Organic Solar Cell

Architectures
PhD Thesis
by
Dipl.Ing. Klaus Petritsch
presented to the
Technisch-Naturwissenschaftliche Fakult¨at
der Technischen Universit¨at Graz (Austria)
under the supervision of
Cavendish Professor, Prof. Richard Friend,
(University of Cambridge, United Kingdom)
and
Univ.Prof.Dr.techn.Dipl.Ing. G¨ unther Leising
(Technische Universit¨at Graz, Austria)
Cambridge and Graz, July 2000
ii
For my dear mother
“If we did all the things we are
capable of doing, we would
astound ourselves.”
Thomas A. Edison
iii
iv
Acknowledgement
I want to thank both Univ.Prof.Dr.G¨ unther Leising and Univ.Prof.Dr. Richard Friend who together made
it possible for me to spend these precious three years in the opto-electronics group at the Cavendish Lab-
oratory in Cambridge (UK) with all its benefits (traveling to conferences, excellent lectures and research
facilities etc..) where I could learn so much not only about science but also life in general.
Thanks to the special relationship between the Cavendish Laboratory and its spin-off company Cambridge
Display Technology (CDT) I had the opportunity to use the excellent AFM microscope and thereby gain
a little insight into the industrial aspect of leading edge technology and research in this exciting new field
- for which I am very grateful.
I am particularly indebted to Dr.A.Lux for her patience when explaining organic chemistry to me and
pointing out which of my “dream molecules” could be synthesised (and how!).
For many interesting and exciting discussions of conjugated polymer science I want to thank J.J. Dittmer,
Dr.M.G. Harrison, Dr.E.Moons, A.Koch as well as Dr.A.N.Ramaprakash who spent many hours to
read and comment on different parts of the manuscript. Furthermore Dr.H.Ago, A.C.Arias, T.Brown,
F.Cacialli, N.Chawdherry, Dr.G.Denton, Dr.M.Granstr¨om, Dr.N.Greenham, P.Ho, Dr.A.K¨ohler, Dr.R.
Lazzaroni, Dr.E.A.Marseglia, Dr.J. de Mello, Dr. G.G.Rozenberg, Dr.H.Sirringhaus, Dr.P.Spearman,
Dr.N.Tessler, S.Thomas, V.Wan and B.Weir.
I would like to thank my mother Gertrude Petritsch and my uncle Arnfried Meyer for their generous
financial support as well as the TMR-Network SELOA.
Last but not least, I want to thank all fellow salseros and tangueros who made even my little free time
highly enjoyable during these intense three years in Cambridge, Graz, London, Paris, Vienna, Faro,
Bologna...
I want to note that, unlike all other results described in this thesis - the experimental part of the Section
on the laminated polymer device in Chapter 6 was carried out by Dr. M. Granstr¨om. However, the
realisation and success of this project was only possible as a result of an excellent collaboration among
the coauthors in Ref.[94] - without each of them this work would not have been possible.
With the exception of the laminated polymer device, more than 95% of the actual experimental and
theoretical work as well as 100% of the writing was performed by the author.
However, the term “we” is used throughout the text to underline the fact that every single result of
the author’s work as presented in this thesis was only possible because of the provision of equipment,
materials and scientific input from others.
This is also reflected in the long list of people mentioned in this acknowledgment and consistent with the
fact that modern research relies on collaboration and teamwork.
v
vi
S U N L I G H T H A R V E S T I N G
Sunlight provides the energy for almost all life on Earth.
It is harvested by plants, including the microscopic al-
gae living in water. Photons of sunlight (represented as
red dots) enter the food-producing regions within the al-
gal cell which act like solar panels. Here the molecules of
the green pigment, chlorophyll, are arranged in clusters
and absorb the energy of the photons. That energy is
used to turn carbon dioxide and water into food, releas-
ing excess oxygen. Algae are eaten by daphnia (water
fleas) and these are caught by roach, small fish which
are preyed on by herons. The heron ultimately depends
on a vast number of algal cells to supply it with food.
The scale of sizes in the painting gradually changes from
the magnified part of a single algal cell, covering the
lower third of the painting, to the Earths atmosphere at
the top of the painting. The oxygen in our atmosphere
was itself created by ancestral forms of algae, beginning
millions of years ago, when they were among the most
advanced life forms on Earth. They changed the ancient
atmosphere of the Earth, making it possible for oxygen-
breathing animals to evolve. Even today, about 90% of
all oxygen production and carbon dioxide use is carried
out by algae living in the seas and freshwaters.
From an oil painting by Glynn Gorick, 113 Hemingford Road,
Cambridge CB1 3BY, UK. c G. Gorick 1994
vii
viii
Contents
1 Motivation - Outline 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 The Combustion of Fossil Fuels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Renewable Energy Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Inorganic Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Organic Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Abbreviations and Synonyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Solar Cells - General 5
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 General Properties of Organic Semiconductors . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 Polarons and Polaron Excitons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.2 P- and n- type Organic Semiconductors . . . . . . . . . . . . . . . . . . . . . . . . 10
Fermi-level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Doping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.3 Structural Properties of Organic Semiconductors . . . . . . . . . . . . . . . . . . . 11
2.2.4 Photogeneration of Free Charges . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Photogeneration due to Oxygen Traps . . . . . . . . . . . . . . . . . . . . . . . . . 12
The Donor/Acceptor Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.5 D/A Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.6 Electrode/Semiconductor Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Current Through Barriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Electrode Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Photovoltaic Characterisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4.1 Power Conversion Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4.2 The Equivalent Circuit Diagram (ECD) . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4.3 Open Circuit Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
The Ideal Solar Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Effect of R
sh
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Effect of Shorts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
The Effect of Photoluminescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
The Upper Limit of V
oc
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4.4 Photocurrent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
The Short Circuit Current I
sc
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
The Optical Filter Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.4.5 Interpretation of IV Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Inorganic Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Organic Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4.6 Space Charges & Traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
ix
3 Single Layer Devices 31
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.1.1 Optical Properties of Organic Photovoltaic Materials . . . . . . . . . . . . . . . . . 31
3.1.2 Performance of Single Layer Devices . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.1.3 Doping of Organic Photovoltaic Materials . . . . . . . . . . . . . . . . . . . . . . . 33
3.1.4 Desired Properties of Organic Photovoltaic Materials . . . . . . . . . . . . . . . . . 34
3.1.5 Electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2 LPPPT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3 MEH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4 PTV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.5 PIF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.6 THPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.7 Per . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.8 Ter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.9 Single Layer Devices - Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4 Double Layer Devices 67
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.2 The Dye/Pigment Double Layer Device (HPc/Per3) . . . . . . . . . . . . . . . . . . . . . 70
4.3 The Dye/Dye Double Layer Device (CuPc/Per) . . . . . . . . . . . . . . . . . . . . . . . 73
5 Blend Layer Devices 75
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.2 HPc+Per . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.3 HPc+PIF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.4 PTV+Per . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.5 PTV+Ter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.6 PTV+THPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.7 PTV+PIF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6 Laminated Layer Devices 95
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.2 The Laminated Polymer Device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
PL Quenching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.3 The Laminated Dye Device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
PL Quenching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
7 Discotic Liquid Crystals as Solar Cell Materials 105
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
7.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.3 Liquid Crystalline Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.4 Photovoltaic Properties of HPc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
7.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
8 Determination of Solar Cell Efficiencies 115
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
8.2 Solar Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
8.3 Simulation of Solar Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
8.3.1 Calculation of the Required AM1.5d Filter . . . . . . . . . . . . . . . . . . . . . . 118
8.3.2 The Simulator Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
8.4 “Numerical” Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
8.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
x
9 Summary and Bibliography 123
9.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
9.1.1 Suggestions for Future Investigations . . . . . . . . . . . . . . . . . . . . . . . . . . 124
9.2 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
10 Appendices 137
10.1 Details of the Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
10.1.1 ITO Substrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
10.1.2 Preparation of the Dye and Polymer Solutions . . . . . . . . . . . . . . . . . . . . 137
10.1.3 Film Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
10.1.4 Sublimation of the Top Electrode . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
10.2 Details of the EQE and IV Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
10.2.1 Note on the Dark Current at 0V . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
10.3 Absorption Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
10.4 Thickness Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
10.5 Thermal Polarisation Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
10.6 List of Publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
10.7 Conferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
10.8 Curriculum Vitae of the Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
xi
xii
List of Figures
1.1 Atmospheric CO
2
and human population growth. . . . . . . . . . . . . . . . . . . . . . . . 1
2.1 Device structure of a typical organic solar cell. . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Energy conversion steps and loss mechanism in a solar cell. . . . . . . . . . . . . . . . . . 6
2.3 The different D/A architectures used in organic solar cells. . . . . . . . . . . . . . . . . . . 7
2.4 IV characteristics related to band energy diagrams. . . . . . . . . . . . . . . . . . . . . . . 8
2.5 Polarons and singlet exciton . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.7 Examples of structures for the different types of organic semiconductors . . . . . . . . . . 11
2.6 Categories of organic semiconductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.8 Organic semiconductor/semiconductor interfaces . . . . . . . . . . . . . . . . . . . . . . . 13
2.9 Band bending of D/A devices with electrode interfaces . . . . . . . . . . . . . . . . . . . . 15
2.10 Solar cell efficiency limits as a function of the semiconductor bandgap. . . . . . . . . . . . 18
2.11 IV characteristics and fillfactor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.12 ECD of a solar cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.13 Extended ECD of a solar cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.14 Optical filter and active layer in device structure. . . . . . . . . . . . . . . . . . . . . . . . 26
2.15 Two scenarios for the formation of a thin active layer . . . . . . . . . . . . . . . . . . . . . 26
2.16 Effect of R
sh
on the IV characteristics 10
4
≤ R
sh
≤ 10
7
Ω - inorganic cell. . . . . . . . . . 27
2.17 Effect of R
sh
on the IV characteristics 10
1
≤ R
sh
≤ 10
4
Ω - inorganic cell. . . . . . . . . . 27
2.18 Effect of R
s
on the IV characteristics. Inorganic cell. . . . . . . . . . . . . . . . . . . . . . 28
2.19 Effect of R
sh
on the IV characteristics 10
7
≤ R
sh
≤ 10
10
Ω - organic cell. . . . . . . . . . . 28
2.20 Effect of R
sh
on the IV characteristics 10
3
≤ R
sh
≤ 10
6
Ω - organic cell. . . . . . . . . . . 29
2.21 Effect of R
s
on the IV characteristics. Organic cell. . . . . . . . . . . . . . . . . . . . . . . 29
2.22 Effect of light intensity on the IV curve in an organic cell. . . . . . . . . . . . . . . . . . . 30
2.23 IV characteristics for space charge limited current with deep and shallow traps . . . . . . 30
3.1 Device architecture of a single layer solar cell. . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2 Absorption versus layer thickness (transparent rear contact) . . . . . . . . . . . . . . . . . 32
3.3 Absorption versus layer thickness (reflective rear contact) . . . . . . . . . . . . . . . . . . 32
3.4 Chemical structure of LPPPT. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.5 EQE and absorption for a thick and a thin LPPPT device. . . . . . . . . . . . . . . . . . 37
3.6 Thickness dependence of EQE (at 536nm) for LPPPT devices . . . . . . . . . . . . . . . . 38
3.7 EQE of LPPP versus thickness at 536nm . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.8 Transmission of the Al electrode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.9 EQE of LPPPT films for different thickness when illuminated through AL. . . . . . . . . 40
3.10 IV characteristics of the 205nm LPPPT device. . . . . . . . . . . . . . . . . . . . . . . . . 40
3.11 IV characteristics of the 185nm LPPPT device. . . . . . . . . . . . . . . . . . . . . . . . . 41
3.12 IV characteristics of the 122nm LPPPT device. . . . . . . . . . . . . . . . . . . . . . . . . 41
3.13 IV characteristics of the 117nm LPPPT device. . . . . . . . . . . . . . . . . . . . . . . . . 41
3.14 IV characteristics of the 93nm LPPPT device. . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.15 IV characteristics of the 67nm LPPPT device. . . . . . . . . . . . . . . . . . . . . . . . . 42
3.16 IV characteristics of the 59nm LPPPT device. . . . . . . . . . . . . . . . . . . . . . . . . 42
3.17 IV characteristics of the 32nm LPPPT device. . . . . . . . . . . . . . . . . . . . . . . . . 42
xiii
3.18 V
oc
and dark current versus film thickness in LPPPT devices. . . . . . . . . . . . . . . . . 43
3.19 R
s
and R
sh
(dark) versus film thickness in LPPPT devices. . . . . . . . . . . . . . . . . . 43
3.20 R
s
and R
sh
(dark and light) versus film thickness in LPPPT devices. . . . . . . . . . . . . 44
3.21 Ratio R
s
/R
sh
as a function of film thickness in LPPPT . . . . . . . . . . . . . . . . . . . 44
3.22 IV characteristics of various film thicknesses in LPPPT. . . . . . . . . . . . . . . . . . . . 45
3.23 Fitted IV curves for LPPPT. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.24 Chemical structure of MEH-PPV. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.25 EQE and absorption for a thick and a thin MEH-PPV device. . . . . . . . . . . . . . . . . 46
3.26 EQE spectra for MEH-PPV thicknesses between 10 and 250nm . . . . . . . . . . . . . . . 47
3.27 EQE of MEH-PPV versus film thicknesses. . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.28 IV characteristics of a 250nm thick MEH PPV cell. . . . . . . . . . . . . . . . . . . . . . . 47
3.29 IV characteristics of a 210nm thick MEH PPV cell. . . . . . . . . . . . . . . . . . . . . . . 47
3.30 IV characteristics of a 54nm thick MEH-PPV cell. . . . . . . . . . . . . . . . . . . . . . . 48
3.31 IV characteristics of a 38nm thick MEH-PPV cell. . . . . . . . . . . . . . . . . . . . . . . 48
3.32 IV characteristics of a 35nm thick MEH-PPV cell. . . . . . . . . . . . . . . . . . . . . . . 48
3.33 IV characteristics of a 10nm thick MEH-PPV cell. . . . . . . . . . . . . . . . . . . . . . . 48
3.34 IV characteristics for varying film thicknesses in MEH-PPV. . . . . . . . . . . . . . . . . . 49
3.35 V
oc
and I
d
in MEH-PPV versus film thickness. . . . . . . . . . . . . . . . . . . . . . . . . 49
3.36 R
sh
and R
s
versus film thickness in MEH-PPV. . . . . . . . . . . . . . . . . . . . . . . . . 50
3.37 R
sh
and R
s
versus film thickness in MEH-PPV. . . . . . . . . . . . . . . . . . . . . . . . . 50
3.38 R
sh
and R
s
versus film thickness in MEH-PPV. . . . . . . . . . . . . . . . . . . . . . . . . 50
3.39 Fitted IV curves for MEH-PPV. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.40 Chemical structure of PTV. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.41 EQE and absorption spectrum of a 90nm thick PTV device. . . . . . . . . . . . . . . . . . 53
3.42 EQE of a 90nm thick PTV device (semi logarithmic plot) . . . . . . . . . . . . . . . . . . 53
3.43 IV characteristics of the PTV device. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.44 IV characteristics of the PTV cell (double logarithmic). . . . . . . . . . . . . . . . . . . . 54
3.45 Chemical structure of PIF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.46 EQE and absorption spectrum of a PIF device. . . . . . . . . . . . . . . . . . . . . . . . . 56
3.47 EQE and absorption of PIF device (semi logarithmic plot) . . . . . . . . . . . . . . . . . . 56
3.48 IV characteristics of the PIF device. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.49 IV characteristics of the PIF device (double logarithmic plot). . . . . . . . . . . . . . . . . 57
3.50 Chemical structure of THPF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.51 EQE and absorption spectrum of the THPF device. . . . . . . . . . . . . . . . . . . . . . 58
3.52 EQE and absorption spectrum of the THPF device(semi logarithmic plot). . . . . . . . . 59
3.53 IV characteristics of the THPF cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.54 Chemical structure of Per. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.55 EQE and absorption for the Per device. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.56 EQE and absorption of the perylene device (semi logarithmic plot). . . . . . . . . . . . . . 61
3.57 IV characteristics of the Per device. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.58 Chemical structure of Ter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.59 EQE and absorption of the Ter-device (semi logarithmic plot). . . . . . . . . . . . . . . . 63
3.60 IV characteristics of the Ter device. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.1 Device architecture of a double layer solar cell. . . . . . . . . . . . . . . . . . . . . . . . . 67
4.2 EQE and OD of the heated HPc/Per device . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.3 HPc/Per3 double layer - architecture and chemical structures . . . . . . . . . . . . . . . . 70
4.4 EQE and OD of the pristine and heat treated ITO/HPc/Per3/Al device . . . . . . . . . . 72
4.5 IV-characteristics of the pristine and heated HPc/Per double layer device. . . . . . . . . . 72
4.6 Chemical structure of CuPc and Per . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.7 Absorption and EQE of the CuPc/Per double layer device . . . . . . . . . . . . . . . . . . 73
4.8 IV Characteristics of the CuPc/Per double layer device . . . . . . . . . . . . . . . . . . . 74
5.1 Solar cell architecture comprising a blend of two organic semiconductors . . . . . . . . . . 75
xiv
5.2 Chemical structure of HPc and Per . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.3 PL efficiency versus blend ratio (HPc+Per) . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.4 AFM phase contrast images of a HPc+Per blend . . . . . . . . . . . . . . . . . . . . . . . 80
5.5 Absorption and EQE of the HPc+Per blend device . . . . . . . . . . . . . . . . . . . . . . 80
5.6 EQE of the HPc+Per blend device and the single components . . . . . . . . . . . . . . . . 81
5.7 IV characteristics of the HPc+Per blend device . . . . . . . . . . . . . . . . . . . . . . . . 81
5.8 Chemical structure of HPc and PIF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.9 Absorption and EQE of the HPc+PIF blend device . . . . . . . . . . . . . . . . . . . . . . 83
5.10 Absorption and EQE of the HPc+PIF blend device . . . . . . . . . . . . . . . . . . . . . . 83
5.11 IV characteristics of the HPC+PIF blend device . . . . . . . . . . . . . . . . . . . . . . . 84
5.12 Chemical structure of Per and PTV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.13 EQE and absorption of the PTV+Per blend device . . . . . . . . . . . . . . . . . . . . . . 85
5.14 EQE of the PTV/PER blend device (semi logarithmic plot) . . . . . . . . . . . . . . . . . 85
5.15 IV characteristics of the PTV+Per blend device . . . . . . . . . . . . . . . . . . . . . . . . 86
5.16 Chemical structure of Ter and PTV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.17 EQE and absorption of the PTV+Ter blend device . . . . . . . . . . . . . . . . . . . . . . 87
5.18 Absorption versus energy of the PTV+Ter device (vibronic modes) . . . . . . . . . . . . . 87
5.19 EQE of the PTV+Ter blend device (semi logarithmic plot) . . . . . . . . . . . . . . . . . 88
5.20 IV characteristics of the PTV+Ter blend device . . . . . . . . . . . . . . . . . . . . . . . . 89
5.21 Chemical structure of PTV and THPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.22 Absorption and EQE of the PTV+THPF blend device . . . . . . . . . . . . . . . . . . . . 90
5.23 EQE of the PTV+THPF blend device (semi logarithmic plot) . . . . . . . . . . . . . . . . 90
5.24 IV characteristics of the PTV+THPF blend device . . . . . . . . . . . . . . . . . . . . . . 91
5.25 Chemical structure of PTV and PIF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.26 Absorption and EQE of the PTV+PIF blend device . . . . . . . . . . . . . . . . . . . . . 92
5.27 EQE of the PTV+PIF blend device (semi logarithmic plot) . . . . . . . . . . . . . . . . . 92
5.28 IV characteristics of the PTV+PIF blend device . . . . . . . . . . . . . . . . . . . . . . . 93
5.29 IV characteristic (dark) of the PTV+PIF blend device (double logarithmic plot). . . . . . 93
6.1 The lamination technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.2 PL-quenching of MCP+POPT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.3 Absorption spectra of MCP and POPT . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6.4 EQE and absorption of the laminated polymer device . . . . . . . . . . . . . . . . . . . . 98
6.5 AFM image of cross-section of the laminated polymer device. . . . . . . . . . . . . . . . . 99
6.6 AFM image of a MCP+5%POPT film. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
6.7 AFM image of a POPT+5%MCP film. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
6.8 IV characteristics of the laminated ITO/POPT/MCP+5%/Al device . . . . . . . . . . . . 100
6.9 IV characteristics of the laminated Au/PEDOT/POPT/MCP+5%POPT/Ca device . . . 100
6.10 V
oc
and I
sc
versus light intensity of the laminated PEDOT/Ca polymer device . . . . . . 101
6.11 PL-quenching HPc+Per . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
6.12 EQE and absorption spectrum of the laminated dye device . . . . . . . . . . . . . . . . . 103
6.13 EQE and absorption spectrum of the laminated dye device . . . . . . . . . . . . . . . . . 103
7.1 Schematic illustration of the solid, liquid crystalline and liquid phase of discotic molecules. 105
7.2 Illustration of the different discotic phases. . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
7.3 Chemical structure of some liquid crystalline phthalocyanines. . . . . . . . . . . . . . . . . 106
7.4 DSC trace of HPc. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.5 Image of digitate star texture and birefringent defects. . . . . . . . . . . . . . . . . . . . . 108
7.6 Sketch of a bend mode as origin for birefringent defects in the mesophase of discotic molecules108
7.7 Image of a focal conic texture under the polarisation microscope. . . . . . . . . . . . . . . 109
7.8 TGA of a series of liquid crystalline phthalocyanines. . . . . . . . . . . . . . . . . . . . . . 109
7.9 TGA of CuPcSC12 under nitrogen and air. . . . . . . . . . . . . . . . . . . . . . . . . . . 110
7.10 Absorption and photoluminescence of HPc. . . . . . . . . . . . . . . . . . . . . . . . . . . 110
7.11 EQE of the thin HPc device, illumination through ITO. . . . . . . . . . . . . . . . . . . . 110
xv
7.12 EQE of the thick HPc cell, illuminated through AL. . . . . . . . . . . . . . . . . . . . . . 111
7.13 EQE of thin HPc device, illumination through Al. . . . . . . . . . . . . . . . . . . . . . . 111
7.14 EQE of the thick HPc device in vacuum and air illuminated through ITO. . . . . . . . . . 112
7.15 IV characteristics of the thin HPc device. . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
7.16 IV characteristics of the thick HPc device. . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
8.1 Definition of Air Mass. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
8.2 Global and direct solar spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
8.3 Energy distribution in AM1.5d spectrum. . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
8.4 Emission of a QTH-lamp and filter transmission spectra. . . . . . . . . . . . . . . . . . . . 118
8.5 Measured and modeled AM1.5d spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
8.6 Transmission of the required, modeled and measured AM1.5d filter. . . . . . . . . . . . . 119
8.7 Measured, modeled and standard AM1.5d spectrum . . . . . . . . . . . . . . . . . . . . . 119
8.11 Solar simulator setup. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
8.8 Deviation of actual filter absorption from the catalogue. . . . . . . . . . . . . . . . . . . . 120
8.9 AM1.5d spectrum and filter calculated from measured filter curves . . . . . . . . . . . . . 120
8.10 Calculated spectrum for new filter combination and AM1.5d . . . . . . . . . . . . . . . . . 121
8.12 Intensity dependence of V
oc
and I
sc
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
10.1 Electrode pattern of the substrate with 8 pixels. . . . . . . . . . . . . . . . . . . . . . . . 138
10.2 Electrode pattern of the substrate with 8 pixels. . . . . . . . . . . . . . . . . . . . . . . . 138
10.3 Typical light intensities versus wavelength in EQE/IV measurements. . . . . . . . . . . . 139
10.4 EQE spectrum and spectral response of the silicon reference diode. . . . . . . . . . . . . . 139
xvi
List of Tables
1.1 List of synonyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 List of abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1 Workfunction of the elements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 Workfunction of molecular electrode materials . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.1 Survey of the absorption properties of various organic semiconductors . . . . . . . . . . . 32
3.2 Survey of important performance parameters of various organic single layer solar cells . . 34
3.3 Survey of solar cell parameters obtained from the IV curves . . . . . . . . . . . . . . . . . 40
3.4 Survey of various solar cell parameters obtained from the IV characteristics . . . . . . . . 49
3.5 Survey of photovoltaic parameters of the investigated organic semiconductors . . . . . . . 65
4.1 Survey of performance parameters of various organic double layer solar cells . . . . . . . . 69
5.1 Survey of performance parameters of various solar cells using organic semiconductor blends 77
6.1 Survey of performance parameters of laminated solar cells. . . . . . . . . . . . . . . . . . 96
7.1 Phase transition temperatures of a series of liquid crystalline phthalocyanine derivatives . 107
7.2 Survey of electron and hole mobilities of important organic and inorganic semiconductors 112
9.1 Overall assessment of the different solar cell architectures incl. pigments . . . . . . . . . . 123
9.2 Overall assessment of the different solar cell architectures for soluble materials only . . . . 123
9.3 Summary of important performance parameters of various organic solar cells including all
types of architectures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
xvii
Chapter 1
Motivation - Outline
1.1 Motivation
1.1.1 The Combustion of Fossil Fuels
The limited supply of todays main energy sources
(oil, coal, uranium) will force us - sooner or later -
to replace most of the currently used power plants
with renewable energy sources. According to recent
predictions [123, 36]
1
, the inevitable permanent de-
cline in the global oil production rate is expected to
start within the next 10-20 years.
Worldwide, oil prices will then rise considerably
favouring the introduction of various renewable en-
ergy sources such as the direct conversion of solar
energy (solar cells), but also others like for example,
hydroelectric- and wind-power systems.
Figure 1.1: The atmospheric concentration of
greenhouse gases like CO
2
have grown significantly
since pre-industrial times. This can be largely at-
tributed to human activities, mostly fossil-fuel use
[136]. Dashed lines are possible (optimistic) future
scenarios [136, 10].
1
As a service to the reader, references are listed in alpha-
betical order in Chapter 9. Hence, they do not appear in
numerical order in the text.
However, the combustion of fossil fuels in the past
has already harmful effects on the delicate balance
of nature on our planet. Today, about 20· 10
12
kg of
carbon dioxide are put into the atmosphere every
year, mainly by burning fossil fuel [17, 136, 252].
Today’s plants are unable to absorb this huge
amount of extra CO
2
. As a result the CO
2
con-
centration in the atmosphere continues to mount
(Fig.1.1) adding considerably to the greenhouse ef-
fect which will increase the global mean surface
temperature - depending on future emission scenar-
ios and the actual climate sensitivity - by another
0.6-7.0
o
C by the year 2100 [136].
Global mean surface temperature has increased
by 0.3-0.6
o
C since the late 19th century and the
global sea level
2
has risen by 10-25cm, most likely
due to human activities [136].
The consequences of this temperature change
have already increased the frequency and severity of
natural disasters [252] and are likely to have more
devastating effects for humans and other life forms
in all parts of Earth within the next decades.
1.1.2 Renewable Energy Sources
Fortunately, we have renewable energy sources
which neither run out nor have any significant
harmful effects on our environment. Naturally,
power plants that use wind, the potential, wave or
tidal energy of water, the heat from the Earth’s
crust or direct solar radiation rely on the local
supply of their primary energy source. Thus, the
amount of power they can supply over a longer
period
3
often depends on geographical and weather
conditions.
It is important to recognise that the installation
of these power systems should always be preferred
as long as they supply more energy throughout their
2
Projected sea level rise from the present to the year 2100
is 15-95cm [136].
3
Using the existing energy storage technology.
1
2 CHAPTER 1. MOTIVATION - OUTLINE
lifetime than they have consumed during their fab-
rication, installation and maintenance. Although
this criterion is fulfilled for all of them their prices
are not yet low enough.
Unless we can soon develop low cost technolo-
gies for renewable energy sources we have to hope
that the world’s governments start to consider the
“costs” of environmental hazards of the majority
of existing power sources in the planning of future
power plants.
Inorganic Solar Cells
At present, solar cells comprising an inorganic semi-
conductor such as mono- and multi-crystalline sili-
con have found markets for small scale devices such
as solar panels on roofs, pocket calculators and wa-
ter pumps. These conventional solar cells can har-
vest up to as much as 24% [237] of the incoming
solar energy which is already close to the theoret-
ically predicted upper limit of 30% [230]
4
. This
illustrates that technologies which allow low fabri-
cation costs - rather than somewhat higher conver-
sion efficiencies - are now desired. One approach
here would be to reduce the amount of silicon by
using thinner films on (cheap) glass substrates.
Today, the production of these solar cells still
requires many energy intensive processes at high
temperatures (400-1400
o
C) and high vacuum con-
ditions with numerous lithographic steps leading to
relatively high manufacturing costs [162].
Organic Solar Cells
Considerably less effort and production energy is
necessary if organic semiconductors are used be-
cause of simpler processing at much lower temper-
atures (20-200
o
C) than the above mentioned inor-
ganic cells.
For example, electro-chemical solar cells using ti-
tanium dioxide in conjunction with an organic dye
and a liquid electrolyte [182] already exceeded 6%
power conversion efficiencies [237] and are about to
enter the commercial market thanks to their rela-
tively low production costs.
Another interesting alternative to inorganic cells
is given by the semiconducting polymers, which
combine the opto-electronic properties of conven-
tional semiconductors with the excellent mechanical
and processing properties of polymeric i.e. ”plastic”
materials. These can be processed from solution at
4
Note that using the more expensive GaAs even higher
power conversion efficiencies (AM1.5) have been confirmed
[237]: 25% (monocrystalline GaAs)and 30% (GaInP/GaAs)
room-temperature onto e.g. flexible substrates us-
ing simple and therefore cheaper deposition meth-
ods like spin or blade coating.
Since the discovery of electro-luminescence in
conjugated polymers [34], this class of materi-
als has been used to build efficient light emitting
diodes [248, 274, 100], field effect transistors [232],
optically-pumped LASERs [249, 250] and photovol-
taic diodes [112, 278, 213, 94].
The advantage of polymeric photovoltaic cells
when compared to electro-chemical cells is predom-
inantly the absence of a liquid electrolyte, which
generates problems with sealing against air, but
also the prospect of even cheaper production us-
ing large area devices and the use of flexible sub-
strates. Possible applications may range from small
disposable solar cells to power smart plastic (credit,
debit, phone or other) cards which can display for
example, the remaining amount, to photo-detectors
in large area scanners or medical imaging and solar
power applications on uneven surfaces.
1.2 Outline
Chapter 2 is aimed to introduce researchers who
are planning to work on organic solar cells into this
very specialised but also interdisciplinary field. The
more general properties of organic semiconductors
can be found in many excellent text books and other
references and are only summarized briefly in the
first part.
The second part gives a more comprehensive in-
sight into the important characteristic solar cell pa-
rameters and links between them. In fact, this
Chapter contains a unique compilation and sum-
mary of “organic solar cell relevant knowledge” that
is consistent with the experience, understanding
and view of the author.
However, since a full theoretical understanding
of organic solar cells is still not possible we have
tried to improve understanding of device physics
by drawing analogies to inorganic cells using the
equivalent circuit diagram and the “traditional” in-
terpretation of current voltage characteristics.
The subsequent Chapters deal with the four
known device architectures: the single layer (Chap-
ter 3), double layer (Chapter 4), blend (Chapter 5)
and the laminated device (Chapter 6).
Each of them begins with a survey of characteris-
tic parameters of already reported devices - includ-
ing the results of this thesis - pointing out specific
advantages and encountered problems.
1.3. ABBREVIATIONS AND SYNONYMS 3
Chapter 7 concerns single layer devices compris-
ing a liquid crystalline semiconductor. The out-
standing properties of discotic liquid crystals jus-
tify the discussion of this device in a Chapter sep-
arated from the single layer device Chapter. It
also comprises a survey of the interesting trans-
port (charge carrier mobility) properties of liquid
crystalline semiconductors as well as the mesogenic
characterization of a series of discotic molecules
from which one has been used to fabricate a sin-
gle layer solar cell.
How solar cell efficiencies can be determined in a
reasonable yet practical way either by setting up a
solar simulator or numerical simulation is discussed
and described in Chapter 8.
Chapter 9 concludes with a summary of the char-
acteristic parameters comprising all four solar cell
architecture, an overall assessment, some sugges-
tions for future investigations and a comprehensive
bibliography.
Details of sample preparation and measurements
as well as a list of publications by the author and a
brief CV can be found in the Appendices in Chapter
10.
The object of this thesis was the investigation of
various types of organic semiconductors (preferably
with low bandgaps) in different solar cell architec-
tures. However, the following findings may be of
particular interest for both experts and newcomers
in the field:
• We have introduced a new device architecture
that combines advantages of double layer and
blend devices and opens exciting new possibil-
ities in device design such as selective doping.
Two laminated devices are discussed in Chap-
ter 6 and published in Refs. [94, 95, 201].
• We have shown for the first time that dye/dye
interfaces can be used for photogeneration of
charges in solar cells - see Section 4.3, Section
5.2 and Section 6.3 as well as publications Refs
[200, 201].
• In Chapter 7 we have shown that dyes with liq-
uid crystalline properties can be used as active
semiconducting components in solar cells. Our
results together with the recent literature in-
dicate that heating into the liquid crystalline
phase is not necessary.
• All devices discussed in the Chapters on dou-
ble layer, blends and laminated structures
show spectral responses covering at least the
wavelength range of visible light. Two de-
vices even had a clear photo-response down to
λ >1000nm - see Section 5.7 and Section 5.3.
• The single layer device comprising PTV shows
a very strong monoton dependency of the EQE
on the wavelength so that the device can be
used as simple colour - or even - wavelength
detector covering the entire visible range - see
Section 3.4.
• We have studied effects of film thickness and
have found that those devices that have the
thinnest films (10-30nm, which is near the es-
timated exciton diffusion length) give the high-
est currents - see Section 3.2 and Section 3.3. A
method to estimate the optimal thickness has
been introduced - see Section 3.2. However, we
have also found that the shunt resistor grows
30 to 100 times faster than the series resistor
with increasing film thickness favouring thicker
films for larger fill factors.
• In Chapter 8 we describe how the standard
solar spectrum can be simulated with a rela-
tively simple setup that can be built in most
laboratories for a fraction of the cost of com-
mercial simulators. We also discuss a method
solely based on intensity dependent photocur-
rent/voltage measurements that can be used to
estimate the AM1.5 efficiency of solar cells.
1.3 Abbreviations and Syn-
onyms
Table 1.1 lists terms which are frequently used as
synonyms in literature and throughout the text.
Note that the terms ”dye” and ”pigment” are fre-
quently used as synonyms in literature - while we
differentiate clearly between these two in this thesis
- see also Section 2.2.3.
Note that, for example, the term IPCE which
is sometimes used in publications on organic solar
cells, can have two very different meanings:
1. Internal power conversion efficiency. This is
the power conversion efficiency when losses due
to reflection and transmission are considered.
Thus its value can be considerably larger than
the external power conversion efficiency.
2. Incident photon to current conversion effi-
ciency. This is often the equivalent to EQE
but could also be the IQE.
4 CHAPTER 1. MOTIVATION - OUTLINE
Table 1.1: List of expressions which are used as syn-
onyms throughout the text or relevant literature.
AM1.5 efficiency solar (power) conversion
efficiency
blend device blend layer device
conjugated polymer semiconducting polymer
external quantum- photon to current-
efficiency conversion efficiency
exciton splitting exciton dissociation
heterojunction donor/acceptor interface
HOMO/LUMO gap bandgap
HPc (pig.or dye) HPcSC12 (dye)
irradiance light intensity
n-type - electron transport -
laminated device laminated layer device
material or acceptor material
photocurrent light current
p-type - hole transport or -
material electron donor material
shunt parallel resistor
Shottky diode single layer device
spectral response photo-response
Moreover, both interpretations of IPCE could re-
fer to monochromatic light (more common) but also
to white light exposure or some other light spec-
trum.
In order to avoid these irritating ambiguities, the
big scientific community that has been research-
ing on inorganic solar cells for many decades usu-
ally quotes the EQE and if required the IQE to-
gether with AM1.5 efficiencies to allow meaningful
comparisons of solar cell performances all over the
world.
Many terms used in organic chemistry or to de-
scribe theoretical objects or concepts can grow very
long and become unreadable. Hence we have had
to use abbreviations most of which are common in
the field. These are listed in Table 1.2.
Table 1.2: List of common abbreviations used
throughout the text.
A electron acceptor
CB conduction band
CN-PPV cyano substituted PPV
CV cyclic voltametry
D electron donor
DSC differential scanning calorimetry
EA electron affinity
ECD equivalent circuit diagram
EQE external quantum efficiency
FF fillfactor
FTO fluorinated tin oxide
HOMO highest occupied molecular orbital
HPc metal free phthalocyanine
IP ionisation potential
ITO indium tin oxide
LED light emitting device
LPPP poly(para-phenylene) type
ladder polymer
LPPPT poly(phenylene-tienylene) type
ladder polymer
LUMO lowest unoccupied molecular orbital
MEH 2-methoxy, 5-(2-ethyl-hexyloxy)
MCP CN substituted MEH-PPV
MWNT multiwall nanotubes
OD optical density
PANI poly(aniline)
Pc phthalocyanine
PEDOT poly(ethylene dioxythiophene)
Per perylene diimide derivative
PIF poly(indenofluorene)
PITN poly(iso-thionaphthene)
PL photoluminescence
PR-TRMC pulse radiolysis time resolved
microwave conductivity
PPP poly(para-phenylene)
PPV poly(para-phenylene vinylene)
PT poly(thiophene)
PTV poly(tienylene vinylene)
PVK poly(vinyl carbazole)
P3HT poly (3-hexylthiophene)
SCE saturated calomel electrode
SMU source measure unit
SR spectral response
TCNQ tetra-cyano-quino-di-methane
TGA thermal gravimetry analysis
TNF trinitrofluorenone
TO tin oxide
TOF time of flight
VB valence band
Chapter 2
Solar Cells - General
2.1 Introduction
The conversion of solar light into electric power re-
quires the generation of both negative and positive
charges as well as a driving force that can push these
charges through an external electric circuit. When
connected to the external electric circuit any electri-
cal device, such as a computer screen or the motor
of a water pump, may then utilise the converted
solar energy.
In fact, a solar cell (Fig.2.1) may be seen as a
solar-light driven electron pump: The maximum
height the electrons can be ”pumped” is equal to
the highest voltage the solar cell can develop. The
maximum current is determined by the pump rate.
Suppose the ”pump” can promote 100 electrons/s
from the valence band (VB) into the conduction
band (CB), the highest possible continuous current
of electrons flowing through the external circuit is
then 100 electrons/s. If the current flow through
the external circuit is reduced by e.g. a load resistor
to - for example - 80 electrons/s, the remaining 20
electrons/s will drop back into the VB before they
can leave the device.
In real semiconductors, these leakage currents are
simply realised by the recombination of the pho-
toexcited charge carriers or using our analogy: the
”pumped” electrons drop back into their VB. These
leakage currents are typically caused by defects
or other deviations from the ideal semiconductor
structure which give rise to the appearance of al-
lowed energy levels within the bandgap.
Only if there are virtually no such defects, radia-
tive recombination, which occurs on a much larger
time scale, will remain as the only decay channel
since it does not require the existence of mid-gap
levels but can occur directly from band to band.
As a consequence, high PL - efficiency can be
regarded as evidence for the absence of the faster
and therefore more efficient non-radiative recombi-
nation channels. The charge carriers will then have
Figure 2.1: Device structure of a typical organic
solar cell. The organic film may comprise one [89]
or more [245] semiconducting layers, a blend [115]
or a combination of these [95]. Figure courtesy of
Ref. [63].
much more time to reach the device electrodes be-
fore they recombine under the emission of light.
The assumption of the absence of nonradiative re-
combination allows predictions for the upper limit
for both the power conversion efficiency of a semi-
conductor with a given bandgap and the open cir-
cuit voltage. These issues are discussed in more
detail in Section 2.4.3.
Fig.2.2 shows the conversion steps of photons into
separated charges as it takes place in an organic so-
lar cell. It also shows the associated loss mechanism
and the related electrical quantity used in the equiv-
alent circuit diagram (ECD). The latter is described
in more detail in Section 2.4.2.
In organic semiconductors, absorption of photons
leads to the creation of bound electron hole pairs
(excitons) rather than free charges. These excitons
carrying energy but no net charge may diffuse to
dissociation sites where their charges can be sepa-
rated. The separated charges then need to travel to
the respective device electrodes, holes to the anode
and electrodes to the cathode to provide voltage and
be available for injection into an external circuit.
5
6 CHAPTER 2. SOLAR CELLS - GENERAL
Figure 2.2: Survey of the specific conversion steps
and loss mechanisms in an organic solar cell. The
symbols in brackets represent the quantities that
allow for the specific loss mechanism in the ECD -
see Section 2.4.2. For effects of doping see Section
2.2.2.
In the following we comment on the individual
conversion steps with regard to the special situation
in organic solar cells:
1. Absorption of photons. In most organic device
only a small portion of the incident light is ab-
sorbed for the following reasons:
• The semiconductor bandgap is too high.
A bandgap of 1.1eV (1100nm) is required
to absorb 77% of the solar radiation on
earth (see Table 8.3 whereas the ma-
jority of semiconducting polymers have
bandgaps higher than 2.0eV (600nm) lim-
iting the possible absorption to about
30%.
• The organic layer is too thin. The typi-
cally low charge carrier and exciton mo-
bilities require layer thickness in the order
of 100nm. Fortunately the absorption co-
efficient of organic materials is generally
much higher than in e.g. Si so that only
about 100nm are necessary to absorb be-
tween 60 and 90% if a reflective back con-
tact is used.
• Reflection. Reflection losses are probably
significant but little investigated in these
materials. Systematic measurements of
photovoltaic materials are desired to pro-
vide knowledge of their impact on ab-
sorption losses. Anti-reflection coatings as
used in inorganic devices may then prove
useful once other losses such as recombi-
nation become less dominant.
2. Exciton diffusion. Ideally, all photoexcited ex-
citons should reach a dissociation site. Since
such a site may be at the other end of the semi-
conductor, their diffusion length should be at
least equal the required layer thickness (for suf-
ficient absorption) - otherwise they recombine
and photons were wasted. Exciton diffusion
ranges in polymers and pigments are typically
around 10nm [111, 113, 52, 88]. However, some
pigments like perylenes are believed to have ex-
citon diffusion lengths of several 100nm [154].
3. Charge separation. Charge separation is
known to occur at organic semiconduc-
tor/metal interfaces, impurities (e.g. oxygen)
or between materials with sufficiently different
electron affinities (EA) and ionisation poten-
tials (IA). In the latter one material can than
act as electron acceptor (A) while the other
keeps the positive charge and is referred to as
electron donor (D) - since it did actually do-
nate the electron to A. If the difference in IA
and EA is not sufficient, the exciton may just
hop onto the material with the lower bandgap
without splitting up its charges. Eventually it
will recombine without contributing charges to
the photocurrent. See also Fig.2.8.
4. Charge transport. The transport of charges is
affected by recombination during the journey
to the electrodes - particularly if the same ma-
terial serves as transport medium for both elec-
trons and holes. Also, interaction with atoms
or other charges may slow down the travel
speed and thereby limit the current - see also
space charge limited current in Section 2.4.2).
5. Charge collection. In order to enter an elec-
trode material with a relatively low workfunc-
tion (e.g. Al, Ca) the charges often have to
overcome the potential barrier of a thin oxide
layer. In addition, the metal may have formed
a blocking contact with the semiconductor so
that they can not immediately reach the metal.
Semiconductor/metal interfaces are discussed
in more detail in Section 2.2.6.
2.1. INTRODUCTION 7
We note that both exciton and charge transport
in organic materials usually require hopping from
molecule to molecule. Thus, close packing of the
molecules is assumed to decrease the width of the
intermolecular barriers and a flat molecular struc-
ture should generally lead to better transport prop-
erties than bulky 3 dimensional molecules
1
. We
note that dense packing also favours a higher ab-
sorption coefficient.
In order to meet these specific requirement for
efficient photon to charge conversion different de-
vice architectures have been developed in the past.
Fig.2.3 shows device designs used for cells compris-
ing two components, an electron donor (D) and an
electron acceptor material (A). Charge separation
occurs at the interface between those two. Ide-
ally the D- material should only be in contact with
the electrode material with the higher workfunction
(typically ITO) and the A-material with the lower
workfunction electrode (typically Al)
2
.
In the following we give a brief summary includ-
ing the strengths and weak points of the four dif-
ferent architectures known today:
1. Single Layer Cell. Single layer structures con-
sist of only one semiconductor material and are
often referred to as Schottky type devices or
Schottky diodes since charge separation occurs
at the rectifying (Schottky) junction with one
electrode. The other electrode interface is sup-
posed to be of ohmic nature. The structure is
simple but absorption covering the entire visi-
ble range is rare using a single type of molecule.
The photoactive region is often very thin and
since both positive and negative photoexcited
charges travel through the same material re-
combination losses are generally high.
2. Blend cell. The strong point of this type is
the large interface area if the molecular mixing
occurs on a scale that allows good contact be-
tween alike molecules (charge percolation) and
most excitons to reach the D/A interface. This
can usually only be partly achieved so the de-
fects of the network structure - particularly the
connectivity with the correct electrode - repre-
sents a technological challenge.
1
PL is often quenched considerably if close packing oc-
curs indicating the presence of faster recombination (loss)
channels. The reason why many aggregated materials can
still be used to give excellent devices is not yet clear. The
gain of improved charge transport may compensate the extra
recombination channel here.
2
See Section 2.2.6 for more information on electrode
materials.
Figure 2.3: Three different types of D/A architec-
tures for organic solar cells. The blend structure
can be obtained by spincoating using a D/A mix-
ture whereas the double layer requires two separate
deposition processes. The laminated device can be
seen as a hybrid between these two.
3. Double layer cell. This structure benefits from
the separated charge transport layers that en-
sure connectivity with the correct electrode
and give the separated charge carriers only lit-
tle chance to recombine with its counterpart.
The drawback is the small interface that al-
lows only excitons of a thin layer to reach it
and get dissociated.
4. Laminated device. This relatively recent type
represents the successful attempt to unify
the advantages of the two structures above.
Charge separation occurs in the blend layer in
the middle that is obtained after laminating the
two separate layers together and charge trans-
port can only occur via the correct transport
layer. This structure also features the useful
options to treat each layer separately (e.g. dop-
ing, physical/chemical conversion) before form-
ing the blend layer and instant encapsulation
between the two substrates. The drawback is
that certain mechanical properties of the or-
ganic semiconductors are required (low glass
transition temperature) to form the intermixed
layer.
In Section 2.2 we discuss the special general prop-
erties associated with organic semiconducting ma-
8 CHAPTER 2. SOLAR CELLS - GENERAL
Figure 2.4: Simple model that relates the IV charac-
teristics qualitatively to energy level diagrams. For
clarity it is assumed that bands are tilted rather
than bent and exciton dissociation occurs through-
out the entire device. Tilted bands represent homo-
geneous electric fields that drive the charge carriers
to the respective electrodes.
terials whereas in Section 2.4 we focus on the pho-
tovoltaic properties. Since solar energy conversion
using solar cells is in principle the same for both or-
ganic and inorganic semiconductors ”general tools”
like the definition of solar cell parameters (FF, V
oc
,
EQE), the concept of an equivalent circuit diagram
(ECD) or the deduction of resistor parameters from
the IV characteristics are the same.
2.2 General Properties of Or-
ganic Semiconductors
Due to e.g. the strong electron-phonon interaction
in organic materials the photo-physics in this class
of materials is often different compared to the inor-
ganic semiconductors and not yet fully understood.
One of the main differences is that photo-
excitation in these materials does not automatically
lead to the generation of free charge carriers, but to
bound electron-hole pairs (excitons) with a binding
energy of about 0.4eV [12, 56, 168].
These excitons need to be split up (or dissoci-
ate) before the charges can be transported through
the film and collected at the electrodes. For ex-
ample, exciton dissociation can occur at a rectify-
ing interface (Schottky contact) in single layer de-
vices or at the interface between an electron-donor
and an electron-acceptor semiconductor material.
The larger this interface area the more excitons can
reach it and dissociate.
In addition, the small diffusion range of the exci-
tons (typically about 10nm [111, 113, 88, 52]) com-
pared to the film thickness necessary to absorb the
major portion of light (typically >100nm) has made
it difficult to reach practically interesting conversion
efficiencies in organic solar cells.
2.2.1 Polarons and Polaron Excitons
The origin of the term polaron for charged quasi-
particles can be found in classical polar crystals
where the associated charge e.g. the electron re-
pulses adjacent electrons while attracting the nuclei
resulting in a polarisation of the lattice in its closer
vicinity.
Moreover, this accompanying cloud of polarisa-
tion causes an increasing effective mass decreasing
the mobility of these quasi-particles. Due to the
dimerisation
3
of the bonds, conjugated systems ex-
hibit neither pure covalent nor pure polar bonds.
In fact, they represent a separate class with re-
gard to their bond nature exhibiting huge electron-
lattice coupling
4
compared to the inorganic solids.
The strong electron-lattice coupling is responsible
for the existence of the quasi-particles such as po-
larons and polaron excitons in conjugated systems.
These quasi particles can be identified by the
additional energy levels [70] associated with them
which appear within the semiconductor bandgap
5
-
see Fig.2.5 - using sensitive detection methods like
photo-induced absorption spectroscopy [163, 197,
198].
Polarons in conjugated systems affect not merely
the polarisation in their vicinity - they even change
the nature of bonds from σ to π and vice versa via
excitation and while traveling. This usually leads
to a more rigid structure in the excited state the so
called quinoid configuration.
Note that photogeneration of charged excita-
tions usually requires separation of charges between
3
The alteration of single and double bonds does not allow
determination of which bond is double and which is single
bond in degenerate semiconducting systems such as most
semiconductors. Thus, the π electron of the double bond
can be seen as delocalised.
4
Excitation of the π electron used for the double bond
has a strong effect on the ”lattice” structure.
5
upon photoexcitation or chemical doping (upon blending
or intercalation with guest material (dopant)).
2.2. GENERAL PROPERTIES OF ORGANIC SEMICONDUCTORS 9
Figure 2.5: Energy levels of the positive and neg-
ative polarons as well as the polaron (singlet) ex-
citon. Arrows with solid lines represent electrons
with their spin orientation, arrows with dashed lines
show possible intra gap transitions that can be used
to detect and identify the excited species.
the chains, the most effective generation is photo-
excitation with a pump beam that is polarised per-
pendicular to the polymer backbone [76].
In the following we give a brief overview about
differences between excitons in the much better un-
derstood inorganic systems and the organic ones:
• Excitons in inorganic solar cells. The binding
energy E
b
of inorganic excitons is estimated to
be around 16meV which means that these exci-
ton become important predominantly at lower
temperatures (where kT becomes small com-
pared to E
b
)[98]. Unlike its constituents par-
ticles, an inorganic exciton is a Boson rather
than a Fermion, with occupation statistics de-
termined by the Bose-Einstein equation.
These excitons can reduce their energy further
if they bind with impurities or defects. Since
they are Bosons, all of them can occupy the
lowest lying state (at once) giving sharp peaks
in the light emitted at low temperatures due to
radiative recombination of the constituent elec-
tron and hole. At high densities and low tem-
peratures, the free exciton gas can condense to
give an electron-hole liquid phase with inter-
esting properties [270].
• Excitons in organic solar cells. The binding en-
ergy E
b
of excitons in organic semiconductors -
in particular for conjugated polymers like PPV
and its derivatives - has been subject to in-
tense debates over the past decade. E
b
values
ranging from very small [158] over intermedi-
ate values (around 0.4eV) [56, 168, 12] to very
high (up to 0.95eV) [49] have been proposed.
Within the framework of this thesis we assume
that binding energy is not larger than about
0.4eV. As a consequence, the energy levels of
the exciton are separated from the CB and VB
by less than about 0.2eV which have been ne-
glected in all band energy diagrams in this the-
sis since the bandgaps of the materials used are
between 1 and 2eV. However, a clear offset be-
tween the HOMO and LUMO levels of the D/A
material is still required for the exciton to dis-
sociate at room temperature.
In the following we give a brief description of dif-
ferent types of excitons
6
:
1. Frenkel exciton: The electron hole pair is con-
fined to not more than one molecular unit
[75, 195].
2. Mott-Wannier exciton: The distance between
electron and hole is much greater than the
spacing between unit cell (here molecular
units) [175, 265]. Their energy levels can be
described as hydrogen-like bound states, cal-
culated with a reduced effective mass.
3. Charge transfer exciton: The exciton extends
over only a few adjacent molecular units [138]
4. Inter-chain excitons: This term is used for
polymeric semiconductors to indicate that the
constituent charges are located on different
polymer chains. It can be regarded as a charge
transfer exciton.
5. Intrachain excitons: This term also refers to
polymeric semiconductors to indicate that the
constituent charges are located on the same
polymer chain. It is believed that intra-
chain excitons represent the main species that
is formed after photoexcitation in conjugated
polymers [78, 79, 54].
We note that there is another related excited
species called “excimer” that can play an impor-
tant role in organic semiconductors. An excimer
can be understood as a dimer that is only stable in
the excited state [205, 138]. Hence, it is only weakly
absorbing and shows little vibronic structure in its
emission spectrum. It can be formed by interaction
of an excited molecule with a molecule in its ground
state. Both singlet and triplet excimers have been
observed in organic molecules such as naphthalenes
or perylenes [205, 104]. An exciplex is an excimer
consisting of two different molecules.
6
Inter- and intrachain excitons may also be seen as sub
groups of the three main types.
10 CHAPTER 2. SOLAR CELLS - GENERAL
2.2.2 P- and n- type Organic Semi-
conductors
Fermi-level
The energetic position of the Fermi-level in semi-
conductors is important for two reasons:
1. Together with the workfunction of the metal
the Fermi level determines whether a blocking
or ohmic contact is formed at the semiconduc-
tor/metal interface.
2. The relative position of the Fermi levels is a
measure for the type of conductivity - whether
the semiconductor conducts preferably holes in
the VB (p-type) or electrons in the CB (n-type)
The energetic position of the Fermi level in a
semiconductor represents the balance between the
concentration of holes and electrons that occupy
allowed energy levels under equilibrium condition
(dark, no applied voltage). If the Fermi-level is
closer to the CB the material is called n-type con-
ductor since more electrons are available for con-
duction than holes - otherwise it is called p-type
conducting.
Thus, the position of the Fermi level can be writ-
ten as a function of both the effective density of
states in the conduction band N
C
and valence band
N
V
as well as the concentration of donors N
D
and
acceptors N
A
:
For n-type semiconductors this gives [162]:
E
F
= E
C
−kT · ln(N
C
/N
D
) (2.1)
whereas for p-type semiconductors the relation is:
E
F
= E
V
−kT · ln(N
V
/N
A
) (2.2)
E
V
and E
C
stand for the top edge of the
VB and the bottom edge of the conduction band
respectively
7
.
When voltage is applied and/or the semiconduc-
tor is illuminated, the concept of a Fermi level can
no longer be applied. Upon e.g. illumination the in-
creased concentration of electrons in the CB would
shift E
f
up while the higher concentration of holes
in the VB would require to shift E
f
down - at the
same time. As a consequence, two separate Fermi
levels - the so called quasi Fermi-levels
8
- are then
introduced to describe the situation under this non
equilibrium condition [98, 162]:
7
E
V
may be seen as the HOMO and E
C
as the LUMO
level in a quasi molecular model.
8
These are also known as Imrefs (Fermi read backwards.)
E
FN
= E
C
+kT · ln(n/N
C
) (2.3)
and
E
FP
= E
V
+kT · ln(N
V
/p) (2.4)
with n and p denoting the concentration of elec-
trons in the CB and holes in the VB. E
FN
and E
FP
are the quasi Fermi-levels that are associated with
the balance of electrons and holes - both exist at
the same time in one semiconductor.
Doping
In inorganic semiconductors doping occurs via the
introduction of dopant atoms that provide addi-
tional free charge carriers at room temperature so
that the extra charges increase the conductivity for
this type of charge carrier. As discussed above, the
Fermi level would be moved towards the CB if extra
electrons are introduced or towards the VB if the
concentration of mobile holes has been increased.
In organic semiconductors “doping” is usually
achieved by the introduction of foreign molecules
rather than atoms
9
. Another possibility to achieve
doping effects in conjugated systems is electro-
chemical oxidation or reduction which can be
achieved and investigated by e.g. cyclic voltame-
try [125, 126, 127, 233].
Moreover, changes of the concentration of mobile
charges are often achieved by trapping of e.g. elec-
trons in the CB so that the concentration of mobile
holes in the VB (relative to the mobile electrons)
increases. That way the Fermi level would move
towards the VB and make the material a ‘doped”
p-type conductor i.e. a better hole than electron
transport material. Since there is no net increase
of mobile charge carriers the conductivity cannot
increase (unless the mobility of charge carriers is
affected).
However, in practice we often find increased con-
ductivity (particularly in light) upon exposure to
e.g. oxygen [268, 43]. This effect may be explained
by the generation of charge carriers due to exciton
dissociation - see discussion below.
Examples for dopants
10
of p-type materials are:
Cl
2
, NO
2
, I
2
, Br
2
, organic molecules like o-chloranil,
and 2,4,7-trinitrofluorenone, H
2
O, O
2
. Other p-
type dopants are TCNQ or also high electron affin-
ity semiconductors like C60 or CN-PPV or perylene
9
Atoms are not used because they are too small. They
would diffuse out of the bulk and eventually neutralize with
other atomic dopants.
10
In decreasing order of effectiveness.
2.2. GENERAL PROPERTIES OF ORGANIC SEMICONDUCTORS 11
diimides. Dopants for n-type materials
11
are e.g. al-
kali metals, phenothiazines and also semiconductors
with low ionisation potentials.
A different form of doping resulting in a truly in-
creased concentration of mobile charge carriers on
the semiconducting molecule and therefore consid-
erably increased conductivity (in both the dark and
under illumination) can be achieved by complexa-
tion with foreign molecules. Such doping molecules
are preferably large (polymers) so that they can-
not diffuse and penetrate into another material (e.g.
in a p/n junction device) where they can possibly
neutralize with another (different) dopant to form a
salt. Examples are doping of PANI with camphor-
sulfonic acid or PEDOT with polystyrene sulphonic
acid [38, 6].
We note that compared to doped inorganic semi-
conductors, organic semiconductors can have very
few mobile charges available - particularly in the
dark - so that extra mobile charges due to illumi-
nation or doping can make a big difference. In fact,
they can sometimes be described better using mod-
els for photo-conducting insulators [138].
More information on doping of molecular semi-
conductors can be found in Refs. [49, 43, 164, 225,
268, 38].
2.2.3 Structural Properties of Or-
ganic Semiconductors
We have seen that the optical and electronic prop-
erties of organic semiconductors can be tailored by
attaching certain functionalities to the conjugated
system. These properties are crucial for the opera-
tion of devices such as solar cells or LEDs - however,
the “mechanical” i.e. processing properties are also
very important.
For example the fabrication of devices via ther-
mal sublimation requires high vacuum conditions
and high temperatures whereas processing from so-
lution can be done at room temperature at ambient
pressure. In addition the fabrication of certain de-
vice architectures such as blends or laminated de-
vices requires much more work or are not even fea-
sible if thermal sublimation techniques have to be
used. Moreover liquid crystalline systems were not
accessible for investigation if only sublimation was
available.
Fig.2.6 shows how organic semiconductors can
be divided into different categories due to their
processing and mechanical properties. We also
11
Note that H
2
, NH
3
can be used to compensate radicals or
impurities and thereby somewhat enhance the photocurrent.
distinguish between molecules that have only a
few (oligomers) or no (monomer) repeat unit and
molecules (polymers) which have more than about
10 repeat units. Oligomers and monomers that ab-
sorb visible light are often called chromophores and
are referred to as dyes if they are clearly soluble or
pigments if they are not
12
.
Liquid crystalline materials have only recently
been investigated for the use in LEDs and organic
solar cells [151, 199, 200]. For a certain tempera-
ture range these material exhibit a phase where the
molecules move like in a liquid but are still able to
sustain or attain a certain structural order among
the molecules as in crystalline materials.
Such molecules have to have a strong asymmetry
to show structural order and are therefore either
rod-like (calamitic) or disc-like (discotic) molecules.
We will discuss these interesting molecules in Chap-
ter 7 in more detail.
Figure 2.7: The “mechanical” properties of a flat
structure like perylene diimides depend to a large
extent on the position, number and properties of
the side chains rather than the constituent atoms
of the main structure. Common solubilising side
chains (R,R
1
, R
2
) are often linear or branched alkyl
or alkoxy groups. The synthesis of the perylene dye
is reported in [60] while more information about the
other shown molecules can be found in Ref. [217]
Fig.2.7 shows that the “mechanical” properties
depend mainly on the position and number of
side chains. Side chains are usually attached to
molecules to introduce or improve solubility. They
12
Note that the terms pigment and dye are sometimes
regarded as synonyms among researchers who are not con-
cerned about solubility properties. Recently, a method has
been found to process many pigments like dyes using strong
acids [283].
12 CHAPTER 2. SOLAR CELLS - GENERAL
Figure 2.6: Organic semiconductors may be divided into three different categories according to their
mechanical i.e. processing properties: insoluble, soluble and liquid crystalline. They may be divided
further into monomers (dyes, pigments) and polymers.
are more successful in solubilising the molecule
the better they can prevent aggregation between
molecules.
Flat molecules in particular have a strong ten-
dency to stick together due to π-π interaction and
form clusters of solids in many solvents. Bulky
side chains can separate theses molecules and make
it easier for solvent molecules to surround the in-
dividual molecules i.e. dissolve them
13
. Typi-
cally, smaller molecules are better soluble and/or
have lower sublimation temperatures but larger
molecules give better films upon spincoating.
2.2.4 Photogeneration of Free
Charges
Photogeneration due to Oxygen Traps
It is widely believed that oxygen can act as trap for
electrons in the CB of organic materials and thereby
lead to a higher number of free holes in the VB. In
this sense oxygen can act as a dopant that favours
p-type conductivity as discussed above. However,
oxygen traps can also act as exciton dissociation
sites and therefore generate free charge carriers in
the presence of excitons. The proposed mechanism
is the following: After a certain period, an electron
in the oxygen trap may further decrease its energy
by decaying to the VB where it recombines with
either a free hole or the hole of an exciton. Thus 3
scenarios may be distinguished:
1. The electron stems from a not yet dissociated
exciton and it recombines with a hole that
13
In liquid crystalline materials solvent molecules are not
even necessary since the cores of the molecules are ”dis-
solved” in their own side chains.
comes from a different exciton. Both exci-
tons are dissociated - but only two charges (the
hole of the first exciton and the electron of the
other) are produced since the other two recom-
bined.
2. Only one recombination partner is associated
with an exciton. This means its dissociation
produces one free charge but still eliminates
two.
3. None of the recombination partners belongs to
an exciton. Two charges are lost, none is gen-
erated.
Note that both electrons and holes can be pro-
duced with the same probability according to the
above list. In any case, the electron in the oxygen
trap still reduces predominantly the mobility of neg-
ative charges thus sustaining its p-dopant character.
We note that basically any defect that leads to an
allowed free energy level between the middle of the
bandgap and the CB may have the same effect as
the here discussed oxygen trap - although other ef-
fects may become more important. For example,
photo-oxidised PPV is believed to contain a good
deal of carbonyl-groups which can facilitate exciton
splitting but may also decrease both electron & hole
as well as exciton transport properties [114, 215].
The Donor/Acceptor Interface
We have seen that photogeneration of charge due to
traps is not very efficient since there is always the
recombination of two charge carriers involved. Here
we describe a more efficient way to generate free
charge carriers - the formation of an electron donor
2.2. GENERAL PROPERTIES OF ORGANIC SEMICONDUCTORS 13
(D)/electron acceptor (A) interface. Although on
a first glance, the concept may be reminiscent of
the p/n junction in inorganic materials, the physical
process are often very different.
Fig.2.8a) shows that the required transfer of
charges can be realised at the interface between
two materials if one material has a higher electron
affinity (EA) whereas the other has a lower ionisa-
tion potential (IP). The one with the larger EA can
accept an electron from the CB of the other and
is therefore called electron acceptor. The material
with the lower IP can accept the hole from the VB
of the contacting semiconductor may therefore be
termed hole acceptor or electron donor since it also
donates the electron to the contacting acceptor.
We note that the offset of IP and EA needs to be
large enough so that the resulting field (the poten-
tial gradient at the junction) can overcome the exci-
ton binding energy which is typically around 0.4eV.
Otherwise charge transfer may occur but the exci-
tons do not split into their constituent charges and
recombine eventually at the D/A interface
14
.
Figure 2.8: The interface between two different or-
ganic semiconductors can either: a) facilitate charge
transfer (D/A interface), b) energy transfer (no ex-
citon splitting). What kind of interface forms de-
pends on the position of the HOMO and LUMO lev-
els and the direction in which band bending occurs.
The latter is determined by the relative position of
the Fermi levels.
Fig.2.8 b) shows that an exciton that carries its
excitation energy can be transferred from the ma-
terial with the higher bandgap to the other if both
the hole and the electron can decrease their energy
thereby. This process of excitation energy transfer
14
This is the situation that is desired for the design of light
emitting diodes (LEDs) to favour radiative recombination at
the interface between an electron and a hole transporting
material[100].
which is accompanied with some energy loss is also
known as Foerster transfer [72] and can be used to
shift the emission band of the exciton towards lower
energies in e.g. LEDs [246].
Foerster transfer also allows the design of a pos-
sibly helpful directionality in an organic solar cell
since it is an one way mechanism, carrying exci-
tons from the the shorter wavelength absorber to
the longer wavelength absorber. This can be help-
ful to concentrate delivered excitons in a spatially
small high field region [267] to enhance dissociation.
We note that placing the higher bandgap absorber
in front of the lower bandgap material to avoid ther-
malisation losses is proposed in Ref. [268].
Whether Foerster transfer or charge transfer oc-
curs with a certain combination of organic semicon-
ductor materials is often difficult to assess - particu-
larly if one or both materials do not fluoresce. How-
ever, if they do, it is helpful to plot PL versus blend
ratio. If the PL of both pure components decreases
upon blending Foerster transfer can be ruled out -
otherwise the PL of one component would enhance
the other. Although, the exciton may still recom-
bine at the D/A interface rather than split into its
constituent charges there is a good chance for the
latter.
Confirmation of exciton dissociation can be ob-
tained by comparing the EQE of a blend solar cell
with equivalent devices comprising only the pure
components. If the EQE of the blend cell is clearly
higher over the entire spectrum exciton splitting at
the D/A interface is most likely. Extra confirmation
may be obtained if the experimental and/or theo-
retical LUMO and HOMO data of the components
also suggest a sufficient offset of the band edges to
split the exciton. We note that the prediction of the
offset of the band edges actually requires the knowl-
edge of the accurate position of the Fermi levels -
since those have to equalize upon contact.
The determination of Fermi levels in organic
semiconductors can be done
15
but there are very
few data in literature [172]. For practical purposes
Fermi-level are often assumed to lie around the mid-
dle of the semiconductor bandgap. HOMO and
LUMO levels are often taken from cyclic-voltametry
(CV) measurements where the current through the
semiconductor gives a peak at the voltage where ox-
idation and reduction occurs [125, 126, 127]. The
oxidation or reduction potential are then regarded
as HOMO and LUMO levels.
We note that this method appears very useful if
15
Using e.g. the Kelvin probe method described in Ref.
[172].
14 CHAPTER 2. SOLAR CELLS - GENERAL
all components can be measured at the same setup
under comparable conditions. Otherwise care has
to be taken if absolute numbers are compared, since
the scan rate, and the assessment of the peak (it is
not clear if the peak or the onset should be taken)
can vary and affect the result considerably. CV val-
ues for films and solution may also differ.
Given that photo excitation results in the for-
mation of excitons with a binding energy consider-
ably higher than kT (26meV), photogeneration of
charge is always assisted by an external factor such
as the presence of traps, a D/A interface or a metal
which can also act as an electron or hole acceptor
or simply provide an electric field that can split the
exciton. Thus photogeneration of charge in these
organic semiconductors can be regarded as extrin-
sic as opposed to inorganic semiconductors where
separation occurs mainly together with photo exci-
tation.
2.2.5 D/A Materials
For the design of D/A solar cells it is important
to know whether a specific material combination is
likely to lead to charge transfer + exciton splitting.
One way to find out is to build blend devices and
look for PL quenching of both components and en-
hanced EQE with respect to the single components
- as discussed above. Much effort, time and money
can be saved if the chemical structure of the chosen
molecules is considered. In the following we provide
some background knowledge that allows guessing
not only the D/A properties but also some other
relevant properties for organic photovoltaic materi-
als:
1. Donor/Acceptor properties. Since a donor can
act as an acceptor for an even stronger donor
- these two terms can not really be used sep-
arately. There are certain functionalities that
are likely to make a material an electron ac-
ceptor with respect to most other materials.
Examples for functionalities that favour elec-
tron acceptor properties are: CN, CF
3
, F,
= O (keto-groups) or diimides (all perylene
derivatives in this thesis are perylene diimides).
These groups are known as electron withdraw-
ing groups if they are attached to an unsatu-
rated (=conjugated) system.
Note that their ability to enhance the elec-
tron affinity (EA) of the entire molecule is not
purely connected to the electron withdrawing
properties of these functionalities. It is essen-
tially a result of both the inductive and the
mesomeric effect.
While the inductive effect is determined by the
electronegativity of the substituents, the me-
someric effect takes into account the influence
of free electron pairs. The latter considers
that many conjugated structures can equally
be drawn with a different double/single bond
i.e. π electron arrangement. The average of
these possible π electron locations can be re-
garded as the overall mesomeric effect of a spe-
cific molecular structure. Examples for elec-
tron acceptors are CN-PPV, CF
3
substituted
PPV and perylene diimides [167, 154]. Most
organic semiconductors behave more like elec-
tron donors, for example PPV, PT, PPP and
phthalocyanines.
2. Bandgap. As a rule of thumb, the bandgap
is lower the larger the conjugated π-system is.
Examples for large π-systems are phthalocya-
nines [117, 118, 119, 154], naphthalocyanines
[171, 140, 37] and perylenes [154, 155, 267].
Sulfur in a conjugated systems also tends to
reduce E
g
. The bandgap can also be small
if a molecule consists of a donor and an elec-
tron acceptor that are connected via a con-
jugated structure to create a “push/pull sys-
tem” as in e.g. poly-methines. Such structures
can accomplish charge transfer sometimes al-
ready in the ground state (e.g. charge trans-
fer salt) or only little extra (light) energy is
required to complete it. Examples for charge
transfer salts are PVK-TNF [154] and for poly-
methines: merocyanines [204, 173, 43].
3. Solubility. Solubility is good if the planar parts
of the conjugated π-systems of the molecules
cannot get too close to each other. Otherwise
they would stick together, driven by their π−π
interaction, to form aggregates i.e. larger clus-
ters i.e. particles which do not contain solvent.
Flexible and bulky side chains or atoms that
stick out of the molecular plane can prevent
molecules from getting to close - see also dis-
cussion in Section 2.2.3.
The general rule is: Molecules dissolve in sol-
vents with similar structure. Polar solvents
like water, alcohol or iso propanol dissolve po-
lar molecules while less polar solvents (e.g.
toluene, xylene) dissolve non-polar molecules.
As a rule of thumb, it can be said that a sol-
vent containing other atoms than carbon and
hydrogen has a good chance to be more polar
- particularly if the structure is asymmetric.
2.2. GENERAL PROPERTIES OF ORGANIC SEMICONDUCTORS 15
4. Photoluminescence (PL). Practice shows that
PL is quenched if the molecules can aggregate.
Consequently, molecules in solution fluoresce
much stronger.
Materials with higher bandgaps tend to show
stronger PL
16
.
Moreover, the presence of impurities or struc-
tural defects usually leads to energy levels
within the bandgap. The energetic decay of
an excited electron can occur faster and there-
fore more efficiently if smaller energy jumps are
required.
Also conjugated systems containing certain
atoms like sulfur tend to give lower PL than
conjugated structures containing only carbon
and hydrogen.
2.2.6 Electrode/Semiconductor In-
terfaces
The theory of contacts between doped semiconduc-
tors and metals was developed in 1938 by Mott
and Schottky [176, 225]. Since then charge injec-
tion from metallic contacts into semiconductors has
been studied extensively and is explained in detail
in the semiconductor literature [220, 241, 138, 205]
to which we refer the interested reader.
However, it is not clear to what extent the devel-
oped models can be applied to organic semiconduc-
tors we only describe the basic concepts as they are
often used to understand and predict the behaviour
of organic semiconductor/ electrode interfaces.
In Fig.2.9 we have drawn the band energy dia-
gram as it is expected for the situation before con-
tact and after contact. It can be seen that - after
contact - band bending occurs at all interfaces and
that the Fermi levels and workfunctions equalize.
The workfunction of the electrodes are drawn as
equal here to show the situation in the short circuit
mode.
The depicted band bending can be qualitatively
predicted by assuming that the edges of the LUMOs
and HOMOs are “pinned” to their absolute energy
values as they were before contact. The bulk of
the semiconductor sustains the same distance to the
equalising Fermi-level.
Interestingly, both the electron affinity and ion-
isation potential of the semiconductors does not
need to change - not even where band bending
occurs - since the corresponding vacuum level fol-
lows both the band curvature and vertical shifts.
16
Systematic intentional interruption of the conjugation
has been shown to increase the PL in PPV considerably []
The direction the band bending occurs depends on
whether the W
f
of the electrode is above or below
the Fermi-level of the semiconductor.
In case the workfunctions of both electrodes are
deeper inside the bandgap than the Fermi-levels of
the materials they are contacting, band bending oc-
curs as depicted in Fig.2.9 b). Note that band bend-
ing at the D/A interface would be the other way
round (blocking) if both Fermi-levels are assumed
to be close to the middle. This may be the case and
desired for LEDs [100].
Photoexcited electrons from the CB of the elec-
tron acceptor (A) encounter a barrier before they
can reach the Al electrode. The same is true
for holes from the VB near the ITO contact.
These types of contacts are referred to as block-
ing contacts. Since these contacts can become non-
blocking upon changing the potential of the elec-
trode they are also called rectifying or Schottky con-
tacts.
Figure 2.9: Energy band diagrams of D/A devices
with the electrode interfaces. a) the situation be-
fore contact. After contact the Fermi levels (dashed
lines)and W
f
equalize and band bending occurs:
b) the formation of a blocking contact for holes
(ITO/D) and electrons (A/Al). c) the formation
of a non-blocking (ohmic) contact for holes (Au/D)
and electrons (A/Ca).
Note that Al forms a blocking contact only with
with the CB of the n-type semiconductor
17
but an
ohmic contact with the VB. The reason why this
17
p type conductivity is assumed for D and n-type for A -
see also discussion in Section 2.2.2
16 CHAPTER 2. SOLAR CELLS - GENERAL
particular interface situation is called rectifying is
because the n-type conductor is supposed to trans-
port electrons in its CB - and for electrons it is
blocking i.e. rectifying. The situation is analog for
the semiconductor/ITO interface.
Since charge separation at the D/A interface
leads to an excess of electrons in the CB of A
(n-type) and holes in the VB of D (p-type), both
charge carriers find barriers if they want to get out
of the semiconductor. Thus currents through this
device are decreased by the contact barriers.
Fig.2.9 illustrates that the situation is improved
if Au and Ca are used as contact materials. The W
f
of both electrodes is further away from the middle of
the bandgap than the semiconductors Fermi-levels.
Thus, after contact, band bending occurs the other
way round so that electrons in the excited state can
reach the Ca electrode and holes the Au electrode
without encountering a barrier.
Practice confirms that organic devices using Au
and Ca instead of ITO and Al show indeed con-
siderably higher open circuit voltages and often
also higher photocurrents and EQE values [114, 94].
However, because of the lower W
f
of Ca compared
to Al, Ca is more prone to oxidation than Al. Gold
electrodes are expensive and have to be very thin to
become transparent and are more difficult to pat-
tern via etching. For the above reasons Al and ITO
electrodes are actually more suitable for research
devices since they can be reproduced easier.
Current Through Barriers
There are two principal mechanisms for the charge
carriers to overcome barriers: Thermionic emission
and quantum mechanical tunneling (field emission).
For thermionic emission over a triangular barrier
of height Φ from a metal into a high-mobility semi-
conductor, the current density (J) is given by
J = A

