ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2017

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1464

Exploring Electronic Processesat the Mesoporous TiO2/Dye/Electrolyte Interface

WENXING YANG

ISSN 1651-6214ISBN 978-91-554-9780-4urn:nbn:se:uu:diva-310191

Dissertation presented at Uppsala University to be publicly examined in Häggsalen,Ångströmlaboratoriet, Lägerhyddsvägen 1, Uppsala, Friday, 10 February 2017 at 10:15 forthe degree of Doctor of Philosophy. The examination will be conducted in English. Facultyexaminer: Professor Gerald Meyer (University of North Carolina at Chapel Hill).

AbstractYang, W. 2017. Exploring Electronic Processes at the Mesoporous TiO2/Dye/ElectrolyteInterface. Digital Comprehensive Summaries of Uppsala Dissertations from the Facultyof Science and Technology 1464. 86 pp. Uppsala: Acta Universitatis Upsaliensis.ISBN 978-91-554-9780-4.

Dye sensitized solar cells (DSSCs) are an attractive way to convert light into electricity. Itsdevelopment requires a detailed understanding and kinetic optimization of various electronicprocesses, especially those occurring at the mesoporous TiO2/dye/electrolyte interface. Thisdissertation work is focused on the exploration of the various electronic processes at thesensitized-electrode/electrolyte interface by using various electrochemical and photochemicalmethods.

Firstly, an alternative redox couple—TEMPO/TEMPO·+ with a relatively high positive redoxpotential—is explored, aiming to reduce the energy loss during the dye regeneration process.Despite the fast dye regeneration, the charge recombination between the electrons in theconduction band of mesoporous TiO2 and the oxidized redox species is found to be the limitingfactor of the device. Further, a more efficient tandem-electrolyte system is developed, leadingto DSSCs with the power conversion efficiency of 10.5 % and 11.7 % at 1 sun and 0.5-sunillumination, respectively. An electron-transfer cascade process during dye regeneration by theredox mediators is discovered to be beneficial. Further stability studies on the device suggestthe crucial role of TiO2/dye/electrolyte interfaces in the long-term stability of cobalt bipyridylelectrolyte-based DSSCs.

On the fundamental level, the local electric field and Stark effects at the TiO2/dye/electrolyteinterface are investigated in various aspects—including the charge compensation mechanism,the factors affecting the electric field strength, as well as its impact on charge transfer kinetics.These results give further insights about the TiO2/dye/electrolyte interface, and contribute to thefurther development and understanding of DSSCs.

Keywords: dye-sensitized solar cells, dye regeneration, Stark effect, the local electric field,cationic effect

Wenxing Yang, Department of Chemistry - Ångström, Physical Chemistry, Box 523, UppsalaUniversity, SE-75120 Uppsala, Sweden.

© Wenxing Yang 2017

ISSN 1651-6214ISBN 978-91-554-9780-4urn:nbn:se:uu:diva-310191 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-310191)

To my parents

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals. Reprints were made with permission from the respective publishers.

I Yang, W., Vlachopoulos, N., Hao, Y., Hagfeldt, A., Boschloo,

G. Efficient Dye Regeneration at Low Driving Force Achieved in Triphenylamine Dye LEG4 and TEMPO Redox Media-tor Based Dye-Sensitized Solar Cells. Phys. Chem. Chem. Phys, 17: 15868–15875 (2015)

II Hao, Y.*, Yang, W.*, Zhang, L., Jiang, R., Mijangos, E., Saygili, Y., Hammarström, L., Hagfeldt, A., Boschloo, G. A Small Electron Donor in Cobalt Complex Electrolyte Sig-nificantly Improves Efficiency in Dye-Sensitized Solar Cells Nature Communication, Accepted. (2016)

III Yang, W., Hao, Y., Ghamgosar, P., Boschloo, G. Thermal Stability Study of Dye-Sensitized Solar Cells with Cobalt Bipyridyl–based Electrolytes. Electrochimica Acta, 213: 879–886 (2016)

IV Yang, W., Pazoki, M., Eriksson, A. I. K., Hao, Y., Boschloo, G. A Key Discovery at the TiO2/Dye/Electrolyte Interface: Slow Local Charge Compensation and a Reversible Electric Field. Phys. Chem. Chem. Phys, 17: 16744–16751 (2015)

V Yang, W., Hao, Y., Vlachopoulos, N., Eriksson, A. I. K., Boschloo G. Studies on the Interfacial Electric Field and Stark Effect at the TiO2/Dye/Electrolyte Interface. J. Phys. Chem. C 120: 22215–22224 (2016)

VI Yang, W., Vlachopoulos, N., Boschloo, G.. Impact of Local Electric Fields on Charge-Transfer Pro-cesses at the TiO2/Dye/Electrolyte Interface ACS Energy Letter, Accepted. (2016) DOI:10.1021/acsenergylett.6b00568

Comments on my contribution:

I am the main responsible person for papers I, III-VI and shared the first authorship for paper II. Paper I: I did all the experiments, analysis of the data, and wrote the manu-script with contribution from all the coauthors. Paper II: I did most of the spectroscopic and electrochemical characteriza-tion and analysis of the data. I wrote the manuscript jointly with Dr. Yan Hao, with contribution from all the coauthors. Paper III: I did most of the experiments, analyzed the data and wrote the manuscript with contribution from all the coauthors. Ghamgosar Pedram conducted the differential pulse voltammetry measurements. Paper IV: I did all the experiments, analyzed the data and wrote the manu-script with contribution from all the coauthors. Paper V: I did most of the experiments, analyzed the data and wrote the manuscript with contribution from all the coauthors. Paper VI: I did all the experiments, analyzed the data and wrote the manu-script with contribution from all the coauthors.

I am also an author or co-author of the following papers, which are not in-cluded in this thesis:

1. Yang, W., Söderberg, M., Eriksson, A. I. K., Boschloo, G. Efficient

Aqueous Dye-Sensitized Solar Cell Electrolytes Based on TEM-PO/TEMPO+ Redox Couple. RSC Adv. 1: 26706–26709 (2015)

2. Yang, W.*, Eriksson, A. I. K*, Oscarsson, J., Edvinsson, T., Rensmo, H., Boschloo, G. Impacts of Deprotonation on Organic D--A Dyes with a Carboxylic Anchoring Group: Physical Properties, Sensitization Process and DSSCs. To be submitted.

3. Yang, W., Boschloo, G. Conductivity Studies of Nanostructured TiO2 Films Permeated with Electrolyte: Comparison Between Dye-sensitized Films and Non-sensitized films Manuscript in preparation.

4. Gao, J., Yang, W., Pazoki, M., Boschloo, G., Kloo, L. Cation-Dependent Photostability of Co(II/III)-Mediated Dye-Sensitized So-lar Cells. J. Phy. Chem. C 119: 24704–24713 (2015)

5. Gao, J., Yang, W., Ahmed, M. E., Govind, K. P., Yuan, F., Jing, D., Per

H. S., István, F., Boschloo, G, Kloo, L. Light-induced Cobalt Tris(bipyridine) Ligand Exchange: A Simple Strategy Solving Diffu-sion-Controlled Hysteresis Towards Efficient and Durable Dye-Sensitized Solar Cells

To be summited.

6. Freitag, M., Yang, W., Fredin, L. A., Luca D., Karlsson K. M., Hagfeldt, A., Boschloo, G. Supramolecular Hemicage Cobalt Mediators for Dye-Sensitized Solar Cells. ChemPhysChem 56: 607–622 (2016)

7. Freitag, M., Giordano, F., Yang, W., Pazoki, M., Hao, Y., Zietz, B.,

Grätzel, M., Hagfeldt, A., Boschloo G. Copper Phenanthroline as a Fast and High-Performance Redox Mediator for Dye-Sensitized So-lar Cells. J. Phys. Chem. C 120: 9595–9603 (2016)

8. Hao, Y., Tian, H., Cong, J., Yang, W., Bora, I., Sun, L., Boschloo, G., Hagfeldt A. Triphenylamine Groups Improve Blocking Behavior of Phenoxazine Dyes in Cobalt-Electrolyte-Based Dye-Sensitized Solar Cells. ChemPhysChem 15: 3476–3483 (2014)

9. Hao, Y., Gabrielsson, E., Lohse, P. W., Yang, W., Johansson, E., Hag-feldt, A., Sun, L., Boschloo, G. Peripheral Hole Acceptor Moieties on an Organic Dye Improve Dye-Sensitized Solar Cell Performance. Advanced Science 2: 1500174 (2015)

Contents

1. Introduction ............................................................................................... 131.1 Background ........................................................................................ 131.2 Photovoltaics ...................................................................................... 141.3 Aim of the thesis ................................................................................ 161.4 Outline of the thesis ........................................................................... 17

2. Dye-sensitized solar cells (DSSCs) .......................................................... 182.1 Working principle .............................................................................. 182.2 Fabrication procedure ......................................................................... 192.3 State-of-art DSSCs and related-materials .......................................... 202.4 Maximum efficiency and spectral losses............................................ 21

3. Meso-TiO2/dye/electrolyte interface ......................................................... 253.1 Semiconductors in the photoelectrochemical cells ............................ 253.2 Nanocrystalline semiconductor-liquid junction ................................. 263.3 Electronic properties and charge transport mechanism ...................... 28

3.3.1 Electronic properties ................................................................... 283.3.2 Charge transport mechanism ...................................................... 30

3.4 The local electric field and Stark effect ............................................. 323.4.1 Double layer ............................................................................... 323.4.2 Double layer at meso-TiO2/dye/electrolyte interface ................. 333.4.3 Stark effect .................................................................................. 35

4. Experimental methods .............................................................................. 384.1 Characterization of single components .............................................. 38

4.1.1 UV-Vis spectroscopy .................................................................. 384.1.2 Electrochemical techniques ........................................................ 38

4.2 Characterization of solar cells performance ....................................... 414.2.1 Current-voltage (I-V) characterization ........................................ 414.2.2 Incident photon to current efficiency (IPCE) characterization ... 42

4.3 Tool-box ............................................................................................. 444.3.1 Small-modulation transient photovoltage technique .................. 444.3.2 Small-modulation transient photocurrent technique ................... 454.3.3 Charge extraction technique ....................................................... 45

4.4 Photoinduced absorption (PIA) spectroscopy .................................... 464.4.1 Time-resolved photo-induced absorption traces ......................... 474.4.2 Photoinduced absorption spectrum ............................................. 47

4.5 Transient absorption spectroscopy (TAS) .......................................... 484.6 Kinetic analysis .................................................................................. 51

4.6.1 Kohlrausch-Williams-Watts (KWW) function ........................... 514.6.2 Half time ..................................................................................... 524.6.3 Double or multiple exponential fittings ...................................... 53

5. Results and discussion .............................................................................. 555.1 Strategies to improve the performance of DSSCs .............................. 55

5.1.1 Alternative one electron redox system ....................................... 555.1.2 One-electron or two-electron transfer agent? ............................. 58

5.2 The stability of cobalt-bipyridyl electrolyte based DSSCs ................ 605.3 The Local Electric Field and Stark Effect .......................................... 63

5.3.1 Charge compensation ................................................................. 645.3.2 Factors affecting the local electric field ...................................... 675.3.3 Effects of local electric field on the interfacial charge transfer .. 71

6. Summary and outlook ............................................................................... 74

7. Populärvetenskaplig sammanfattning på svenska ..................................... 76

8. Acknowledgements ................................................................................... 79

8. References ................................................................................................. 81

Abbreviations

ACN Acetonitrile

CV Cyclic Voltammetry

Co(bpy)32+/3+ tris(2,2’-bipyridine) cobalt(II/III)

DSSCs Dye-sensitized solar cells

D149 5-[[4-[4-(2,2-Diphenylethenyl)phenyl]-1,2,3-3a,4,8b-hexahydrocyclopent[b]indol-7-yl]methylene]-2-(3-ethyl-4-oxo-2-thioxo-5-thiazolidinylidene)-4-oxo-3-thiazolidineacetic acid

D--A Donor- conjugated linker-acceptor

EC Conduction band edge energy

EF Fermi level

EF,redox Fermi level of the redox couple

EV Valence band edge energy

FF Fill factor

FTO Fluorine doped tin oxide

HL Helmholtz layer

HOMO Highest occupied molecular orbital

IPCE Incident photon to current efficiency

IV Current-voltage

JSC Short-circuit current density

MPN 3-methoxypropionitrile

LEG4 3-{6-{4-[bis(2’,4’-dibutyloxybiphenyl-4-yl)amino-]phenyl}-4,4-dihexyl-cyclopenta-[2,1-b:3,4-b’]dithiophene-2-yl}-2-cyanoacrylic acid

LEG4/TiO2 LEG4 sensitized TiO2

LUMO Lowest unoccupied molecular orbital

Meso-TiO2 Mesoporous TiO2

PIA Photo induced absorption

TAS Transient absorption spectroscopy

TPAA Tris(4-methoxyphenyl)amine

TEMPO 2,2,6,6-Tetramethyl-1-piperidinyloxy

VOC Open-circuit voltage

13

1. Introduction

1.1 Background During the last hundred years, the world has experienced dramatic changes due to industrialization, with substantial improvements in the living stand-ards of human beings’; however, the man-made climate changes also have become more severe than ever. For example, global warming promotes the rising of the sea level at an estimated rate of 0.13 inches (3.2 millimeters) per year since the early 1990s.(1) As a result, thousands of coastline cities, including Venice, are now at a risk of disappearing. These detrimental changes mostly originate from the tremendous consumption of fossil fuels, including oil, coal and natural gas, which release substantial amounts of CO2 into the air after combustion, accounting for more than 60% of the green house effects.(2) To change the projecture of the man-made climate changes, we must promote eco-friendly industrial production and revolutionize the current lifestyle through technological innovations, e.g., electrical vehicles and smart indoor construction. Ultimately, a sustainable society cannot be created without the development of clean and renewable energy sources.

Figure 1. The global energy consumption structure in 2014. Source: BP statistical review of world energy

At the moment, fossil fuels, unfortunately, still dominate the global energy supply, accounting for more than 86% (Figure 1). The large investments on renewable energies, especially in the developed countries, are improving the

14

energy structure, but it is still not sufficient. A large part of the CO2 emission may transfer to the developing countries through the global trade(3), where the energy structure is still based on the fossil fuels. Meanwhile, with the emerging of economies with a large population, the emission of CO2 would increase unavoidably due to the higher life standard in these regions. We are therefore at an age that demands urgent international collaboration and tech-nological development to maintain a sustainable earth for the future genera-tions.

Compared with other renewable energies, e.g.,, wind, biomass and nucle-ar power, solar energy is widely considered as the most promising energy resource; it is clean, safe and by far the largest exploitable resource, provid-ing more energy in 1 hour to the earth than all of the energy consumed by humans in an entire year.(4) There are many ways to harvest solar energy, e.g., solar thermal plants(5) and water splitting devices(6, 7). However, pho-tovoltaics, which convert solar energy into electricity, are one of the most promising means of solar energy conversion; the generated electricity can be easily integrated into the well-established electronic grid systems and readily distributed.

1.2 Photovoltaics In 1839, Edmund Becquerel, a French physicist, discovered the photovoltaic effect while experimenting with an electrolytic cell made up of two metal electrodes.(8) Since then, several approaches have been investigated for harvesting solar energy and turning it into electricity. As a result of enor-mous efforts, various photovoltaic systems(9, 10) are available nowadays, including traditional silicon solar cells(11), thin-film solar cells(12), organic solar cells(13), dye-sensitized solar cells (DSSCs)(14) as well as the recently emerging perovskite solar cells(15, 16). The efficiencies of these technolo-gies span a large range as shown in Figure 2. Some of them have already achieved great success and are being commercialized, e.g., the silicon solar cells; some may have important application in spacecraft, e.g., GaAs solar cells; while others may be still at the laboratory stage, e.g., perovskite solar cells.

