binda in annorlunda och inte genom bryggbindning, som på den ordnade ytan. Bryggbindning är en stabil adsorptionsstruktur som är svårare att bryta ned än andra adsorptionsgeometrier av format. Dessa oordnade ytor har också genom sin oordnande struktur fler adsorptionssäten för hydroxyler och vat-ten vilket underlättar för nedbrytningen av format. För prover som belyses under atmosfäriska förhållanden i syntetisk luft finns det ett överflöd av syre och vatten i den omgivande miljön vilket gör att nedbrytningen går mycket fortare på dessa ytor.

En framtida fortsättning på detta arbeta skulle kunna vara att med samma metoder studera adsorption och fotokemi av format på enkristaller av anatas. Detta är av intresse då de flesta tillämpningar använder nanopartiklar av anatas då den generellt anses ha större fotokatalytisk aktivitet. Detta medför dock en del experimentella utmaningar eftersom enkristaller av anatas inte

104

finns tillgängligt med stor ytarea. Normalt har dessa en ytarea på 25 mm2 till skillnad från de 400 mm2 hos de rutila enkristallerna som använts i denna avhandling. Detta gör att mätuppställningen måste vara optimal genom att implementera olika mättekniska lösningar för att kunna detektera de absor-berade molekylerna på dessa små enkristaller.

Figur 9.2. Adsorption av myrsyra på en reducerad rutil (110) enkristall (r-TiO2), en oordnad rutil (110) enkristall (sp-TiO2) och stora rutila nanopartiklar som har en stor andel <110> facetter mätt med p-polariserat ljus.

Figur 9.3. Adsorption av myrsyra på en reducerad rutil (110) enkristall (r-TiO2), en oordnad rutil (110) enkristall (sp-TiO2), en tunn nanokristallin film av rutil TiO2 (np-TiO2) och rutila nanopartiklar.

105

10 Acknowledgements

First of all I want to express my gratitude to my main supervisor Professor Lars Österlund for his support, encouragement and trust in me to follow my own ideas. I would also like to thank my co-supervisors Professor Gunnar Niklasson for his profound knowledge and kindness and Professor Claes Göran Granqvist for his support and advices. I am also indebted to the divi-sion of solid state physics for providing me with the opportunity to pursue a PhD degree as a part of my regular work. Anna Maria Lundin foundation is also gratefully acknowledged for travel grants to conferences during my PhD studies.

I would also like to thank all my collaborators for many interesting and rewarding research projects: Professor Kersti Hermansson and Dr. Shuanglin Hu for sharing your knowledge about DFT calculations. Professor Ewa Wäckelgård, Dr. Shuxi Zhao and all the other members of the Digespo pro-ject for all the fun. Dr. Vaclav Štengl, Dr Snejana Bakardjieva, Dr Jiri Henych and Dr Petra Ecorchard for all your help and for making my visits to your institute so pleasant. Dr Malin Johansson and Professor Sten-Eric Lind-quist for interesting discussions about photoelectrochemistry.

I would also like to thank David Langhammer for fruitful discussions about IRRAS, metal oxides and for providing useful tips about crystal visu-alisation software. Thanks to Professor Tomas Edvinsson and Jakob Thyr for providing deep insights into Raman spectroscopy. Thanks also to Dr. Annica Nilsson and Professor Arne Roos for their knowledge about optical meas-urements.

I would also like to thank all the past and present members of the solid state division for contributing to a very warm and joyful working environ-ment and all the lunch room guests for providing interesting discussions at the lunch table. Thanks also to Dr. José Montero for sharing funny stories and always being helpful and Dr. Erik Johansson for the occasional lunch session where different aspects of family life and research were pondered. I would also take this opportunity to thank all my friends for making life in Uppsala so pleasant.

Finally I would like to thank my family for all help and support. Tack mamma och pappa för att ni alltid tar er tid och hjälper till med stort som smått. Storasyster med familj för alla trevliga besök och samtal om allt mel-lan himmel och jord. Min underbara fru för allt. Isabella och Anja för att ni alltid får mig att le.

107

11 Bibliography

[1] U.S. Energy Information Administration, International energy outlook 2017, U.S Energy Information Administration, Washington D.C., 2017.

[2] U.S. Department of Energy, 2011 Buildings energy data book, U.S. Department of Energy, Washington D.C., 2011.

[3] P. Wolkoff, C. K. Wilkins, P. A. Clausen and G. D. Nielsen, "Organic compounds in office environments - sensory irritation, odor, measurements and the role of reactive chemistry", Indoor Air, 2006, 16, 7–19.

[4] P. Wolkoff, P. A. Clausen, P. A. Nielsen and L. Molhave, "The danish twin apartment study; part I: Formaldehyde and long-term VOC measurements", Indoor Air, 1991, 1, 478–490.

[5] T. Salthammer and M. Bahadir, "Occurrence, dynamics and reactions of organic pollutants in the indoor environment", Clean - Soil, Air, Water, 2009, 37, 417–435.

[6] T. Salthammer, "Very volatile organic compounds: An understudied class of indoor air pollutants", Indoor Air, 2016, 26, 25–38.

[7] H. E. Dawson and T. McAlary, "A compilation of statistics for VOCs from post-1990 indoor air concentration studies in North American residences unaffected by subsurface vapor intrusion", Ground Water Monitoring and Remediation, 2009, 29, 60–69.

