ACTAUNIVERSITATISUPSALIENSISUPPSALA2006

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 188

Interfaces in Dye-Sensitized Oxide /Hole-Conductor Heterojunctionsfor Solar Cell Applications

ERIK JOHANSSON

ISSN 1651-6214ISBN 91-554-6575-7urn:nbn:se:uu:diva-6892

O, hur härligt majsol ler, o, hur härligt majsol ler. I de blåa höjder, sjungas vårens fröjder, och sin doft var blomma ger. O, hur härligt majsol ler, o, hur härligt majsol ler, o, hur härligt, härligt majsol ler.

F.Kuhlau

List of Papers

I. Electronic and Molecular Surface Structure of Ru(tcterpy)(NCS)3 and Ru(dcbpy)2(NCS)2 Adsorbed from Solution onto Nanostructured TiO2 – a Photoelectron Spectroscopy Study, E. M. J. Johansson, M. Hedlund, H. Siegbahn, H. Rensmo, J. Phys. Chem. B, 109 (47) (2005) 22256-22263

II. Frontier Electronic Structure of Ru(tcterpy)(NCS)3 and Ru(dcbpy)2(NCS)2 - a Photoelectron Spectroscopy Study, E. M. J. Jo-hansson, M. Hedlund, M. Odelius, H. Siegbahn, H. Rensmo, in manuscript

III. Electronic and Molecular Surface Structure of a Polyene-diphenylaniline Dye Adsorbed from Solution onto Nanoporous TiO2,E.M.J Johansson, T. Edvinsson, M. Odelius, D.P. Hagberg, L. Sun, A. Hag-feldt, H. Siegbahn and H. Rensmo, in manuscript

IV. Photovoltaic and Interfacial Properties of Heterojunctions Contain-ing Dye-sensitized Dense TiO2 and Triarylamine derivatives, E.M.J. Jo-hansson, P.G. Karlsson, M. Hedlund, D. Ryan, H. Siegbahn and H. Rensmo, in manuscript

V. Interfacial Properties of Photovoltaic TiO2/dye/PEDOT-PSS Hetero-junctions, E. M. J. Johansson, A. Sandell, H. Siegbahn, H. Rensmo, B. Mahrov, G. Boschloo, E. Figgemeier, A. Hagfeldt, S. K. M. Jönsson, M. Fahlman, Synth. Met. 149 (2005) 157-167

VI. Interfacial properties of the nanostructured dye-sensitized solid het-erojunction TiO2/RuL2(NCS)2/CuI, P.G. Karlsson, S. Bolik, J. H. Richter, B. Mahrov, E.M.J. Johansson, J. Blomquist, P. Uvdal, H. Rensmo, H. Sieg-bahn, A. Sandell, J. Chem. Phys. 120(23) (2004) 11224-11232

VII. Electronic structure and mechanism for conductivity in thiophene oligomers and regioregular polymer, E. Johansson, S. Larsson, Synth. Met. 144 (2004) 183-191

Comments on my own participation

The papers presented here are based on teamwork. I had the main re-sponsibility for experiments, data analysis and manuscripts in paper I-V. In paper VI I took part in the experimental work and participated in the evaluation of the data. In paper VII I had the main responsibility for the calculations and the manuscript.

Contents

Introduction.....................................................................................................9Solar energy................................................................................................9The solar cell ..............................................................................................9The present thesis .....................................................................................11

1. The nanoporous dye-sensitized solar cell .................................................121.1. The function ......................................................................................121.3. Efficiency limitations ........................................................................15

2. Techniques ................................................................................................172.1. Electron spectroscopy techniques .....................................................17

2.1.1. Photoelectron Spectroscopy.......................................................172.1.2. X-ray absorption spectroscopy ..................................................202.1.3. Resonant spectroscopy...............................................................212.1.4. Experimental setup ....................................................................23

2.2. Photovoltaic measurements...............................................................252.2.1. IPCE measurements...................................................................252.2.2. I-V measurements......................................................................26

2.3. Theoretical methods ..........................................................................262.3.1. Hückel method...........................................................................262.3.2. Hartree-Fock equations and Density functional theory .............26

2.4. Sample preparations ..........................................................................272.4.1 Preparation of TiO2 electrodes....................................................272.4.2 Dye-sensitization ........................................................................272.4.3 Hole-conductor preparations ......................................................28

3. Summary of the results .............................................................................293.1. Dyes and TiO2/dye interfaces............................................................303.3. TiO2 / dye / hole-conductor interfaces ...............................................373.2. Polythiophene and oligothiophene hole-conductors .........................42

5. Outlook .....................................................................................................45

4. Populärvetenskaplig sammanfattning .......................................................464.1. Inledning............................................................................................464.2. Den nanoporösa infärgade solcellen..................................................474.3. Experiment och resultat.....................................................................48

5. Acknowledgements...................................................................................49

6. References.................................................................................................50

9

Introduction

Solar energy Today, as much as 80% of all energy consumed worldwide comes from fos-sil fuels and about 10% from nuclear power [1]. Considering the large envi-ronmental impact, such as global warming and acidic rain of fossil fuels, the problems of nuclear waste storage for a few hundred thousands years, danger of nuclear accidents and increased risk for nuclear weapon production, re-newable energy sources such as solar-, bio-, wind- and waterpower are pref-erable. Most of the solar energy used today utilizes thermal conversion (e.g. solar collectors). The growing use of electricity and also the possibility to easily transport the electrical energy makes the direct conversion from solar energy to electrical energy (solar cells) advantageous for many applications and gives potential for increase of the solar cell usage.

To get an idea of the area of solar cells that is needed to cover the energy demand, we make a small example. The solar energy that shines on one square meter in Sweden is about 1000 kWh/year. The total amount of elec-trical energy used for electrical installations in households (e.g. refrigerator, lamps, oven ) is 19.5 TWh in Sweden [2]. The need in households is there-fore about 2000 kWh/year/person, which for example could be supplied in a 10 m2 area of solar cells with 20% efficiency. This small calculation shows that it is practically possible to cover the electrical energy demand in house-holds with solar cells.

One large energy demand is cooling. This seems to be the perfect use of solar cells, since they work best during the warmest time of the day and of the year. The energy demand is usually highest during the day, which makes solar cells better to cover the energy peak demand than for example nuclear power [3]. In some countries the daytime is short in the winter, in these cases there may be a need for energy storage, such as batteries.

The solar cell The solar cell may be described as a device that works as a battery as long as it is kept under illumination. Edmund Bequerel is usually credited as the first to demonstrate the photovoltaic effect in 1839, when a small photocreated voltage was measured [4,5]. The American inventor Charles Fritts fabricated

10

a solar cell in 1883 based on selenium [6]. The efficiency was below 1% but this type of solar cells was used as light sensor for cameras. Researchers at the Bell Laboratories made a breakthrough in 1954 when they achieved solar cells with an overall solar energy to electrical energy efficiency of 6% [7]. The solar cells were based on silicon and today, after 50 years, the solar cell market is still completely dominated by silicon solar cells due to their good efficiency and stability. However, the production cost of the silicon solar cells is rather high. Elements from group III and IV in the periodic table have also been used for solar cells. The material cost for solar cells using these materials is also high but efficiencies of about 30% can be reached [4]. Another type of solar cells, so called thin film solar cells emerged during the 1970s. The materials used are for example CdTe and Cu(In,Ga)Se2 (CIGS). These materials have a large absorption of visible light and the absorbing layer can therefore be made thin, typically about 1 m. The small amount of material needed to produce these solar cells, together with an effective con-version makes these cells promising for low cost production. The efficiency of the best thin film solar cells is today about 20% [4].

Solar cells based on molecular materials show properties that may sub-stantially reduce the production cost. In these solar cells molecules are used to absorb the light and in some cases also to transport charges. The type of molecular solar cell that this thesis focuses on appeared in the beginning of the 1990s. Brian O’Regan and Michael Grätzel obtained about 7% efficiency with a new solar cell design, the nanoporous dye-sensitized solar cell (DSSC) [8]. The solar cell was a photo-electrochemical solar cell using a liquid electrolyte. The working electrode was made of nanoporous TiO2covered with a monolayer of dye molecules. Recently DSSCs in which the liquid electrolyte is replaced with solid hole conducting materials have been constructed [9-12]. The efficiencies of these solar cells are however still not as high as for the liquid DSSC.

Important issues for the production of solar cells are for example produc-tion cost, energy amount to produce it, efficiency and lifetime. The energy to produce crystalline silicon solar cells is high and it takes years before the solar cells themselves have produced the same amount of energy. This and the high production cost make it probable that other solar cell technologies will grow. Various forms of crystalline silicon together share about 90% of the total solar cell market today and amorphous silicon has about 8% of the market [4]. At present there are however several attempts to set up produc-tion lines for CdTe and CIGS solar cells [4]. The energy and large scale pro-duction costs for the DSSCs are considered to be very low. Short lifetime and low stability at high temperatures have been problems, but recent devel-opment has made the DSSC more stable and increased the lifetime [13,14].

11

The present thesis The focus of this thesis is to further develop the understanding of the dye-sensitized solar cells (DSSC) and especially the solid state DSSC. Different techniques were used that will be described in more detail in section 2. With electron spectroscopy techniques detailed information about the surface structures of the materials on a molecular level was obtained, for example how the materials in the photoactive interface bind to each other and how they interact. Also, how the electronic structure important in the photo-conversion process differs for a series of dyes and hole-conductors.

Methods that more directly measure the photovoltaic properties of the DSSC interfaces were also used. Measurements of the current and voltage characteristics under illumination using a lamp that simulates the solar spec-trum give for example information about the total energy conversion effi-ciency. IPCE (incident photon to current conversion efficiency) measure-ments give for example information of the spectral response of the solar cell. Theoretical models were used to investigate the electronic structure of the dyes and also to study the mechanisms of charge transfer in one hole con-ducting material.