However, of great importance is that solar technologies demand diversity, with respect to not only the greatest efficiency and stability at a cost as low as possible, but also to more subtle properties, e.g., transparency and flexi-bility. The eventual applicability of these technologies depends on a large variety of factors, e.g., political will, costs, durability and marketing. From a scientific point of view, advancing and developing the current technology is important to promote photovoltaics, but elucidating and establishing the

F

igur

e 2.

The

rec

ord

effi

cien

cies

for

dif

fere

nt ty

pe o

f so

lar

cells

. Sou

rce:

Nat

iona

l Ren

ewab

le E

nerg

y L

abor

ator

y.

http

://w

ww

.nre

l.gov

/ncp

v/ (

retr

ieve

d 20

16-0

9-29

)

16

general scientific principles will certainly benefit the society in even longer term.

Figure 3. The solar spectrum, taken from ASTM G173-03 reference spectra.

This thesis is dedicated to dye-sensitized solar cells (DSSCs). Its basic prin-ciple was originally investigated by several groups, notably by those of Ger-ischer(17, 18) and Memming(19, 20), in 1960-1985. In 1991, O’Regan and Gräztel, following the initial research in Lausanne on dye-sensitization of both colloidal TiO2 dispersions and porous TiO2 electrodes in the late 1980s and early 1990s(21, 22), made a breakthrough using colloidal mesoporous TiO2 (meso-TiO2), leading to a remarkable solar power conversion efficiency of (~7%).(23) Nowadays, with continuous research efforts, the record DSSCs performance has been boosted up to a record efficiency of 14%.(24) Although still not as efficient as the silicon solar cells, DSSCs have several advantages including low cost, ease of fabrication and efficient operation under indoor light condition. The solar cells can also be made from dyes with different colors, which therefore allow designs that are more authentic. The future application of these devices stipulates the overcome of the effi-ciency bottleneck. To achieve this, we still need better fundamental under-standing of these devices.(25)

1.3 Aim of the thesis The dye-sensitized mesoporous TiO2/electrolyte interface is the most im-portant interface in DSSCs, where multiple electronic processes, including charge separation, transportation, recombination, etc., occur in the simulta-

17

neous and complicated way. The overall aim of the thesis is to understand these electronic processes occurring at the interface.

More specific aims are:

Optimization of the dye regeneration process in order to reduce the recombination loss and improve the solar cell performance.

Characterization of electronic changes of the photoanode during the stability test in order to understand factors limiting the stabil-ity

Investigation of the local electric field, generated by photon in-duced electron accumulation within TiO2 and the surrounding cat-ions, and Stark effect at this interface.

Although these studies are conducted on the meso-TiO2/dye/electrolyte in-terface, the achieved insights are also expected to be applicable to a wider range of sensitized-semiconductor/electrolyte interface.

1.4 Outline of the thesis Chapter 2 introduces the device structure of a DSSC, together with its fabri-cation procedures, working principles and state-of-art components. It is fol-lowed by a discussion on the maximum achievable efficiency that illustrates the key limiting factors of the performance of the DSSCs.

Chapter 3 discusses some of the most important fundamental features of the TiO2/dye/electrolyte interface. These aspects are of extreme importance for a deep understanding of the unique operation principles of DSSCs. The special focus is devoted to the discussions of the concept of the local electric field and Stark effect, relevant to the results of this thesis.

Chapter 4 briefly introduces the experimental techniques used in the the-sis.

Chapter 5 summarizes the key experimental results obtained during my PhD studies. The attached articles at the end of the thesis describe these re-sults into details.

Finally, Chapter 6 gives the summary and outlook for the future work.

18

2. Dye-sensitized solar cells (DSSCs)

2.1 Working principle

Figure 4. (a) Schematical picture of a typical DSSC (b) The electronic processes occurring in the DSSCs. Black solid line: the desired electron transfer processes; Red dash lines: the back recombination reactions in DSSCs.

The device architecture of a typical liquid DSSC is schematically shown in Figure 4a, which consists of a dye-sensitized mesoporous TiO2 (meso-TiO2) electrode, a redox electrolyte and a counter electrode (more about the vari-ous components will be discussed in the Section 2.3). Its general working principle is demonstrated in Figure 4b, as described below:

Dye excitation and electron injection into meso-TiO2: When the DSSCs are exposed to light illumination, the dye molecules anchored to the meso-TiO2 absorb the photons and get excited. This excitation process pro-motes electrons from the highest occupied molecular orbital level (HOMO) of the dye to its lowest unoccupied molecular orbital level (LUMO) (Step 1). The electrons in the LUMO level will then be rapidly injected into the con-duction band of the meso-TiO2, due to the strong electronic coupling be-tween the dye and the conduction band of TiO2 (Step 2).

Electron transport: The injected electrons will move through the meso-TiO2 network by diffusion and multiple-trapping processes, and eventually,

(a) (b)

19

will be collected at the back contact through the external circuit (Step 3) (this transport mechanism will be described into details in Section 3.2.).

Redox mediators: When the electrons are transferred to the counter elec-trodes, they will reduce the oxidative redox species to the reductive redox species (Step 4). The reduced redox species will then diffuse to the meso-TiO2/dye/electrolyte region and regenerate the oxidized dye by Step 5.

By completing the aforementioned processes, the solar-to-electricity cycle finishes, where a photon has been converted into an energetic electron ready for producing electricity.

Back reactions: However, the above steps only described the desired processes in DSSCs. In the real operation, there are also several undesired reactions happening at the same time. The excited dye molecules may lose its energy before injecting electrons into the TiO2, e.g., through direct radia-tive or non-radiative recombination (Step 6). Moreover, the oxidized redox species in the electrolyte (Step 7) or the oxidative dye molecules (Step 8) may capture the electrons within the meso-TiO2 during its transport. Similar-ly, the electrons passing through the FTO glasses can also be captured by the oxidative redox species (not shown for clarity). All these undesired process-es result in a loss of electrons before their accomplishment of the entire so-lar-to-electricity cycle. As a result, the photon energy from the sun is con-verted into the thermal energy dissipating back to the environment instead of producing the electricity.

2.2 Fabrication procedure

Figure 5. Schematic picture of the typical fabrication processes of DSSCs.

20

The state-of-art fabrication procedure of DSSCs in our lab is schematically shown in the Figure 5, with the preparation details described below.

Working electrodes: Briefly, the cleaned fluorine-doped tin oxide (FTO) glasses are firstly treated with TiCl4 aqueous solution in order to grow a thin layer of TiO2 on top of FTO glasses. This step is to reduce the surface roughness of FTO and prevent the back reaction of electrons from the bare FTO to oxidized species in the electrolyte. Then, a mesoporous TiO2 (meso-TiO2) layer is screen-printed onto the FTO glasses and sintered. Finally, the film is treated again with TiCl4 solution. This procedure has been well stud-ied and shown significant effects on the DSSCs performance mainly due to its control of the conduction band position.(26) After sintering, the elec-trodes are put into a dye bath solution for sensitization (0.2 mM dye in ACN solution), normally overnight.

Counter electrodes: Cleaned FTO glass pieces are normally coated with conducting materials, like platinum, polymers or graphene.

Assembling working and counter electrodes: After preparing both working electrodes and counter electrodes, the solar cells are fabricated by sandwiching them and sealing them by use of a thermoplastic Surlyn frame.

Injection of electrolytes: The electrolyte solution is vacuum-injected into the solar cells through a predrilled hole on the backside of the counter elec-trode. The back hole is then sealed to avoid any solvent evaporation.

Silver contacts: The silver pastes are finally put on top of the glasses, in order to improve the connection between the DSSCs and the measurement clamps, and also to decrease the resistivity of the systems.

2.3 State-of-art DSSCs and related-materials Short comments about the state-of-art components of DSSCs will be given in this section, with reference to excellent reviews available in the literature.(14) The latest certificated efficiency of DSSCs is 11.9%(27) from Sharp cooperation in 2012, with designed area 1.00 cm2. Recently, with the advance of the cobalt-based electrolyte, the record DSSCs efficiency has been reported to reach 13% in 2014(28) and later advanced to 14.3% in 2015(24) mainly by engineering the sensitization film by efficient sensitizers or the cosensitization technique(29).

Mesoporous network: The dominant photoanode applied in DSSCs is still the meso-TiO2 as it was initially used in 1991.(23) Efforts have been dedicated to look for materials with higher electron mobility (e.g., ZnO(30, 31)) or better controlled morphology (e.g., nanowire(32, 33)) in order to optimize the electron transportation through the working electrode. Howev-er, the champion DSSCs have been so far only fabricated with the meso-

21

TiO2. These results imply that a single optimized parameter of the materials is often not sufficient to guarantee the good performance of DSSCs due to its complexity. The desired electrode materials should possess, at least, suitable band edge energy, high electron mobility, large surface dye coverage and slow recombination kinetics.

Dyes: There are plenty of efficient dyes emerging from development of DSSCs. The most efficient ones include ruthenium dyes(34, 35), triphenyl-amine-based dyes(36), porphyrin dye(28), etc. Nowadays, the metal-free organic dyes have achieved comparable or even higher efficiency than the traditional ruthenium dyes(36–38), which corresponds to one of the largest achievement of material development in DSSCs during the last decades.

Redox mediators: The redox mediators in the electrolyte undertake the regeneration of the oxidized dye at the TiO2/dye/electrolyte interface. Then, the oxidized redox mediators diffuse to the counter electrode, where they are converted back to the reduced form. The I-/I3

- redox mediator has dominated the DSSCs for more than 20 years. From 2010, the cobalt bipyridyl-based redox mediator proves to be capable of producing DSSCs with comparable efficiencies. From then on, with continuous research efforts, the cobalt-based electrolytes have now become the dominating electrolyte components, which give the above 13% and 14.3% record efficiency for liquid based DSSCs.

Counter electrodes: The counter electrodes of DSSCs have varied a lot in the literature. The platinized counter electrode was originally used in 1991; it is, though, still one of the most efficient counter electrodes. Alterna-tive counter electrodes have been proposed, e.g., graphene nanoplatelets(39) and PEDOT(40). They have been reported to be superior the platinized counter electrode in certain systems; however, the use of the counter elec-trode across the literature, even from the same group, differs from publica-tion to publication. Its functionality depends on the redox systems used.

In recent years, there is an emerging interest in the development of solid-states DSSCs(41), where the liquid electrolyte is replaced with a solid-state hole transport material. It is promising for fabrication of flexible solar cells and avoids the leakage problems caused by the liquid electrolyte. The state-of-art efficiency is around 8.5%, with use of an inorganic hole transport ma-terial SnPbI3.(42)

2.4 Maximum efficiency and spectral losses All the undesired processes in section 2.2 cause energy losses in DSSCs. Through proper optimization, their impact would be lowered but hardly re-moved. Nevertheless, it is still very inspiring to ask what the maximum

22

achievable efficiency of DSSCs is, if all the undesired reactions can be re-moved.

Figure 6. Spectral loss mechanism. (a) the loss of photon flux: only photons with higher energy than the absorption band edge will be absorbed, while the rest of the photons are lost. The blue line shows the integrated maximum current as a function of the absorption edge, assuming a 100% photon to electron conversion efficiency. Take the absorption band edge at 800 nm for example, the maximum achievable current is 27. 27 mAcm-2 at AM 1.5 G illumination. (b) the loss of hot carriers ener-gy. The absorbed photons with energy higher than the bandgap (hot carriers) will be rapidly relaxed to the band edge. This causes additional energy loss in the system. The left panel shows the case of silicon solar cells, and the right panel shows the case of DSSCs. In DSSCs, instead of relaxing to the absorption band edge, there is additional loss due to the dye injection and regeneration processes, Eloss,inj and Eloss,reg, respectively. CB: conduction band edge. VB: valence band edge.

In 1961, William Shockley and Hans-Joachim Queisser, on the basis of vari-ous thermodynamic considerations, firstly calculated the maximum efficien-cy of solar cells based on a single p-n junction to be around 33%.(43) Here, one can only consider the spectral loss to estimate the maximum achievable efficiency of DSSCs, which will be certainly higher than the real achievable efficiency, but it illustrates the importance to reduce certain energy losses.

(b)

(a)

23

The spectral loss is calculated by the consideration that: (a) only photons with energy higher than that of the band edge will be absorbed (Figure 6a) and (b) photons with energy higher than the bandgap will be rapidly relaxed to the band edge. For the traditional p-n solar cells (Figure 6b left), the effi-ciency is therefore limited to ~ 44% (sometimes called ultimate efficiency) and calculated by the equation 1:

(1)

where Emax and Emin are the maximum and minimum photon energy in the solar spectrum, respectively. Eedge is the photon energy at the absorption band edge. (E) is the photon flux.

However, the case in DSSCs is very different, where the electrons and holes are not only relaxed to their absorption band edge, i.e. HOMO and LUMO, respectively (Figure 6b right). Instead, the electron will relax to the conduction band edge of TiO2, while the hole will relax to the oxidized level of the electrolyte-dissolved redox mediator. Consequently, additional energy losses, Eloss, will be caused by the required driving forces during the injec-tion Eloss,inj and regeneration Eloss,reg processes, i.e. Eloss = Eloss,inj +

Eloss,reg. The ultimate efficiency can now be estimated by the equation 2:

(2)

Figure 7a shows the theoretical ultimate efficiency of DSSCs as a function of the bandgap of the dye molecules at different Eloss. The increase of Eloss leads to a significant decrease of the efficiencies of DSSCs irrespective of the absorption onset of the dye molecules. For instance, by simply increasing the energy loss from around 0 eV to 1.2 eV, the ultimate efficiency of DSSCs drops from 50% to 14% (Figure 7b).

24

Figure 7. (a) The ultimate efficiency calculated for DSSCs as a function of the ener-gy loss during the injection and regeneration processes (b) the maximum ultimate efficiency versus Eloss.

In reality, this analysis is limited because the absorption in molecules is not the same as those in the case of semiconductor. The former normally has a Gaussian type absorption profile, while the latter has a clear band-edge on-set. This will lower the calculated ultimate efficiency for systems based on molecular light absorbers, although co-sensitization from two dyes may im-prove the spectral coverage. Nevertheless, the above analysis still clearly demonstrate that reducing the energy loss due to charge injection and regen-eration processes is essential to improve the maximum achievable efficiency of DSSCs.

(a)

(b)

25

3. Meso-TiO2/dye/electrolyte interface

3.1 Semiconductors in the photoelectrochemical cells The concept of sensitized-semiconductor systems originates from the re-search on the photoelectrochemical systems, initiated in the 1950-1960s, where the semiconductor acts as both a light harvesting electrode as well as the electron transport material. Taking an n-type photoelectrode for example, after absorption of photons, the photogenerated minority carrier, hole, will be driven directly towards the surface of the electrode and participate in an electrochemical oxidation reaction. In turn, the photogenerated majority carrier, electron, will be driven to the semiconductor bulk by diffusion and/or migration, and ultimately to the external circuit, to contribute to cur-rent flow. Figure 8 shows several common used semiconductors in the early studies of the photoelectrochemical cells. However, these systems are often limited by several problems. The most severe one is their self-corrosion due to the photogenerated holes (photocorrosion), which can be oxidative in nature. Several large bandgap oxide semiconductors, like TiO2, are not sus-ceptible to photocorrosion; but, they absorb a relative low fraction of sun-light, in the ultraviolet and in eventually the near visible (around 400-420 nm), therefore limiting their practical use.

Figure 8. Positions of several semiconductors in contact with aqueous electrolyte at pH 1 used in photoelectrochemical cell. The conduction band is colored red while the valence band is colored green. The data is taken from the literature.(44)

26

Later, the concept of sensitization was introduced (17), where the sensitizer anchored on the semiconductor takes the task of light harvesting while the oxide substrate acts solely as the electron transporting layer. A big advantage of this system is the tunability of the absorption profiles of dye molecules through appropriate synthetic routines. Among various electrode materials, TiO2 was one of the first investigated photoelectrode materials with respect to both direct photoexcitation and sensitization; in the latter aspect, it has proven to be the most efficient one so far, mostly due to its stability and suitable band alignment. The present thesis is based on dye-sensitized TiO2. However, the following discussion and achieved insights should be also applicable to other sensitized-semiconductor systems.