T
2
e
−Φ
kT
(2.5)
where A

is the effective Richardson constant and
T the absolute temperature
If an electric field is applied across the barrier the
barrier height decreases by
∆Φ =

q
3
E

0
(2.6)
where E is the applied electric field, q the el-
ementary charge and
0
the dielectric constant of
vacuum.
In a semiconductor with a low mobility, it is nec-
essary to take into account the diffusion of carri-
ers within the barrier region back towards the con-
tact. The analysis is complicated and depends on
the type of barrier assumed. A full treatment is
given by Kao and Hwang [138].
At sufficiently low temperatures , or for large bar-
riers at high fields, emission due to quantum me-
chanical tunneling through the barrier (field emis-
sion), can become important. For a Schottky bar-
rier the current is given by
J ∝
Φ
b
+qV
Φ
e
K
Φ
3/2
(Φ+qV )
1/2
(2.7)
where V is the applied voltage and K a material
constant.
The essential characteristic of field emission is
that the current is insensitive to temperature, but
strongly dependent on the applied field (voltage).
Since the tunneling rate depends also strongly on
the width of the barrier
18
, thermal excitation can
increase the tunneling current. For a Schottky bar-
rier the current can then be written as
J = J
0
e
qV/nkT
(2.8)
where J
0
is essentially a constant for low doping
levels and n ≈ 1 at high temperatures [101].
2.3 Electrode Materials
In organic devices the workfunction of the elec-
trode materials is very important since it deter-
mines together with the LUMO/HOMO and Fermi-
level of the semiconductor whether the electrode
forms an ohmic or a blocking contact for the re-
spective charge carrier (holes in VB, electrons in
CB). Moreover a large difference in workfunction of
the electrode materials can increase the V
oc
consid-
erably.
In Table 2.1 we have listed the periodic table of
the elements with its values of W
f
- as far as it was
known in Ref. [170]. The values in Table 2.1 are
valid only for poly-crystalline materials. However,
many numbers for single crystals which depend on
its crystallographic orientation have also been re-
ported in Ref. [170].
Common electrode materials for the electron col-
lecting contact (low W
f
required) of organic solar
cells are Al, Ca, In, Ag whereas for the hole col-
lecting contact high workfunction materials like Au
are preferred. For both solar cells and LEDs one
18
This is not considered in Eq. (2.7)
2.3. ELECTRODE MATERIALS 17
Table 2.1: Workfunction of the elements. The numbers refer to the workfunction in eV of the elements
in poly-crystalline form..
IA IIA IIIB IVB VB VIB VIIB VIII IB IIB IIIA IVA VA VIA
Li Be B C N O
2.9 4.98 4.45 5.0 - -
Na Mg Al Si P S
2.75 3.66 4.28 4.85 - -
K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se
2.30 2.87 3.5 4.33 4.3 4.5 4.1 4.5 5.0 5.15 4.65 4.33 4.2 5.0 3.75 5.9
Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te
2.16 2.59 3.1 4.05 4.3 4.6 - 4.71 4.98 5.12 4.26 4.22 4.12 4.42 4.55 4.95
Cs Ba La Hf Ta W Re Os Ir Pt Au Hg Ti Pb Bi Po
2.14 2.7 3.5 3.9 4.25 4.55 4.96 4.83 5.27 5.65 5.1 4.49 3.84 4.25 4.22 -
Fr Ra Ac Th Pa U
- - - 3.4 - 3.63
Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb
2.9 - 3.2 - 2.7 2.5 3.1 3.0 - - - - -
contact has to be at least partly transparent. Semi
transparency can be obtained if the sublimed metal
e.g. Au is not much thicker than about 15-20nm
compared to about 50-100nm which are typical val-
ues for non transparent contact. We note that the
sheet resistance of such a thin layer can be consid-
erably increased compared to a 50 to 100nm thick
layer and possibly add to the series resistor of the
cell significantly. Also the optical properties change
considerably in the thickness range from 10-20nm
so that devices with slightly different thickness may
not be immediately comparable. For these reasons
so called conducting glasses are often used.
Particularly Indium Tin Oxide (ITO) which is
a degenerated semiconductor comprising a mixture
of In
2
O
3
(90%)and SnO
2
(10%) with a bandgap of
3.7eV and a Fermi-level between 4.5 and 4.9eV is
widely used. The large bandgap allows no absorp-
tion of wavelengths longer than about 350nm. The
material can be highly conducting if there is excess
of In due to a lack of oxygen - so that In acts as n-
type dopant leading to very low sheet resistances for
already only 100nm thick layers of ITO. ITO cov-
ered quartz substrates are commercially available
since they are widely used as conducting glasses in
the liquid crystal screen industry.
The sheet resistance of these ITO substrates is
lower, the thicker the ITO layer is. Typical thick-
ness of ITO in our devices is 100nm. Values smaller
than 5 Ω/ square are available. The transmis-
sion properties of the thicker ITO substrates do
not change much since the material does not ab-
sorb in the visible but interference effects cause sig-
nificant spectral dependence of the transmission.
Very thick ITO layers (several 100nm - microns)
are problematic because of the increased absolute
surface roughness
19
which can cause shorts in the
thin organic films.
We note that ITO itself can be used as anti-
reflection layer [51] as active semiconductor [122].
The surface can be modified via plasma etching[271,
82] or the semiconducting molecules can even be
chemically attached to it[178].
Other conducting glasses are Tin oxide and In-
dium oxide. Alternatively, conjugated polymers
with absorption in the whole visible range can be
used if they are doped so that the allowed energy
levels move deep into the bandgap and create an
absorption minimum in the visible region. Exam-
ple for such polymers are PEDOT [193, 263, 55] and
PITN [272, 222, 191]. An additional advantage of
PEDOT is that it can be processed from (water)
solution and therefore easily be spincoated.
Moreover, other water- insoluble materials may
be spun on top of it
20
. Some important soluble or
transparent conductors are listed with their work-
function and references in Table 2.2.
19
Considering the typical thickness of organic devices (30-
150nm) the substrate surface should ideally be smooth on a
nm scale.
20
Thicker layers of PEDOT can be obtained by repeated
spincoating since it seems poorly soluble in water once the
film is formed.
18 CHAPTER 2. SOLAR CELLS - GENERAL
Table 2.2: Molecular electrode materials for organic
solar cells. The abbreviations are listed in Chapter
1.
material W
f
(eV) comment
PANI sol. polymer[38]
PEDOT 5.2-5.3eV [33] sol. polymer
transp.[55, 39]
ITO 4.4-4.9[188, 144]
21
transp.
FTO 4.84eV [172] transp.
TO [273] transp.
2.4 Photovoltaic Characteri-
sation
2.4.1 Power Conversion Efficiency
In practice the photoexcited electron can decrease
its potential energy by losing energy to phonons
until it reaches the lowest lying level in the CB
- the LUMO. Since the phonon energy dissipates
into heat this processed is known as thermalisation.
As a consequence of thermalisation the semicon-
ductor bandgap is often regarded as a measure for
the achievable voltage. The higher the bandgap the
higher the voltage can be.
On the other hand, a low bandgap material can
absorb more photons and thus increase the num-
ber of photogenerated charge carriers i.e. the pho-
tocurrent. The lower the bandgap, the higher the
photocurrent. Hence, there must be an optimal
bandgap for a given illumination spectrum. Shock-
ley and Queisser were the first who calculated the
maximum power conversion efficiency for a semicon-
ductor with a given bandgap assuming only radia-
tive recombination and the solar radiation o earth.
They obtained a value of 30% for a semiconductor
with a bandgap of 1.12eV like Si. How the maxi-
mum conversion efficiency varies with the bandgap
can be seen in Fig.2.10.
The semi empirical limits take into account re-
alistic loss mechanism e.g. by assuming realistic
values for the fillfactor. AM0 is the solar spectrum
above earth’s atmosphere while AM1.5 is the stan-
dard spectrum at sea level - see Chapter 8. The
graphs show that for both the terrestrial and the
space solar spectrum a semiconductor bandgap be-
tween 1.3 and 1.5eV allows reaching power conver-
sion efficiencies around 30%.
Since a high load resistor reduces the current flux
the charges need more time to get out of the semi-
conductor. This means that recombination can take
Figure 2.10: Power conversion efficiency limits as
a function of the semiconductor bandgap. The
solid lines are semi empirical limits assuming AM0
(=space) and AM1.5 (=terrestrial) solar radiation;
the dashed line is based purely on thermodynamic
considerations for blackbody solar cells under AM0
radiation. The graph shows that for both AM0 and
AM1.5 condition a semiconductor bandgap between
1.3 and 1.5eV gives the highest power. Figure cour-
tesy of Ref.[97].
place increasingly and the extracted external cur-
rent decreases. This behaviour can be seen in the
fourth Quadrant of the IV characteristic in Fig.2.11.
Thus considering the voltage dependence of the
IV curve - the maximum power is the maximum
product of I and V that can be found amongst the
data points in the fourth quadrant. This maximum
area is larger the more the IV curve resembles a
rectangle with the area V
oc
x I
sc
. The ratio between
these two areas represents a measure of the quality
of the shape of the IV characteristics:
FF ≡
(IV )
max
I
sc
V
oc
(2.9)
thus:
P
max
= (IV )
max
= V
oc
· I
sc
· FF (2.10)
The higher FF the more the IV characteris-
tics resembles a constant current source with a
maximum voltage and the higher is the electric
power that can be extracted. The voltage/current
(V
p
, I
p
)combination that gives the largest power
rectangle is called the maximum power point. Thus,
any appliance connected to a solar cell can utilize
2.4. PHOTOVOLTAIC CHARACTERISATION 19
the maximum output power only if it’s supply volt-
age is around V
p
or in other words: The load resistor
R
l
= V
p
/I
p
. Otherwise power would be wasted in
heating the series resistor (V < V
p
) or via increased
current losses through the ideal diode and shunt re-
sistor (V > V
p
).
Figure 2.11: Current versus applied voltage of a
solar cell. The extracted current is negative. The
fourth quadrant represents the voltage and current
that is generated by the cell. An externally applied
voltage is necessary to obtain data points in the first
and third quadrant.
In order to describe the power conversion effi-
ciency η of a solar cell the maximum output power
P
max
has to be related to the power of the inci-
dent light P
light
. Using Eq. (2.10) to express P
max
considering the wavelength dependence of the pa-
rameters involved we can write:
η(λ) ≡
I
sc
(λ) · V
oc
(λ) · FF(λ)
P
light
(2.11)
Because of the wavelength and in-
tensity dependence
22
power conversion efficiencies
are only meaningful for a given spectral distribu-
tion and intensity. This can be the solar spectrum
or the spectrum of the artificial illumination for in-
door applications (for e.g. pocket calculators).
In order to compare efficiencies of solar cells, solar
radiation standards have been defined in the past.
The most common standard at present is the AM1.5
spectrum which can be approached by commercial
solar simulators. We note that, if international
22
the current usually scales linearly with the light intensity
but Voc
also increases slightly. The latter is neglected in Eq.
(2.11).
recognition of a solar power conversion efficiency
number of a cell is desired, the cell should be mea-
sured by one of the internationally recognised insti-
tutions that offer solar cell efficiency measurements
such as the National Renewable Energy Laborato-
ries (NREL) in Golden (USA) or the Fraunhofer
Institut for Solar Cell research in Freiburg (Ger-
many).
In case its not feasible to send a cell to such an
institute, it would still be very helpful to estimate
the solar power efficiency of a solar cell rather than
quoting monochromatic power efficiencies (or other)
to allow at least rough comparisons among different
research groups. We note that in Chapter 8 we
describe how an inexpensive solar simulator can be
built or - alternatively - η can be estimated from
intensity dependence measurements of the spectral
response.
Figure 2.12: ECD of a solar cell. The circuit con-
sists of the following ideal components: A current
source I
L
that considers the light-generated current,
a diode that accounts for the nonlinear voltage de-
pendence and a shunt as well as a series resistor.
Also shown is a load resistor R and its voltage drop
and current. The current arrows point into the di-
rection the holes flow - according to the standard in
electronics. The current I is negative if V > V
oc
and
it flows “into” the device - otherwise it is positive.
2.4.2 The Equivalent Circuit Dia-
gram (ECD)
ECDs are frequently used to describe the electric
behaviour of more complex semiconductor devices
with a network of ideal
23
electrical components such
as diodes, current or voltage sources and resistors.
23
A real current source for example always has losses due
to a shunt and series resistor.
20 CHAPTER 2. SOLAR CELLS - GENERAL
Fig.2.12 shows the ECD that is typically used for
inorganic solar cells. Although the specific physi-
cal processes in organic semiconductors may be dif-
ferent and therefore lead to other parameters, the
principal loss mechanisms are the same and we may
therefore apply the same circuit. We suggest to as-
sociate the following (more general) processes with
the single components in the ECD:
• The current source generates current I
L
upon
illumination. I
L
= number of dissociated ex-
citons /s = number of free electron/hole pairs
immediately after generation - before any re-
combination can take place.
• The shunt resistor R
sh
is due to recombination
of charge carriers near the dissociation site (e.g
D/A interface). Provided the series resistor R
s
is at least one order of magnitude lower than
R
sh
, it may also include recombination fur-
ther away from the dissociation site e.g. near
the electrode. Otherwise an extra shunt resis-
tor R
sh2
has to be considered - see Fig.2.13.
R
sh
can be derived by taking the inverse slope
around 0V:
R
sh
≈ (
I
V
)
−1
(2.12)
This is because at very small voltages the diode
D is not conducting and the current driven by
the external voltage (positive or negative) is
only determined by R
sh
+ R
s
with R
sh
(typi-
cally) being much larger.
• The series resistor R
s
considers conductivity
i.e. mobility of the specific charge carrier in
the respective transport medium. For exam-
ple mobility of holes in a p-type conductor or
electron donor material. The mobility can be
affected by space charges and traps or other
barriers (hopping). R
S
is also increased with a
longer traveling distance of the charges in e.g.
thicker transport layers. R
s
can be estimated
from the (inverse) slope at a positive voltage
> V
oc
where the IV curves becomes linear:
R
S
=≈ (
I
V
)
−1
(2.13)
This is because at high positive external volt-
ages V the diode D becomes much more con-
ducting than R
sh
so that R
S
can dominate the
shape of the IV curve.
• ideal diode D = voltage dependent resistor that
takes into account the asymmetry of conduc-
tivity due to the built in field in D/A cell (dif-
ference between the acceptor LUMO and the
donor HOMO) or the nature of the semicon-
ductor electrode interface (blocking contact)
in single layer cells. This diode is responsible
for the nonlinear shape of the IV curves. The
diode characteristic is not necessarily Shockley
type. Note that the IV characteristic of the
ideal diode D is only equal to the IV charac-
teristic of the entire cell (circuit) if R
s
= 0Ω
and R
sh
= ∞Ω.
• solar cell voltage V. The cell can generate a
voltage between 0 and V
oc
depending on the
size of the load resistor. In order to obtain
IV curve data in other voltage ranges (V <
0 and V
oc
< V ) in the IV curve an external
voltage source is required. We note that also
the voltage drop across a load resistor - the
range between 0 and V
oc
- can be simulated
by the same voltage source so that the entire
range can be scanned by applying an external
voltage.
Note that the current for V > V
oc
and the extra
current for V < 0V is delivered from the exter-
nal voltage source. The external voltage source
can then act as a current amplifier to boost
the photo sensitivity
24
but the actual photon
to current conversion efficiency (EQE) of the
solar cell or photodiode can not really be in-
creased.
These are the components of a ECD with which
we can associate the most important effects in solar
cells of all types. However, a more comprehensive
ECD for organic devices - see Fig.2.13 may comprise
the following extra components:
• another diode D
2
. The formation of an ex-
tra blocking contact (e.g for holes at the ITO
electrode) can affect the IV curve in the third
quadrant or even lead to FF values < 0.25. We
note that the normal blocking contact for elec-
trons at ITO and holes at Al allows charge in-
jection into the electrodes if a sufficiently high
negative voltage is applied. This diode is al-
ways present and not considered in any of the
ECDs drawn here.
• the capacitor C. The capacitor that takes into
account charging/discharging and other time
24
Provided the current increases with applied voltage.
2.4. PHOTOVOLTAIC CHARACTERISATION 21
Figure 2.13: Extended ECD of a solar cell. The
formation of a counter diode D
2
may sometimes
need to be considered as well as another shunt R
sh
that connects the two electrodes directly. The ca-
pacitor accounts for accumulation i.e. storage ef-
fects of charges and may be of particular interest
for AC measurements i.e. photocurrent measure-
ments with chopped light.
dependent effects that can be significant since
the contact area A can be large and the dis-
tance between the electrodes d is very small:
C =
A
d
• an extra shunt resistor R
sh2
. It considers
shorts due to pinholes or significant conductiv-
ity of the bulk material. It may also account for
recombination losses near the electrodes. The
effect of R
sh2
is considered in R
sh
if R
s
is con-
siderably smaller than any of the two shunts.
We note that current and resistor values depend
on the illuminated area and need to be related to it
if comparisons between cells with different illumina-
tion areas are desired
25
. Note that the dark current
depends on the actual device area - regardless of the
size of the light spot.
Using the simple ECD in Fig.2.12 and Kirchoff’s
laws for current knots and voltage loops, we can
formulate the following relation:
I
sh
. .. .
(I
L
−I
d
−I) R
sh
= V +IR
s
(2.14)
which can be transformed into
25
Within this thesis the illuminated area was typically
around 4mm
2
(only slightly smaller than the active area)
which allows immediate comparison of the absolute values.
I(
R
S
R
sh
+ 1) = I
L
−I
d

V
R
sh
(2.15)
if we now assume that the Shockley diode equa-
tion describes the voltage dependence of the current
I
d
through the ideal diode D
I
d
= I
0
· (e
V −IRs
nkT/q
−1) (2.16)
and replace I
d
in Eq. (2.15) we obtain
I = (I
L

V
R
sh
)
R
sh
R
sh
+R
s
−I
0
R
sh
R
sh
+R
s
(e
V −IRs
nkT/q
−1)
(2.17)
If we now use
R
sh
+R
s
R
sh
= 1 +
R
s
R
sh
(2.18)
we can write Eq. (2.17) as:
I =
I
L

V
R
sh
1 +
Rs
R
sh

I
0
1 +
Rs
R
sh
· (e
V −IRs
nkT/q
−1)
(2.19)
This equation will be used later when we discuss
what can be learned from the specific shapes of IV
characteristics. Even though the current I in Eq.
(2.19) appears also in the exponent which requires
numerical methods to find solutions, we can already
see the following:
Inorganic solar cells have R
s
values
26
between
0.3Ω and a few Ohms and R
sh
larger than 1000Ω
so that the ratio R
s
/R
sh
is very small and the term
1 + R
s
/R
sh
is virtually 1 and can therefore be ne-
glected in Eq. (2.19). However, even in silicon solar
cells the voltage drop across R
s
represents an im-
portant loss factor and has to be considered as well
as the loss current through the shunt. Note that
both R
s
and R
sh
still appear in Eq. (2.19) even
though 1 + R
s
/R
sh
an be neglected. However, the
situation can be very different for organic cells.
Organic cells very often suffer from high series
resistors and - in addition - relatively small R
sh
values. As a consequence not only the losses due
to both resistors are higher than in silicon cells but
also the term 1 + R
s
/R
sh
can become significantly
different from 1 and has to be considered in Eq.
(2.19).
As a result, the shape of the IV curves and the
absolute currents are more affected by these resistor
values in organic than in inorganic cells.
26
Related to 1cm
2
22 CHAPTER 2. SOLAR CELLS - GENERAL
2.4.3 Open Circuit Voltage
In the preceding Section we have deduced a formula
that allows calculating the output current of a solar
cell. In this Section we derive an expression for the
open circuit voltage V
oc
and consider the effects of
the single components of the ECD. We will also
discuss how PL efficiency can positively affect the
solar cell performance and show how the upper limit
for V
oc
can be estimated.
The Ideal Solar Cell
First we consider the case of an ideal solar cell that
comprises only the ideal
27
diode D in the dark so
that R
s
= 0 and R
sh
= ∞ in Eq. (2.19) turns into:
I
d
= I
0
(e
qV/nkT
−1) (2.20)
The current through such a cells or photodiode
is determined only by the current through the ideal
D - here represented by the Shockley equation. For
positive voltages I increases exponentially.
Upon illumination the light generates a photo
current I
L
that is simply superimposed (added)
upon the normal rectifying IV characteristics of the
diode D:
I = I
0
(e
qV/nkT
−1) −I
L
(2.21)
The addition of I
L
results in a region of the fourth
quadrant where electrical current and voltage can
be extracted from the solar cell. The highest voltage
in this quadrant develops at the electrodes when
I
L
just manages to cancel the dark current. Thus,
given a constant I
L
the V
oc
is higher the smaller the
dark current is - note that I
0
determines the height
of the characteristics. Canceling of I
d
by means of
I
L
can be considered in Eq. (2.21) by setting I = 0.
V
oc
can than be derived quantitatively using:
V
oc
=
nkT
q
ln(
I
L
I
0
+ 1) (2.22)
Note that the effect of the I
L
/I
0
ratio is relatively
small. Reduction of I
0
by a factor 10x increases V
oc
only by 25mV ·ln(10) (58mV). Typical V
oc
of silicon
cells under solar conditions are around 550mV.
Effect of R
sh
A formula for V
oc
that considers the influence of
R
sh
can be deduced from Eq. (2.19) after setting
27
We keep the ideality factor in this equation to consider
different shapes. However, the diode D is still ideal in the
sense that there are no Ohmic losses.
the output current to zero (I ≡ 0) to give
28
:
V
oc
=
nkT
q
ln(
I
L
−V
oc
/R
sh
I
0
+ 1) (2.23)
This equation relates the maximum output volt-
age not only to the light generated current I
L
but
also to the reverse saturation current and the shunt
resistor. Now, also the formula for V
oc
becomes
transcendent and requires numerical modeling to
find solution. The value for V
oc
still depends on
the ratio I
L
/I
0
but now I
L
is decreased by the pres-
ence of a finite R
sh
whereas R
s
is unimportant since
there is no current flowing through it that can cre-
ate a voltage drop (loss).
Both the ideal diode D and R
sh
are now the com-
ponents that determine V
oc
: Suppose R
sh
is not
very high and the device is in the dark. If we ap-
ply a positive voltage across the cell electrodes we
create a voltage drop across R
sh
that is equal to
the voltage V
d
across the ideal diode D. The cur-
rent that can pass through the diode D at V
d
is
determined by its IV characteristic
29
. The sum of
the currents through D and R
sh
yields the current
through the electrodes of the solar cell for a given
applied voltage.
Upon illumination, the current source generates
the current I
L
some of which passes through the
diode where a voltage drop is generated that is big
enough to allow the rest of I
L
to go through R
sh
-
if the electrodes are open. The same voltage can be
measured with a voltmeter with high internal resis-
tor across the device electrodes and is then termed
open circuit voltage V
oc
.
Note that n which determines the shape of the
IV curve stands outside the logarithm in Eq. (2.23)
and has therefore a stronger influence than a varia-
tion of R
sh
or I
0
. The latter controls the ”height”
of the IV curve of D. However, since R
sh
can vary
considerably it can seriously decrease V
oc
if V
oc
/R
sh
is not much smaller than I
L
. We note that I
0
can
be related to PL efficiency as discussed later in this
thesis.
Effect of Shorts
Organic solar cells have often values for series resis-
tance which are more than 3 orders of magnitude
28
There is no influence of Rs since Rs can only affect the
voltage when there is a current flowing through it.
29
Note that the IV characteristics of the diode D is not
the IV curve one obtains when measuring the entire device
(with R
sh
and Rs)! For inorganic solar cells it is assumed
to be the Schottky diode characteristics - while for organic
cells it depends on many material and device parameters.
2.4. PHOTOVOLTAIC CHARACTERISATION 23
larger. This means that currents measured in short
circuit mode still have to get through a large re-
sistor which can seriously limit the output current
particularly if also R
sh
is low. Moreover, since a
short across the electrodes represented by R
sh2
in
Fig.2.13 represents a voltage divider in conjunction
with R
s
according to Eq. (2.24) R
s
can actually
reduce V
oc
if R
s
reaches values in the order of R
sh2
.
Thus ”shorts” do not need to have very low resis-
tances to smaller V
oc
considerably - they just need
to approach the value of R
s
which can be as big as
10
7
Ω in some small area organic devices!
V
ocs
= V
oc
·
R
sh2
R
s
+R
sh2
(2.24)
This also makes voltage measurements across the
device electrodes more critical. Voltmeters need to
have internal resistors larger than 10
7
Ω to exclude
effects on the circuit that lead to lower readings.
Analogously, recombination that takes place near
the electrodes - represented by R
sh
may be more
effective in decreasing V
oc
than recombination in
the middle of the device if R
s
is so high.
The Effect of Photoluminescence
Any type of recombination
30
represents a less mech-
anism for the photoexcited charge carriers in a so-
lar cell. However, non-radiative recombination af-
fects the photoexcited charge carriers statistically
much faster than the unavoidable radiative decay
that does not require any semiconductor defects.
Hence, if there is only radiative recombination the
charge carriers have much better chances to reach
the electrodes and circumvent the radiative decay
by entering an external circuit.
If there is no external circuit (open circuit mode)
the charges accumulate at the electrodes to build up
V
oc
before they decay showing their maximum PL
efficiency. Indeed, it has been shown that the PL of
e.g. chlorophyll is about 3% in a living plant, and
30% if the chlorophyll molecules are separated from
the rest of the electron transport chain, which cor-
responds to operation under load and open circuit
mode, respectively [208, 234].
In other words, a highly fluorescent material can
have relatively low charge carrier mobilities and still
generate a high V
oc
and good photocurrents. Ac-
cording to Ref. [234] radiative recombination Φ
PL
can be considered as a correction factor for the re-
verse saturation current I
0
in the diode character-
istics
30
Before passing through the external circuit.
I
d
=
I0
. .. .
I
01
Φ
PL
·(e
V
nkT/q
−1) (2.25)
Note that the Shottky equation remains un-
changed if the photoluminescence efficiency Φ
PL
=
1 i.e. 100%. Within this thesis we call I
01
the PL
independent reverse saturation current. Thus the
expression Eq. (2.23) to derive V
oc
can be written
as
V
oc
=
nkT
q
ln(
I
l
−V
oc
/R
sh
I
01

PL
+ 1) (2.26)
The reverse saturation current I
0
= I
01

PL
can
be obtained experimentally by modeling of the IV
curve and gives indeed about 10
3
larger values for
e.g. silicon cells (Φ
PL
≈ 10

4) than for cells using
organic semiconductors (Φ
PL
≈ 10

2). Indeed the
more fluorescent porous silicon has been reported
to show lower I
0
values being consistent within Eq.
(2.25). The higher PL (10
−1
) in GaAs predicts V
oc
values of 1V under 1 sun which is also in good agree-
ment with the observations [234].
We note that non-radiative recombination losses
affect both I
0
and the shunt resistor R
sh
. For very
fluorescent materials the I
0
is small and R
sh
should
be high.
In fact, polymers and dyes with higher PL yield
appear to give high V
oc
values like PPV (1.2V),
MEH-PPV (1.3V) in contrast to poorly fluores-
cent materials like PIF (V
oc
< 0.02V) and HPc
(V
oc
<0.2V). So called non-fluorescent materials
may, however, still give reasonable voltages as long
as their optical and charge transport properties are
very good and the PL is not too low. For exam-
ple, the low PL of e.g Si (10
−3
to 10
−4
) is still high
enough to result in a I
L
/I
0
allowing V
oc
to reach
0.6V [234]. Similar considerations hold for the very
good photovoltaic performance of certain organic
material with relatively low PL efficiencies.
The Upper Limit of V
oc
In the previous Section we have discussed how
R
s
, R
sh
and in particular I
0
can affect the value
for V
oc
. In this Section we discuss whether there is
an upper limit for V
oc
and what might determine it.
We have seen that according to Eq. (2.26) the
voltage is higher the lower I
0
and the higher I
L
is.
In virtually all inorganic cells and many organic
ones I
L
scales linearly with the light intensity over
many orders of magnitude. Possible upper lim-
its with regard to excitation intensity may only be
24 CHAPTER 2. SOLAR CELLS - GENERAL
reached if the charge carrier density approaches crit-
ical values which would typically only occur at un-
realistic high excitation intensities.
What determines in practice the absolute value of
I
0
in inorganic devices has been reported: major-
ity charge carrier and dopant concentration, charge
carrier diffusion lengths and recombination veloci-
ties [97].
Here we focus on a more general approach to find
the V
oc
upper limit so that it may be applicable to
the organic solar cell structures described in this
thesis.
Ideally the potential difference between the
ground state and the excited state of an electron
after photo-excitation can be regarded as a driv-
ing force i.e. maximum voltage. Thus the maxi-
mum photo-voltage that a semiconductor can de-
velop should increase linearly with the the excita-
tion energy. In other words, a 2eV photon should
be able to supply a voltage of 2V.
In practice, in both organic and inorganic semi-
conductors photoexcited electrons can decrease
their potential energy very quickly using the high
density of allowed states in the CB until they reach
the lowest unoccupied molecular orbital (LUMO).
Since the lost potential energy is usually converted
into phonons i.e. heat this process is called thermal-
isation. As a consequence the difference between
the HOMO and LUMO levels (the bandgap) can
be seen as a theoretical upper limit for V
oc
. Prac-
tice confirms that indeed inorganic solar cells made
of higher bandgap materials develop higher V
oc
and
the maximum achieved V
oc
has never exceeded the
width of the bandgap.
For organic solar cells, we also find higher V
oc
for higher bandgap materials. As discussed earlier
a high V
oc
may be favored by a lower reverse sat-
uration current I
0
due to the frequently low non-
radiative recombination in these more fluorescent
materials. However, it has been observed for many
organic solar cells that a higher workfunction differ-
ence of the electrodes ∆W
f
can also increase V
oc
.
This is often explained by assuming that ∆W
f
in-
troduces a “built in field” across the thin semicon-
ducting layer that helps to extract the photoexcited
charge carriers.
Qualitatively, as it can be seen from Fig.2.4, a
larger ∆W
f
results in either a stronger tilting of the
semiconductor band for voltages < V
oc
or a stronger
bend bending near the electrode semiconductor in-
terface if there are enough charges to cancel the
field
31
within a fraction of the bulk thickness. Thus,
31
A gradient of a potential in space can be treated as a
low W
f
for the negative and high W
f
for the posi-
tive hole collecting electrode often helps to increase
V
oc
. The drawback is that the use of low W
f
ma-
terials such as Ca often requires the sublimation of
a protection layer on top because of the increased
tendency to form an oxide layer when exposed to air
that represents an extra obstacle for charge carrier
extraction.
On the other hand shorts may be avoided more
efficiently which also helps to develop higher volt-
ages at the electrodes. We note that in single layer
devices ∆W
f
> E
g
is not expected to increase V
oc
above E
g
since charge injection from the LUMO
and HOMO into the electrodes is limited.
In D/A cells
32
, however, ∆W
f
> E
g
may result in
V
oc
larger that the “combined bandgap” E
gDA
since
the internal field at the D/A interface can assist
the extraction of the charges through the “high” ly-
ing electrode workfunction. V
oc
larger than E
g
has
been observed in a laminated device as discussed in
Chapter 6.
As a result of these considerations the upper limit
for V
oc
in single layer devices is expected to be the
bandgap whereas in D/A cells ∆W
f
of the elec-
trodes or the bandgap - whichever is higher - deter-
mines the maximum photogenerated voltage.
In case of ∆W
f
> E
gDA
the built in field due
to the offset of LUMOs (∆LUMO) and HOMOs
(∆HOMO) between the two components has to
be large enough to provide the extra driving force.
Hence, the maximum V
oc
can be written as:
V
oc
≤ |LUMO
A
−HOMO
D
|+∆LUMO+∆HOMO
(2.27)
if only positive numbers are used.
Note that ∆LUMO+∆HOMO has to be greater
than the exciton binding energy to split the excitons
in D/A devices.
The above formula can be written as
V
oc
≤ E
g
D
+E
gA
−|LUMO
A
−HOMO
D
| (2.28)
Thus, it can be expected that the bandgaps of
the semiconducting components determine the (the-
oretical) upper limit of D/A cells provided the ∆W
f
is large enough.
We note that the consideration of the unavoidable
radiative recombination losses already reduces the
upper limit for V
oc
in Si from the bandgap value
(1.1V) to about 0.8V [234].
field.
32
In case of D/A devices the voltage relevant bandgap is
the difference between the LUMO of A and the HOMO of D.
2.4. PHOTOVOLTAIC CHARACTERISATION 25
Summary
According to the described model a high V
oc
at a
given light intensity is expected if:
1. I
0
is low. The reverse saturation current is of-
ten low if the PL efficiency is high, the bandgap
is high and the conductivity is low.
2. The shape of the IV curve is not too steep. The
dark current can then only be canceled by the
photocurrent at higher (V
oc
) voltages.
3. R
sh
is clearly larger than the DC resistor value
of the ideal diode D.
Note that not all requirements to obtain a high
V
oc
coincide with the requirements to obtain a high
short circuit current. For example, a large I
sc
is
favored if the bandgap is small and the conductivity
is high. However, both V
oc
and I
sc
are high if R
s
is
small and R
sh
is high.
2.4.4 Photocurrent
The Short Circuit Current I
sc
The current I
sc
under short circuit condition (R
L
=
0 - see Fig.2.13) can be derived from Eq. (2.19) by
setting V ≡ 0. If we also consider the influence of
the PL as discussed above we can write:
I
sc
=
I
L
−I
sh
. .. .
I
L
1 +
Rs
R
sh

I
d
. .. .
I
0

PL
1 +
Rs
R
sh
(e
IRs
nkT/q
−1) (2.29)
From the ECD in Fig.2.12 can be seen that the
short circuit current is simply the light current I
L
(photogenerated charge carriers/s i.e. number of
split excitons/s) reduced by the current through the
shunt I
sh
and the diode D, I
d
:
I
sc
= I
L
−I
sh
−I
d
(2.30)
This is the highest current that can be extracted
from a solar cell
33
. The light current increases lin-
early with the illumination intensity. In order to ex-
tract a quantity that describes how efficiently pho-
tons are converted into charges in the external cir-
cuit, I
sc
has to be related to light intensity. This
leads to the definition of the spectral response:
33
Even if an external voltage is applied, only the photo-
sensitivity can be amplified - any current higher than Isc
current originates in the external voltage source.
SR(λ) =
J
sc
(λ)
Φ(λ)
(2.31)
with Φ being the light intensity per illuminated
area
34
(W/m
2
) and J
sc
representing the current
density (A/m
2
) in short circuit mode.
The spectral response (SR) thus gives the cur-
rent in the external circuit per watt incident light
(A/W). This quantity is frequently used to char-
acterise the light sensitivity of photodiodes since
the product SR · J
sc
gives immediately the light
intensity
35
.
For the characterisation of solar cells a quantity
that considers the actual fraction of the incident
photons that can be converted into electrons in an
external circuit is of higher interest:
EQE ≡
number of electrons in external circuit
number of incident photons
(2.32)
EQE stands for external quantum efficiency and
represents, together with the power conversion effi-
ciency η, the most interesting parameter of a solar
cell. The EQE can be derived from the spectral
response considering that the energy of a photon
E
p
= hc/λ with h being Planck’s constant, c the
speed of light and q the elementary charge using
the following formula
36
:
EQE(λ) = SR(λ) ·
hc