[8] L. Zhang, C. Steinmaus, D. A. Eastmond, X. K. Xin and M. T. Smith, "Formaldehyde exposure and leukemia: A new meta-analysis and potential mechanisms", Mutation Research/Reviews in Mutation Research, 2009, 681, 150–168.

[9] J. Sundell, "On the history of indoor air quality and health", Indoor Air, 2004, 14, 51–58.

[10] G. B. Smith and C. G. Granqvist, Green nanotechnology: Solutions for sustainability and energy in the built environment, CRC Press, Boca Raton, 2011.

[11] K. Straif, A. Cohen and J. Samet, Air polution and cancer, ARC Scientific Publication No 161, International Agency for Research on Cancer, Lyon, 2013.

[12] World health Organization, “9 out of 10 people breathe polluted air, but more countries are taking action, News release 2nd May 2018, WHO, Geneva.

[13] S. Gligorovski and J. P. D. Abbatt, "An indoor chemical cocktail.", Science (New York, N.Y.), 2018, 359, 632–633.

[14] Ø. Svanes, R. J. Bertelsen, S. H. L. Lygre, A. E. Carsin, J. M. Anto, B. Forsberg, J. M. García-García, J. A. Gullón, J. Heinrich, M. Holm, M. Kogevinas, I. Urrutia, B. Leynaert, J. M. Moratalla, N. Le Moual, T. Lytras, D. Norbäck, D. Nowak, M. Olivieri, I. Pin, N. Probst-Hensch, V. Schlünssen, T. Sigsgaard, T. D. Skorge, S. Villani, D. Jarvis, J. P. Zock and C. Svanes, "Cleaning at home and at work in relation to lung function decline and

108

airway obstruction", American Journal of Respiratory and Critical Care Medicine, 2018, 197, 1157–1163.

[15] N. Rahimi, R. A. Pax and E. M. Gray, "Review of functional titanium oxides. I: TiO2 and its modifications", Progress in Solid State Chemistry, 2016, 44, 86–105.

[16] M. Wang, Z. Hou, A. A. Al Kheraif, B. Xing and J. Lin, "Mini review of TiO2-based multifunctional nanocomposites for near-infrared light-responsive phototherapy", Advanced Healthcare Materials, 2018, 1800351.

[17] A. Fujishima, T. N. Rao and D. A. Tryk, "Titanium dioxide photocatalysis", Journal of Photochemistry and Photobiology C: Photochemistry Reviews, 2000, 1, 1–21.

[18] Y. Bai, I. Mora-Seró, F. De Angelis, J. Bisquert and P. Wang, "Titanium dioxide nanomaterials for photovoltaic applications", Chemical Reviews, 2014, 114, 10095–10130.

[19] M. Jacoby, "The future of low-cost solar cells", Chemical & Engineering News, 2016, 94, 30–35.

[20] M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi, E. A. Santori and N. S. Lewis, "Solar water splitting cells", Chemical Reviews, 2010, 110, 6446–6473.

[21] Y. Ma, X. Wang, Y. Jia, X. Chen, H. Han and C. Li, "Titanium dioxide-based nanomaterials for photocatalytic fuel generations", Chemical Reviews, 2014, 114, 9987–10043.

[22] A. Kudo and Y. Miseki, "Heterogeneous photocatalyst materials for water splitting", Chemical Society Reviews, 2009, 38, 253–278.

[23] H. Wang, L. Zhang, Z. Chen, J. Hu, S. Li, Z. Wang, J. Liu and X. Wang, "Semiconductor heterojunction photocatalysts: Design, construction, and photocatalytic performances", Chemical Society Reviews, 2014, 43, 5234–5244.

[24] J. Low, B. Cheng and J. Yu, "Surface modification and enhanced photocatalytic CO2 reduction performance of TiO2: A review", Applied Surface Science, 2017, 392, 658–686.

[25] A. Mills and S. Le Hunte, "An overview of semiconductor photocatalysis", Journal of Photochemistry and Photobiology A: Chemistry, 1997, 108, 1–35.

[26] A. Mills, R. H. Davies and D. Worsley, "Water purification by semiconductor photocatalysis", Chemical Society Reviews, 1993, 22, 417–425.

[27] O. Legrini, E. Oliveros and A. M. Braun, "Photochemical processes for water treatment", Chemical Reviews, 1993, 93, 671–698.

[28] M. R. Hoffmann, S. T. Martin, W. Choi and D. W. Bahnemann, "Environmental applications of semiconductor photocatalysis", Chemical Reviews, 1995, 95, 69–96.

[29] O. Carp, C. L. Huisman and A. Reller, "Photoinduced reactivity of titanium oxide", Progress in Solid State Chemistry, 2004, 32, 33–177.

[30] J. Peral, X. Domènech and D. F. Ollis, "Heterogeneous photocatalysis for purification, decontamination and deodorization of air", Journal of Chemical Technology and Biotechnology, 1997, 70, 117–140.

[31] A. T. Hodgson, D. P. Sullivan and W. J. Fisk, Evaluation of Ultra-Violet Photocatalytic Oxidation (UVPCO) for Indoor Air Applications: Conversion of Volatile Organic Compounds at Low Part-per-Billion Concentrations, Lawrence Berkely National Laboratory, Berkeley, 2005.

[32] A. Fujishima, X. Zhang and D. A. Tryk, "TiO2 Photocatalysis and related surface phenomena", Surface Science Reports, 2008, 63, 515–582.