3.2 Nanocrystalline semiconductor-liquid junction The work function is defined as the energy required to remove one electron from the solid and/or liquid to a point in the vacuum immediately outside the surface. For example, Figure 9 shows the work function of the semiconduc-tor (s) and electrolyte (e), respectively. Classically, due to their difference in the work function, the contact of a semiconductor and an electrolyte solu-tion shall develop an interfacial junction, where electrons move from the low work function side (high energy) to the high work function side (low energy). Depending on the relative working function of the semiconductor and the electrolyte, three types of junction can be formed: accumulation, depletion and inversion. This effect is similar to the Mott-Schottky junction, describing the junction between the semiconductor and the metal.

Figure 9. The band-alignment between flat TiO2 (n-type) and the electrolyte junc-tion before and after contact. HL stands for the Helmholtz layer.

27

Here, a flat TiO2 electrodewhere the Fermi level of the semiconductor before contact lies higher than in the electrolyteis taken as an example to explain the formation of the space charge layer in the depletion case1, while other cases can be deduced in a similar way. Upon contact, electrons will move from the TiO2 side to the electrolyte side; as a result, there will be an interface junction with the depletion of electrons in the TiO2 side and the accumulation of electrons in the electrolyte side. The effects of the above processes can be calculated by solving the Poisson and continuity equation. The result is that there will be an electric field gradually disappearing into the semiconductor. The layer in the semiconductor at the proximity of elec-trolyte where an electrical field is present is called space-charge layer. In a band diagram, this means that the conduction and the valence band will bend in the way as schematically shown in Figure 9b, i.e. so-called band-bending, and the potential drop is sc.

Figure 10. The band-alignment between meso-TiO2 and electrolyte junction before and after contact.

However, the case changes dramatically when the diameter of the nanoparti-cle becomes smaller than the width of the space charge layer (Figure 10). In such case, the free electrons within the nanoparticle will be completely de-pleted. The potential energy difference at the edge and the center of the na-noparticle will be described by the equation 3 derived by solving the Pois-son-Boltzmann equation(45–47):

(3)

1 For other cases, the reader is referred to the standard semiconductor electrochemistry book.(20)

(a) (b)

28

where kB is the Boltzmann constant, T is the temperature, e is the elementary charge, LD is the Debye length, a characteristic parameter in a solid indicat-ing how far the electric field will be screened and r is the diameter of the nanoparticles.

By assuming that dielectric constant and the ionized donor concentration are 1302 and 1017 cm-3, respectively, LD was calculated to be 30 nm(46). The voltage drop within a spherical nanoparticle with r = 10 nm will be only 0.5 mV. The virtual absence of an electric field creates important differences between a mesoscopic and a non-porous, flat-TiO2 electrode. As a result, the electron transport processes through the mesoporous films is controlled by the diffusion process(48), in strong contrast with the drift-and-diffusion mode in classical thin film photovoltaics. These transport processes will be further discussed in section 3.3.2.

3.3 Electronic properties and charge transport mechanism 3.3.1 Electronic properties

The available energy states within semiconductors are not continuous but merged into bands, i.e. the valence band and the conduction band. In be-tween them is a forbidden band, where no electronic states are allowed. The electron concentration in the conduction band, nC, can be expressed by the equation 4:

(4)

(5)

where N(E) is the density of states. kB is the Boltzmann constant, f(E) is the Fermi-Dirac distribution, T is the absolute temperature and EC is the conduc-tion band edge energy

2 The dielectric constants of anatase and rutile TiO2 have been also previously reported to be 46 (99) and 89 (100), respectively. The reported values in the literature vary depending on methods and conditions.(101)

29

Under the assumption of , the Fermi-Dirac distribu-tion becomes the Boltzmann distribution, so that the equation 4 becomes integratable, with nC expressed as:

(6)

therefore,

(7)

where NC is effective densities of states.

In the case of nanocrystalline nanoparticles, there are large amounts of trap states (Figure 11a, also called localized states) within the bandgap due to its large surface area. The physical origin of these traps is not entirely clear, but can be attributed from crystal defects (e.g., oxygen vacancies(49)), impuri-ties(50), surface treatment(51) and chemical surroundings(52). It is widely accepted that the density of these trap states, g(E), fall exponentially with energy below the conduction band, described by equation 8(53):

(8)

where Nt,0 is the effective density of states of localized states at the conduc-tion band edge and T0 is a characteristic parameter of the trap distribution.

Figure 11. Electronic states within mesoporous TiO2. (a) The distribution of elec-tronic states suggested by the multi-trapping model, where an exponential distribu-tion of trap states exists within the band gap. (b) the dependence of trap distribution on the characteristic temperature, T0.

(a) (b)

30

With use of the zero-kelvin approximation, the total charge density in the trap states can be calculated from(54):

(9)

This expression is surprisingly similar to the Equation 6. The essential fact is that the temperature is now a characteristic temperature T0 instead of the absolute temperature T. The slope between ln nt and EF is now kBT0 rather than kBT. The range of T0 is normally reported to between 600 K and 1200 K(55), and have direct impacts on the distribution of trap states (Figure 11b). There are alternative ways in the literature to derive the relationship between nt and EC - EF, e.g., the distribution of the trap states may be sug-gested to increase from the redox electrolyte(31), instead of exponentially decreasing from the conduction band edge as given above; but the eventual formula is rather similar.(56) The presence of the trap states may seem to be trivial in terms of charge accumulation in TiO2; but they do have profound effects on the electronic processes happening at meso-TiO2/dye/electrolyte interface, as discussed below.

3.3.2 Charge transport mechanism In the standard solar cells, e.g., p-n junction silicon solar cells, the charge transport occurs by both diffusion and electric drift, with the latter driven by the build-in electric field at the p-n junction interface. By contrast, as pointed out in the section 3.1, the finite dimension of nanoparticles results in a negli-gible electric field within the nanoparticles. Further, there is no macroscopic electric field across the TiO2 film, due to the effective screening of the elec-tric field by ions in the highly conductive electrolyte.(57) Therefore, the charge transport within the mesoporous film has been suggested to occur mainly via a diffusion process, which can be described by the continuity equation:

(10)

where D and stand for the diffusion coefficient and electron lifetime of the electrons, respectively, n0 is the equilibrium electron concentration of TiO2 film in the dark condition and g(x, t) stands for the generation of electrons in the system.

31

A direct proof of the above arguments was obtained by predicting the steady-state photo-response spectrum of DSSCs with the above equations, which showed great consistency with the experimental results.(48)

Figure 12. Schematic illustration of effects of trap filling on the charge transport and charge recombination within meso-TiO2. (a) High light intensity: the Fermi level is high, with the deep traps filled. (b) Low light intensity: the Fermi level is low, with the deep traps unfilled. The deeper the trap states are, the longer time it will take for electrons to de-trap back to the conduction band, where all the electronic transport and recombination occurs.

Later, it was further found that D and depends strongly on the light intensi-ty, where the multiple trapping are suggested to play important roles in the interpretation of the data, as described in the literature.(58–60) These charge transportation and recombination processes are suggest to occur through a random-walking at different trap states.(60) Firstly, the electrons injected into meso-TiO2 will be rapidly trapped into the localized states below the conduc-tion band edge. Then, it will stay there shortly and then de-trap back into the conduction band. Only the detrapped electrons, i.e. free electrons, will con-tribute to the conduction of electricity, or undergo recombination. The deeper the trap states, the longer time it takes for the electrons to de-trap. By trapping-and-detrapping several times, the electrons will eventually be collected by the outside circuit or recombined with oxidative centers in the film. One conse-quence of the trapping-and-de-trapping process is that the effective diffusion coefficient of electrons will be higher if the Fermi level lies higher within meso-TiO2with all the deep states filled, de-trapping takes less energy and gets more frequent. Meanwhile, since electrons move easily toward their re-combination sites, their electron lifetime will be shorter (Figure 12a). On the contrary, when the light intensity is lower, the electrons will be situated in the deep lying states (Figure 12b) and, as a result, it therefore takes more thermal energy and longer time to de-trap; the apparent diffusion coefficient will be smaller, but the electron lifetime will be longer. More strict physical and mathematical explanations can be found in the literature.(58, 61)

32

3.4 The local electric field and Stark effect Of particular interest in this thesis is the investigation of the local electric field and Stark effect3 at TiO2/dye/electrolyte interface. Herein, Stark effect refers to the shift of the dye absorption spectrum induced by the electric field. In this context, it is suggested to be generated between electron accu-mulated within TiO2, injected from excited dyes, and the surrounding cati-ons.

3.4.1 Double layer During the operation of DSSCs, the nanocrystalline TiO2 particle is nega-tively charged because of electron accumulation. From an electrostatic point of view, a charged spherical nanoparticle generates an electric field in the surrounding dielectric medium(62). In the electrolytic environment, the elec-tric field is normally rapidly screened by the free moving ions in the electro-lyte; this creates the so-called electric double layer, which can be described by the Gouy-Chapman-Stern model(63), a model widely used in describing solid and liquid interfacial double-layer structure (Figure 13).

Figure 13. Schematic picture of the double layer structure at the solid-electrolyte interface according to the Gouy-Chapman-Stern model. Here the solid is TiO2, nega-tively charged.

According to this model, the arrangement of the interfacial ions can be ap-proximated as the series connection of an electrical capacitor, Helmholtz capacitance, CH, and a diffuse capacitor, Cd, at the solid-electrolyte interface. The former can be imaged as the close packaging of positive and negative charge across a dielectric medium. The latter one is caused by the interplay 3Stark effect may also refer to the splitting of atomic spectrum by the external electric field.

33

between the tendency of electrolytic species to be attracted or repelled from the solid surface, and the tendency of randomization due to the thermal pro-cesses in solution. The overall capacitance of the double layer can be ex-pressed as:

(11)

In the highly concentrated electrolyte, Cd becomes very large correspond-ing to a low thickness of this layer, of the same order as the ion diameter.(63) Therefore, the overall capacitance is essentially determined by CH only. For example, in a 1 M 1:1 electrolyte, the characteristic thickness of the diffuse layer is only few Å, beyond which the interface electric field will vanish. The above theories have worked out very well at the normal solid-electrolyte interface.

3.4.2 Double layer at meso-TiO2/dye/electrolyte interface Due to the fact that the densely packed dye layer could impede the free movement of ionic species towards the surface, the double layer structure at meso-TiO2/dye/electrolyte interface is still not widely established in litera-ture.

Figure 14. Schematic picture of the Helmholtz layer structure at TiO2/dye/electrolyte interfaces suggested by Gregg et al.(64, 65). The outplane of Helmholtz layer was shown to be affected by the dye structure and nature of the ions in the electrolyte.

Previously, Brain A. Gregg et al.(64, 65) have qualitatively suggested that the dye should be at least partially or even completely within the Helmholtz layer (Figure 14). Their arguments were mainly based on the observation that the redox potential of dyes on the sensitized-TiO2 films followed the trend of conduction band shifts of TiO2 caused by the pH modulation; on the

34

contrary, the redox potential of dyes in the solution was independent of pH. Recently, the characterized Stark effects also strongly indicate that dye mol-ecules are at least partially located within the Helmholtz layer, which will be discussed in the following section of the thesis. However, the exact distribu-tion of the electric field still represents one of the challenging questions in the study of these interfaces, due to the entangled factors, e.g., dielectric properties of the solution, dye structures, dye coverage, the identity and con-centrations of ions in the electrolyte, etc.(64)

Figure 15. Schematic illustration of the Stark effect at TiO2/dye/electrolyte inter-face. (a) the dye used in this thesis and its dipole moments: ground state g, excited states, e and the change: = e-g. (b) the mechanism of the effects of the electric field on the absorption spectrum of the dye molecules anchored to the TiO2 surface. Vectors are marked with bold symbols. The direction of molecular dipole moment is defined from negative to positive, while the electric field is defined from the posi-tive to negative.

35

3.4.3 Stark effect Since 2010, a local electric field has been suggested to be presented at the TiO2/dye/electrolyte interface, created by the electrons accumulated into meso-TiO2 and the surrounding cations.(66, 67) The electron accumulation can be achieved either by a potentiostatic control of the potential of the films(68, 69) or by electron injection from the excited dyes(68, 70–72). One of the noticeable effects of this electric field is the shift of the absorption spectrum of the dye. Spectroscopically, the shift manifests itself as a bleach signal in the difference absorption spectrum even after the complete regener-ation of the oxidized dye.

A simple but helpful way to interpret this effect is schematically shown in Figure 15. The dye used in this thesis is called LEG4, which has a Donor--Acceptor structure (Figure 15a). It has a ground-state dipole moment (g

excited-state dipole moment (e), and the change in the dipole moment ( upon excitation, respectively. When applying an electric field across the dye layer, the HOMO and LUMO level of the dye will be shifted to a different extent due to the dipole-electric interaction. Consequently, the energy gap between the HOMO-LUMO will increase or decrease to h, depending on the relative relationship between the direction of the electric field direction E and the direction of (Figure 15b). By assuming there is little change in the line shape of the absorption spectrum before and after applying the electric fieldi.e. the electric field has no effects on the transition dipole momentthe absorption spectrum will therefore be red or blue-shifted simply due to the aforementioned shift in the HOMO-LUMO gap.

Theoretically, W. Liptay et al. (73, 74) has described quantitatively the shift of the absorption spectra of molecules upon application of an electric field. However, the complete analytical results are rather tedious(73, 74). An alternative simpler approach can be to express the absorbance A of the dye at the frequency v into its Taylor expansion, truncated to the second-order terms (67, 75–78):

(12)

where A(v, E) is the absorption of dye at frequency v when the dye is ex-posed into an electric field, E; A(v, 0) is the absorption of dye at v when E=0, i.e. no electric field; h is the Planck constant; dA/dv and d2A/dv2 are the first derivatives and the second derivative of the absorption spectrum against the frequency. is the dipole moment difference of ground state and the excited state of the dye and is the polarizability difference of the ground state and excited state dyes. For simplicity’s sake, the direction of both E and have been considered to be normal to the plane of the substrates.

36

Figure 16. Effects of the electric field on the absorption spectrum of the dye-sensitized TiO2 demonstrated by the electroabsorption measurement. (a) Schematic picture of the sample configuration, where an controllable electric field is generated between the FTO substrate and a transparent silver film evaporated on top of the poly(methyl methacrylate) (PMMA) layer. (b) The change of the absorption spec-trum of the dye when varying the electric field across the dye layer. The red and blue represents the positive and negative electric field on the TiO2 surface, respec-tively. The dashed line is the 1st order derivative of the absorption spectrum of LEG4. (c) the scatter plot of the amplitude of A at 550 nm versus applied voltage with the linear fit (black line).

One can therefore see that the change of the absorption spectrum A, A=A(v, E) - A(v,0), can be contributed by the linear and quadratic terms in relationship with the electric field (equation 12). In practice, for dye-sensitized films, the change has been found to be only linearly related to the electric field strength(67, 79), with an example shown in Figure 16. There-fore, the quadratic contribution is negligible, and the equation 12 can be simplified into the following equation 13 or 14:

(13)

(a)

(b) (c)

37

(14)

According to the laser spectroscopic(66) and PIA spectroscopy measure-ments(67, 68), the electric field across the dye in DSSCs has been previously estimated in the order of ~ 1 MV/cm by use of equation 14. These results indicate that the dye molecules are partially situated within the Helmholtz layer, in accordance with the previous results by Brain A. Gregg et al.(64, 65) mentioned in section 3.4.2.

38

4. Experimental methods

4.1 Characterization of single components 4.1.1 UV-Vis spectroscopy When shining light through matter, the incident energy of the light is partial-ly absorbed due to its interaction with the material. This absorption process can be quantified through the equation:

(15)

where A() is defined as absorbance, I0() and I() are the intensity of the light before and after passing through the sample, respectively.