Using a combination of the different methods investigations of how a change on a molecular level affects the efficiency of the solar cell was fol-lowed. Small changes were shown to affect the solar cell efficiency signifi-cantly. Increased knowledge of why and how these molecular changes affect the efficiency is important for further optimization of DSSC solar cells.

12

1. The nanoporous dye-sensitized solar cell

1.1. The function In the dye-sensitized solar cell three different materials are responsible for the function in the conversion of light to electric energy. A semiconducting material transparent to visible light, a monolayer of dye molecules, and a material that can give electrons to the oxidized dye molecules, a so called hole-conductor material. A schematic picture of the function is shown in figure 1. Photons are absorbed by the dye molecules, which then are excited. The excited dye injects an electron into the semiconductor material. This process is extremely fast and takes only pico or femto seconds [15-18]. The electrons are transported through the semiconductor material to the back-contact. The dye molecules are left in an oxidized state and takes up elec-trons from the hole-conducting material to become regenerated. A photopo-tential is built up between the semiconductor and the hole-conductor, which can then be used to perform electric work.

The nanostructured DSSC differ from the conventional solar cells in that the light absorption is separated from the charge transport. This can be an advantage, since it is then possible to make the dye molecule optimal for absorption of light (and not charge transport), and the charge transport mate-rials optimal for charge transport (and not absorption of light), which gives a larger freedom in design of the most effective material combination.

13

Figure 1. Schematic picture of the photoinduced processes in the DSSC. The picture shows the electronic energy levels most important for the solar to electric energy conversion process; The conduction band (C.B) and valence band (V.B) of the semi-conductor, the highest occupied molecular orbital (HOMO) and the lowest unoccu-pied molecular orbital (LUMO) of the dye molecule and the oxidation redox-potential of the hole-conductor material (E/E+). Dye molecule absorbs a photon and one electron is excited from the HOMO-level to the LUMO-level1. The excited dye injects an electron into the semiconductor material2. The electrons are transported through the semiconductor material to the back-contact. The dye molecules are left in an oxidized state and takes up electrons from the hole-conducting material to become regenerated3. The hole-conductor material (E/E+) transports charges from the dye molecules to the contact.

A semiconductor electrode material that has been often used is TiO2 [8, 19-23]. This is a large bandgap semiconductor, which means that it is almost insulating and that it does not absorb visible light. The semiconductor elec-trode can be made porous to make the area of the absorbing dye layer larger. Semiconductor particles that have a size of about 10 nm, are then sintered together to an electrode of about 10 m in thickness. In this way the elec-trode gets a huge surface area and although there only is a monolayer of dye molecules on each particle, the light absorption is large. Compared to a pla-nar surface with dye molecules which have an absorption of less than 1% of the incoming light, a 10 m thick nanostructured electrode has a 1000 time larger surface area and the dye-sensitized electrode become very dark in colour, see figure 2.

C.B

V.B

HOMO

LUMO

E/E+

e-

e-

1.

2.

3

Load

14

Figure 2. Dye-sensitized nanoporous TiO2 on conducting glass.

The most common dyes used so far in the DSSC are Ru-polypyridine com-plexes, which result in solar cells that show high energy conversion efficien-cies [8, 19-23]. The dye molecules often contain anchor groups to attach to the TiO2 surface and the dyes studied in this thesis have carboxylic acid an-chor groups. The absorption spectra of the dye molecule should, to as large extent as possible, coincide with the solar spectrum in order to absorb much sunlight. The photo-conversion yield of DSSCs using the Ru-polypyridine dyes studied here is almost 100 % up to a wavelength of about 700nm [22, 23].

A liquid electrolyte with a suitable choice of redox couple has been used as an efficient material to give electrons to the oxidized dye molecules. The liquid electrolyte fills the nanoporous electrode effectively and transports charges fast enough to the dye molecules. The liquid electrolyte is usually based on a tri-iodide/iodide redox couple, which has a redox potential suit-able for regenerating the dye. The most effective DSSC has been made with this liquid electrolyte. However, to have a liquid electrolyte in the nanostruc-tured solar cell may give incapsulation problems, which reduce the lifetime of these cells. Also the redox potential of the tri-iodide/iodide redox-couple is not energetically optimal for a high photovoltage, although it efficiently regenerates the dye molecules.

Solid hole-conductor materials have been used as interesting alternatives to the liquid electrolyte [9-12]. Replacing the liquid electrolyte with a solid material may lead to practical advantages, although the conversion efficien-cies are still not as high as with a liquid electrolyte. Triarylamine derivative

15

molecules have successfully been used as solid hole-conductor materials [9, 24-28] and the efficiency have been more than 4% [26-28]. Also CuI, CuSCN and conducting polymers [10-11, 29-33] have been used as solid hole-conductor materials.

1.3. Efficiency limitations The maximum efficiency that can be obtained using a single bandgap photo-conversion device is about 30 % [4]. This is due to the loss of photons that have lower energy than the bandgap energy and loss of part of the energy from photons having higher photon energy than the bandgap energy. To obtain the maximum efficiency the bandgap energy should be about 1.3 eV [4], which corresponds to an absorption edge wavelength of about 1 m.

After light absorption in a DSSC the charges must be separated effec-tively. After the excitation the electron is transferred into the TiO2 material and the hole into the hole-conductor material as described above. If this process is too slow the electron and hole recombine and the efficiency will be lower. Therefore the materials have to be designed to work together. The electronic levels of the three materials must have the right energies in order for advantageous charge transfer reactions to dominate, which prevent the loss energy in unwanted charge transfer reactions between the materials.

The energies of the electronic levels should also be in favour for a high voltage. For example the LUMO-level of the dye should be close the con-duction band of TiO2 and the HOMO-level of the hole-conductor should be close to the HOMO-level of the dye. In this way the highest possible voltage from the cell can be obtained. It should however be kept in mind that the materials affect each other when put together and the energy levels are then changed.

Also, for a large efficiency, the separated electron and hole must travel all the way to the contacts without recombining with an opposite charge. The separation of electrons and holes in two different materials is an advantage of the DSSC, compared to standard photovoltaic cells where both holes and electrons are present in the same material, since the recombination in the dye-sensitized cell therefore only can occur over the interface between the materials. The recombination of charges in the interface between the materi-als during the transport through the nanoporous material will not only affects the photo current but will also affect the photovoltage, since it depends on the number of charges in the TiO2 and the hole-conductor.

For a DSSC with a well working absorption material and charge transport material, the efficiency therefore largely depends on the electron transfer rates at the interfaces between the different materials (semiconductor, dye and hole-conductor). The transfer rates are closely connected to the energy-level matching between the materials at the interface as well as the geomet-

16

rical structure. Insight into the geometrical and the electronic structure of the interfacial regions is therefore useful in optimizing the cell [34-44] and has been a main objective in the present work.

17

2. Techniques

2.1. Electron spectroscopy techniquesElectron spectroscopy methods are used to experimentally study the elec-tronic structure of materials. The electronic structure can, for example, give information about what elements are present in the sample and their chemi-cal state. The electron spectroscopy techniques used in this thesis are shortly presented below.

2.1.1. Photoelectron Spectroscopy Photoelectron spectroscopy, PES (also X-ray photoelectron spectroscopy, XPS) is based on the photoelectric effect [45]. Radiation with a well-known energy ionizes the material and the kinetic energy of the ejected electrons is measured, see figure 3.

Figure 3. Schematic picture of process for PES.

In a photoelectron spectroscopy measurement, the number of electrons is recorded as a function of their kinetic energy. The binding energy (Eb) with respect to the Fermi level is then given by Eb=h -Ek- , where h is the en-ergy of the incoming photon, Ek is the kinetic energy of the outgoing elec-

Vacuum

18

tron and the workfunction, which is the energy difference between the Fermi level and the vacuum level. The number of measured electrons is plot-ted as function of the binding energy in a PES spectrum, see figure 4. This spectrum provides information on what elements are present in the sample and their chemical state.

8000

6000

4000

2000

0

Inte

nsity

/arb

.uni

ts

600 400 200 0Binding energy /eV

Figure 4. PES spectrum of a dye-sensitized electrode.

The PES spectrum can be loosely divided in two regions; electronic levels with high binding energies (core electrons) and electronic levels with low binding energy (valence electrons). Core electrons are strongly bound to the nucleus (high binding energies) and do not directly participate in the chemi-cal bonding between atoms. The orbitals of the core electrons therefore have nearly atomic character also in solid materials and molecules. Thus, all at-oms have their unique set of core electron binding energies, which can be used for chemical analysis of the sample.

The exact positions of the core levels depend on the chemical state of the probed atoms and are therefore slightly shifted. The so called chemical shifts are in the order of a few electron volts. To a first approximation the shifts can be interpreted in terms of potential models [46]. In such models, redis-tributions among the valence electrons due to the surrounding atoms result in electrostatic potential energy changes close to the nucleus. For example, negative contribution to the electrostatic potential close to the nucleus, due to valence electron transfer from the surrounding to the atom considered, will give a lowering of the core electron binding energy. That is, it takes less energy to create a core hole on the atom having higher negative charge den-sity localized on it.

19

However, to fully understand a PES spectrum one has to realize that PES measures the energy difference between the initial state and the final state. The initial state energy is the total energy of a state with N electrons and the final state energy is the energy of the state with N-1 electrons. The electrons in the final state are affected by the core hole, giving rise to relaxation ef-fects. The energy of the final state is therefore the energy of the state with N-1 electrons that have relaxed. The relaxation effects of the nuclei position are much slower and the distances between nuclei can therefore be considered to be constant during the emission process. Calculations that include the final state effect can be very important to understand the chemical shifts and are therefore useful in the determination of the chemical state of an atom.

The valence electrons are directly involved in the formation of chemical bonds. The interpretation of valence spectra is therefore more difficult, since the molecular orbitals or bands usually consist of contributions from many atoms. Comparison with calculations is therefore often required to under-stand the valence spectra.