109

[33] M. A. Henderson, "A surface science perspective on TiO2 photocatalysis", Surface Science Reports, 2011, 66, 185–297.

[34] U. Diebold, "The surface science of titanium dioxide", Surface Science Reports, 2003, 48, 53–229.

[35] A. L. Linsebigler, G. Q. Lu and J. T. J. Yates, "Photocatalysis on TiO2 surfaces - Principles, mechanisms, and selected results", Chemical Reviews, 1995, 95, 735–758.

[36] J. Balajka, M. A. Hines, W. J. I. DeBenedetti, M. Komora, J. Pavelec, M. Schmid and U. Diebold, "High-affinity adsorption leads to molecularly ordered interfaces on TiO2 in air and solution", Science, 2018, 361, 786–789.

[37] A. Di Paola, E. García-López, G. Marcí and L. Palmisano, "A survey of photocatalytic materials for environmental remediation", Journal of Hazardous Materials, 2012, 211–212, 3–29.

[38] F. Fresno, R. Portela, S. Suárez and J. M. Coronado, "Photocatalytic materials: recent achievements and near future trends", Journal of Materials Chemistry A, 2014, 2, 2863–2884.

[39] A. Mattsson and L. Österlund, "Adsorption and photoinduced decomposition of acetone and acetic acid on anatase, brookite, and rutile TiO2 nanoparticles", Journal of Physical Chemistry C, 2010, 114, 14121–14132.

[40] Y. Cui, H. Du and L.-S. Wen, "Doped-TiO2 Photocatalysts and synthesis methods to prepare TiO2 films", Journal of Materials Science and Technology, 2008, 24, 675–689.

[41] S. H. Szczepankiewicz, J. A. Moss and M. R. Hoffmann, "Slow surface charge trapping kinetics on irradiated TiO2", The Journal of Physical Chemistry B, 2002, 106, 2922–2927.

[42] T. Berger, M. Sterrer, O. Diwald, E. Knözinger, D. Panayotov, T. L. Thompson and J. T. Yates, "Light-induced charge separation in anatase TiO2 particles", The Journal of Physical Chemistry B, 2005, 109, 6061–6068.

[43] D. Wang, L. Zhao, D. Wang, L. Yan, C. Jing, H. Zhang, L.-H. Guo and N. Tang, "Direct evidence for surface long-lived superoxide radicals photo-generated in TiO2 and other metal oxide suspensions", Physical Chemistry Chemical Physics, 2018, 20, 18978–18985.

[44] A. V. Emeline, V. N. Kuznetsov, V. K. Ryabchuk and N. Serpone, "On the way to the creation of next generation photoactive materials", Environmental Science and Pollution Research, 2012, 19, 3666–3675.

[45] M. Grätzel, "Photoelectrochemical cells", Nature, 2001, 414, 338–344. [46] A. Fujishima, K. Hashimoto and T. Watanabe, TiO2 photocatalysis:

Fundamentals and applications, BKC, Tokyo, 1999. [47] V. Augugliaro, V. Loddo, G. Palmisano and L. Palmisano, Clean by light

irradation: Practical applications of supported TiO2, Royal Society of Chemistry, 2010.

[48] M. Takeuchi, K. Sakamoto, G. Martra, S. Coluccia and M. Anpo, "Mechanism of photoinduced superhydrophilicity on the TiO2 photocatalyst surface", The Journal of Physical Chemistry B, 2005, 109, 15422–15428.

[49] J. T. Yates, "Photochemistry on TiO2: Mechanisms behind the surface chemistry", Surface Science, 2009, 603, 1605–1612.

[50] Pilkington, https://www.pilkington.com/en-gb/uk/householders/types-of-glass/self-cleaning-glass, (accessed 6 September 2018).

[51] E. Boonen and A. Beeldens, "Photocatalytic roads: From lab tests to real scale applications", European Transport Research Review, 2013, 5, 79–89.

110

[52] T. Maggos, A. Plassais, J. G. Bartzis, C. Vasilakos, N. Moussiopoulos and L. Bonafous, "Photocatalytic degradation of NOx in a pilot street canyon configuration using TiO2-mortar panels", Environmental Monitoring and Assessment, 2008, 136, 35–44.

[53] D. H. Hanaor and C. Sorrell, "Review of the anatase to rutile phase transformation", Journal of Materials Science, 2011, 46, 855–874.

[54] H. Sato, S. Endo, M. Sugiyama, T. Kikegawa, O. Shimomura and K. Kusaba, "Baddeleyite-type high-pressure phase of TiO2", Science, 1991, 251, 786–788.

[55] L. Gerward and J. Staun Olsen, "Post-rutile high-pressure phases in TiO2", Journal of Applied Crystallography, 1997, 30, 259–264.

[56] A. Navrotsky, J. C. Jamieson and O. J. Kleppa, "Enthalpy of transformation of a high-pressure polymorph of titanium dioxide to the rutile modification", Science, 1967, 158, 388–389.

[57] J. C. Jamieson and B. Olinger, "Pressure-temperature studies of anatase, brookite, rutile, and TiO2(II): A discussion", The American Mineralogist, 1969, 54, 1477–1481.

[58] H. Zhang and J. F. Banfield, "Understanding polymorphic phase transformation behavior during growth of nanocrystalline aggregates: Insigths from TiO2", Journal of Physical Chemistry B, 2000, 104, 3481–3487.