The absorption of light by dissolved molecules in a dilute solution can be described by the Beer-Lambert law as:

(16)

where c is the concentration of molecules in the solvent, M (mol L-1), l is the distance of the optical path (cm) and () is the extinction coefficient of the molecule, for example a dye (M-1 cm-1).

() is an important parameter in the evaluation of the ability of dye to harvest light. The organic dyes used in this thesis normally have an in the order of 105 to 106 M-1 cm-1, due to the strong coupling of the D--A struc-ture.

4.1.2 Electrochemical techniques Electrochemical techniques can be used to provide important properties of the materials used in DSSCs, e.g., the redox potential, the diffusion coeffi-cient of the redox couple in solution and the HOMO level of the dye on sen-sitized films. Figure 17 shows a schematical setup of a typical electrochemi-cal measurement, where gold is used as the working electrode, a graphite rod as the counter electrode and Ag/AgCl (1 M LiCl in ethanol) as the reference electrode.

39

Figure 17. Schematic setup of a typical electrochemical measurement. WE: the working electrode; RE: the reference electrode; CE: the counter electrode.

During the potentiostat measurement, a voltage with a specific waveform is applied on the working electrodewith respect to the voltage of the refer-ence electrodewhile the current response is simultaneously recorded. This current response is a measurement of the properties of the investigated redox species. Of essential importance in this technique are two fundamental rela-tionships: Nernst’s equation and Fick’s law. The former describes the fun-damental relationship between the electrode potential E, the formal potential E0

’4 of the redox species, the oxidized species concentration, [O] and the

reduced species concentration, [R] (equation 17). The latter describes the diffusion properties of the system. Overall, it is the change of the local redox species concentration, governed by Nernst’s equation, and the diffusion pro-cesses, governed by the Fick’s Law, that determine the current response.(80)

(17)

where F is the Faraday constant.

4.1.2.1 Cyclic voltammetry (CV) In this measurement, a cyclic linear-sweep voltage (Figure 18a) is applied on the working electrode, while the current response is simultaneously rec-orded. Normally, a peak can be found both in the cathodic and anodic direc-

4 The formal potential can be related with the standard potential E0 by (63):

where fox and fred are the activity of the oxidative and reductive species, respective. Unless otherwise mentioned, the redox potential mentioned in this thesis is referred to the formal potential.

40

tion as shown in Figure 18b. The formation of peaks is caused two convolu-tional effects: (a) the tendency of increase of current due to the higher driv-ing force when scanning to larger overpotential and (b) the tendency of de-crease of current due to the depletion of redox species around the electrode with time. The redox potential of the investigated species can be taken from the average value of the anodic and cathodic peak potential, Ea and Ec, re-spectively. Ideally, the peak separation between Ea and Ec should be 59 mV for one-electron reversible reaction at room temperature.(80) In the real measurement, this peak separation is often larger than that due to various reasons, e.g., the Ohmic drop of the system and the irreversibility of the reaction.

Figure 18. (a) The voltage waveform applied in the CV measurements (b) typical results of CV measurements at different scan rates.

The peak current ip can also be analytically expressed for a Nernstian reac-tion into the equation:

(18)

where n is the number of electrons involved in the reaction, A is the surface area, D and c is the diffusion coefficient and the concentration of the diffu-sion species, respectively, and v is the scan rate.

By scanning the CV at different scan rates, one can therefore measure the diffusion coefficients of the redox species by the above equation. Alterna-tively, one could also use a microelectrode to determine the diffusion coeffi-cient of the redox species; in such a case, the spherical diffusion of redox species around the microelectrode impedes the development of a depletion layer of the redox species. The limiting currents are therefore determined by the diffusion speed of the redox species. The diffusion coefficient, D, is re-lated to the plateau current, Ilim, in the following equation:

(19)

where r is the radius of the microelectrode.

(a) (b)

41

4.2 Characterization of solar cells performance 4.2.1 Current-voltage (I-V) characterization The current-voltage characterization is the “touchstone” of a solar cell, which evaluates its ability to convert light into electricity. During the meas-urement, a voltage scan is applied between the working electrode and the counter electrode of the solar cell, while the current is simultaneously rec-orded with a source meter. This measurement can be performed either in the dark or under illumination.

Figure 19. A typical I-V curve (black line, the left coordinate) and the calculated output power curve (blue line, the right coordinate). Point A stands for the maxi-mum power point of the solar cell.

Figure 19 shows a typical I-V curve of a DSSC under illumination, demon-strating the relationship between its voltage and current. The output power, P(V), of the solar cells can be calculated by the equation: P(V) = VI, and plotted as the blue line. The output power is a function of the voltage. The optimal operation position of DSSCs is therefore point A, where P(V) reach-es the maximum value (Pmax). The fill factor, FF, can be defined through the equation 20, indicating the “squareness” of the I-V curve.

(20)

therefore, the efficiency of the solar cells can be calculated by:

(21)

42

where Pin stands for the power of the incoming light. For standard measure-ments, this power is 100 mWcm-2 at AM 1.5 condition.

4.2.2 Incident photon to current efficiency (IPCE) characterization Although the I-V measurement mentioned above can assess the performance of the device, there are still two main shortages: (a) the detailed photo-response of the devicethe conversion efficiency at a specific wave-lengthis not resolved and (b) the recorded current may be distorted or imprecisely measured due to other electronic processes (e.g., capacitance current) in the solar cells or spectral mismatch between the light source and the AM 1.5 spectrum. The latter issue becomes more severe when using a fast scan. Due to these limitations, an IPCE measurement, also sometimes known as the external quantum efficiency measurement, is important in the assessement of the solar cell performance.

In the IPCE measurement, monochromatic light is focused onto the solar cell under short-circuit condition. The incident photon flux can be precisely measured using a calibrated photodetector. Meanwhile, the short-circuit current of the DSSCs, ISC(A), is recorded, which indicates the amount of electrons generated.

Figure 20. A typical IPCE spectrum illuminated from the working electrode side of the DSSCs.

The IPCE under the illumination of a specific wavelength can therefore be expressed into the equation:

(22)

43

When scanning the whole visible spectrum, one can therefore obtain the detailed photo-response of the device at each single wavelength. Integration of IPCE over the sun spectrum can therefore give the overall photocurrent, which should be similar with the JSC measured in the I-V measurements. Figure 20 shows a typical IPCE spectrum of a DSSC. This IPCE shows that the solar cell have maximum IPCE value at only around 42%, implying more than half of the incoming photons are not effectively converted into electrons as a photocurrent. In addition, the photoresponse also disappears at 650 nm. Due to the low conversion efficiency and limited photoresponse range, one would imagine the solar cell performance is going to be low.

One could further formulate the factors affecting the IPCE into: the light harvesting efficiency of the dyes, LHE(), the injection efficiency of excited dyes into the conduction band of TiO2, inj() and the collection efficiency of the injected electrons through external circuits, cc(). In this case, one can express IPCE into the equation 23:

(23)

The LHE() can be normally measured by the absorptance of the solar cells. inj() is normally 1 in the DSSCs. Therefore, the IPCE value reflects direct-ly the charge collection efficiency of the systems, which is determined by the competitive processes of charge transport, charge recombination and dye regeneration. Meanwhile, this charge collection efficiency can be reflected by the diffusion length of the system, defined as the square root product of the diffusion coefficient and electron lifetime of the cells: . There-fore, the IPCE spectra, measured by the illumination of the device from both the front side and the backside, can be used to calculate the diffusion length. The basic calculation is based on solving the continuity equation of the solar cells under short circuit condition with proper boundary conditions.(48) The IPCE from front side and backside are described by the equation:

(24)

(25)

where d is the thickness of the film, is the reciprocal absorption length and L is the diffusion length of the solar cell.

44

It should be emphasized that IPCE is the characteristic measurement of the solar cells at photo-stationary, steady state. Therefore, the charge collec-tion should be the result of all the microscopic processes happening in the system.

4.3 Tool-box Knowing the I-V curve and IPCE of a solar cell is sufficient to evaluate its functionality of converting light into electricity. However, limited infor-mation is achieved with respect to other physical properties of the device, e.g., the electron lifetime, the transport time and the conduction band posi-tion. These parameters explain the electronic properties of the solar cells and therefore are crucial for our understanding. In this thesis, they are mostly investigated through a homemade electronic system named as “Tool-box”, which integrates a bunch of techniques, including small-modulation transient photovoltage, small-modulation transient photocurrent and charge extraction techniques, as described below.

Figure 21. Schematic mechanism of the electron lifetime and transport time meas-urement in the tool-box.

4.3.1 Small-modulation transient photovoltage technique In this measurement (Figure 21), the solar cell is kept at open circuit under illumination of a LED, whose intensity is controlled by the applied bias volt-age. Normally, a small square-wave modulated voltage (e.g., 10 mV) is ap-plied on the base voltage of the LED in order to create a perturbed square-wave illumination. In response, the VOC of the DSSC will increase and de-crease correspondingly, normally in an exponential way. The rising and de-creasing time constants are defined as the electron lifetime of the solar cells through the equation (81):

45

(26)

Due to the dependence of electron lifetime of DSSCs on its voltage as de-scribed in section 3.3.2, the measurement is typically conducted under dif-ferent bias illumination. The measured electron lifetime normally decrease in an exponential way with the voltage increase (Figure 22).

Figure 22. Typical electron lifetime results. The electron lifetime decrease exponen-tially with the increase of fermi level within TiO2, EF. Here: EF = VOC+ Eredox.

4.3.2 Small-modulation transient photocurrent technique In this measurement (Figure 21), a solar cell is illuminated and perturbed by a LED in the similar fashion as the small-modulation transient photovoltage measurement mentioned above. However, the solar cell is kept at short cir-cuit, and JSC is followed instead of VOC. The rise and decrease time of the photocurrent, due to the modulation of light, are defined as a characteristic response time, resp, which can be expressed into:

(27)

where tr is the electron transport time. e is the electron lifetime. Since tr is normally of one order smaller than e (82), the measured resp is

normally treated as the transport time: resp ≈ tr.

4.3.3 Charge extraction technique In this measurement, the DSSC is illuminated by a LED and achieve a stable voltage at open circuit. Then, the illumination is turned off, with the connec-tion simultaneously switched to short circuit. The current response of the

46

DSSC is recorded and integrated, which gives the amount of electrons stored at this specific VOC. By use of eq. 7 or eq. 9, one can therefore correlate the amount of electrons extracted with the fermi level and the conduction band edge of the system.

Here, all the electrons are assumed to be extracted without loss. However, due to the recombination loss, the actual extracted electrons could be under-estimated in the system with fast recombination kinetics.

4.4 Photoinduced absorption (PIA) spectroscopy The absorption of a complete DSSC is contributed from all its components, e.g., TiO2, dye, and the electrolyte species. When a DSSC is exposed under a pulsed illumination, the concentration of its various species could be dis-turbed due to light excitation, therefore resulting in a change in its absorp-tion spectrum. Meanwhile, the transient species are often associated with specific characteristic absorption features. Therefore, monitoring the optical change can give information regarding the concentration of the species in the DSSCs, as well as their kinetics.

Figure 23. Schematic picture of PIA spectroscopy applied in this thesis

PIA spectroscopy measures the optical change of a system when applying a square-wave light modulation and the general experimental setup for PIA spectroscopy is shown in Figure 23. It is essentially a pump-probe system. The optical probe is a continuous halogen lamp, and the pump can be a LED or laser. In the measurement, one can monitor the change of the transmit-tance before and after the laser pulse. Depending on the type of measure-ment, the optical changes can be either measured in the time domain, or alternatively, measured in the frequency domain by a lock-in amplifier.

47

4.4.1 Time-resolved photo-induced absorption traces In time-resolved photo-induced absorption experiments, the absorption change of the samples caused by square-wave excitation is monitored with time. Figure 24 shows a typical optical change of the DSSC monitored at 750 nm during the square-wave excitation. With the light on and off, the absorption signal at 750 nm is found to increase and decrease, correspond-ingly. With proper spectral assignment, this signal is attributed to the intra-band absorption of electrons within TiO2, as the redox species in the electro-lyte regenerates the oxidized dyes. Therefore, one can follow the concentra-tion change of accumulated electrons within TiO2 by monitoring these sig-nals.

Similarly, one could also apply electrical voltage between two electrodes of a DSSC and then monitor the optical change. With the electrolyte inside DSSCs, the counter electrode can act as a quasi-electrode. Therefore, this electrical modulation can be regarded as an electrochemical way of tuning the electron concentration within TiO2. Further information can be found in the results part of Paper III.

Figure 24. The typical response of a DSSC in the time-resolved PIA measurement. Detection wavelength: 750 nm.

4.4.2 Photoinduced absorption spectrum As shown in Figure 24, the PIA signal is normally rather small, in the order of 10-3 - 10-4 and therefore has low signal to noise ratio (SN). To improve the SN, one can use the lock-in detection technique of a lock-in amplifier. Brief-ly, an AC periodic excitation is applied on the sample, while the AC source is also simultaneously synced with a lock-in amplifier, generating a refer-ence signal. During the detection, only signals with the same frequency as the AC will be picked up, by passing through a phase-sensitive detection circuit, which, as a result, excludes the electronic noises effectively.

48

In this measurement, the lock-in detection technique mentioned above is used, and the absorption changes of a sample are measured over a wave-length range. Figure 25 shows a typical PIA spectrum for DSSCs with the inert (-Inert) and redox (-T) electrolyte. In the case of an inert electrolyte (without redox couple in the electrolyte), laser excitation of the sample cre-ates a population of oxidized dye molecules, as a result of the equilibrium between charge injection and recombination.(83) Therefore, one can observe spectroscopic features belonging to the oxidized dye, with characteristic peaks at 670 nm and 770 nm. With the redox electrolytes (with the redox couple in the electrolyte), the oxidize dyes will be rapidly regenerated by redox mediators in the electrolyte. Consequently, the characteristic absorp-tion of the oxidized dyes disappeared. By comparing the difference between the PIA with the inert and redox electrolyte, one can roughly estimate the regeneration efficiency of the oxidized dye by the redox couples.

Figure 25. The typical PIA spectra for the LEG4-sensitized TiO2 in the presence of the inert electrolyte (LEG4-Inert) and the redox electrolyte (LEG4-T).

4.5 Transient absorption spectroscopy (TAS) TAS is a pump-probe technique similar to PIA spectroscopy introduced above, while a short laser pulse is used, instead, to perturb the system. In this thesis, TAS is mainly used for the investigation of the photophysical pro-cesses at the TiO2/dye/electrolyte interface, which occur at the time range of ns to ms. Practically, one can either measure the change of the absorption spectra at different delay times after the laser excitation. Alternatively, one can also monitor the absorption change at a specific wavelength with time to obtain the kinetic information.

49

Figure 26a shows a typical example of a transient absorption spectrum of the LEG4/TiO2 film in contact with 0.1 M CaClO4 in ACN (Blue line). In a separate spectroelectrochemical study, the absorption spectrum of the oxi-dized LEG4 (Figure 26b) was measured with two characteristic absorption peaks at ~ 670 nm and 770 nm. By comparing these two spectra, one can readily conclude the generation of oxidized LEG4 after laser excitation (Figure 26a, black dash line). There is also another spectral contribution from the Stark effect (red line), as discussed in Section 3.4.4. The simulated spectrum (cyan line) based on these two contributions are in good consist-ence with the measured results, confirming the above assignment of the

Figure 26. (a) The transient absorption spectrum of LEG4-sensitized TiO2 film in contact with the inert electrolyte (0.1 M LiClO4 in ACN) after ~ 10 ns laser excita-tion and its spectral decomposition (b) the spectra recorded during the electrochemi-cal oxidization of LEG4 in the ACN.

(a)

(b)

50

spectral contribution. (The mismatch at wavelength below 500 nm is at-tributed to the low detection sensitivity, resulting from the strong ground-state absorption).