The intensity of a peak in a photoelectron spectrum depends on a number of parameters and the dependence can be written as,

S),,,I(I

where is the differential cross section for ejection of electrons from the relevant orbital, is the surface density of the corresponding element and is the mean free path of the electrons in the sample. The spectrometer-function (S) is the intensity dependence of the electron-optical transport from the sample to the detector. The differential cross section depends on the photon energy, the angle between the photon polarization vector and the direction of the photoelectron. The mean free path is a function of the kinetic energy of the photoelectron and the dependence on kinetic energy follows the so called universal curve, see ref. 47. The spectrometer-function is de-pendent on the kinetic energy of the electrons.

PES is a very surface sensitive technique. This is due to the short mean free path of the electrons in the material. The radiation penetrates several hundreds of Ångström into the material, but the electron elastic escape depth is very small. For the photon energies normally used in this thesis, the mean free path is only 5-25 Å. In a PES experiment the mean free path ( of the emitted electrons depends on their kinetic energy as mentioned above. Thus the surface sensitivity can be changed by varying the photon energy. The minimum mean free path and thus the highest surface sensitivity, is obtained at electron kinetic energies of about 50 eV. The surface sensitivity can also be changed by varying the angle of the sample surface towards the detector. This results in different pathlengths in the material for the emitted electrons that hit the detector.

20

2.1.2. X-ray absorption spectroscopy In X-ray absorption spectroscopy, XAS, electrons from a core level are ex-cited to unoccupied levels by absorption of photons, see schematic picture in figure 5.

Figure 5. Schematic picture of process for XAS.

The excitation leaves the atom with a core hole and thereby in a highly ex-cited and unstable state. The core hole is thereafter rapidly filled with an electron from an outer level. The excess energy is released either as a photon (radiative decay) or is transferred to a secondary electron (Auger electron) that leaves the atom. For the core levels studied here secondary electron decay dominates the deexcitation of the core hole, which means that an elec-tron is emitted in the decay process. The number of absorbed photons is pro-portional to the number of core holes and thus to the number of emitted sec-ondary electrons. The absorption spectrum can therefore be recorded by measuring the number of secondary electrons while changing the photon energy.

The absorption probability of X-rays in matter varies smoothly with the photon energy except at certain energies, absorption edges, corresponding to excitation into unoccupied bound states. These resonant features are often referred to as Near Edge X-ray Absorption Fine Structure (NEXAFS), see figure 6.

Vacuum

21

Inte

nstiy

/arb

. uni

ts

408eV406404402400398396Photon Energy /eV

BD

Figure 6. N1s NEXAFS structure of one of the most effective dyes (BD) in the dye-sensitized solar cell, see molecular structure in summary of the results.

When exciting electrons from a core level to valence levels, the unoccupied levels located on the core hole site are primarily probed. It is therefore pos-sible to distinguish contributions from different species in the unoccupied levels. The transition obeys dipole selection rules, which for example means that p orbitals are probed by excitation from an s core orbital. It should also be kept in mind when analysing the NEXAFS spectra that the core hole will affect the outcome of the measurement.

2.1.3. Resonant spectroscopy Resonant Photoemission Spectroscopy, RPES, refers to photoelectron spec-troscopy using a photon energy that coincides with an excitation energy of a core electron to an unoccupied state (i.e with a NEXAFS peak). In a simpli-fied picture we make the following description of the resonant process: after excitation the core hole is deexcited and the excess energy is transferred to an electron that leaves the atom, so called autoionization. The autoionisation may occur via two types of decay: participator and spectator decay, see fig-ure 7.

22

Figure 7. Schematic picture of decay processes in resonant photoemission.

In participator decay, the excited electron takes part in the decay process leading to a final state single hole configuration with charge +1. The final states are in this case the same as obtained in a normal PE valence band spectrum, having a hole in one of the valence levels. The intensities of the valence band peaks, which contain contributions from the elements where the core excitation took place are however enhanced. Since the final state in the resonant participator decay is the same as in the nonresonant photoemis-sion the participator decay signal is constant in the binding energy. Spectator decay, on the other hand, is a process where the excited electron remains in its excited state during the decay process, the core hole is filled by a valence electron, and another valence electron is emitted. The final state in this case is a two-hole one particle state with charge +1 and the signal is constant in kinetic energy.

Due to the enhancement of the valence band peaks, which contain contri-butions from the elements where the core excitation took place, RPES can be used for studies of element specific contributions to the valence orbitals. Constant initial state spectroscopy (CIS) implies that the excitation energy is scanned, normally over the NEXAFS region, while recording the intensity of a PES spectrum. In the present work, CIS was used to separate x-ray absorp-tion spectra of two different peaks in the valence spectrum of a dye mole-cule. The integration over the two different peaks over a range of excitation wavelengths gave information of the orbital composition of the peaks, see summary of papers or paper II. The CIS spectrum of a dye molecule (BD, see molecular structure in summary of results) is shown in figure 8.

Vacuum

Participator decay

Vacuum

Spectator decay

23

398

400

402

3020

100

500

1000

1500

2000

500

1000

1500

2000

Binding energy/ eVPhoton energy /eV

Figure 8. CIS spectrum of BD dye molecule, see summary of results for molecular structure.

RPES can also be used to extract dynamic information on electronic proc-esses that occur within the lifetime of the core hole, typically a few femto-seconds [15, 48-49].

2.1.4. Experimental setupThe PES experiments need high energetic light for the emission of electrons. Several types of light sources can be utilized: characteristic X-rays (e.g. AlK , MgK ), UV (e.g HeI ) and synchrotron radiation (SR). All electron spectroscopy spectra in this thesis where recorded using SR at the Swedish National Laboratory MAX-lab in Lund [50].

Free charges that are accelerated emit electromagnetic radiation. In a syn-chrotron, bunches of electrons are moving in an ultra high vacuum storage ring, which consists of straight sections and corners where magnets change the electron path. When the magnets bend the electron path the electrons are accelerated and they emit light. To enhance the emission of light, devices called wigglers and undulators are inserted into the straight sections. These constitute arrays of magnets that bend the electron path a number of times, which gives higher light intensity. The vertical distance between the mag-netic arrays can be varied, which will change the frequency of the intensity maximum.

24

The intensity from a SR source is often much higher compared to the characteristic X-ray radiation, which makes the measurement faster or with improved statistics. Another advantage using SR is that photon energies can be changed continuously, which makes for example XAS and resonant PES possible.

The emitted light radiates from the synchrotron storage ring into a vac-uum tube to a monochromator, which is used to select light with the pre-ferred photon energy. The selected light is then transmitted via tubes to the measurement chamber. As described in the PES section above, the light irra-diates the sample and electrons are emitted. The kinetic energies of the emit-ted electrons are measured using an electron analyser. To allow for the emit-ted electrons to reach the analyser, the experiments are performed in a vac-uum chamber with a pressure below 10-6 Torr. The experimental station of beamline I411 at maxlab, where most of the PES experiments in this thesis were performed, is shown in figure 9.

Figure 9. Experimental endstation BeamLine I411 at MAX-lab.

The analyser consists of an electron lens, two hemispheres and a multichan-nel detector. The electron lens collects and accelerates or retards the elec-trons before they reach the two hemispheres. A voltage is set between the two hemispheres and electrons within a specified energy range will reach the multichannel detector. The acceleration or retardation voltages of the elec-tron lens are varied and the number of electrons at different voltages is measured using the multichannel detector.

25

2.2. Photovoltaic measurements 2.2.1. IPCE measurements The incident photon to current conversion efficiency (IPCE) is a measure of the number of useful electrons generated per incident photon. The IPCE can be calculated according to:

Pj

IPCE ph1240

Where jph is the photocurrent density, P is the light power density and is the wavelength in nanometer.

The photocurrent of the solar cell is measured during illumination with monochromatic light. In an IPCE spectrum the IPCE value is given as a function of the wavelength. The IPCE equipment used in paper IV is shown in figure 10.

Figure 10. IPCE setup used in paper IV. From left: Lamp, IR-filter, lens, mono-chromator, sample-holder and contact.

26

2.2.2. I-V measurements In the I-V measurements presented in this thesis, the current versus the volt-age over the device is measured. This can be done both with and without illumination from a light source. The I-V spectrum of a solar cell at illumina-tion gives for example information about the photovoltage without any load, the highest photocurrent and also the maximum solar- to electric energy conversion efficiency. The power that a solar cell provides is the product of its current and voltage under operating conditions. Under short-circuit condi-tions the current is maximal but the voltage is zero and no power is provided. On the other hand, under open-circuit conditions, voltage is maximal but no current flows and the power is again zero. Somewhere between these ex-tremes, the product of voltage and current is maximized and therefore also the power.

2.3. Theoretical methods 2.3.1. Hückel method To calculate energy levels and orbitals in very large molecules with pi-bonds, for example conducting polymers one may use the Hückel method [51]. In the Hückel method molecular wavefunctions are constituted from the 2pz orbitals of the atoms in the molecule (2pz-orbitals orthogonal to the molecule plane) by linear combination.

In the Hückel method, one approximates: 1) The overlap of orbitals from different atoms is zero. 2) The coupling (the interaction between atoms) is zero if the atoms are

not closest neighbours.

This is called the tight-binding approximation.

2.3.2. Hartree-Fock equations and Density functional theory More complete theories to solve the Schrödinger equation of atoms and molecules are used by many computational scientists. However, in all cases of computation of larger molecules approximations are needed. In the Born-Oppenheimer approximation the nuclei are assumed to be static, which al-lows for separation of the electron’s and the nuclei’s Schrödinger equations. Moreover, in the so-called orbital approximation the many-electron wave functions are built up by one-electron wavefunctions. To satisfy the Pauli principle the wavefunction is written as a Slater determinant. When such a determinantal wavefunction is combined with the variational principle, it turns out that the optimum one-electron wavefunctions must satisfy the Har-tree-Fock equation.

27

In Density functional theory (DFT) the electron density rather than the wavefunction is in focus. The central idea is that there is a relationship be-tween the total electronic energy and the overall electronic density. Hohen-berg and Kohn showed that the ground state energy of a system is uniquely defined by the electron density.