[59] K.-R. Zhu, M.-S. Zhang, J.-M. Hong and Z. Yin, "Size effect on phase transition sequence of TiO2 nanocrystal", Materials Science and Engineering A, 2005, 403, 87–93.

[60] A. Mattsson, C. Lejon, V. Štengl, S. Bakardjieva, F. Opluštil, P. O. Andersson and L. Österlund, "Photodegradation of DMMP and CEES on zirconium doped titania nanoparticles", Applied Catalysis B: Environmental, 2009, 92, 401–410.

[61] R. E. Tanner, Y. Liang and E. I. Altman, "Structure and chemical reactivity of adsorbed carboxylic acids on anatase TiO2 (001)", Surface Science, 2002, 506, 251–271.

[62] C. Lejon and L. Österlund, "Influence of quantum confinement, surface stress, and zirconium doping on the Raman vibrational properties of anatase TiO2 nanoparticles", Journal of Raman Spectroscopy, 2011, 42, 2026–2035.

[63] T. Ohsaka, F. Izumi and Y. Fujiki, "Raman spectrum of anatase, TiO2", Journal of Raman Spectroscopy, 1978, 7, 321–324.

[64] K. Momma and F. Izumi, "VESTA 3 for three-dimensional visualization of crystal, volumetric and morphology data", Journal of Applied Crystallography, 2011, 44, 1272–1276.

[65] R. T. Downs and M. Hall-Wallace, "The American Mineralogist crystal structure database", American Mineralogist, 2003, 88, 247–250.

[66] J. Schneider, M. Matsuoka, M. Takeuchi, J. Zhang, Y. Horiuchi, M. Anpo and D. W. Bahnemann, "Understanding TiO2 photocatalysis : Mechanisms and materials", Chem. Rev., 2014, 114, 9919–9986.

[67] L. Österlund, in Solid State Chemistry and Photocatalysis of Titanium Dioxide, eds. M. K. Nowotny and J. Nowotny, Trans Tech Publications, Stafa-Zürich, Switzerland, 2010, pp. 203–219.

[68] L. Pan, J. J. Zou, S. Wang, X. Y. Liu, X. Zhang and L. Wang, "Morphology evolution of TiO2 facets and vital influences on photocatalytic activity", ACS Applied Materials and Interfaces, 2012, 4, 1650–1655.

111

[69] Y. Wang and C. Wöll, "IR spectroscopic investigations of chemical and photochemical reactions on metal oxides: bridging the materials gap", Chemical Society Reviews., 2017, 46, 1875–1932.

[70] M. Xu, Y. Gao, E. M. Moreno, M. Kunst, M. Muhler, Y. Wang, H. Idriss and C. Wöll, "Photocatalytic activity of bulk TiO2 anatase and rutile single crystals using infrared absorption spectroscopy", Physical Review Letters, 2011, 106, 138302.

[71] T. Ohsawa, I. V. Lyubinetsky, M. A. Henderson and S. A. Chambers, "Hole-mediated photodecomposition of trimethyl acetate on a TiO2 (001) anatase epitaxial thin film surface", Journal of Physical Chemistry C, 2008, 112, 20050–20056.

[72] T. Ohno, K. Sarukawa and M. Matsumura, "Crystal faces of rutile and anatase TiO2 particles and their roles in photocatalytic reactions", New Journal of Chemistry, 2002, 26, 1167–1170.

[73] N. Roy, Y. Sohn and D. Pradhan, "Synergy of low-energy {101} and high-energy {001} TiO2 crystal facets for enhanced photocatalysis", ACS Nano, 2013, 7, 2532–2540.

[74] Z. Zhao, Z. Li and Z. Zou, "Surface properties and electronic structure of low-index stoichiometric anatase TiO2 surfaces", Journal of Physics: Condensed Matter, 2010, 22, 175008.

[75] X. Ma, Y. Dai, M. Guo and B. Huang, "Relative Photooxidation and photoreduction activities of the {100}, {101}, and {001} surfaces of anatase TiO2", Langmuir, 2013, 29, 13647–13654.

[76] C.-L. Pang, R. Lindsay and G. Thornton, "Structure of clean and adsorbate-covered single crystal rutile TiO2 surfaces", Chemical Reviews, 2013, 113, 3887–3948.

[77] Y. W. Chung, W. J. Lo and G. A. Somorjai, "Low energy electron diffraction and electron spectroscopy studies of the clean (110) and (100) titanium dioxide (rutile) crystal surfaces", Surface Science, 1977, 64, 588–602.

[78] B. J. Morgan and G. W. Watson, "A DFT + U description of oxygen vacancies at the TiO2 rutile (110) surface", Surface Science, 2007, 601, 5034–5041.

[79] Q. Cuan, J. Tao, X. Q. Gong and M. Batzill, "Adsorbate induced restructuring of TiO2 (011)-(2×1) leads to one-dimensional nanocluster formation", Physical Review Letters, 2012, 108, 1–5.

[80] H. Takahashi, R. Watanabe, Y. Miyauchi and G. Mizutani, "Discovery of deep and shallow trap states from step structures of rutile TiO2 vicinal surfaces by second harmonic and sum frequency generation spectroscopy", Journal of Chemical Physics, 2011, 134, 154704.