(2.33)
Note that the EQE gives higher values for shorter
wavelengths when compared with the SR spectrum.
This is because the spectral response relates to the
energy of photons whereas the EQE refers only to
the number of particles. Since the output current is
determined by the number of electrons that can be
“pumped” into the CB regardless of their energy,
the EQE represents a true measure for the photon
to current conversion efficiency in contrast to SR.
The EQE can be converted into the internal
quantum efficiency IQE if only the fraction of the
actually absorbed photons are considered:
IQE(λ) =
EQE(λ)
1 −R(λ) −T(λ)
(2.34)
with R(λ) denoting the fraction of reflected light
and T(λ) the fraction of the transmitted light.
34
The hardly used but accurate term is: irradiance.
35
Light intensity measurements are an important applica-
tion for photodiodes.
36
Note that hc/q = 1240WnmA
−1
.
26 CHAPTER 2. SOLAR CELLS - GENERAL
The IQE is very helpful and frequently used in in-
organic devices to investigate the physical processes
in the semiconductor material. In contrast to inor-
ganic solar cells where the complete bulk absorbs
light and contributes to the photocurrent, organic
devices have typically large regions that act only as
optical filter.
In addition, interference effects would have to be
considered because of the very thin films compared
to inorganic cells. This makes the interpretation of
IQE spectra in organic cells only more speculative
rather than simpler and therefore generally less use-
ful at the present time. The EQE is therefore the
preferred quantity in the devices discussed in this
thesis.
The Optical Filter Effect
The region in the semiconductor that allows pho-
toexcited excitons to dissociate and the separated
charges to reach their respective electrons is often
called the ”active region” - see Fig.2.14. Thus, only
light absorbed in the active region can contribute
to the photocurrent. The layer that absorbs light
before it reaches the active region
37
is the actual
optical filter.
Figure 2.14: In organic devices very often only a
thin layer (active layer) near back electrode is elec-
tronically active (contributes to the photocurrent).
Depending on the thickness and absorption coeffi-
cient a significant fraction of photons can be ab-
sorbed in the semiconductor bulk (optical filter) be-
fore this region is reached. Figure from [63].
Fig.2.15 shows two scenarios for the formation of
an active region smaller than the bulk which can
lead to an optical filter effect:
37
Electrode/polymer interfaces often act as dissociation
site but also polymer/polymer interfaces or impurities
(dopants, oxygen molecules) can act as such - see also Section
2.2.2.
1. Exciton dissociation throughout the entire bulk
but small mobility of one charge carrier. Often
the mobility of the hole is considerably higher
than the mobility of the electron. Thus, free
electrons have to be generated close to the neg-
ative electrode for charge collection.
Even if the excitons have large diffusion
lengths, a considerable fraction of the pho-
togenerated excitons would dissociate too far
away from the negative electrode. Thus the
width of the active layer is determined by the
distance electrons can travel L
e
in the specific
material i.e. the electron mobility - see Fig.2.15
a).
2. Exciton dissociation occurs only near one elec-
trode interface. If exciton splitting occurs due
to a strong local field (band bending), excitons
that are generated outside the band bending
area have to diffuse into this region to supply
charges. Although small hole and electron mo-
bilities can still become limiting factors it is
usually the exciton diffusion length that deter-
mines the width of the active layer in this case
- see Fig.2.15 b).
Figure 2.15: Two scenarios that can lead to the for-
mation of an optical filter: a) exciton dissociation
occurs throughout the entire bulk but one charge
carrier (here the electron) can only travel a very
short distance before it recombines. b) exciton dis-
sociation occurs only near the (here negative, Al)
electrode due to a local high field. Thus a small ex-
citon diffusion length determines the width of the
active layer - provided the charge carrier traveling
distance is sufficiently high.
Only if the active region extends across the entire
bulk i.e. the bulk is thin enough, most of the disso-
2.4. PHOTOVOLTAIC CHARACTERISATION 27
ciated excitons can generate a photocurrent so that
the EQE essentially follows the absorption features.
Due to short exciton diffusion lengths most or-
ganic cells have a bulk thickness that is consider-
ably larger than the active layer (to avoid shorts).
Thus these EQE spectra show a strong filter effect
resulting in EQE maxima at wavelengths where the
optical density is low and EQE minima where the
OD is too high to allow many photons to reach the
active region.
Such a spectrum is termed antibatic spectrum in
contrast to the symbatic spectrum that shows no
offset between absorption and EQE features [120].
Note that a symbatic spectrum can be obtained if
the device is illuminated through the other contact
- if it is sufficiently transparent.
In Section 3.2 we describe a method to estimate
how thick a semiconducting layer can be before it
shows the conversion losses due to the optical filter
effect.
Figure 2.16: Effect of R
sh
(ranging from 10
4
to
10
7
Ω) on the shape of the IV characteristic. A large
R
sh
can reduce both the fillfactor and V
oc
but not
I
sc
. The dashed curve is the dark characteristic at
10
6

2.4.5 Interpretation of IV Charac-
teristics
Inorganic Solar Cells
The higher charge carrier mobility in inorganic
semiconductors leads to sufficiently long diffusion
lengths which allow most charge carriers to reach
the electrodes before they undergo intense recom-
bination - even though the devices are about 1000
times thicker. In addition R
sh
is typically more
than 3 orders of magnitude larger than R
s
. This
not only leads to a negligible term 1 +
Rs
R
sh
in Eq.
(2.19) and Eq. (2.29) but also makes it much easier
to distinguish the effects of the two resistors in the
IV characteristics.
In addition the reverse saturation current I
0
is
typically larger than in organic devices - probably
due to the higher PL efficiency in the latter - as
discussed in Section 2.4.3. The values for R
sh
and
R
s
have been estimated from the IV curve data as
described in Section 2.4.2.
Fig.2.16 shows how the IV characteristic changes
if the shunt resistor varies between 10
4
and 10
7
Ω assuming the shown
38
values for R
s
, I
L
, I
0
. If
R
sh
> 10
6
Ω the shape (FF,V
oc
) of the IV curves
remains virtually unchanged and the current shows
no significant field dependence for negative bias.
Figure 2.17: Effect of R
sh
(ranging from 10
1
to
10
4
Ω) on the shape of the IV characteristics. V
oc
is virtually zero and R
sh
extends its “linearisation
effect” into the first quadrant. The photocurrent
(I
sc
−I
0
) is too small to be visible on this scale.
However, clearly smaller R
sh
values have enor-
mous bad effects on the IV curve. The slope (field
dependence) in the third quadrant increases consid-
erably, V
oc
approaches zero and the fillfactor reaches
its theoretical minimum of 0.25 (Ohmic - linear)
39
38
The currents given in this Section are related to an area
of 1mm
2
- which is sufficiently close to our experimental con-
ditions (A≈ 4mm
2
) to get a “feel” for the absolute numbers.
39
Lower values than the one for ohmic behaviour may be
28 CHAPTER 2. SOLAR CELLS - GENERAL
very quickly. However, the short circuit current is
not affected since the current through the shunt can
be neglected if R
s
<< R
sh
.
Figure 2.18: Effect of R
s
on the shape of the IV
characteristics assuming the same values for an in-
organic solar cell as above with R
sh
= 10
6
Ω. R
s
reduces FF and the slope in the first quadrant when
it reaches the same order of magnitude as R
sh
. R
s
can reduce the short circuit current but not V
oc
.
The slope in the third quadrant is still dominated
by R
sh
.
Note that the reverse slope around 0V or in the
third quadrant is equal to the value of R
sh
. This
can easily be seen from the ECD - Fig.2.12. Under
reverse bias the diode D is blocking and R = R
sh
+
R
s
only determine the voltage dependence of the
current (I = R
−1
U). Since R
s
is much smaller than
R
sh
it is only R
sh
that determines the slope R
−1
.
Fig.2.17 shows how even smaller R
sh
values af-
fect the IV curve. The slope in the third quadrant
increases even further and the “linearisation effect”
of R
sh
extends to about 0.3 V in the first quadrant.
Note that V
oc
is virtually zero. Note that the values
for R
sh
> 10
4
Ω that have been shown in Fig.2.16
do look exactly the same on this scale.
Fig.2.18 shows what happens if the series resistor
increases assuming the same parameters as above
and R
sh
= 10
6
Ω: While the slope in the third
quadrant remains unchanged the slope in the first
quadrant starts to decrease considerably when R
s
reaches the same order of magnitude as R
sh
. The ef-
fect of the large R
s
even extends into the 3 quadrant
obtained if another diode (with reverse orientation) is formed
(=counterdiode).
thereby decreasing the fillfactor to its minimum.
Moreover, if the two resistors have similar values
the IV curve is dominated by their ohmic character-
istics (the inverse slope in the 1 quadrant is equal to
R
s
) and the ideality factor or voltage dependence
of D may then have little influence. Note that -
in contrast to the effect of R
sh
- the short circuit
current can decrease but the open circuit voltage
can not be affected at all - since there is no current
through R
s
at V
oc
.
Figure 2.19: Effect of R
sh
(ranging from 10
7
to
10
10
Ω) on the shape of the IV characteristics as-
suming typical dark and light currents for an or-
ganic solar cell. The effect of R
sh
is qualitatively
the same as for the inorganic device in Fig.2.16 but
1000 times larger resistor values are required to give
the same shape.
Organic Solar Cells
Lower mobilities and smaller charge carrier concen-
tration lead to considerably smaller currents in most
organic solar cells. In fact, both the dark
40
and light
currents are about 1000 times smaller (I
0
≈ 1pA
and I
sc
≈ 1nA) so that the light current in these
devices is about as big as the dark current in sil-
icon devices while the dark current is still about
1000 times smaller so that according to Eq. (2.22)
a similar V
oc
is expected.
We note that the dark current may be related to
the fraction of radiative recombination so that high
PL-efficiency - as observed in many organic semi-
conductors - is expected to give a low dark current.
40
The dark current I
d
scales with I
0
2.4. PHOTOVOLTAIC CHARACTERISATION 29
This is discussed in more detail in the Section about
V
oc
(Section 2.4.3).
Figure 2.20: Effect of R
sh
(ranging from 10
3
to
10
6
Ω) on the shape of the IV characteristics as-
suming typical dark and light currents for an or-
ganic solar cell. The effect of R
sh
is qualitatively
the same as for the inorganic device in Fig.2.17 but
1000 times larger resistor values are required to give
the same shape.
Fig.2.19 shows the effect of a decreasing R
sh
on
the IV curve using current values typical for organic
solar cells. In order to show the influence of only
R
sh
we assumed an unrealistic low value (1Ω) for
R
s
.
Even lower values for R
sh
are considered in
Fig.2.20. Note that the IV curve for values larger
than 10
6
Ω - like e.g. all curves shown in Fig.2.19 -
look exactly the same on this scale.
Both figures show qualitatively the same effect
of a decreasing R
sh
: a steeper slope in the third
quadrant and eventually reduction of V
oc
. However,
although the I
sc
/I
d
ratio is exactly the same the
R
sh
value has to be 1000 times higher to cause the
same effect. For example, an R
sh
as low as 10
5
Ω in
the inorganic cell (Fig.2.16) is necessary to decrease
V
oc
by 50% whereas a value of 10
8
Ω can show the
same effect in the organic device (Fig.2.19).
Fig.2.21 shows the effect of R
s
with currents typ-
ical for organic cells and assuming R
sh
= 10
9
Ω.
Again, we observe the same effects as in for the inor-
ganic device upon increasing: decrease of the slope
in the first quadrant and reduction of I
sc
. How-
ever, the output current is 1000 times smaller and
the resistor value 1000 times higher - compare with
Figure 2.21: Effect of R
s
on the shape of the IV
characteristics assuming the same values for an or-
ganic solar cell as above with R
sh
= 10
9
Ω. R
s
decreases FF and the slope in the first quadrant if
it reaches the same order of magnitude as R
sh
. R
s
can decrease the short circuit current but not V
oc
.
The slope in the third quadrant is still dominated
by R
sh
.
Fig.2.18.
Fig.2.22 shows that - apart from high R
s
and/or
R
sh
values - a low light current can cause both a low
FF and a low V
oc
. Thus, IV curves should only be
compared if light intensities are similar and FF and
V
oc
values (mW/cm
2
) should be quoted together
with light intensity numbers.
In fact, IV curves of inorganic solar cells are usu-
ally measured under standard (AM1.5) conditions
to allow direct comparisons. The problem with
many organic solar cells is that they may not be
stable enough under such high light intensities and
that depending on the material, device type and
measurement condition, IV curves - particularly of
very inefficient devices - can vary significantly from
pixel to pixel or even over time
41
.
As a guideline it may be said that the light cur-
rent should be at least 3 orders of magnitude higher
than the dark current before the fillfactor can be
regarded as light intensity independent and used
for direct comparisons. Otherwise, the potential fill
factor and V
oc
for realistic conditions may be un-
derestimated considerably.
41
We note that unlike many IV curves the EQE spectra
are usually very well reproducible.
30 CHAPTER 2. SOLAR CELLS - GENERAL
Figure 2.22: Effect of I
sc
on the IV characteristics of
an organic solar cell. The dashed curve is the dark
characteristic. It can be seen that low I
sc
values can
result in a considerably smaller fillfactor and also
lower V
oc
values compared to higher I
sc
numbers.
2.4.6 Space Charges & Traps
The comparatively low charge carrier mobilities in
most organic semiconductors
42
in conjunction with
the need to drive high currents through the device
can lead to accumulation of charges in the semi-
conductor bulk. These excess charges build up a
space charge that creates a field which reduces the
throughput i.e. increases the transit time of charges
between the device electrodes. This phenomenon is
termed space charge limited current (SCLC).
Provided there are no traps, the total current (I)
through a semiconductor considering space charge
formation as a function of voltage (V) can be ex-
pressed as [138]:
I = qp
0
µ
p
V
d
. .. .
ohmic
+
9
8
µ
p
V
2
d
3
. .. .
SCLC
(2.35)
p
0
is the density of the thermally generated free
carriers, µ is the mobility, d is the semiconductor
thickness, the dielectric constant of the semicon-
ductor.
As long as the density of the the thermally gener-
ated free carriers p
0
is dominant the current follows
Ohm’s law. Otherwise the current increases with
the square of the voltage according to the second
term in Eq. (2.35). This second term is known as
42
Typical values are below 10
−4
cm
2
V
−1
s
−1
Figure 2.23: Schematic log-log plot showing the
shape of the current-voltage characteristics for
space charge limited current with deep and shallow
traps.
the Mott and Gurney equation for trap free space
charge limited currents. The transition from the
ohmic to the space charge conduction here is very
smooth.
If the material contains shallow traps the res-
idence time inside the traps is shorter than the
charge carrier transit time. The space charge is
then composed of the mobile free carriers and those
carriers that are thermally freed from the shallow
traps. The trap free Mott and Gurney equation will
still be applicable, if the mobility µ is multiplied by
a factor F defined as the proportion of free charge
carriers, giving an effective mobility:
µ
eff
= Fµ (2.36)
Once the applied electrical field is strong enough
to release the charge carriers from the traps, the
current will return to obey the trap free Mott and
Gurney equation - see Fig.2.23.
If the material contains deep traps the residence
time inside the traps is longer than the charge car-
rier transit time. The SCLC depends then on both
the density of traps and their distribution inside
the bandgap. The transition from the ohmic con-
duction to the trap free condition occurs sharper
than for shallow traps - see Fig.2.23.
Chapter 3
Single Layer Devices
3.1 Introduction
In its simplest form, an organic solar cell consists
of a single polymer layer sandwiched between two
different electrode materials as depicted in Fig.3.1.
Solar photons can enter the cell through the trans-
parent contact (ITO) and create excitons upon ab-
sorption in the organic film. Not all photons will be
absorbed in the thin film - some will be reflected by
the back contact and have another chance to be ab-
sorbed in the film before they can leave the device
again through the ITO layer.
Figure 3.1: Device architecture of a single layer so-
lar cell as it is used throughout this thesis. The
organic film (dye, pigment, polymer) is sandwiched
between the transparent ITO covered quartz sub-
strate and a sublimed Al layer. The ITO and Al
electrodes are sometimes replaced by other mate-
rials in the cited literature. The thickness of the
organic layer is typically between 40 and 200nm.
3.1.1 Optical Properties of Organic
Photovoltaic Materials
Table 3.1 gives insight into the absorption proper-
ties of some important organic semiconductors. The
(half peak) absorption coefficient α
1/2
are listed to-
gether with the wavelength range that exceeds this
value to give a “feel” for the spectral dependence of
the absorption.
We have also calculated the percentage of ab-
sorbed photons α
10
using α
1/2
for a 10nm thin
1
film. Note that the percentage of absorbed photons
in a 10nm film in the cited range would require in-
tegration of the absorption over the given range.
The quoted number for α
10
is therefore somewhat
underestimated
2
if it wants to be interpreted as ab-
sorption in the given range.
PPV being the only precursor polymer in the list
has clearly the highest absorption coefficient fol-
lowed by its well known soluble derivative MEH-
PPV that has to accommodate also the bulky and
optically non active side-chains in the film. Inter-
estingly the absorption coefficient of MEH-PPV is
still twice as large as the other soluble polymers in-
dicating it can arrange its aromatic system in a very
compact fashion.
The metal-free phthalocyanine has virtually the
same α
1/2
regardless if eight bulky thio-alkyl
chains are attached (HPc(dye)) or nothing at all
(HPc(pig)) as in the insoluble version. The pery-
lene diimide pigments (Per(pig)), however, is some-
what stronger absorbing when compared with the
soluble derivative (Per(dye)). This is actually ex-
pected because the absence of bulky side chains in
the pigment should favour denser packing. In any
case, it appears as if the differences between absorp-
tion properties of dyes and pigments do not differ
tremendously.
On the other hand, the introduction of a metal
like Cu in the phthalocyanine core can lead to an
increase of the absorption. Pigments also tend to
have slightly broader absorption bands.
In Chapter 2 we have discussed this in context
with the exciton diffusion range and the importance
to reach dissociation sites. In single layer devices
1
10nm were chosen since this is the typical thickness of
the active layer i.e. exciton diffusion length - see Section 2.1.
2
For example: 50% of α
1/2
would have to be added if the
absorption spectrum is approximated as being triangular.
31
32 CHAPTER 3. SINGLE LAYER DEVICES
Table 3.1: Survey of the absorption properties of
various organic semiconductors. α
1/2
is half the
value of the absorption peak in the visible. The
range parameter refers to the part of the absorp-
tion band that exceeds this value. α
10
is the per-
centage of absorbed photons using α
1/2
in a 10nm
film without and with (brackets) a fully reflective
rear contact.
material α
1/2
range α
10
– nm
−1
nm %
PPV 0.017 320-480 16(29)
MEH-PPV 0.013 440-560 12(23)
CuPc(pig) [43] 0.012 570-740 11(21)
Per(pig) [133] 0.010 450-590 10(18)
Per(dye) 0.007 450-560 7(13)
CN-PPV 0.006 410-550 6(11)
MCP 0.006 370-510 6(11)
HPc(dye) 0.004 610-750 4(8)
HPc(pig) [68] 0.004 590-??? 4(8)
such a dissociation site is typically the low work-
function electrode (Al)/semiconductor interface
3
.
Only if the organic material can be regarded as
good electron acceptor material, the high workfunc-
tion electrode (ITO)/semiconductor interface can
become active.
The absorbed fraction A as a function of layer
thickness d in Fig.3.2 and Fig.3.3 have been derived
using Lambert Beer’s law to express the fraction of
transmitted light T yielding:
A = 1 −
T
. .. .
e
−αdr
(3.1)
with α being the absorption coefficient (here
α
1/2
) and r=1 for the transparent rear contact or
r=2 for the fully reflective one.
It is helpful to know that the reciprocal of α gives
the layer thickness that is necessary to absorb 63%
(1-1/e) light. The layer can be half as thin if a
reflective rear contact is used - compare with Fig.3.2
and Fig.3.3.
However, interference and multiple reflections
have been neglected above. We note that interfer-
ence effects may be more important to consider in
double layer, blend or laminated device structures.
In these structures the active layer can either be
3
Interestingly oxygen molecules that can also act as dis-
sociation site do not seem to increase the active layer signif-
icantly - see Chapter 2 and end of this Section.
Figure 3.2: Percentage of absorbed photons in a
film for various absorption coefficients 0.005 < α <
0.04nm
−1
as a function of film thickness.
placed in an optimal distance from the electrode
using a second layer as optical spacer material [213]
or the active layer can be made thick enough to al-
low a significant fraction of the visible light-wave to
enter the device.
Figure 3.3: Percentage of absorbed photons in a
film for various absorption coefficients 0.005 < α <
0.04nm
−1
as a function of film thickness assuming
a reflective rear contact.
Since the active layer is usually located next to
the reflective electrode interface and not much more
than 10nm thick with a refractive index of about 2,
interference with the reflected waves is not expected
to cause dramatic changes of the exciton density in
a 50nm single layer structure. More information on
interference effects that may be used to optimise
reflection
4
and absorption properties in solar cells
- particularly in multi-layer structures (considering
4
Anti-reflection coatings like the SiO
2
, Ta
2
O
5
or MgF
2
layer in silicon solar cells [162] may also be employed for
organic cells.
3.1. INTRODUCTION 33
also the about 100nm ITO film)- can be found in
Ref. [281, 282].
The inability to absorb more photons in the 10nm
layer represents a main loss mechanism in single
and double layer organic solar cells. According to
Fig.3.2 and Fig.3.3 even with the best absorbing
materials (α
1/2
≈ 0.02nm
−1
- see Table 3.1) and
reflective rear contacts, 70-80% of the photons in
the absorption range are lost for the photovoltaic
conversion process if the active layer is only 10nm.
We note that 80% of the solar radiation on earth
is distributed over wavelengths ranging from 400
to 1200nm - see Table 8.3. Indeed, most currently
available organic semiconductors loose more than
60% of the (mainly) low energy solar photons be-
cause their bandgap is too large - independently of
the thickness of their active layer i.e. exciton dif-
fusion range. Thus, the development of new low
bandgap materials is highly desirable.
Fortunately, organic semiconductors have very
high absorption coefficients compared to their inor-
ganic counterparts so that the extremely thin active
layer of 10nm can still absorb more than 20-30% of
the light in the absorption range.
For example, the peak absorption coefficient of
pure silicon at room temperature is 0.0004nm
−1
at
600nm
5
. Thus, a 10nm Si layer would only absorb
0.8% of light at 600nm - even with a reflective rear
contact.
3.1.2 Performance of Single Layer
Devices
A survey of the photovoltaic performance of various
single layer materials is given in Table 3.2. Blends
with a non conjugated polymer binder molecules are
treated as single layer here. PEDOT in conjunc-
tion with PPV here is heavily doped and therefore
treated as a conductive transparent electrode film
rather than a second semiconductive layer.
It can be seen that EQE values of 62 and 73%
have been achieved with voltages of 1.1 and 0.45V
many years ago. However, these cells have still some
way to go to become practically interesting since
e.g. illumination through the poorly transparent
indium electrode was required or sub-linear inten-
sity dependence of the photocurrent decreases the
actually achieved solar power conversion efficiency
down to 2.5% [164]. The recently published doped
pentacene cell in Ref. [224] shows a solar power
conversion efficiency of 2.4%.
5
In fact, α
Si
varies over 3 orders of magnitude ranging
from 0.00001nm
−1
at 1000nm to 0.01nm
−1
at 400nm [98].
As can be seen from Table 3.2 cells using pig-
ments have reached the highest EQE, V
oc
and fill
factors. However, although they have been investi-
gated intensely in the past three decades and they
have replaced many inorganic photo-receptors in xe-
rograpy - they could not quite conquer the field of
commercial solar cells.
Even though some can now be processed - with
some effort from (acidic) solution [283] - or dis-
persed in binder materials [164] and thereby cir-
cumvent thermal sublimation, film formation is still
considerably simpler, cheaper and less harmful us-
ing dyes, polymers or liquid crystals which are sol-
uble in common and less harmful solvents. More
information on coating techniques - especially for
pigments can be found on page 295 in Ref.[29] and
in Ref. [268] as well as references therein.
3.1.3 Doping of Organic Photovol-
taic Materials
It is well known that molecular substances can
have a ”doping” effect on organic semiconductors
[268, 231]. It has been found - and we can confirm
this from our own experience - that often exposure
to ambient air of only a few seconds is required
6
to
obtain a photoresponse in hole transport materials
i.e. good electron donors
7
. This relates to mate-
rials like PPV, MEH-PPV, PTV, phthalocyanines,
merocyanines and porphyrins.
Note that in phthalocyanines, for example, it is
found that the influence of the central metal ions on
the photovoltaic performance strictly follows their
oxygen binding ability - since oxygen acts here as
dopant. It was found that the photovoltaic effi-
ciency of metal containing phthalocyanines follows
the following order [231]:
PcMg > PcBe > PcFe > PcCu > PcCo (3.2)
In other words: the easier they bind oxygen the
higher the photocurrent.
Dopants with even greater electronegativity than
oxygen, such as halogens, NO
2
, o-chloranil or 2,4,7-
trinitrofluorenone result in an even greater enhance-
ment of the photoresponse[43] in such hole trans-
porters.
6
Typically there are marked concentration effects: Ex-
posure over many hours might be required before the pho-
tocurrent reaches its maximum which is then followed by
slow decrease.
7
Good electron donors have a high lying LUMO = small
electron affinity.
34 CHAPTER 3. SINGLE LAYER DEVICES
Table 3.2: Survey of important performance parameters of various organic single layer solar cells. The
abbreviation for materials are listed in Chapter 1. Best parameters parameters are printed bold while
italic print marks devices discussed later in this Chapter.
EQE Voc FF range device comments year
% V % nm – –
73 0.45 33 400-800(?) In/HPc+polymer/NESA pigment in polymer [43] 1981
62 1.1 33 400-800(?) Al/HPc+polymer/SnO2 pigment in polymer [43] 1979
?? ?? ?? 400-800(?) doped pentacene pigment [224] 2000
33 1.2 25 400-800(?) Al/merocyanine/Ag pigment [43] 1978
14 0.7 39 400-800(?) Al/CuPc/Au pigment [43] 1981
3 0.6 34 400-800(?) Al/merocyanine/Au pigment [43] 1978
1 1.2 20 400-500 PPV polymer [168] 1994
0.7 0.32 25 400-700(?) Cr/chlorophyll-a/Hg dye [43] 1975
0.5 1.0 23 400-500 Al/PPV+PEDOT/ITO polymer [8] 1998
0.15 0.85 25 400-800(?) Al/MgPc/Au pigment [43] 1974
0.1 0.7 25 400-750 Al/THPF/ITO polymer 2000
0.1 0.1 25 400-560 Al/LPPPT/ITO polymer 2000
0.09 0.5 27 400-570 Al/CN-MEH-PPV/ITO polymer [114] 1997
0.08 0.19 25 400-600 Al/Per/ITO dye 2000
0.07 0.7 25 400-570 Al/MEH-PPV/ITO polymer [196] 2000
0.05 0.37 25 400-750 Al/PTV/ITO polymer 2000
0.03 < 0.02 25 400-650 Al/P3HT/ITO polymer [63] 2001(?)
0.015 0.1 25 400-650
+
In/HPc/Au pigment [68] 1978
0.007 0.15 25 400-820 Al/HPc/ITO liquid crystal 2000
0.005 < 0.02 25 400-1000 Al/PIF/ITO polymer 2000
On the other hand, organic electron transporters
can be doped effectively with H
2
, NH
3
or phenoth-
iazine.
In general, the photocurrent is much more af-
fected by the presence of dopants than the dark cur-
rent. Although exciton dissociation should there-
fore occur throughout the whole film, optical filter
effects that indicate a thick electronically passive
layer are often found. We consider therefore that
excitons can dissociate at e.g. oxygen molecules
and increase the concentration of one type of free
charge carrier - while the other one is trapped in
the dopant within the inactive layer. This would
locally improve the charge transport properties but
does not add to the photocurrent.
A high photocurrent requires both the generation
of a free electron in the CB and a free hole in the
VB. In that way they can both meet after passing
through the external circuit. Consequently, molec-
ular doping can be seen as essentially different from
D/A blend (or laminated) structures since the lat-
ter allow both charge carriers to be free to travel
to the electrodes through their percolation paths.
Only then, the entire layer - and not only a layer
as thin as the exciton diffusion length (≈ 10nm) -
can become active. For more information on oxygen
doping see Chapter 2.
Since the presence of oxygen appears to enhance
the EQE in single layer devices we have exposed all
our single layer device samples to air for at least
3 days. Hence, different concentrations of water &
oxygen - due to different air exposure times of our
samples - should not have a significant effect on
our results. We note that the adsorption of water
molecules also affects the photovoltaic performance
[50].
3.1.4 Desired Properties of Organic
Photovoltaic Materials
A very interesting review of photovoltaic, fluores-
cence and redox properties of 60 semiconducting
molecules - most of which are (soluble) merocyanine
derivatives - is given in Ref. [204]. These devices
were structured like Al
8
/dye/Ag and reached quan-
tum efficiencies of up to 16%. In the following we
list the desired properties for photovoltaic molecules
as they have been identified in Ref. [204]. The list
is followed and complemented by comments found
in other studies.
1. The chromophoric
9
core of the dye must be es-
sentially planar and free form steric crowding
8
A AlO
2
O
3
layer on top of Al was considered.
9
Chromophoric core = the pi conjugated part of the
molecule is responsible for the optical properties in the visi-
ble and near IR range.
3.1. INTRODUCTION 35
effects which tend to twist the dye, predispos-
ing it to photo-isomerisation or internal conver-
sion, instead of sensitisation by electron injec-
tion. Small groups projecting out of the plane
of the chromophore
10
play a critical role in “in-
dexing” the packing of adjacent molecules to
obtain the desired aggregate structure.
2. The dye must be a stable material and a strong
absorber.
3. The performance of photovoltaic dyes can be
correlated with their chemical parameters. In
the configuration of Al/Al
2
O
3
/dye/Ag, the ad-
hesion of the dye to the Al surface is criti-
cal and can be analyzed by studying the chro-
matography of the dyes on the Al substrates.
The electron rich end of the dye molecule
should be chemisorbed at the Al surface. For
example, dyes with COOH groups have a much
stronger bond to the Al surface and result al-
most always in much higher photovoltaic quan-
tum yields.
4. The reduction potentials of the dues should be
more negative than -1.3V vs. SCE for efficient
electron injection through the Al
2
O
3
barrier.
5. Dyes should not be subject to protonation or
deprotonation by ambient materials in the PH-
range from 3 to 10.
6. Dyes having central conjugated chains and
steric crowding are to be avoided since such
materials can readily degrade excitons to heat
by torsional internal conversion. High PL of
the isolated molecules in e.g. fluid solutions
signal the effect of less torsional stress and is
therefore preferable.
7. The quadrupolar charge distribution of e.g.
squaric acid dyes may pack into a “brick-
stone” i.e. J-aggregate structure which seems
to favour mobility of low energy excitons. Me-
rocyanines on the other hand pack in deck of
cards or stacks showing only up to 15% pho-
tovoltaic quantum efficiency in their main ab-
sorption band but up to 100%
11
on the short
wavelength edge of the absorption band.
10
Chromophore = coloured molecule
11
Photovoltaic quantum efficiency numbers in Ref. [204]
consider transmission and reflection losses and are more
likely identical with the internal quantum efficiency (IQE)
- which gives higher numbers - rather than the usually used
EQE. This is discussed in Chapter 2.
8. Low bandgap materials seem to have difficul-
ties to maintain a large negative reduction po-
tential i.e. a low lying LUMO level. This would
mean that low bandgap electron acceptors are
harder to find i.e. synthesise.
We are not aware that any of these early find-
ings have been proven wrong but many have been
confirmed, complemented and further elaborated as
can be seen from the following list of comments:
ad 1): The finding that a flat, rigid
12
, aromatic
molecule favors close packing and good photovoltaic
properties - in particular high exciton mobility - has
been recognized in other publications [268, 267, 154,
245].
The importance of this point may become obvi-
ous by referring to the fact that C.Tang actually
included a claim in his patent regarding the use of
flat large aromatic molecules as photo-active mate-
rials in solar cells. In Ref. [154] it is stated that the
inter-planar distance should be short (≈ 3.5
˚
A) be-
tween photo-conductive molecules to achieve good
performance. We note that close packing also in-
creases the absorption coefficient. However, we also
want to emphasise that exciton mobility looses its
importance in e.g. blend structures - so that even
spheric molecules like C60 can act as excellent elec-
tron acceptor and transport materials.
ad 3): We note that COOH groups play indeed an
important role in bonding sensitiser molecules like
the ruthenium dye in electrochemical photovoltaic
cells to the TiO
2
surface [182].
ad 6): While PL in solution provides information
about the isolated molecule and its ability to dissi-
pate excitation energy through e.g. intra-molecular
vibrations, PL studies of organic films can provide
further insight into a material’s photovoltaic prop-
erties.
High PL in the solid state has been found to be
desirable since it means that the life time of exci-
tons is relatively long and allows excitons to diffuse
over longer distances. If there is little or no PL, this
would indicate that other - faster - decay channels
for excitons are present. These fast decay channels
reduce the time an exciton has to reach a dissocia-
tion site.
The effect of PL efficiency in the solid state on
the reverse saturation current and the theoretically
12
Rigid molecules are preferred to flexible structures since
they have lower probability for those internal conversions
that dissipate the energy carrying species before transport
can occur.
36 CHAPTER 3. SINGLE LAYER DEVICES
achievable open circuit voltage is discussed in more
detail in Chapter 2 and Ref. [234].
ad 8): If low bandgap materials tend to have
higher lying LUMO levels, then they are more suit-
able as electron donor (D) materials rather than
electron acceptors (A) - when used in combination
with another semiconducting component to form a
D/A system.
In fact the combination phthalocyanine & pery-
lene representing a system with a low bandgap elec-
tron donor and a high bandgap electron acceptor
works very well. The other possibility using a low
band gap electron acceptor with high bandgap elec-
tron donor gives indeed only very low photocurrents
(e.g. HPc+PIF in Chapter 5 or MEH-PPV+PIF
[202]).
Figure 3.4: Chemical structure of the polymer used
in this Chapter. It is ladder poly(para-phenylene
thienylene) or short: LPPPT. The bridge between
the phenylene-rings enforce a planar structure de-
spite the proximity of the thiophene and phenylene
units. R=-1,4-C6H4-n-C10H21 and n=20-25.
Hence, the conclusion is that a low bandgap ma-
terial in a D/A solar cell is likely to act as electron
donor and thus require a partner with high elec-
tron affinity to enable charge transfer i.e. exciton
splitting at their interface.
3.1.5 Electrodes
The workfunction and adsorption properties of elec-
trodes have a clear effect on the performance of or-
ganic solar cells. The latter has to do with band
bending due to equilibration of the workfunctions
with the Fermi-levels of the semiconductors upon
contact - as discussed in Chapter 2.
In the following Sections of this Chapter we have
always used the same combination of electrodes
(ITO and Al) to decrease the number of changing
parameters. Moreover, both electrodes have been
proven to give more reliable and reproducible re-
sults than most other materials under laboratory
conditions.
However, higher open circuit voltages and some-
times higher photocurrents can be obtained using
electrodes like e.g. Au and Ca instead of ITO and
Al.
3.2 LPPPT
It has been found that a poly(para-phenylene)-type
ladder polymer exhibits many interesting properties
such as high PL not only in solution (90%) but also
in the film (50%) enabling the fabrication of efficient
organic multi-colour LEDs [246] and even optically
pumped blue solid state lasers [279].
The unique performance of this polymer is likely
related to a very low concentration of electronically
active traps such as structural defects or impuri-
ties [96] and the high charge carrier mobility (up to
10
−3
cm
2
/V s for holes at room-temperature) [128].
However, the bandgap of this polymer with 2.7eV
is too high to absorb a large fraction of the solar
spectrum.
A well established method to shift the absorp-
tion onset towards lower energies is the replacement
of benzene by electron rich hetero-arylene moieties
such as 2,5 thienylene units.
For example the bandgap of PPV could be low-
ered by about 0.5eV by replacing every benzene unit
with a 2,5 thienylene unit to give poly-thienylene
vinylene (PTV) [275, 137]. The reason for this ef-
fect is the lower aromatic resonance energy of the
thiophene building block when compared to that of
benzene. This results in a reduced π − π

transi-
tion energy between the benzoid ground state and
the quinoid excited state.
Indeed by replacing every other phenylene unit
with a thiophene ring the structure of LPPPT - see
Fig.3.4 - the low bandgap version of LPPP could
be obtained. The bandgap in solution was lowered
by 0.45eV compared to LPPP - which means a red-
shift by about 85nm.
Both polymers exhibit a dominating 0-0 tran-
sition and an extraordinary small Stoke’s shift of
35meV (LPPP) and 17meV (LPPPT) [74]. The lat-
ter (small stokes shift) is a consequence of the geo-
metrically fixed, planar structure which allows only
minimal changes of the polymer geometry during
the transition from the ground to the excited state
i.e. photoexcitation [160].
We note that planarity of aromatic systems is
also suspected to favour charge and energy (exci-
3.2. LPPPT 37
ton) transport as discussed in Chapter 2 and below.
The well resolved vibronic structure allows better
attribution of EQE features to absorption features
- which we were exploiting by studying thickness
effects in this special material.
Experimental
The chemical structure of the polymer used in this
Section is shown in Fig.3.4. LPPPT has been pro-
vided by and synthesised at the MPI-Mainz. De-
tails of the synthesis can be found in Ref [74].
LPPPT films were prepared by spin coating from
chloroform solution onto ITO covered quartz sub-
strates (12x12mm) after filtering (0.45µm). In or-
der to obtain a wide range of different film thick-
nesses, solutions with concentrations of 10, 15, 20
and 30 mg/ml have been prepared and spincoated
at 1600 and 3200rpm.
More details on the preparation of substrates and
the EQE/IV measurements can be found in Chapter
10.
Figure 3.5: Features of the EQE spectrum of a thin
(33nm) LPPPT device do almost coincide with the
absorption features. The very thick (205nm) device
shows not only smaller absolute numbers but also a
distinct offset to the absorption i.e. strong antibatic
behaviour.
Results and Discussion
Absorption and EQE Spectra
Fig.3.5 shows the EQE spectrum of a thick and a
very thin LPPPT device when illuminated through
the ITO contact as well as the linear absorption
curve. The absorption spectrum shows two domi-
nant peaks at 535nm (2.32eV) and 495nm (2.51eV)
and a shoulder at 460nm (2.70eV). The energetic
spacing of 190eV is constant and can be attributed
to the dominant vibronic (C-C) mode that has
been observed in the absorption of films of other
higher ordered polymers such as stretch oriented
PPV [108] or the methyl substituted ladder type
polymer comprising only benzene rings [197, 198].
The suppression of the rotational degree of free-
dom between adjacent benzene rings in the latter re-
sults in a particularly small Stoke’s shift, high photo
- and electro-luminescence quantum yields but also
in a clearly resolved vibronic structure in both PL
and absorption [238, 239, 197].
The EQE spectra of both the thick and the thin
device show roughly the same features (the vibronic
peaks 0-0 and 0-1); however, there are some differ-
ences:
• The EQE of the thinner device is almost “in
phase” with the absorption while the thick one
shows a clear red shift by 20nm.
• The EQE of the thicker device increases while
The EQE of the thin device decreases in the
shorter wavelength region.
• The second EQE peak (at 500nm) is virtually
as high as the first which is not the case for the
thick device.
• The EQE of the thinner device reaches values
that are about 5 times larger than those for the
thick one.
The peaks in the photoresponse of the thick cell
occur not exactly at wavelengths with absorption
maxima. In fact the second photoresponse maxi-
mum coincides with the absorption minimum be-
tween the two dominant vibronic peaks. This has
been observed in organic solar cells before and is
sometimes termed as antibatic behaviour [120] - see
also Chapter 2.
The peak shift
13
can be quantitatively explained
with the so called optical filter effect[120]- see Chap-
ter 2.
Only if the bulk is thin enough, most of the disso-
ciated excitons can generate a photocurrent so that
the EQE essentially follows the absorption features.
This seems to be approached for the 30nm thin de-
vice since there is only a small peak offset left. In
the thick device only light which is absorbed less -
for example in the onset or the minimum between
the vibronic peaks in the absorption spectrum - can
13
but not the shape of the EQE peak
38 CHAPTER 3. SINGLE LAYER DEVICES
create excitons in the active region so that absorp-
tion maxima actually turn out as EQE minima and
vice versa.
Assuming the filter effect
14
is the main factor that
determines the offset between absorption and EQE
peaks we can estimate the maximum thickness of
the organic film d
opt.
that can be used in single
layer devices without losses due to the optical fil-
ter effect. First we need to know the wavelength
λ
e
of the EQE peak in the onset of the antibatic
absorption spectrum. The optical density (OD) at
this position together with the OD at the absorp-
tion peak (Fig.3.5) and the film thickness d
0
of the
measured EQE spectrum can then be taken to get
and estimate for the optimal thickness d
opt.
using:
d
opt.
=
OD(λ
e
)
OD(λ
0
)
· d
0
(3.3)
This equation is solely based on Lambert Beer’s
Law:
I(λ)
I
0
(λ)
= e
−α(λ)d
= 10
−OD(λ)
(3.4)
(with I
o
being the incident and I the transmitted
intensity) and the resulting simple relation:
d ∝ OD (3.5)
The error of the optimal thickness ∆d
opt.
due to
the error of the input variables is:

∆dopt
dopt

2
=

∆OD(λ
e
)
OD(λe)

2
+

∆OD(λ
0
)
OD(λ
0
)

2
+
+

∆d
0
d
0

2
(3.6)
Using the data for the 205nm thick device and es-
timated errors (OD(λ
e
) = 0.0696±45%, OD(λ
0
) =
0.22 ±2% and d
0
= 205nm±5%) we obtain:
d
opt
= 65 ±30nm (3.7)
The large error of 30nm or 45% for OD(λ
e
) is
due to the steep slope in the absorption onset and
the relatively broad peak in the EQE spectrum of
the thick device.
We obtain a considerably smaller error and value
for d
opt
using the data for the thinnest investigated
device (10nm): OD(λ
e
) = 0.201 ±12%, OD(λ
0
) =
0.22 ±2% and d
0
= 33nm±15%. Here we obtain:
d
opt
= 30 ±6nm(20%) (3.8)
14
See Section 2.4.4
This is a more reasonable result considering that
we know already that d
opt
has to be (slightly)
smaller than 33nm. Given that we are trying to
avoid the optical filter effect we can justify to use
the lowest number within the error bars and calcu-
late the mean in case of several results. This would
yield in our case:
d
opt
= (36 + 24)/2 = 30nm (3.9)
Figure 3.6: The EQE of LPPPT at 536nm when
illuminated through ITO decreases according to
the layer thickness of 33,59,67,93,127,117 (dashed
line),163,168 and 205nm.
Generally, d
opt.
may be regarded as an upper limit
of the exciton diffusion range for this material. The
actual value might be smaller if e.g. there is an
effect of other dissociation sites such as impurities
or dopants
15
in addition to the electrode-polymer
interface.
We also need to consider the high electric field
across the thin layer due to the difference in
workfunctions of the electrodes (0.5V/30nm≈ 2 ·
10
5
V cm
−2
). Hence, we cannot be sure that the ex-
citon diffusion length is as long as 30nm but at least
we know that the absorbed light is most efficiently
used in a device with about 30nm thickness.
On the other hand, a relatively large diffusion
length in LPPPT is not surprising knowing that pla-
nar structures are expected to favour exciton trans-
port. In any case, the large value for d
opt.
compared
to about 10nm for MEH-PPV devices makes this
material an interesting candidate for use in D/A
cells.
15
oxygen can act as such in electron donors
3.2. LPPPT 39
Figure 3.7: EQE versus film thickness when illu-
minated (λ=536nm) through ITO (circles) and Al
contact (squares). The solid line in the circles is a
fit with a polynom of 3rd order whereas the solid
line in the squares is a linear fit. The dashed line
illustrates a possible trend.
How the shape of the EQE spectra varies with
layer thickness is shown in more detail in Fig.3.6.
The antibatic behaviour of the 205nm thick device
gradually changes into a more and more symbatic
response by going to thinner cells. This also holds
true for the short wavelength range (400-500nm).
Only the 117nm thick device seems to behave dif-
ferently.
Optical interference effects between the incident
and the reflected beam from the Al rear contact
may be super-imposed here. In fact, numerical sim-
ulation using the described model and taking into
account interference effects showed that there is in-
deed a dip around this wavelength in the thickness
dependence curve of the EQE for such a device
[266].
We plotted the experimental data of the EQE
at 536nm versus thickness in Fig.3.7. The figure
clearly shows the trend to higher EQE for thinner
devices as well as the dip at 117nm.
The figure also shows the curve for EQE at 536nm
obtained upon illumination through the semitrans-
parent Al contact taking into account the transmis-
sion losses through the Al layer (Fig.3.8). We ob-
tain then even higher values than with illumination
through ITO. This is expected since there should
be no losses due to optical filtering
16
.
16
The slight increase of the EQE for thinner devices is
probably due to improved electric properties such as a
Figure 3.8: Optical transmission spectrum of the
15nm thick Al layer i.e. electrode of the LPPPT
devices discussed in this Chapter.
Hence, this curve displays the maximum EQE
that is achievable for a single layer device of this
material using ITO/Al electrodes regardless of film
thickness or which side of the cell is illuminated.
Thus, we can expect that the strong increase
of EQE with thinner devices when illuminated
through ITO will slow down considerably when
reaching the trend line from the Al illumination
curve. The thickness at this point can be assumed
to coincide with optimal layer thickness d
opt
. As
expected in the described picture, the shape, peak
positions and absolute value of the EQE do not vary
considerably when illuminated through Al
17
- see
Fig.3.9.
IV Characteristics
The IV characteristics in the dark and under
monochromatic illumination (550nm) of the thickest
device (205nm) is shown in Fig.3.10. Even though
the open circuit voltage is relatively small (106mV)
this device shows a good rectification i.e. diode be-
haviour: The current that can pass through the de-
vice at 0.4V is about 15 times (=rectification ratio)
higher than at -0.4V.
Dark and light characteristics have virtually the
same shape i.e. lie on top of each other at this
scale. The parameter Id should not be confused
with the reverse saturation current I
o
which can
only be obtained via numerical modeling of the IV
curves. Id here is the actual current measured at
smaller series resistant.
17
The small variation seen in this figure may well be due
to random errors in the measurement.
40 CHAPTER 3. SINGLE LAYER DEVICES
Figure 3.9: EQE of the LPPPT films when illu-
minated through the semitransparent Al contact
(not corrected for the Al transmission) for differ-
ent thickness (same data as in Fig.3.7-full circles).
Shape and ratio of the features do not vary signif-
icantly with film thickness. The highest EQE was
obtained with a 93nm thick film.
0V
18
.
With decreasing film thickness - see Fig.3.11 to
Fig.3.16, the rectification ratio also decreases un-
til the IV characteristics of the thinnest device -
see Fig.3.17 resembles more the characteristic of an
ohmic resistor (constant slope) with virtually no
rectification and only a small (≈ 30mV) open cir-
cuit voltage. The steep slope reveals a significantly
higher conductivity (under illumination and in the
dark) in comparison to the thick device.
A survey of parameters obtained from the IV
18
For the suspected origin of Id see Chapter 10
Table 3.3: Survey of solar cell parameters obtained
from the IV curves and the EQE (at 536nm) for
LPPPT films of varying thickness.
d EQE V
oc
I
d
R
sh
R
s
nm 10
−3
% mV pA MΩ MΩ
32 132 28 33 16(17) 4.8(5.0)
59 85 105 31 83(200) 7.5(11)
67 77 126 31 150(530) 6.3(9.5)
93 64 141 26 170(700) 7.5(10)
117 36 176 18 250(1400) –
122 52 120 38 200(500) 6.3(8.8)
185 26 99 27 250(600) 7.5(11)
205 15 106 10 – –
Figure 3.10: IV characteristics of the 205nm thick
LPPPT device.
curves with different film thickness is given in Table
3.3. Although there might be a trend towards lower
dark currents with increasing thickness and increas-
ing V
oc
it is not completely conclusive. However, a
correlation between dark current and V
oc
seems to
emerge though if the data are plotted on a suitably
chosen scale - see Fig.3.18. The conductivity in the
dark seems to be high enough to reduce the open
circuit voltage.
Larger open circuit voltages appear to require a
shunt higher than about 80MΩ here. However, the
table also shows that V
oc
does not necessarily in-
crease for even higher shunts.
A very interesting result is that R
s
does increase
relatively slowly (11kΩ/nm) with film thickness
whereas the shunt increases more than 100 times
faster (1.5MΩ/nm) - see Fig.3.19. While the in-
crease of R
s
can be qualitatively explained by the
fact that charge carriers have simply more semicon-
ducting material to travel through to reach the elec-
trodes, the distinct growth of the shunt is harder to
understand.
In Chapter 10 we have stated that recombination
losses somewhere between the dissociation site and
the electrodes can be associated with a small R
s
.
However, it is not obvious why the absolute num-
ber of recombinations should go up if the charges
travel through less material. It would be easier to
accept that the number of recombinations per sec-
ond increased since the charges need more time to
get out of the thicker semiconductor.
Fig.3.20 shows that the shunt also increases for
thicker films when measured in the dark - although
there seems to be some consolidation for d>60nm.
3.2. LPPPT 41
Figure 3.11: IV characteristics of the 185nm thick
LPPPT device.
The thickness dependence of R
s
appears unchanged
in the dark with absolute numbers about 30% larger
than under illumination.
R
s
is expected to decrease significantly with in-
creasing light intensity simply because more charge
carriers become available for the conduction pro-
cess. A considerable decrease of R
s
is required to
allow both high voltages as well as high photocur-
rents and thus high power conversion efficiencies
[206, 214].
In Chapter 2 we have seen that the ratio R
s
/R
sh
not only affects the fillfactor strongly but also the
output current in a solar cell. According to Eq.
(2.19) the current through a solar cell is reduced by
the factor β:
β = 1 +R
s
/R
sh
(3.10)
In Fig.3.21 we have plotted β versus film thick-
ness. The figure shows that the factor β is close
to 1 for film thicknesses larger than about 60nm.
For thinner films the reduction factor can increase
from 1.04 (60-200nm) to at least 1.3 (30nm) in light.
Although this does not affect the photocurrent dra-
matically it indicates that there is an additional ef-
fect for devices thinner than 60nm in LPPPT.
In the preceding Section we have already seen
that with decreasing thickness the EQE is indeed
rising.
In order to see the effect of film thickness on the
shape of the IV curves, we have plotted the light IV
curves in a single graph - see Fig.3.22. The graph
illustrates clearly the distinct changes of slopes and
open circuit voltages.
Figure 3.12: IV characteristics of the 122nm thick
LPPPT device.
Figure 3.13: IV characteristics of the 117nm thick
LPPPT device.
The increasing R
sh
can be recognized immedi-
ately from the decreasing slope in the 3 quadrant
(negative voltage range here) although there are ex-
ceptions in the expected order.
As we have seen in Fig.3.18 before the V
oc
reaches
a maximum value for the 117nm thin layer and not
for the thickest device as it may be expected from
the flat slope in Fig.3.22.
In fact, V
oc
of the 32nm device is as small as
28mV whereas the highest voltage in LPPPT is
176mV (117nm). Such a large difference cannot
be explained with the effect of the R
s
/R
sh
- which
essentially works as a voltage divider. According
to Eq. (2.24) the value for R
s
/R
sh
= 0.3 for this
device can only account for a reduction of V
oc
by
about 30%. Hence, apart from a low shunt, there
42 CHAPTER 3. SINGLE LAYER DEVICES
Figure 3.14: IV characteristics of the 93nm thick
LPPPT device.
Figure 3.15: IV characteristics of the 67nm thick
LPPPT device.
must be another not yet known reason to cause the
low open circuit voltage in this device.
Fig.3.23 shows again the dark and light character-
istics of the ITO/LPPPT/Al device (59nm thick).
It also shows the fit with Eq. (2.19) that could be
applied for the region -0.1V to 0.4V using the fol-
lowing fit parameters:
1. ideality factor n=1.10
2. reverse saturation current I
0
= 10pA
3. resistors in the dark: R
sh
= 18 · 10
7
Ω, R
s
=
2.3 · 10
7