Calculations are performed using the Kohn-Sham equation, which is analogous to the Hartree-Fock equation except for the exchange-correlation potential term. The exchange correlation potential is calculated by assuming an approximate form of the dependence on the electron density. To solve the Kohn-Sham equation a self-consistent approach is taken. An initial guess of the density is fed into the equation from which a set of orbitals can be de-rived, leading to an improved value for the density, which is then used in the second iteration and so on until convergence is achieved within a chosen criterion.

2.4. Sample preparations 2.4.1 Preparation of TiO2 electrodes Three different preparation methods for TiO2 electrodes have been used. The nanoporous TiO2 layers used in paper I-III were prepared from a paste with nano-crystals of TiO2 [8]. The paste was spread out on a conducting glass substrate to a film with a thickness of a few micrometer (doctor blading). The paste was dried and then the substrate was sintered in 450oC in air for 30 min.

The second type of TiO2 used in paper IV-V was made by spray-coating on conducting glass (fluorine doped tin dioxide). The conducting glass were heated to 450oC and a solution of titanium-tetra(isopropoxide), acetyl-acetone and ethanol was sprayed onto the hot glass [52]. The resulting TiO2layers were rather compact and X-ray diffraction (XRD) analysis showed that the crystal structure was anatase.

The third preparation method used in paper VI was Chemical Vapour Deposition (CVD), which is a synthetic process in which the chemical con-stituents react in the vapour phase near or on the heated substrate to form a solid deposit. The TiO2 films were grown on a pre-oxidized Si (111) by means of ultra-high vacuum metal-organic chemical vapor deposition (UHV-MOCVD) using titanium-tetra(isopropoxide) as a precursor. The re-sulting films were thin nanocrystalline anatase TiO2 layers [53].

2.4.2 Dye-sensitization The dye-sensitization was made in two different ways. It is not possible to evaporate the Ruthenium dye molecules used in the thesis. Therefore, in

28

paper I and VI, to reduce the level of contamination the dye was deposited by immersing the sample into an ethanol solution of the dye in an Argon filled glove-box. Hot TiO2 samples were transferred into the glovebox and after the dye-sensitization the samples were transferred from the glovebox to the PES vacuum chamber using an Argon filled transfer module. In paper III-V the dye-sensitization were performed outside the glove-box in a similar way. The compact layers were sensitized shorter times (minutes) than the nanoporous layers since it takes longer time (hours) for the dye-molecules to fill up all the nanoporous surface.

2.4.3 Hole-conductor preparations Three different methods were used to prepare the hole-conductor layers. The solid state triarylamine hole-conductors in paper IV and the CuI layer in paper VI were deposited by thermal evaporation. The triarylamine molecules were put in a glass nozzle with a tungsten wire around that was heated. It was then possible to observe when the molecules melted and when the depo-sition was initiated. The CuI was evaporated from a Ta-pocket that was heated to about 400oC.

The second preparation method was spin coating that was used in paper V. A drop of the conducting polymer was put on the substrate that was then rotated to form a thin layer of the polymer.

The third method was to use the triarylamine hole-conductors in the sol-vated state. The molecules were then solvated in a solvent and used as a liquid electrolyte, see paper IV.

29

3. Summary of the results

In the general presentation of the solar cell function (section 1), it was noted that the final efficiency is largely affected by the electron transfer processes between the materials at the dye-sensitized interface. The transfer rates are closely connected to the electronic structure of the materials at the interface as well as to their combined molecular structure. A schematic picture of the dye-sensitized solar cell emphasizing the importance of the electronic and molecular structure is shown in figure 11.

Figure 11. Schematic picture of the interfaces in the dye-sensitized solar cell.

In the present thesis different techniques have been used to obtain informa-tion of the different materials and specifically of the electronic and molecu-lar structure in the interface between them. In paper I-III different dyes and the molecular surface structure of the dye/TiO2 interface is investigated in detail using electron spectroscopy techniques. In paper IV-VI the interface as well as the photovoltaic properties of heterojunctions of dye-sensitized TiO2and different hole-conductor materials is investigated. The charge transfer mechanism is theoretically studied for a hole-conducting material (conduct-ing polymer) in paper VII.

Ru

e-e-

N

NN

N

Electronic structure Molecular structure

Interfacial properties

TiO2Dye

Hole conductor

Solar cell

e-

e-

30

3.1. Dyes and TiO2/dye interfaces Sensitization of nanoporous TiO2 with different Ru-polypyridine complexes, e.g. BD, N719 and N3, see molecular structures in figure 11, has resulted in solar cell systems having very good photon to current conversion efficiency in the visible region [8,19-23]. In the IR region the photoresponse of black dye (BD) reaches about 100nm further than N3 and N719 [8,19-23]. The total energy conversion efficiency is, however, rather similar.

Figure 12. Molecular structure of BD, N719 and N3 dyes.

In paper I the molecular and electronic structure of BD, N719 and N3 ad-sorbed on TiO2 is studied using PES. Both N3 and N719 are investigated in order to measure the effects of different protonation of the complex.

Different protonation of the dye molecule have been found to have an ef-fect on the properties of the solar cell. Changing some of the counterions from protons to tetrabutylammonium (TBA+) results in higher open-circuit voltages, lower short circuit currents and higher fillfactors which result in higher efficiency of the solar cells with TBA+ counter-ions [54,55]. The amount of TBA+ counter ions relative to the amount of dye molecules on the surface was measured by PES on the N1s core level. The results showed that the number of TBA+ counterions on the dye sensitised TiO2 is lower than expected from the concentration in the dye solution. This agrees with meas-urements for N719 on TiO2 using other methods [56].

Core level PES also gave information of the binding structure of the three dyes to the surface. The binding geometry is still under debate and several suggestions of the binding configuration have been presented [35-41, 55, 57-62]. The results in paper I indicated a mixture of binding configurations, see figure 13.

NN

NN

Ru NCS

NCSCOOH

COO TBA

COOH

COO TBA

- +

- +

N719

NN

NN

Ru NCS

NCSCOOH

COOH

COOH

COOH

N3

NN

RuN

NCS

NCS

NCS

BD

-+

COOH

COO

-+ COO

TBA +

TBA

TBA

31

Figure 13. A mix of two different binding configurations of N719 that are suggested from results in paper I.

Specifically, a comparison of the O1s spectra for the three dyes indicates that the interactions through the carboxylate groups with the TiO2 surface are very similar. As has been observed previously, the O1s spectrum of N3 on TiO2 indicates that the N3 molecules anchor to the TiO2 surface through two of the four carboxylate/carboxylic acid groups, i.e., two of the four carbox-ylic acid groups are still present also when adsorbed at the surface. Also, for N719 and BD on TiO2, -OH units in carboxylic acid groups are observed and photon energy dependent measurements indicate that they are positioned on a larger distance from the TiO2 surface than carboxylate oxygen. This indi-cates that N719 and BD are anchored to the TiO2 surface through carboxy-late groups partly having TBA+ as counter ions.

The S2p spectra showed an extra contribution in the spectra for the monolayers of the dyes on TiO2 compared to the spectra of the multilayers, see figure 14. Mainly based on the structure of the S2p core level spectra a model is proposed in which a fraction of the -NCS groups interact with the TiO2 surface through the sulfur atoms, see figure 13. Interestingly, in this mixed binding configuration the fraction of NCS-groups interacting with the TiO2 surface is larger for N3 (30%) than for N719 (10%). For the BD dye the fraction is estimated to 15%. This mixing in binding configurations is important when analyzing the mechanism of the energy conversion in the solar cell since differences in binding configurations is expected to affect charge transfer rates at the dye/TiO2 interface.

N N

N

Ru N

NN

CS

C

S

OH

O

OOH

O O O O

N N

NN

Ru

N

NC

C

S

S

O O O O

HO O OOH

32

Figure 14. S2p spectra of dyes on TiO2 (left) and multilayers of the dyes (right).

Valence photoemission spectroscopy shows that the binding energy of the highest occupied orbitals of BD is lower than the binding energy of the high-est occupied orbitals of N3 and N719 when adsorbed onto TiO2, see figure 15. Interestingly, the relative energy level matching of N3 versus TiO2 and N719 versus TiO2 are similar, indicating that the presence of TBA+ does not affect the energy matching between the HOMO level and the substrate.

Figure 15. Schematic figure of the frontier electronic structure of the investigated dye molecules as obtained by PES, showing the peaks obtained in the occupied electronic structure of the dyes in relation to the bands of TiO2. The energy level matching of the highest occupied electron structure of the dyes (in the bandgap of TiO2) relative the TiO2 substrate is similar for N3 and N719 but differs for BD and D5 (about -0.3 eV and +0.2 eV relative N3, respectively). The energy difference between the occupied electron structure and the unoccupied electron structure may be approximated by the optical absorption peak energy for the dyes as indicated in the figure.

Inte

nsity

/arb

. uni

ts

166eV 165 164 163 162 161 160Binding energy /eV

BD

N719

N3

S2pmultilayers

Inte

nsity

/arb

.uni

ts

166 165 164 163 162 161 160Binding energy /eV

BD

N719

N3

S2p

C.B

V.B

TiO2/N3 TiO2/N719 TiO2/BD TiO2/D5

33

In paper II the frontier electron structures of BD and N719 multilayers are investigated. The valence electronic structures in the binding energy range 10-0 eV of the multilayers of the dyes are shown in figure 16. Structures observed in the valence spectra of multilayers of BD and N719 dyes show large variations in relative intensity when varying the photon energy from high to low energies. Based on the values from theoretical cross-section these variations indicate that the structure at lower binding energy (the two peaks in figure 16) have large Ru4d character in both BD and N719. Contri-butions to these structures from the different ligands were also investigated (see below).

Inte

nsity

/arb

. uni

ts

10eV 8 6 4 2 0Binding energy /eV

BD

N719

Ru-N(NCS)

Ru-N(bpy)

Ru-N(terpy)

Figure 16. Valence electron structure of BD and N719.