Figure 27 shows a typical example of a kinetic decay measurement at 750 nm—characteristic absorption from the oxidized LEG4—when the LEG4/TiO2 film is in contact with an inert electrolyte (0.1 M LiClO4 in ACN). In this case, the regeneration of the oxidized dye can be only achieved by the recombination between TiO2(e

-) and the oxidized dyes. Therefore, the monitored absorption change at 750 nm as a function of time indicates the kinetics of the charge recombination. In DSSCs, the kinetics of this decay show strong dependence on the light intensity (Figure 27). There-fore, the comparison of the kinetic data demands measurements conducted at similar laser excitation intensities (or similar electron densities in the TiO2).

Figure 27. The kinetic decay of the oxidized dye in the inert electrolyte measured at different laser intensities. (a) Original data (b) Normalized decay data.

(a)

(b)

51

4.6 Kinetic analysis The acquired kinetic traces normally need to be quantified to make most use of the data. Generally, two types of information can be achieved by fitting the kinetic traces: a) a time constant, describing the speed of the change; b) a kinetic model, describing the physical mechanism. For example, if a reaction is 1st order in nature, the change of a specific species’ concentration, C, with time can be expressed by:

(28)

(29)

Therefore, a linear relationship between the lnC and t can normally serve as an indicator that the reaction is a first order reaction. Similarly, for a second order reaction, the 1/C is linear to t.

However, in practice, the kinetic traces can rarely be fitted into the simple predefined models discussed above. Herein, some of the most widely applied methods in literature, as well as in this thesis, are described below with short comments.

4.6.1 Kohlrausch-Williams-Watts (KWW) function KWW function is well known as the stretched exponential function (equation 30). It can be imagined as a linear superposition of simple exponential decays, where the rate constants have a Levy distribution.(72, 84) In the studies of dye-sensitized solar cells, the KWW function has been frequently used to characterize the recombination processes between oxidized dyes and electrons within TiO2, due to its excellence in fitting the data (this process is normally hard to fit with the single exponential or double exponential decay).(85–87)The average time constant, , can be calculated from equation 31.

(30)

(31)

where 0 is a characteristic time constant, β is a characteristic parameter, representing the “width” of the distribution of 0 and Г is the gamma func-tion.

52

Practically, the change of fitting conditione.g., the selected data range, initial position, etc.,can sometimes lead to different without justification, although an apparent good fitting. In this thesis, this method has been used mostly to fit the kinetic traces as a numerical tool, however, for the sole pur-pose of estimating the decay half time (see infra vide).

4.6.2 Half time The half time, t1/2, is defined as the time required for the kinetic traces to relax to half of its initial value, which therefore serves as a characteristic parameter to quantify the kinetics. It is simple and easy to interpret, and mostly used in the thesis. Previous studies have shown that the decay t1/2 of the charge recombination process in DSSCs, i.e. between the TiO2(e

-) and oxidized dyes, shall decrease with the light intensities in a power relation-ship (Eq. 32)(55, 88). This phenomenon is attributed to a transport-limited charge-recombination mechanism.(60, 88) According to this model, elec-trons conduct random walk within TiO2 between different trap states, and the waiting time for the electrons at a specific trap is associated with the depth of the trap, i.e. the energy difference between the trap state and the conduc-tion band edge. Once reaching the recombination center, it will react with the oxidized species immediately.

(32)

However, one limitation of the half time method is that it can only be capa-ble to indicate the speed of the decay from 1 to 0.5 (after normalization), while it cannot characterize speed of the decay after the half time. Practical-ly, the definition of the time zero has strong dependence on the time resolu-tion of the instrument. The time zero will be more precisely defined when using a short oscilloscope time window with fast response and high time resolution. In contrast, the measured time zero will be prolonged when using oscilloscope with a larger time window. As a result, the decay half-time may be measured with large artifacts. For example, Figure 28a shows that the decay half-time would be 0.53 s and 2.35 s for the oscilloscope with a time window of 10 s and 1000 s, respectively, which correspond to ~ 4 times difference. Similar situation is also observed at lower light intensities with the slower decay (Figure 28b). In order to avoid this problem, the ki-netic traces can be firstly measured with different time windows and then combined together, in order to ensure both a long observation time and high time resolution.

53

Figure 28. Relationship between the choice of the time window of oscilloscope and the estimated half times at: (a) the high light intensity: ~ 2.2 mJ/Pulse and (b) the low light intensity: 0.28 mJ/Pulse.

4.6.3 Double or multiple exponential fittings The double or multiple exponential methods fit the kinetic traces with a line-ar summation of several single exponential terms. In the double exponential fitting, for example, the kinetic process can be described by equation 33. The averaged time constants, , for the kinetics can be defined by the equation 34:

(33)

(34)

where An stands for the “weight” of the decay, with a specific time constant tn.

(a)

(b)

54

Although its wide application and success in the literature, the calculated may fail to reflect the kinetics when either t1 or t2 are extremely long, and therefore overestimate the time constant. In this case, a logarithm-averaged method has also been reported(89), and proven to work nicely:

(35)

55

5. Results and discussion

In this chapter, the results are summarized into three sections: (a) the strate-gies to improve the performance of DSSCs, (b) stability of cobalt-bipyridyl electrolyte based DSSCs and (c) the local electric field and Stark effect. All these results are strongly associated with the TiO2/dye/electrolyte interfaces. They are presented only to address the essential questions: why, what has been done and what are the results and implication, with the details in-cluded in the attached articles.

5.1 Strategies to improve the performance of DSSCs 5.1.1 Alternative one electron redox system From 1991 to 2010, I-/I3

- was the dominant redox couple during the devel-opment of the DSSCs, with a redox potential at ~ 0.35 V vs NHE in organic solvent.(14) However, the regeneration of the oxidized dyes by I-/I3

- involves complicated processes(90) and require large driving force, in the order of 0.5 to 0.6 eV for the ruthenium-based dyes(14), which therefore limits the achievable maximum efficiencies of DSSCs. Although tremendous efforts have been dedicated to explore other redox species, no alternative redox couples were as competent as I-/I3

-, mainly due to the large recombination losses.

A breakthrough occurred in 2010 when Sandra Feldt et al.(82) combined an organic dye named D35 with a previously reported cobalt-bipyridyl redox mediator. The blocking effects of the alkyl chains on the donor part of D35 improved dramatically the electron lifetime of DSSCs, which lead to, for the first time, the development of an alternative redox system with comparable performance as I-/I3

-. However, it is not perfect. The regeneration processes of the oxidized dye by the cobalt redox mediator still demands a large driv-ing force; at least, 0.4 eV is required for more than 85% dye regeneration.(91) Further search for alternative redox species, combining lower driving force, efficient dye regeneration and low recombination loss, is therefore one of the most promising ways to boost the efficiency of DSSCs further.

56

Figure 29. Materials used in Paper I. (a) the molecular structures of various com-pounds and (b) the energy diagram of the compounds investigated, which compares the HOMO level of the dyes and the redox potential of the mediators.

In Paper I, I have tried to develop the DSSCs with an organic redox cou-pleTEMPO/TEMPO+ (Figure 29) with a redox potential at ~ 0.89 V vs NHE. This very positive redox potential holds promise of reducing the ener-gy loss during the dye regeneration process. Previously, the redox couple has been combined with an indoline dye named D149(Figure 29), which achieved sound results, but suffering from serious recombination.(92) I therefore tried to use an organic dye called LEG4 (Figure 29), with the simi-lar alkyl chains as D35, as a sensitizer to develop the TEMPO/TEMPO+-based DSSCs. The aim was to reduce the recombination in a similar fashion as the D35-cobalt bipyridyl system mentioned above, therefore improving the DSSCs performance.

Figure 30. Solar cell performance of the LEG4 (red line) and D149 (blue line) sensi-tized TEMPO/TEMPO+ (-T) based DSSCs. (a) Current–voltage curve measured at 100 mW cm-2 AM 1.5 G illumination. (b) Incident photon to current conversion efficiency (IPCE).

(a) (b)

57

After the optimization, LEG4-sensitized DSSCs indeed showed an improved solar cell performance (5.4%) compared with that of D149-sensitized (3.2%) DSSCs (see Figure 30) when using TEMPO/TEMPO+ as a redox couple. However, the detailed characterizations suggested that this improvement is not from the reduction of recombination loss. Figure 31 showed that the electron lifetime of DSSCs with TEMPO/TEMPO+ (-T), regardless of LEG4 or D149 as sensitizer, is dramatically lower than that with the cobalt electro-lyte (-Co). Similar conclusion is also achieved by the simulation of the IPCE spectral from both the front and back side illumination; it leads to the con-clusion that the diffusion length of electrons is ~ 2.8 m for LEG4 with co-balt electrolyte, while only ~ 0.5 m for both LEG4 and D149 in combina-tion with TEMPO electrolytes. Further spectroscopic studies suggested that the improvement of LEG4-sensitized DSSCs originate from faster regenera-tion compared with D149-sensitized DSSCs (Figure 32), with the calculated regeneration efficiency of 87% and 68%, respectively.

Figure 31. Electron lifetime for D149–T, LEG4–T, and LEG4–Co at different elec-tron quasi-Fermi levels in mesoporous TiO2 under various bias illuminations. The quasi-Fermi level, EF, was calculated from the VOC, using the relationship: EF = Eredox+ VOC.

Figure 32. Transient absorption traces for (a) LEG4–Inert and LEG4–T. Excitation wavelength: 530 nm, detection wavelength: 750 nm (b) D149–Inert and D149–T. Excitation wavelength: 530 nm, detection wavelength: 825 nm.

58

These results show that the energy loss during the dye regeneration pro-cess is still reducible. Meanwhile, further surface passivation is, however, required in order to reduce the recombination loss.

5.1.2 One-electron or two-electron transfer agent? As shown in Section 5.1.1, application of redox species with very positive redox potential is often associated with fast recombination of electrons with the oxidized species, even if ensuring enough dye regeneration. In Paper II, An alternative way to improve the efficiency of DSSCs was demonstrated, by simply adding an organic donor molecules with fast regeneration kinetics, tris(p-anisyl)amine (TPAA) (Eredox = 0.79 V vs NHE , Figure 33), into the cobalt bipyridyl electrolyte.

Figure 33. Chemical structures of TPAA and Co(bpy)3

2+/3+ used in the study.

Figure 34. Comparison of DSSCs performance with (-TPAA/Co) and without addi-tion (-Co) of TPAA in the cobalt electrolyte. The cosensitized film is sensitized both with LEG4 and Dyenamo Blue dyes.(29) (a) I-V curves under 100 mW cm-2 AM 1.5G illumination (solid lines) and in darkness (dashed lines) (b) IPCE.

Figure 34a shows the comparison of DSSCs performance with and without addition of TPAA in the cobalt electrolyte. Remarkably, the efficiency of the LEG4-sensitized DSSCs improved from 7.2% to 9.1% only by the addition of TPAA. Further application of a co-sensitized film (where two dyes are combined together to harvest more light), led to the development of the

59

DSSCs with efficiency of 10.5% and 11.7%, at 1 sun and 0.5 sun, respec-tively. These improvements were largely contributed from the increase of JSC, with VOC of the solar cells only slightly affected. This conclusion is also confirmed by the IPCE results (Figure 34b), where higher IPCE was achieved with the addition of TPAA in the electrolyte.

Detailed electronic characterization demonstrates that the addition of TPAA barely affect the charge accumulation of the DSSCs (Figure 35a); instead, the electron lifetime was improved significantly (Figure 35b). Fur-ther spectroscopy techniques, including PIA, nanosecond laser and ultrafast laser spectroscopies, were combined to elaborate the various electron trans-fer processes in the DSSCs with the addition of TPAA. Figure 36 summa-rizes these various processes and their corresponding time constants. Briefly, after dye excitation and electron injection, the oxidized dyes are found to be regenerated firstly by the TPAA molecules within ~ 100 ps to 10 ns. The oxidized TPAA+ is then regenerated by the Co(bpy)3

2+ within ~ 14 s, be-fore recombining with electrons from the TiO2 (~ 63 s). In other words, the addition of TPAA has introduced an electron transfer intermediate step dur-ing the regeneration process of DSSCs, which ensure both a fast dye regen-eration and reduce the recombination speedin combination, these effects eventually improve the DSSCs performance dramatically.

Figure 35. Charge extraction (a) and Electron lifetime (b) measurements as a func-tion of the voltage for LEG4-sensitized DSSCs dye employing the standard cobalt (Co, black dot) and the standard cobalt with addition of TPAA (TPAA/Co, red dot), respectively

Further experiments have shown that the overall VOC of the solar cells are still controlled by the quasi-Fermi level of electrons within the TiO2 and the redox potential of Co(bpy)2+/3+ in these DSSCs. According to the analysis in section 2.4, the addition of TPAA should therefore not change the achieva-ble maximum efficiency of DSSCs. However, the prolonged electron life-time does significantly enhance the solar cells performance by reducing the nonradioactive recombination loss.

60

Figure 36. The overall charge transfer scheme in the DSSCs fabrication with TPAA/Co electrolyte (path 1) and Co electrolyte (path 2). Regeneration and recom-bination time constants were summarized from both nanosecond and femtosecond measurements.

As a final remark, one may notice the similarity of the current system with the I-/I3

- electrolyte, which possesses multiple electron transfer steps (shown below) during dye regeneration process as well.(90) Those sequential charge transfer processes also enable the fast dye regeneration by an intermediate state (Step 1 and 2); then the intermediate states react rapidly with the elec-trolytic species (Step 3) rather than recombining with the electrons in TiO2. These similarities suggest that addition of electron donors in the normal redox electrolyte could be a universal and effective strategy to develop DSSCs and should be further exploited.

5.2 The stability of cobalt-bipyridyl electrolyte based DSSCs The stability of DSSCs is widely concerned due to its importance for com-mercialization. This concern is more severe when using cobalt bipyridyl as the redox mediators, where ligand dissociation is suspected to accelerate the degradation of DSSCs. In Paper III, the stability of cobalt bipyridyl-based DSSCs was investigated. Different Lewis-bases additives, widely used in DSSCs (Figure 37), are added into the cobalt electrolyte, considering their potential effects on the ligand exchange or dissociation reaction of Co(bpy)3

3+/2+. We conducted the stability test by storing the DSSCs at 70 C

61

in the dark, which corresponds to the ISOS-D-2 protocols(93) used for thin film solar cells. This condition is chosen in order to avoid the complicated factors that might be induced by the light soaking. The results therefore rep-resent the stability of DSSCs under only the thermal stress. Meanwhile, a non-volatile solvent 3-methoxypropionitrile (MPN) is used to avoid the sol-vent evaporation, although also leading to a low DSSCs performance, ~ 3%.

Figure 37. Structure of cobalt tribipyridine complex and the additives MBI, TBP and BPY used in the present study. The counter ion for the cobalt complex is PF6

-.

Figure 38. Change of the performance parameters of the investigated DSSCs during the stability test in 50 days (a) VOC (b) JSC (c) FF (d) Efficiency. The solar cells were tested at 70 C in dark.

62

Figure 38 shows the evolution of the photovoltaic parameters of DSSCs with different additives in the electrolytes during 50 days. Noticeably, the de-crease of the photovoltage is the main degradation reason for all the investi-gated DSSCs, with more than 100 mV drop during the test. The photocurrent firstly increased to a maximum value at around 30 days and then decreased gradually; the fill factor showed opposite trends as the photocurrent.

Figure 39. Change of the charge extraction measurements of DSSCs at open circuit during the stability test. (a) A typical evolution example from the MBI-BPY cells. The arrow indicates the direction of the shift. All the other cells have similar pat-terns shown in the supporting information of Figure S1in Paper III. (b) Comparison of the VOC of DSSCs at the same QOC = 210-5 C during the test for four electrolytes.