4. resistors under illumination (0.1mW/cm
2
):
R
sh
= 8.8 · 10
7
Ω, R
s
= 1.8 · 10
7
Ω and I
L
=
1610pA
Figure 3.16: IV characteristics of the 59nm thick
LPPPT device.
Figure 3.17: IV characteristics of the 32nm thick
LPPPT device.
Note that the resistor values shown in this list
are somewhat different from the values shown in
Table 3.3. In the latter the numbers were derived
directly by taking the slopes of the characteristics
at 0V (R
sh
) and the highest voltages around 0.5V
(R
s
) whereas in the former the formalism consid-
ered effects of n, I
0
, I
L
. The results for the shunt
are fairly similar and the values for R
s
are about 4
times higher in the fit.
We consider the numbers for R
s
values which are
derived from the slope around 0.5V to be more ac-
curate than the value obtained from fitting the en-
tire curve since R
s
really starts to dominate only at
higher voltages
19
.
In any case, the comparison of these two methods
indicates that the error for R
s
is much higher than
the one for R
sh
. This is because of both the steeper
slope at higher voltages which results in stronger
19
Unless the IV curve becomes clearly linear for higher
voltages the (inverse) slope can only represent a upper limit
of Rs - see Chapter 2.
3.2. LPPPT 43
Figure 3.18: Open circuit voltage and dark current
seem to be “out of phase” when plotted versus film
thickness on the shown scales.
dependence and voltage fluctuations of the source
measure unit in the low mV range.
According to the discussion on the photolumines-
cence effects we can now take the reverse saturation
current together with the PL efficiency of LPPPT
of about 1.2% [35, 63] to obtain a value for the PL
independent component of the reverse saturation
current according to Eq. (2.25):
I
01
= I
0
· Φ
PL
= 12pA· 0.012 = 0.12pA (3.11)
The value for I
01
is close to the 0.091pA obtained
for the later discussed MEH-PPV. Thus, it can not
be ruled out here that I
0
depends only or at least
predominantly on the PL efficiency. This would
then indicate that the maximum performance of
both materials LPPPT and MEH-PPV showing the
used PL efficiency has been closely approached in
these devices.
The higher EQE numbers of thinner devices
would possibly disappear if the optical losses like
the filter effect were considered or - in other words
- the internal quantum efficiency (IQE) was calcu-
lated.
A more detailed study comprising a higher num-
ber of samples and systematic IV analysis is nec-
essary to confirm this possibility and explain the
observed variation of V
oc
with thickness.
Summary
In this Section we have found for LPPPT that:
Figure 3.19: Series and shunt resistor as obtained
from the slopes in the IV curves under illumination
for varying film thickness.
• Thickness is an important parameter. The op-
timal thickness can be estimated from the ab-
sorption spectrum and an antibatic EQE spec-
trum.
• The EQE increases from 0.029% for the thick-
est (205)devices to 0.135% for the thinnest
(32nm) device.
• The highest EQE is almost twice as high as the
best i.e. thinnest (10nm) MEH-PPV device.
• The highest EQEs were achieved for the
thinnest cells where the optical filter effect has
virtually disappeared. Since the entire 30nm
layer contributes to the photocurrent - the ex-
citon diffusion length could be as high as 30nm
in this material - if the polymer/Al interface
has to be reached.
• Since LPPPT is a very large flat molecule but
without any other obvious outstanding struc-
tural properties the high EQE in this mate-
rial and the lack of an optical filter effect sup-
ports the suspicion that planar rigid structures
favour exciton diffusion.
• Although the EQE increases with thinner de-
vices without reaching a maximum in the ob-
served range (30-200nm) the V
oc
appears to
reach a maximum value at 117nm.
• Both shunt and series resistor increase with
increasing film thickness but R
sh
grows more
44 CHAPTER 3. SINGLE LAYER DEVICES
Figure 3.20: Series and shunt resistor as obtained
from the slopes in the IV curves in the dark and
under illumination for varying film thickness.
than 100 times faster (1.5MΩ/nm) than R
s
un-
der illumination (λ = 536nm, 0.1W/cm
2
) for
an area of 4mm
2
.
• The PL efficiency is rather low (1.2%).
• The PL efficiency multiplied with the reverse
saturation current (12pA) gives approximately
the same number for LPPPT (0.12pA) as for
MEH-PPV (0.091pA). This is consistent with
a strong correlation between the IV curve i.e
the photovoltaic performance and the PL as
discussed in Chapter 2 and Ref. [234].
• Films of LPPPT which are thinner than about
60nm show a strong increase of EQE, more sim-
ilar R
s
and R
sh
values, a fast decreasing V
oc
,
a fast decreasing R
sh
in the dark. Hence, the
physics in these thin devices seems to be domi-
nated by other effects which are not yet under-
stood.
3.3 MEH
After the discovery of EL in conjugated polymers
[34] the next step towards inexpensive device man-
ufacturing was the design of a molecule with similar
semi-conducting properties but even easier proces-
sibility.
At that time PPV films could only be obtained
via spincoating a precursor solution and a subse-
quent heat conversion step at about 200C for 5 to
Figure 3.21: Current reduction factor β as a func-
tion of film thickness in LPPPT - calculated using
the numbers under illumination.
10hrs
20
.
Already in 1991 researchers in Santa Barbara
(CA, USA) were successful at obtaining a PPV
derivative with branched saturated side chains
[124]. These bulky side chains allow solvent
molecules to penetrate into the space between the
polymer chains thus allowing device fabrication us-
ing simple spin-casting techniques ([32] and Refer-
ences therein). These side chains are also not con-
jugated so they do not alter the semiconducting or
electronic properties significantly.
This was the birth of MEH-PPV, probably the
worlds first soluble semiconducting polymer. Sin-
gle layer MEH-PPV cells have already been sub-
ject to photocurrent spectroscopy [120], impedance
and photo-voltage spectroscopy [121] as well as deep
level transient spectroscopy [92].
The availability of a soluble semiconductor not
only opened the way towards cheaper processing via
numerous coating methods - even inkjet printing
21
has already been reported [19] - but also made it
possible to investigate blends with other semicon-
ductors such as C60 while maintaining the advan-
tages of solution processing.
Photo-induced electron transfer from MEH-PPV
onto the fullerene C60 has been discovered and ex-
ploited in solar cells reaching EQE as high as 29%
for monochromatic light [278]. Shortly afterwards,
blends of MEH-PPV with its more electro-negative
derivative CN-PPV that also benefit from photo-
20
Recently, another deposition method which can be ap-
plied for PPV such as laser-ablation has been found and in-
vestigated [240].
21
This technique could be used to fabricate not only com-
puter or video screens but also photosensitive CCD arrays.
3.3. MEH 45
Figure 3.22: 3rd and 4th quadrant IV characteris-
tics of LPPPT devices under light. The numbers
represent the film thickness.
induced charge transfer with 6% monochromatic
EQE have been reported[112].
Here we investigate to what extent the thickness
(between 10 and 250nm) of the MEH-PPV layer
effects the photovoltaic properties.
Experimental
We fabricated ITO/MEH-PPV/Al photodiodes
with different thickness of the MEH-PPV layer
ranging between 10 and 250nm. MEH-PPV was
synthesised by H. Rost, at the Melville Laboratory
for Polymer Synthesis in Cambridge. information
on the synthesis can be found in Ref. [181, 124]
and references therein.
The MEH-PPV films were spincoated in air af-
ter filtering (5µm) from chloroform solution (1 - 6
mg/ml) onto ITO coated glass substrates (Balzers).
Unless stated otherwise all devices where measured
when illuminated through the ITO contact. We
measured the thickness of the thicker films (50 to
250nm) with a profilometer (Dektak) after 2 days
drying in air in darkness which is necessary to ob-
tain films hard enough to avoid the profilometer
needle to scratch the polymer considerably. To-
gether with the absorption data we calculated the
absorption coefficient which we used in order to con-
firm the thickness of the thinner films.
More details on sample preparation and the used
setup can be found in Chapter 10.
Figure 3.23: IV characteristics (solid lines) in the
dark and under illumination of a ITO/LPPPT/Al
device (59nm thick) and calculated curves (dashed
lines) using Eq. (2.19).
Results and Discussion
Absorption and EQE Spectra
Fig.3.25 shows the linear absorption of MEH-PPV
together with the EQE spectra for two different
thicknesses. The absorption spectrum reveals es-
sentially one broad feature with the peak at 505nm
(2.46eV). The onset of absorption is not very sharp,
starting at 600nm which corresponds to a bandgap
of 2.07eV.
According to Table 8.3 a sufficiently thick MEH-
PPV film can absorb about 28% of the solar radia-
tion. Note that the bandgap has been significantly
reduced with respect to the insoluble PPV.
For the EQE spectra of these two devices we
found that:
• The thickest (250nm) device shows an EQE
peak in the onset of the absorption spectrum
while the EQE peak coincides with the absorp-
tion peak for the thinnest (10nm) cell.
• The EQE of the thin cell increases again for
shorter wavelengths after reaching its maxi-
mum at the absorption peak whereas the op-
tical density continues to fall. Such behaviour
is unexpected for an otherwise symbatic EQE
spectrum and indeed in contrast to the results
in e.g. LPPPT where the 30nm device keeps
following the absorption.
46 CHAPTER 3. SINGLE LAYER DEVICES
Figure 3.24: Chemical structure of the polymer
used in this Chapter. It is a poly(2-methoxy-5-(2-
ethyl-hexyloxy)-1,4-phenylene vinylene - referred to
as MEH-PPV.
• The EQE peak of the 10nm device is almost 5
times as high as for the 250nm cell.
The EQE values for thicknesses ranging from 10
to 250nm with some intermediate values are plotted
in Fig.3.26. All films show an increasing EQE in
the short wavelength range - although the slope is
flatter for medium thickness (35, 54nm) but higher
again for the 10nm film.
Apart from the raising EQE of the 10nm film,
all spectra show very similar features and trends as
observed in the LPPPT film - see previous Section.
The EQE increases clearly with thinner films.
However, if we apply Eq. (3.3) to estimate the
thickness of the active layer from the antibatic EQE
of the 35nm cell, we obtain:
d
opt
= 26 ±7nm (3.12)
which is 17nm higher than the value found for
PPV [113, 111]. The large error stems from the
uncertainty of the peak position of the 35nm film
in conjunction with the steep slope of OD which
results in a relative error of λ
e
= 22%. Note that
the 54nm film actually has its peak closer to the
absorption peak than the 35nm film. We consider
this mainly to be due to the low resolution of the
EQE scan (10nm) but also random errors related to
the photocurrent measuring process.
A plot of the EQE versus film thickness taken
at λ =505nm shows how the efficiency increases by
nearly a factor 50 from 0.0014% for a thickness of
250nm to 0.064% for 10nm - see Fig.3.27.
The fact that the efficiency is highest in a layer
as thin as 10nm is particularly interesting in the
Figure 3.25: EQE spectrum of a thick (250nm) and
very thin (10nm) MEH-PPV cell. Also shown is the
absorption of a 38nm MEH-PPV film.
view of other discoveries: For example, it has been
shown that non radiative energy transfer - in other
words - the transfer of an exciton from an excited
molecule to the metal electrode can generate surface
plasmons. The energy transfer onto these plasmons
represents an effective decay channel particularly if
the excitons are close (≈20nm) to the metal film
[40]. More recently, effective photo- and electro-
luminescence quenching has been found in organic
films near (<20nm) metal electrodes [14, 15, 103].
Both these earlier findings would let us expect
lower EQEs in such thin films since the lost excitons
are believed to be transferred onto the metal where
their energy dissipates into heat rather than being
split into their constituent charges. Hence, the high
EQE in our thin devices may be affected by the
generation of surface plasmons but the gain of a
smaller filter effect and higher field due to the small
distance between the two electrodes seems to over-
compensate those quenching effects.
3.3. MEH 47
Figure 3.26: EQE spectra for MEH-PPV devices
with thicknesses ranging from 10 to 250nm.
Figure 3.27: EQE versus film thicknesses for the
MEH-PPV devices ranging from 10 to 250nm. The
dashed line is a guide to the eye.
Figure 3.28: IV light (solid line) and dark (dashed
line)characteristics of a 250nm thick MEH-PPV de-
vice.
Figure 3.29: IV light (solid line) and dark (dashed
line)characteristics of a 210nm thick MEH-PPV de-
vice.
48 CHAPTER 3. SINGLE LAYER DEVICES
Figure 3.30: IV light (solid line) and dark (dashed
line)characteristics of a 54nm thick MEH-PPV de-
vice.
Figure 3.31: IV light (solid line) and dark (dashed
line)characteristics of a 38nm thick MEH-PPV de-
vice.
Figure 3.32: IV light (solid line) and dark (dashed
line)characteristics of a 35nm thick MEH-PPV de-
vice.
Figure 3.33: IV light (solid line) and dark (dashed
line)characteristics of a 10nm thick MEH-PPV de-
vice.
3.3. MEH 49
Table 3.4: Survey of various solar cell parame-
ters obtained from the IV curves and the EQE (at
536nm) for MEH-PPV films of varying thickness.
d EQE V
oc
I
d
R
sh
R
s
nm % mV pA MΩ MΩ
10 0.064 650 31 200(2400) 7.5(8.0)
35 0.039 730 14 500(11000) 40(74)
38 0.047 730 16 400(12000) 14(23)
54 0.039 700 9 380(18000) 40(270)
210 0.004 500 6 1400(21000) 100(170)
250 0.001 420 7 2400(24000) 54(220)
IV Characteristics
Figure 3.34: IV characteristics under illumination
for different film thicknesses in MEH-PPV. The
numbers represent the thickness in nm.
Fig.3.29 to Fig.3.33 show the IV characteristics
for various film thicknesses in darkness and under il-
lumination (through ITO) with λ =550nm at about
0.2mWcm
−2
. The dark current I
d
is the actually
measured current at 0V in darkness and should
not be confused with the reverse saturation current
which can only be obtained by fitting the IV curve.
For more information on the dark current which is
an experimental artifact see Chapter 10.
In contrast to the findings in LPPPT films the
IV characteristics in the dark for MEH-PPV do not
change their shape much for different thicknesses in
the negative voltage range (3rd quadrant). Apart
from the thinnest device the IV curves for positive
voltages (1 quadrant)under illumination reveal two
different slopes. The slope at lower voltages is al-
ways steeper than the other. The latter also de-
creases with thickness while the other remains more
or less constant. The changes of the shape of the
IV curves under illumination can be assessed more
easily in Fig.3.34.
Figure 3.35: V
oc
and I
d
in MEH-PPV versus film
thickness. The lines are guides to the eye.
Table 3.4 gives a survey of important cell param-
eters obtained from these IV curves plus the EQE
values at λ =505nm. The R
sh
and R
s
values were
taken from the inverse slopes around 0V and the
steeper slope in the 1st quadrant. A rough compar-
ison with the table for LPPPT reveals that both
the shunts (R
sh
) and R
s
in MEH-PPV are gener-
ally about an order of magnitude higher and V
oc
by about a factor 6. The dark currents I
d
are only
slightly smaller than for LPPPT. However, the EQE
is reduced by a factor of 2-3 with respect to the lad-
der type polymer.
These results can be understood if we consider
the equivalent circuit diagram ECD - as discussed in
Chapter 2 qualitatively: Higher R
s
results in lower
photocurrents and a large shunt favours higher V
oc
.
The slightly smaller dark currents I
d
are also con-
sistent with the higher V
oc
in MEH-PPV.
In Fig.3.35 we have plotted V
oc
and dark current
versus film thickness. As in LPPPT there seems
to be an opposite trend between V
oc
and I
d
: For
similar film thickness the V
oc
is smaller if I
d
is higher
and vice versa.
The series and shunt resistors are plotted ver-
sus film thickness in Fig.3.36. As in the LPPPT
films we find that the shunt increases much faster
50 CHAPTER 3. SINGLE LAYER DEVICES
Figure 3.36: Series and shunt resistor versus film
thickness in MEH-PPV cells. The lines are linear
fits.
(8MΩ/nm) than the series resistor (250kΩ/nm).
Considering the average size of R
s
≈ 40MΩ in
MEH-PPV the increase for a device with e.g. poly-
mer film thickness of 100nm is 25MΩ which is about
60% is relatively small.
Very interesting and somewhat surprising is that
R
sh
increases very quickly (350MΩ/nm) and virtu-
ally linearly up to a thickness of at least 54nm. The
shunts in light and darkness are plotted in Fig.3.38.
However, for larger thicknesses we find consoli-
dation at around 23000MΩ. We note that the dif-
ferences in dark currents around 0V for the thicker
films become less than 1pA which is close to the
resolution limit of the source measure unit. Thus
the shunts for these two thick films do contain a
large relative error and can therefore only give and
indication
22
.
In Chapter 2 we have seen that the ratio R
s
/R
sh
does not only affect the fill factor but also the out-
put current of a solar cell. According to Eq. (2.19)
the current through a solar cell is reduced by the
factor β
β = 1 +R
s
/R
sh
(3.13)
In Fig.3.37 we have plotted β versus film thick-
ness. The figure shows that - unlike the situation
in LPPPT where we found a strong increase for
d< 60nm - β in MEH-PPV is not much different
from 1 for thicknesses ranging from 10 to 250nm.
Hence, the R
s
/R
sh
ratio is small enough so that
22
Similar considerations could apply for LPPPT and thick-
ness values larger than 120nm.
Figure 3.37: Reduction factor β versus film thick-
ness in MEH-PPV under illumination (λ = 550nm)
and in the dark.
reasonable fill factors should be possible for higher
light intensities.
Figure 3.38: Shunt resistor of MEH-PPV devices in
the dark and under illumination versus film thick-
ness. The solid lines are linear fits while the dotted
line represents a possible saturation.
Fig.3.39 shows again the dark and light character-
istics of the ITO/LPPPT/Al device (59nm thick).
It also shows the fit with Eq. (2.19) that could be
applied for the region -0.4V to 1.2V using the fol-
lowing fit parameters:
1. ideality factor: n=4.54
3.3. MEH 51
Figure 3.39: Experimental IV curves (solid lines)
in the dark and under illumination of a ITO/MEH-
PPV/Al device (38nm thick) and fits (dashed lines)
using Eq. (2.19).
2. reverse saturation current: I
0
= 0.7pA (light),
I
0
= 0.1pA (dark)
3. resistors in the dark: R
sh
= 1130MΩ, R
s
=
1.3MΩ
4. resistors under illumination: (0.2mW/cm
2
):
R
sh
= 400MΩ, R
s
= 2.0MΩ and I
L
= 1720pA
Note that the resistor values shown in this list
are somewhat different from the values shown in
Table 3.4. In the latter the numbers were derived
directly by taking the slopes of the characteristics
at 0V (R
sh
) and the highest voltages around 0.9V
(R
s
) whereas in the former the formalism consid-
ered effects of n, I
0
, I
L
. The results for the shunt
are very similar but the values for R
s
are about 10
times larger in the fit.
We consider the numbers for R
s
values which are
derived from the slope around 0.9V to be more ac-
curate since R
s
really starts to dominate only at
higher voltages
23
.
Note that in contrast to the fit for the LPPPT
device we had to choose two different values for I
0
in MEH-PPV to obtain a reasonable fit.
According to the discussion on the photolumines-
cence effects in Chapter 2 we can now take the re-
verse saturation current I
0
together with the PL ef-
23
Unless the IV curve becomes clearly linear for higher
voltages the (inverse) slope can only represent a upper limit
of Rs - see Chapter 2.
ficiency of MEH-PPV of about 13% [63]) we obtain
a value for the PL independent component of the
reverse saturation current according to Eq. (2.25) -
of:
I
01
= I
0
· Φ
PL
= 0.7pA· 0.13 = 0.091pA (3.14)
The value for I
01
is close to the 0.12pA obtained
for LPPPT. Thus, it can not be ruled out here that
I
0
depends only or at least predominantly on the
PL efficiency. This would then indicate that the
maximum performance of both materials LPPPT
and MEH-PPV showing the used PL efficiency has
been closely approached in these devices.
A more detailed study comprising a higher num-
ber of samples and systematic IV analysis is nec-
essary to confirm this possibility and explain the
observed variation of V
oc
with thickness.
Summary
In this Section we have found for MEH-PPV that:
• Thickness is an important parameter. The op-
timal thickness can be estimated roughly from
the absorption spectrum and an antibatic EQE
spectrum.
• The EQE increases from 0.0014% for the thick-
est (250nm) devices to 0.064% for the thinnest
(10nm) device.
• The highest EQE efficiencies were achieved for
the thinnest cells where the optical filter ef-
fect has virtually disappeared. Since the entire
10nm layer contributes to the photocurrent -
the exciton diffusion length can be at least as
high as 10nm in this material - if the poly-
mer/Al interface has to be reached.
The maximum active layer was estimated to
be around 26±7nm which is an estimate for
the upper limit for the exciton diffusion length
in MEH-PPV.
• For similar thicknesses devices with lower dark
current I
d
shows higher open circuit voltages.
• Both resistors increase with increasing film
thickness, but R
sh
grows more than 30times
faster (8.0MΩ/nm) than R
s
(250kΩ/nm) un-
der illumination (λ = 550nm, 0.2W/cm
2
) for
an area of 4mm
2
.
• The PL efficiency multiplied with the reverse
saturation current (0.7pA) gives approximately
52 CHAPTER 3. SINGLE LAYER DEVICES
the same number for MEH-PPV (0.091pA) as
for LPPPT (0.12pA). This is consistent with
a strong correlation between the IV curve i.e
the photovoltaic performance and the PL as
discussed in Chapter 2 and Ref. [234].
The following observations are in contrast to the
findings for LPPPT:
• V
oc
in MEH-PPV is generally about 6 times
higher and V
oc
also tends to increase with thin-
ner films.
• The dark current in MEH-PPV is virtually
voltage independent for negative voltages and
the thickness range 30 to 250nm. Only the
10nm device shows clear dependence for (neg-
ative) bias larger than 0.6V.
• R
s
is on average about 15 times smaller than
R
sh
so that the output current reduction factor
β is not very different from 1 for all thicknesses.
• The EQE of the thinnest MEH-PPV device fol-
lows the main absorption band but increases
again in the blue, while the absorption still de-
creases.
3.4 PTV
With an EQE of about 1% and V
oc
of 1.2V, PPV has
shown some of the highest solar cell parameters for
polymers in single layer devices. Unfortunately, the
high bandgap (≈ 2.4eV ) does not allow sufficient
absorption of photons in the visible range. This is
not only an obstacle for the application as photo-
detector material but also for the use in solar cells.
Figure 3.40: Chemical structure of Poly[3,4-bis-
(S)-(2-methylbutylthio) thienylenevinylene], PTV.
Here n ≈ 120 (Mn)
With a bandgap of 2.4eV, only about 15% of
the solar light on earth could be absorbed in a
very thick film of PTV - see Table 8.3. The in-
troduction of alkoxy substituents could lower the
bandgap by about 0.4eV and lead to good solu-
bility in many solvents. One of these derivatives
(MEH-PPV) has been investigated in the previous
Section and showed broader wavelength sensitivity
but otherwise poorer photovoltaic properties than
PPV.
In this Section we investigate a molecule where
the benzene ring of MEH-PPV is replaced by a thio-
phene ring and the alkoxy by the stronger electron
donating thioalkyl chain. Note that replacement of
carbon atoms with sulfur in a conjugated systems
generally decreases the bandgap. Thus we obtain a
polythiophene (PT) derivative with ethylene link-
ages PTV - see Fig.3.40.
If we compare PTV with poly-thiophene deriva-
tives, we find that the introduction of ethylene link-
ages between the thiophene rings should lead to a
more rigid structure with less rotational disorder
and lower aromatic character
24
and therefore higher
electron delocalisation. Hence the bandgap should
be even lower than in many PTs.
Moreover, the side-chains are chiral and can
therefore induce interesting optical properties such
as the ability to turn the plane of electromagnetic
waves (optical activity) by means of aggregation of
the main chains into chiral superstructures. This
has already been observed for other semiconduct-
ing polymers [156].
Experimental
PTV has been synthesised via the McMurry reduc-
tive polymerisation by Francesca Goldoni, Univer-
sity of Technology (Eindhoven, The Netherlands).
Details of the synthesis, solvatochromic effects and
circular dichroism (a bisignated sigal has been ob-
served for this polymer) can be found in Ref.[90, 91].
The molecular weight of the polymer is high but
also shows strong polydispersity (Mn = 38600, Mw
= 146680). The number of repeat units can there-
fore be calculated as 120 (Mn) and 470 (Mw).
The polymer was spincoated from chloroform
solution (15mg/ml)onto pre-etched ITO coated
quartz substrates with 2000rpm in a laminar flow
box.
Absorption and photocurrent measurements were
performed in air. More details on the preparation of
substrates, devices and the EQE/IV measurements
can be found in Chapter 10.
24
Directly linked aromatic rings tend to twist
3.4. PTV 53
Results and Discussion
Absorption and EQE Spectra
Fig.3.41 shows the EQE spectrum of an ITO/PTV
(70nm)/Al device when illuminated through the
ITO contact as well as the linear absorption curve.
The absorption spectrum shows features of the vi-
bronic structure - similar to the ones observed in
solution[90]. The low bandgap suggests well or-
dered regions with extended conjugation along the
polymeric backbone. The onset of absorption is at
around 720nm which corresponds to a bandgap of
1.73eV. According to Fig.8.3 a sufficient thick layer
of a material with such a bandgap can absorb about
42% of the solar light (ranging from 300 to 720nm).
Figure 3.41: The EQE of a 90nm thick PTV de-
vice is very low in the main absorption region but
reaches values similar to MEH-PPV in the blue
range.
The EQE spectrum as shown in Fig.3.41 does not
reveal any response in the region of the main ab-
sorption band (not even at the absorption onset)
but reaches its maximum EQE at around 0.07%
around 400nm. This number is very similar to peak
EQEs of good single layer materials such as MEH-
PPV (see Ref.[196]) or other soluble polythiophenes
[202] but clearly lower than the 1% found in PPV
[168]. Such a behaviour, neither a symbatic re-
sponse nor a clear sign of an optical filter effect is
rather unusual.
A possible explanation may have to consider the
special optical properties (optical activity) due to a
chiral superstructure
25
or simply be the result of a
25
A bisignated circular dichroism spectrum was found for
aggregates in solution [90]
combination of two extrinsic charge generation pro-
cesses such as a photoactive Al/polymer interface
and the presence of exciton dissociation sites such
as oxygen distributed within the bulk.
Figure 3.42: The semi logarithmic plot of the EQE
spectrum of the same device as in Fig.3.41reveals
that the photoresponse starts in fact with the onset
of light absorption and increases nearly by a factor
of 10 every 100nm. We found that EQE ∝ λ
−9
fits
very well for λ > 480nm
The EQE plotted on a logarithmic scale -
Fig.3.42- reveals a nearly single exponential increase
of the EQE starting from the absorption onset at
720nm. The photoresponse covers the full visible
range with virtually unequivocally correlated EQE
values distributed over 4 orders of magnitude. The
photocurrent increases by about an order of mag-
nitude when illuminated with e.g. 500nm instead
of 600nm. As a consequence, this device can be
used as a colour or wavelength detector for a wave-
length range as wide as 400 to 800nm (the full visi-
ble range). For comparison: Tada et al. had to use
the photoresponse of a thick and a thin layer of a
poly thiophene derivative (Poly(3-hexylthiophene)
to obtain a colour sensor with similar sensitivity
but much narrower spectral range (550 and 630nm)
[243].
IV Characteristics
The IV characteristics in the dark and under
monochromatic illumination (420nm) is shown in
Fig.3.43. Even under the low illumination intensity
(0.05mW/cm
2
) the device develops an open circuit
voltage of 370mV with a fill factor of ≈ 30%.
Another interesting result is that despite the rea-
54 CHAPTER 3. SINGLE LAYER DEVICES
Figure 3.43: IV characteristics of the PTV device in
the dark (dashed line) and when illuminated with
λ =420nm (solid line).
sonable V
oc
and relatively good fill factor, this de-
vice shows no strong rectification ratio since there
is no clear off state current. It looks more as if a
counter-diode has been formed with a even lower
breakthrough voltage than in forward bias. This
has probably to do with the low bandgap that is
more likely to lead to low interface barriers for
both forward and reverse bias. From the slopes
in Fig.3.43 we have extracted the following resistor
values:
• light: R
s
= 200MΩ and R
sh
=3000MΩ
• dark: R
s
= 200MΩ and R
sh
=6400MΩ
The series resistor is about 10 times higher than
in MEH-PPV or LPPPT - possibly because of the
lower light intensity here. A study of R
s
as a func-
tion of light intensity would be desirable as part of a
future project. The smaller difference between the
shunt in dark and light compared to LPPPT and
MEH-PPV may also be due to the lower illumina-
tion intensity here.
We note that the R
s
values taken from the first
(around 0.45V) or the third quadrant (counter-
diode, around -0.25V) are virtually identical.
Fig.3.44 shows a double logarithmic plot of the
same IV characteristics as in Fig.3.43. There are
two different slopes fitting the dark characteristics
in the positive bias voltage range: For voltages up to
Figure 3.44: A double logarithmic plot of the dark
characteristics. Ohm’s law seems to fit for lower
voltages while the I ∝ V
4
relation may be due to
charges released from deep and/or shallow traps.
0.2V, current and voltage obey Ohm’s law of direct
proportionality.
For voltages larger than 0.2V up to 1V we find
an I ∝ V
4
relationship. According to the Mott-
Gurney equation [138, 231] for trap free space
charge limited current the current through the
diode obeys:
I ∝
V
d
(3.15)
for low voltages and
I ∝
V
2
d
3
(3.16)
for high voltages
However, the presence of deep and/or shallow
traps can lead to an even stronger dependence in
a limited range of V before returning to obey the
trap free equations above. The range from 0.2 to 1
V in Fig.3.44 may display the trap dominated re-
gion.
Summary
In this Section we have found that:
• A semiconducting polymer with a bandgap
as small as 1.76eV can be used to built a
ITO/polymer/Al cell.
3.5. PIF 55
• The EQE is very low at long wavelengths but
very sensitive to the illumination wavelength
and obeys EQE ∝ λ
1/9
between 480 to 720nm.
This enables direct determination of the wave-
length of incident light from the photocurrent
using the spectral response curve. Thus, this
device can be used as colour sensor covering
the full visible range.
• The cell shows a relatively good fill factor for
a single layer polymer cell of about 30% with
an open circuit voltage of 370mV
• The device shows low breakthrough voltages
for both negative and positive voltages. Thus
it can be used to amplify positive and negative
voltages.
• The EQE reaches 0.07% at 400nm.
• The EQE spectrum is neither antibatic nor
symbatic - possibly due to the optical activity
in conjunction with a chiral superstructure
• Resistor values derived from the IV curves
at 0.05mW/cm
2
are: R
s
= 200MΩ and
R
sh
=3000MΩ and in the dark: R
s
= 200MΩ
and R
sh
=6400MΩ
3.5 PIF
Up to now, all polymers that have been discussed
in this Chapter can be used as electron donors be-
cause of their small ionisation energy i.e. high lying
HOMO level. Fig.3.45 shows the chemical structure
of a polymer (PIF) that has the potential to be both
a low bandgap absorber as well as an electron ac-
ceptor with respect to many other polymers.
The short distance between the fairly large re-
peat units causes a twisting angle of 33
o
[210]. The
distinct twist between the monomer units and the
short but bulky side chains allow for plenty of space
between the molecules so that it dissolves readily in
many common solvents.
Despite the distortion of the conjugated backbone
this molecule can still have a very low bandgap
thanks to the large π system of the single repeat
units. The quinoid character was expected to lead
to relatively high electron affinity which would clas-
sify this material as an electron acceptor with re-
gard to the majority of other conjugated polymers.
In this Section we will discuss the absorption and
photovoltaic properties of ITO/PIF/Al devices.
Figure 3.45: Chemical structure of poly indeno fluo-
rene, referred to as PIF. The quinoid character, the
large π - system and the bulky side chains make this
material a candidate for a low bandgap electron ac-
ceptor. n = 20-25.
Experimental
PIF has been synthesised by H. Reisch and U.
Wiesler at the MPI in Mainz, Germany. Details
about the synthesis can be found in Ref. [210].
The polymer was spincoated from chloroform solu-
tion (8mg/ml) onto pre-etched ITO coated quartz
substrates at 2000rpm in a laminar flow box. Ab-
sorption and photocurrent measurements were per-
formed in air. More details on the preparation of
substrates, devices and the EQE/IV measurements
can be found in Chapter 10. The thickness of the
organic layer was about 30nm.
Results and Discussion
Absorption and EQE Spectra
Fig.3.46 shows the absorption spectrum of PIF to-
gether with the EQE of an ITO/PIF/Al device
(20nm thick). The absorption peak is at 800nm
whereas the onset of absorption is around 1100nm
which corresponds to a semiconductor bandgap of
1.13eV. A sufficiently thick device of a semiconduc-
tor with this material can absorb 77% of the direct
solar radiation on earth - see Fig.8.3.
Note that the bandgap is virtually identical to the
one of crystalline silicon. Thus, this might be the
56 CHAPTER 3. SINGLE LAYER DEVICES
Figure 3.46: EQE and absorption spectrum of a
PIF device when illuminated through ITO and mea-
sured in air. The whole bulk is active since we
observe a symbatic behaviour. The photoresponse
covers the same range as commercial silicon pho-
todetectors (400-1100nm)
first photodiode based on a soluble semiconductor
that covers the same spectral range as commercial
silicon photodiodes. Organic semiconductors with
very low bandgaps often lack photo-stability.
However, with this material we did not observe
any signs of photo-degradation before or during the
measurements where the polymer was exposed to
light intensities of up to 0.2mW/cm
2
for several
minutes in air. Although these intensities are still
relatively low, the device was working and could
serve, for example, as photo-detector at such inten-
sities.
The EQE spectrum closely follows the absorption
indicating that the entire bulk contributes to the
photocurrent. The absolute EQE numbers (0.003%
at 800nm), however, are about an order of magni-
tude lower than in e.g. MEH-PPV.
A semi logarithmic plot - see Fig.3.47 - confirms
that the photoresponse coincides very well not only
with the peaks but also with the onset of absorp-
tion. The EQE is again higher for the shorter wave-
lengths. We want to note here that the error of the
EQE spectrum between 400 and 500nm can be as
high as ±50%. This is because the signal to noise
ratio is very low due to the low emission intensity
of the QTH-lamp and the low EQE of PIF in this
region.
Figure 3.47: The semi logarithmic plot of the same
data as in Fig.3.46 shows that absorption and EQE
are very much “in phase” even at the onset at
1100nm
IV Characteristics
The dark and light characteristics of the same PIF
device are shown in Fig.3.48. Despite a significant
I
sc
, there is neither a significant V
oc
nor a reason-
able rectification ratio. Both the dark and the light
characteristics appear almost symmetrical and on
top of each other at the scale shown.
Note that the current through this device is about
10 times higher than in MEH-PPV so that the slope
around 0V is actually very steep i.e. the material
is very conductive already without applying a volt-
age. Thus, the relatively few photogenerated charge
carriers cannot change the shape of the dark char-
acteristic significantly.
From the slopes in Fig.3.49 we have extracted the
following resistor values:
• light: R
s
= 1.7MΩ,R
s
= 5.7MΩ (negative volt-
age range) and R
sh
=66MΩ
• dark: R
s
= 1.7MΩ,R
s
= 5.7MΩ (negative volt-
age range) and R
sh
=83MΩ
Note that according to the discussion in Chapter
2 all R
s
values obtained from the slope in the IV
curve can only represent upper limits. The values
are considerably smaller than in e.g. MEH-PPV
which may indicate either partial shorts due to pin-
holes or high conductivity due to dopants or rather
low barriers in the pathway of the charges. The lack
of a reasonable V
oc
is most likely related to the low
shunt resistor.
3.6. THPF 57
Figure 3.48: IV characteristics of the PIF device
(32nm thick). Dark and light curve coincide on this
scale. The device was illuminated at λ=800nm
The extraordinary low bandgap among today’s
polymers (1.13eV) may well lead to low elec-
trode/polymer - interface barriers for both injection
and extraction of charges and therefore lead to the
formation of a counter-diode as in the low bandgap
polymer PTV. Such a steep slope with a low V
oc
seems to occur more often in electron transport
materials particularly if they have a low bandgap
[150, 64].
In Fig.3.49 we plotted the dark curve of Fig.3.48
on a double logarithmic scale. The IV curve follows
the Mott-Gurney equation (Child’s Law) for trap
free materials (I ∝ V
2
) between 0.1 and 1V.
Summary
In this Section we have found that:
• The conjugated polymer PIF with a bandgap
as low as 1.13eV can be used to manufacture
a photovoltaic cell with a photoresponse cover-
ing the same range as a commercial crystalline
silicon based solar cells.
• The EQE of this material reaches 0.003% at its
absorption peak (800nm)
• There is only little rectification and a negligible
V
oc
(smaller 20meV) - most likely due to the
low shunt resistor.
• Resistor values derived from the IV curves
at 0.2mW/cm
2
and λ = 800nm are: R
s
=
Figure 3.49: The IV curve (filled circles) fits the
Mott and Gurney equation (Child s law) for trap
free materials (I ∝ V
2
, solid line) and shows no
ohmic behaviour as expected for such low voltages
(I ∝ V , dashed line). There are also no clear
signs for deep or shallow traps. The steep slope i.e.
strong voltage dependence of the current around 0V
seems to occur more often in electron transport ma-
terials.
1.7MΩ,R
s
= 5.7MΩ (negative voltage range)
and R
sh
=66MΩ and in the dark: R
s
=
1.7MΩ,R
s
= 5.7MΩ (negative voltage range)
and R
sh
=83MΩ
• Child’s law for trap free materials fits the IV
characteristics between 0.1 and 1V
3.6 THPF
In this Section we are going to investigate a polymer
with a molecular structure which has a number of
interesting peculiarities - see Fig.3.50: A dialkoxy
substituted phenylene ring is coupled via a triple
bond to a thiophene with a pyrazine fused to it.
Moreover, two fluoro substituted phenylene rings
extend the the conjugated system further.
The triple bond does not break the conjugation
in these poly-ynes
26
; in fact, extended conjugation
has been found in both organic and organo-metallic
poly-ynes [142]. The large and electron rich π sys-
tem should be reasonably flat and therefore result
in a particular low bandgap - see below.
26
The term -yne denotes the triple bond.
58 CHAPTER 3. SINGLE LAYER DEVICES
Figure 3.50: Chemical structure of the polymer
THPF used in this Chapter. n ≈ 22
In addition, the introduction of the fluoro substi-
tuted phenylene could increase the electron affinity
significantly and make it - together with the low
bandgap - an interesting candidate as electron ac-
ceptor in D/A solar cells. However, the spatial sep-
aration from the thieno-pyrazine via the phenylene
could weaken their effect to some extent.
From the comparison with a derivative that lacks
the fluor atoms, we know that their effect on e.g.
the bandgap is negligible - which is consistent with
the findings for MEH-PPV and its cyano substi-
tuted analogue
27
.
In any case, low bandgap acceptor materials are
particularly promising for the use in D/A solar
cells but still very hard to find among polymeric
semiconductors. More information on organic and
organo-metallic poly-ynes can be found in Refs.
[141, 146, 147, 184].
Experimental
THPF has been provided and synthesised by
M. Khan, University Chemical Laboratory (Cam-
bridge, UK). Details of the synthesis can be found in
Ref. [143]. The polymer was spincoated from chlo-
roform solution (16mg/ml) onto pre-etched ITO
coated quartz substrates with 2000rpm in a lami-
nar flow box.
Absorption and photocurrent measurements were
performed in air. More details on substrate, device
preparation and the EQE/IV measurements can be
found in Chapter 10.
27
Cyano (CN) groups are like fluor atoms strong electron
withdrawing and therefore often used to increase the electron
affinity of an organic compound.
Results and Discussion
Absorption and EQE Spectra
Fig.3.52 shows the linear absorption spectrum to-
gether with the EQE spectrum of a 110nm thick
ITO/THPF/Al cell. The onset of absorption is at
720nm (1.72eV). A sufficiently thick layer with this
bandgap can absorb about 48% of the direct solar
radiation on earth - see Fig.8.3. The absorption
spectrum shows a dominant peak at 620nm and a
shoulder at 680nm.
Figure 3.51: The EQE of an ITO/THPF/Al device
reaches 0.15% around 400nm. However, in the main
absorption range (500-700nm) the efficiency is only
0.003% which is an order of magnitude lower than
in analogous MEH-PPV cells.
The EQE exceeds values of 0.1% at 400nm which
is more than twice as much as typical MEH-PPV
devices. The shape as well as the absolute num-
bers with the considerably high peak EQE around
400nm resemble the EQE spectrum found in the
PTV device. Note that in both materials the EQE
in the main absorption band is considerably lower
than near 400nm.
Unlike the results in PTV devices, the semi log-
arithmic plot - see Fig.3.51 - shows no exponen-
tial dependence on the wavelength for THPF. The
plateau around 600nm makes an application as
colour sensor clearly more difficult. In addition, the
shoulder in the EQE curve at 700nm might indicate
the presence of the optical filter effect.
IV Characteristics
The IV characteristics of the same device is
shown in Fig.3.53. The IV curves show ohmic
3.6. THPF 59
Figure 3.52: A semi logarithmic plot of the EQE
spectrum in Fig.3.52 reveals that the EQE follows
the absorption - although some signs of an internal
filter effect are visible.
behaviour between -1 and +1V but develop
a reasonable open circuit voltage of 600mV
(at λ =552nm, 0.2mW/cm
2
) and 700mV (at
λ =440nm, 0.06mW/cm
2
). We can also see that
the EQE can be doubled by increasing the reverse
bias voltage by 1V.
Figure 3.53: IV dark (dashed line) and light charac-
teristics (solid lines - shown here for two excitation
wavelengths) of the THPF device. The behaviour
is ohmic.
From the slopes in Fig.3.53 we have extracted the
following resistor values:
• light: R
s
< 6300MΩ (λ =552nm),R
s
<
1400MΩ (λ =440nm) and R
sh
=6300MΩ
(λ =552nm), R
sh
=1400MΩ (λ =440nm)
• dark: R
s
< 17000MΩ and R
sh
=17000MΩ
Note that according to the discussion in Chapter
2 all R
s
values obtained from the slope in the IV
curve can only represent upper limits. This is par-
ticularly important here since the voltage range is
probably not large enough to exceed the onset of the
diode breakthrough voltage - so that both the shunt
and the series resistors are equal here. However, R
s
could be considerably smaller.
The values are somewhat higher than in e.g.
MEH-PPV which indicates the absence of signifi-
cant shorts through pinholes and thus account for
the good open circuit voltage.
Note that the shunt (and the upper limit for R
s
)
at the wavelength with the lower light intensity is al-
most 5 times smaller as is the short circuit current.
Considering that the OD for the two wavelengths
is roughly the same (0.51 at 440nm and 0.37 at
552nm) the higher conductivity at lower light in-
tensity is rather surprising.
However, the unusually big difference between
the EQE in the main absorption and around 400nm
suggests that there may be other wavelength depen-
dent effects dominating. This would require further
investigation preferably involving light intensity de-
pendent EQE/IV for different wavelengths in future
studies.
Summary
In this Section we have found for ITO/THPF/Al
devices:
• The EQE is relatively low (0.003%) in the main
absorption band (500-700nm) but reaches more
than 0.1% around 400nm.
• The photo-responsive range covers a broad
wavelength range (400 to 800nm).
• A relatively high open circuit voltage of up to
700mV.
• The EQE can be doubled by applying a reverse
bias voltage of 1V.
• Resistor values derived from the IV curves
at 0.2mW/cm
2
and λ = 552nm are: R
s
<
6300MΩ and R
sh
=6300MΩ and in the dark:
R
s
< 17000MΩ and R
sh
=17000MΩ
60 CHAPTER 3. SINGLE LAYER DEVICES
3.7 Per
The most frequently used perylene pigments are
diimides and bis imidazoles of perylene-3,4,9,10-
tetracarboxylic acid. They show good light and
water fastnesses and are best known for their use
as colourants for automobile paints and plastics.
Figure 3.54: Chemical structure of the soluble pery-
lene diimide (Per) used in this Section.
Also photo-conducting properties of perylenes
have been recognized as early as 1972 when Re-
gensburger and Jakubowski studied the xerogra-
phic behaviour of a number of perylenes in bi-layer
devices[218]. Their motivation to investigate these
molecules were likely driven by the fact that many
perylene diimides are strongly absorbing in nearly
the entire visible range and that they were avail-
able. We note that perylene bis-imidazole absorb
at longer wavelengths relative to those of perylene
diimide due to the extended π-conjugation[154].
Later it was discovered that charge separation in
perylenes primarily takes place at an interface with
a hole accepting molecule unless the exciton decays
non-radiatively to the ground state [154], [244].
Furthermore, it has been found that the photo-
sensitivity of perylenes is higher, the more planar
the molecule is [154].
The following formulation can be found in claim
17 of Chin Tang’s Patent on a multi-layer or-
ganic photovoltaic device which underlines the im-
portance of size
28
and planarity of photovoltaic
molecules in general - not only perylene derivatives
[245]:“...materials comprising a compound contain-
ing a generally planar polycyclic nucleus having a
surface area of at least about 40 square
˚
A and a
width of at least about 5
˚
A.” In this basic patent
he also mentions that electron-acceptor compounds
should have at least 7 aromatic rings and 8 for elec-
28
The actual perylene structure (4 aromatic rings fused
together around a central ring) measures about 7x6
˚
A
tron donors to avoid shorts due to pinholes in the
about 40 to 50nm thin films.
Following these guidelines, he managed to fabri-
cate one of the best organic solar cells until today
using a double layer comprising two large and flat
aromatic molecules such as a perylene bis-imidazole
together with a copper phthalocyanine [244].
Interestingly at least one device in the patent was
even better than in Ref. [244]: A NESATRON
29
/
diimide-perylene/Ag device with V
oc
=440mV, Ise =
3.0V, FF = 0.6 and a power conversion efficiency as
high as 1.0% under AM2 conditions. The particular
diimide perylene used in this cell had a phenyl ring
coupled to both nitrogen atoms.
However, in order to fabricate thin film solar cells
with pigments, energy intensive and expensive ther-
mal sublimation has to be employed since their sol-
ubility in common solvents is very low.
Only very recently (1998), chemists at Xerox
have found a method to process many hetero
cyclic pigment classes from solution to facilitate
cheaper manufacturing [283]. In this Lewis-acid
pigment solubilisation (LAPS) process, the pig-
ments are mixed with a solution of a Lewis-acid in
nitromethane. The solubilisation involves coordi-
nation of the acid with the pigment molecules’ het-
ero atoms to form a soluble electron donor-acceptor
complex which is later removed by washing. Unfor-
tunately this method does not work for all pigments
and the acid might attack the pigment and/or the
electrode material causing other complications.
In this Section we investigate a photovoltaic cell
using a perylene derivative (a dye) that allows for
simple processing via spincoating from solution.
Other soluble perylenes have been reported, most
of which also exhibit photoluminescence efficiencies
around 100% in solution [69].
We note that in the meanwhile also soluble pery-
lene containing polymers are available [217]. One
has already been employed successfully in LED
structures to produce white light by means of exci-
tation energy and charge transfer from a blue emit-
ting ladder type polymer in a blend [246].
Experimental
The perylene dye has been purchased from Synthec
(Germany). Details of the synthesis can be found in
Ref.[60]. The polymer was spin-coated from chlo-
roform solution (8mg/ml)
30
onto pre-etched ITO
29
NESATRON is a brandmark for a material with similar
properties as ITO - mainly used in the 1970s
30
According to Ref [60] up to 70mg(!) can be dissolved in
1ml.
3.7. PER 61
coated quartz substrates with 2000rpm in a lami-
nar flow box.
We obtained very tractile films of reasonable
thickness (about 100nm) - with no spin marks but
the surface appeared rough in contrast to the shiny
surface of e.g. MEH-PPV films. Absorption and
photocurrent measurements were performed in air.
More information on sample and device preparation
can be found in Chapter 10.
Results and Discussion
Absorption and EQE Spectra
Fig.3.55 shows the EQE spectrum of a 40nm thick
ITO/Per/Al cell when illuminated through the ITO
as well as the linear absorption curve. We find vir-
tually no antibatic behaviour so that the two main
features can be found in both spectra: A dominant
peak at 495nm and another peak at 540nm. In ad-
dition, the EQE reveals another local maximum at
370nm.
Figure 3.55: The EQE of this Per device follows the
absorption spectrum closely reaching numbers com-
parable to MEH-PPV but covering a larger spectral
range.
The lack of an optical filter effect upon illumi-
nation through ITO in this 40nm thick device is
consistent with the assumption that perylene can
be regarded as n-type semiconductor: N-type semi-
conductors form an rectifying i.e. active interface
with ITO rather than Al which is expected to re-
sult in a symbatic EQE spectrum when illuminated
through ITO - regardless of the actual exciton dif-
fusion length.
Further confirmation of the n-type nature of this
particular organic dye could be obtained from field
effect measurements following the argumentation in
Ref. [134]. In the same reference a very similar
perylene diimide has been classified as n-type semi-
conductor by means of the symbatic response
31
and
field-effect-transistor IV curves.
Figure 3.56: The semi logarithmic plot of the data
in Fig.3.55 reveals an onset of the photoresponse
even before the onset of absorption (as determined
by our absorption setup). Photo-thermal deflection
(PDS) measurements may be useful to investigate
this further.
The EQE reaches about 0.08% which is about
twice as high as MEH-PPV cells with the same
thickness. This is still between one and two orders
of magnitude smaller than the values for perylene
diimide pigments published in Ref. [245]. Thus,
the same side chains that facilitate solubility pre-
vent the molecules from close packing as in pigment
films. Hence, strong inter molecular interaction
seems to favour efficient charge transport and/or
exciton generation.
The logarithmic plot in Fig.3.56 reveals that the
onset of the EQE spectrum does not exactly co-
incide with the onset in the absorption at 620nm.
Significant photoresponse can already be found at
about 700nm. Photoresponse before the absorption
onset has also been found in THPF - earlier in this
Chapter. Both materials have electron withdraw-
ing groups that are expected to lower the LUMO
to make them stronger electron acceptors - see also
Chapter 2.
31
when illuminated through SnO
2
62 CHAPTER 3. SINGLE LAYER DEVICES
IV Characteristics
The light and dark characteristics of this device is
shown in Fig.3.57. We found very large currents in-
dicating high conductivity and a rectification ratio
of about 20 at ±1V. Note that similar to the pre-
viously described PTV and PIF devices there is a
relatively small negative breakthrough voltage indi-
cating the presence of a counter-diode.
From the slopes in Fig.3.57 we have extracted the
following resistor values:
• light: R
s
= 0.9MΩ and R
sh
=50MΩ
• dark: R
s
= 0.9MΩ and R
sh
=60MΩ
Both the series and the shunt resistors are more
than an order of magnitude smaller than in e.g.
MEH-PPV or LPPPT. This would probably ac-
count for the relatively small V
oc
and the high pos-
itive dark current. The value for the series resistor
has been taken from the slope in the negative volt-
age range - since the slope there is smaller than the
one in forward bias representing the “lower” upper
limit for R
s
.
Figure 3.57: IV dark (dashed line) and light
(λ =550nm, solid line) characteristics of the Per
device. A clear rectification for voltages above 1V
is observed.
We note that the formation of crystals has been
observed in Per/P3HT blends but also to a smaller
extent in single layers [62, 63]. Crystal formation
is a known problem with dye films and can lead to
pinholes and partial shorts - which might be - at
least partly - responsible for the high conductivity
here.
In addition, the dark and the light characteris-
tics have similar shapes even in the negative bias
region. This means that the EQE cannot be in-
creased by increasing the negative bias voltage as
in MEH-PPV or THPF. Note that photocurrent
multiplication as large as 10 000 times has been
observed in other perylene diimides sandwiched be-
tween Au electrodes [133].
Summary
In this Section we have found that:
• A soluble perylene (Per) derivative can be used
to fabricate an ITO/Per/Al solar cell from
solution avoiding the usual high temperature
sublimation steps hitherto used for perylene
derivatives.
• The EQE reaches up to 0.08% which is twice as
much as comparable MEH-PPV devices but 10
to 100 times less than the best pigment devices.
Hence, compact packing of molecules (as with
pigments or using the LAPS method) seems to
be required to obtain the best efficiencies.
• The Per/ITO interface is the “active” inter-
face which is consistent with the n-type semi-
conducting properties of other known perylene
derivatives.
• The V
oc
(190mV) of the dye device is somewhat
smaller than in pigment devices (280-530mV) -
although different illumination conditions and
electrode material have to be considered.
• The EQE onset appears to be significantly be-
fore the absorption onset which requires further
clarification in future studies.
• A counter-diode has been formed similar to
PTV and PIF devices.
• Resistor values obtained from the slopes in the
IV curves are:
– light: R
s
= 0.9MΩ and R
sh
=50MΩ
– dark: R
s
= 0.9MΩ and R
sh
=60MΩ
3.8 Ter
As we have mentioned earlier, there is some evi-
dence that large polycyclic aromatic hydrocarbons
3.8. TER 63
Figure 3.58: Chemical structure of the soluble ter-
rylene derivative, Ter, investigated in this Section.
are likely to give good photovoltaic conversion effi-
ciencies - see e.g. Ref. [245].
The reason for that may be found in the possibil-
ity of a more compact arrangement of the molecules
in the thin films that leads to the following list of
advantages:
• The relatively undisturbed π system of planar
molecules favours exciton and charge transport
as well as a low bandgap.
• The higher molecular density makes the occur-
rence of pinholes i.e. shorts more unlikely.
• The higher concentration of π - electrons per
unit volume increases the absorption coeffi-
cient.
Well-known examples for large polycyclic aro-
matic hydro-carbons are phthalocyanines and
perylenes. The “pure” perylene structure can be
regarded as two peri-condensed naphthalene com-
pounds. Theoretically three or more naphthalene
compounds can be condensed together to form a
series of oligo-rylenes leading to a rylene-polymer.
Such a polymer i.e. large rylene should have a van-
ishing bandgap indicating the possibility of intrinsic
electrical conductivity (metallic behaviour) accord-
ing to theoretical predictions - see Ref. [145] and
references therein. In fact a decreasing bandgap has
been found for a series of increasing rylene units:
terrylene (≈ 620nm or 2.02eV), quaterrylene (≈
800nm or 1.58eV) and pentarylene (≈ 950nm or
1.32eV) [145].
The cited bandgaps have been obtained for thin
films after thermal sublimation. These oligo-rylenes
were also amenable to spectroscopic measurements
in solution (1,4 dioxane) thanks to substitution with
four tert-butyl side chains [145].
We note that such a terrylene derivative has
played an important role in nano-science where sin-
gle molecule spectroscopy was possible even at room
Figure 3.59: The EQE of a Ter-device shows a very
broad spectral response (UV to near IR) reach-
ing high numbers in the short wavelengths range
(around 1%).
temperature [153, 71]. In the foreseeable future,
the application of, for example, heat to a single
molecule and thereby the initiation of an optically
detectable chemical reaction affecting the PL of a
single molecule may lead to optical data storage on
a molecular scale.
In this Section we investigate the photovoltaic
properties of this well soluble terrylene diimide
derivative as active material between ITO and Al.
Experimental
The terrylene has been synthesised by Stefan Becker
at the Max Planck Institut (Mainz, Germany). De-
tails of the synthesis can be found in Ref. [16].
The polymer was spin-coated from chloroform so-
lution (6mg/ml) onto pre-etched ITO coated quartz
substrates with 1600rpm in a laminar flow box. We
note that the solubility (< 1mg/ml in Chloroform)
of the quaterrylene of the same series (same side-
chains) was not sufficient to produce continuous
films via spincoating.
Absorption and photocurrent measurements were
performed in air. More details about the sample i.e.
device preparation and the IV/EQE measurements
can be found in Ref. Chapter 10.
Results and Discussion
Absorption and EQE Spectra
Fig.3.59 shows the EQE spectrum of an ITO/Ter
(≈ 15nm thick) /Al cell when illuminated through
64 CHAPTER 3. SINGLE LAYER DEVICES
Figure 3.60: The IV dark (dashed line) and light
(solid line, λ = 650nm, 0.28mW/cm
2
) characteris-
tics of this Ter device could only be recorded for a
very small voltage range due to the steep slope i.e.
high conductivity (330kΩ).
the ITO as well as the linear absorption curve. The
absorption spectrum of the Ter film is very broad
(410 to 700nm) and featureless apart from a little
dip at 630nm.
The EQE spectrum resembles the spectrum ob-
served in PTV with no clear antibatic peak in the
absorption onset but reaching the highest EQE in
the UV range with the local absorption minimum.
In fact the EQE is surprisingly high for a single
layer cell with values above 0.01% throughout the
visible spectrum reaching about 4% in the UV.
However, since the IV curve suggests a very con-
ducting and/or strongly pinhole dominated and
therefore partly shorted film, we have to take into
account that the EQE might be affected here by
small fluctuations in the voltage applied by the
source measure unit. For example, a voltage change
of 1mV can lead to an increase of current of 3nA
equivalent to change in EQE of nearly 1% here.
Generally, both the HOMO and the LUMO is ex-
pected to become lower with increasing π system i.e.
the terrylene should be a better electron acceptor
than the perylene.
IV Characteristics
The light and dark characteristics of this device is
shown in Fig.3.60. The very steep slope that should
represent the shunt resistor (here only 330kΩ) sug-
gests that the device is considerably shorted via pin-
holes in the only 15nm thin film. This is not sur-
prising considering the observations of crystal for-
mation even to a larger extent than in previously
investigated related molecule Per [64]. The partial
shorts would also explain the lack of a detectable
open circuit voltage 10mV.
However, upon illumination the current at 0V
changes from +1000pA to - 97pA. The positive dark
current is most likely an artifact related to a slight
ground potential of the source measure unit. Note
that an offset of only 1mV would change the cur-
rent by about 3nA here. IV curves with very high
conductivities and small open circuit voltages tend
to be observed more often in electron conductors
like PIF and others [150, 64].
Summary
In this Section we have found that:
• A soluble terrylene derivative can be used to
produce a photocurrent. The EQE around
500nm is even in the same order of magnitude
as in e.g MEH-PPV reaching higher values of
up to 1% around 400nm.
However, since we could not obtain a de-
tectable V
oc
we believe the device performance
may be affected by local shorts due to the for-
mation of crystals causing a resistance of only
about 900kΩ. As a consequence, the IV and
EQE characteristics of this particular device
have therefore to be interpreted with caution.
3.9 Single Layer Devices -
Summary
In Table 3.5 we have listed the most important so-
lar cell parameters for the different organic semi-
conductors investigated in this Chapter. Many
of these materials will be investigated further in
blend and/or double layer devices in the subsequent
Chapters.
Where possible, we have chosen a device with a
“typical” characteristic. The EQE value of the Ter
device has to be treated with caution since the de-
vice showed non-ideal IV and EQE characteristics.
The list shows that only devices with shunt re-
sistors above 100MΩ and dark currents under 30pA
do develop a V
oc
larger than about 200mV. We also
have to consider that the values for R
s
are only
upper limits and that the actual values could be
considerably smaller.
3.9. SINGLE LAYER DEVICES - SUMMARY 65
Table 3.5: Survey of photovoltaic parameters of the investigated organic semiconductors. The number in
brackets are obtained from the dark characteristics. d stands for the thickness of the organic layer. EQE
values were taken at peak wavelength.
material d EQE V
oc
I
d
R
sh
R
s
comment
– nm 10
−3
% mV pA MΩ MΩ –
Ter 15 4000 0 1000 0.9(0.9) 0.9(0.9) dye
THPF 110 100 600 6 6300(17000) 6300(17000) polymer
LPPPT 59 85 105 31 83(200) 7.5(11) polymer
Per 40 80 190 825 50(60) 0.9(0.9) dye
PTV 90 70 370 0.8 3000(6400) 200(200) polymer
MEH-PPV 35 45 730 14 500(11000) 40(74) polymer
PIF 30 3 0 116 66(83) 1.7(1.7) polymer
Hence, a suspected correlation of the EQE with
the series resistor or the ratios between R
s
and R
sh
cannot be ruled out here.
The list also reveals that both polymers and dyes
give peak EQE values of approximately the same
size ranging from 0.04 to 0.10% - with MEH-PPV
being near the lower end in terms of the photo-
current but on the high end with respect to the
open circuit voltage.
66 CHAPTER 3. SINGLE LAYER DEVICES
Chapter 4
Double Layer Devices
4.1 Introduction
In the previous Chapter we have found that the
main losses in single layer structures are due to
short exciton diffusion lengths and recombination
of the excited charge carriers. The short exciton
diffusion length causes a very thin active region of
the device whereas recombination losses before and
after charge separation can reduce the shunt resis-
tor considerably.
Figure 4.1: Device architecture of a double layer
photovoltaic cell. The electron acceptor A and elec-
tron donor D are sandwiched between the transpar-
ent ITO contact and an Al layer.
With the introduction of an electron acceptor
layer between the active material and the nega-
tive electrode (Al), both the exciton diffusion range
and the poor shunt resistor can be improved - see
Fig.4.1.
Provided the offset between the HOMO and
LUMO levels of both materials upon contact are
sufficient to split excitons, the separated charges
can travel to the respective electrode with very
small chances of meeting a recombination partner.
The active region is now extended to the exciton
diffusion length of both materials.
Moreover, the use of two different semiconduc-
tors allows tuning of the absorption spectrum for
e.g. maximum coverage of the spectrum of the light
source. The latter is not only desired for optical
photodetectors but also essential if high solar cell
efficiencies are required.
In fact, absorption of photons from 300 to
1000nm is required to utilise 70% of the solar ra-
diation on earth. Only about 22% of the solar radi-
ation can be exploited with a single layer material
like MEH-PPV.
One of the first organic double layer solar cells
is the one reported by Tang in 1985 [244]. He
sublimed two pigment films, a copper phthalocya-
nine and a perylene derivative on top of each other,
sandwiched between an ITO and Ag contact, and
obtained one of the hitherto most efficient organic
photovoltaic devices. The separation of the charge
carriers resulted in a higher shunt and low series
resistor such that the fillfactor was comparable to
commercial Si photodiodes.
Since the absorption of the perylene derivative
and the Cu-phthalocyanine complement each other
very well, the device showed a good EQE over
the full visible range. The open circuit voltage at
AM1.5 condition (75mW/cm
2
) was 450mV.
Although a rough similarity to the P/N junction
in inorganic solar cells can often be drawn, care has
to be taken using such comparison. However, in
both cases equalisation of the Fermi-levels occurs at
the interfaces between semiconductors and/or met-
als. In both organic and inorganic solar cells the
equalisation of Fermi levels leads to a significant
potential gradient that can drive the photoexcited
charge carriers (minority carriers) through the rest
of the bulk towards the electrodes.
However, as discussed in Chapter 2 the nature of
doping in organic semiconductors can be very dif-
ferent from the usual doping in inorganic materi-
als. Although, the Fermi-levels are still important
the offset of the valence band (VB) and conduction
67
68 CHAPTER 4. DOUBLE LAYER DEVICES
band (CB) at the interface after equalisation of the
Fermi-levels is a crucial factor for the function of
these devices. This is largely because there are ex-
citons in organic devices that need to be split at the
first place.
Thus, a device with sufficient VB/CB offset can
split excitons and separate charges efficiently if both
materials are e.g. p-type conductors. As a con-
sequence, it is more appropriate to speak about a
D/A -rather than P/N junction in case of organic
solar cells. Note that the term heterojunction is also
found in the literature but we prefer D/A junction
in this thesis since a junction between two differ-
ent semiconductors does not necessarily lead to the
required splitting of the excitons.
Figure 4.2: Band energy diagrams for a D/A dou-
ble layer device in short circuit mode. Left: the
built-in field due to ∆W
f
of the electrodes leads
to tilted bands if only few free charge carriers are
present. Right: For higher charge carrier concen-
tration the bands can remain flat in the bulk but
blocking contacts may be formed
The figure on the left in Fig.4.2 shows the band
energy diagram for a double layer device with a low
concentration of free charge carriers. The difference
of workfunctions of the electrodes ∆W
f
can cause
the bands to tilt, as shown, which creates a constant
field across the bulk. Although band bending at
the electrodes can still occurs, it may considered
insignificant in such devices
1
.
The figure on the right in Fig.4.2 shows the situa-
tion for a device with a concentration of free charge
carriers that is high enough to compensate the built
in field due the electrode workfunctions within a
fraction of the layer thickness. As a consequence,
accumulation and depletion of charges near the elec-
trodes can cause distinct band bending
2
.
According to this picture the driving force i.e.
1
Here we have also neglected the band bending at the
D/A junction.
2
We have drawn the formation of blocking contacts,
ohmic contacts are more likely to form using Au and Ca
instead of ITO and Al.
V
oc
is mainly determined by the offset at the D/A
junction although the increased built in field (tilted
bands) can help to transport charges through the
bulk [223].
In Table 4.1 we have listed D/A double layer
devices from the literature. The device with