The N1s unoccupied states were investigated with XAS (X-ray absorption spectroscopy) over the N1s absorption edge. The N1s-NEXAFS structure of BD and N719 is shown in figure 17. The appearance of two resonances with different intensity in the BD and N719 spectra at about 400 eV is explained by the fact that BD and N719 contain two inequivalent nitrogens coordinated to the Ru center. By comparison with calculations the resonance at lower energy (about 399 eV) is assigned to the pyridine (ter/bi) units while the feature at higher energy (about 400 eV) is assigned to the NCS ligands.

34

Inte

nstiy

/arb

. uni

ts

408eV406404402400398396Photon Energy /eV

BD

N719

N(NCS)

N(terpy)

N(bipy)

Figure 17. N1s-NEXAFS spectra of BD and N719.

Since the N1s-NEXAFS of the BD and the N719 molecules contain two structures that have very different characters (one localized on the NCS ligand and one localized on the ter/bipyridin ligand) a CIS spectra can be used to map the NCS and ter/bipyridine contributions to the valence states. In figure 18 the CIS spectra over the first two valence electronic peaks in BD and N719 are shown.

Figure 18. Valence spectra at different photon energies (CIS spectra) over the first valence peaks in BD (left) and N719 (right). Lighter is higher intensity. The circle shows the peak in intensity at a photon energy close to 400 eV, in the region of the first valence structure peak of the dye.

401

400

399

398

Pho

ton

ener

gy /e

V

543210-1 Binding energy /eV

401

400

399

398

Pho

ton

ener

gy /e

V

543210-1 Binding energy /eV

35

At the binding energy corresponding to the first valence peak, the CIS spec-tra show a peak in intensity at a photon energy close to 400 eV. Comparing with the NEXAFS spectra, this photon energy corresponds to the NCS con-tribution in the NEXAFS spectra for both N719 and BD. This means that the peak at lowest binding energy has partly a NCS character both in N719 and BD and much less ter/bipyridine character. The experimental results indicat-ing large NCS contribution in the highest occupied valence peak is supported by calculations of the two molecules and earlier calculations also show simi-lar results [63,64].

Finally an investigation on a TiO2/dye interface comprising a purely or-ganic dye is discussed, see paper III. The interest for metal free, organic dyes with high extinction coefficient has grown in recent years and an efficient and easily synthesized polyene-diphenylaniline dye has recently been re-ported [65]. In this dye (D5) the diphenylaniline moiety acts as an electron donor and the cyanoacetic moiety acts as the electron acceptor and as an-choring groups for attachment on the TiO2, see molecular structure in figure 19.

Figure 19. molecular structure of D5.

In paper III the electronic structure of D5 adsorbed from solution on nanoporous TiO2 was investigated by photoelectron spectroscopy (PES), X-ray absorption spectroscopy (XAS) and resonant photoelectron spectroscopy (RPES). The results were also compared to calculations based on density functional theory. The N1s core level spectrum show two different contribu-tions assigned to nitrogen in the triarylamine (TAA) moiety and the cyano (NC) moiety in the molecule. When adsorbed on TiO2 the intensity ratio of the two contributions differs from the molecular formula of the dye. The TAA moiety has a larger contribution than the NC moiety, see figure 20 (left).

N

SCOOH

CN

36

Figure 20. Left: N1s spectra of D5 on TiO2. The peak at higher binding energies is assigned to the triarylamine moiety and the peak at lower binding energies is as-signed to the NC moiety. Right: Valence spectra of D5 and N3.

Since PES is a very surface sensitive method, the difference in relative inten-sity indicates that the dye molecules are bonded to the surface with the NC moiety close to the TiO2 substrate while the TAA unit is sticking out from the surface. The surface sensitivity of PES may be varied by changing the photon energy as described above and the results in figure 20 supports the suggested binding geometry.

In figure 20 to the right, the valence electronic structure of TiO2 sensi-tized with D-TAA and TiO2 sensitized with N3 is shown. The outermost valence structure of D5 show a peak at 1.6 eV and is shifted about 0.2 eV towards higher binding in comparison to the peak for N3, see also figure 15. When comparing the spectra, the structure of D5 is also clearly broader and contains a small structure at 2.2 eV (i.e. separated 0.6 eV from the large peak). The appearance of two separated structures was also observed in the theoretical calculation of the dye molecule.

N1s-NEXAS was used to probe the nitrogen contribution in the unoccu-pied density of states, see figure 21. By comparing the experimental spectra of a different model molecule, we could assign the different structures. In particular the spectra contain two strong resonances at lower photon energy originating from the NC moiety and two resonances at higher energy origi-nating from the triarylamine moiety. Calculations could reproduce the ex-perimental XAS spectra and confirmed the assignment of the different peaks.

An attempt to further delineate the valence electronic structure using RPES is also shown in figure 21. An integrated CIS spectrum was recorded for the outermost electronic structure 0-3 eV and compared to the N1s-NEXAFS spectrum. Calculations show that the outermost valence region only has a minor nitrogen contribution (mainly TAA). The low intensity of the TAA signal compared to the CN signal is therefore surprising and possi-

Inte

nsity

/arb

. uni

ts

402eV 400 398 396Binding energy /eV

758 eV

540 eV

Inte

nsity

/arb

. uni

ts

2eV 1 0 -1Binding energy /eV

D5 N3

37

bly indicates electron transfer before the core-hole decay. However, a com-parison with a multilayer measurement and better statistics is needed to make any certain conclusions about such dynamics in the system.

Inte

nsity

/arb

. uni

ts

415eV410405400Photon energy /eV

N1-NEXAFS CIS 0-3 eV

Figure 21. N1s-NEXAFS and CIS (integrated 0-3 eV B.E.) spectra for D5 on TiO2.

3.3. TiO2 / dye / hole-conductor interfaces In paper IV-VI combinations of different materials are assembled to form model systems for solid state DSSC. TiO2/dye/ hole-conductor heterojunc-tions were prepared and the interfacial and photovoltaic properties were studied. Three different hole-conductor materials were investigated. Tri-arylamine derivatives, a conducting polymer and CuI. All systems showed photovoltaic properties and the IPCE spectra for three of the systems are shown in figure 22.

Figure 22. IPCE spectra of systems with the three different types of hole-conductors used. From left PEDOT-PSS, HC2 (one of the triarylamine molecules) and CuI.

3.5x10-3

3.0

2.5

2.0

1.5

1.0

0.5

0.0

IPC

E

800700600500400wavelength /nm

TiO2/ N719/ HC2

3.5x10-3

3.0

2.5

2.0

1.5

1.0

0.5

0.0

IPC

E

800700600500400Wavelength /nm

TiO2/ N719/ PEDOT-PSS

3.5x10-3

3.0

2.5

2.0

1.5

1.0

0.5

0.0

IPC

E

800700600500400Wavelength /nm

TiO2/ N719/ CuI

38

The series of triarylamine molecules that were used as hole-conductors are shown in figure 23. Similar molecules have successfully been used as solid hole-conductor materials [9, 24-28] and the efficiency have in some cases been more than 4% [26-28].

Figure 23. Molecular structure of the triarylamine derivative molecules used in paper IV.

The series of triarylamine molecules in figure 23 were used to investigate the influence of differences in the hole-conductor material on the photovoltaic properties. Both solid state and liquid state heterojunctions with the tri-arylamine molecules were investigated. In the solid state heterojunctions the hole-conductor molecules were deposited by thermal evaporation on the substrate and in the liquid state heterojunctions the hole-conductor molecules were solvated in an organic solvent. In figure 24 the current versus voltage characteristics of the solid state heterojunctions are shown. The heterojunc-tion with HC2 and HC3 showed higher photocurrents than HC1 and the photovoltage was highest in the heterojunction containing HC3.

N

NN

N

N

NN

NN N

HC1 HC2 HC3

39

25x10-3

20

15

10

5

0

Cur

rent

mA

/cm

2

0.70.60.50.40.30.20.10.0Voltage /V

HC3 HC2 HC1

Figure 24. Current versus voltage characteristics under illumination for solid state heterojunctions with triarylamine molecules as holeconductors.

The photovoltaic properties of the heterojunction largely depend on the in-terfacial structure between the different materials (semiconductor, dye and hole-conductor) as mentioned above. From comparisons of PES valence structure measurements of the interfaces with triarylamine molecules it was possible to observe the energy-level matching of the different materials. It was concluded that in the solid state system having lowest photo-current yields, the hole-conductor molecule (HC1) has an energy level matching to the dye molecule that is disadvantageous for fast electron transfer. In the systems with higher efficiency the energy-level matching is probably more favorable for electron transfer to the dye, due to an extra contribution in the highest occupied structure of the hole-conductor at about 0.5 eV lower bind-ing energy than the peak of the dye, see figure 26.

Figure 25. Valence electronic spectra of dye-sensitized TiO2 with HC1, HC2 and HC3.

Inte

nsity

/arb

. uni

ts

3eV 2 1 0 -1Binding energy /eV

TiO2/dye TiO2/dye/thin HC1 TiO2/dye/HC1 TiO2/dye/thick HC1

Inte

nsity

/arb

.uni

ts

3eV 2 1 0 -1Binding energy /eV

TiO2/dye TiO2/dye/thin HC2 TiO2/dye/HC2 TiO2/dye/thick HC2

Inte

nsity

/ ar

b.un

its

3 2 1 0 -1Binding energy /eV

TiO2/dye TiO2/dye/thin HC3 TiO2/dye/HC3 TiO2/dye/thick HC3

40

The PES investigation of the systems with the different triarylamine hole-conductor molecules also gave information on the geometric interfacial structure. Based on the N1s spectra, the different hole-conductor molecules were found to have different preferred locations on the surface. The solid state interfaces that showed a more advantageous location of the hole-conductor molecules also showed better photovoltaic properties. The unfa-vorable position of the HC1 and partly also the HC2 molecules between the dye molecules on the TiO2 substrate probably affect charge recombination and thereby the short-circuit current (HC1) and the fill factor (HC2). It was therefore found that in the heterojunction with HC1, that showed low photo-current, both the geometric and energetic properties were unfavorable.