[81] Y. Yamamoto, K. Nakajima, T. Ohsawa, Y. Matsumoto and H. Koinuma, "Preparation of atomically smooth TiO2 single crystal surfaces and their photochemical property", Japanese Journal of Applied Physics, Part 2: Letters, 2005, 44, L511–L514.

[82] X. Mao, Z. Wang, X. Lang, Q. Hao, B. Wen, D. Dai, C. Zhou, L. M. Liu and X. Yang, "Effect of surface structure on the photoreactivity of TiO2", Journal of Physical Chemistry C, 2015, 119, 6121–6127.

[83] X. Torrelles, G. Cabailh, R. Lindsay, O. Bikondoa, J. Roy, J. Zegenhagen, G. Teobaldi, W. A. Hofer and G. Thornton, "Geometric structure of TiO2 (011) (2×1)", Physical Review Letters, 2008, 101, 185501.

112

[84] Y. Yamaguchi, H. Onishi and Y. Iwasawa, "Catalytic decomposition Reaction of formic acid on an Ar+-bombarded TiO2 (110) surface: Steady-state kinetics and microscopic surface structure", Journal of the Chemical Society: Faraday Transactions, 1995, 91, 1663–1668.

[85] Z. Dohnalek, I. Lyubinetsky and R. Rousseau, "Thermally-driven processes on rutile TiO2 (110)-(1x1): A direct view at the atomic scale", Progress in Surface Science, 2010, 85, 161–205.

[86] N. G. Petrik and G. A. Kimmel, "Reaction kinetics of water molecules with oxygen vacancies on rutile TiO2 (110)", Journal of Physical Chemistry C, 2015, 119, 23059–23067.

[87] S. Wendt, R. Schaub, J. Matthiesen, E. K. Vestergaard, E. Wahlström, M. D. Rasmussen, P. Thostrup, L. M. Molina, E. Laegsgaard, I. Stensgaard, B. Hammer and F. Besenbacher, "Oxygen vacancies on TiO2 (110) and their interaction with H2O and O2: A combined high-resolution STM and DFT study", Surface Science, 2005, 598, 226–245.

[88] S. Fischer, A. W. Munz, K.-D. Schierbaum and W. Göpel, "The geometric structure of intrinsic defects at TiO2 (110) surfaces: An STM study", Surface Science, 1995, 337, 17–30.

[89] X. Cui, B. Wang, Z. Wang, T. Huang, Y. Zhao, J. Yang and J. G. Hou, "Formation and diffusion of oxygen-vacancy pairs on TiO2 (110)-(1x1)", The Journal of Chemical Physics, 2008, 129, 44703.

[90] X. Q. Gong and A. Selloni, "First-principles study of the structures and energetics of stoichiometric brookite TiO2 surfaces", Physical Review B - Condensed Matter and Materials Physics, 2007, 76, 1–11.

[91] T. R. Esch, I. Gadaczek and T. Bredow, "Surface structures and thermodynamics of low-index of rutile, brookite and anatase - A comparative DFT study", Applied Surface Science, 2014, 288, 275–287.

[92] U. Diebold, N. Ruzycki, G. S. Herman and A. Selloni, "One step towards bridging the materials gap: Surface studies of TiO2 anatase", Catalysis Today, 2003, 85, 93–100.

[93] A. G. Thomas, W. R. Flavell, A. K. Mallick, A. R. Kumarasinghe, D. Tsoutsou, N. Khan, C. Chatwin, S. Rayner, G. C. Smith, R. L. Stockbauer, S. Warren, T. K. Johal, S. Patel, D. Holland, A. Taleb and F. Wiame, "Comparison of the electronic structure of anatase and rutile TiO2 single-crystal surfaces using resonant photoemission and x-ray absorption spectroscopy", Physical Review B - Condensed Matter and Materials Physics, 2007, 75, 1–12.

[94] Y. He, A. Tilocca, O. Dulub, A. Selloni and U. Diebold, "Local ordering and electronic signatures of submonolayer water on anatase TiO2 (101)", Nature Materials, 2009, 8, 585–589.

[95] P. Scheiber, M. Fidler, O. Dulub, M. Schmid, U. Diebold, W. Hou, U. Aschauer and A. Selloni, "(Sub)Surface mobility of oxygen vacancies at the TiO2 anatase (101) surface", Physical Review Letters, 2012, 109, 1–5.

[96] D. T. Payne, Y. Zhang, C. L. Pang, H. H. Fielding and G. Thornton, "Creating excess electrons at the anatase TiO2 (101) surface", Topics in Catalysis, 2017, 60, 392–400.

[97] K. Bourikas, C. Kordulis and A. Lycourghiotis, "Titanium dioxide (anatase and rutile): Surface chemistry, liquid-solid interface chemistry, and scientific synthesis of supported catalysts", Chemical Reviews, 2014, 114, 9754–9823.

[98] G. S. Herman and Y. Gao, "Growth of epitaxial anatase (001) and (101) films", Thin Solid Films, 2001, 397, 157–161.

113

[99] R. E. Tanner, A. Sasahara, Y. Liang, E. I. Altman and H. Onishi, "Formic acid adsorption on anatase TiO2 (001)-(1×4) thin films studied by NC-AFM and STM", Journal of Physical Chemistry B, 2002, 106, 8211–8222.