has been measured at only 0.015 mW/cm
2
so
that significantly higher values for V
oc
may be
achieved if measured under intensities in the order
of 0.5mW/cm
2
.
The table shows that the highest efficiencies have
been achieved with insoluble molecules (pigments).
Pigments have been successfully used as photo-
receptors and charge transport media in colour
copying machines and related devices (Xerography)
replacing many inorganic materials
3
.
However, the prospect of easier processing by
e.g. spin or blade coating from non-acidic solution
4
,
lamination or even inkjet printing, makes soluble
molecules technologically more interesting. The ta-
ble also shows that technically pigments can already
be replaced by polymers, dyes and even liquid crys-
tals but at the expense of lower efficiency.
We note that only recently, a modification of the
double layer device, the laminated device structure
(see Chapter 6), allowed semiconducting polymers
to exceed in performance values for the best pig-
ment devices [94].
Both Pcs and Per pigments show very high sta-
bility, good absorption and good photo-conductive
properties which makes them the probably most in-
vestigated photovoltaic materials in the past [268,
43, 29, 154, 231, 267].
On the other hand, the easier processible soluble
derivatives have attracted the interest of researchers
more recently - driven by the fast progress of poly-
mer based devices and the discovery of outstanding
electronic properties of liquid crystalline (LC) ma-
terials [228, 226, 227, 184, 253, 254, 255, 259, 260,
262, 1, 2].
Although the successful synthesis of perylene
dyes has been reported already 18 years ago [209],
phthalocyanine dyes and polymers of both have
only been synthesised in the more recent past - see
Refs above.
We note that many other molecules that have
been proven suitable as photo-active pigments in so-
lar cells, may also be of utmost interest to be chem-
ically modified so that they can be used as dyes, liq-
3
In 1993 more than 90% of the xerographic photo-
receptors were made of organic photo-conductors [154]
4
Recently a method was found to dissolve certain pig-
ments in acids [283]. Although, this simplifies film deposi-
tion, the use of acids introduces other problems.
4.1. INTRODUCTION 69
Table 4.1: Survey of performance parameters of various organic double layer solar cells. The abbreviation
for materials are listed in Chapter 1. Polymers and dyes in this table were processed from solution
whereas pigment films have been formed by thermal sublimation. The EQE for devices marked with
+
was estimated from the information given in the respective papers. Per denotes the soluble perylene
derivative while Per1,2 and 3 refer to different perylene pigments. Best parameters are printed bold
while results discussed in this Chapter are printed italic.
EQE V
oc
FF range materials comments year
% V % nm D/A –
30
+
0.5 65 400-800 CuPc/Per1 pigment/pigment [244] 1985
23 0.5 ≈30 400-700+ HPc1/Per2 pigment/pigment [131] 1990
23 0.3

40

400-600 PEDOT/PEOPT/C60 polymer/pigment [213] 1998
9 0.8 48 400-560 PPV/C60 polymer/pigment [111] 1996
6 1.0 60 400-620 PPV/Per3 polymer/pigment [113] 1996
≈ 3
+
1.0 23 400-620 PPyV/P3HT polymer/polymer [242] 1997
2 1.0 51 400-590 PEDOT/PPV/Per polymer/dye [9] 1999
0.6 0.12 25 400-800 CuPc2/Per dye/dye [200] 2000
0.4 0.35 20 400-850 HPc/Per3 liquid crystal(dye)/pigment [199] 1999
uid crystals or polymers in solar cells. Among those
should be mentioned: squaraines, quinacridone,
azulenium, but also charge transfer complexes like
PVK:TNF [154] and certainly the large number of
different merocyanines and azo pigments [154, 29].
Many dyes are commercially available from suppli-
ers for the copy industry (Xerography) like Syn-
thec/Synthone in Germany but also laser dye man-
ufacturers like Kodak.
The following we list some ideas that have been
published to improve the performance of double
layer photovoltaic structures:
1. It has been shown that the device efficiency of
a double layer (PEOPT/C60) is strongly con-
nected to the geometry and optical field distri-
bution [213].
Choosing suitable electrodes and thicknesses of
the involved layers in such a device allows to
optimize the optical field taking into account
interference effects at the active interface and
thus enhance the exciton concentration within
a layer with thickness comparable to the short
exciton diffusion length.
2. F. Zhu and J. Singh have published an opti-
cal analysis method to optimise the structure
of multi-layer thin film solar cells with respect
to the thickness of the layers and the electrode
materials [281, 282]. Their model takes into
account multiple reflections of multi-junction
structures composed of absorbing and non ab-
sorbing materials.
They have applied their analysis on e.g.
glass/ITO/p/i/n/rear-contact where p,i, and
n stands for the doped and intrinsinc layer of
amorphous silicon. Their method may be of
interest for application on organic multi-layer
structures to maximise the effects of the inci-
dent radiation.
3. Another possibility to optimise organic double
layer cells is to place the material with the
higher bandgap (e.g. a perylene derivative)
next to the transparent (ITO) contact and the
lower bandgap material (e.g. a phthalocya-
nine) behind. Although, a higher workfunc-
tion material than Al might then be required to
avoid the formation of blocking contacts, such
structure can show better efficiencies than the
reverse order of layers [268, 187].
As possible reasons
5
for such an improvement
may also be considered: longer exciton diffu-
sion lengths in the high bandgap material or
simply lower thermalisation losses of the pho-
toexcited carriers in the first layer.
5
We note that in Ref. [187] Foerster energy transfer fa-
cilitating exciton diffusion to the active junction is given as
explanation for better performance in these devices.
70 CHAPTER 4. DOUBLE LAYER DEVICES
Figure 4.3: The figure shows the chemical structure of the two organic semiconductors (HPc, Per3)
and the architecture of the double layer device structure. Charge separation typically occurs at the
HPc/Per3 interface so that electrons and holes can travel individually without a high chance to meet a
recombination partner. Thus, only the exciton diffusion range determines the width of the active region
in such a structure. One device has been heated and slowly cooled in order to obtain the alignment of the
discotic molecules of HPc, as shown here, which is expected to affect the charge and exciton transport
properties. The characteristics (IV, EQE) of this device were compared with the figures of a non-heated
reference cell.
4.2 The Dye/Pigment Double
Layer Device (HPc/Per3)
As mentioned above the small exciton diffusion
length in most organic semiconductors represents a
serious limitation for double layer structures. More
recently researchers have discovered not only long
diffusion lengths (several 100nm) for excitons, but
also particularly high mobilities (0.1-0.5cm
2
Vs
−1
)
for charges in discotic liquid crystalline materials
[1, 57, 21].
It is the strong intermolecular interaction of the
flat disc-like molecules which acts as driving force
to induce a higher degree of order. The forma-
tion of such a highly ordered phase requires a cer-
tain degree of mobility of the molecules themselves
which can be found in the liquid crystalline phase
(mesophase).
Indeed, PR-TRMC studies have revealed that
both solid (crystalline) and mesophases of many
discotic molecules allow have higher charge carrier
mobilities than in most pigments or polymers [57].
The same study revealed also that the crystalline
phase of the same liquid crystalline materials shows
even higher numbers than the mesophase of a given
material. Our own studies of HPc in single layer de-
vices are consistent with these improved transport
properties at room temperature- see Chapter 3.
In practice liquid crystalline molecules had plenty
of opportunities to rearrange themselves during pu-
rification steps (e.g. column chromatography, re-
crystallisation) or simply during the last synthe-
sis steps which typically involves solutions. More-
over, materials with transition temperatures <≈
40
o
may have seen sufficiently high temperatures
during their storage or transport period prior to
their use. The decreased motion (vibrations) of the
molecules at lower temperatures i.e. the solid phase
allows even closer packing and even less disorder
once the molecules have arranged themselves.
Although the PR-TRMC studies seem very con-
clusive there are also signs for poorer transport
properties in the crystalline phase. For example,
alkoxy substituted metal free phthalocyanine shows
very high PL (50%) at low temperatures (4.2K)
which is considerably quenched at the transition to
the mesophase (372K) [21]. This quenching effect
is interpreted as increased mobility of the excitons
in the mesophase so that they can now reach more
nonradiative decay sites
6
.
In other words it is a priori not clear whether
or not the effort to orient the columns perpendic-
ular to the electrodes pays off in terms of photo-
6
We note that breaking the conjugation(reducing the ex-
citon diffusion length) of PPV chains by the introduction of
acetylene units has also increased the PL[]
4.2. THE DYE/PIGMENT DOUBLE LAYER DEVICE (HPC/PER3) 71
voltaic performance. One advantage of liquid crys-
talline phases is that they are self-repairing and can
thereby minimise defects that can act as recombi-
nation sites. In any case, already the high solubility
that allows simple spin-coating like normal dyes - is
technologically very interesting.
After characterisation of a number of mesogenic
phthalocyanines - see Chapter 7 we found that only
one compound, HPc, forms a liquid phase while all
the others started to decompose before the isotropic
liquid can be formed. The particularly high mobil-
ity of the molecules in the liquid phase allows the
formation of the perpendicular arrangement at the
transition to the discotic phase as observed for HPc
in Chapter 7 or for a similar derivative in Ref.[152].
HPc, is also soluble enough to allow spin coating
from solution and sufficiently viscous to form excel-
lent films on the ITO substrates despite the rela-
tively low molecular mass compared to polymers.
During the transition from the solid to the liquid
phase the surface tension leads to the destruction of
the continuous thin film. As a consequence, instead
of a thin film, much thicker droplets which are sur-
rounded by areas of no or very little material are
created. Hence, subsequent sublimation of the Al
top contact inevitably results in a shorted device.
In order to avoid these shorts the electron acceptor
Per was sublimed prior to the deposition of the Al
top contact.
Thus, this device may in fact be regarded as a
hybrid between a single layer device (ITO/Per3/Al)
and a double layer device (ITO/HPc/Per3/Al) with
varying thickness of both organic layers. Hence, it
is not possible to probe the effect of alignment via
comparison with a pristine device directly
7
.
However, such a device has not been investigated
yet and it should at least be possible to see to
what extent the two materials can contribute to the
photocurrent
8
in such a hybrid structure.
After spin-coating HPc onto ITO, we heated the
phthalocyanine until it forms an isotropic liquid
(clearing point =292
o
C) and cooled it slowly down
to room temperature - according to the procedure
described in Chapter 3. After that we sublimed a
thin layer of Per3 as electron acceptor and transport
layer. The double layer device was then completed
7
Note that according to Ref.[57] see Table 7.2 the dif-
ference of charge carrier mobilities in HPc at the solid to
mesophase transition is only very small. Also the exciton
diffusion length is likely much longer than the layer thick-
ness - see Chapter 3 and Ref.[21]. Thus a dramatic change
in performance is not expected even if a non hybrid double
layer could be fabricated.
8
Such a distinction is possible since Per3 does not absorb
around 650-850nm but HPc does.
by the sublimation of an Al layer as top contact.
Then we compare the external quantum efficiency
(EQE) of this device with the same device architec-
ture which has not been heat treated.
Experimental
Films of HPc have been fabricated in air by spin
coating from solution onto indium tin oxide (ITO)
coated glass substrates. HPc was dissolved in
chloroform (30mg/ml) and filtered with 0.1mm
PTFE-filter. One HPc film on ITO was heated to
292
o
C (20
o
C /min) under a constant nitrogen flow
and slowly (5
o
C/min until 270
o
C, then 20
o
C/min)
cooled down to room temperature using a hot stage
(Linkam).
The transition into the liquid phase as well as
the formation of a few linear defects - see Chap-
ter 3 (visible with crossed polarisers) has been ob-
served simultaneously under the microscope (Vick-
ers) which has been adapted for temperature-
dependent polarisation microscopy. Exposure to
the illumination source (tungsten) has been min-
imised to avoid photo-degradation effects. The Al
top contact was obtained through thermal evapora-
tion. The thickness of the HPc film after spincoat-
ing was 450nm and the Per3 film was about 20nm.
Current- voltage curves and photocurrents were
measured using a Keithley 237 source measure unit.
Monochromatic illumination was provided by the
output of a tungsten lamp dispersed by a Bentham
M300 single-grating monochromator. Quantum ef-
ficiencies were determined by normalisation with a
calibrated silicon photodiode at the sample posi-
tion. The sample was kept under vacuum (< 10
−5
mbar) during the measurements. Reflection losses
were neglected for all measurements. Absorption
spectra were taken with a UV-Vis spectrometer (HP
8453).
Results and Discussion
Absorption and EQE Spectra
Fig.4.4 shows that - for the pristine device - illu-
mination through ITO results in an EQE spectrum
with maxima where the absorption is low and EQE
minima at absorption peaks. This suggests that
a considerable part of the HPc layer absorbs light
but does not contribute to the photocurrent (opti-
cal filter effect). The thickness of the active layer
can be estimated using Eq. (3.3) and gives about
180 ± 90nm. This value is in good agreement with
72 CHAPTER 4. DOUBLE LAYER DEVICES
Figure 4.4: EQE and optical density of the pristine
(solid lines) and the heat treated (dashed lines) de-
vice. Illumination through ITO.
the 200±80nm obtained from the measurements of
the single layer HPc - see Eq. (7.2).
Figure 4.5: IV-characteristics in light at 570nm and
0.19mW/cm
2
of the heated device and the pristine
device.
Fig.4.4 shows also that the heat treated device
only reaches about 1/10 of the EQE of the pristine
device. The heated device using the same mate-
rials and structure was heated until it turned into
a liquid at 292
o
C and slowly cooled down to room
temperature to obtain the perpendicular orienta-
tion of the discotic molecules. Interestingly the ab-
sorption spectrum of the heated device is broader
in the range where HPc absorbs only (650-850nm).
Upon illumination through the semitransparent
Al contact (transmission about 2%) the EQE of the
pristine device follows the features of the absorption
spectrum (not shown here, see [199]). This is ex-
pected in this device since the Per3 layer is rather
thin (about 20nm) so that a filter affect due to a
passive Per3 layer can not be very strong.
If we consider the transmission losses of the Al
contact the EQE reaches up to about 1%. This
number is not very different from the EQE obtained
in double layer structures containing at least one
polymer layer - see Table 4.1. The photo-response
spectrum of this device is very broad. The V
oc
is 350mV which is slightly smaller than in devices
made of insoluble molecules.
The low OD of the heated device is expected since
droplets have been formed leading to very thick but
also very thin layers. The latter dominate the ab-
sorption spectrum leading to the shown low OD
numbers. The redistribution of oscillator strength
into the low energy region may be due to the high
ordered phase of HPc.
IV Characteristics
The IV curves in vacuum for both the pristine and
the heated device are shown in Fig.4.5. The pristine
device shows only little rectification but an V
oc
of
350mV.
The heated device shows higher currents for small
voltages and also little rectification. The V
oc
is only
about 50mV. In addition the heated device reveals
a stronger voltage dependence in both reverse and
forward bias. This may be explained by the for-
mation of a “counter”-diode in conjunction with a
vanishing HPc layer for parts of the device.
We note that the shape of the IV curves can
be changed considerably upon exposure to air ap-
proaching the shape of the pristine device [199]. In-
creased conductivity of the thick HPc droplets may
here increase the contribution of HPc. However, the
parts of the device which contain only Per3 seem to
be responsible for the low V
oc
. Interestingly the
photocurrent of the heated cell reaches a maximum
after about 4 hours exposure to air [199].
Summary
In this Section we have shown that:
• A layer of discotic liquid crystals can be used
to manufacture a working D/A double layer
photovoltaic device.
• Heat treatment to align the discotic columns
perpendicular to the electrodes could not im-
prove the efficiency.
• The efficiency (EQE) of the pristine device is
between 0.5 and 1% which is about 30 times
smaller than comparable devices using only
pigments. The V
oc
was 350mV - similar to typ-
ical pigment devices.
4.3. THE DYE/DYE DOUBLE LAYER DEVICE (CUPC/PER) 73
4.3 The Dye/Dye Double
Layer Device (CuPc/Per)
Figure 4.6: Chemical structure of the two dyes used
in this section: A perylene tetracarboxylic acid-bis
imide with aliphatic side chains (Per) and a copper
phthalocyanine with polar side chains (CuPc).
The D/A double layer cell reported by Tang who
sublimed insoluble CuPc and perylene derivatives
onto ITO with a silver top electrode is still one of
the most efficient organic solar cells [244].
In this Section we investigate a double layer solar
cell using soluble derivatives (see Fig.4.6) of Tang’s
pigments. That way we could circumvent the ther-
mal sublimation process and simply manufacture
the device at room temperature via spincoating.
The essential differences to Tangs pigments are
the polar side chains attached to the phthalocya-
nine structure and the short but bulky aliphatic
side chains of Per. Since these side chains are not
electronically active, we might expect that these
molecules would have very similar electronic prop-
erties to those in Ref. [244]. These side chains do
alter the mechanical properties since they prevent
aggregation in solution although they do not induce
liquid crystalline phases here.
However, they are also likely to change the in-
termolecular arrangement in the solid (film) which
can affect both charge and exciton transport and
the optical properties of the films. Indeed, we have
found that the absorption coefficient at the absorp-
tion maximum of HPc is about 20% smaller than in
typical phthalocyanine pigment films.
Experimental
The chemical structures of the two dyes used in
this section are shown in Fig.4.6. Per and CuPc
are commercially available (Synthec, Aldrich). The
double layer device (Fig.4.1) was fabricated by spin
coating onto an ITO/quartz substrate from a so-
lution containing 30mg/ml CuPc in distilled wa-
ter followed by 10min annealing at 150
o
C in air.
Per/chloroform solution (30mg/ml) was spin coated
on top of the CuPc layer with subsequent sublima-
tion of the Al electrode. The Per film was about
60nm thick and the CuPc film 15nm. The CuPc
solution is not very viscous despite the high concen-
tration and did not only give very thin films with
moderate quality. The device was prepared in air
but measured after 1 hour in vacuum.
Results and Discussion
Absorption and EQE Spectra
Fig.4.7 shows the EQE spectrum together with the
optical absorption spectrum of the dye-based dou-
ble layer cell. Because of the very thin CuPc layer
(15nm) the optical density is less than 0.1 at wave-
lengths where Per does not absorb (650-800nm).
However, even though only about 10% of the inci-
dent photons around 700nm are absorbed the EQE
spectrum is almost constant between 450nm and
720nm. The EQE roughly follows the absorption
spectrum suggesting the absence of significant filter
effect.
The EQE around 700nm seems to be enhanced
compared to the rest of the spectrum. This indi-
cates that unlike the 15nm thin CuPc layer the Per
layer can not facilitate dissociation of all absorbed
photons. Hence, the thickness of the active region,
which is the exciton diffusion range of both the Per
and the CuPc layer, is likely shorter than the 60nm
thick Per film.
Figure 4.7: Linear optical absorption and EQE of
the double layer device.
A more efficient device may be obtained by re-
74 CHAPTER 4. DOUBLE LAYER DEVICES
ducing the thickness of Per and increasing the thick-
ness of CuPc. The highest EQE would then be ex-
pected when both layers are as thick as their respec-
tive exciton diffusion lengths. Note that the fairly
symbatic response spectrum indicates that the exci-
ton diffusion length of CuPc is possibly higher than
15nm. We also note that the CuPc film also con-
tains Na
+
ions which may affect the electronic prop-
erties.
IV Characteristics
The (IV) characteristics in the dark and under illu-
mination are shown in Fig.4.8. This device devel-
ops a relatively low open circuit voltage of 120mV.
V
oc
for other double layer devices at comparable in-
tensities is typically a few hundred mV. Although
there is only little rectification for all shown curves
certain features are interesting: The linear IV char-
acteristics for small positive voltages indicates that
the series resistant affects the shape and FF in this
region. It even extends into the negative voltage
range so that the EQE at 0V is directly affected
and reduced by resistive losses.
Figure 4.8: IV characteristics in the dark and
under illumination at 610nm and an intensity of
0.25mW/cm
2
. The thickness of the CuPc layer
was about 15nm and the Per layer was about 60nm
thick.
An analysis of the slopes at -0.3V and +0.5V
gives values of R
sh
=1.4 · 10
7
Ω and R
s
=0.7 · 10
7
Ω.
Although the R
s
is not unusually high
9
the shunt
is relatively low. In Chapter 2 we have seen that
the effect of the series resistor becomes important
if the shunt is not at least an order of magnitude
9
We found values around 0.210
7
Ω in MEH-PPV devices
higher. The reason for a low shunt resistor can be
recombination at the interface or local shorts.
We note that Per tends to form crystals that may
easily penetrate the thin CuPc film and thereby
favour a lower
10
R
sh
. The small shunt compared to
R
s
may also be responsible for the low V
oc
. More-
over the hyper-linear increase of current for negative
bias suggests the presence of a diode with reverse
polarity (=“counter”-diode).
We note that according to the IV curve the EQE,
judged from the difference between dark and light
characteristics, can be more than doubled if a small
negative bias voltage is applied. This can be used
if the device is used as a photo-diode with a broad
spectral response - or as solar cell with e.g. Ca
instead of Al electrodes.
Summary
In this Section we have found that:
1. A whole new class of organic semiconductors -
small soluble organic molecules (dyes) - can be
used to built double layer photovoltaic cells.
2. The efficiency of dyes in these first studies is
about an order of magnitude lower than e.g.
polymer devices and about 30 times lower than
the pigment double layer structures.
3. Both R
s
and R
sh
appear to play an important
role as limiting factors in these devices. The
surprisingly high R
s
for a double layer struc-
ture may be due to larger intra-molecular bar-
riers (hopping distance is increased by the in-
sulating side chains) that limit both the charge
and exciton transport in the investigated mate-
rial. Thus, further studies that focus on reduc-
ing R
s
by e.g. doping are necessary to estimate
the potential of these molecules.
4. The device shows a very broad spectral re-
sponse for a solution processed cell. This may
be exploited if used as a photodiode - where
the sensitivity can also enhanced by at least a
factor 2 by applying a small bias voltage.
5. The exciton diffusion length of Per is smaller
than 60nm whereas the number for CuPc is
likely larger than 15nm.
10
Also the Na
+
ions may contribute to the higher
conductivity.
Chapter 5
Blend Layer Devices
5.1 Introduction
In the previous Chapter we have seen that the in-
troduction of an electron acceptor (A) i.e. electron
transport layer between the negative electrode (Al)
and the electron donor material (D) can increase
the EQE with respect to single layer architectures
1
.
In such double layer structures exciton dissocia-
tion takes place at the D/A interface. At the same
time the acceptor material transports the electrons
to the negative electrode (Al) while the hole accep-
tor provides a pathway for the holes to the positive
electrode (ITO). Both are spatially well separated
so that mutual recombination of the photoexcited
charges during their journey out of the device is
reduced.
In addition, the thickness of the active layer can
be increased compared to single layer structures
since excitons created within their diffusion length
from both sides of the D/A interface can now con-
tribute to the photocurrent. Furthermore, the spec-
tral range of the photoresponse can be greatly in-
creased if the absorption spectra of the components
(D and A) complement each other.
In this Chapter we discuss another D/A-device
architecture: Blends i.e. mixtures of donor and ac-
ceptor materials (Fig.5.1). Since both blend and
double layer structures use two materials that can
form a charge separating D/A interface they share
many advantages over single layer devices - for ex-
ample: separate charge transport and a thicker ac-
tive layer.
However, the main advantage of the blend device
over double layers is that the active layer can be
considerably thicker than the sum of the exciton
diffusion lengths in the D and A material as long
as the blend comprises an interpenetrating network
1
Note that not everything could be improved, for exam-
ple, the open circuit voltage is generally higher in single layer
devices than in double layers.
Figure 5.1: Solar cell architecture comprising a
blend of two organic semiconductors as used in this
Chapter. Thicknesses of the blends vary between
20 and 120nm.
with domains not larger than twice the exciton dif-
fusion length. The latter is typically only about
20nm (2x10nm).
If we consider both the spectral dependence of
the absorption as well as the absorption coefficient
of the majority of presently available organic semi-
conductors, the layer thickness has to be at least
150nm of the pure material to approach 100% ab-
sorption over a broader wavelength range.
Even with a very low bandgap, a single layer can
only use about 7% and a double layer about 14%
of the incident light - assuming an exciton diffusion
length of 10nm in the constituent material(s).
Consequently, since blends could have active re-
gions as thick as necessary to absorb most light,
they have the potential to utilise 15 times more
photons then typical single layer structures and 7
times more than double layer devices.
However, blend devices have also drawbacks as
can be seen from the following list:
1. If processing from solution is desired, blends
require high solubility of both components
2
2
Note that double layer have similar restrictions: they
require at least one component to be insoluble in the other
components solvent or high thermal stability in order to use
75
76 CHAPTER 5. BLEND LAYER DEVICES
2. For maximum performance, blends require a
defect-free interpenetrating network for effi-
cient percolation of charges to the electrodes
i.e. phase separation on a scale near or smaller
than the double exciton diffusion length in the
respective material.
3. Despite the larger active region, most blends
tend to develop lower open circuit voltages
than single layer structures - for reasons which
are not well understood to date.
4. The random network of a blend device bene-
fits strongly from a driving force that directs
the separated charges from the D/A interface
to the respective electrodes. The difference
between the workfunctions of the electrodes
and/or an externally applied electric field (bias
voltage) can represent such a driving force.
As a consequence, blends generally show
stronger field (voltage) dependence of the pho-
tocurrent which often decreases the fill factor
and may cause lower V
oc
. Double layer struc-
tures, on the other hand, are less dependent
on external fields and do therefore not always
benefit from an extra driving force.
In Table 5.1 we give a survey of the four key pa-
rameters for solar cell efficiency in various blend
structures extracted from our own results and from
literature.
Note that already two of these parameters (sen-
sitivity range, open circuit voltage) reach i.e. even
surpass the numbers of their silicon counterparts
using the blend architecture. However, the EQE
of the best organic cell (29%) is yet only a third
and the best fillfactor in blends (44%) about 50%
of typical commercial silicon solar cells. The blend
of MEH-PPV with the hardly soluble C60 could
only be accomplished in a very strong solvent (1,2
dichlorobenzene). Since C60 is virtually insoluble in
most known solvents we regard it as pigment within
this thesis - in contrast to the C60 derivative in Ref.
[278] which is readily soluble in more common sol-
vents and therefore rather a dye than a pigment.
The table also illustrates that there are problems
to realise all of these performance parameters in
the same device.
In the following list we will comment on many of
these structures in order to point out specific nov-
elties and strengths
3
:
thermal sublimation for film deposition.
3
Here we follow the historical order
• The first report on D/A solar cells with con-
jugated polymer blends has been reported by
Halls et. al. [112]. Blends comprising an elec-
tron accepting (CN-PPV) and an electron do-
nating (MEH-PPV) polymer have shown EQE
values of up to 6% which boosted the EQE with
respect to single layer cells by about two orders
of magnitude.
The EQE of the single layer structures is
0.001% (CN-PPV) and 0.04% (MEH-PPV) re-
spectively. Phase separation on the scale of
10-100nm was demonstrated and a V
oc
of 0.6V
with fill-factors of 25% were obtained. The low
fill-factor is most likely due to the a relatively
small shunt that also causes the field depen-
dence of the light current.
However, the shunt must be large enough to
allow the device to develop the relatively high
open circuit voltage (0.6V). Another problem
here is the higher polarity of the electron trans-
porting CN-PPV that favours it’s accumula-
tion on the positive ITO electrode
4
during the
spincoating process [114].
• The highest EQE values (29%) among blend
devices have been achieved with a blend be-
tween MEH-PPV and a soluble C60 (Fullerene)
derivative as electron acceptor sandwiched be-
tween ITO and Al [278]. Photoexcited charge
transfer from the polymer donor to the C60
is about 1000 times faster
5
than radiative and
nonradiative recombination[83]. Thus the D/A
interface between MEH-PPV and C60 serves as
an efficient exciton dissociation site.
Under high monochromatic light intensities
(20mW/cm
2
) this device develops a V
oc
of
0.68V using ITO and Al electrodes but - inter-
estingly - only little more (0.82V) with Ca in-
stead of Al. Unfortunately, “internal resistive”
losses
6
do not allow the use of thicker devices
to circumvent the 40% transmission and reflec-
tion losses at the absorption maximum in these
devices. The MEH-PPV/C60 blends also show
the field dependent photocurrent that leads to
low fill-factors (here still only about 30%) typ-
ically encountered in blend devices.
Despite the mentioned obstacles for further im-
provement, the achieved performance with the
4
Molecular dipoles on the ITO surface attract the polar
CN groups.
5
Radiative and nonradiative decay of singlet excitons oc-
curs typically between 100 and 1000ps [102].
6
Probably related to charge and exciton transport prob-
lems through the C60 network.
5.1. INTRODUCTION 77
Table 5.1: Survey of performance parameters of various solar cells comprising at least one organic com-
ponent in a blend. The abbreviations for materials are listed in Chapter 1. Best parameters are printed
bold while results of this work are in italic.
EQE V
oc
FF range materials comments year
% V % nm – –
29 0.5 ≈25 400-550 MEH-PPV+C60 polymer+dye [278] 1995
23 0.4 ≈33 400-550 MEH-PPV/C60 polymer+pigment [83] 1998
12 0.5 26 400-640 MEH-PPV+CdSe polymer+nanocrystals [102] 1996
11 0.35 41 400-620 P3HT+Per polymer+dye [65] 2000
6 0.6 25 400-600 MEH-PPV+CN-PPV polymer+polymer [112] 1995
1.8 0.9 23 400-500 PPV+MWNT polymer+nanotubes [3] 1999
1.0 0.1 25 400-800 HPc+Per liquid crystal+dye [200] 2000
0.7 0.4 44 400-600 MEH-PPV+Per polymer+dye [62] 1999
0.4 0.2 25 400-850 PTV+Per polymer+dye 2000
0.4 0.12 25 400-900 PTV+Ter polymer+dye 2000
0.2 0.6 25 400-800 PTV+THPF polymer+polymer 2000
0.02 <0.01 25 400-1000 PTV+PIF polymer+polymer 2000
0.0007 <0.01 25 400-1100 HPc+PIF polymer+liquid crystal 2000
C60 cells was sufficient to stimulate commercial
interest in organic solar cells. We note that a
further breakthrough towards economically at-
tractive organic devices has been achieved with
the introduction of the laminated device struc-
ture - see Chapter 6.
• Blends of MEH-PPV with CdSe nanocrystals
7
,
sandwiched between ITO and Al, have also
been used to fabricate solar cells. The devices
show EQE values up to 12%, V
oc
of 0.5V and
again the strong field dependence leading to a
fillfactor no larger than 0.26 [102].
These nanocrystals turn out to be good elec-
tron acceptors with the extra advantage that
their absorption properties can be tailored by
varying their size during the synthesis; for ex-
ample, upon changing their diameter from 6 to
2nm, the energy gap can be increased from 2.0
to 2.6eV. Apart from the small shunt resistant
that causes the field dependence, charge trans-
port between the nanocrystals seems to be the
limiting factor in these cells.
• More recently, blends of MEH-PPV and a pery-
lene dye (Per) as electron acceptor have been
investigated. The IV characteristics show that
a fill-factor as high as (44%) can be achieved
in blends despite the relatively low V
oc
of 0.4V
and an EQE of about 1% [62]. Phase sensitive
7
Both can be dissolved i.e. dispersed in chloroform
AFM images revealed the formation of a large
electron conducting network consisting of very
long (µm) Per crystals in addition to phase sep-
aration on a scale of 50nm upon annealing.
The surprising fact that the formation of such
large crystals does still allow the device to per-
form reasonably well, opens new possibilities
- particularly with respect to the use of non
polymeric semiconductors such as dyes which
are prone to form crystals in the solid state.
Indeed blends with 2µm long nanotubes have
been made - see below.
• A blend between a polymer (PPV) and multi-
wall carbon nanotubes (MWNT’s) has been
investigated with respect to its photovoltaic
properties in Ref. [3]. Although the MWNTs
were as long as 2µm (and only 10nm in di-
ameter) and show relatively high conductiv-
ity (8S/cm) the devices developed the highest
open circuit voltage in our list of blend devices
(0.9V) instead of the expected shorts in the
only about 200nm thin films.
This confirms the feasibility of fabrication of
working solar cells using materials containing
or forming large crystals - consistent with the
results and conclusions in Ref. [62] as men-
tioned above. The EQE however, is only about
twice as high as in the pure PPV device. This
small but significant improvement is considered
to be due to the high workfunction of the nan-
78 CHAPTER 5. BLEND LAYER DEVICES
otubes (5.1eV) and/or the increased interface
area due to the formation of an interpenetrat-
ing network.
• The first D/A blend between two dyes with one
showing liquid crystalline properties at room-
temperature has been published [200] and will
be discussed in detail in this Chapter.
• Blends with the low bandgap semiconductor
PIF allowed for the first time to extend the
spectral response of an organic D/A blend be-
yond the 1000nm range. The EQE however
is yet only modest (0.02%) and in one case
(EQE=7 · 10
−4
%) even smaller than the sin-
gle layer device. Both devices will be discussed
later in this Chapter.
For all these blend structures discussed above,
strong photoluminescence quenching has been
found, ranging from a factor of 5 for MEH-
PPV+CN-PPV blends to a factor of 10
4
for MEH-
PPV+C60 blends. This indicates that the vast ma-
jority (over 80%) of photoexcited excitons can reach
a dissociation site (D/A interface) within their dif-
fusion length and dissociate before they can recom-
bine under the emission of light (PL).
Hence, the dominant loss mechanism has to be
on the way of the charges to their respective elec-
trodes. The charge transport to the electrodes can
be affected by:
• inter-molecular barriers (e.g. long distances for
the hopping process)
• intra-molecular barriers (e.g. distortion of con-
jugation along the polymer backbone by e.g.
twisting of aromatic units)
• super-molecular structural defects (e.g. “dead
ends” of aggregates in blends)
• formation of blocking contacts at the electrodes
• charge carrier traps that can lead to recombi-
nation or built up space charges that limit the
current
Large flat aromatic molecules like perylenes,
phthalocyanines or ladder-type structures
8
like
LPPPT have a good chance to overcome the prob-
lems described in the first 3 points in the list above.
8
Polymers that are planarised through bridges between
i.e. aromatic units are said to resemble a “ladder” along
their backbone.
It is also noteworthy that low bandgap materials
are more likely to form non-blocking (ohmic) con-
tacts with metals simply because the workfunction
of the electrode material is less likely to lie within
the bandgap. On the other hand, trap levels in the
bandgap of a low bandgap polymer are likely rela-
tively close to both the HOMO and the LUMO so
that they can facilitate fast i.e. efficient recombina-
tion representing “deep traps”.
In this Chapter we investigate the photovoltaic
properties of blends comprising the materials dis-
cussed in Chapter 3.
The specific combinations of semiconducting ma-
terials has been chosen with the aim to accomplish
both charge transfer which would indicate a D/A
interface and a extended photoresponse covering at
least the visible range (400-800nm).
With the exception of Per (2.1eV, 600nm)
all components have low bandgaps ranging from
1.77eV (700nm) to 1.12eV (1100nm)and most of
them are flat aromatic molecules. We note that
HPc also exhibits liquid crystalline properties as
discussed in Chapter 7.
5.2 HPc+Per
In this Section we explore the possibilities to form
a D/A blend network in a photovoltaic device using
the two small soluble molecules (dyes) Per and HPc
- see Fig.5.2.
Both molecules have been investigated in both
single and double layer devices earlier in this thesis.
Although there are reports of dye doped
9
polymers
[276] and and mixtures of D/A pigments [264, 251],
we are not aware of any reports of photovoltaic D/A
blend structures using solely dyes before this work.
The reason for the lack of published work on this
interesting subject in literature is presumably re-
lated to difficulties in synthesising of soluble semi-
conducting molecules with the desired optical and
electronic properties. Presently, various soluble
derivatives of pigments (dyes) with promising pho-
tovoltaic properties can be purchased from suppli-
ers that specialise in chemicals for xerography e.g.
Syntec/Synthone in Germany.
Two well soluble small molecules like Per and
HPc are likely to mix on a molecular level to
form a solid solution rather than showing phase
9
Note that mixtures containing less than about 10% by
weight - sometimes even more - of the guest material are
often regarded as “doped” materials in the field of organic
semiconductor devices.
5.2. HPC+PER 79
Figure 5.2: Chemical structure of the two dyes
investigated in this Section: A perylene tetracar-
boxylic diimide with aliphatic
11
side chains (Per)
and a metal free phthalocyanine with aliphatic side
chains (HPc).
segregation
12
as usually observed between poly-
mers. Such an intimate mixture may then be re-
garded as a “new” material rather than a blend with
a large D/A interface.
However, it is not clear a priori to what extent
percolation paths are present in such solid solutions.
Charge carriers might be able to find their way to
the respective electrodes via intermolecular hopping
between like molecules even if they are not always
in direct contact with each other.
Our hope was that these specific dyes would allow
sufficient percolation of charges through molecular
clusters (small aggregates) since both show strong
aggregation as individual components. We know
that Per has a tendency to form crystals during
spincoating from solution or upon heat treatment
[62] while HPc is liquid crystalline at room temper-
ature and forms even aggregates in solutions that
quench the PL - see Chapter 7.
However, we did not know if the dyes still prefer
to form aggregates with like molecules
13
and if the
size of the aggregates would be on a suitable scale.
Too large aggregates for example would quench ex-
citons before they reach the D/A interface whereas
too small aggregates may not provide percolation
paths. In the following we try to shed light on these
questions.
12
Phase segregation in dyes often happens at higher con-
centrations resulting in the formation of crystals.
13
They might aggregate with the other material and/or
form exciplexes.
Experimental
PL efficiency measurements were performed using
an integrating sphere coupled to a CCD spectrom-
eter (Oriel) via a liquid light guide. Details of the
procedure can be found in Ref. [169].
Images were recorded using an atomic force mi-
croscope (NanoScope IIIa Dimension 3100) oper-
ated in the tapping mode which is located at and
provided by Cambridge Display Technology (CDT).
Figure 5.3: The absolute photoluminescence effi-
ciency of the Per+HPc blend film on a quartz sub-
strate plotted as a function of composition. The
excitation wavelength was 488nm (1.0mW light in-
tensity). The line shown represents a guide to the
eye. HPc does not fluoresce in the film due to ag-
gregation (π-stacking).
The 1:1 (weight) blend device was fabricated by
spin coating onto an indium tin oxide (ITO) coated
quartz substrate from chloroform solution contain-
ing 30mg/ml HPc and 30mg/ml Per followed by
sublimation of the Al electrode. IV characteristics
and photocurrent measurements were done in vac-
uum (< 10
−5
mbar).
The chemical structure of the dyes investigated
in this Section are shown in Fig.5.2. Per was used
as received from Synthec (Germany) while HPc has
been synthesised and purified by G.G. Rozenberg
and A. Lux as reported elsewhere [165].
The liquid crystalline properties of HPc are dis-
cussed in Chapter 7 and have been published in
Ref. [199]. We note that a double layer device with
a sublimed perylene derivative is discussed in Chap-
ter 4.
80 CHAPTER 5. BLEND LAYER DEVICES
Figure 5.4: AFM images (phase contrast) of a 1:1 (by weight) blend of HPc and Per. Left: Scan size is
2x2 microns. Right: Scan size is 200x200nm. The lack of contrast suggests that there are no crystals or
aggregates between about 20-1000nm. Thus possible aggregates can only be smaller than about 20nm.
Results and Discussion
PL-Quenching
Photoexcited excitons can decay either radia-
tively showing photoluminescence (PL) or dissipate
their energy via direct conversion into heat. Fig.5.3
shows that in a film of 100% Per about 7% of the
excitons decay under emission of light. This ratio
can be considerably decreased upon the addition of
HPc. For example, a 50% Per blend shows virtually
no PL indicating that most excitons are affected by
the presence of HPc. The photoexcited excitons can
undergo several processes:
• recombination at the interface
• energy transfer to the material with the lower
bandgap followed by radiative or nonradiative
recombination
• charge transfer (exciton splitting) at the inter-
face
We consider the last process (charge transfer) to
occur in these devices since we found enhanced pho-
tocurrents in the blend compared to devices made of
the single components Fig.5.6. Moreover, the inter-
face of similar phthalocyanine and perylene deriva-
tives has been recognized as the active site for ex-
citon dissociation elsewhere [244, 267].
We note that the PL quenching shown in Fig.5.3
for Per+HPc is more than one order of magnitude
Figure 5.5: Linear absorption and EQE of the blend
device.
higher than in the 1:1 (by weight) blend between
MEH-PPV and CN-PPV [112]. Since the aggre-
gates in the latter are between 10 and 100nm, the
strong PL quenching in our dye blend is consistent
with the formation of aggregates on a scale <20nm.
In fact, if the dyes do not form aggregates at all, the
blend may be regarded as a solid solution (= blend
on a molecular scale) rather than a phase segregated
mixture.
5.2. HPC+PER 81
Figure 5.6: EQE of the HPc+Per blend device and
the single components. The blend show clearly en-
hanced EQE over the entire range indicating the
presence of a D/A interface and percolation paths
to the electrodes.
Absorption and EQE Spectra
The EQE and optical absorption spectrum of the
1:1 (by weight) HPc+Per blend cell are plotted in
Fig.5.5 showing that the cell is able to utilise light
from the entire visible spectrum. The absorption
band between 600nm and 850nm corresponds to
HPc whereas Per starts to absorb at around 650nm
reaching its peak at 500nm.
From the fact that the EQE spectrum follows the
absorption features we can infer that most of the
bulk of the cell contributes to the generation of the
external current. The EQE spectrum reaches its
peak (≈ 1 %) at around 500nm which exceeds the
EQE of a double layer cell made with very similar
materials (Chapter 4) by a factor of 2. The onset of
the EQE is around 900nm as in the single layer and
double layer devices comprising these components.
IV Characteristics
The (IV) characteristics in the dark and under il-
lumination are plotted in Fig.5.7. The open circuit
voltage is 110mV and the fill factor 25 percent un-
der these conditions. Both the dark and the light
current depend highly on the applied field. The de-
vice shows little rectification in the depicted voltage
range.
A voltage and light dependent device like a photo-
diode can be formed if exciton splitting occurs (here
Figure 5.7: IV characteristics of the HPc+Per blend
in the dark and under illumination with light at
650nm and an intensity of 0.25mW/cm
2
. The thick-
ness of this device was about 70nm.
at the D/A interface) and the built in field drives
the separated charges through percolation paths to
the different electrode materials.
Indeed, both the light and the dark IV curves in
Fig.5.7 show a weak but significant diode behaviour
under forward bias. However, the experiment re-
veals a strong linear voltage dependence under neg-
ative bias. Thus we conclude that the shunt is too
small which may be the case for the following rea-
sons:
1. A short has been formed through a pinhole that
results in a shunt resistor approaching the value
of the series resistor.
2. There is significant recombination of the sepa-
rated charges during their journey through the
fine interpenetrating network
Since we did not find any pinholes under the AFM
and the device is not particularly thin (70nm) we
may exclude the presence of shorts due to pinhole
formation.
Instead, the formation of a very fine network
may account for the observed strong PL quench-
ing, the enhanced photocurrents and the high se-
ries resistor
14
. Hence, the separated charge carriers
have many opportunities to recombine at this large
but presumably not very conducting D/A interface.
14
Which is caused by the “fine” percolation paths.
82 CHAPTER 5. BLEND LAYER DEVICES
This loss mechanism could be assisted by the pres-
ence of trap levels in the bandgap.
Summary
In this Section we have found that:
• Photovoltaic devices with a photoresponse
ranging from 400 to 800nm can be built us-
ing a blend between two soluble derivatives of
a perylene diimide and a phthalocyanine.
• The high solubility of both molecules appears
to facilitate the formation of a large D/A in-
terface - possibly even a solid solution - that
enables efficient exciton splitting.
• Such a very fine interpenetrating network
(<20nm )is probably also responsible for the
small shunt resistor since it offers increased
possibilities for recombination during the jour-
ney of the separated charges.
• The small shunt may be the main reason for
the relatively low V
oc
of 110mV and fill-factor
(0.25).
5.3 HPc+PIF
Experimental
Fig.5.8 shows the chemical structure of the organic
semiconductors used for the device fabrication in
this Section. The liquid crystalline properties of
HPc are discussed in Chapter 7 and have been
published in Ref. [165]. PIF has been synthe-
sised by H. Reisch and U. Wiesler at the MPI in
Mainz, Germany [210]. HPc has been synthesised
by G.G. Rozenberg and A. Lux, Melville Labora-
tory for Polymer Synthesis, Cambridge (UK) as re-
ported elsewhere [165].
We note that a double layer device with a sub-
limed perylene derivative is discussed in Chapter 4
of this thesis and published in Ref. [199].
The ITO/HPc+PIF(1:1 by weight)/Al devices
were fabricated by spin coating onto an ITO coated
quartz substrate from chloroform solution contain-
ing 30mg/ml HPc and 10mg/ml Per followed by
sublimation of the Al electrode. More details about
preparation of substrates and the setup can be
found in Chapter 10.
Figure 5.8: Chemical structure of the two organic
semiconductors investigated in this Section: A
poly-indenofluorene with short aliphatic side chains
(PIF) and a metal free (M=H
2
) phthalocyanine
(HPc) with thio-alkyl side chains (R = C
12
H
25
).
While HPc is very flat, the short distance between
the repeat units of PIF causes a 33
o
distortion be-
tween adjacent indenofluorene moieties [210].
Results and Discussion
Absorption and EQE Spectra
The EQE and optical absorption spectrum of the
HPc+PIF blend cell are plotted in Fig.5.9.
The features around 440, 680 and 720nm are due
to the HPc absorption whereas the peak at 800nm
can be found in the absorption of PIF - see also
Chapter 3. Both components have a small absorp-
tion coefficient around 550nm. However, since the
absorption onset is indeed around 850nm for HPc
and at 1040nm for PIF, the broad peak in the in-
frared (1040nm) is probably due to enhanced ag-
gregation of PIF in the presence of HPc rather than
poor background correction or interference effects in
the absorption measurement. Another reason why
the latter can be ruled out is that we observe a sig-
nificant photocurrent in the absorption onset.
We point out that the absorption range of this
device is larger than the range of the spectrometer
using a silicon based CCD array.
The EQE spectrum follows the absorption fea-
tures very well indicating that the entire bulk - in-
volving both components - contributes to the gen-
eration of the external current. However, the EQE
does not reach more than 6.5· 10
−4
% which is lower
than in devices comprising the pure components
only: 3 · 10
−3
% in HPc as well as in PIF - see
5.3. HPC+PIF 83
Figure 5.9: Linear absorption and EQE of the
HPc+PIF blend device.
Fig.5.10.
The following reasons may be considered to ex-
plain such low external quantum efficiencies:
1. Pinholes in the device that can cause shorts
when filled with e.g. Al during the sublimation
of the electrode.
2. A defect-rich charge transport network in the
blend. Such defects could be “dead ends”, iso-
lated areas, accumulation of a homolayer on
the wrong electrode. In addition, the blend
ratio of 1:1 (weight) may be far from the per-
colation threshold
15
.
3. The offset between the LUMO and HOMO lev-
els between PIF and HPc is not larger than the
exciton binding energy
16
which is required to
separate the charges i.e. to enable photoex-
cited charge transfer at the D/A interface.
4. The formation of new recombination channels
for the photoexcited species upon mixing the
two components.
ad 1): Shorts caused by e.g. pinholes in the or-
ganic film can be ruled out since the IV character-
istics shows very low currents in the dark - even
under bias voltage (Fig.5.11).
ad 2): In MEH-PPV+CN-PPV blends the EQE
values are at least two orders of magnitude higher
than the devices made of the single components -
15
which is difficult to estimate here
16
≈ 0.4eV according to Ref. [56, 168, 12]
Figure 5.10: EQE of the HPc+PIF blend device and
the single components. The blend shows a signifi-
cantly lower EQE than each component - indicating
that there is no charge transfer between HPc and
PIF.
despite serious network defects such as the forma-
tion of a homolayer of CN-PPV
17
near the wrong
electrode (ITO) [114].
In addition, network defects such as “dead ends”
cannot account for EQE numbers lower than the
single components because extra exciton splitting
should always lead to enhanced photocurrents com-
pared to the single layer. This is like introducing
oxygen in a single layer device using a hole transport
material - see Chapter 2. Molecular oxygen would
be the electron acceptor with probably no percola-
tion path at all - and is still better than the oxygen
free material. Hence insufficient network formation
cannot easily be blamed alone for the poor perfor-
mance of the HPc+PIF blend.
ad 3): The quinoid structure of PIF and the
known donor properties of phthalocyanine deriva-
tives suggest that the difference between the HOMO
and LUMO levels of HPc and PIF should be suffi-
cient to allow charge transfer. However, since both
materials do not show enough PL we cannot test for
PL quenching and have therefore no indication that
excitons are split at the interface between these two
molecules.
ad 4): In fact, the combination of low bandgaps
in both components and the (probably) different
electron affinities may bring the LUMO of PIF rel-
17
The more polar CN-PPV is attracted by the polar ITO
surface.
84 CHAPTER 5. BLEND LAYER DEVICES
Figure 5.11: IV characteristics in the dark (dashed
line) and under illumination with light at 700nm
and an intensity of 0.35mW/cm
2
. The thickness of
this device was about 40nm.
atively wide into the middle of the HPc bandgap.
PIF with its low LUMO may then capture an elec-
tron to act as a “deep trap” and allow recombina-
tion with a hole in the HOMO of HPc (electron back
transfer).
The mirror process, photoexcitation of an elec-
tron from the HPc directly into the LUMO of PIF
may then account for the new absorption band
around 1100nm. Note that a pure PIF film does
not absorb wavelengths longer than about 1000nm
- see Chapter 3 and [221]. Recombination via this
“deep trap”, PIF, could then even outnumber the
extra photoexcited charge carriers if exciton split-
ting actually occurs at the PIF/HPc.
IV Characteristics
The (IV) characteristics in the dark and under illu-
mination are plotted in Fig.5.11. The dark current
is small and relatively independent of the voltage
in both forward and reverse bias in the depicted
range
18
. Upon illumination the device generates a
small but significant photocurrent that can be in-
creased considerably (up to 15 times) by applying
an electric field. We can identify a threshold (≈
0.5V) under negative bias where the photocurrent
enhancement seems to begin. The device, however,
does not develop a significant open circuit voltage
18
This rules out the presence of a short - as discussed
above.
(<10mV).
The device also shows only little rectification in
the depicted voltage range.
Summary
In this Section we have found that:
• A working photovoltaic device comprising a 1:1
blend with low bandgap dye and a low bandgap
polymer has been built.
• The device shows a photoresponse in a wave-
length range as wide as 400 to 1100nm which
is virtually identical to commercial photo-
detector systems based on silicon.
• The blend between such low bandgap materials
with different electron affinities may open new
decay channels for the photoexcited species
that lead to EQEs that are lower than in de-
vices with only the single component. This
may be seen as a first hint for the lower lim-
its of the size of the semiconductor bandgap in
organic solar cells.
• The EQE can be tremendously increased by
applying -2V bias voltage.
5.4 PTV+Per
Experimental
The chemical structure of the organic semiconduc-
tors discussed in this Section are shown in Fig.5.12.
Per was purchased from Synthec (Germany) while
PTV has been synthesised by F. Goldoni, Univer-
sity of Technology (Eindhoven, The Netherlands).
Details of the synthesis, solvatochromic effects and
circular dichroism can be found in Ref. [90].
We note that single layer devices with both com-
ponents are discussed in Chapter 3 of this thesis.
The 1:1 (weight) blend device was fabricated by
spin coating onto an indium tin oxide (ITO) coated
quartz substrate from chloroform solution contain-
ing 15mg/ml PTV and 8mg/ml Per followed by sub-
limation of the Al electrode. More details about
preparation of substrates and the EQE/IV setup
can be found in Chapter 10. The thickness of the or-
ganic layer was measured with a Dektak profilome-
ter and found to be 80nm.
5.4. PTV+PER 85
Figure 5.12: Chemical structure of the two organic
semiconductors discussed in this Section: A poly-
thienylene vinylene with chiral aliphatic side chains
(PTV) and a perylene diimide with aliphatic side
chains (Per).
Results and Discussion
Absorption and EQE Spectra
The EQE and optical absorption spectrum of the
1:1 (by weight) PTV+Per blend cell are plotted in
Fig.5.13. The two peaks around 500nm correspond
to the absorption of a single Per layer whereas the
shoulder at 650nm is due to an absorption maxi-
mum of PTV. The relatively strong and broad band
between 800 and 1000nm is most likely an artifact
and probably related to the low signal to noise ratio
that affects the background correction.
However, since we find a significant photocurrent
down to about 900nm, there must be some absorp-
tion in that region. We note, that we find a more
pronounced - and probably real feature with a fine
structure in the absorption onset of the PTV+Ter
device discussed in a later Section of this Chapter.
The EQE reaches 0.44% in short circuit mode
and 8% under 3V negative bias. The EQE spec-
trum follows the main absorption bands. Since the
EQE values for single layer devices made of the pure
components PTV - see Fig.5.14 are clearly lower,
we can assume that charge separation takes place
at the PTV/Per interface. Interestingly, the EQE
spectrum of the blend extends further into the in-
frared than the EQE of the single components. This
indicates that there is a long absorption tail in the
blend film.
We consider that the presence of the rigid and
planar Per molecules may favour a more planar
arrangement of the PTV molecules which are ex-
Figure 5.13: EQE of the PTV+Per blend device
with 0 and -3V bias and its absorption spectrum.
pected to form a chiral superstructure
19
in the pure
film. Such a more planar molecular arrangement
may then lead to a lower bandgap due to increased
conjugation compared to the pure PTV molecule.
In addition enhanced aggregation or even crystalli-
sation may take place due to the flatter alignment
which has been suggested to be responsible for en-
hanced exciton splitting leading to enhanced pho-
tocurrents in the absorption onset in MEH-PPV
[120].
Figure 5.14: EQE of the PTV+Per blend device
with 0,-1.5 and -3V bias and the single layer device
efficiencies at 0V.
19
because of the chiral side chains [90]
86 CHAPTER 5. BLEND LAYER DEVICES
IV Characteristics
The IV characteristics in the dark and under illu-
mination are plotted in Fig.5.15. The open circuit
voltage is 150mV and the fill factor 0.25. The dark
current is very low and virtually field independent
over a wide voltage range as observed in many D/A
blend devices. The light current, however, is highly
field dependent following Ohm’s law between +3
and -3V.
What is remarkable here is that the shunt of the
dark IV curve decreases by at least
20
3 orders of
magnitude from 1.3 · 10
10
Ω to 1.3 · 10
7
Ω upon illu-
mination with light while there is no indication for
a significant decrease of R
s
.
Figure 5.15: IV characteristics of the PTV+Per
blend in the dark and under illumination with light
at 550nm and an intensity of 0.15mW/cm
2
. The
thickness of the organic film is 80nm.
From the observation that photoexcitation re-
duces predominantly the shunt rather than R
s
we
consider that a new recombination channel near the
D/A interface opens under illumination. The low
shunt can also explain the moderate open circuit
voltage (150mV).
The relatively high EQE (despite the low shunt)
suggests efficient exciton splitting at the D/A in-
terface. The linear dependence of the photocurrent
shows that the current is not trap limited. The
ohmic IV characteristics of the light IV curve does
not allow any rectification. We note that the recti-
fication ratios under light are also low for other or-
ganic semiconductor blends MEH-PPV/CN-PPV,
20
Note that the ohmic characteristics as obtained for the
light IV curve represents the sum of both Rs and R
sh
as
discussed in Chapter 2 so that R
sh
is possibly even smaller
than 1.3 · 10
7
Ω.
MEH/CdSe and MEH/C60 mixtures with low C60
ratios [112, 102, 83].
In any case, the strong field dependence of the
photocurrent can be used to increase the photo-
sensitivity i.e. EQE by a factor of 20 (reach-
ing about 8%) by applying either -3V (photodiode
mode) or +3V (photo-resistor mode).
Summary
In this Section we have found that:
• The external quantum efficiency of a photovol-
taic device comprising a 1:1 blend between a
small soluble perylene derivative (Per) and a
poly-thiophene derivative (PTV) is consider-
ably higher than the EQE of the pure single
layer devices.
• Thanks to the low bandgap of PTV the photo-
voltaic cell shows a very broad photoresponse
surpassing the visible wavelength range.
• The EQE can be increased by a factor of 20
(reaching 8%) by applying a negative or posi-
tive voltage of 3V.
• The distinct difference between the dark and
the light IV characteristics makes it possible to
operate this device not only as a photo-diode,
solar cell but also as light dependent resistor
(photo-resistor).
• In short circuit mode the solar cell reaches an
EQE up to 0.4% with V
oc
of 150mV
• The shunt resistor decreases by at least 3
orders of magnitude upon illumination with
monochromatic (550nm) light of 0.15mW/cm
2
.
5.5 PTV+Ter
Experimental
The chemical structure of the organic semiconduc-
tors used in this Section are shown in Fig.5.16. PTV
was synthesised by F. Goldoni at The University of
Technology (Eindhoven, The Netherlands). Details
of the synthesis can be found in Ref. [90]. Ter was
synthesised by S. Becker at the MPI (Mainz, Ger-
many). More information about the synthesis of
various perylene-, terrylene-, quaterrylene-, deriva-
tives including polymers can be found in Ref. [217].
Single layer devices comprising the pure com-
pounds have been investigated in Chapter 2.
5.5. PTV+TER 87
Figure 5.16: Chemical structure of the two organic
semiconductors discussed in this Section: A poly-
thienylene vinylene with with chiral aliphatic side
chains (PTV) and a terrylene diimide (Ter) with
aliphatic side chains.
The 1:1 (by weight) blend device was fabricated
by spin coating onto an ITO coated quartz sub-
strate from chloroform solution containing 15mg/ml
PTV and 6mg/ml Ter followed by thermal subli-
mation of the Al electrode. The thickness of the
organic film is 32nm. More details about prepara-
tion of substrates and the setup can be found on
Chapter 10.
Results and Discussion
Absorption and EQE Spectra
The optical absorption spectrum and EQE at 0V
and -1.5V of the 1:1 PTV+Ter blend cell are plotted
in Fig.5.17. Both PTV and Ter have very similar
absorption spectra with features at 600 and 660nm
and the onset at 750nm. The spectrum of the blend
in Fig.5.17 shows qualitatively the same features
but they are much more pronounced than in the
single compounds - see Chapter 3.
Also interesting is the occurrence of a broad ab-
sorption band between 780 and 1100nm where none
of the single components absorbs at all. We note
that we have observed a similar absorption band in
the PTV+Per blend as discussed in the previous
Section in this Chapter. The band here is different
from the one in PTV+Per in that the band here is
stronger and structured.
In fact, the same data plotted on an energy scale
(Fig.5.18)reveal a constant energetic spacing typi-
cal for vibronic modes. We found peaks at 1.237,
Figure 5.17: EQE of the PTV+Ter (32nm thick)
blend device with 0 and -1.5V bias and its absorp-
tion spectrum. The EQE at 0V is multiplied by 20
for clarity.
1.320, 1.402, 1.485, 1.567, 1.649, 1.732 which corre-
spond to an energetic spacing of 83meV. The two
main peaks are at 1.878 and 2.043eV with a spac-
ing of 165meV - which is 83 times 2. We consider
that these two peaks reflect the 0-0 and 0-1 vibronic
transition with the 0-0 peak being the dominant fea-
ture.
Figure 5.18: Linear absorption of the PTV+Ter
blend versus energy. Two vibronic modes can be
found: 165meV and 82meV.
A similar vibronic mode of 50meV in addition
to the dominant C-C mode of the polymer back-
bone (180meV) has been found in the absorption
88 CHAPTER 5. BLEND LAYER DEVICES
spectrum of the well defined poly-phenylene deriva-
tive (m-LPPP). Both the occurrence of such well
resolved vibronic structures and a dominant 0-0
transition have only been found either in small
molecules or well ordered conjugated polymers like
m-LPPP and PPV [198, 197, 163]. The latter had
to be especially synthesised [109, 110] or stretch ori-
ented [108] to obtain the required degree of order
i.e. maximum conjugation length
21
.
However, regardless of the vibronic structure such
a band in the absorption onset may be caused by
e.g.:
1. The introduction of sub-bandgap levels via un-
intentional doping by e.g. the presence of im-
purities.
2. Interference effects in the thin film.
3. Aggregation effects.
Since the band is very strong and does usually not
appear in films of comparable optical properties we
consider the first two possibilities (doping, inter-
ference) to be very unlikely. However, they could
be ruled out by performing photo thermal reflec-
tion (PDS) measurements as part of a more detailed
study of these types of blends.
We believe it is more likely that the π − π in-
teraction of the large planar Ter molecules favours
a more planar (less disturbed) molecular arrange-
ment of the PTV molecules thus leading to longer
effective conjugation lengths and enhanced aggre-
gation and the observed vibronic structure in the
polymer. Although this could explain the obser-
vation of both the well structured absorption spec-
trum and the broad absorption band around 900nm
further structural investigation - such as the one
mentioned above - are desirable to confirm such a
possibility.
From the fact that the EQE spectrum follows es-
sentially the absorption features we can infer that
most of the bulk of the cell contributes to the gen-
eration of the external current. However, in the
range between 400 and 500nm the EQE with 0V
increases despite the local decrease of absorption in
this range.
We note that we found the same effect in the
PTV+Per blend discussed in a previous Section. In
fact the pure PTV single layer shows a monoton in-
crease of EQE with shorter wavelengths despite a
local absorption minimum at 420nm - see Chapter
21
Different lengths of conjugated segments on a polymer
chain smear out the absorption spectrum - unless they are
all longer than the effective conjugation length.
Figure 5.19: EQE of the PTV+Ter blend device
with 0 and -1.5 bias voltage and the single layer
device efficiency of PTV at 0V. EQE data for the
pure Ter film may not be representative due to the
formation of shorts - see Chapter 3.
3. In other words, photons carrying higher ener-
gies contribute more efficiently to the photocurrent
already in the pure PTV (containing oxygen and
water) than photons with lower energy. Hence, al-
though we cannot identify the exact nature of this
effect, we consider that the PTV component is re-
sponsible for this effect.
Another interesting observation is that photons
absorbed in the low energy absorption band con-
tribute to the photocurrent the more, the higher
their photon energy (Fig.5.19). In fact, 1.55eV
(800nm) photons are almost 100 times more effi-
cient in contributing to the photocurrent than pho-
tons with only 1.38eV (900nm). Under negative
bias voltage the sensitivity to photon energy de-
creases considerably in both the high energy (400 to
600nm) and the low energy range (800 to 900nm).
IV Characteristics
The IV characteristics in the dark and under illumi-
nation are plotted in Fig.5.20. The device develops
an open circuit voltage of 120mV upon illumination
with light at 600nm.
The dark current is low and stays virtually con-
stant under negative bias whereas the light current
depends strongly on the applied field. Similar to
the PTV+Per device discussed earlier, the photo-
sensitivity can be enhanced by a factor of 20 by
5.6. PTV+THPF 89
Figure 5.20: IV characteristics of the PTV+Ter
blend in the dark and under illumination with light
at 600nm. The thickness of this device was about
32nm.
applying a voltage of -1.5V.
Note that strong electric field dependence depen-
dence of the photocurrent under negative bias is
often observed in blend devices - especially if the
open circuit voltage is relatively low (< 300mV at
typical light intensities around 0.2mW cm
2
). This
has probably to do with a tendency towards low
shunt resistors in blends since separated charges
have more opportunities to recombine within the
large D/A network.
In contrast to the PTV+Per cell the PTV+Ter
device shows some rectification in the dark but vir-
tually none under illumination.
Summary
In this Section we have found that:
• The external quantum efficiency of a device
comprising a 1:1 blend between a soluble terry-
lene derivative (Per) and a soluble polythieny-
lene vinylene derivative (PTV) is considerably
higher than in the pure PTV device.
• The device shows a photoresponse exceeding
the visible wavelength range (400-900nm)
• The photoresponse can be increased by a factor
of 20 reaching 2.4% at 670nm by applying a
negative voltage of 1.5V.
• The device can operate as a photodiode, solar
cell and a photo-resistor.
• The EQE reaches up to 0.4% with an open cir-
cuit voltage of 120mV.
• We find little rectification (5 at ± 1.8V) in the
dark but virtually none under illumination.
• The absorption of the blend shows a sharp vi-
bronic mode (165meV) with a dominant 0-0
transition indicating high molecular order i.e.
long effective conjugation of the PTV in the
presence of Ter.
We also found an extra absorption band at low
energies (1.2 to 1.8eV) with a vibronic mode
(83meV).
5.6 PTV+THPF
Experimental
Figure 5.21: Chemical structure of the two organic
semiconductors discussed in this Section: A poly-
thienylene vinylene with chiral aliphatic side chains
(PTV) and a fluorinated poly-yne, THPF.
The chemical structure of the organic semicon-
ductors discussed in this Section are shown in
Fig.5.21. PTV was synthesised by F. Goldoni at the
University of Technology (Eindhoven, The Nether-
lands). Details of the synthesis can be found in Ref.
[90]. THPF was synthesised by M. Khan, Imperial
College London. The synthesis has not been pub-
lished yet but some information about D/A interac-
tion in fluoro-substituted phenylene containing or-
ganic and organo-metallic poly-ynes can be found in
90 CHAPTER 5. BLEND LAYER DEVICES
Ref. [141]. The average molecular weight is 15440
corresponding to n=22 repeat units [142].
Single layer devices comprising the pure com-
pounds have been investigated in Chapter 3 of this
thesis.
The 1:1 (by weight) blend device was fabricated
by spin coating onto an ITO coated quartz sub-
strate from chloroform solution containing 15mg/ml
PTV and 16mg/ml THPF followed by sublimation
of the Al electrode. The organic film has been mea-
sured with a Dektak profilometer and found to be
50nm thick. More details about preparation of sub-
strates and the setup can be found in Chapter 10.
Results and Discussion
Absorption and EQE Spectra
The EQE and optical absorption spectrum of the
1:1 (by weight) PTV+THPF blend cell are plotted
in Fig.5.22. Both components have very similar ab-
sorption spectra with features at 600 and 660nm -
see also Chapter 3. They do not absorb light with
wavelengths longer than 800nm. Since there is no
photocurrent beyond 800nm we consider the band
around 900nm to be an artifact or due to interfer-
ence effects.
The low bandgap of THPF shows that the triple
bond is not breaking the conjugation of the back-
bone - thus charge transport along the backbone
should be efficient. A sufficiently thick layer of such
a blend can absorb 54% of the solar radiation on
earth - see Table 8.3.
Figure 5.22: Linear absorption and EQE of the
PTV+THPF blend device. The thickness of the
organic film is 50nm.
As observed in the pure PTV device and other
PTV blends the EQE increases for shorter wave-
lengths and reaches values greater than 0.1%
around 400nm.
The semi logarithmic plot of the EQE (Fig.5.23)
shows clearly that the photoresponse actually starts
at around 800nm and increases over more than four
orders of magnitude when reaching shorter wave-
lengths. The blend shows about 10 times higher
numbers than in the pure PTV device but only lit-
tle (2x) more than in the pure THPF cell. The
EQE numbers between 800 and 900nm are not sig-
nificant due to the low signal to noise ratio in this
region so that the actual bandwidth of operation of
this device is 400 to 800nm.
However, the small improvement of the EQE
compared to the pure THPF device may be ex-
plained in that charge separation is not particu-
larly effective at the interface between those two
molecules. This would mean that THPF is not a
very strong electron acceptor as expected because
of the electron withdrawing fluor atoms that couple
to the conjugated system.
It may be that the rather long distance between
the fluor atoms and the conjugated polymer back-
bone - they are separated not only by the thiophene
but also the pyrazine and phenyl units - weakens
the electronic effect. Such a weak effect of the fluor
atoms is also consistent with the observation of an
unchanged bandgap
22
regardless whether the fluor
atoms in this particular systems are present or not
[142].
Figure 5.23: EQE of the PTV+THPF blend device
and the single layer device efficiencies at 0V.
22
The bandgap in both systems was found to be around
1.7eV (730nm) in solution.
5.7. PTV+PIF 91
IV Characteristics
The IV characteristics in the dark and under illu-
mination are plotted in Fig.5.24. This device devel-
ops a relatively high open circuit voltage (550mV)
which is close to the difference in workfunctions of
the electrodes. Similar open circuit voltages have
been observed for e.g. MEH-PPV+CN-PPV blend
devices under comparable conditions [112].
Indeed, the open circuit voltage in blend devices
appears mainly to be determined by the workfunc-
tion of the electrodes and increases only very little
with light intensity [102]. The IV characteristics re-
semble the curves obtained for blends of PTV with
Per and Ter of the previous Section but with the ex-
ception that the (light) currents are lower and V
oc
is higher.
Figure 5.24: IV characteristics of the PTV+THPF
blend in the dark and under illumination with light
at 550nm and an intensity of 0.2mW/cm
2
. The
thickness of this device was about 50nm.
The shunt - calculated from the slope in the 3
quadrant - decreases considerably (from 10
10
Ω to
6· 10
8
Ω). Note that the (inverse) slope of the ohmic
IV curve in Fig.5.24 equals R
s
+ R
sh
. There is no
significant rectification in the investigated range (-
3V to +3V).
Another result often found in blends is the low
dark current even under higher negative bias voltage
- as opposed to the rather voltage dependent light
current.
Summary
In this Section we have found that:
• The external quantum efficiency of a device
comprising a 1:1 blend between THPF and a
polythiophene derivative (PTV) is about two
times higher than the EQE of the more effi-
cient single material device (THPF).
• The device shows a photoresponse covering the
entire visible wavelength range (400-800nm).
• The photoresponse can be increased by a factor
of 4 by applying a negative voltage of 2V.
• Thanks to a linear IV curve the device can op-
erate as a photodiode, solar cell and a photore-
sistor.
• The EQE reaches up to 0.2% with V
oc
of
550mV which is close to the difference of work-
functions of the electrodes.
• We find some rectification (5) at ±1.8V in the
dark but virtually none under illumination.
• The dark current is very small and voltage in-
dependent in contrast to the photocurrent.
5.7 PTV+PIF
Experimental
The chemical structure of the organic semiconduc-
tors used in this Section are shown in Fig.5.25. PTV
was synthesised by F. Goldoni at the University of
Technology (Eindhoven, The Netherlands). Details
of the synthesis can be found in Ref. [90]. PIF has
been synthesised by H. Reisch and U. Wiesler at
the MPI in Mainz, Germany [210].
Single layer devices comprising the pure com-
pounds have been investigated in Chapter 3 of this
thesis.
The 1:1 (weight) PTV+PIF blend device was fab-
ricated by spin coating onto an indium tin oxide
(ITO) coated quartz substrate from chloroform so-
lution containing 15mg/ml PTV and 8mg/ml PIF
followed by thermal sublimation of the Al electrode.
The thickness of the organic film was determined
with a profilometer (Dektak) and found to be 55nm.
More details about preparation of substrates and
the setup can be found in Chapter 10.
Results and Discussion
Absorption and EQE Spectra
The EQE and optical absorption spectrum of the
1:1 PTV+PIF blend cell are plotted in Fig.5.26.
The broad absorption band at 600nm stems from
PTV whereas the peak at 800nm with the band
92 CHAPTER 5. BLEND LAYER DEVICES
Figure 5.25: Chemical structure of the organic
semiconductors discussed in this Section: A poly-
thienylene vinylene with chiral aliphatic side chains
(PTV) and a poly indenofluorene (PIF).
between 900 and 1100nm is due to the absorption
of PIF - see Chapter 3.
Figure 5.26: Linear absorption and EQE of the
PTV+PIF blend device. The thickness of the or-
ganic layer is 50nm.
The onset of absorption of PTV is around 790nm.
Since the absorption spectra of the low bandgap
material PTV and the very low bandgap semicon-
ductor PIF complement each other, this device can
utilize light with wavelengths ranging from less than
400 to up to 1100nm.
A sufficiently thick layer of such a blend can ab-
sorb 77% of the solar radiation on earth which is
comparable to commercial silicon based photode-
tectors and solar cells - see also 8.3. As observed in
both the pure PTV and the pure PIF device as well
as other PTV blends the EQE increases for shorter
wavelengths reaching its maximum around 400nm
(0.014%).
Figure 5.27: EQE of the PTV+PIF blend device
and the single layer device efficiencies at 0V.
The semi logarithmic plot of the EQE (Fig.5.27)
reveals that the photoresponse of the blend really
starts at 1100nm. From the symbatic photore-
sponse we can exclude charge separation at the
ITO/blend interface. The figure also shows that the
EQE of the blend is higher than both pure compo-
nents only around 600nm.
Thus, we can conclude that photo-induced charge
transfer takes place at the D/A interface but there
must be another decay channel present that causes
the device to lower its performance - possibly wave-
length independent.
An alternative
23
and more likely possibility to ex-
plain the partial enhancement of the blend com-
pared to the single layer is the presence of partial
shorts i.e. a low shunt resistor. This would be con-
sistent with the results of the IV measurements -
see following Section.
IV Characteristics
The IV characteristics in the dark and under illu-
mination are plotted in Fig.5.28.
Although the device shows a more than signif-
icant photoresponse we believe that the device is
23
Another possibility is that the organic layer of PIF device
was close to its optimal thickness - whereas thickness effects
of the double layer here are not investigated or optimised.
5.7. PTV+PIF 93
Figure 5.28: IV characteristics of the PTV+PIF
blend in the dark and under illumination with light
at 550nm and an intensity of about 0.2mW/cm
2
.
partly shorted by e.g. pinholes for the following
reasons:
1. The dark current
24
is relatively high (more
than 20 times higher near 0V than in the other
blend devices discussed in this thesis). Also the
current is relatively voltage dependent in both
the dark and under illumination which results
indeed in a low shunt resistor R
sh
= 4 · 10
7
Ω.
We note that R
s
= 10
6
Ω (derived from the
slope in the first quadrant).
2. The open circuit voltage is expected to reach
about 0.5V (the workfunction difference be-
tween the electrodes), but is much lower here
(≈10mV)
Fortunately, the shunt resistor is still high enough
to avoid complete cancelling of the photogenerated
current so that at least a qualitative evaluation of
this blend is possible.
A double logarithmic plot of the dark IV curve
reveals tree distinct slopes (regimes)- see Fig.5.29.
The first slope is clearly Ohmic, the second is ∝ V
2
and the third ∝ V
3
. The quadratic dependence of
the current may be due to shallow traps while the
third slope may represent the transition to the trap
free condition - see also Chapter 2.
24
This is the current at 0.00V which is driven by potential
fluctuation of the source measure unit which are in the mV
range.
Figure 5.29: Double logarithmic plot of the dark IV
curve of the PTV+PIF device (forward bias only).
Three different slopes fit the curve suggesting the
presence of shallow traps - see text.
We note that the low shunt resistance requires
higher fields across the device to release the trapped
charges but should not affect this analysis otherwise
.
Summary
In this Section we have found that:
• The external quantum efficiency of a solar cell
comprising a 1:1 blend between PIF and a poly-
thiophene derivative (PTV) can reach at least
0.015% but can most probably increased con-
siderably if partial shorts can be avoided.
• The photoresponse is - at least partly - en-
hanced with respect to devices comprising only
the pure component suggesting photoexcited
charge transfer between the two components
• The device shows a very wide photoresponse
exceeding the visible wavelength range (400-
1000nm).
• We found evidence for the presence of shallow
traps from the analysis of the IV characteris-
tics.
94 CHAPTER 5. BLEND LAYER DEVICES
Chapter 6
Laminated Layer Devices
6.1 Introduction
In the previous chapters we have seen that single
layer devices can be considerably improved by in-
troducing a second semiconducting component re-
sulting in a D/A type solar cell.
Both types of D/A cells, double layer and blend
structures, have certain strengths but also disad-
vantages. For instance, the fabrication of a blend
device requires solubility of both components in the
same solvent - while the situation is reversed for
the double layer structure. Unfavourable dimen-
sions and/or defects of the interpenetrating net-
work structure are the main problem in blends
whereas short exciton diffusion lengths limit the
performance of double layers severely.
Moreover, both types would benefit a lot if there
was a possibility for separate treatment such as dop-
ing or heating of the specific D or A layer in order
to modify the electrical and optical properties. Re-
cently we have published a new device architecture
i.e. manufacturing method that allows to realise
all these possibilities in a surprisingly simple way
[94, 200, 95].
The basic idea is to deposit both the D and A
layer on separate electrodes (on substrates) before
they are laminated together to obtain a D/A double
layer device structure. The lamination process can
lead to an improved D/A interface when compared
to both a standard double layer and even a blend.
This is because molecules in each layer have the
opportunity to diffuse into the other layer during
the lamination process.
The quality of the D/A interface can be con-
trolled by heating, applying pressure or exposing
the layers to solution vapor before, during or after
the fabrication. The so obtained network structure
is less likely to result in both isolated areas and -
thanks to the separate charge transport layers - the
(unwanted) connectivity with the wrong electrode.
Figure 6.1: Scheme of the manufacturing method
and architecture used to obtain the laminated solar
cell. After lamination, charge-separation can oc-
cur at the interface of the hole and electron trans-
port layers. Here we investigate both a device com-
prising polymeric semiconductors and another using
dyes.
Both of which represent serious problems in normal
blend devices.
In the following we list the strong points of the
laminated structure:
1. Improved D/A network structure.
2. Relatively simple fabrication - from solution
applying the technologically well known lam-
ination technique.
3. Opportunity to treat each layer separately to
improve the electronic and optical properties
by e.g. doping or heat treatment.
4. Automatic encapsulation of the two semicon-
ducting layers between the substrates.
5. The quality of the D/A interface may be con-
trolled by treating the layers before, during and
after the lamination process
95
96 CHAPTER 6. LAMINATED LAYER DEVICES
Table 6.1: Survey of performance parameters of laminated solar cells.
EQE V
oc
FF range materials comments year
% V % nm – –
30 1.3 33 400-700 POPT/MCP polymer/polymer [94] 1998
0.2 0.37 23 400-900 HPc/Per liquid crystal(dye)/dye [200] 2000
Among the listed advantages of laminated devices
we want to emphasize that the possibility of sepa-
rate treatment of the D and the A layer represents a
crucial point if commercially interesting efficiencies
are to be achieved [277].
In practice, the following limitations using the
lamination technique have to be considered:
1. Both components have to be compatible i.e.
preferably dissolve in the same solvent and
form aggregates not larger than about twice
the exciton diffusion length - similar to the re-
quirements for a blend structure, see Chapter
5.
2. Both material need to be sufficiently“soft” to
enable lamination. Many polymers can be
heated above a certain temperature (glass tran-
sition) where they become soft and “rubber-
like”. Liquid crystalline phases are particularly
“soft”.
3. The curvature of the substrates can become im-
portant since the films are only about 100nm
thin and good contact over the entire area is de-
sired. Particularly flat ITO covered glass sub-
strates or flexible substrates are commercially
available and may be used to overcome such
problems.
4. Dust particles or others (crystallites..) between
the electrodes could deteriorate the quality of
the interface.
In the following sections we investigate first how
soluble polymers can be used to obtain a device
whose performance exceeds even the best pigment
based device in terms of solar efficiency. The second
device in this Chapter comprises the two soluble
small molecules HPc and Per.
A survey of the performance parameters of both
devices - which represent the only reported struc-
tures of this type in literature to date - is given in
Table 6.1. Note that - despite some advantages of
the used dyes over the polymers as discussed in Sec-
tion 6.3, the performance of the laminated dye de-
vice was surprisingly poor. In fact, it was even lower
than the blend structure using the same molecules.
6.2 The Laminated Polymer
Device
In this section we investigate a laminated device
structure using two polymers (Fig.6.2) that have
been chosen because of the following properties:
1. PL quenching experiments indicate exciton
splitting at the polymer/polymer interface - a
crucial requirement for building a D/A solar
cell.
2. Both polythiophenes and polyphenylene-
vinylenes have been showing good photovoltaic
properties within the class of conjugated poly-
mers [112, 213].
3. The bandgap of the spincoated film of POPT
can be lowered by about 100nm upon moderate
heat treatment. Together with MCP virtually
the entire visible range can be covered which
increases efficiency.
4. Both materials show suitable mechanical and
film forming properties together with the re-
quired thermal stability.
Experimental
Current voltage curves and photocurrents were
measured using a Keithley 237 source measure unit.
Monochromatic illumination was provided by the
output of a tungsten lamp dispersed by Bentham
M300 single-grating monochromator. Quantum ef-
ficiencies were determined by normalisation with
a calibrated silicon photodiode in the sample po-
sition. Absorption spectra were measured with a
Hewlett Packard 8453 UV-Vis spectrometer. Re-
flection losses were neglected for all measurements.
6.2. THE LAMINATED POLYMER DEVICE 97
The chemical structures of the polymer semicon-
ductors used in this work are shown in Fig.6.2.
MEH-CN-PPV or MCP is a fluorescent cyano sub-
stituted derivative of poly(p-phenylene vinylene)
with a large electron affinity as a result of the elec-
tron withdrawing cyano-groups [174].
The device was fabricated in a nitrogen-filled
glove box (oxygen<3ppm). One half of the de-
vice comprises the electron acceptor material(MCP)
spincoated from filtered (0.45µm) chloroform solu-
tion (5mg/ml) on Ca which was sublimed on a glass
substrate. The other half of the device comprises
the hole acceptor material (POPT) spincoated from
filtered (0.45µm) chloroform solution onto a con-
ducting polymer (PEDOT, commercially available)
film. PEDOT was spincoated from water onto a
transparent gold layer to decrease the sheet resis-
tance of PEDOT. Note that PEDOT and Au have
approximately the same workfunction.
The gold layer was sublimed on a glass substrate.
We added 5% (by weight) POPT to the MCP solu-
tion which has improved the efficiency considerably.
The half with the POPT layer was heated to 200
o
C
under vacuum before the device was laminated to-
gether by applying a gentle pressure while one half
was still at elevated temperature.
Heating of the polymer device prior to lamina-
tion was required for two reasons: 1) to introduce a
transition of POPT into the more crystalline phase
[18, 4] which absorbs more in the red region of the
spectrum 2) to exceed the glass transition temper-
atures so that both polymers become “soft”.
The total thickness of the semiconducting lay-
ers was about 80nm and the active area 2.5mm
2
.
PL-efficiencies were measured using an integrat-
ing sphere and an argon ion laser as described
previously[169]. The AFM images were obtained
using phase sensitive detection, and were taken with
a NanoScope IIIa Dimension 3100 (Digital Instru-
ments Inc, Santa Barbara).
Results and Discussion
PL Quenching
Fig.6.2 shows that the single components are fluo-
rescent and the PL of both is quenched upon addi-
tion of a small amount of one polymer to the other.
This indicates efficient exciton dissociation in both
polymers when blended.
Energy transfer between the blend partner as an
alternative explanation has been ruled out
1
since
1
Only in the unlikely case that energy transfer is followed
Figure 6.2: Chemical structures of the polymers
used in the polymer double-layer blend and PL ef-
ficiency as a function of the blend ratio.
the PL of both materials is quenched in the blend.
Such a ”bath tub” shape of the PL versus blend
ratio plot may therefore be taken as indication for
charge transfer at the polymer/polymer interface -
particularly if the electron affinity of one component
(here MCP) is higher.
Note that MCP with 45% PL efficiency is very
fluorescent. In other words, as much as 45% of
all excitons generated in MCP recombine radia-
tively in the pure film. Upon the addition of only
5% POPT the number of radiative recombination
is reduced to only 15%. This means a good deal
of the photoexcited excitons is now split into its
constituent charges with the holes trapped on the
POPT molecules and extra free electrons in the con-
duction band of MCP.
Note that PL quenching ratios of 100 to about
10000 have been reported for blends involving rela-
tively small molecules whereas quenching ratios in
polymers are typically (here:20) about one order or-
der of magnitude lower [112, 278, 62].
In order to sustain the photo-induced charge gen-
eration mechanism over time the trapped hole in
POPT eventually has to recombine with an elec-
tron. From the fact that we do observe “continuous
PL quenching” - not only in the first few ms - we
can infer that these trapped holes do find a recom-
bination partner such as e.g. the electrons in the
CB of adjacent MCP molecules.
by strong non-radiative recombination, a similar “bath tub”
shape may be obtained.
98 CHAPTER 6. LAMINATED LAYER DEVICES
Note that this charge generation mechanism is
not as efficient as the one at D/A interfaces of dou-
ble layer structures where both charges of an exciton
can travel to the electrodes. In fact the generation
mechanism here resembles more the situation as we
find it with oxygen traps distributed throughout an
organic semiconductor - as discussed in Chapter 2.
The main difference to oxygen induced charge
generation is that unlike the oxygen atoms, POPT
is a semiconductor with a sufficiently low bandgap
and can therefore also produce excitons. This re-
sults in the generation of extra charge carriers in
the host system - here electrons in MCP, which con-
tribute to the photocurrent and the reduction of the
series resistor.
Figure 6.3: Absorption spectra of the pure MCP
and POPT layer. Upon heat treatment the POPT
film can shift its absorption onset by about 100nm
towards the infrared, POPT