To investigate how the dye molecule affect the properties of a solid state heterojunction a series of dyes were used in heterojunctions with TiO2 and a conducting polymer in paper V. Ruthenium-polypyridine dyes [22, 66-68] were used as light absorbing materials and PEDOT-PSS (poly(3,4-ethylenedioxythio-phene)-poly(4-styrenesulfonate)), as organic hole-conducting material. The polymer consists of PEDOT [69,70] which is ‘doped’ (partly oxidized), to become highly conductive, and charge compen-sated with PSS [71]. Since PEDOT-PSS is relatively transparent [72-74] it is an advantage to use it in the DSSC, where the light absorption should be carried out mainly by the dye, and the polymer should only be used for transport of electrons to the dye. The electronic structure of PEDOT-PSS has been investigated previously [75,76].

The heterojunctions were shown to have photovoltaic properties, with the dye absorbing the light, the TiO2 acting as an electron conducting material and PEDOT-PSS acting as a hole-conducting material. The polymer systems were compared to a photoelectrochemical system in which the PEDOT-PSS was replaced by a liquid electrolyte containing tri-iodide/iodide. One of the polymer systems had higher IPCE values compared to the tri-iodide/iodide liquid electrolyte system with the same dye. The IPCE of PEDOT-PSS sys-tems with the different dyes are shown in figure 27 (left). The stability of the most effective TiO2/dye/PEDOT-PSS heterojunction in paper V was also studied. In figure 27 (right) the IPCE maximum value is measured a number of times for more than a month. The maximum IPCE decreases the first two weeks, but stays after that at a stable value over the rest of the time. This shows that it is possible to make rather stable devices using conducting polymers, without sealing of the solar cell.

41

Figure 26. Left: The IPCE of PEDOT-PSS systems with different dyes. For dye structures, see paper V. Right: The stability of the IPCE maximum value for the most effective TiO2/dye/PEDOT-PSS heterojunction in paper V, studied for some time.

In order to allow for a PES study of the specific interactions of the interfac-ing materials the dye/TiO2 and dye/PEDOT-PSS interfaces were investigated separately.

Specifically, it was concluded that the interaction with the dye clearly af-fects the structure of PEDOT-PSS, both with respect to the surface composi-tion of PSS relative to PEDOT and with respect to the chemical state of the sulphur in the polymers.

Moreover, a comparison of the valence band spectra of the two different interfaces (dye/TiO2 and dye/PEDOT-PSS) indicated that the energy level structure of the dyes compared to the substrate is different for the two sur-faces. Thus, in the combined energy level picture under dark conditions, the energy levels in TiO2 relative to the energy levels in PEDOT-PSS depend on the dye, see figure 27. A comparison of the open-circuit voltages for the heterojunctions with different dyes showed that the voltages depend on the dyes and that the largest voltages were in fact measured for the dyes that in the combined energy level picture gave the highest difference between the energy levels of TiO2 and PEDOT-PSS.

0.8

0.6

0.4

0.2

0.0

IPC

E %

800700600500Wavelength /nm

a b c d e

0.8

0.6

0.4

0.2

0.0

IPC

E%

403020100Days

42

Figure 27. Schematic picture of the PEDOT-PSS/dye/TiO2 heterojunction (in the middle) in combination with PES spectra of the PEDOT-PSS/dye interface (to the left) as well as the dye/TiO2 interface (to the right) for a number of dyes. The dyes in the PEDOT-PSS/dye interfaces are from the right: V, IV, III, II, I. The dyes in the dye/TiO2 interfaces are from the left: V, IV, III, II, I. The molecular structures of the dyes are shown in paper V.

Heterojunctions with CuI as hole-conductor material was prepared in paper VI. In this heterojunction the TiO2 layer was prepared in-situ under ultra high vacuum conditions. The dye layer was deposited in an Argon filled glove-box to minimize contamination and after dye-sensitization the samples were transferred from the glovebox to the PES vacuum chamber using an Argon filled transfer module. The CuI was deposited onto the dye-sensitized film with thermal evaporation of CuI powder. Similarly to the investigation on the triarylamine molecules, different thickness of the CuI layer were in-vestigated to study the interaction between the materials in the heterojunc-tion.

It was found that CuI forms particles, which interact with a fraction of the NCS groups of the dye. Also, bond breaking between the dye and TiO2 oc-curs, which suggests that the CuI displaces the dye molecules leading to a film consisting of a compressed dye layer separated by CuI particles. There is then direct contact between the CuI and TiO2 in some points, which may be a disadvantage in these cells. In fact attempts to observe any photovoltage from systems prepared in this way failed, while it still was possible to obtain a photocurrent, see figure 22.

3.2. Polythiophene and oligothiophene hole-conductors To understand how the holes are transferred away from the dye-sensitized interface to the back-contact, the hole-conduction mechanism of a conduct-ing polymer was investigated theoretically in paper VII. The charge localiza-tion on the thiophene polymer chain was investigated as well as how the

B.E(eV)

e-

e-h

e-

e

Counts (arb. Units) Counts (arb. Units)

Dye/PEDOT- Dye/TiO2

B.E(eV)

0 0

5

5

43

properties of the thiophene chains and the properties of the ordering in dif-ferent directions in an ordered oligo- and polymer thiophene material affect the conduction.

When an electron is taken away from the polymer chain, the calculation indicates that the positive charge localizes to a region of the chain, see figure 28. The region has the size of about 5 thiophene units and the geometry (bond lengths) of the chain also adjust in this region. The change in energy due to this change in geometry is called the reorganization energy. This geometrical reorganization energy should not be confused with the electronic relaxation energy that was discussed earlier in the PES section.

0 200 400-0.04

-0.02

0

atom number

char

ge (e

)

a

Figure 28. Charge localization in the positive ion of polythiophene including 375 carbon and sulphur atoms in the chain.

Progress in organic electronics has emphasized the need of theoretical mod-els for the conductivity that are applicable in the case of localized charges [77,78]. A theoretical model should explain the dependence of conductivity on temperature, carrier density and the strength of the applied electric field.

In the model for conductivity reported in paper VII the conduction of holes on the microscopic scale is treated as a hopping process (phonon as-sisted tunnelling) between sites [79]. This hopping process is described using Marcus theory [80], where the rate of hopping partly depends on the reor-ganization energy and the coupling. The reorganization energy is a material constant that is equal in all directions and the coupling is a measure of the tunnelling probability to a site in a certain direction and is therefore direc-tional dependent. In the model an applied electric field is treated as a change in the driving force for the electron transfer.

This theoretical approach results in a linear electric field dependence of the conductivity (ohmic behaviour) at low electric fields, which is expected. The model covers both an activated region at lower temperatures and a metal

44

like conductivity at high temperatures. No insulator-to-metal transition is necessary to obtain this dependence, contrary to what is often expressed. The peak conductivity temperature depends on both the reorganization energy and the coupling. With a large coupling the activation energy disappears and the system passes into a metallic region without activation energy.

45

5. Outlook

Molecular based solar cells are attractive partly due to the substantially re-duced production cost compared to inorganic solar cell. Progress in synthesis of organic materials and inorganic nano-materials for display applications makes the number of useful materials very large. Therefore, the functional properties in the molecular solar cells can be fine-tuned. Also by combina-tion of different materials unique properties can arise. The large range of possible material combinations makes it probable to find material combina-tions that have energy conversion efficiencies comparable to the best inor-ganic solar cells. The understanding of the physics of molecular solar cells is, however, still at a primitive stage compared to inorganic solar cells. Fur-ther fundamental experimental studies are therefore required to reach the goals of high efficiency and stability [81].

The work in the present thesis has shown the possibilities to study the dyes and the surface structure in dye-sensitized solar cells on a molecular level using PES. By comparing these investigations to measurements that more directly measure the photovoltaic properties of the DSSC interfaces it was possible to follow how a change on the molecular level affects the photocurrent and photovoltage. A further step in this development is to measure the photovoltaic properties under highly controlled conditions. An initial test has been made, where a solid state heterojunction was prepared in vacuum. PES was used to measure the surface properties and then after de-positing a contact, the photocurrent and photovoltage was measured in vac-uum. Also a compact IPCE setup is being constructed that will be attached to the vacuum chamber extending the possibilities for such vacuum measure-ments. With this approach we have good control of the heterojunction. The structure can be defined and built up systematically for fundamental studies. It is interesting also to perform long time photovoltaic measurements in dif-ferent atmospheres under maximum control to study what mechanisms that lead to degradation of the solar cell. In the future one can also think of com-binations of PES and time resolved two-beam experiments where controlled molecular and electronic surface structures can be investigated for studies of dynamics. Measurements of the dynamics with element specific methods will then give new insight in the charge transfer processes and in combina-tion with the detailed molecular structure, the steps toward further efficiency improvements can be outlined.

46

4. Populärvetenskaplig sammanfattning

4.1. InledningIdag kommer hela 80% av all konsumerad energi i världen från fossila bränslen som olja och kol och ungefär 10% från kärnkraft [1]. Med tanke på de negativa miljöeffekter som detta medför, t.ex. försurning, växthuseffekt, problemet med lagring av använt kärnkraftsbränsle och faran med kärn-kraftsolyckor, så är det bättre att använda förnyelsebara energikällor som solceller, solvärme, bioenergi, vindkraft och vattenkraft. Av de nämnda energikällorna är solcellerna den minst utnyttjade och har en stor potential att utvecklas.

Till exempel innehåller solljuset på två kvadratmeter i Sverige tillräckligt mycket energi för att tillgodose behovet av hushållsel (t.ex. spis, ugn, kyl-skåp, frys, lampor, TV) för en person. Med solceller som har en verknings-grad på ca 20% krävs det alltså ungefär 10 kvadratmeter solceller per person, vilket gör det fullt möjligt att tillgodose behovet av hushållsel med solceller.