[100] D. Matthey, J. G. Wang, S. Wendt, J. Matthiesen, R. Schaub, E. Laegsgaard, B. Hammer and F. Besenbacher, "Enhanced bonding of gold nanoparticles on oxidized TiO2 (110)", Science, 2007, 315, 1692–1696.

[101] A. Mattsson, S. Hu, K. Hermansson and L. Österlund, "Infrared spectroscopy study of adsorption and photodecomposition of formic acid on reduced and defective rutile TiO2 (110) surfaces", Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films, 2014, 32, 061402.

[102] Y. Du, Z. Dohnálek and I. Lyubinetsky, "Transient mobility of oxygen adatoms upon O2 dissociation on reduced TiO2 (110)", Journal of Physical Chemistry C, 2008, 112, 2649–2653.

[103] L.-M. Liu, P. Crawford and P. Hu, "The interaction between adsorbed OH and O2 on TiO2 surfaces", Progress in Surface Science, 2009, 84, 155–176.

[104] L. M. Liu, B. McAllister, H. Q. Ye and P. Hu, "Identifying an O2 supply pathway in CO oxidation on Au/TiO2 (110): A density functional theory study on the intrinsic role of water", Journal of the American Chemical Society, 2006, 2, 4017–4022.

[105] Z. Zhang, Y. Du, N. G. Petrik, G. A. Kimmel, I. Lyubinetsky and Z. Dohnálek, "Water as a catalyst: Imaging reactions of O2 with partially and fully hydroxylated TiO2 (110) surfaces", Journal of Physical Chemistry C, 2009, 113, 1908–1916.

[106] Y. Du, N. A. Deskins, Z. Zhang, Z. Dohnálek, M. Dupuis and I. Lyubinetsky, "Imaging consecutive steps of O2 reaction with hydroxylated TiO2 (110): Identification of HO2 and terminal OH intermediates", Journal of Physical Chemistry C, 2009, 113, 666–671.

[107] M. A. Henderson, W. S. Epling, C. H. F. Peden and C. L. Perkins, "Insights into photoexcited electron scavenging processes on TiO2 obtained from studies of the reaction of O2 with OH groups adsorbed at electronic defects on TiO2 (110)", Journal of Physical Chemistry B, 2003, 107, 534–545.

[108] A. C. Papageorgiou, N. S. Beglitis, C. L. Pang, G. Teobaldi, G. Cabailh, Q. Chen, A. J. Fisher, W. A. Hofer and G. Thornton, "Electron traps and their effect on the surface chemistry of TiO2 (110).", Proceedings of the National Academy of Sciences of the United States of America, 2010, 107, 2391–6.

[109] W. S. Epling, C. H. F. Peden, M. A. Henderson and U. Diebold, "Evidence for oxygen adatoms on TiO2 (110) resulting from O2 dissociation at vacancy sites", Surface Science, 1998, 412–413, 333–343.

[110] T. Cremer, S. C. Jensen and C. M. Friend, "Enhanced photo-oxidation of formaldehyde on highly reduced o-TiO2(110)", The Journal of Physical Chemistry C, 2014, 118, 29242–29251.

[111] C. L. Pang, R. Lindsay and G. Thornton, "Chemical reactions on rutile TiO2 (110).", Chemical Society reviews, 2008, 37, 2328–2353.

[112] J. M. Hollas, Modern Spectroscopy, John Wiley & Sons Ltd, Chichester, 3rd Ed., 1996.

[113] D. Willock, Molecular Symmetry, Wiley, Hoboken, NJ, USA, 2009. [114] P. M. Morse, "Diatomic molecules according to the wave mechanics. II.

Vibrational levels", Physical Review, 1929, 34, 57–64. [115] J. C. Maxwell, "A dynamical theory of the electromagnetic field",

Philosophical Transactions of the Royal Society of London, 1865, 155, 459–512.

114

[116] F. Wooten, Optical properties of solids, Academic Press Inc, New York, 1st edn., 1972.

[117] S. A. Francis and A. H. Elllison, "Infrared Spectra of monolayers on metal mirrors", Journal of the Optical Society of America, 1959, 49, 131–138.

[118] R. G. Greenler, "Infrared study of adsorbed molecules on metal surfaces by reflection techniques", The Journal of Chemical Physics, 1966, 44, 310–315.

[119] W. N. Hansen, "Electric fields produced by the propagation of plane coherent electromagnetic radiation in a stratified medium", Journal of the Optical Society of America, 1968, 58, 380–390.

[120] W. N. Hansen, "Reflection spectroscopy of adsorbed layers", Symposia of the Faraday Society, 1970, 4, 27–35.

[121] M. Xu, Y. Gao, Y. Wang and C. Wöll, "Monitoring electronic structure changes of TiO2 (110) via sign reversal of adsorbate vibrational bands", Physical Chemistry Chemical Physics, 2010, 12, 3649–3652.

[122] D. C. Cronemeyer, "Electrical and optical properties of rutile single crystals", Physical Review, 1952, 87, 876–886.

[123] E. M. Assim, "Optical constants of TiO1.7 thin films deposited by electron beam gun", Journal of Alloys and Compounds, 2008, 463, 55–61.

[124] Y. J. Chabal, "Surface infrared spectroscopy", Surface Science Reports, 1988, 8, 211–357.

[125] J. D. E. McIntyre and D. E. Aspnes, "Differential reflection spectroscopy of very thin surface films", Surface Science, 1971, 24, 417–433.