.
The described scenario is consistent with the ob-
served incomplete PL quenching for higher concen-
trations (e.g. 20%)of the guest molecule (Fig.6.2):
When the concentration of the guest material ap-
proaches a certain threshold, it becomes more diffi-
cult to get rid of the trapped charge since there are
more and more like molecules in the neighborhood
carrying the same trapped charges and no recombi-
nation partner.
The addition of 5% of MCP to the POPT layer
reduces the fraction of radiatively recombining ex-
citons from 5 to 3%. This is about the same PL
reduction factor as for the quenching of the PL of
pure MCP.
However, pure POPT shows already only 5%
photoluminescence. Thus, the relatively few extra
charges due to exciton splitting in the blend are not
expected to make much of a difference. Indeed, we
found that the addition of 5% MCP to POPT could
not improve the device performance [94].
Figure 6.4: External quantum efficiency and ab-
sorption spectrum of the laminated polymer cell.
Absorption and EQE Spectra
Fig.6.3 shows the absorption spectra of MCP and
POPT films. The absorption of the POPT film can
be considerably improved (broadened) upon ther-
mal annealing. This has been observed before in the
same molecule by Bergren et al. [18] and is proba-
bly related to a temperature induced structural re-
organisation of the regio-regular POPT molecules.
Fig.6.4 shows the linear absorption as well
as the EQE spectrum of the laminated device
(ITO/PEDOT/POPT+5%MCP/MCP/Ca). The
absorption spectra of the heated POPT and the
MCP film complement each other well so that rela-
tively high OD can be achieved over a wide spectral
range. The EQE follows the shape of the absorption
spectrum nicely so that the entire semiconducting
bulk can be regarded as active layer. The EQE is
generally high and reaches 28% near the peak of the
absorption (480nm).
Fig.6.5 shows the cross section of a laminated
structure: There is inter-penetration between the
two layers following the lamination and annealing
procedure, on a length scale of 20-30nm. The cut
to reveal the cross section was made at low temper-
ature. We note that finer scale inter-penetration is
not expected to be revealed in these images. These
(and the other) AFM images are shown in phase
contrast mode, most of which is due to differences in
viscoelastic properties of the two polymers; height
differences are no more than 5nm.
Fig.6.6 and Fig.6.7 show an in-plane image of the
two semiconducting polymers mixed with 5% of the
other component. Apart from the formation of the
6.2. THE LAMINATED POLYMER DEVICE 99
Figure 6.5: AFM phase contrast image of a cross-section of a laminated MCP/POPT+5%MCP polymer
interface without metal electrodes sandwiched between polyester substrates. An increased D/A interface
(outlined on the right picture) on a 40nm scale can be seen.
Figure 6.6: AFM phase contrast image of a
MCP+5%POPT film. The larger spots are most
likely islands of POPT clusters in the MCP matrix.
shown clusters or islands - finer scale mixing is sus-
pected.
We note that a comparable device that does not
contain 5% POPT in the MCP film showed only
EQE values around 5% which is comparable to
“conventional” polymer blend devices [112].
Figure 6.7: AFM phase contrast image of a
POPT+5%MCP film. The larger spots are most
likely islands of MCP clusters in the POPT matrix.
IV Characteristics
The IV characteristics of the laminated polymer
cell with ITO and Al contacts in both the dark and
under illumination is shown in Fig.6.8. The open
circuit voltage of 600mV is comparable to good dou-
ble layer devices and close to the difference of work-
functions of the electrodes.
100 CHAPTER 6. LAMINATED LAYER DEVICES
Figure 6.8: IV-Characteristics of the laminated
polymer cell with ITO and Al contacts in the dark
(dashed line) and under monochromatic illumina-
tion with λ=480 nm, illuminated area = 2.5mm
2
,
irradiance = 35mW/cm
2
(solid line).
From the slope in the first and third quadrant
we obtain an upper limit for the series resistor of
7· 10
4
Ω and a shunt resistor of approximately 10
7
Ω.
These resistor values are in the order of inorganic
cells of comparable dimensions - see Chapter 2.
Hence, the both the high photocurrent and open
circuit voltage with a relatively good FF - is not sur-
prising. However, it is the shunt and not the series
resistor that limits both the photocurrent and the
fillfactor in this device. Interestingly, the current
in the region between -1 and -1.5V shows particu-
larly little voltage dependence for which we have no
conclusive explanation yet.
The IV characteristics of the laminated polymer
cell with Au/PEDOT and Ca contacts in both the
dark and under illumination is shown in Fig.6.9.
The open circuit voltage reaches now 1.3V whereas
the photocurrent and the fillfactor (0.33) remains
the same as in the ITO/Al device. The light
characteristic can be fitted well with the Schock-
ley equation (see Chapter 2) using R
s
= 1 · 10
5
Ω,
R
sh
= 2 · 10
7
Ω, I
L
= 114nA, I
0
= 6pA and n=5.8.
Both resistor values are only little better than
in the laminated polymer device with ITO and Al
contacts. However, the open circuit voltage is con-
siderably higher and the shape of the IV curve is
now clearly dominated by the two resistor values
(R
sh
and R
s
). The high value for n (and the low
I
0
) was necessary to fit the high V
oc
.
The dark characteristics remains unchanged and
Figure 6.9: IV-Characteristics of the laminated
polymer cell with Au/PEDOT and Ca contacts in
the dark (dashed line) and under monochromatic
illumination with λ=480 nm, illuminated area =
2.5mm
2
, irradiance = 35 mW/cm
2
(solid line). Cir-
cles represent data points while the solid line shows
the modeled curve using the Shockley equation - see
Chapter 2.
the increasing difference between light and dark
curve (associated with a not negligible R
sh
) allows
to double the EQE or spectral response if -2V are
applied. With such bias voltage at least 58% can
be achieved in the photodiode mode.
In Fig.6.10 we plotted the open circuit voltage
and the photocurrent versus intensity (irradiance)
of the incident light. The figure shows that the pho-
tocurrent increases linearly with light intensity over
at least 6 orders of magnitude. V
oc
also increases
with light intensity but slower and levels out around
2.2V. In fact, the slope of V
oc
can be fitted well with
V (E) = 2.1V + 0.206V · ln(E) (6.1)
if we ignore the offset value of 2.1V we can see
that Eq. (6.1) resembles the expression as derived
from the Schockley equation in Chapter 2:
U = n
kT
q
· ln(
I
L
I
0
+ 1) (6.2)
with kT/q=26mV. Using nkT/q = 0.206V we
obtain a diode ideality factor n=8.0 which is rea-
sonably close to the n=5.8 obtained from the IV
curve. The large offset value of 2.1V may be caused
by the driving force that can be associated with
6.3. THE LAMINATED DYE DEVICE 101
Figure 6.10: Dependence of V
oc
and I
sc
on the light
intensity (E) at λ=488nm for the laminated PE-
DOT/Ca polymer solar cell. I
sc
∝ E
1.02
while
V
oc
∝ ln(E) up to 20 mW/cm
2
and saturates then
around 2.2V.
the difference of workfunctions of the electrodes
(W
f,Au
= 5.1eV, W
f,Ca
= 2.9eV).
Under the assumption that I
sc
and V
oc
scale
with intensity in the same way for all wavelengths
we calculated a power-conversion efficiency for the
polymer solar cell under solar conditions on earth
(AM1.5d, 76mW/cm
2
) of 1.9 % - for details see
Chapter 8.
This number is remarkable since it is almost twice
as high as the hitherto best organic devices with
reported solar efficiencies [244] and was achieved
even without sublimation of an active layer.
Summary
In this Section we have shown that:
1. The lamination method allows the fabrication
of organic solar cells that surpass the power ef-
ficiencies of earlier organic device architectures
such as sublimed double-layers or blends, fea-
turing many technological advantages.
2. Selective treatment of the two laminated lay-
ers, for example, annealing or ”doping” of the
organic semiconducting components in a D/A
structure is now possible and can lead to high
EQE and V
oc
numbers.
3. The laminated polymer device can utilise light
over a broad spectral range (400-750nm).
4. The workfunction difference of the electrode
materials seems to determine the achievable
V
oc
in this device.
5. The shunt resistor rather than R
s
are respon-
sible for remaining EQE and fillfactor losses in
this device.
6. The electronic transport properties of the elec-
tron acceptor material could be improved con-
siderably upon mixing a small amount (5%) of
the donor material into it.
We note that the method of lamination is already
well established in (general) polymer & plastics in-
dustry and often realised as a roll to roll process
using self supporting sheets.
Nevertheless, before it comes to its first industrial
applications, such as large-area photosensitive ar-
rays for scanners and medical imaging etc., further
efforts to improve charge transport and absorption
are necessary and possible stability problems will
have to be addressed.
6.3 The Laminated Dye De-
vice
In this section we investigate a laminated device
structure using soluble small molecules (dyes) in-
stead of polymers. We have chosen the two solu-
ble derivatives of a metal-free phthalocyanine (HPc)
and a perylene diimide (Per) - see Fig.6.11.
A combination of these two materials looked par-
ticularly promising to achieve good performance
values using the new technique for the following rea-
sons:
1. Both materials are relatively small molecules
which dissolve readily in the same solvent.
Thus inter-diffusion after lamination may lead
to a very fine and therefore efficient interpene-
trating network sandwiched between to trans-
port layers.
2. Insoluble derivatives of these molecules have
shown good performance in both single layer
and double layer solar cells.
3. Both molecules have high light-fastness and
thermal stability.
4. The liquid crystalline properties of HPc may
favour the lamination process.
102 CHAPTER 6. LAMINATED LAYER DEVICES
5. Efficient PL quenching has been observed
suggesting efficient exciton quenching at the
dye/dye interface.
6. The photovoltaic properties of HPc as a sin-
gle layer may be improved by introducing more
charge free charge carriers by the addition of a
few % Per.
Experimental
Current voltage curves and photocurrents were
measured using a Keithley 237 source measure unit.
Monochromatic illumination was provided by the
output of a tungsten lamp dispersed by Bentham
M300 single-grating monochromator. Quantum ef-
ficiencies were determined by normalisation with
a calibrated silicon photodiode in the sample po-
sition. Absorption spectra were measured with a
Hewlett Packard 8453 UV-Vis spectrometer. Re-
flection losses were neglected for all measurements.
The chemical structures of the dyes used in
this work are shown in Fig.6.11. Per is com-
mercially available from Synthec/Synthone (Ger-
many) while HPc has been synthesised as re-
ported elsewhere[165]. The electron-accepting dye
(Per) was spincoated from chloroform solution (15
mg/ml) onto indium tin oxide (ITO) covered glass
substrate. The hole accepting dye (HPc) was spin-
coated from chloroform solution (35 mg/ml) onto a
50nm thick Al layer which was thermally sublimed
on glass. We also added 10 percent (by weight) Per
to the HPc solution, which - as discussed below -
increased the performance significantly. Both solu-
tions were filtered with 0.1mm PTFE filters.
The lamination was performed under gentle pres-
sure in a laminar flow box after exposing both
halves of the device for 15 min to chloroform va-
por.
Results and Discussion
PL Quenching
If exciton splitting occurs in a blend with at least
one fluorescent material, it is usually faster than its
radiative decay and can therefore be observed as
photoluminescence (PL) quenching.
Fig.6.11 shows that the relative high PL of pure
Per is almost completely quenched if the blend con-
tains more than about 20% HPc. This means that
virtually all excitons in the blend film are affected
by the presence of HPc molecules suggesting a good
Figure 6.11: Chemical structure of the dyes used
(R=C
12
H
25
, M=H
2
) and photoluminescence effi-
ciency as a function of the blend ratio. Radiative re-
combination of Per is strongly reduced for all blend
ratios indicating charge-separation at the HPc/Per
interface. HPc does not fluoresce in the solid state.
solution and/or a sufficiently long diffusion range
for excitons in Per.
The pure HPc layer does not show any detectable
PL at room-temperature. However, PL efficiency
of about 50% has been reported in a metal-free Pc
with alkoxy substituents at low temperature (4K).
The low temperature PL decreases with rising tem-
perature until a sharp drop occurs at the solid
to mesophase transition where the PL is virtually
quenched completely [21]. This sharp drop has been
associated with an increase of the exciton diffusion
length in the mesophase that enables more excitons
to reach quenching sites.
We note that our photovoltaic measurements of
cells involving HPc in different architectures sug-
gest exciton diffusion lengths of around 200nm at
room-temperature - see Chapter 3 and Chapter 4.
In addition, PR-TRMC measurements indicate very
high charge carrier mobilities in the crystalline and
the mesophase of HPc.
Because of the absence of PL in the solid state,
it is not possible to probe for quenching of the PL
of HPc upon addition of Per. As a consequence,
we cannot completely exclude that transfer of the
exciton from the high bandgap material Per to the
lower bandgap material HPc is taking place.
However, the investigation of the HPc+Per
blends in Chapter 5 indicates that charge transfer
6.3. THE LAMINATED DYE DEVICE 103
takes place at the interface of HPc/Per mixtures.
In the same Chapter we show AFM images that are
consistent with the formation of a very fine network
with these molecules. Note that, if the network is
too fine, the connectivity between molecules may
not be very good leading to a high recombination
rate i.e. low shunt and high series resistor.
Figure 6.12: External quantum efficiency (EQE)
and absorption spectrum of the laminated dye cell.
Absorption and EQE Spectra
The EQE of this device is plotted together with the
absorption spectrum in Fig.6.12. The cell is able
to harvest light from the UV to the near infrared
(900nm). The EQE spectrum is strongly antibatic
in the wavelength range where only HPc absorbs
(600-900nm) indicating the presence of a filter effect
due to a too thick HPc layer. The thickness of HPc
is about 400nm to avoid shorts in this soft material.
The EQE approaches 0.3% at the short wave-
length end of the visible range. This value is slightly
smaller than in a comparable double layer device
(0.42%) and only about a third when compared
with the standard blend device (1.0%).
IV Characteristics
The IV characteristics of the laminated dye cell is
shown in Fig.6.13. The slope in the first quadrant
can be associated with a series resistor of 0.5 · 10
8