Charge extraction measurements for all the DSSCs demonstrated a left-shifted charge extraction curves during the degradation process. Figure 39a shows the result from MBI-BPY cells as a typical example. According to the discussion in section 2.2.1 (equation 9), this shift suggests a positive shift of the conduction band position of TiO2. All the band shifts are summarized in Figure 39b, in the order of TBP (105 mV) MBI (82 mV) TBP-BPY (80 mV) MBI-BPY (66 mV), accounting for the majority reason of the photo-voltage decrease. Further electron lifetime measurements also indicated a decreased electron lifetime during the degradation, in the order of TBP-BPY (87.5%) MBI (78%) MBI-BPY (70%) TBP (67%) to its initial life-time. These electronic characterizations indicate that both the positive shift of the conduction band edge and the decrease of the electron lifetime con-tribute to the decrease of the voltage of DSSCs, while the former dominates.

We further conducted electrochemical experiments to characterize the change of the electrolyte species during the thermal stress. Collectively, those results suggested a decreased concentration of Co(bpy)3

3+ during the stability tests, while the Co(bpy)3

2+ are relatively stable (Figure 40).

63

Figure 40. Differential pulsed voltammetry of Co(bpy)3

2+/3+ before and after heating at 70 C for a week, respectively. (a) 1 mM Co(bpy)3

2+ in 0.1M TBAPF6 in MPN (b) 1 mM Co(bpy)3

3+ in 0.1M TBAPF6 in MPN.

Importantly, the differential pulse voltammetry measurement, although it may not reflect the exact change of the electrolytes in the real working con-dition, do indicate a superior stability of Co(bpy)3

2+ compared with Co(bpy)3

3+. However, the instability of Co(bpy)33+ in the electrolyte are not

the dominant factor limiting the DSSCs performance in this study, evidently from the small change in the photocurrent. Thereby, other reasons for the conduction band shift and reduced electron lifetime during the thermal heat-ing period should be explored. It was postulated here that dye desorption during the thermal stress would probably cause these results. Previous stud-ies have shown that reduction of surface dye coverage by an amount of only 5% could lead to dramatic change of the electron lifetime and band shift.(94) Unfortunately, attempts to monitor the UV-Vis absorption of the film during the degradation processes did not give convincing evidences for this argu-ment yet, possibly due to the small changes and variation in the solar cells fabrication. Other reasons like Li+ intercalation under thermal stress are also possible. The exact role of these Lewis bases are also not entirely sure yet, but they all give qualitatively the same trend as discussed above.

As a final remark, the performance of DSSCs in this study is rather low, which may not reflect the changes of JSC in more efficient DSSCs. Further work should try to investigate systems with high photocurrents in order to evaluate the contribution of electrolytic species to the overall stability.

5.3 The Local Electric Field and Stark Effect The third part of the thesis focuses on the investigation of the local electric filed and Stark effect at the TiO2/dye/electrolyte interface.

64

5.3.1 Charge compensation The discovery of the local electric field at the TiO2/dye/electrolyte(66, 67) interface and the resulting Stark effects (with details in Section 3.4.2) triggers intriguing questions about various processes at TiO2/dye/electrolyte interface. One of the most obvious ones is: how do the cations in the electrolyte respond to this local electric field? Previously, this question has been investigated mainly using laser spectroscopies(72, 87, 95), with the results indicating that charge compensation occurs in the time scale of s-ms. In Paper IV, I have tried to investigate this question by use of the photoinduced absorption (PIA) spectroscopy. These methods correspond to the investigation of charge com-pensation processes with the steady-state electron concentration within meso-TiO2. Also, PIA spectroscopy investigates the charge compensation in DSSCs at longer time scale, compared with the laser spectroscopy.

Figure 41. (a) UV-Vis absorption spectrum of the LEG4-sensitized TiO2 film meas-ured in air (blue line) and the scaled 1st order derivative of the absorption spectrum (dA/d) (red line). (b) The change of the absorption spectrum, A, of LEG4/TiO2 film in the spectroelectochemical measurement, during a linear potential scan from 0 V to 0.8 V. (c) PIA spectrum of LEG4-sensitized DSSCs in the presence of Co(bpy)3

2+/3+ in the electrolyte.

LEG4 is used as the sensitizer in this study which has been characterized to have a Stark bleach peak at ~ 580 nm, taken from the 1st derivative of the ab-sorption spectrum (Figure 41a). Spectroelectrochemical measurements

(a) (b)

(c)

65

(Figure 41b) show that LEG4+ has two characteristic absorption peaks at ~ 670 nm and 770 nm. Figure 41c shows a typical PIA spectrum of LEG4-sensitized DSSCs, which clearly showed a bleach at around 580 nm, confirm-ing the presence of the local electric field and Stark effect as reported in the previous studies.(67) Meanwhile, the absence of characteristic absorption peaks at 670 nm and 770 nm in PIA spectrum implied the efficient dye regen-eration by the redox species in the electrolyte. Therefore, no contribution is expected from the oxidized dye or ground state dye bleach to the PIA spec-trum.

Figure 42. Time-resolved photo-induced absorption of LEG4-sensitized DSSCs. monitored at (a) 580 nm and (b) 750 nm. Square wave laser modulation (530 nm) with f = 0.33 Hz was applied.

Figure 43. Time-resolved photoinduced absorption of LEG4-sensitized DSSCs monitored at 580 nm, with the A normalized at the minimum peak position to be – 100%: (a) with addition of 0.1 M LiClO4 in the electrolyte (b) without addition of LiClO4 in the electrolyte. The error range was based on the average from four solar cells with the same electrolyte. Every 20th data point was shown for clarity.

(a)

(b)

66

Figure 42 displays the most important experiments. Here we use a time-resolved PIA measurement. The white probe light is a continuous halogen lamp, while the pump light is a square-wave modulated green laser (530 nm, 4.33 Hz). Upon turning on the laser, the absorption at 750 nm rapidly in-creased and stabilized within ~ 0.3 s (Figure 42b). This absorption is due to the intraband absorption of electrons injected from dye molecules into TiO2.(96) Upon turning off the laser, the absorption signal decays back to zero within subseconds. The above processes stand for the charging and discharging of meso-TiO2 with light on and off, respectively.

However, the signal at 580 nm, characteristic of the Stark bleach, behaves strikingly different (Figure 42a). When the laser is turned on, after the initial increase within ~ 0.1 s, the Stark bleach then surprisingly decreased with a rather long time constant and eventually reached a plateau with 20% of bleach signal left. When the laser is turned off, the signal at 580 nm showed a reversed positive value, then decaying slowly back to zero. The initial in-crease at 580 nm is due to electron accumulation in meso-TiO2, which de-velops the local electric field and increases Stark effect. However, the fol-lowing slow decay and reversible behavior are rather intriguing. We further did control experiments by use of voltage modulation to accumulate elec-trons within TiO2, which can therefore exclude any contribution of oxidized dyes, and observed similar results (Paper IV). Moreover, without Li+ in the electrolyte, the slow decay and reversible behavior are significantly reduced (Figure 43).

Based on these observations, a slow and reversible charge compensation mechanism was proposed to occur at these interfaces, schematically shown in Figure 44: (a) the initial condition, (b) after electron injection, a local electric field is created between the elections accumulation within TiO2 and the surrounding cations. Sequentially, the cations will move slowly towards TiO2 in response to the electric field, resulting in the decrease of the electric field and lower Stark bleach signal, (c) a plateau is achieved at the end of charge compensation, indicating the equilibrium state of the system, and (d) upon turning off the light, electrons within TiO2 will be lost by recombining with the oxidized species in the electrolyte; The residual excess cations in the surface therefore create an electric field pointing into the electrolyte and cause the positive A. Finally, the cations diffuse back to the electrolyte and the system recover to the initial condition.

Notably, the charge compensation are rather slow, ~ ms to s, suggested by the above mechanism. Cations may have to lose its solvation shell in order to complete the charge compensation, which requires large activation energy.

67

Figure 44. Schematic pictures of the slow and reversible charge compensation mechanism proposed in Paper VI.

5.3.2 Factors affecting the local electric field In Paper V, I focused on the investigation of factors contributing to the strength of the local electric field at various cases of charge accumulation in meso-TiO2. These factors were systematically investigated at three condi-tions: a) surface cations adsorption b) potentiostatic accumulation of elec-trons and c) photoinduced electron injection into TiO2.

Small cations are well known for their surface affinity to the meso-TiO2. For example, H+ adsorption to TiO2 has resulted in the famous Nernstian shift, i.e. a 59 mV/pH positive shift of the band edge energy.(18, 97) Here, the absorption spectrum of LEG4-sensitized TiO2 film, abbreviated as LEG4/TiO2, is found to be gradually red-shifted during a Li+ titration exper-iment (Figure 45a). The change of the absorption spectrum can be well de-scribed by the Stark effect in the presence of a positive electric field (point-ing towards to the electrolyte direction). We therefore estimated the electric field by equation 14, with the results indicating that the field strength is in the order of ~ 1 MV/cm (Figure 45b).

By contrast, application of a negative potential (ranging from 0 V to -1 V) on LEG4/TiO2 accumulates electrons within TiO2 by a potentiostatic way, and was shown to create the local electric field and Stark effect in the similar order.

68

Figure 45. (a) UV-Vis absorption spectra of LEG4-sensitized TiO2 films during LiClO4 titration in ACN; the inset graph shows the difference absorption spectrum, A; (b) the change of A at 580 nm versus the LiClO4 concentration in ACN (left ordinate axis). The right ordinate axis shows the corresponding electric field at LEG4/TiO2/electrolyte interface according to Equation 14.

Further, we investigated the factors contributing to the Stark bleach in the case of photoinduced electron injection by PIA spectroscopy. As described in the Section 5.3.1, the absorption signal at 750-800 nm is attributed to the intraband absorptions of electrons injected into TiO2, while the absorption at 580 nm is from the Stark bleach caused by the electric field generated by those electrons.

In the case of varying concentration of TBP, Co(bpy)2+/3+ and light inten-sities in the DSSCs, normalization of the bleach signal at 750 nm to the sig-nal at 580 nm resulted in an constant value within experimental errors (Figure 46). These results can be explained by the fact that the Stark bleach is proportional to the local electric field and therefore is proportional to the electron concentration, AStark E e-. (Equation 14)

However, the situation is different in the case of varying the dye cover-ages on the meso-TiO2 and Li+ concentration in the electrolyte. Figure 47

69

shows the results of varying Li+ concentration as an example. Higher Li+ concentration in the electrolyte results in the increase of the absorption sig-nal at 750-800 nm, namely higher electron concentration. This is explained by the fact that Li+ can shift the conduction band in the positive direction, therefore resulting in better electron injection and better charge accumula-tion. Strikingly, the Stark bleach at 580 nm decreased with the higher elec-tron concentrations. Normalization of A at 750 nm to 580 nm therefore results in a clear nonlinear curve, significantly different with the case of TBP.

Figure 46. Effects of TBP concentration on the PIA spectrum of LEG4-sensitized TiO2 films. Electrolyte composition: 0.22 M Co(bpy)3(PF6)2, 0.05 M Co(bpy)3(PF6)3 in ACN and varying TBP concentration. (a) Averaged original PIA spectra of three samples vs. wavelength in the 480-800 nm range. The inset graph shows the zoomed-in PIA spectra between 750 nm to 800 nm. (b) Normalized PIA at 580 nm against PIA at 800 nm at the different TBP concentration.

Figure 47. Effects of LiClO4 concentration on the PIA spectrum of LEG4 sensi-tized-films. (a) Averaged original PIA of three samples with various LiClO4 concen-trations in the 480 nm to 800 nm range. The inset graph shows the zoomed-in spec-tra between 750 nm to 800 nm and (b) Normalized PIA at 580 nm against PIA at 800 nm at the different Li+ concentration.

70

The implication of the abnormal observation, the nonlinearity of the normal-ization curve, was further quantified by calculating the apparent interfacial capacitance by the equation:

(36)

where Q is the change of surface charge caused by the injection; E is the change of the electrical field calculated above; d is the length of the dye molecule, assumed to be 2 nm.

Here, Q is attributed to the accumulated electrons in meso-TiO2, calcu-lated from the electron absorption using the Beer-Lambert law. E can be calculated from the Stark bleach amplitude through Equation 14. Therefore, the results demonstrated that the surface capacitance improved nonlinearly and then saturated with the increase of Li+ concentration in the electrolyte (Figure 48). Classical capillary measurements have shown that the capaci-tance of the metal surface is strongly affected by the electrolyte ions concen-tration.

Figure 48. Helmholtz capacitance of electrolyte/LEG4/TiO2 interface vs Li+ concen-tration. The detailed descriptions of the calculation of the electron density, the elec-tric field strength and the interfacial capacitance are given in Paper V.

These results further confirm that the dye molecules on TiO2/dye/electrolyte are at least partially located within the Helmholtz layer(64–67), therefore vulnerable to the change of local environment. Changes of absorption spec-trum could serve as important indicator to the local field strength and direc-tion. Of great importance is that the Helmholtz layer capacitance is suggest-ed to change as a result of varying the cationic environment.

71

Figure 49. Schematic diagram of the implication regarding TiO2/dye/electrolyte interface in Paper V. Note: dye molecules are at least partially located within the Helmholtz layer (HL), therefore the outlayer of HL is drawn movable.

5.3.3 Effects of local electric field on the interfacial charge transfer A vital aspect of studying the local electric field and Stark effect is to eluci-date their impacts on the interfacial charge transfer processes, especially considering the strong strength of the field in the order of MV cm-1.

In Paper VI, I therefore investigated the correlation between the local electric field and interfacial charge transfer processes. Different cations (Li+, Na+, Mg2+ and Ca2+) were used to tune the electric field strength after elec-tron accumulation within meso-TiO2. Previously electrochemical studies have shown that the local electric field present at approximately 20 elec-trons/TiO2 particle are in the order of ECa2+(2.2 MV cm-1) EMg2+(1.8 MV cm-1) ELi+(1.3 MV cm-1) ENa+(1.1 MV cm-1).(69, 98) In this study, the transient absorption spectroscopic study has qualitatively confirmed the above order but with a slight difference: ECa2+ EMg2+ ENa+ ELi+ (Figure 50).

72

Figure 50. Comparison of the TAS of different cations, in which ∆A is normalized at 800 nm.

Two charge transfer processes were investigated: (a) the charge recombina-tion between electrons within TiO2 and the oxidized dye; (b) the regenera-tion of the oxidized dye by the redox mediator. These processes are studied under a large range of laser excitation intensities, in order to exclude the possible effects of electron concentrations on the kinetics.(55, 85) Mean-while, the decay halftimes are used to characterize the rate constants.

Figure 51. Comparison of the acquired half-time as a function of the electron densi-ty in TiO2 (a) the inert electrolyte: 0.1 M Li+ /Na+/Mg2+/Ca2+ perchlorate salt in ACN (b) the redox electrolyte: 0.1 M Li+ /Na+/Mg2+/Ca2+ perchlorate salt, 0.22 M Co(bpy)3(PF6)2 and 0.05 M Co(bpy)3(PF6)3 in ACN. The inset diagrams show the corresponding electron transfer processes measured. The shaded areas represent the 95% confidence region of the fittings.

Figure 51a shows that cationic environments affect noticeably the recom-bination processes, regardless of the electron concentration in TiO2, in the order t0.5,Li+ t0.5,Na+ t0.5,Mg2+ t0.5,Ca2+. Further correlation between the elec-tric field strength and the time constants demonstrates a linear relation-

a b

73

shipi.e. higher electric field resulting in smaller rate constants (Figure 52a). In contrast, the regeneration processes are relatively unaffected (Figure 51b and Figure 52b).

Figure 52. The relationship between the charge recombination time constants (a) and regeneration time constants (b) with its normalized electric field (a.u.) estimated from Figure 50.

These results implied that the effects of the local electric field on the interfa-cial charge transfer are spatially confined, possibly because the regeneration process occur far from the surface, and therefore the electric field has been effectively screened. Meanwhile, when the electron transfer process occur within the field, the driving force or the coupling between the dye molecule and the semiconductor can be changed, therefore altering the interfacial ki-netics observed.