[126] S. Baroni and R. Resta, "Ab initio calculation of the macroscopic dielectric constant in silicon", Physical Review B, 1986, 33, 7017–7021.

[127] M. R. Pederson, T. Baruah, P. B. Allen and C. Schmidt, "Density-functional-based determination of vibrational polarizabilities in molecules within the double-harmonic approximation: Derivation and application", Journal of Chemical Theory and Computation, 2005, 1, 590–596.

[128] D. Karhanek, T. Bucko and J. Hafner, "A density-functional study of the adsorption of methane-thiol on the (111) surfaces of the Ni-group metals: II. Vibrational spectroscopy", Journal of Physics: Condensed Matter, 2010, 22, 265006.

[129] M. Gajdoš, K. Hummer, G. Kresse, J. Furthmüller and F. Bechstedt, "Linear optical properties in the projector-augmented wave methodology", Physical Review B, 2006, 73, 45112.

[130] A. Mattsson, S. Hu, K. Hermansson and L. Österlund, "Adsorption of formic acid on rutile TiO2 (110) revisited: An infrared reflection-absorption spectroscopy and density functional theory study", Journal of Chemical Physics, 2014, 140, 34705.

[131] The scientific papers of sir William Herschel vol II, The Royal Society and the Royal Astonomical Society, London, 1912.

[132] P. H. Hindle, in Handbook of near-infrared analysis, eds. D. A. Burns and E. W. Ciurczak, CRC Press, Boca Raton, 3rd edn., 2007.

[133] W. F. Meggers, William Weber Coblentz 1873—1962: A biographical memoir, National Academy of Sciences, Washington D. C., 1967.

[134] W. D. Lawson, S. Nielsen, E. H. Putley and A. S. Young, "Preparation and properties of HgTe and mixed crystals of HgTe-CdTe", Journal of Physics and Chemistry of Solids, 1959, 9, 325–329.

[135] A. Dazzi and C. B. Prater, "AFM-IR: Technology and applications in nanoscale infrared spectroscopy and chemical imaging", Chemical Reviews, 2017, 117, 5146–5173.

115

[136] W. D. Perkins, "Fourier transform-infrared spectroscopy: Part l. Instrumentation", Journal of Chemical Education, 1986, 63, A5.

[137] W. D. Perkins, "Fourier transform infrared spectroscopy. Part II. Advantages of FT-IR", Journal of Chemical Education, 1987, 64, A269.

[138] J. B. Pendry, Low energy electron diffraction, Academic Press Inc, London, 1974.

[139] N. Begum, A. S. Bhatti, F. Jabeen, S. Rubini and F. Martelli, in Nanowires, ed. P. Prete, IntechOpen, London, 2010, pp. 189–230.

[140] G. Renaud, R. Lazzari and F. Leroy, "Probing surface and interface morphology with grazing incidence small angle X-ray scattering", Surface Science Reports, 2009, 64, 255–380.

[141] A. Mattsson, M. Leideborg, K. Larsson, G. Westin and L. Österlund, "Adsorption and solar light decomposition of acetone on anatase TiO2 and niobium doped TiO2 thin films", Journal of Physical Chemistry B, 2006, 110, 1210–1220.

[142] A. Mattsson, M. Leideborg, L. Persson, G. Westin and L. Österlund, "Oxygen diffusion and photon-induced decomposition of acetone on Zr- and Nb-doped TiO2 nanoparticles", Journal of Physical Chemistry C, 2009, 113, 3810–3818.

[143] A. Linsebigler, G. Lu and J. T. Yates, "CO chemisorption on TiO2 (110): Oxygen vacancy site influence on CO adsorption", The Journal of Chemical Physics, 1995, 103, 9438–9443.

[144] Z. Dohnálek, J. Kim, O. Bondarchuk, J. Mike White and B. D. Kay, "Physisorption of N2, O2, and CO on fully oxidized TiO2 (110)", Journal of Physical Chemistry B, 2006, 110, 6229–6235.

[145] N. G. Petrik and G. A. Kimmel, "Adsorption geometry of CO versus Coverage on TiO2 (110) from s- and p-polarized infrared spectroscopy", The Journal of Physical Chemistry Letters, 2012, 3, 3425–3430.

[146] M. Kunat, F. Traeger, D. Silber, H. Qiu, Y. Wang, A. C. van Veen, C. Wöll, P. M. Kowalski, B. Meyer, C. Hättig and D. Marx, "Formation of weakly bound, ordered adlayers of CO on rutile TiO2 (110): A combined experimental and theoretical study", The Journal of Chemical Physics, 2009, 130, 144703.

[147] C. Rohmann, Y. Wang, M. Muhler, J. Metson, H. Idriss and C. Wöll, "Direct monitoring of photo-induced reactions on well-defined metal oxide surfaces using vibrational spectroscopy", Chemical Physics Letters, 2008, 460, 10–12.

[148] S. Hu, Z. Wang, A. Mattsson, L. Österlund and K. Hermansson, "Simulation of IRRAS spectra for molecules on oxide surfaces: CO on TiO2 (110)", Journal of Physical Chemistry C, 2015, 119, 5403–5411.

[149] R. Lindsay, S. Tomić, A. Wander, M. García-Méndez, G. Thornton, S. Tomić, A. Wander, M. García-Méndez and G. Thornton, "Low energy electron diffraction study of TiO2 (110)(2x1)-[HCOO]-", The Journal of Physical Chemistry C, 2008, 112, 14154–14157.