whereas the inverse slope in the 3 quadrant gives a
shunt of 0.1 · 10
8
Ω. Although the value for R
s
can
only represent an upper limit (since the slope could
Figure 6.13: IV characteristic in the dark (dashed
line) and under illumination (solid line) of the lam-
inated dye device.
increase for higher voltages) the low fillfactor and
flat continuation of the curve after leaving the third
quadrant suggests that R
s
is truly large.
We note that according to the considerations of
IV curves in Chapter 2 a series resistor that is larger
than the shunt would dominate the entire IV curve.
The IV curve would then become completely linear
with a inverse slope according to the value R
s
+R
sh
.
Since this is not the case here we consider that
there is an effect of a counter-diode that increases
the slope in the negative voltage range. This would
also comply with the small fillfactor < 0.25. The
photocurrent to dark current ratio can be increased
by a factor three upon applying a bias voltage of
-1V.
Summary
In this Section we have shown that:
• Small soluble molecules (dyes) can be used to
manufacture a laminated solar cell.
• The efficiency could not be increased with re-
spect to neither the double layer nor the blend
cell.
• The laminated dye cell can harvest photons
ranging from 400 to about 900nm over the
entire visible range with EQE comparable
to good polymeric single layer devices like
ITO/MEH-PPV/Al.
104 CHAPTER 6. LAMINATED LAYER DEVICES
• The EQE can be increased by a factor 3 in the
photodiode mode by applying -1V bias voltage.
Chapter 7
Discotic Liquid Crystals as Solar
Cell Materials
7.1 Introduction
Molecules in the solid phase have virtually no op-
portunity to move and orient themselves with re-
spect to one another. They have a certain degree
of positional and orientational order that cannot be
changed in this phase. However, if the same solid
melts to a liquid both types of order are lost com-
pletely; the molecules can now move and tumble
randomly.
The situation is different when a solid melts to
a liquid crystal
1
: The positional order is then lost
- since the single molecules are free to move in the
same fashion as in a liquid - but at least some of
the orientational order remains or even increases.
Fig.7.1 illustrates how both orientational and posi-
tional order may be improved since the molecules
are free to move and to orient themselves. This
can be the case for disc shaped (discotic) aromatic
molecules since their strong π−π interaction acts as
driving force that favours aggregation in stacks i.e.
columns
2
. The disc like shape also favours hexago-
nal arrangement of the columns.
However, Fig.7.2 shows that the degree of or-
der within such a liquid crystalline phase can vary
and mesophases with a different degree of positional
order can form. The nematic phase for example
can occur with e.g. molecules that have one or
more atoms sticking out of the middle of the disc
which limits the achievable degree of positional or-
der. Note that even these molecules are still able
1
The first man who discovered this special phase of matter
was F. Reinitzer, an Austrian botanist. He noted the double
melting behaviour of cholesterol acetate and cholesteryl ben-
zoate [219]. Reinitzer collaborated with O. Lehmann who
was the first to use the term liquid crystal [159].
2
S. Chandrasekhar reported the first observation of ther-
motropic mesomorphism of a disc like molecule (benzene-
hexa-n-alkanoate) [44].
Figure 7.1: Schematic illustration of the solid, liq-
uid crystal, and liquid phases that can be formed by
disc-shaped molecules. The degree of order of the
solid phase depends on its (thermal)history. The
liquid crystalline phase allows self-organisation of
the molecules in a limited temperature range (dis-
cotic phase).
to sustain a certain degree of orientational order
since all discs can still arrange such that they face
the common substrate plane - which is sufficient to
distinguish the phase clearly from a liquid.
In 1994, Adam et. al. reported high charge car-
rier (hole) mobility in the liquid crystalline phase
of a discotic molecule (hexa-hexyl-thio tripheny-
lene, HTTP) based on time of flight (TOF) mea-
surements [2]. The mobility in HTTP increased by
a factor of 10 at the transition from the liquid to
the hexagonal discotic phase and gains further two
orders of magnitude in a higher ordered discotic
phase.
With the exception of organic single crystals [139]
the reported mobility (0.1cm
2
V
−1
s
−1
) for photo-
induced charge carriers of HTTP was higher than
105
106 CHAPTER 7. DISCOTIC LIQUID CRYSTALS AS SOLAR CELL MATERIALS
Figure 7.2: Schematic illustration of three different
types of liquid crystalline discotic phases. The or-
dered columnar phase is expected to show the best
transport properties.
for any other organic system reported at that time.
The distance between adjacent discs in a column
was found to be 3.6
˚
Awhereas the distance between
the columns was 21.7
˚
A.
As a consequence, charge (and exciton) transport
is expected to occur preferably along the columns
whereas the mobility in directions perpendicular to
column axes can be more than 500 times smaller [23]
since the excited species have to tunnel through a
saturated hydro-carbon mantle (the insulating side
chains). Such columns may be used to transport
efficiently not only charge carriers but also excitons
and ions [262].
The discotic phases may even be frozen in, sim-
ply upon suitable cooling or become temperature
independent if the molecules are polymerised i.e.
cross-linked in the mesophase [262]. Moreover, both
hole and electron transporting discotic liquid crys-
tals can already be prepared [23].
Columnar discotic molecules (triphenylenes) have
already been used in LEDs where they showed
strong reduction of the onset voltage upon perpen-
dicular alignment of the columns [53]
Because of these interesting opportunities molec-
ular wires made of discotic liquid crystals are con-
sidered a promising alternative to conjugated poly-
mers which can also be seen as molecular wires since
transport occurs most efficiently along their conju-
gated backbone.
The HTTP molecule in which the high mobil-
ity was first measured is a relatively small conju-
gated system and absorbs not much light in the
visible range. Larger aromatic molecules like ph-
thalocyanine pigments with their strong absorption
and high stability despite the low bandgap have
been proven very successful in the past as semi-
conducting layers in photovoltaic cells and therefore
have been intensely investigated over many decades
[268, 43, 29, 154, 231, 267, 59, 67].
A. Lux and G.G. Rozenberg have synthesised ph-
thalocyanine derivatives which carry 8 thio-alkyl
side chains (Fig.7.3) to facilitate solubility and pos-
sibly liquid crystallinity [165]. These particular
derivatives are subject of our investigation in this
Chapter.
Figure 7.3: Chemical structure of the liquid crys-
talline phthalocyanines studied in this thesis.
Although their monomers are well known in liter-
ature, these soluble and probably liquid crystalline
phthalocyanines have either not been synthesised
before or not investigated in terms of their liquid
crystalline and photovoltaic properties. CV data
and aggregation effects have been reported for sim-
ilar derivatives in Ref. [180]).
In order to exploit the full potential of our
molecules in photovoltaic devices it was important
to know the ”mechanical” properties such as film
forming properties, solubility but also the temper-
ature range of the mesophases. Only then can we
orient the columns such that their axes points per-
pendicular to the electrodes.
Thus a hotstage with a polarisation microscope
and video camera to record images of the developed
textures was set up and access to the equipment for
essential thermal characterisation (TGA, DSC) of
other research groups has been arranged.
After a description of the experimental details in
the next section we present and discuss the results
of the liquid crystalline properties. The obtained
information is then used to fabricate two double
7.3. LIQUID CRYSTALLINE PROPERTIES 107
layer devices, both comprising a liquid crystalline
phthalocyanine (HPc) with a perylene pigment as
electron acceptor. One device was heat treated to
orient the columns perpendicular to the electrodes
while the other one was kept at room tempera-
ture. These results are discussed in Chapter 4 which
treats double layer devices structures.
In this Chapter we investigate the photovoltaic
properties of a simple device comprising a single
layer of the liquid crystalline semiconductor sand-
wiched between ITO and AL.
7.2 Experimental
Current voltage curves and photocurrents were all
measured in air using a Keithley 237 source mea-
sure unit. Monochromatic illumination was pro-
vided by the output of a tungsten lamp dispersed
by Bentham M300 single-grating monochromator.
External quantum efficiencies were determined by
normalisation with a calibrated silicon photo diode
in the sample position. A more detailed descrip-
tion of how we calculated the EQE can be found in
Chapter 10. Absorption spectra were taken using
a Hewlett Packard 8453 UV-Vis spectrometer. Re-
flection losses were neglected for all measurements.
The chemical structure of the dyes used in this
section are shown in Fig.7.3. Information on the
synthesis of the different phthalocyanine’s can be
found in Refs. [165, 269, 105, 73]. All materials
were purified carefully by column chromatography
and the structures established by IR, NMR and
mass spectroscopy. HPc was spincoated from chlo-
roform solution (90 mg/ml for the thick (450nm)
and 30mg/ml for the thin (80nm)layer) onto either
an ITO covered quartz substrate to built a device or
a spectrosil substrate for an optical measurements
(polarisation microscopy). The solution was filtered
with a 0.1µm disposable PTFE-filter.
The spin-speed for the thick layer was 1600rpm
and 2000rpm for the thin film. The Al top contact
for the two devices was obtained via thermal sub-
limation. The oriented HPc film was obtained by
heating to 292
o
C (20K/min) under constant nitro-
gen flow and then slowly (5K/min until 270
o
C, then
20K/min) cooled down to room temperature using
the hotstage (Linkam).
DSC and TGA measurements were performed
with a Perkin Elmer DSC7 at a heating rate of
10K/min. In order to perform polarisation mi-
croscopy, an adapted Vickers microscope has been
equipped with a hotstage (Linkam) and a set of po-
larisers.
Table 7.1: Phase transition temperatures (
o
C) of
a series of liquid crystalline phthalocyanine deriva-
tives. All temperatures have been determined us-
ing DSC scans and there was no contradiction with
the results from polarisation microscopy. Brackets
indicate transitions upon cooling. K=crystalline,
D=discotic=liquid crystalline, I= isotropic (liquid).
semicond. molecule K D I
H
2
PcSC
12
• 20 (4) • 292 •
CoPcSC
12
• 52(20) • – •
CuPcSC
8
• 79(-) • – •
CuPcSC
12
• 63(9) • – •
CuPcSC
18
• 52(45) • – •
7.3 Liquid Crystalline Prop-
erties
As a result of our investigations of these phthalo-
cyanines in the hotstage under the polarisation mi-
croscope together with DSC analysis, we could de-
termine the transition temperatures as summarised
in Table 7.1. Since all mesophases were found to
be very viscous at lower temperatures not all DSC
peaks could be confirmed under the polarisation mi-
croscope.
Figure 7.4: DSC trace of HPc. The peaks at 292
and 288
o
C indicating the transition to the liquid
phase are clearly visible here - which is fairly rare
for discotic phthalocyanines.
The transition temperatures of both the crys-
talline (K) to the discotic phase (D) and the discotic
(=liquid crystalline) to the isotropic liquid (I)
3
of
3
Which were in fact only found for HPc.
108 CHAPTER 7. DISCOTIC LIQUID CRYSTALS AS SOLAR CELL MATERIALS
the octakis(alkylthio)- substituted phthalocyanines
investigated here are generally lower than those of
their analogues with alkoxy side chains [73] and de-
crease even further with increasing chain length as
reported for similar derivatives in Ref. [253].
Figure 7.5: The image on the top (A) shows the
digitate stars as they appear under parallel polar-
izers upon cooling from the liquid phase of HPc.
Image B shows the same section but with crossed
polarizers which reveals the linear birefringent de-
fects that indicate perpendicular orientation of the
columns. Both pictures were taken at 280
o
C after
cooling from the liquid phase.
Note that even lower transition temperatures can
be achieved generally by replacing the thio alkyl
side-chains with alkyl or alkoxy chains with ethy-
lene oxide units [152].
For HPc the K→D transition temperature is re-
duced to such an extent that the mesophase can
form already at room temperature (20
o
C). It is also
the only material for which the transition to the
isotropic liquid phase - the clearing temperature -
could be detected by DSC (see Fig.7.4).
Figure 7.6: Proposed molecular arrangement of dis-
cotic molecules to explain the formation of birefrin-
gent defects: The main deformation mode is a bend
mode [253]. The observation of such defects may be
taken as indication for perpendicular alignment of
the columns with respect to the substrate plane.
For all other materials the mesophases are stable
up to about 290-350C (see also the TGA results be-
low) where they start to decompose before a liquid
phase can be detected.
Upon cooling HPc from the isotropic liquid at a
rate of 10K/min, digitate stars appear under paral-
lel polarizers (Fig.7.5 A). Such a texture has been
observed for alkoxy-substituted PCs [253, 152].
As they grow they coalesce to a homeotropi-
cally aligned sample whereby some linear defects
are formed, Fig.7.5. These appear as bright bire-
fringent lines in the dark background under crossed
polarisers.
The origin of these linear defects has been pro-
posed in Ref. [253] whereas other defects are stud-
ied in Refs [31, 30]. The birefringent lines can be
the results of a deformation mode as illustrated in
Fig.7.6.
Heating of a sample of the cobalt derivative
(CoPcSC12) to 280
o
C revealed another, more com-
mon, characteristic texture of a columnar phase, the
focal conic texture, see Fig.7.7. The shown texture
was retained during cooling down to room temper-
ature.
The thermal stability of the phthalocyanines was
investigated by thermogravimetric analysis (TGA)
at a heating rate of 10K/min, see Fig.7.8. For ex-
ample, initial decomposition of CuPcSC12 under
nitrogen atmosphere occurs at about 320
o
C, then
7.4. PHOTOVOLTAIC PROPERTIES OF HPC 109
Figure 7.7: The focal conic texture of CoPcSC12
as it is observed under crossed polarizers at 280
o
C
after cooling from the liquid phase.
the main decay takes place at 380
o
C. The mass loss
around 400
o
C corresponds exactly to the relative
mass loss of the eight side chains.
In air, see Fig.7.8, the temperatures for initial and
main decomposition are decreased by about 40
o
C.
We note that heating and cooling of both the de-
vice samples and samples for optical microscopy was
typically done with the hotstage purged with ni-
trogen to minimize possible degradation effects at
higher temperatures.
The absorption spectra of the metal-free phthalo-
cyanine in chloroform shows two intense transitions
(Q bands) at 702 and 730 nm (see Fig.7.10) which
are about 30 nm red-shifted compared with their
alkoxy-substituted analogues [253].
The Q bands appear generally less intense and
are blue-shifted in more polar (ethyl acetate) as well
as in less polar (toluene, n-hexane) solvents due to
formation of aggregates[166].
Photoluminescence is only detectable in very
good solvents such as chloroform and only at con-
centrations lower than 10
−3
mg/ml since aggrega-
tion quenches the fluorescence drastically. Fig.7.10
shows both PL and absorption of such a dilute so-
lution.
We note that in the following section we use HPc
at room temperature in a single layer solar cell
whereas double layer devices comprising a LC phase
where the columns are oriented parallel to the sub-
strate are investigated in Chapter 4.
For all liquid crystalline phthalocyanines the
intra-molecular charge transfer step occurs on a
pico-second time scale even in the liquid crystalline
Figure 7.8: TGA of a series of phthalocyanine with
different chain lengths measured under nitrogen.
phase [57]. Although the PR-TRMC method does
not allow to differentiate between mobilities of elec-
trons and holes, TOF measurements on discotic ma-
terials identified the hole as majority charge carrier.
From the fact that range of mobilities for a large
variety of side chains spans less than an order of
magnitude can be inferred - as expected - that the
aromatic core of the phthalocyanine mainly deter-
mines the transport properties.
7.4 Photovoltaic Properties
of HPc
Absorption and EQE Spectra
Fig.7.11 shows the linear absorption spectrum of
an ITO/HPc/Al device (80nm thick) when illumi-
nated through the ITO contact as well as the linear
absorption spectrum. The Q band here is consid-
erably smoother and absorbs stronger in the long
wavelength range compared to the one in dilute so-
lution in Fig.7.10. Such behaviour is usually as-
signed to strong aggregation as it is expected for
this mesogenic compound.
The onset of absorption is around 900nm which
corresponds to a semiconductor bandgap of 1.38eV.
A sufficiently thick layer of HPc can therefore ab-
sorb up to 64 % of the solar radiation on earth - see
table in Chapter 8.
The EQE in Fig.7.11 follows the absorption spec-
trum closely suggesting that the whole bulk of 80nm
contributes to the photocurrent when illuminated
through the ITO contact. The somewhat pro-
110 CHAPTER 7. DISCOTIC LIQUID CRYSTALS AS SOLAR CELL MATERIALS
Figure 7.9: TGA of CuPcSC12 under nitrogen and
air. Even in air the material is stable up to about
300
o
C.
nounced EQE in the absorption onset is probably
due to enhanced exciton dissociation in aggregates
as suggested in a study of photocurrent peaks in
MEH-PPV [120] and not a signature of an optical
filter effect.
The other EQE features coincide well with the
absorption
4
and there is no offset at the short wave-
length side of the Q band.
Fig.7.12 shows the EQE and absorption upon il-
lumination through ITO for a much thicker device
(450nm). The first absorption peak (0-0 transi-
tion) appears stronger than in the thin film. This
is more likely an artifact in conjunction with the
background correction during the absorption mea-
surement and therefore not significant.
However, here in this rather thick device, the
EQE shows the familiar antibatic behaviour as it is
usually observed in p-type semiconductors between
ITO and Al when illuminated through ITO.
At the beginning of Chapter 3 we describe a
method how the maximum thickness d
opt.
of the
active region can be estimated from the absorption
and EQE spectrum using the formula:
d
opt.
=
OD(λ
e
)
OD(λ
0
)
· d
0
(7.1)
We have chosen to take the required input param-
eters from Fig.7.12 (thick device) since the shape
EQE of the thin device does not allow a clear de-
4
Note that a slight offset may be caused by a non hori-
zontal baseline in the EQE spectrum.
Figure 7.10: Absorption and photoluminescence of
a very dilute solution of HPcSC12 in chloroform.
Figure 7.11: The EQE of the thin HPc device (in
air) follows the absorption features when illumi-
nated through ITO.
termination at what wavelength we ought to take
the value for OD(λ
e
).
From Fig.7.12 we extracted the following data:
OD(λ
e
) = 0.55 ± 0.2, OD(λ
0
) = 1.24 ± 0.01 and
d
0
= 450 ±30nm). Using Eq. (7.1) we obtain:
d
opt
= 200 ±80nm (7.2)
The large error for OD(λ
e
) is due to the steep
slope and the relatively broad peak in the EQE
spectrum of the thick device. To avoid the oc-
currence of a filter effect we suggest not to exceed
120nm in this material.
Note that the absolute EQE values for both HPc
devices are surprisingly low - only about a tenth of
the EQE in e.g. a ITO/MEH-PPV/Al device - con-
sidering the large active range of at least 120nm for
7.4. PHOTOVOLTAIC PROPERTIES OF HPC 111
Figure 7.12: The EQE (solid line: 5hrs in air;
dashed line: in vacuum) of the thick HPc cell shows
antibatic behaviour when illuminated through the
ITO contact.
HPc. We believe that a considerably higher series
resistor due to a low concentration of free charge
carriers may cause the low photocurrent in HPc.
This possibility will be considered in the subsequent
paragraph focusing on the IV characteristics.
The EQE spectra of both HPc devices - see
Fig.7.14 and Fig.7.13, follow the absorption as ex-
pected if illumination occurs through the semitrans-
parent Al contact.
However, the EQE of the thicker device is then
about 3 times higher than the EQE of the 80nm
device. This suggests that the active region is actu-
ally around 240nm, favouring a larger value within
the error bars for the maximum active range as ob-
tained above.
According to the discussion in Chapter 2 the filter
effect in the thick device (ITO/HPc/Al, illumina-
tion through Al) can be caused either by a limiting
exciton diffusion range or a limiting electron mobil-
ity (or both). Although both the charge and exci-
ton transport are likely increased due to the strong
π − π interaction, the high value for the mobility

e
+ µ
h
) value obtained from PR-TRMC in Ref.
[57] do not allow to exclude the possibility that the
electron drift range is still a limiting factor in these
devices. Hence, we can only say that the exciton
diffusion range plus the electron drift range in these
devices is around 200nm±80nm.
We want to note that this number means that the
exciton can be over an order of magnitude larger
than in typical polymers and is largely consistent
with the suggested several hundred nanometers as
suggested in Ref. [21] for a very similar compound,
Figure 7.13: The EQE of the thin (d=80nm)device
(solid line), illuminated through the Al contact and
measured after 5hrs in air follows the linear absorp-
tion features (dashed line)
HPcOC12.
The lower absorption coefficient of HPc allows a
10nm layer of MEH-PPV to absorb as much light as
the 80nm HPc film. Thus, since both films are thin-
ner than their maximum active region they should
give approximately the same photocurrent i.e. EQE
numbers - provided all other parameters are com-
parable.
Charge Carrier Mobilities
In Table 7.2 we give a survey over published charge
carrier mobilities in organic semiconductors includ-
ing several liquid crystals. Recently published val-
ues using pulse radiolysis time resolved microwave
conductivity (PR-TRMC) reveals even higher mo-
bilities in the crystalline phase of many discotic
materials [57]. The results are promising in that
the compound we used for fabrication of solar cells
showed the highest mobility for the discotic phase
among all investigated compounds in this reference.
The latter includes other metal free phthalocya-
nines as well as porphyrins and triphenylenes with
various lengths and types of side chains [57]. In
the crystalline phase the HPc derivative with 8 in-
stead 12 carbon atoms gave the highest mobility
(0.54cm
2
V
−1
s
−1
).
However, note that HPcSC12 is in its liquid crys-
talline phase starts to form already at 20
o
C accord-
ing to our results whereas the K →D transition oc-
curs only at 77
o
C according to Ref. [57]. Provided
our number is true, significantly higher values for
the mobility were expected for slightly lower tem-
112 CHAPTER 7. DISCOTIC LIQUID CRYSTALS AS SOLAR CELL MATERIALS
Table 7.2: Survey of electron (e) and hole (h) mobilities of important organic and inorganic semiconduc-
tors. For abbreviations see Chapter 1. Σ denotes the sum of charge carrier mobilities.
material type mobility source/method comment
- cm
2
V s
−1
- -
Si-monocr. h/e 450/1500 [241] inorganic
Si-amorph e 0.1-1.0 [85] inorganic
HPcS8 (K-phase) Σ 0.54 PR-TRMC [57] liquid crystal
HPcS8 (D-phase) Σ 0.16 PR-TRMC [57] liquid crystal
HHTS6 (K-phase) h 0.26 PR-TRMC [57] liquid crystal
HHTS6 (H-phase) h 0.1 TOF [2] liquid crystal
HHTS6 (D-phase) h 0.007 PR-TRMC [57] liquid crystal
HPcS12 (K-phase) Σ 0.25 PR-TRMC [57] liquid crystal
HPcS12 (D-phase) Σ 0.22 PR-TRMC [57] liquid crystal
MgPc h 0.1 [87] pigment
PcOx Σ 2 · 10
−2
PR-TRMC [262] pigment
HPc h 1 · 10
−3
FET [13] pigment
PPV h 1 · 10
−4
TOF [168] polymer
PT(alkyl) h 9 · 10
−5
[85] polymer
PT h 1 · 10
−5
[85] polymer
Figure 7.14: The EQE of the thick (d=450nm) HPc
device (dashed line, in vacuum) increased when ex-
posed to air for 5 hrs (solid line). The device was
illuminated through the Al contact.
peratures since the value for the crystalline phase
was measured at room-temperature.
It is worth considering that the mobilities ob-
tained in PR-TRMC measurements usually refer to
the maximum that could possibly be achieved in a
two electrode device with a perfectly orthogonally
oriented mono-domain of a discotic material [57].
Hence, one may conclude that the sample of HTTP
in the TOF experiments of Ref. [2] must have been
close to perfectly aligned, with the columnar axes
orthogonal to the electrode surface and with com-
plete columnar integrity across the entire 30µ elec-
trode gap corresponding to as much as 10
5
macro
cyclic units - a really thin molecular wire.
We note that rapid charge transport along self as-
sembling graphite nanowires has recently been re-
ported in the discotic molecule tetrakis coronene
[58].
IV Characteristics
A numerical analysis of the IV curves in Fig.7.16
shows that both the series resistant and the shunt
of the HPc devices are considerably higher (R
s

2 · 10
9
Ω, R
sh
≈ 1 · 10
10
Ω under illumination
5
and
air) than for the MEH-PPV devices (R
s
≈ 2 · 10
6
Ω,
R
sh
≈ 4 · 10
8
Ω under illumination).
Thus, it is likely that the very large R
s
is respon-
sible for the low dark and light currents. From the
dark and light IV curves we can immediately see
that the extra charge carriers due to photoexcita-
tion indeed make a big difference.
7.5 Summary
In this Chapter we have shown that:
5
Resistor values in the dark are: Rs ≈ R
sh
≈ 3 · 10
11
Ω.
7.5. SUMMARY 113
Figure 7.15: IV characteristics of the thin HPc
device, after 5hrs in air, when illuminated (λ =
700nm)through the ITO contact (dashed line) and
in the dark (solid line).
1. It is possible to spin coat a monomer with a
considerably smaller molecular mass than poly-
mers like a phthalocyanine from non acidic so-
lutions to built a working photovoltaic cell
2. The maximum thickness of the active region
in a HPc film is estimated to lie around 200 ±
80nm. This value is over an order of magni-
tude larger than for many polymers including
MEH-PPV. Thus we can confirm that higher
mobilities for charge carriers and/or excitons
are possible in this material [57].
3. The low EQE values compared to e.g. MEH-
PPV may be due to the very high series resistor
in our devices. The reason for the high series
resistor is probably a low concentration of free
charge carriers.
4. The liquid crystalline phases of a series of ph-
thalocyanines with varying center atoms and
different lengths of side chains has been de-
termined. Only one compound forms a liquid
phase before decomposition. The same com-
pound HPc is liquid crystalline already at room
temperature (> 20
o
C).
We want to emphasize that already the high sol-
ubility of the phthalocyanine derivatives represents
an exciting novelty among organic solar cells with
the potential of considerably simple and cheaper de-
vice processing compared to the usual sublimation
of pigments.
Figure 7.16: The IV characteristics (λ = 700nm) of
the thick HPc cell shows a linear behaviour under
both forward and reverse bias when measured in
vacuum. After 5hrs exposure to air the conductivity
in forward direction increases showing to different
slopes. Dark conductivity is very low and constant.
Further studies are desired to investigate the pos-
sibilities to increase the conductivity of HPc by the
introduction of extra charge carriers via doping.
The EQE may be thereby be increased consider-
ably and make these materials also interesting can-
didates as active components in D/A systems. We
also propose to study devices with discotic mate-
rials which are not necessarily liquid crystalline at
room temperature but still well soluble.
114 CHAPTER 7. DISCOTIC LIQUID CRYSTALS AS SOLAR CELL MATERIALS
Chapter 8
Determination of Solar Cell
Efficiencies
8.1 Introduction
The ultimate method to test the overall perfor-
mance of solar cells is outdoor exposure. How-
ever, reliable and reproducible results can only be
obtained with solar simulators whose illumination
conditions are designed to match an internationally
standardised solar spectrum.
Commercially available solar simulators usually
use a xenon arc lamp in conjunction with a spec-
tral correction filter which approaches the terrestrial
solar spectrum such that it is good enough for many
applications [183].
Here we discuss to what extent a simulator based
on a quartz tungsten halogen-lamp (QTH) can be
used as a much cheaper (≈ 1/10) alternative. In
particular, researchers in the field of organic solar
cells may benefit from this alternative since their
cells are usually not optimised for high stability and
may therefore be reluctant to send them to an in-
stitute where a commercial simulator is available.
We also discuss and apply a numerical method
that allows estimation of the power efficiency num-
ber that would be obtained if the cell was exposed
to the solar light on Earth. The calculation is based
on spectral response and intensity dependence mea-
surements.
8.2 Solar Radiation
Radiant energy from the sun is vital for life on our
planet. It determines the surface temperature of the
Earth as well as supplying virtually all the energy
for natural processes both on its surface and in the
atmosphere. Every second the sun emits the energy
of 3.90·10
26
J into space. Considering that the mean
distance from Earth to sun is 150 · 10
9
m our planet
receives the energy of 1395J/m
2
, every second [66].
The sun is essentially a sphere of gas heated by a
nuclear fusion reaction at its center with estimated
temperature of up to 20 000 000K [97].
Figure 8.1: The path length of the solar radiation
through the Earth’s atmosphere in units of Air Mass
(AM) increases with the angle from the zenith. It
increases from unity for 0
o
(zenith) to 1.5 for 48
o
and 2.0 for 60
o
. The AM1.5 spectrum is the pre-
ferred standard spectrum for solar cell efficiency
measurements in literature.
However, this is not the temperature that deter-
mines the characteristic electromagnetic emission
from the sun since most of the intense radiation
from the sun’s deep interior is absorbed after a few
millimeters (near the center) and by a layer of neg-
ative hydrogen ions
1
near the surface.
1
This ion can exist because the single electron of the neu-
tral hydrogen atom does not completely screen the positive
proton [280].
115
116 CHAPTER 8. DETERMINATION OF SOLAR CELL EFFICIENCIES
Figure 8.2: The two most commonly used standard spectra, AM1.5 direct and AM1.5 global. The global
spectrum comprises the direct plus the diffuse sunlight. The difference is mainly in the shorter wavelength
region since molecular scattering scales with λ
4
. Water (H
2
O), ozone (O
3
) and carbondioxide (CO
2
) are
the main absorbers of the solar radiation in Earth’s atmosphere
Figure 8.3: Distribution of energy (in percent) within the AM1.5d spectrum. The Integral from 300 to
2500nm was taken as 100%. For example, silicon solar cells with a bandgap of 1.13eV (1100nm) can
maximally absorb 77% of the terrestrial solar energy. Taking into account the region with the highest
energy content (400-900nm, see values in the diagonal) a bandgap of 1.4eV is already sufficiently small
to absorb 61% of the terrestrial solar radiation or 80% of the value for Silicon.
8.2. SOLAR RADIATION 117
The accumulation of heat in this layer sets up
convective currents that transport the excess en-
ergy through the optical barrier into the so called
photosphere.
The temperature within the photosphere is
only 5800K and emits an essentially continuous
spectrum
2
of electromagnetic radiation closely ap-
proximating that expected from a black body at
this temperature [280].
This is the solar spectrum we receive outside the
Earth’s atmosphere which is also known as extrater-
restrial or Air Mass 0 (AM0) spectrum.
About 70% of the solar radiation in space can
reach the surface after penetrating Earth‘s atmo-
sphere. Apart from the reflection of clouds the rea-
sons for such attenuation are [86]:
• Scattering by molecules in the atmosphere
(Rayleigh scattering) which is most effective at
short wavelengths.
• Scattering by aerosols and dust particles
2
A fine structure appears due to absorption in the cool
peripheral solar gas(Fraunhofer lines)
• Absorption by the atmosphere and its con-
stituent gases but mainly oxygen, O3, H2O,
CO2
The actual degree of attenuation through the
Earth’s atmosphere is highly variable. It depends
mainly on two factors:
• Fraction of diffuse radiation scattered from the
sky (and clouds) and surroundings
• Length of the light path through the atmo-
sphere
The fraction of diffuse radiation received by a
horizontal surface can reach 10 to 20% even in clear,
cloudless skies. For completely covered skies most
of the radiation will be diffuse.
Hence, for days between the sunny and cloudy
extremes mentioned above, about 50% of the so-
lar light will be diffuse [97]. Diffuse sunlight gener-
ally has a different spectral composition from direct
sunlight - it will be richer in the shorter or “blue”
wavelengths.
The standardised terrestrial spectra for direct
and global sunlight are shown in Fig.8.2 where the
global spectra comprises the direct plus the diffuse
components.
The length of the light path through the atmo-
sphere also undergoes changes during the day time
- see Fig.8.1, but also depends on latitude and time
of year. In any case, the path is shortest when the
sun is directly overhead (zenith). The ratio of any
actual path length to this minimum value is known
as the optical air mass (AM). AM is unity when the
sun is directly overhead or larger according to
AM =
1
cos θ
(8.1)
where θ is the angle measured from the zenith.
Hence, when the sun is 48 degrees off overhead,
the radiation is AM1.5. The easiest way to estimate
the air mass in practice is to measure the length of
the shadow s cast by a vertical structure of height
h using:
AM =

1 + (
s
h
)
2
(8.2)
Consequently, as opposed to the situation outside
the Earth’s atmosphere, terrestrial sunlight varies
greatly both in intensity and spectral composition.
In order to allow meaningful comparison between
the performances of different solar cells tested at
different locations, a (preferably: terrestrial) stan-
dard has to be defined and measurements referred
118 CHAPTER 8. DETERMINATION OF SOLAR CELL EFFICIENCIES
to this standard. The most widely used terrestrial
standard at the time of writing are the AM1.5 direct
and AM1.5 global spectrum as shown in Fig.8.2.
8.3 Simulation of Solar Radi-
ation
Depending on the application, different parts of the
solar spectrum become more important. For ex-
ample, biological testing (sun screens, plant growth
etc.) requires accurate simulation of the ultravio-
let especially in the UV-B region, whereas for solar
energy conversion the range between the visible to
the near infrared light is the important part since
it contains most of the solar power - see Fig.8.3.
The emission spectrum of a 100W QTH-lamp op-
erating at 3200K colour temperature has an emis-
sion peak at 900nm - see Fig.8.4. It has a smooth
continuous output over 350 to 2000nm. In order to
determine solar efficiencies of organic solar cells we
focus on the range between 400 to 1000nm which
contains about 70% of the solar energy.
Figure 8.4: Emission spectrum of a 100W - QTH
lamp (solid line) and transmission spectrum of the
two coloured Schott-glass filters FG6 (6mm thick)
and KG4 (3mm thick) which are used to shift the
blackbody emission peak from 900nm (3200K) to
550nm (5800K) to approximate the AM1.5d spec-
trum.
8.3.1 Calculation of the Required
AM1.5d Filter
In order to approach the AM1.5d spectrum, the
emission peak at 900nm of the QTH lamp, operat-
ing at 3200K, has to be shifted towards 550nm by
Figure 8.5: The actually measured spectrum (solid
line) resembles the shape of the modeled spectrum
(full circles). However, there is a significant differ-
ence - which we investigated further - see text.
suppressing the IR emission using suitable optical
filter.
The transmission curves of the two selected
coloured glass filter are also shown in Fig.8.4.
The spectral composition X(λ) of the light from
the quartz halogen tungsten lamp QTH(λ) after
passing through the two filters F1 and F2 with
their transmission characteristics T
1
(λ) and T
2
(λ)
is given by
X(λ) = QTH(λ) · T
d
1
20
1
(λ) · T
d
2
20
2
(λ) (8.3)
with d
1
and d
2
being the thickness in mm. The
denominator 20 is introduced since we used the cat-
alogue data which were related to 20mm thick fil-
ters for the actual calculation. The best fit with the
AM1.5d spectrum using integer thickness numbers
(in mm) was found for filter 1 (FG6) with 6mm and
filter 2 (KG4) with 3mm.
Fig.8.6 shows that this particular combination of
filters approaches the desired AM1.5d filter quite
well. The required filter transmission spectrum
F
requ.
(λ) can be derived using
F
requ.
(λ) =
AM1.5d(λ)
QTH(λ)
(8.4)
All lenses in the setup (Fig.8.11) are made of
optical crown glass providing a constant optical
throughput over the 380 to 1000nm with less than
10% loss. Hence, they should not affect the spectral
shape of the final light spot.
Using an ORIEL CCD spectrometer we measured
the QTH lamp emission after the filters and lenses
8.3. SIMULATION OF SOLAR RADIATION 119
Figure 8.6: Transmission spectra of the required,
modeled and measured AM1.5d filter. There is a
clear deviation of the measured transmission from
the expected (calculated) curve particularly in the
400nm to 500nm range by about 50%.
(Fig.8.5). Apart from the measured spectrum
3
the
figure also shows the AM1.5d spectrum. The spec-
tra match well, in fact it appears comparable to the
spectrum of commercial solar simulators [183].
However, there is some intensity lost in the UV
and IR which was not predicted by the calculation
(also shown in Fig.8.7). In order to understand the
reason for this discrepancy we measured the trans-
mission spectra of the filters as we received them
and found a distinct deviation from the data ac-
cording to the catalogue that have been used for
the numerical simulation - see Fig.8.6 and Fig.8.8.
A new optimisation procedure using the new
(more accurate) transmission data of the filters
suggested 8mm FG6 and 4mm KG4 - see Fig.8.9
and Fig.8.10. It can be seen, although the UV
range could be simulated better, the overall spectral
match with the AM1.5d spectrum has not improved
significantly. In addition the maximum available in-
tensity would be further reduced which can only be
compensated by using a stronger QTH-lamp and/or
a smaller beam size.
Hence, the spectrum in Fig.8.7 represents our
best result to simulate the AM1.5d spectrum. We
believe that the spectrum provided by such a sim-
ulator is sufficient to allow reasonable comparisons
of solar cell performance between different research
3
Such an intense spectrum can be measured using e.g.
metal coated neutral density filter with a known transmission
spectrum for correction
Figure 8.7: The calculated (see Eq. (8.3)) spectrum
is shown here together with the standard spectrum
(AM1.5d). The somewhat different measured spec-
trum (QTH+filter) has been shifted towards higher
irradiance compared to Fig.8.5 to obtain a reason-
able match almost as good as with a commercial
simulator. The four squares represent irradiance
levels measured with a powermeter to confirm the
absolute values.
groups - particularly in the field of organic solar
cells.
8.3.2 The Simulator Setup
The chosen optical arrangement of the simula-
tor can be seen in Fig.8.11. The light generated
from the QTH lamp is collimated after the first
lens, passes through the above described filter sys-
tem with subsequent concentration after the second
lens.
The irradiance level of the final spot depends,
apart from the optical throughput (lenses, filter),
on the chosen focal distance of the two lenses as
well as the actual sample position (since the final
beam is divergent) and the distance x of the QTH
lamp from the first lens.
Sample position and distance x has been ad-
justed to give the required irradiance (0.11mW
cm
−2
nm
−1
)using a calibrated photodiode masked
with an interference filter at 550nm- taking into ac-
count its spectral width (6nm) and actual transmis-
sion factor (0.55).
To ensure that the sample cell always sees the
correct (solar) intensity despite the divergent beam,
a stripe of white paper was (temporarily) inserted
120 CHAPTER 8. DETERMINATION OF SOLAR CELL EFFICIENCIES
Figure 8.11: The solar simulator setup comprises the QTH-lamp (QTH), biconvex lenses (52mm diameter)
made of optical crown glass L1 (f=10cm), L2 (f=4cm) the two filters F1 (FG6, 6mm thick) and F2 (KG4,
3mm thick) plus the iris B to define the spot size. The projection of a paper stripe (1cm wide) served as
sample positioning tool
5
allowing to reproduce the sample position within ±1mm and irradiance within
about 5%. Variation of the distance x allowed for fine tuning of the irradiance level.
Figure 8.8: We found that the actual supplied fil-
ter transmission curves (solid lines)deviate up to
20% from the catalogue data (dotted lines). For
3x2mm FG6 filter this means a deviation of up to
1 − 0.8
3
= 0.5 (50%) which is in good agreement
with the deviation found in Fig.8.6.
into the light path at a position
6
where its sharp
projected image has the desired irradiance thereby
defining the sample position. In this way, any sam-
ple solar cell or calibration device such as e.g. a
thermopile could be positioned within ±1mm.
The irradiance can still be controlled i.e. fine ad-
justed via the distance x. Variation of the voltage
and/or current of the QTH lamp allows mainly for
6
which was just behind the first lens here
Figure 8.9: Using the measured filter data from
Fig.8.8 we found that 4mm of KG4 and 8mm of
FG6 would approximated the required filter better.
However, transmission will then be reduced further
by a factor of 2 - compare with Fig.8.6.
changes in colour temperature of the filament and
thereby tune the spectral composition. With the
described setup we could obtain the intensity of 1
sun with a spot diameter of 8mm which could al-
most illuminate all pixels of our standard devices
(12x12mm) at once.
8.4 “Numerical” Simulation
An alternative way to estimate the solar power effi-
ciency for a solar cell is to take the short circuit cur-
rent (I
sc
) and open circuit voltage (V
oc
) data from
the spectral response measurements at low illumina-
8.4. “NUMERICAL” SIMULATION 121
Figure 8.10: The simulated spectrum using the new
filter combination approaches the standard AM1.5d
spectrum much better in the short wavelength range
than the previous - see Fig.8.6.
tion intensities, and scale them up to the standard
AM1.5d condition.
The I
sc
and V
oc
extrapolated to the solar light
intensities can then be used to calculate the power
efficiency that is expected under AM1.5d condition
by integrating over the entire solar spectrum.
This procedure requires the intensity dependence
of the current and voltage and should ideally be
performed for a large number of wavelengths. In
addition IV curves should be taken for all these
wavelengths to allow for changes of the fillfactor.
However, even when performed carefully and
with much consideration this method can certainly
not replace solar simulator measurements since it
relies on a number of assumptions which may not
apply for the concerned cell types.
The main problem is that we do not really know
if we can simply integrate (add) the number of
photons at higher intensities without considering
Coulomb-Coulomb interaction of the charge carri-
ers. We have tested this calculation on a standard
silicon diode (HAMAMATSU) and obtained a rea-
sonable efficiency of about 5% suggesting it s appli-
cable at least to silicon - solar cells.
However, inorganic solar cells generally have 1000
times higher charge carrier mobilities and function
in a different way.
Here we demonstrate the procedure with data
taken from the laminated solar cell described in
Chapter 6. The intensity dependence of V
oc
and I
sc
Figure 8.12: V
oc
and I
sc
versus intensity (λ =
488nm). The current goes virtually linear while the
voltage saturates around the workfunction differ-
ence of the electrodes.
over a few orders of magnitude is plotted in Fig.8.12.
The photocurrent varies virtually linearly with the
irradiance E, hence we can write:
I
sc
(λ) ∼ E(λ) (8.5)
In other words, the current scales by the same
factor as the irradiance such that
I
sc
AM1.5d
(λ) =
E
AM1.5d
(λ)
E
m
(λ)
· I
scm
(λ) (8.6)
with AM1.5d indicating quantities under AM1.5d
condition and m indicating the actually measured
quantity obtained during the spectral response mea-
surements.
The wavelength dependent V
oc
values for solar
irradiance levels were taken from Fig.8.12. Fill-
factors were derived from IV curves measured with
irradiance E
m
and found to be virtually indepen-
dent of the excitation wavelength (FF=0.33). Now
we can compute the power efficiency under AM1.5d
condition expected for this cell using:
η
AM1.5d
(λ) =
J
sc
(λ) · V
oc
AM1.5d
(λ) · FF
AM1.5d
(λ)
E
AM1.5d
(λ)
(8.7)
with
J
scAM1.5d
(λ) =
I
sc
AM1.5d
(λ)
A
(8.8)
where A is the illuminated area of the device
(5mm
2
).
122 CHAPTER 8. DETERMINATION OF SOLAR CELL EFFICIENCIES
Since the spectral width of the light used for the
SR measurements was set to 10nm - the same value
as the chosen spectral resolution we could obtain
the integral by simple adding all obtained values
together. We took the mean values of the open
circuit voltage and FF and applied:
η
AM1.5d

(
λ2
¸
λ1
J
sc
AM1.5d
(λ)) · V
oc
· FF
P
AM1.5d
(8.9)
with P
AM1.5d
≡ 76.6mW being the sum (integral)
of the light power over the entire AM1.5d spectrum
and A as the area of the sample cell (5mm
2
).
The result for the power efficiency with this
method was η
AM1.5d
= 1.84%.
Alternatively we could calculate the power per
device area J
sc
AM1.5d5
· V
oc
for every wavelength in-
terval before we built the integral (sum) as in:
η
AM1.5d

(
λ2
¸
λ1
J
sc
AM1.5d
(λ) · V
oc
(λ)) · FF
P
AM1.5d
(8.10)
Using this equation we obtained a slightly higher
value η
AM1.5d
= 1.87%.
Since both methods can be justified equally, we
decided to take the mean of both numbers so that
our final result (after rounding) is η
AM1.5d
= 1.9%.
The latter is the number cited in Ref. [94].
We note that a similar method has been applied
to calculate η
AM1.5d
of a double layer cell using a
perylene diimde and a ZnPc [214]. The calculated
value (1.05%) in this reference was in satisfying
agreement with the result using a solar simulator.
In order to justify the integration over the wave-
length range in Eq. (8.10) the short circuit current
I
sc
should continue its linearity up to the full solar
intensity of 76mW/cm
2
- which is the case here.
We also want to note that building the sum of
I
sc
7
over the relevant wavelength range (here about
400nm) means that the resulting intensity i.e. num-
ber of photons/area can be about 2 orders of magni-
tude higher than the intensity extracted from 10nm
of the solar spectrum.
8.5 Summary
In this Chapter we have shown:
7
with a spectral width of 10nm
• A good match with the standard spectrum for
direct solar radiation on Earth (AM1.5d) can
be achieved with a solar simulator using a QTH
lamp. Although clearly more tedious and in-
convenient to operate in comparison with com-
mercial simulation equipment, such a simulator
may still be interesting as a low cost evalua-
tion device for researchers investigating solar
cells with potential stability problems and/or
(in comparison with inorganic cells) low con-
version efficiencies such as organic solar cells.
In any case, if the cell reaches commercially
interesting efficiencies ( 5%) we recommend to
do outdoor performance tests and/or entrust
internationally accepted solar energy research
and test laboratories such as the Fraunhofer
Institute for Solar Energy (FISE) in Freiburg,
Germany or the National Renewable Energy
Lab (NREL) in Golden, USA with their more
sophisticated simulation facilities.
• An AM1.5d efficiency value for a silicon solar
cell has been calculated by extrapolating the
V
oc
and I
sc
values from the spectral response
measurements to solar condition. By applying
the same method to a laminated polymeric so-
lar cell we obtained a value for the power con-
version efficiency under AM1.5d conditions of
1.9 %.
Although this method appears to give reason-
able results it would be desirable to investigate
the theoretical assumptions (integration over
the spectrum, wavelength independent fillfac-
tor, extrapolation of the intensity dependency
etc.) further. However, it may be considered
as another possibility to characterise and com-
pare the performance of organic photovoltaic
devices among researchers in this field.
Chapter 9
Summary and Bibliography
9.1 Summary
The aim of this work has been the investigation of
different classes of organic semiconducting materi-
als using various types of solar cell architectures
with respect to their photovoltaic properties.
In Table 9.3 we have summarised the photovoltaic
key parameters of the devices investigated in this
thesis or reported by other research groups. It can
be seen that, on average, devices using pigments
(still) perform better than polymer, dye and liquid
crystal containing devices.
Among devices that include at least one compo-
nent which is a pigment, the type of cell architec-
ture does only little affect the overall performance -
see Table 9.1. Note that only monochromatic EQE
values are considered here - cells comprising two or
more molecular components have a better chance
to absorb over wide spectrum to give higher solar
efficiencies (AM1.5) than single layer structures.
Table 9.1: Overall assessment (∗=low, ∗∗=medium,
∗ ∗ ∗=high) of the different solar cell architectures.
Devices must include at least one pigment compo-
nent.
including pigments
device architecture EQE V
oc
FF
– % V %
single layer ∗ ∗ ∗ ∗ ∗ ∗ ∗∗
double layer ∗ ∗ ∗ ∗∗ ∗ ∗ ∗
blend ∗ ∗ ∗ ∗∗ ∗∗
laminated – – –
It is important to keep in mind that pigments - in
contrast to polymers and dyes - can not simply be
processed from solution and that the best pigment
devices suffer from severe problems related to the
electrode materials [164, 224].
If only (the production friendly) soluble semicon-
ductors are considered (see Table 9.2), the EQE but
also the overall performance can vary considerably
for different device architectures. The following or-
der (best device architecture on top) for solar cell
overall performance has been found:
1. laminated device
2. blend device
3. double layer device
4. single layer device
Table 9.2: Overall assessment of the different solar
cell architectures. Soluble semiconductors only.
only soluble materials
device architecture EQE V
oc
FF
– % V %
single layer ∗ ∗ ∗ ∗ ∗
double layer ∗∗ ∗ ∗ ∗ ∗
blend ∗ ∗ ∗ ∗∗ ∗∗
laminated ∗ ∗ ∗ ∗ ∗ ∗ ∗∗
If we focus only on the EQE of soluble materials,
we find that double layer structures give about 10-
100 times higher numbers than single layers whereas
blends and the laminated structure allow a further
enhancement by another factor of 5 to 10.
If all types of organic semiconductors are consid-
ered, we can see that each architecture seems to
reach highest numbers for a different evaluation pa-
rameter:
• Highest EQE
1
in single layer devices
1
at peak wavelength
123
124 CHAPTER 9. SUMMARY AND BIBLIOGRAPHY
• Highest V
oc
in laminated devices
• Highest FF in double layer devices
• Highest SR-range in blend devices
We note that the laminated device structure
which is particularly efficient among soluble semi-
conductors has been introduced and investigated as
part of this work. This technologically interesting
structure opens many new possibilities in terms of
combination of semiconductors and semiconductor
treatment such as selective doping or heating.
Moreover, we have investigated, for the first time,
dye/dye and liquid crystal/dye interfaces with re-
spect to their photovoltaic properties.
9.1.1 Suggestions for Future Investi-
gations
In future, the investigation and exploitation of the
effects of doping using organic semiconductors or
gases like oxygen as dopants will most likely allow
further enhancements of efficiencies.
Another interesting subject to explore will be the
use of photonic bandgap structures [203] to extend
the range of available low bandgap materials.
Such photonic structures may even be realised in
an highly ordered structure consisting of co-block
polymers.
9.1. SUMMARY 125
Table 9.3: Summary of important performance parameters of various organic solar cells including all
types of architectures. The abbreviation for materials are listed in Chapter 1. In D/A cells the donor
material is followed by the acceptor material. For clarity the electrode materials who were always Al and
ITO in D/A devices have been omitted. Best parameters are printed using bold characters while italic
letters mark devices investigated in this thesis. EQE values were taken at peak wavelength.
EQE Voc FF range device comments year
% V % nm – –
Single Layer Devices
73 0.45 33 400-800(?) In/HPc+polymer/NESA pigment in polymer [164] 1981
62 1.1 33 400-800(?) Al/HPc+polymer/SnO2 pigment in polymer [43] 1979
36
1
0.96
1
0.47
1
400-850 Al/pentacene/ITO pigment [224] 2000
33 1.2 25 400-800(?) Al/merocyanine/Ag pigment [43] 1978
14 0.7 39 400-800(?) Al/CuPc/Au pigment [43] 1981
3 0.6 34 400-800(?) Al/merocyanine/Au pigment [43] 1978
1 1.2 20 400-500 PPV polymer [168] 1994
0.7 0.32 25 400-700(?) Cr/chlorophyll-a/Hg dye [43] 1975
0.5 1.0 23 400-500 Al/PPV+PEDOT/ITO polymer [8] 1998
0.15 0.85 25 400-800(?) Al/MgPc/Au pigment [43] 1974
0.1 0.7 25 400-750 Al/THPF/ITO polymer 2000
0.1 0.1 25 400-560 Al/LPPPT/ITO polymer 2000
0.09 0.5 27 400-570 Al/CN-MEH-PPV/ITO polymer [114] 1997
0.08 0.19 25 400-600 Al/Per/ITO dye 2000
0.07 0.7 25 400-570 Al/MEH-PPV/ITO polymer [196] 2000
0.05 0.37 25 400-750 Al/PTV/ITO polymer 2000
0.03 < 0.02 25 400-650 Al/P3HT/ITO polymer [63] 2001(?)
0.015 0.1 25 400-650
+
In/HPc/Au pigment [68] 1978
0.007 0.15 25 400-820 Al/HPc/ITO liquid crystal 2000
0.005 < 0.02 25 400-1000 Al/PIF/ITO polymer 2000
Double Layer Devices
30
+
0.5 65 400-800 CuPc/Per1 pigment/pigment [244] 1985
23 0.5 ≈30 400-700+ HPc1/Per2 pigment/pigment [131] 1990
23 0.3