74

6. Summary and outlook

The development of DSSCs demands better optimization and understanding of various light-induced electronic processes at the TiO2/dye/electrolyte interface. Alternative redox couples with high redox potential are still one of the most promising ways to further promote the efficiency of DSSCs due to its reduction of energy loss during dye regeneration. Organic redox couples, for example TEMPO, hold promise due to its fast dye regeneration kinetics, but it also resulted in the dramatic charge recombination between electrons in TiO2 (or even from FTO glass) and the oxidized redox species (Paper I). Further promise relies on the better surface passivation to reduce the recom-bination, but is very challenging. One possible alternative way is to intro-duce an electron-transfer cascade during the regeneration process (Paper II). The intermediate step can reduce recombination by the optimized but com-plicated kinetics. For the future work, new organic molecule donors should be further explored with the concept of these two redox mediator systems.

The long-term stability is the key to the commercialization of DSSCs. The multi-components nature of DSSCs restricts a simple mechanism of degradation. The stability of dye/TiO2/electrolyte interface deserves more attention, as suggested by Paper III.

At a fundamental level, the recent discovered local electric field at this in-terface intrigues various questions. The suggested slow charge compensation (Paper IV) and change of the interfacial capacitance (Paper V) demonstrates the abnormal features of these interfaces. In addition, this field has shown to play a significant role in the determination of the interfacial electron transfer processes (Paper VI), with spatially confined effects. The charge recombina-tion is accelerated with stronger local electric field, while the dye regenera-tion shows negligible dependence on the strength of the local electric field. The key feature for improved understanding is to quantify the relationship between the dye structure, electric field strength and the charge transfer ki-netics and to check the universality of the above conclusion. In order to achieve that, more dyes with specific structures should be designed, e.g., varying the packing density or changing the distance of charge transfer unit from the surface. Meanwhile, up to now, the effect of the electric field is still referred to the entire molecule. However, the spatial distribution of the local electric field is not clear yet, which could be provided by combining more

75

diverse spectroscopy, e.g., Raman and infrared. All these understandings would provide insights about this fundamental important interface, therefore guiding the rational design of dyes and interfacial engineering for the further development of DSSCs.

76

7. Populärvetenskaplig sammanfattning på svenska

Bakgrund Just nu består majoriteten av den energi vi behöver, för att värma huset, köra transporter och industri, av fossila bränslen, det vill säga olja, gas och kol. Dessa förbränningsprocesser frigör stora mängder av CO2 i atmosfären, lik-som små partiklar som orsakar allvarliga luftföroreningar och eventuella konstgjorda klimatförändringar, det vill säga den globala uppvärmningen. För att bygga ett hållbart samhälle måste vi därför hitta en renare, förnybar energikälla. Bland de olika förnybara energikällor som finns att välja på idag, är solenergi otvivelaktigt den mest attraktiva, på grund av sin renhet och stora kapacitet. Energin som vi får från solen på en timme är tillräckligt för att förse hela mänsklighetens energibehov under ett år.

Figure 53. Jämförelse mellan plantorna och solceller för omvandling av solenergi till kemi eller elektricitet

Växter har använt solenergi i miljontals år; bladen absorberar fotoner från solens ljus, och omvandlar den till biomassa och bioenergi (Figur 53). I denna avhandling arbetade jag med en viss typ av solceller, kallade färgäm-nessensitiserade solceller (Dye sensitized solar cells på engelska, förkortat

77

DSSC), som omvandlar solenergin till elektricitet genom att imitera hur växterna fungerar.

Grätzelsolcell

Figure 54. En schematisk bild av en typisk färgämnessensiterad solcell som under-söktes i avhandlingen. [Ox] och [Red] står för de oxiderade och reducerade redox formema.

Den grundläggande strukturen hos solcellen visas i Figur 54. Den består av tre delar:

1. En ljusabsorberande och omvandlande elektrod (arbetselektrod) Denna elektrod består av ett tunt lager med färgämnessensiterad TiO2, där ljus absorberas av färgämnesmolekylerna och omvandlas till elektroner, redo att transporteras. Färgämnesmolekylerna kan vara från naturen, t.ex. den mörkblå färgen från blåbär. I denna avhandling används effektivare färgäm-nesmolekyler, utformade av organiska kemister, för att underlätta ljus-absorptionen och förbättra solcellensprestandan. TiO2, som också används som vitt färgämne i både målarfärg och tandkräm fungerar som elektron-transporterande nätverk. Denna elektrod innefattar gränssnitte mellan TiO2/färgämne/elektrolyt, som utgör kärnan i solcellen.

2. Elektrolytsystemet. Elektrolyten är gjord av ett redoxpar , dvs ett par bestående av den oxide-rade, [Ox], och reducerade , [Red] versionen av en molekyl. Dessutom ingår flera andra ingredienser som är fördelaktiga för solcellens prestanda. Redox-

78

paret transporterar elektroner genom en diffusionsprocess, som tar elektro-nerna från motelektroden (se nedan) till arbetselektroden. Elektrolytsystemet ändrar också arbetselektrodens egenskaper genom mer subtila interaktioner som har stora effekter på solcellens prestanda.

3. Motelektroden Motelektroden är där den oxiderade [Ox] versionen av redoxparet i elektro-lyten omvandlas till den reducerade [Red] versionen. Arbetet med avhandlingen Spontant flödar elektroner alltid frånhöga energinivåea till låga. Därmed förekommer både önskade och oönskade reaktioner i systemet. Den övergri-pande inriktningen på avhandlingen är därför att förstå och kontrollera flödet av elektronerna i systemet, i syfte att skapa mer effektiva solceller. Specifikt fokuserar avhandlingen på förståelsen av de elektroniska processer som sker vid det färgämnessensiterade halvledare/elektrolyt gränssnittet. Frågorna som ställs inkluderar: Hur kan man styra sammansättningen av redoxsyste-met för att vägleda elektronernas flöde så att det blir så optimalt som möjligt (Paper I och II)? Vad är de begränsande faktorerna för solcellens prestanda under lång tid (Paper III)? Svaren på dessa två frågor har direkt implikation på solcellens effektivitet och hållbarhet. Samtidigt finns det mer grundläg-gande frågor när det gäller det färgämnessensiterade halvledare/elektrolyt gränssnittet. Nyligen upptäcktes ett lokalt elektriskt fält vid den färgämnes-sensiterade elektrodytan, men fältets funktion och dess potentiella effekter på systemet är oklart. Den andra delen av avhandlingen fokuserar därmed på att utreda de elektroniska processer som är relaterade till det elektriska fältet (Paper IV, V och VI). Dessa studier kommer därmed att ge ytterligare vik-tiga insikter och förståelse angående färgämnessensiterade halvle-dare/elektrolyt gränssnitt.

79

8. Acknowledgements

My PhD started when Lei Yang forwarded my CV to Prof. Anders Hagfeldt, and then followed by the interview with Erik Johansson and Gerrit Boschloo. It was right after I finished my master thesis related to polymer chemistry, and I had negligible knowledge about semiconductors and elec-trochemistry, not even to mention DSSCs. It is their great trust that opens me up this opportunity and adventure.

First, I would like to thank my main supervisor, Gerrit. Thank you for teaching me enormous things: how to do research, how to define and answer scientific questions. Thanks for all the patience, freedom and enormous sup-port throughout my PhD.

Many thanks to Anders, without whom my PhD would not start in Upp-sala at all. Thank you for always being positive, encouraging and inspiring, giving lots of generous help.

Thank Erik for always keeping the door open and answer my various questions.

I must thank my master of electrochemistry, my unofficial supervisor Nick Vlachopoulos. Thank you for keeping teaching me electrochemistry and revising my manuscripts, regardless if you are part of the projects, re-gardless if you are in Uppsala or in Lausanne.

Also, my master of chemicals, Leif Häggman. Thanks for all the secret "gifts", the home-designed apertures, as well as all the jokes, kindness, openness and curiosity, which I am impressed and inspired.

I would also like to thank Yan Hao, for teaching me how to make solar cells when I started as well as the continuous nice collaboration. Thank you for all the companies and cares during my PhD.

I owe enormous thanks to many great colleges: Anna Eriksson, for all the help in my writings and being an excellent swimming teacher; Kerttu Aitola for sharing so many thoughts and ideas towards life and science, and her cares in the life; Malin Johansson, for her extreme openness to collabo-ration and trying new stuffs; Marina Freitag, for organizing many group activities, parties and lots of nice collaboration; Hanna Ellis, for answering my enormous questions regarding Sweden and life in Sweden; Roger Jiang, for all the critical thought, constructive suggestion and all the enjoyable dai-ly lunch conversation, Kári Sveinbjörnsson, for all the organization and

80

help in the lab, as well as your openness; Meysam Pazoki, Dongqin Bi, Sagar Motilal Jain, for sharing the office and all the deep thoughts; Luca D’Amario, for all the help with the laser; Valentina leandri, for the fun time together when I first came to the group; Ulrika Jansson, for taking care of many administrative stuffs; Michael Wang, for all the legendary stories and experience. Also, many thanks to Xiaoliang Zhang, Jinbao Zhang, Victor, Zahra, Rodrigo, Huimin Zhu, Lei Yang for their kind help in the lab and in life, and all the nice time together.

I would also like to thank Haining for many inspiring conversations about science and career; thank Tomas Edvinsson for all the nice collabora-tion, for teaching me how to conduct DFT calculation and sharing many discussions and passions about science; thank Prof. Håkan Rensmo for organizing me to MAX-IV, as well as Johan Oscarsson who took great care of me there.

I am grateful to stay in many dynamic and interdisciplinary research envi-ronments. Thanks all the current and previous CMD members, especially Prof. Lars Kloo, for always being positive and motivational. Thanks the physical chemistry department, especially Prof. Leif Hammarström, for all the nice research environment, wonderful lectures and seminars. I am proud of being a member of these groups.

Also, thank Magnus Söderberg for his excellent summer work, Edgar Mijangos, for the nice collaboration on the TPAA paper.

I am grateful to all the funding and scholarships that support my study and attendance of the conferences.

I would also like to thank many of my Chinese friends for their caring and wonderful time together, especially, Meiyuan Guo, Teng Zhang, Shihuai Wang, Lei Zhang and Mingzhi Jiao.

Finally, I would like to thank my parents. I have been away from home to study for more than 10 years. Their support and understanding also accom-panies me; without them, I can never make it. Also, thank my girlfriend, Xia, for her endless love and support.

When writing this acknowledgement, I realized it could never be com-pleted. Thanks for all the love and cares I received, all the great people I met in the past years. I am sincerely grateful to all of them.

81

8. References

1. A. Cazenave et al., The rate of sea-level rise. Nat. Clim. Chang. 4, 358–361 (2014).

2. D. a Lashof, D. R. Ahuja, Relative contributions of greenhouse gas emissions to global warming. Nature. 344, 529–531 (1990).

3. G. P. Peters, E. G. Hertwich, CO2 Embodied in International Trade with Implications for Global Climate Policy. Environ. Sci. Technol. 42, 1401–1407 (2008).

4. N. S. Lewis, D. G. Nocera, Powering the planet: Chemical challenges in solar energy utilization. Proc. Natl. Acad. Sci. 103, 15729–15735 (2006).

5. D. Mills, Advances in solar thermal electricity technology. Sol. Energy. 76, 19–31 (2004).

6. W. J. Youngblood, S.-H. A. Lee, K. Maeda, T. E. Mallouk, Visible Light Water Splitting Using Dye-Sensitized Oxide Semiconductors. Acc. Chem. Res. 42, 1966–1973 (2009).

7. M. G. Walter et al., Solar Water Splitting Cells. Chem. Rev. 110, 6446–6473 (2010).

8. E. L.Lewitschnik, Solrevolution. Number 9789155801557 in 20009615 Naturskayddsfören,Årgång 106.

9. N. S. Lewis, Science, in press, doi:10.1126/science.aad1920. 10. N. S. Lewis, Toward Cost-Effective Solar Energy Use. Science. 315, 798–801

(2007). 11. M. A. Green, Solar cells: operating principles, technology, and system

applications (Prentice-Hall, Inc.,Englewood Cliffs, NJ, 1982). 12. P. Jackson et al., New world record efficiency for Cu(In,Ga)Se2 thin-film solar

cells beyond 20%. Prog. Photovoltaics Res. Appl. 19, 894–897 (2011). 13. T. M. Clarke, J. R. Durrant, Charge Photogeneration in Organic Solar Cells.

Chem. Rev. 110, 6736–6767 (2010). 14. A. Hagfeldt, G. Boschloo, L. Sun, L. Kloo, H. Pettersson, Dye-sensitized solar

cells. Chem. Rev. 110, 6595–663 (2010). 15. M. M. Lee, J. Teuscher, T. Miyasaka, T. N. Murakami, H. J. Snaith, Efficient

Hybrid Solar Cells Based on Meso-Superstructured Organometal Halide Perovskites. Science. 338, 643–647 (2012).

16. H.-S. Kim et al., Lead Iodide Perovskite Sensitized All-Solid-State Submicron Thin Film Mesoscopic Solar Cell with Efficiency Exceeding 9%. Sci. Rep. 2, 591 (2012).

17. H. Gerischer, M. E. Michel-Beyerle, F. Rebentrost, H. Tributsch, Sensitization of charge injection into semiconductors with large band gap. Electrochim. Acta. 13, 1509–1515 (1968).

18. H. Gerischer, The impact of semiconductors on the concepts of electrochemistry. Electrochim. Acta. 35, 1677–1699 (1990).

82

19. R. Memming, F. Schröppel, U. Bringmann, Sensitized oxidation of water by tris(2,2′-bipyridyl) ruthenium at SnO2 electrodes. J. Electroanal. Chem. Interfacial Electrochem. 100, 307–318 (1979).

20. R. Memming, Semiconductor Electrochemistry (Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2015).

21. P. Liska, N. Vlachopoulos, M. K. Nazeeruddin, P. Comte, M. Graetzel, cis-Diaquabis(2,2’-bipyridyl-4,4’-dicarboxylate)ruthenium(II) sensitizes wide band gap oxide semiconductors very efficiently over a broad spectral range in the visible. J. Am. Chem. Soc. 110, 3686–3687 (1988).

22. N. Vlachopoulos, P. Liska, J. Augustynski, M. Graetzel, Very efficient visible light energy harvesting and conversion by spectral sensitization of high surface area polycrystalline titanium dioxide films. J. Am. Chem. Soc. 110, 1216–1220 (1988).

23. B. O’Regan, M. Grätzel, A low-cost, high-efficiency solar cell based on dye-sensitized colloidal TiO2 films. Nature. 353, 737–740 (1991).

24. J. Yang, A. Banerjee, S. Guha, Triple-junction amorphous silicon alloy solar cell with 14.6% initial and 13.0% stable conversion efficiencies. Appl. Phys. Lett. 70 (1997).

25. L. M. Peter, The Grätzel Cell: Where Next? J. Phys. Chem. Lett. 2, 1861–1867 (2011).

26. P. M. Sommeling et al., Influence of a TiCl4 Post-Treatment on Nanocrystalline TiO2 Films in Dye-Sensitized Solar Cells. J. Phys. Chem. B. 110, 19191–19197 (2006).

27. A. Hinsch, Status of the Dye Solar Cell Technology (DSC) As a Guideline for Further Research. 28th Eur. PV Sol. Energy Conf. Exhib., 1–9 (2013).

28. S. Mathew et al., Dye-sensitized solar cells with 13% efficiency achieved through the molecular engineering of porphyrin sensitizers. Nat. Chem. 6, 242–247 (2014).

29. Y. Hao et al., ACS Appl. Mater. Interfaces, in press, doi:10.1021/acsami.6b09671.

30. T. P. Chou, Q. Zhang, G. E. Fryxell, G. Cao, Hierarchically structured ZnO film for dye-sensitized solar cells with enhanced energy conversion efficiency. Adv. Mater. 19, 2588–2592 (2007).