[150] B. E. Hayden, A. King and M. A. Newton, "Fourier transform reflection-absorption IR spectroscopy study of formate adsorption on TiO2 (110)", Journal of Physical Chemistry B, 1999, 103, 203–208.

[151] D. I. Sayago, M. Polcik, R. Lindsay, R. L. Toomes, J. T. Hoeft, M. Kittel and D. P. Woodruff, "Structure determination of formic acid reaction products on TiO2 (110)", The Journal of Physical Chemistry B, 2004, 108, 14316–14323.

116

[152] M. Aizawa, Y. Morikawa, Y. Namai, H. Morikawa and Y. Iwasawa, "Oxygen vacancy promoting catalytic dehydration of formic acid on TiO2 (110) by in situ scanning tunneling microscopic observation", Journal of Physical Chemistry B, 2005, 109, 18831–18838.

[153] F. P. Rotzinger, J. M. Kesselman-Truttmann, S. J. Hug, V. Shklover and M. Grätzel, "Structure and vibrational spectrum of formate and acetate adsorbed from aqueous solution onto the TiO2 rutile (110) surface", Journal of Physical Chemistry B, 2004, 108, 5004–5017.

[154] A. Mattsson and L. Österlund, "Co-adsorption of oxygen and formic acid on rutile TiO2 (110) studied by infrared reflection-absorption spectroscopy", Surface Science, 2017, 663, 47–55.

[155] S. Wendt, P. T. Sprunger, E. Lira, G. K. H. Madsen, Z. Li, J. O. Hansen, J. Matthiesen, A. Blekinge-Rasmussen, E. Laegsgaard, B. Hammer and F. Besenbacher, "The role of interstitial sites in the Ti3d defect state in the band gap of titania.", Science, 2008, 320, 1755–1759.

[156] H. Onishi, T. Aruga and Y. Iwasawa, "Switchover of reaction paths in the catalytic decomposition of formic acid on TiO2 (110) surface", Journal of Catalysis, 1994, 146, 557–567.

[157] L.-Q. Wang, K. F. F. Ferris, A. N. N. Shultz, D. R. R. Baer and M. H. H. Engelhard, "Interactions of HCOOH with stoichiometric and defective TiO2 (110) surfaces", Surface Science, 1997, 380, 352–364.

[158] H. Uetsuka, M. A. Henderson, A. Sasahara and H. Onishi, "Formate adsorption on the (111) surface of rutile TiO2", Journal of Physical Chemistry B, 2004, 108, 13706–13710.

[159] L. Österlund, in Solar Hydrogen and Nanotechnology, ed. L. Vayssieres, Wiley, Singapore, 2010, pp. 189–230.

[160] D. S. Muggli and J. L. Falconer, "Role of lattice oxygen in photocatalytic oxidation on TiO2", Journal of Catalysis, 2000, 191, 318–325.

[161] D. S. Muggli and J. L. Falconer, "Parallel pathways for photocatalytic decomposition of acetic acid on TiO2", Journal of Catalysis, 1999, 187, 230–237.

[162] M. Xu, H. Noei, M. Buchholz, M. Muhler, C. Wöll and Y. Wang, "Dissociation of formic acid on anatase TiO2 (101) probed by vibrational spectroscopy", Catalysis Today, 2012, 182, 12–15.

[163] M. Setvin, M. Buchholz, W. Hou, C. Zhang, B. Stöger, J. Hulva, T. Simschitz, X. Shi, J. Pavelec, G. S. Parkinson, M. Xu, Y. Wang, M. Schmid, C. Wöll, A. Selloni and U. Diebold, "A multitechnique study of CO adsorption on the TiO2 anatase (101) surface", Journal of Physical Chemistry C, 2015, 119, 21044–21052.

[164] U. Diebold, N. Ruzycki, G. S. Herman and A. Selloni, "One step towards bridging the materials gap: surface stuides of TiO2 anatase", Catalysis Today, 2003, 85, 93–100.

[165] G. Rupprechter, "Sum frequency generation and polarization-modulation infrared reflection absorption spectroscopy of functioning model catalysts from ultrahigh vacuum to ambient pressure", Advances in Catalysis, 2007, 51, 133–263.

[166] B. L. Frey, R. M. Corn and S. C. Weibel, in Handbook of vibrational spectroscopy, eds. J. Chalmers and P. R. Griffiths, John Wiley & Sons, 2002, vol. 2, pp. 1042–1056.

117

[167] C. Hess, E. Ozensoy, C. W. Yi and D. W. Goodman, "NO dimer and dinitrosyl formation on Pd (111): From ultra-high-vacuum to elevated pressure conditions", Journal of the American Chemical Society, 2006, 128, 2988–2994.

[168] M. A. Ramin, G. Le Bourdon, N. Daugey, B. Bennetau, L. Vellutini and T. Buffeteau, "PM-IRRAS investigation of self-assembled monolayers grafted onto SiO2/Au substrates", Langmuir, 2011, 27, 6076–6084.

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1726

Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science andTechnology, Uppsala University, is usually a summary of anumber of papers. A few copies of the complete dissertationare kept at major Swedish research libraries, while thesummary alone is distributed internationally throughthe series Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology.(Prior to January, 2005, the series was published under thetitle “Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology”.)

Distribution: publications.uu.seurn:nbn:se:uu:diva-362326

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2018