Kylning står för en stor del av el-energiförbrukningen i världen. Detta verkar vara en intressant tillämpning för solceller, som ger mest energi då det är som varmast och soligast. Men för att man ska kunna utnyttja energin vid tidpunkter då solen inte lyser krävs att det finns möjlighet att lagra sol-energin. Detta kan idag göras med batterier. En annan möjlighet för energi-lagring kan vara att omvandla den elektriska energin från solcellerna till kemisk energi, t.ex. att omvandla vatten till syrgas och vätgas. Sedan kan man använda bränsleceller för att omvandla den kemiska energin tillbaka till elektrisk energi.

Idag utgör kristallina kiselsolceller mer än 90% av den totala solcells-marknaden [4]. Den höga produktionskostnaden och energimängden för att producera kristallina kiselsolceller gör att andelen av andra typer av solceller förmodligen kommer att öka. En variant med stor potential som uppfunnits nyligen är den nanoporösa infärgade solcellen, vilken är den typ av solcell som har undersökts i den här avhandlingen. Energin att producera nano-porösa infärgade solceller förväntas vara låg och likaså produktionskostna-den vid massproduktion. Kort livstid och dålig stabilitet vid höga temperatu-rer har dock varit problem med den här typen av solcell.

47

4.2. Den nanoporösa infärgade solcellen I den nanoporösa infärgade solcellen är tre olika material viktigast för om-vandlingen från ljusenergi till elektrisk energi. Ett lager av färgämnesmole-kyler som absorberar ljus och därefter ger elektroner till en genomskinlig halvledare, vanligtvis titandioxid. Ett så kallat hålledarmaterial ger elektro-ner till de färgämnesmolekyler som har avgett elektroner till titandioxiden. På så sätt får titandioxiden ett överskott av elektroner och hålledarmaterialet ett underskott av elektroner, vilket ger upphov till en spänning som kan an-vändas till elektriskt arbete, se figur 29.

Figure 29. Till vänster en schematisk bild över den nanoporösa infärgade solcellens olika delar. Till höger en schematisk bild av gränssnittet mellan materialen på mole-kylär nivå där en triarylamin-baserad molekyl används som hålledarmaterial och ett ruteniumkomplex som färgämne.

Titandioxidlagret består av nanokristallina titandioxidpartiklar med en stor-lek på ca 10 nanometer, som är värmda eller pressade ihop till ett poröst skikt med en tjocklek på cirka 10 mikrometer. Ett molekyllager av färgämnet fastnar på titandioxidytan då titandioxidskiktet doppas i en lösning av färg-ämnet. Den inre ytan av det porösa titandioxidskiktet är enorm, ungefär tu-sen gånger större än en plan ickeporös yta. Detta gör att titandioxid lagret får en mörk färg efter infärgningen, trots att varje molekyllager av färgämnet absorberar mindre än en procent av det infallande ljuset, se figur 2 sidan 14.

Effektiviteten hos solcellen begränsas av en rad faktorer som till stor del är beroende av gränssnittet mellan de olika materialen. Information om hur de olika materialen är bundna till varandra och påverkar varandra i gränssnit-tet är därför viktiga för att kunna förbättra solcellen ytterligare.

Hålledarmaterial GlasTitandioxidmed färgämne

Ru

ee

Ljus

Hålledare Färgämne

N

NN

N

48

4.3. Experiment och resultat I detta arbete användes olika metoder för att studera solcellerna. Med foto-elektronspektroskopi kunde ytorna av materialen undersökas på en moleky-lär nivå. Till exempel studerades hur färgämnena är bundna till titandioxid ytan, hur de olika materialens yttersta elektronstruktur ser ut, vilken är aktiv i omvandlingen av ljus energi till elektrisk energi och hur materialen i solcel-len samverkar och påverkar varandra.

Dessutom användes metoder som mer direkt mäter de fotoelektriska egenskaperna hos solcellerna. Mätningar av ström och spänning på solcel-lerna vid belysning av en lampa som utsände ljus som liknade solljus, gav t.ex. information om den totala verkningsgraden. IPCE (incident photon to current conversion efficiency) mätningar gav t.ex. information om antalet användbara laddningar som varje foton som träffade solcellen gav upphov till. Teoretiska metoder användes också för att få ökad förståelse för hur hålledarmaterialet leder laddningar och hur färgämnenas elektronstruktur ser ut.

Med en kombination av de olika metoderna var det möjligt att följa hur en förändring på molekylär nivå påverkar effektiviteten hos solcellen. Även mycket små förändringar visade sig påverka solcellens effektivitet väsentligt. Den ökade förståelsen för varför och på vilket sätt dessa molekylära föränd-ringar påverkar effektiviteten ger insikt i hur bättre solceller ska konstrueras.

49

5. Acknowledgements

First I would like to thank my supervisors Håkan Rensmo, Anders Sandell and Hans Siegbahn for support and for sharing knowledge and experience, for believing in me, learning me a lot of science and how to put the experi-mental results in words; My co-workers in our group Anders Henningsson, Boriss Mahrov, Emma Kristensen, Jan Richter, Maria Hedlund, Patrik Karls-son, Ylvi Alfredsson, Ulrik Gelius and Malin Johansson for being great friends that makes it fun to work; Mats Fahlman and Stina Jönsson at the department of Science and Technology in Norrköping, Michael Odelius at the department of Physics at Stockholm University; Declan Ryan; Sven Larsson at the departmernt of Physical Chemistry at Chalmers in Göteborg; Egbert Figgemeier at University of Basel Switzerland; Gerrit Boschloo, Tomas Edvinsson, Jarl Nissfolk, and Anders Hagfeldt at the Center of Mo-lecular Devices at the Royal Institute of Technology in Stockholm for nice cooperation in a friendly atmosphere; All nice persons at the Physics and Physical Chemistry departments in Uppsala and at MAX-lab in Lund; the Swedish Science Council (VR), the Göran Gustafsson Foundation, and the Swedish Foundation for Strategic Research (SSF) for financing; My friends outside work; My parents, brothers. Special thanks to my dearest Sara, you are best.

50

6. References

1. Renewables in Global Energy Supply, International Energy Agancy, IEA, 2006. 2. Energiförsörjningen i Sverige 2003, Energimyndigheten, 2003. 3. K. W. J. Barnham, M. Mazzer, B. Clive, Nature materials 5 (2006) 161-164.4. A. Goetzberger, C. Hebling, H-W Schock, Materials Science and Engi-neering R 40 (2003), 1-46. 5. M. Grätzel, Nature 2001, 414, 338. 6. M. A. Green, Physica E, 14, 2002, 11. 7. D. M. Chapin, C. S. Fuller, G. L. Pearson, J. Appl. Phys. 1954, 25, 676. 8. B. O’Regan and M. Grätzel, Nature (London) 353 (1991) 737-740. 9. U. Bach, D. Lupo, P. Comte, J. E. Moser, F. Weissörtel, J. Salbeck, H. Spreitzer, M. Grätzel, Nature 395 (1998) 583-585. 10. Y. Saito, N. Fukuri, R. Senadeera, T. Kitamura, Y. Wada, S. Yanagida, Electrochem. Commun. 6 (2004) 71-74. 11. G. R. A. Kumara, S. Kaneko, M. Okuya, K. Tennakone, Langmuir, 18 (2002) 10493. 12. K. Tennakone, G. R. R. A. Kumara, A. R. Kumarasinghe, K. G. U. Wi-jayantha, P. M. Sirimanne, Semicond. Sci. Technol. 10, (1995) 1689. 13. P. Wang, S. M. Zakeeruddin, R. Humphry-Baker, J. E. Moser, M. Grät-zel, Adv. Mater, 2003, 15, 24. 14. P. Wang, S. M. Zakeeruddin, , J. E. Moser, M. K. Nazeeruddin, T. Seki-guchi, M. Grätzel, Nat. Mater. 2003, 2, 402. 15. J. Schnadt, P. A. Bruhwiler, L. Pattey, J. N. O’Shea, S. Södergren, M. Odelius, R. Ahuja, O. Karis, M. Bässler, P. Persson, H. Siegbahn, S. Lunell, N. Mårtensson, Nature 418 (2002) 620-623. 16. R. Eichberger, F. Willig, Chem. Phys.141 (1990) 159-173. 17. Y. Tachibana, J. E. Moser, M. Grätzel, D. R. Klog, J. R. Durrant, J. Phys. Chem. 100 (1996) 20056. 18. G. Benkö, P. Myllyperkiö, J. Pan, A. P. Yartsev, V. Sundström, J. Am. Chem. Soc. 125 (2003) 1118-1119. 19. A. Hagfeldt, B. Didriksson, T. Palmqvist, H. Lindström, S. Södergren, H. Rensmo, S.-E. Lindquist, Sol. Energy Mater.Sol. Cells, 31, (1994), 481-488. 20. M. Gratzel, Nature 421 (2003) 586-587. 21. G. Boschloo, J. Lindström, E. Magnusson, A. Holmberg. A. Hagfeldt, J. Photochemistry and Photobiology A:Chemistry 148, 2002, 11-15.