31. M. Quintana, T. Edvinsson, A. Hagfeldt, G. Boschloo, Comparison of dye-sensitized ZnO and TiO2 solar cells: Studies of charge transport and carrier lifetime. J. Phys. Chem. C. 111, 1035–1041 (2007).

32. B. Liu, E. S. Aydil, Growth of oriented single-crystalline rutile TiO2 nanorods on transparent conducting substrates for dye-sensitized solar cells. J. Am. Chem. Soc. 131, 3985–3990 (2009).

33. M. Adachi et al., Highly Efficient Dye-Sensitized Solar Cells with a Titania Thin-Film Electrode Composed of a Network Structure of Single-Crystal-like TiO2 Nanowires Made by the “Oriented Attachment” Mechanism. J. Am. Chem. Soc. 126, 14943–14949 (2004).

34. M. K. Nazeeruddin et al., Conversion of light to electricity by cis-X2bis(2,2’-bipyridyl-4,4’-dicarboxylate)ruthenium(II) charge-transfer sensitizers (X = Cl-, Br-, I-, CN-, and SCN-) on nanocrystalline titanium dioxide electrodes. J. Am. Chem. Soc. 115, 6382–6390 (1993).

35. H. Nusbaumer, J. Moser, S. M. Zakeeruddin, M. K. Nazeeruddin, M. Grätzel, Co II (dbbip)2

2+ Complex Rivals Tri-iodide/Iodide Redox Mediator in Dye-Sensitized Photovoltaic Cells. J. Phys. Chem. B. 105, 10461–10464 (2001).

83

36. E. Gabrielsson et al., Convergent/Divergent Synthesis of a Linker-Varied Series of Dyes for Dye-Sensitized Solar Cells Based on the D35 Donor. Adv. Energy Mater. 3, 1647–1656 (2013).

37. Z. Yao et al., Donor/Acceptor Indenoperylene Dye for Highly Efficient Organic Dye-Sensitized Solar Cells. J. Am. Chem. Soc. 137, 3799–3802 (2015).

38. T. Horiuchi, H. Miura, K. Sumioka, S. Uchida, High efficiency of dye-sensitized solar cells based on metal-free indoline dyes. J. Am. Chem. Soc. 126, 12218–9 (2004).

39. X. Wang, L. Zhi, K. Müllen, Transparent, conductive graphene electrodes for dye-sensitized solar cells. Nano Lett. 8, 323–7 (2008).

40. H. Ellis et al., PEDOT counter electrodes for dye-sensitized solar cells prepared by aqueous micellar electrodeposition. Electrochim. Acta. 107, 45–51 (2013).

41. U. Bach et al., Solid-state dye-sensitized mesoporous TiO2 solar cells with high photon-to-electron conversion efficiencies. Nature. 395, 583–585 (1998).

42. I. Chung, B. Lee, J. He, R. P. H. Chang, M. G. Kanatzidis, All-solid-state dye-sensitized solar cells with high efficiency. Nature. 485, 486–489 (2012).

43. W. Shockley, H. J. Queisser, Detailed Balance Limit of Efficiency of p-n Junction Solar Cells. J. Appl. Phys. 32, 510 (1961).

44. M. Grätzel, Photoelectrochemical cells. Nature. 414, 338–344 (2001). 45. W. J. Albery, P. P. N. Bartlett, The transport and kinetics of photogenerated

carriers in colloidal semiconductor electrode particles. J. Electrochem. Soc. 131, 315–325 (1984).

46. B. O’Regan, J. Moser, M. Anderson, M. Graetzel, Vectorial electron injection into transparent semiconductor membranes and electric field effects on the dynamics of light-induced charge separation. J. Phys. Chem. 94, 8720–8726 (1990).

47. L. M. Peter, Dye-sensitized nanocrystalline solar cells. Phys. Chem. Chem. Phys. 9, 2630–42 (2007).

48. S. Soedergren, A. Hagfeldt, J. Olsson, S.-E. Lindquist, Theoretical Models for the Action Spectrum and the Current-Voltage Characteristics of Microporous Semiconductor Films in Photoelectrochemical Cells. J. Phys. Chem. 98, 5552–5556 (1994).

49. C. Di Valentin, G. Pacchioni, A. Selloni, Reduced and n-Type Doped TiO2 : Nature of Ti3+ Species. J. Phys. Chem. C. 113, 20543–20552 (2009).

50. A. M. Czoska et al., The nature of defects in fluorine-doped TiO2. J. Phys. Chem. C. 112, 8951–8956 (2008).

51. P. Tiwana, P. Parkinson, M. B. Johnston, H. J. Snaith, L. M. Herz, Ultrafast terahertz conductivity dynamics in mesoporous TiO2: Influence of dye sensitization and surface treatment in solid-state dye-sensitized solar cells. J. Phys. Chem. C. 114, 1365–1371 (2010).

52. S. Nakade et al., Influence of TiO2 Nanoparticle Size on Electron Diffusion and Recombination in Dye-Sensitized TiO2 Solar Cells. J. Phys. Chem. B. 107, 8607–8611 (2003).

53. H. Wang, L. M. Peter, Influence of electrolyte cations on electron transport and electron transfer in dye-sensitized solar cells. J. Phys. Chem. C. 116, 10468–10475 (2012).

84

54. T. T. O. Nguyen, L. M. Peter, H. Wang, Characterization of Electron Trapping in Dye-Sensitized Solar Cells by Near-IR Transmittance Measurements. J. Phys. Chem. C. 113, 8532–8536 (2009).

55. J. R. Durrant, S. A. Haque, E. Palomares, Towards optimisation of electron transfer processes in dye sensitised solar cells. Coord. Chem. Rev. 248, 1247–1257 (2004).

56. R. Mohammadpour, A. Irajizad, A. Hagfeldt, G. Boschloo, Comparison of Trap-state Distribution and Carrier Transport in Nanotubular and Nanoparticulate TiO2 Electrodes for Dye-Sensitized Solar Cells. ChemPhysChem. 11, 2140–2145 (2010).

57. A. Zaban, A. Meier, B. A. Gregg, Electric Potential Distribution and Short-Range Screening in Nanoporous TiO2 Electrodes. J. Phys. Chem. B. 101, 7985–7990 (1997).

58. P. R. F. Barnes et al., Interpretation of Optoelectronic Transient and Charge Extraction Measurements in Dye-Sensitized Solar Cells. Adv. Mater. 25, 1881–1922 (2013).

59. J. Bisquert, F. Fabregat-Santiago, I. Mora-Seró, G. Garcia-Belmonte, S. Giménez, Electron Lifetime in Dye-Sensitized Solar Cells: Theory and Interpretation of Measurements. J. Phys. Chem. C. 113, 17278–17290 (2009).

60. J. Nelson, Continuous-time random-walk model of electron transport in nanocrystalline TiO2 electrodes. Phys. Rev. B. 59 (1999), pp. 15374–15380.

61. J. Bisquert, V. S. Vikhrenko, Interpretation of the Time Constants Measured by Kinetic Techniques in Nanostructured Semiconductor Electrodes and Dye-Sensitized Solar Cells. J. Phys. Chem. B. 108, 2313–2322 (2004).

62. E. Hendry, M. Koeberg, B. O’Regan, M. Bonn, Local Field Effects on Electron Transport in Nanostructured TiO2 Revealed by Terahertz Spectroscopy. Nano Lett. 6, 755–759 (2006).

63. Allen J.Bard, L. R.Faulkner, Electrochemical Methods: Fundamentals and Applications (Wiley Subscription Services, Inc., A Wiley Company, second edi., 2001).

64. A. Zaban, S. Ferrere, B. A. Gregg, Relative Energetics at the Semiconductor/Sensitizing Dye/Electrolyte Interface. J. Phys. Chem. B. 102, 452–460 (1998).

65. A. Zaban, S. Ferrere, J. Sprague, B. A. Gregg, pH-Dependent Redox Potential Induced in a Sensitizing Dye by Adsorption onto TiO2. J. Phys. Chem. B. 101, 55–57 (1997).

66. S. Ardo, Y. Sun, A. Staniszewski, F. N. Castellano, G. J. Meyer, Stark Effects after Excited-State Interfacial Electron Transfer at Sensitized TiO2 Nanocrystallites. J. Am. Chem. Soc. 132, 6696–6709 (2010).

67. U. B. Cappel, S. M. Feldt, J. Schöneboom, A. Hagfeldt, G. Boschloo, The influence of local electric fields on photoinduced absorption in dye-sensitized solar cells. J. Am. Chem. Soc. 132, 9096–101 (2010).

68. W. Yang, Y. Hao, N. Vlachopoulos, A. I. K. Eriksson, G. Boschloo, Studies on the Interfacial Electric Field and Stark Effect at the TiO2/Dye/Electrolyte Interface. J. Phys. Chem. C. 120, 22215–22224 (2016).

69. R. M. O’Donnell, R. N. Sampaio, T. J. Barr, G. J. Meyer, Electric Fields and Charge Screening in Dye Sensitized Mesoporous Nanocrystalline TiO2 Thin Films. J. Phys. Chem. C. 118, 16976–16986 (2014).

85

70. K. Oum et al., Ultrafast dynamics of the indoline dye D149 on electrodeposited ZnO and sintered ZrO2 and TiO2 thin films. Phys. Chem. Chem. Phys. 14, 15429–37 (2012).

71. P. G. Johansson, A. Kopecky, E. Galoppini, G. J. Meyer, Distance dependent electron transfer at TiO2 interfaces sensitized with phenylene ethynylene bridged RuII-isothiocyanate compounds. J. Am. Chem. Soc. 135, 8331–8341 (2013).

72. R. M. O’Donnell, S. Ardo, G. J. Meyer, Charge-Screening Kinetics at Sensitized TiO2 Interfaces. J. Phys. Chem. Lett. 4, 2817–2821 (2013).

73. W. Liptay, Dipole moments and polarizabilities of molecules in excited electronic states (ACADEMIC PRESS, INC., 1974), vol. 1.

74. W. Liptay, Electrochromism and Solvatochromism. Angew. Chemie Int. Ed. English. 8, 177–188 (1969).

75. R. Luchowski, S. Krawczyk, Electroabsorption (Stark effect) spectroscopy of monomeric purine and pyrimidine bases. Chem. Phys. 314, 309–316 (2005).

76. N. Ohta, Y. Ogata, S. Okazaki, I. Yamazaki, External electric field effects on absorption and fluorescence spectra of indocarbocyanine in Langmuir-Blodgett films. Chem. Phys. Lett. 244, 355–362 (1995).

77. E. JALVISTE, N. OHTA, Theoretical foundation of electroabsorption spectroscopy: Self-contained derivation of the basic equations with the direction cosine method and the Euler angle method. J. Photochem. Photobiol. C Photochem. Rev. 8, 30–46 (2007).

78. S. G. Boxer, Stark realities. J. Phys. Chem. B. 113, 2972–2983 (2009). 79. W. Yang, M. Pazoki, A. I. K. Eriksson, Y. Hao, G. Boschloo, A key discovery

at the TiO2/dye/electrolyte interface: slow local charge compensation and a reversible electric field. Phys. Chem. Chem. Phys. 17, 16744–16751 (2015).

80. A. J. Bard, A. B. Bocarsly, F. R. F. Fan, E. G. Walton, M. S. Wrighton, The concept of Fermi level pinning at semiconductor/liquid junctions. Consequences for energy conversion efficiency and selection of useful solution redox couples in solar devices. J. Am. Chem. Soc. 102, 3671–3677 (1980).

81. A. Zaban, M. Greenshtein, J. Bisquert, Determination of the electron lifetime in nanocrystalline dye solar cells by open-circuit voltage decay measurements. Chemphyschem. 4, 859–64 (2003).

82. S. M. Feldt et al., Design of organic dyes and cobalt polypyridine redox mediators for high-efficiency dye-sensitized solar cells. J. Am. Chem. Soc. 132, 16714–24 (2010).

83. G. Boschloo, A. Hagfeldt, Photoinduced absorption spectroscopy as a tool in the study of dye-sensitized solar cells. Inorganica Chim. Acta. 361, 729–734 (2008).

84. M. N. Berberan-Santos, E. N. Bodunov, B. Valeur, Mathematical functions for the analysis of luminescence decays with underlying distributions 1. Kohlrausch decay function (stretched exponential). Chem. Phys. 315, 171–182 (2005).

85. K. C. D. Robson, K. Hu, G. J. Meyer, C. P. Berlinguette, Atomic level resolution of dye regeneration in the dye-sensitized solar cell. J. Am. Chem. Soc. 135, 1961–71 (2013).

86. C. L. Ward, R. M. O’Donnell, B. N. DiMarco, G. J. Meyer, Kinetic Resolution of Charge Recombination and Electric Fields at the Sensitized TiO2 Interface. J. Phys. Chem. C. 119, 25273–25281 (2015).

86

87. R. N. Sampaio, R. M. O’Donnell, T. J. Barr, G. J. Meyer, Electric Fields Control TiO2 (e– ) + I3

– → Charge Recombination in Dye-Sensitized Solar Cells. J. Phys. Chem. Lett. 5, 3265–3268 (2014).

88. J. Nelson, R. E. Chandler, Random walk models of charge transfer and transport in dye sensitized systems. Coord. Chem. Rev. 248, 1181–1194 (2004).

89. L. Zhang et al., Ultrafast and slow charge recombination dynamics of diketopyrrolopyrrole–NiO dye sensitized solar cells. Phys. Chem. Chem. Phys. 18, 18515–18527 (2016).

90. G. Boschloo, A. Hagfeldt, Characteristics of the iodide/triiodide redox mediator in dye-sensitized solar cells. Acc. Chem. Res. 42, 1819–26 (2009).

91. S. M. Feldt, G. Wang, G. Boschloo, A. Hagfeldt, Effects of Driving Forces for Recombination and Regeneration on the Photovoltaic Performance of Dye-Sensitized Solar Cells using Cobalt Polypyridine Redox Couples. J. Phys. Chem. C. 115, 21500–21507 (2011).

92. Z. Zhang, P. Chen, T. N. Murakami, S. M. Zakeeruddin, M. Grätzel, The 2,2,6,6-Tetramethyl-1-piperidinyloxy Radical: An Efficient, Iodine- Free Redox Mediator for Dye-Sensitized Solar Cells. Adv. Funct. Mater. 18, 341–346 (2008).

93. M. O. Reese et al., Consensus stability testing protocols for organic photovoltaic materials and devices. Sol. Energy Mater. Sol. Cells. 95, 1253–1267 (2011).

94. M. Pazoki, P. W. Lohse, N. Taghavinia, A. Hagfeldt, G. Boschloo, The effect of dye coverage on the performance of dye-sensitized solar cells with a cobalt-based electrolyte. Phys. Chem. Chem. Phys. 16, 8503–8 (2014).

95. B. N. DiMarco, R. M. O’Donnell, G. J. Meyer, Cation-Dependent Charge Recombination to Organic Mediators in Dye-Sensitized Solar Cells. J. Phys. Chem. C. 119, 21599–21604 (2015).

96. G. Boschloo, A. Hagfeldt, Photoinduced absorption spectroscopy of dye-sensitized nanostructured TiO2. Chem. Phys. Lett. 370, 381–386 (2003).

97. H. Gerischer, Neglected problems in the pH dependence of the flatband potential of semiconducting oxides and semiconductors covered with oxide layers. Electrochim. Acta. 34, 1005–1009 (1989).

98. R. M. O ’donnell et al., Photoacidic and Photobasic Behavior of Transition Metal Compounds with Carboxylic Acid Group(s). J. Am. Chem. Soc. 138, 3891–3903 (2016).

99. L. A. Harris, A Titanium Dioxide Hydrogen Detector. J. Electrochem. Soc. 127, 2657 (1980).

100. F. A. Grant, Properties of Rutile (Titanium Dioxide). Rev. Mod. Phys. 31, 646–674 (1959).

101. A. Wypych et al., Dielectric Properties and Characterisation of Titanium Dioxide Obtained by Different Chemistry Methods. J. Nanomater. 2014, 1–9 (2014).

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1464

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-310191

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2017