51

22. M. K. Nazeeruddin, A. Kay, I. Rodicio, R. Humphry-Baker, E. Muller, P. Liska, N. Vlachopoulos, M. Grätzel, J. Am. Chem. Soc. 115 (1993) 6382-6390. 23. M. K. Nazeeruddin, P. Péchy, T. Renouard, S. M. Zakeeruddin, R. Humphry-Baker, P. Comte, P. Liska, L. Cevey, E. Costa, V. Shklover, L. Spiccia, G. B. Deacon, C. A. Bignozzi , M. Grätzel, J. Am. Chem. Soc. 2001, 123, 1613-1624. 24. J. Hagen, W. Schaffrath, P. Otschik, R. Fink, A. Bacher, H.W. Schmidt, D. Haarer, Synth. Metals 89 (1997) 215-220. 25. C. Jäger, D. Haarer, B. Peng, M. Thelakkat, Appl. Phys. Lett. 85 (2004) 6185-6187. 26. L. Scmidt-Mende, U. Bach, R. Humphry-Baker, T. Horiuchi, H. Miura, S. Ito, S. Uchida, M. Grätzel, Adv. Materials 17 (2005) 813-815. 27. H. J. Snaith, S. M. Zakeeruddin, L. Schmidt-Mende, C. Klein, M. Grät-zel, Angew. Chem. Int. Ed. 44 (2005) 6413-6417. 28. L. Schmidt-Mende, S. M. Zakeeruddin, M. Grätzel, Appl. Phys. Lett. 86 (2005) 013504. 29. B. O’Regan, F. Lenzmann, R. Muis, J. Wienke, Chem. Mater. 14 (2002) 5023. 30. Y. Saito, T. Kitamura, Y. Wada, S. Yanagida, Synth. Met. 131 (2002) 185-187. 31. G. P. Smestad, S. Spiekermann, J. Kowalik, C. D. Grant, A. M. Schwartzberg, J. Zhang, L. M. Tolbert, E. Moons, Sol.Energy.Mat. Sol. Cells, 76 (2003) 85-105. 32. C. D. Grant, A. M. Schwartzberg, G. P. Smestad, J Kowalik, L. M. Tol-bert, J. Z. Zhang, Synth. Met. 132(2) (2003) 197-204. 33. S. Spiekermann, G. Smestad, J. Kowalik, L. M. Tolbert and M. Grätzel, Synth. Met. 121 (2001) 1603-1604. 34. H. Rensmo, S. Södergren, L. Pattey, K. Westermark, L. Vayssieres, O. Kohle, P.A. Brühwiler, A. Hagfeldt, H. Siegbahn, Chem. Phys. Lett. 274 (1997) 51-57. 35. H. Rensmo, K. Westermark, S. Södergren, O. Kohle, P. Persson, S. Lu-nell, H. Siegbahn, J. Chem. Phys. 111 (1999) 2744-2750. 36. K. Westermark, H. Rensmo, H. Siegbahn, K. Keis, A. Hagfeldt, L. Oja-mae, P. Persson, J. Phys. Chem. 106 (2002) 10102. 37. K. Westermark, H. Rensmo, J. Schnadt, P. Persson, S. Södergren, P.A. Brühwiler, S. Lunell and H. Siegbahn, J. Chem. Phys. 285 (2002) 167. 38. K. Westermark, H. Rensmo, H. Siegbahn, K. Keis, A. Hagfeldt, L. Oja-mae, P. Persson, J. Phys. Chem. B 106(39) (2002), 10102-10107. 39. P. G. Karlsson, S. Bolik, J. H. Richter, B. Mahrov, E. M. J. Johansson, J. Blomquist, P. Uvdal, H. Rensmo, H. Siegbahn and A. Sandell, J. Chem. Phys., 120(23), 2004, 11224-11232. 40. B. Mahrov, G. Boschloo, A. Hagfeldt, H. Siegbahn, H. Rensmo, J. Phys. Chem. B, 2004, 108, 11604-11610.

52

41. E.M.J. Johansson, A. Sandell, H. Siegbahn, H. Rensmo, B. Mahrov, G. Boschloo, E. Figgemeier, A. Hagfeldt, S.K.M.Jönsson, M. Fahlman, Synth. Met. 149 (2005) 157-167. 42. E.M.J. Johansson, M. Hedlund, H. Siegbahn, H. Rensmo, J. Phys. Chem. B, 109 (47) (2005) 22256-22263. 43. G. Liu, A. Klein, A. Thissen, W. Jaegermann, Surface Science, 539, 2003, 37-48. 44. Y. V. Zubavichus, Yu. L. Slovokhotov, M. K. Nazeeruddin, S. M. Zakeeruddin, M. Grätzel and V. Shklover, Chem. Mater. 14, 2002, 3556-3563. 45. A. Einstein, Ann. Phys. 1905, 17, 132. 46. H. Siegbahn and L. Karlsson, Photoelectron Spectroscopy, Springer Ver-lag: Berlin Heidelberg, 1982, Vol.XXXI. 47. S. Hüfner, Photoelectron Spectroscopy, 3rd edition, Springer-Verlag (2003).48. O. Björneholm, A. Nilsson, A. Sandell, B. Hernnas, N. Martensson, Phys. Rev. Lett. 68 (12), 1892 (1992). 49. P. A. Bruhwiler, A. J. Maxwell, A. Nilsson, R. L. Whetten, N. Martens-son, Chem. Phys. Lett. 193(5), 311 (1992). 50. a) S. Svensson, J. –O. Forsell, H. Siegbahn, A. Ausmees, G. Bray, S. Södergren, S. Sundin, S.J. Osborne, S. Aksela, E. Nömmiste, J. Jauhiainen, Mjurvansuu, J. Karvonen, Pbarta, W.R. Salaneck, A. Evaldsson, M. Lögd-lund, and A. Fahlman, Rev. Sci. Instr 67 (1996) 2149. b) M. Bässler, J-.O. Forsell, O. Björneholm, R. Feifel, M. Jurvansuu, S. Ak-sela, S. Sundin, S.L. Sorensen, R. Nyholm, A. Ausmees, S. Svensson, J. El. Spectr. Rel. Phen. 101-103 (1999) 953. 51. E. Hückel, Zeits. Physik. 60 (1930) 423. 52. L. Kavan, M. Grätzel, Electrochim. Acta, 40 (1995) 643. 53. A. Sandell, M. P. Andersson, M. K. J. Johansson, P.G. Karlsson, Y. Al-fredsson, J. Schnadt, H. Siegbahn, P. Uvdal, Surf. Science 530 (2003) 63-70. 54. M. K. Nazeeruddin, S. M. Zakeeruddin, R. Humphry-Baker, M. Jirouek, P. Liska, N. Vlachopoulos, V. Shklover, C. H. Fisher, M. Grätzel, Inorg. Chem. 1999, 38, 6298-6305. 55. M. K. Nazeeruddin, R. Humphry-Baker, P. Liska and M. Grätzel, J. Phys. Chem. B, 2003, 107, 8981-8987. 56. M. K. Nazeeruddin, M. Amirnasr, P. Comte, J. R. Mackay, A. J. McQuillan, R. Houriet, M. Grätzel, Langmuir, 2000, 16, 8525-8528. 57. A. Fillinger and B. A. Parkinson, J. Elec.Chem. Soc., 146(12), 1999, 4559-4564. 58. K.S. Finnie, J. R. Bartlett and J. L. Woolfrey, Langmuir, 14, 1998, 2744-2749. 59. K. Murakoshi, G. Kano, Y. Wada, S. Yanagida, H. Miyazaki, M. Ma-tsumoto, S. Murasawa, J. Electroanalytical Chem. 396 (1995) 27-34.

53

60. L. Patthey, H. Rensmo, P. Persson, K. Westermark, L. Vayssieres, A. Stashans, Å. Petersson, P. A. Bruhwiler, H. Siegbahn, S. Lunell, N. Mårtens-son, J. Chem. Phys. 110 (1999) 5913. 61. J. Schnadt, A. Henningsson, M. P. Andersson, P. G. Karlsson, P. Uvdahl, H. Siegbahn, P. A. Bruhwiler, A. Sandell, J. Phys. Chem. B, 108(10), 2004, 3114-3122. 62. V. Shklover, Yu. E. Ovchinnikov, L. S. Braginsky, S. M. Zakeeruddin and M. Grätzel, Chem. Mater. 1998, 10, 2533-2541. 63. H. Rensmo, S. Lunell, H. Siegbahn, J. Photochem. Photobio. A: Chemis-try 114 (1998) 117-124. 64. F. Aiga and T. Tada, J. Mol. Structure, 658, 2003, 25-32. 65. D.P. Hagberg, T. Edvinsson, T. Marinado, G. Boschloo, A. Hagfeldt and L. Sun, Chem. Comm, in press. 66. O. Kohle, theses no.1532, Lausanne, EPFL, 1996. 67. O. Kohle, S. Ruile, and M. Grätzel, Inorg. Chem. 35 (1996) 4779. 68. S. Ruile, O. Kohle, P. Pechy, and M. Grätzel, Inorg. Chim. Acta 261 (1997) 129. 69. F. Jonas and L. Schrader, Synth. Met. 41-43 (1991) 831. 70. G. Heywang and F. Jonas, Adv.Mater. 4 (1992) 116. 71. A.N. Aleshin, S.R. Williams, A.J. Heeger, Synth. Met. 94 (1998) 173-177.72. M. Dietrich, J. Heinze, G. Heywang, F. Jonas, J. Elec. Chem. 369 (1994) 87-92.73. Q.Pei, G.Zuccarello, M. Ahlskog, O.Inganäs, Polymer 35 (1994) 1347. 74. Y. Cao, G.Yu, C. Zhang, R. Menon, A.J. Heeger, Synth. Met. 87 (1997) 171-174. 75. G. Greczynski, Th. Kugler, M. Keil, W. Osikowicz, M. Fahlman, W. R. Salaneck, J. Elec. Spec. And Rel. Phen.121 (2001) 1-17. 76. S.K.M. Jönsson, J. Birgerson, X. Crispin, G. Greczynski, W. Osikowicz, A.W. Denier van der Gon, W.R. Salaneck, M. Fahlman, Synth. Met. 139 (2003) 1-10. 77. N. F. Mott, Metal-Insulator Transitions, Taylor and Francis, 1990. 78. J. Jortner, J. Am. Chem. Soc. 102 (1980) 6676. 79. T. Holstein, Ann. Phys. (NY) 8 (1959) 325, 343. 80. R. A. Marcus, J. Chem. Phys. 24 (1956), 966-978. 81. Current Opinion in Solid State and Materials Science, 6(1) (2002) 87-95.

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 188

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 through theseries 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-6892

ACTAUNIVERSITATISUPSALIENSISUPPSALA2006