40

400-600 PEDOT/PEOPT/C60 polymer/pigment [213] 1998
9 0.8 48 400-560 PPV/C60 polymer/pigment [111] 1996
6 1.0 60 400-620 PPV/Per3 polymer/pigment [113] 1996
≈ 3
+
1.0 23 400-620 PPyV/P3HT polymer/polymer [242] 1997
2 1.0 51 400-590 PEDOT/PPV/Per polymer/dye [9] 1999
0.6 0.12 25 400-800 CuPc2/Per dye/dye [200] 2000
0.5 0.35 20 400-850 HPc/Per3 liquid crystal(dye)/pigment [199] 1999
Blend Devices
29 0.5 ≈25 400-550 MEH-PPV+C60 polymer+dye [278] 1995
23 0.4 ≈33 400-550 MEH-PPV/C60 polymer+pigment [83] 1998
12 0.5 26 400-640 MEH-PPV+CdSe polymer+nanocrystals [102] 1996
11 0.35 41 400-620 P3HT+Per polymer+dye [65] 2000
6 0.6 25 400-600 MEH-PPV+CN-PPV polymer+polymer [112] 1995
1.8 0.9 23 400-500 PPV+MWNT polymer+nanotubes [3] 1999
1.0 0.1 25 400-800 HPc+Per liquid crystal+dye [200] 2000
0.7 0.4 44 400-600 MEH-PPV+Per polymer+dye [62] 1999
0.4 0.2 25 400-850 PTV+Per polymer+dye 2000
0.4 0.12 25 400-900 PTV+Ter polymer+dye 2000
0.2 0.6 25 400-800 PTV+THPF polymer+polymer 2000
0.02 <0.01 25 400-1000 PTV+PIF polymer+polymer 2000
0.0007 <0.01 25 400-1100 HPc+PIF polymer+liquid crystal 2000
Laminated Devices
30 1.3 33 400-700 POPT/MCP polymer/polymer [94] 1998
0.2 0.37 23 400-900 HPc/Per liquid crystal/dye [200] 2000
[1] IQE instead of EQE, illumination conditions are: AM1.5 (100mW/cm
2
) resulting in η = 2.4%
126 CHAPTER 9. SUMMARY AND BIBLIOGRAPHY
Bibliography
[1] D. Adam, F. Closs, T. Frey, D. Funhoff, D. Haarer,
H. Ringsdorf, P. Schuhmacher and K. Siemensmeyer
”Transient photoconductivity in a discotic liquid crys-
tal.” Phys.Rev.Lett. 70, 457-460 (1993)
[2] D. Adam, P. Schuhmacher, J. Simmerer, L. Haeus-
sling, K. Siemensmeyer, K. H. Etzbach, H. Rings-
dorf and D. Haarer ”Fast photoconduction in the
highly ordered columnar phase of a discotic liquid crys-
tal (hexhexylthio-triphenylene).” Nature 371, 141-143
(1994)
[3] H. Ago, K. Petritsch, M.S.P. Shaffer, A.H.Windle and
R.H. Friend ”Composites of carbon nanotubes and con-
jugated polymers for photovoltaic devices” Adv.Mat.
11, 1281-1285 (1999)
[4] M. R. Andersson, D. Selse, M. Berggren, H. Jaervi-
nen, T. Hjertberg, O. Inganaes, O. Wennerstroem and
J. E. Oesterholm ”Regioselective polymerization of 3-
(4-octylphenyl)thiophene with FeCl3.” Macromolecules
27, 6503-6506 (1994)
[5] M. R. Andersson, M. Berggren, O. Inganaes, G.
Gustafsson, J. C. Gustafsson-Carlberg, D. Selse, T.
Hjertberg and O. Wennerstroem ”Electroluminescence
from substituted polytiophenes:From blue to near in-
frared.” Macromolecules 28, 7525-7529 (1995)
[6] A.C. Arias, M. Granstrom, D.S. Thomas, K.
Petritsch and R.H. Friend. ”Doped conducting-
polymer-semiconducting-polymer interfaces: their use
in organic photovoltaic devices” Phys.Rev.B - Cond.
Matt. 60, 1854-1860 (1999)
[7] A.C. Arias, M. Granstrom, K. Petritsch and R.H.
Friend. ”Organic photodiodes using polymeric anodes”
Synth. Met. 102, 953-954 (1999)
[8] A.C. Arias, K. Petritsch, M. Granstrom, J.J. Dittmer,
E.A. Marseglia and R. H. Friend ”Efficient Polymeric
Photovoltaic Devices. ” Proc. CBECIMAT (1998)
[9] A.C. Arias, K. Petritsch, M. Granstrom, J. J. Dittmer,
E. A. Marseglia and R. H. Friend ”Efficient Polymeric
Photovoltaic Devices.” Proc. CBECIMAT (1998)
[10] R. Ash Incredible Comparisons, DK, London (1997)
[11] U. Bach et. al. ”Solid state dye sensitized mesoporous
TiO
2
solar cells with high photon to electron conversion
efficiencies” Nature 395, 583-585 (1998).
[12] S. Barth and H. Baessler ”Intrinsinc photoconduction
in PPV-Type conjugated polymers.” Phys.Rev.Lett.
79, 4445-4448 (1997)
[13] Z. Bao, A. J. Lovinger and A. Dodabalapur ”Highly or-
dered vacuum deposited thin films of metallophthalo-
cyanines and their aplications in field effect transis-
tors.” Adv.Mat. 9, 42 (1997)
[14] H. Becker, A. Lux, A. B. Holmes and R. H. Friend ”PL
and EL quenching due to thin metal films in conjugated
polymers and polymer LEDs. ” Synth.Met. 85, 1289-
1290 (1997)
[15] H. Becker, P. L. Burns and R. H. Friend ”Effect of
metal films on the photoluminescence and electrolu-
minescence of conjugated polymers. ” Phys.Rev.B 56,
1893-1905 (1997)
[16] S. Becker et al. to be published.
[17] P. Benett, ”Earth: The Incredible Recycling Machine”,
Wayland (Publishers) Ltd, East Sussex (1993)
[18] M. Bergren et al. ”Thermal control of near infrared and
visible electroluminescence from substituted polythio-
phenes” Appl.Phys.Lett. 65, 1489-1491 (1994)
[19] J. Bharathan, Y. Yang, Appl. Phys. Lett. 72, 2660
(1998)
[20] J. Billard, J.C. Dubois, H.T. Nguyen, A.Zann
Nouv.J.Chim. 2, 535 (1978)
[21] G. Blasse, G. J. Dirksen, A. Meijerink, J. F.
Van der Pol, E. Neeleman and W. Drenth ”Lu-
minescence and energy migration in the solid state
and in the ordered columnar mesophase of periph-
erically octa-n-dodecoxy-substituted phthalocyanine.”
Chem.Phys.Lett. 154, 420-424 (1989)
[22] N. Boden, R. J. Bushby and J. Clements ”Mecha-
nism of quasi one dimensional electronic conductivity in
discotic liquid crystals.” J.Chem.Phys. 98, 5920-5931
(1993)
[23] N. Boden, R. C. Borner, R. J. Bushby and J. Clements
”First observation of an n-doped quasi one dimensional
electronically-conducting discotic liquid crystal hex-
akis(hexylthio)tricycloquinazinole (HTTQ).” J. Am.
Chem. Soc. 116, 10807-8 (1994)
[24] N. Boden, R. J. Bushby and J. Clements ”Electron
transport along molecular stacks in discotic liquid
crystals.” J.Mater.Science: Mat.Electronics 5, 83-88
(1994)
[25] N. Boden, R. J. Bushby and A. N. Cammidge ”Triph-
enylene based discotic liquid crystalline polymers: A
universial, rational synthesis. ” J. Am. Chem. Soc.
117, 924-927 (1995)
127
128 CHAPTER 9. SUMMARY AND BIBLIOGRAPHY
[26] N. Boden, R. J. Bushby, J. Clements and R. Luo ”Char-
acterization of the cationic species formed in p-doped
discotic liquid crystals.” J.Mater.Chem. 5, 1741-1748
(1995)
[27] N. Boden, R. Bissell, J. Clements and B. Movaghar
”Discotic liquid crystals. References for charge and ex-
citon transport and mobilities.” Liquid Crystals - To-
day 6, (1996)
[28] N. Boden, R. J. Bushby, A. N. Cammidge, S. Duck-
worth and G. Headdock ”Halogenation of tripheny-
lene based discotic liquid crystals: towards a chiral nu-
cleus.” J.Mater.Chem. 7, 601-606 (1997)
[29] P.M. Borsenberger & D.S. Weiss ”Organic Photorecep-
tors for Imaging Systems” Dekker, New York (1993)
[30] Y. Bouligand ”Geometry of non smectic hexagonal
mesophases.” J. Physique 41, 1297-1306 (1980)
[31] Y. Bouligand ”Defects and textures of hexagonal dis-
cotics.” J. Physique 41, 1307-1315 (1980)
[32] D. Braun and A. J. Heeger ”Visible light emission from
semiconducting polymer diodes.” Appl.Phys.Lett. 58,
1982 (1991)
[33] T. M. Brown, J. S. Kim, R. H. Friend, F. Cacialli, R.
Daik and W. J. Feast. ”Built-in field electroabsorption
spectroscopy of polymer light- emitting diodes incorpo-
rating a doped poly(3,4-ethylene dioxythiophene) hole
injection layer” Appl.Phys.Lett. 75, 1679-1681 (1999)
[34] J. H. Burroughes, D. D. C. Bradley, A. R. Brown, R.
N. Marks, K. Mackay, R. H. Friend, P. L. Burns and A.
B. Holmes ”Light emitting diodes based on conjugated
polymers.” Nature 347, 539 (1990)
[35] F. Cacialli, unpublished results.
[36] C. J. Campbell and J. H. Laherrere ”The end of cheap
oil.” Scientific American March, 60 (1998)
[37] A. N. Cammidge, I. Chambrier, M. J. Cook, A. D.
Garland, M. J. Heeney and K. Welford ”Octaalkyl- and
octaalkoxy-2,3 naphthalocyanines.” J.Porphyrins and
Phthalocyanines 1, 77-86 (1997)
[38] Y. Cao, P. Smith and A. J. Heeger ”Counter-ion in-
duced processibility of conducting polyaniline and of
conducting poly-blends of poly-aniline in bulk poly-
mers.” Synth.Met. 48, 91-97 (1992)
[39] S. A. Carter, M. Angelopoulos, S. Karg, P. J. Brock
and J. C. Scott ”Polymeric anodes for improved poly-
mer LED performance. ” Appl.Phys.Lett. 70, 2067-
2069 (1997)
[40] R. R. Chance, A. Prock and R. Silbey Adv. Chem.
Phys. 37, 1 (1978)
[41] G.A. ChamberlainP.J. Cooney and S. Dennison ”Pho-
tovoltaic properties of merocyanine solid state photo-
cells” Nature 2
¯
89 45-47 (1981)
[42] G. A. Chamberlain ”Organic PN junction solar cells.”
Mol.Cryst.Liq.Cryst. 93, 369-379 (1983)
[43] G. A. Chamberlain ”Organic solar cells: A review.”
Solar Cells 8, 47-83 (1983)
[44] S. Chandrasekhar, B. K. Sadashiva and K. A. Suresh
”Liquid crystals of disc-like molecules.” Pramana 9,
471-480 (1977)
[45] S. Chandrasekhar ”Liquid crystals of disc -like
molecules.” Phil. Trans R. Soc. Lond. A 309, 93-103
(1983)
[46] D. M. Chapin, C. S. Fuller and G. L. Pearson ”A new
silicon p-n junction photocell for converting solar radi-
ation into electrical power. ” J.Appl.Phys. 25, 676-677
(1954)
[47] N. Chawdhury, M. Younus, P. R. Raithby, J. Lewis
and R. H. Friend ”The effect of air exposure on the
photocurrent in a conjugated platinum poly-yne. ”
Opt.Mat. 9, 498-501 (1998)
[48] G. Cheek, A. Genis, J. B. DuBow and V. R. P. Ver-
necker ”Antireflection properties of ITO on silicon for
photovoltaic applications.” Appl.Phys.Lett. 35, 495-
497 (1979)
[49] M. Chandross, S. Mazumdar, S. Jeglinski, X.
Wei. Z.V. Vardeny, E.W. Kwock and T.M. Miller
J.Phys.:Condens.Matter 6, 1379-1394 (1994)
[50] N. Chawdhury, M. Younus, P. R. Raithby, J. Lewis
and R. H. Friend ”The effect of air exposure on
the photocurrent in a conjugated platinum poly-yne.”
Opt.Mat. 9, 498-501 (1998)
[51] G. Cheek, A. Genis, J. B. DuBow and V. R. P. Ver-
necker ”Antireflection properties of ITO on silicon for
photovoltaic applications.” Appl.Phys.Lett. 35, 495-
497 (1979)
[52] V. Choong, Y. Park, Y. Gao, T. Wehrmeister, K.
Muellen, B. R. Hsieh and C. W. Tang ”Dramatic PL
quenching of PPV oligomer thin films upon submono-
layer Ca deposition.” Appl.Phys.Lett. 69, 1492-1494
(1996)
[53] T. Christ, B. Gluesen, A. Greiner, A. Kettner, R.
Sander, V. Stuempflen, V. Tsukruk and J. H. Wen-
dorff ”Columnar discotics for LEDs.” Adv.Mat. 9, 48-
52 (1997)
[54] J. Cornil, A.J. Heeger, and J.L. Bredas, Chem. Phys.
Lett. 272, 463-470 (1997)
[55] R. Corradi and S. P. Armes ”Chemical synthesis of
poly(3,4-ethylenedioxy thiophene) ” Synth.Met. 84,
453-454 (1997)
[56] P. G. d. Costa and E. M. Conwell ”Excitons and
the band gap in PPV.” Am.Phys.Soc.-Rap.Comm.48,
1993-1997 (1993)
[57] A.van de Craats, P.G Schouten, J.M. Warman ”Charge
Transport in Discotic Liquid Crystalline Materials
Studied by Pulse-Radiolysis Time Resolved Microwave
Conductivity (PR-TRMC)” J.Jap.Liqu.Cryst.Soc. 2,
12-27 (1998)
[58] A.van de Craats, J.M. Warman, K. Muellen, Y. Geerts,
J.D. Brand ”Rapid charge transport along self assem-
bling graphite nanowires” Adv.Mat.10, 36-38 (1998)
9.1. SUMMARY 129
[59] G. M. Delacote, J. P. Fillard and F. J. Marco ”Electron
injection in thin films of copper phthalocyanine.” Solid
State Communications 2, 373-376 (1964)
[60] S. Demmig and H. Langhals ”Leichtloesliche, lichtechte
Perylen- Fluoreszenzfarbstoffe.” Chem.Ber. 121, 225-
230 (1988)
[61] J.J. Dittmer, K. Petritsch, E.A. Marseglia, R.H.
Friend, H. Rost and A.B. Holmes ”Photovoltaic prop-
erties of MEH-PPV/PPEI blend devices” Synth.Met.
102, 879 (1999)
[62] J.J. Dittmer, P. Lazaroni, P.Leclere, P. Moretti, M.
Granstrom K. Petritsch, E.A. Marseglia, R.H. Friend,
H.Rost and A.B. Holmes ”Crystal Network Formation
in Organic Solar Cells” Sol.En.Mat.Sol.Cells 61, 53-61
(2000)
[63] J.J. Dittmer to be published
[64] J.J. Dittmer, private communication
[65] J.J. Dittmer, E.A. Marseglia, R.H. Friend submitted to
Adv.Mat.
[66] Fachlexikon der Physik, Verlag Harri Deutsch, 2. Au-
flage (1989)
[67] F. R. Fan and L. R. Faulkner ”Photovoltaic effects of
metal free and Zn phthalocyanines.” J.Chem.Phys. 69,
3334 (1978)
[68] F. R. Fan and L. R. Faulkner ”Photovoltaic effects of
metal free and Zn phthalocyanines.” J.Chem.Phys. 69,
3341-3349 (1978)
[69] L. Feiler, H. Langhals and K. Polborn ”Synthesis of
perylene- 3,4 dicarboximides - Novel highly photostable
fluorescent dyes.” Liebigs Ann., 1229-1244 (1995)
[70] Fesser, Bishop and Campbell ”Optical absorption from
polarons in a model of polyacetylene” Phys.Rev.B 27,
4804-4825 (1983)
[71] L. Fleury, B. Sick, G. Zumhofen, B. Hecht, U.P. Wild
Mol.Phys. 95, 1333 (1998)
[72] T. Foerster, ”Fluoreszenz Organischer Verbindungen”,
Vandenhoeck und Ruprecht, Goettingen (1951)
[73] W. T. Ford, L. Summer, W. Zhu, Y.H. Chang, P. Um,
K.H. Choi, P.A. Heiney, N.C. Maliszewskyi New J.
Chem., 18 495 (1994)
[74] M. Forster, K.O. Annan, U. Scherf, ”Conjugated Lad-
der Polymers Containing Thienylene Units” Macromol.
32 3159 (1999)
[75] J. Frenkel, Phys. Rev. 38, 309-320 (1931)
[76] R. H. Friend, D. D. C. Bradley and P. D. Townsend
”Photoexcitation in conjugated polymers.” J. Phys. D:
Appl. Phys. 20, 1367 (1987)
[77] R. H. Friend, J. H. Burroughes and D. D. C. Bradley
”Electroluminescent Devices. ” Patent Nr: US5247190
(1993)
[78] R. H. Friend, G. J. Denton, J. J. M. Halls, N. T. Harri-
son, A. B. Holmes, A. Koehler, A. Lux, S. C. Moratti,
K. Pichler, N. Tessler and K. Towns ”Electronic pro-
cesses of conjugated polymers in semiconductor device
structures. ” Synth.Met. 84, 463-470 (1997)
[79] R. H. Friend, G. J. Denton, J. J. M. Halls, N. T.
Harrison, A. B. Holmes, A. Koehler, A. Lux, S. C.
Moratti, K. Pichler, N. Tessler, K. Towns and H. F.
Wittmann ”Electronic excitations in luminescent con-
jugated polymers. ” Solid State Communications 102,
249-258 (1997)
[80] Fu Y, H. Cheng and R. L. Elsenbaumer ”Electron Rich
Thienylene - Vinylene low bandgap polymers.” Chem.
Mater. 9, 1720-1724 (1997)
[81] A. Fujii, A. A. Zakhidov, V. V. Borovkov, Y. Ohmori
and K. Yoshino ”Organic photovoltaic cell with donor-
acceptor double heterojunction.” Jpn. J. Appl. Phys.
35, L1438-1441 (1996)
[82] S. Fujita, T. Sakamoto, K. Ueda, K. Ohta and S. Fujita
”Surface treatment of ITO substrates and its effects on
initial nucleation processes of diamine films.” Jpn. J.
Appl. Phys. 36, 350-353 (1997)
[83] J. Gao, F. Hide and H. Wang ”Efficient photodetec-
tors and photovoltaic cells from composites of C60 and
conjugated polymers: photoinduced electron transfer.”
Proc. ICSM96 (Snowbird) (1996)
[84] J. Gao, Yu G and A. J. Heeger ”Polymer PIN Junction
Solar cells.” Adv.Mat. 9, 692-695 (1998)
[85] F. Garnier, G. Horowitz, X. Z. Peng and D. Fichou
”Structural basis for high carrier mobility in conjugated
oligomers.” Synth.Met. 45, 163-171 (1991)
[86] P.R. Gast, ”Solar Radiation”, in Handbook of Geo-
physics, ed. C.F. Campen et al.,p14-30 (1960).
[87] A. K. Ghosh, D. L. Morel, T. Feng, R. F. Shaw and
C. A. Rowe ”Photovoltaic and rectification properties
of Al/Mg-phthalocyanine/Ag Schottky-barrier cells.”
J.Appl.Phys. 45, 230-236 (1974)
[88] A. K. Ghosh and T. Feng ”Merocyanine organic solar
cells. ” J.Appl.Phys. 49, 5982-5989 (1978)
[89] A.K. Ghosh and T. Feng ”Photovoltaic Device con-
taining an Organic Layer” US-Patent Nr.: US4127738
(1978)
[90] F. Goldoni, R.A.J. Janssen and E.W. Meijer ”Synthesis
and characterization of new poly(thienylenevinylene)s
with thioether side chains” Polymer Preprints 39, 1049
(1998)
[91] F. Goldoni, D. Iarossi, A. Mucci, L. Schenetti, M. J.
Zambianchi, J. Mater. Chem. 7, 593 (1997)
[92] H. L. Gomez, P. Stallinga, H. Rost, A. B. Holmes, M.
G. Harrison and R. H. Friend ”Analysis of deep levels in
a phenylenevinylene polymer by transient capacitance
methods. ” Appl.Phys.Lett. 22, (1999)
[93] P. L. Gourley ”Photonic lattices block light.” Scientific
American March, 43 (1998)
130 CHAPTER 9. SUMMARY AND BIBLIOGRAPHY
[94] M. Granstrom, K. Petritsch, A. C. Arias, A. Lux, M. R.
Andersson and R. H. Friend ”Laminated fabrication of
polymeric photovoltaic diodes. ” Nature 395, 257-260
(1998)
[95] M. Granstrom and K. Petritsch: Patent Nr.:
WO9949525A1: Multilayer photovoltaic or photocon-
ductive devices; Filing date: Febr.02. (1999)
[96] W. Graupner, G. Leditzky, G. Leising, U. Scherf Phys.
Rev.B 54 7610 (1996)
[97] M.A. Green. Solar Cells - Operating Principles , Tech-
nology and System Applications. University of New
South Wales, Kensington, (1992).
[98] M.A. Green Silicon Solar Cells. Advanced Principles
and Practice. University of New South Wales, Sydney,
(1995).
[99] B. A. Gregg ”Evolution of photophysical and photovol-
taic properties of perylene bisphenetylimide films upon
solvent vapor annealing ” J.Phys.Chem. 100, 852-859
(1996)
[100] N. C. Greenham, S. C. Moratti, D. D. C. Bradley, R.
H. Friend and A. B. Holmes ”Efficient light emitting
diodes based on polymers with high electron affinities.”
Nature 365, 628 (1993)
[101] N. C. Greenham and R. H. Friend ”Semiconductor
device physics of conjugated polymers. ” Solid State
Physics 49, 1 (1995)
[102] N. C. Greenham, X. Peng and A. P. Alivisatos ”Charge
separation and transport in conjugated -polymer (MEH
PPV)/CdSe -nanocrystal composites studied by PL
quenching and photoconductivity.” Phys.Rev.B 54,
17628 (1996)
[103] J. Gruener, M. Remmers and D. Neher ”Direct de-
termination of the emission zone in a polymer LED.”
Adv.Mat. 9, 964-968 (1997)
[104] M.S. Gudipati, J. Phys. Chem. 98, 9750-9763 (1994)
[105] A. G. Guerek and O.E Bekaroglu ”Octakis(alkylthio)-
substituted (Cu, Co, Ni, Zn) phthalocyanines and their
interactions with silver and palladium ions. J. Chem.
Soc. Dalton Trans 1419, (1994)
[106] J. Hagen, W. Schaffrath, P. Otschik, R. Fink, A.
Bacher, H. W. Schmidt and D. Haarer ”Novel hybrid
solar cells consisting of inorganic nanoparticles and an
organic hole transport material.” Synth.Met. 89, 215-
220 (1997)
[107] A. Hagfeldt, B. Didriksson, T. Palmqvist, H. Lind-
stroem, S. Soedergren, H. Rensmo and S. E. Lindquist
”Verification of high efficiencies for the Graetzel-cell.”
Sol.En.Mat.Sol.Cells 31, 481-488 (1994)
[108] T. W. Hagler, K. Pakbaz, K. F. Voss and A. J.
Heeger ”Enhanced order and electronic delocalization
in conjugated polymers oriented by gel processing.”
Phys.Rev.B 44, 8652 (1991)
[109] D. A. Halliday, P. L. Burn, R. H. Friend, D. D. C.
Bradley, A. B. Holmes and A. Kraft ”Extended Pi con-
jugation in PPV from a chemically modified precursor
polymer.” Synth.Met. 55-57, 954-959 (1993)
[110] D. A. Halliday, B. L. Burn, D. D. C. Bradley, R. H.
Friend, O. M. Gelsen, A. B. Holmes, A. Kraft, J. H.
F. Martens and K. Pichler ”Large changes in opti-
cal response through chemical pre ordering of PPV.”
Adv.Mat. 5, 40 (1993)
[111] J. J. M. Halls, K. Pichler, R. H. Friend, S. C. Moratti
and A. B. Holmes ”Exciton diffusion and dissocia-
tion in a PPV/C60 heterojunction photovoltaic cell”
Appl.Phys.Lett. 68, 3120-3122 (1996)
[112] J. J. M. Halls, C. A. Walsh, N. C. Greenham, E.
A. Marseglia, R. H. Friend, S. C. Moratti and A. B.
Holmes ”Efficient photodiodes from interpenetrating
polymer networks.” Nature 376, 498 (1995)
[113] J. J. M. Halls and R. H. Friend ”The photovoltaic
effect in a PPV/perylene heterojunction.” Synth.Met.
85, 1307-1308 (1996)
[114] J. J. M. Halls ”Photoconductive properties of conju-
gated polymers” PhD thesis, Cambridge (1997)
[115] J.J.M. Halls and R.H. Friend ”Photoresponsive de-
vice with a photoresponsive zone comprising a polymer
blend.” US Patent 5670791 (1997)
[116] M. Hanack ”Synthesis and properties of con-
ducting bridged macrocylic metal complexes. ”
Mol.Cryst.Liq.Cryst. 105, 133-149 (1984)
[117] M. Hanack, A. Beck and H. Lehmann ”Syntheses of liq-
uid crystalline phthalocyanines.” Communications 703-
705 (1987)
[118] M. Hanack and M. Lang ”Conducting stacked metal-
lophthallocyanines and related compounds.” Adv.Mat.
6, 819-833 (1994)
[119] M. Hanack, K. Duerr, A. Lange, J. Osio Barcina, J.
Pohmer and E. Witke ”Synthesis of low band gap poly-
mers - a challenge for organic chemists.” Synth.Met. 71,
2275-2278 (1995)
[120] M. G. Harrison, J. Gruener and G. W. C. Spencer
”Analysis of the photocurrent action spectra of MEH-
PPV photodiodes.” Phys.Rev.B 55, 7831-7849 (1997)
[121] M. G. Harrison, J. Gruener and G. W. C. Spencer
”Investigations of organic electroluminescent diodes
by impedance spectroscopy, photo-impedance spec-
troscopy and modulated photovoltage spectroscopy ”
Synth.Met. 76, 71-75 (1996)
[122] K. S. Harsha Sree, K. J. Bachmann, P. H. Schmidt,
E. G. Spencer and F. A. Thiel ”n-indium tin oxide/p-
indium phosphide solar cells.” Appl.Phys.Lett. 30, 645-
646 (1977)
[123] C. B. Hatfield ”Oil back on the global agenda. Perma-
nent decline in global oil is virtually certain to begin
within 20 years. ” Nature 387, 121 (1997)
[124] A.J.Heeger et al. Appl. Phys. Lett. 58 1982-1984
(1991)
[125] J. Heinze ”Cyclovoltametry - die ”Spektroskopie” des
Elektrochemikers ” Angewandte Chemie 96, 823 (1984)
9.1. SUMMARY 131
[126] J. Heinze, J. Mortensen, K. Muellen and R. Schenk
”The charge storage mechanism of conducting poly-
mers: A voltametric study on defined soluble oligomers
on the phenylene-vinylene type. ” J. Chem. Soc.,Chem.
Commun 701-703 (1987)
[127] J. Heinze, M. Stoerzbach, J. Mortensen and poly-
mers(conducting) ”Electrical and theoretical studies on
the redox properties of conducting polymers. ” Ber.
Bunsenges. Phys. Chem. 91, 960-967 (1987)
[128] D. Hertel, U. Scherf, H. Baessler Adv. Mat. 10 1119
(1998)
[129] F. Hildebrandt, J. A. Schroeter, C. Tschierske, R. Fes-
tag, M. Wittenberg and H. Wendorff ”Formation of
columnar mesophase by rod-like molecules: Facial am-
phiphilic p-terphenyl derivatives.” Adv.Mat. 9, 564-567
(1997)
[130] M. Hiramoto, M. Suezaki and M. Yokoyama ”Effect of
thin gold layer on the photovoltaic properties of tandem
organic solar cell.” Chem. Lett 327-330 (1990)
[131] M. Hiramoto, Y. Kishigami and M. Yokoyama ”Doping
effect on the two layer organic solar cell.” Chem. Lett
119-122 (1990)
[132] M. Hiramoto, T. Katsume and M. Yokoyama ”Light
amplification in a new light transducer combining an
organic LED with photoresponsive organic pigment
film. ” Opt. Rev. 1, 82-84 (1994)
[133] M. Hiramoto, T. Imahigashi and M. Yokoyama ”Pho-
tocurrent multiplication in organic pigment films”
Appl.Phys.Lett. 64, 187-189 (1994)
[134] G. Horowitz, F. Kouki, P. Spearman, D. Fichou,
C. Nogues, X. Pan and F. Garnier ”Evidence for n-
type conduction on a perylene tetracarboxylic diimide
derivative.” Adv.Mat. 8, 242-244 (1997)
[135] K. Hunger, W. Herbst ”Industrielle Organische Pig-
mente: Herstellung, Eigenschaften, Anwendungen.” 1st
edition, VCH, Weinheim (1987).
[136] Intergovernmental Panel on Climate Change (IPCC)
”Second Assessment Report - Climate Change 1995”,
(1995) Web site: www.meto.gov.uk
[137] K.Y. Jen, H. Eckhardt, T.R. Jow, L.W. Shackette,
R. Elsenbaumer J. Chem. Soc. Chem. Commun 215
(1988)
[138] K. C. Kao and W. Hwang ”Electrical Transports in
Solids - with Particular Reference to Organic Semicon-
ductors” Pergamon Press, Oxford (1981)
[139] N. Karl and Ziegler, J.Chem.Phys.Lett. 32, 438-442
(1975)
[140] M. Katayose, S. Tai, K. Kamijima, H. Hagiwara and N.
Hayashi ”Novel silicon naphthalocyanines: Synthesis
and molecular arrangement in thin films.” J.Chem.Soc.
Perkin Trans. 2, 403-409 (1992)
[141] M. Khan et. al. Synth.Met. 101, 246-247 (1999)
[142] M. Khan personal communication (1999)
[143] M. Khan, unpublished results.
[144] J. S. Kim, M. Granstrm, R. H. Friend, N. Johansson,
W. R. Salaneck, R. Daik, W. J. Feast and F. Cacialli.
”Indium-tin oxide treatments for single- and double-
layer polymeric light-emitting diodes: the relation be-
tween the anode physical, chemical, and morphological
properties and the device performance”, J.Appl.Phys.
84, 6859-6870 (1998)
[145] K. H. Koch and K. Muellen ”Synthesis of tetra alkyl-
substituted oligo (1,4-naphthylene)s and cyclization to
soluble oligo(peri-naphtylene)s. From the perylene to
pentarylene. ” Chem.Ber. 124, 2091-2100 (1991)
[146] A. Koehler, H. F. Wittmann, R. H. Friend, M. S. Khan
and J. Lewis ”Enhanced photocurrent response in pho-
tocells made with platinum-poly-yne/C60 blends by
photoinduced electron transfer. ” Synth.Met. 77, 147
(1996)
[147] A. Koehler, H. F. Wittmann, R. H. Friend, M. S. Khan
and J. Lewis ”The photovoltaic effect in a platinum
poly-yne. Effect of triplet excitons on the photogener-
ation of charge carriers ” Synth.Met. 67, 245 (1994)
[148] A. Koehler, J. Gruener, R. H. Friend, K. Muellen and
U. Scherf ”Photocurrent measurements on aggregates
in LPPP. additional response in the yellow part (at 2.25
eV) attributed to aggregates. ” Chem.Phys.Lett. 243,
456 (1995)
[149] A. Koehler, D. A. Dos Santos, D. Beljonne, Z. Shuai,
J. L. Bredas, A. B. Holmes, A. Kraus, K. Muellen and
R. H. Friend ”Charge separation in localized and delo-
calized electronic states in polymeric semiconductors.”
Nature 392, 903-906 (1998)
[150] A. Koehler, private communication.
[151] K. Kogo, T. Goda, M. Funahashi and J. Hanna ”Po-
larized light emission from a calamitic liquid crystalline
semiconductor doped with dyes. ” Appl.Phys.Lett. 73,
1595-1597 (1998)
[152] J. M. Kroon, R. B. M. Koehorst, M. Van Diik, G. M.
Sanders and E. J. R. Sudhoelter ”Self assembling prop-
erties of non-ionic tetraphenyl - porphyrins and discotic
phthalocyanines carrying oligo(ethylene oxide) alkyl or
alkoxy units.” J.Mater.Chem. 7, 615-624 (1997)
[153] F. Kulzer, F. Koberling, T. Christ, A. Mews and T.
Bache Chem. Phys. 247, 23 (1999).
[154] K. Y. Law ”Organic photoconductive materials: Re-
cent trends and developments. ” Chem.Rev. 93, 449-
486 (1993)
[155] K. Y. Law ”Organic photo conductive materi-
als for xerographic photo receptors.” Handbook
Org.Cond.Mol.Polym. 1, 488-545 (1997)
[156] B. M. W. Langeveld-Voss, E. Peteer, R. A. J. Janssen
and E. W. Meijer, Synth.Met. 84, 611 (1997)
[157] H. Langhals Heterocycles 40, 477 (1995)
[158] C.H. Lee, G. Yu, D. Moses and A.J. Heeger, Phys.
Rev. B 49, 2396-2407 (1994)
[159] O. Lehmann, Z.Phys.Chem. 4, 462 (1889)
132 CHAPTER 9. SUMMARY AND BIBLIOGRAPHY
[160] U. Lemmer, S. Heun, R.F. Mahrt, U. Scherf, M. Hop-
meier, U. Siegner, E.O. Goebel, K. Muellen, H. Baessler
Chem. Phys. Lett. 240 373 (1995)
[161] C.C. Leznoff, A.B.P. Lever (Eds), ”Phthalocyanines,
Properties and Applications” VCH Weinheim, Vol. 2
(1993)
[162] H. J. Lewerenz and H. Jungblut ”Photovoltaic -
Grundlagen und Anwendungen” Springer, Berlin, Hei-
delberg, New York (1995)
[163] E.J.W. List, J. Partee, J. Shinar, U. Scherf, K. Mullen,
E. Zojer, K. Petritsch, G. Leising and W. Graup-
ner ”Localized triplet excitations and the effect of
photo-oxidation in ladder-type poly(p-phenylene)s and
oligo(p-phenylene)s” Phys. Rev. B in print (2000)
[164] R. O. Loutfy, J. H. Sharp, C. K. Hsiao and Ho R ”Ph-
thalocyanine organic solar cells.” J.Appl.Phys. 52, 5218
(1981)
[165] A. Lux, G.G. Rozenberg, K. Petritsch, S.C. Moratti,
A.B. Holmes and R.H. Friend, ”A series of novel liquid
crystalline octakis(alkylthio)- substituted phthalocya-
nines” Synth.Met. 102, 1527-1528 (1999)
[166] A. Lux, private communication.
[167] A. Lux, A. B. Holmes, R. Cervini, J. E. Davies, S. C.
Moratti, J. Gruener, F. Cacialli and R. H. Friend ”New
CF3 substituted PPV type oligomers and polymers for
use as hole blocking layers in LEDs ” Synth.Met. ?
(1997)
[168] R. N. Marks, J. J. M. Halls, D. D. C. Bradley, R. H.
Friend and A. B. Holmes ”The photovoltaic response
in PPV thin film devices.” J. Phys.: Cond. Mat. 6,
1379-1394 (1994)
[169] J. C. Mello, F. Wittmann and R. H. Friend ”An im-
proved experimental determination of external photo-
luminescence quantum efficiency. ” Adv.Mat. 9, 230-
232 (1997)
[170] H. B. Michaelson ”The workfunction of the elements
and its periodicity ” J.Appl.Phys. 48, 4729-4733 (1977)
[171] L. Q. Minh, T. Chot, N. N. Dinh, N. N. Xuan, N. T.
Binh, D. M. Phuoc and N. T. Binh ”A new organic pho-
toconductor in the near infrared region of diode lasers.
absorption maxima between 780nm and 850nm. Solu-
ble in alpha-chlornaphthalene (?). ” Phys.Stat.Sol.(a)
101, K143 (1987)
[172] E. Moons, M. Eschle and M. Graetzel ”Determina-
tion of the energy diagram of the dithioketopyrrole /
SnO2:F heterojunction by surface photovoltage spec-
troscopy ” Appl.Phys.Lett. 71, 3305-3307 (1997)
[173] D. L. Morel, E. L. Stogryn, A. K. Ghosh, T. Feng, P. E.
Purwin, R. F. Shaw, C. Fishman, G. R. Bird and A. P.
Piechowski ”Organic photovoltaic cells” J.Phys.Chem.
88, 923-933 (1984)
[174] S. C. Moratti et al, ”High electron affinity polymers
for LEDs.” Synth. Met. 71 2117-2120 (1995)
[175] N. F. Mott, Trans. Faraday Soc. 34 500-506 (1938)
[176] N. F. Mott ”Note on the contact between a metal and
an insulator or semiconductor. ” Proc.Cambr.Phil. Soc.
34, 568-572 (1938)
[177] M. Mueller, F. Morgenroth, U. Scherf, T. Soczka-
Guth, G. Klaerner and K. Muellen ”The benzene ring
as modulus: Synthesis and supramolecular ordering of
novel conjugated polymers. ” Phil. Trans R. Soc. Lond.
A 355, 715-726 (1997)
[178] F. Nuesch, L. Si-Ahmed, B. Francois and L. Zuppiroli
”Derivatized electrodes in the construction of organic
LEDs.” Adv.Mat. 9, 222-225 (1997)
[179] C. K. Ober and G. Wegner ”Polyelectrolyte surfactant
complexes in the solid state: Facile building blocks for
self organizing materials.” Adv.Mat. 9, 17-31 (1997)
[180] A. R. Oezkaya, A. G. Guerek, A. Guel and
Bekaroglu Oe ”Electrochemical and spectral prop-
erties of octakis(hexylthio)-substituted phthalocya-
nines.” Polyhedron 16, 1877-1883 (1997)
[181] Ohnishi et al. Synth. Met. 57 4174-4179 (1993)
[182] B. O’Regan and M. Graetzel ”A low cost, high effi-
ciency solar cell based on dye-sensitized colloidal TiO2
films.” Nature 353, 737 (1991)
[183] ORIEL Corporation, ”Solar Simulation”, Solar Simu-
lator Product Guide, (1993).
[184] L. Oriol and J. L. Serrano ”Metallomesogenic poly-
mers. ” Adv.Mat. 7, 348-369 (1995)
[185] B. Ottenbourgs, H. Paulussen, P. Adriaensens,
D. Vanderzande and J. Gelan ”Characterization of
poly(isothianaphthene) derivatives and analogs by us-
ing solid state C13 NMR ” Synth.Met. 89, 95-102
(1997)
[186] K. Pakbaz, C. H. Lee, A. J. Heeger, T. W. Hagler
and D. McBranch ”Nature of the primary photoexcita-
tion in poly arylene vinylene ” Synth.Met. 64, 295-306
(1994)
[187] P. Panayotatos, G. Bird, R. Sauers, A. Piechowski and
S. Husain ”An aproach to the optimal design of p-n
heterojunction solar cells using thin film organic semi-
conductors.” Solar Cells 21, 301-311 (1987)
[188] Y. Park, V. Choong, Y. Gao, B. R. Hshieh and
C. W. Tang ”Work function of ITO transparent
conductor measured by photoelectron spectroscopy.”
Appl.Phys.Lett. 68, 2699-2701 (1996)
[189] Y. Park, V. Choong, E. Ettedgui, Y. Gao, B. R. Hsieh,
T. Wehrmeister and K. Muellen ”Energy level bending
and alignment at the interface between Ca and a pheny-
lene vinylene oligomer. ” Appl.Phys.Lett. 69, 1080-1082
(1996)
[190] Parrott ”Thermodynamics of solar cell efficiency ”
Sol.En.Mat.Sol.Cells 25, 73-85 (1992)
[191] H. Paulussen, D. Vanderzande and J. Gelan ”The
synthesis and characterization of soluble poly(iso
thia naphthene)- derivatives.” Synth.Met. 84, 415-416
(1997)
9.1. SUMMARY 133
[192] D. Pede, E. Smela, T. Johansson, M. Johansson and
O. Inganaes ”A general purpose conjugated polymer
device array for imaging.” Adv.Mat. 10, 233-237 (1998)
[193] Q. Pei, G. Zuccarello, M. Ahlskog and O. Inganaes
”Electrochromic and highly stable PEDOT switches
between opaque blue black and transparent sky blue.”
Polymer 35, 1347-1351 (1994)
[194] J. B. Pendry ”Photonic band structures.” J. Modern
Optics 41, 209-229 (1994)
[195] R. Peierls, Ann. Phys. 13, 905-952 (1932)
[196] K. Petritsch and R. H. Friend ”Ultrathin Organic Pho-
tovoltaic Devices. ” Synth.Met. 102, 976 (1998)
[197] K. Petritsch, W. Graupner, G. Leising and U. Scherf
”Photoinduced absorption in a Poly(para-phenylene)
ladder type polymer.” Synth.Met. 84, 625-626 (1997)
[198] K. Petritsch ”Optical Modulation Spectroscopy in Or-
ganic and Inorganic Semiconductors” Diploma Thesis,
Technical University Graz (1996)
[199] K. Petritsch, R.H. Friend, A. Lux, G. Rozenberg, S.C.
Moratti and A.B. Holmes ”Liquid crystalline phthalo-
cyanines in organic solar cells” Synth.Met. 102, 1776
(1999)
[200] K. Petritsch, J.J. Dittmer, E.A. Marseglia A. Lux,
G. Rozenberg, S.C. Moratti and A.B. Holmes and
R.H. Friend ”Dye based Donor/Acceptor Solar Cells”
Sol.En.Mat.Sol.Cells 61/1, 63-72 (2000)
[201] K. Petritsch, M. Granstrom, A.C. Arias, R.H. Friend,
A. Lux, G.G. Rozenberg, S.C. Moratti, A.B. Holmes
and M.R. Andersson ”Towards inexpensive solar cells”
Proc. 2nd World Conference and exhibition on photo-
voltaic solar energy conversion Vienna, 1998
[202] K. Petritsch et. al, unpublished results
[203] J. B. Pendry ”Photonic band structures.” J. Modern
Optics 41, 209-229 (1994)
[204] A.P. Piechowski, G.R. Bird, D. L. Morel and E.L.
Stogryn ”Desirable properties of photovoltaic dyes ”
J.Phys.Chem. 88, 934-950 (1984)
[205] M. Pope and C.E Swenberg, Electronic Processes in
Organic Crystals, Clarendon Press, Oxford (1982)
[206] M. B. Prince ”Silicon Solar Energy Converters.”
J.Appl.Phys. 26, 534-540 (1955)
[207] H. Quante, P. Schlichting, U. Rohr, Y. Geerts and
K. Muellen ”Novel perylene-containing polymers. ”
Macromol. Chem.Phys. 197, 4029-4044 (1996)
[208] E. Rabinowitch, ”Photosynthesis”, Wiley, New York
(1969)
[209] A. Rademacher, S. Maerkle and H. Langhals ”Loes-
liche perylen- Fluoreszenzfarbstoffe mit hoher photo-
stabilitaet.” Chem.Ber. 115, 2927-2934 (1982)
[210] H. Reisch, U. Wiesler, U. Scherf and N. Tuy-
tuylkov ”Poly(indenofluorene)(PIF), a novel low band
gap polyhydrocarbon” Macromolecules 29, 8204-8210
(1996)
[211] L. S. Roman, I. A. Huemmelgen, F. C. Nart, L. O.
Peres and de Sa E L ”Determination of electro-affinity
and ionization potential of conjugated polymers via
Fowler-Nordheim tunneling measurements: theoretical
formulation and application to PPV ” J.Chem.Phys.
105, 10614-620 (1996)
[212] L. S. Roman, M. R. Andersson, T. Yohannes and O.
Inganaes ”Photodiode performance and nano-structure
of polythiophene/c60 blends. ” Adv.Mat. 9, 1164-1168
(1997)
[213] L. S. Roman, W. Mammo, L. A. A. Pettersson, M. R.
Andersson and O. Inganaes ”High quantum efficiency
polythiophene/C60 photodiodes. ” Adv.Mat. 10, 774-
777 (1998)
[214] J. Rostalski and D. Meissner ”Monochromatic versus
Solar Efficiencies of Organic Solar Cells” Solar Energy
Mater. Solar Cells 61, 87-95 (2000)
[215] L. J. Rothberg, M. Yan, S. Son, M. E. Galvin, E. W.
Kwock, T. M. Miller, H. E. Katz, R. C. Haddon and
F. Papadimitrakopoulos ”Intrinsinc and extrinsinc con-
straints on PPV electroluminescence.” Synth.Met. 78,
231-236 (1996)
[216] L. J. Rothberg, M. Yan, A. W. P. Fung, T. M. Jedju, E.
W. Kwock and M. E. Galvin ”Photogeneration mecha-
nism and exciton binding energy in phenylenevinylene
polymers.” Synth.Met. 84, 537-538 (1997)
[217] H. Quante, P. Schlichting, U. Rohr, Y. Geerts and
K. Muellen ”Novel perylene-containing polymers. ”
Macromol. Chem.Phys. 197, 4029-4044 (1996)
[218] R. J. Regensburger, J. J. Jakubowski; U.S. Patent
3,904,407 (1975)
[219] F. Reinitzer, Monatsh. Chem. 9, 421 (1888)
[220] E. H. Rhoderick and R. H. Williams, Metal-
Semiconductor Contacts, Clarendon Press, Oxford
[221] M. Samoc, A. Samoc, B. L. Davies and U. Scherf ”Sat-
urable absorption in PIF - a picket fence polymer. ”
Opt.Lett. in print (1998)
[222] D. J. Sandman ”Semiconducting polymers and their
solid state properties.” TRIP 2, 44 (1994)
[223] Savenije et al. Phys.Rev.Lett
[224] J.H. Schon, C. Kloc, E. Bucher, B. Batiogg, ”Efficient
organic photovoltaic diodes based on doped pentacene”
Nature 403 408-410 (2000)
[225] W. Schottky, Naturwiss. 26, 843 (1938)
[226] A. J. Schouten, J. M. Warman, M. P. de Haas, M. A.
Fox and H. L. Pan ”Charge migration in supramolecu-
lar stacks of peripherally substituted porphyrins.” Na-
ture 353, 736-737 (1991)
[227] P. G. Schouten, J. F. Van der Pol, J. W. Zwikker,
W. Drenth and S. J. Picken ”Peripherally octasubsti-
tuted phthalocyanines with branched alkoxy chains.”
Mol.Cryst.Liq.Cryst. 195, 291 (1991)
134 CHAPTER 9. SUMMARY AND BIBLIOGRAPHY
[228] P. G. Schouten, J. M. Warman, M. P. de Haas, C.
F. van Nostrum, G. H. Gelinck, R. J. M. Nolte, M. J.
Copyn, J. W. Zwikker, M. K. Engel, M. Hanack, Y. H.
Chang and W. T. Ford ”The effect of structural modi-
fications on charge migration in mesomorphic phthalo-
cyanines.” J. Am. Chem. Soc. 116, 6880-6894 (1994)
[229] W. Shockley ”The Theory of p-n Junctions in Semi-
conductors and p-n junction transistors. ” Bell Syst.
Techn. Journ. 28, 435-489 (1949)
[230] W. Shockley and H. J. Queisser ”Detailed balance limit
of efficiency of PN junction solar cells ” J.Appl.Phys.
32, 510 (1961)
[231] J. Simon and J. J. Andre ”Molecular Semiconductors.”
Springer Verlag, Heidelberg (1985)
[232] H. Sirringhaus, N. Tessler and R.H. Friend, Science
280, 1741 (1998)
[233] T.A. Skotheim ”Handbook of Conducting Polymers”,
New York & Basel, Vol1 (1986)
[234] G. Smestad and H. Ries ”Luminescence and current
voltage (IV) characteristics of solar cells and optoelec-
tronic devices.” Sol.En.Mat.Sol.Cells 25, 51-71 (1992)
[235] G. Smestad ”Testing of dye sensitized TiO2 solar cells
II: Theoretical voltage output and photoluminescence
efficiencies.” Sol.En.Mat.Sol.Cells 32, 237-288 (1994)
[236] G. Smestad, C. Bignozzi and R. Argazzi ”Test-
ing of dye sensitized TiO2 solar cells I: Experimen-
tal photocurrent output and conversion efficiencies.”
Sol.En.Mat.Sol.Cells 32, 259-272 (1994)
[237] Solar Cell Efficiency Tables (Version12), Progress
in Photovoltaics: Research and Applications 7
(1998). Website (University of New South Wales):
www.pv.unsw.edu.au/eff/
[238] Stampfl, W. Graupner, G. Leising and U. Scherf ”Pho-
toluminescence and UV-VIS absorption study of PPP-
type ladder polymers.” Journal of Luminescence 63,
117-123 (1995)
[239] Stampfl, S. Tasch, G. Leising and U. Scherf ”Quantum
efficiencies of electroluminescent poly para phenylenes
” Synth.Met. 71, 2125-2128 (1995)
[240] M. Stevens, B. Weir, G. Denton and R.H. Friend
”Laser Ablation of Conjugated Polymers” Synth. Met
101, 234-235 (1999)
[241] S. M. Sze ”Physics of Semiconductor Devices.”, John
Wiley & Sons, New York (1981)
[242] K. Tada, M. Onada, A. A. Zakhidov and K. Yoshino
”Characteristic of poly(p-pyridyl vinylene)/poly (3-
alkyl-thiophene) heterojunction photocell.” Jpn. J.
Appl. Phys. 36, L306-L309 (1997)
[243] K. Tada and K. Yoshino ”Conducting polymer color
sensor.” Jpn. J. Appl. Phys. 36, L1351-53 (1997)
[244] C. W. Tang ”Two-layer organic photovoltaic cell. ”
Appl.Phys.Lett. 48, 183 (1986)
[245] C. W. Tang ”Multilayer organic photovoltaic elements.
” US Patent 4,164,431 (1981)
[246] S. Tasch, E. J. W. List, O. Ekstroem, W. Graup-
ner, G. Leising, P. Schlichting, U. Rohr, Y. Geerts,
U. Scherf and K. Muellen ”Efficient white light emit-
ting diodes realized with new processable blends of
conjugated polymers. ” Appl.Phys.Lett. 71, 2883-2885
(1997)
[247] S. Tasch, C. Brandstaetter, F. Meghdadi, G. Leising,
G. Froyer and L. Athouel ”Red green blue light emis-
sion from a thin film electroluminescence device based
on parahexaphenyl.” Adv.Mat. 9, 33-36 (1997)
[248] N. Tessler, N. T. Harrison, R. H. Friend Adv.Mat.10
10, 64 (1998)
[249] N. Tessler, G. J. Denton and R. H. Friend ”Lasing
from conjugated polymer microcavities.” Nature 382,
695 (1996)
[250] N. Tessler, P. L. Burns, H. Becker and R. H. Friend
”Suppressed angular color dispersion in planar micro-
cavities. ” Appl.Phys.Lett. 70, 556-558 (1997)
[251] T. Tsuzuki , Y. Shirota, J. Rostalski, D. Meissner ”The
Effect of Fullerene Doping on Photoelectric Conversion
Using Titanyl Phthalocyanine and a Perlene Pigment”
Sol.En.Mat.Sol.Cells 61, 1-8 (2000)
[252] United Nations Environment Programme (UNEP)
”Global Environment Outlook (GEO)-2000”, Earth-
scan Publications Ltd., London (2000). Web site:
www.unep.org
[253] J. F. Van der Pol, E. Neeleman, J. W. Zwikker,
R. J. M. Nolte, W. Drenth, J. Aerts, R. Visser and
S. J. Picken ”Homologous series of liquid-crystalline
metal free and copper octa-n-alkoxy phthalocyanines.”
Liq.Cryst. 6, 577-592 (1989)
[254] J. F. Van der Pol, E. Neeleman, J. C. Van Mil-
tenburg, J. W. Zwikker, R. J. M. Nolte and W. Drenth
”A POLYMER with the mesomorphic order of liq-
uid crystalline phthalocyanines (Polymerization in the
mesophase).” Macromolecules 23, 155-162 (1990)
[255] J. F. Van der Pol, M. P. de Haas, J. M. Warman and
W. Drenth ”Sudden conduction (10E-3 to 0.1 S/m)at
the solid to mesophase transition of octa n alkoxy sub-
stituted phthalocyanines.” Mol.Cryst.Liq.Cryst. 183,
411-420 (1990)
[256] J. F. Van der Pol, M. P. de Haas, J. M. War-
man and W. Drenth ”Erratum - Contributions
from dipolar relaxation to microwave conductivity. ”
Mol.Cryst.Liq.Cryst. 195, 307-308 (1991)
[257] J. A. E. H. Van Haare, L. Groenendaal, H. W. I.
Peerlings, E. E. Havinga, J. A. J. M. Vekemans, R.
A. Janssen and E. W. Meijer ”Unusual redox behav-
ior of oligoheteroaromatic compounds: An Increasing
first oxidation potential with increasing conjugation
length.” Chemistry of Materials 7, 1984 (1995)
[258] J. A. E. H. Van Haare ”A general procedure toward
colorless transparent conducting polymer films.”PhD
thesis (1997)
[259] C. F. van Nostrum and R. J. M. Nolte ”Supramolecu-
lar architectures from phthalocyanine building blocks.”
Macromol.Symp. 77, 267-276 (1994)
9.1. SUMMARY 135
[260] C. F. van Nostrum, A. W. Bosman, G. H. Gelinck, P.
G. Schouten, J. M. Warman, A. P. M. Kentgens, A.
C. Devillers, A. Meijerink, S. J. Picken, U. Sohling, A.
J. Schouten and J. M. Nolte ”Supramolecular struc-
ture, physical properties, and Langmuir Blodget film
formation of an optically active liquid crystalline ph-
thalocyanine.” Chem. Eur. J. 1, 171-182 (1995)
[261] C. F. van Nostrum, S. J. Picken, A. J. Schouten and R.
J. M. Nolte ”Synthesis and supramolecular chemistry
of novel liquid crystalline crown ether substituted ph-
thalocyanines: Toward molecular wires and molecular
ionoelectronics.” J. Am. Chem. Soc. 117, 9957-9965
(1995)
[262] C. F. van Nostrum ”Self assembled wires and chan-
nels.” Adv.Mat. 8, 1027-1030 (1996)
[263] J.A.E.H. Van Haare ”A general procedure toward col-
orless transparent conducting polymer films.” - Title of
thesis: Redox states of conjugated oligomers and poly-
mers.” PhD thesis (1997)
[264] S. C. Veenstra, G. G. Malliaras, H. J. Brouwer, F.
J. Esselink, V. V. Krasnikov, P. F. van Hutten, J.
Wildeman, H. T. Jonkman, G. A. Sawatzky and G.
Hadziioannou ”Sexithiophene+C60 blends as model
systems for photovoltaic devices. ” Synth.Met. 84, 971-
972 (1997)
[265] G.H. Wannier Phys. Rev. 52, 191-197 (1937)
[266] V. Wan unpublished results.
[267] J. B. Whitlock, P. Panayotas, G. D. Sharma, M. D.
Cox, R. R. Sauers and G. R. Bird ”Investigations of
materials and device structures for organic semicon-
ductor solar cells.” Optical Engineering 32, 1921-1933
(1993)
[268] D. Woehrle and D. Meissner ”Organic solar cells”
Adv.Mat. 3, 129 (1991)
[269] D. Woehrle, M. Eskes, K. Shigehara, A. Yamada, Syn-
thesis 194 (1993)
[270] J.P. Wolfe and A. Mysyrowicz, ”Excitonic Matter”
Scientific American 250 No.3, 70 (1984)
[271] C.C. Wu , C.I. Wu , J. C. Sturm and A. Kahn ”Surface
modification of ITO by plasma treatment: An effective
method to improve the efficiency, brightness, and relia-
bility of organic LEDs” Appl.Phys.Lett. 70, 1348-1350
(1997)
[272] F. Wudl, M. Kobayashi and A.J. Heeger J.Org.Chem.
49 3382 (1984)
[273] Y. P. Yadava, G. Denicolo, A. C. Arias, L. S. Roman
and I. A. Huemmelgen ”Preparation and characteri-
zation of transparent conducting tin oxide (TO) film
electrodes by chemical vapour deposition from reactive
thermal evaporation of SnCl2.” Mat.Chem.Phys. 48,
263-267 (1997)
[274] Y. Yang, Q. Pei and A. J. Heeger ”Efficient blue poly-
mer light emitting diodes from a series of soluble PPP
derivatives. J.Appl.Phys. 79, 934-939 (1996)
[275] S. Yamada, S. Tokito, T. Tsutsui, S. Saito J. Chem.
Soc., Chem Commun. 1448 (1987).
[276] K. Yoshino, K. Tada, M. Hirohata, H. Kajii, Y. Hi-
ronaka, N. Tada, Y. Kaneuchi, M. Yoshida, A. Fu-
jii, M. Hamaguchi, H. Araki, T. Kawai, M. Ozaki, Y.
Ohmori, M. Onada and A. Zakhidov ”Novel properties
of molecularly doped conducting polymers and junction
devices. ” Synth.Met. 84, 477-482 (1997)
[277] K. Yoshino, K. Tada, A. Fujii, E. M. Conwell and A. A.
Zakhidov ”Novel photovoltaic devices based on donor-
acceptor molecular and conducting polymer systems.”
IEEE trans. Electron Devices 44, 1315-1324 (1997)
[278] Yu G, J. Gao, C. Hummelen Jan, F. Wudl and A. J.
Heeger ”Polymer photovoltaic cells: Enhanced efficien-
cies via a network of internal donor acceptor hetero-
junctions. ” Science 270, 1789 (1995)
[279] C. Zenz, W. Graupner, S. Tasch, G. Leising, K.
Muellen, U. Scherf Appl. Phys. Lett 71 2566 (1997).
[280] M. Zeilik and E.v.P. Smith ”Introductory: Astronomy
and Astrophysics”, 2nd Edition, CBS College Publish-
ing (1987)
[281] F. Zhu and J. Singh ”On the optical design and
thin film amorphous silicon solar cells.” Sol.En.Mat.
Sol.Cells 31, 119-131 (1993)
[282] F. Zhu, P. Jennings, J. Cornish, G. Hefter and K.
Luczak ”Optimal optical design of thin film photovol-
taic devices. ” Sol.En.Mat.Sol.Cells 49, 163-169 (1997)
[283] ”Method lets pigments be processed like dyes.” Chem-
istry and Engineering News 21, 56 (1998)
136 CHAPTER 9. SUMMARY AND BIBLIOGRAPHY
Chapter 10
Appendices
10.1 Details of the Sample
Preparation
Unless stated otherwise, the preparation of sub-
strates and devices that have been used in the
Chapters of this thesis has been performed as de-
scribed in the following Sections.
10.1.1 ITO Substrates
The ITO substrates have been patterned by remov-
ing 2.5mm wide stripes of ITO from either side via
etching in 20% HCl. The results is a conducting
area of the form of a 7mm wide stripe in the middle
of the substrate which is used as electrode. During
the etching process the stripe in the middle was pro-
tected with a solid self adhesive photoresist sheet
1
.
Note that the photoresist was only used as protec-
tive layer during etching and was illuminated with a
UV lamp (before etching) for 2min so that it can be
readily removed in two mechanical cleaning steps:
rubbing the ITO surface with cotton buds using ace-
tone and isopropanol (IPA).
The ITO substrates were then cleaned in ultra-
sonic baths with acetone (5min) followed by IPA
(5min) in a laminar flow box.
10.1.2 Preparation of the Dye and
Polymer Solutions
In order to obtain the desired concentrations of the
polymer or dye solutions (typically between 2 and
30mg/ml), we were weighing the required amount
of powder using a micro-scale in small vials (<2ml).
1
Later we used substrates which we received already pre-
patterned. However, when received from the supplier/cutter
the substrates were hold in place with a scotch tape. We
used the same (pre-)cleaning procedure to remove remaining
parts of the tape for these pre patterned substrates as for
the home etched ones.
The solvent - usually chloroform - was added us-
ing disposable poly-propylene syringes and medium
size needles. We did not use syringes with standard
rubber plunges since the rubber dissolves quickly in
chloroform which could lead to unknown effects in
the organic film
2
.
The solution was then shaked using an automatic
roller device for typically more than an hour while
avoiding to touch the vial caps
3
.
The solution was then filtered using polypropy-
lene filters with pore sizes between 0.1 and 2.5µm
- depending on the size of the molecule. Small
molecules like dyes (Per or HPc) could be filtered
using 0.1µm while polymers with high molecular
weight require larger pore sizes. For example MEH-
PPV may only be filtered through 2.5µm. The re-
quired minimal pore size can be estimated if the
average size (length) of the specific molecule is cal-
culated. Since we preferred to use fresh solutions
so the required amount of solvent was usually less
than 1ml. We found that about 3 films can be spun
with 0.25ml solvent using 5 drops per substrate.
10.1.3 Film Deposition
The substrate was briefly (1-3s) rinsed with (IPA)
and subsequently blow dried (≈ 10s) with nitro-
gen gas. About 5 drops of the solution were then
dropped onto the patterned ITO substrates using
disposable glass cuvettes - in a laminar flow box.
Spin-speeds were typically 2000rpm with spinning
times of 60s. The spun film was then transferred
into the evaporation chamber where the top con-
tact - usually Al - was sublimed.
2
We could find significant absorption and PL of the dis-
solved rubber in solution.
3
The vial caps had teflon layers inside which should be
fairly resistant against chloroform - however, the long term
exposure might allow solvent vapor to reach the vial cap
which is actually made of some less resistant plastic material.
137
138 CHAPTER 10. APPENDICES
10.1.4 Sublimation of the Top Elec-
trode
In order to complete the photovoltaic device, the
top electrode - usually Al - was thermally sublimed
onto the organic film. The pattern of the top elec-
trodes was initially that shown in Fig.10.1 - so that
8 different electrodes allowed contacting of 8 dif-
ferent pixels on one substrate. Later the pattern
was changed to 6 electrodes (pixels) to obtain larger
pixel areas.
The larger pixel areas have been chosen so that
the whole light beam could illuminate the electrode.
Since the intensity of the whole light beam is the
same as measured with the silicon photodiode the
actual device area does not need to be known to
calculate the EQE. This allowed more accurate de-
termination of the EQE compared to the previously
method where the pixels have been smaller than
the light beam. In the latter, the actual size of the
pixel and photodiode had to be determined which
increased the error-bar of the final EQE number.
The more accurate method was chosen since some
devices (like the laminated polymer cell) reached
high EQE numbers where higher accuracy was de-
sired.
Typical pressures during evaporation were
around 10
−6
mbar.
The sublimation-mask used was made of
brass with 9 substrate positions. When semi-
transparency (T > 20%) was required the Al film
was between 10 and 20nm thin - otherwise the thick-
ness was between 50 and 100nm.
Figure 10.1: Planar view of the substrate - electrode
pattern with the 8 pixel mask. The vertical stripe
in the middle represents the ITO area (=bottom
electrode). The substrate dimensions are 12x12mm.
Figure courtesy of Ref. [63].
10.2 Details of the EQE and
IV Measurements
After the sublimation of the top electrode the sub-
strates where taken to the setup and dots of silver-
paste solution were applied where the wires of the
sample holder contact the electrodes in order to en-
sure a good electrical connection. Unless stated oth-
erwise, all devices have been exposed to air for at
least 24hrs prior to the measurement. After this pe-
riod air molecules are assumed to have penetrated
the entire bulk so that all devices can be compared
even though some have seen air for one or two days
longer than others.
In this way we were able to obtain fairly repro-
ducible results. The fabrication of a set of 9 sub-
strates - each containing 6 to 8 devices or pixels
can be finished within 12hrs including dissolving the
molecules and patterning the ITO.
Figure 10.2: Experimental setup for the EQE and
IV measurements. Figure courtesy of Ref. [63].
The experimental setup used for the EQE and
IV measurements is shown in Fig.10.2. The light
source is a 100W Bentham quartz tungsten halo-
gen lamp with a spectral emission range larger than
350nm to 1300nm and an intensity peak at 880nm.
The lamp is driven by a Bentham 505 current sta-
bilised power supply. The white light was dispersed
by a Czerny-Turner Bentham M300 single grating
monochromator using a 1200 lines/mm grating and
a blaze wavelength of 500nm.
The sample holder allowed automatic contacting
10.2. DETAILS OF THE EQE AND IV MEASUREMENTS 139
of the 8 or 6 pixels + the ITO contact using spring
contacts made of berylium-copper. The holder can
be closed to apply a vacuum and is still optically
accessible through quartz windows. A remotely
controlled shutter allowed convenient switching be-
tween light and dark measurements for both the IV
and the SR measurements. A switchboard serves to
switch between different electrodes/pixels.
The setup was initially designed and built by
Anna Koehler and later modified by the author.
A 580nm filter was used for the range from 600-
1200nm to block the higher orders of the monochro-
mator grating.
We note that the photocurrent can vary consid-
erably between different pixels for various reasons
(shorts, quality of the contact, poor film quality
etc.). A quick method to find the best working
pixel is to expose the entire substrate to a white
light source (e.g. a 100W desk lamp) and switch
between the pixels while observing the differences
between dark and light current
4
.
Figure 10.3: Typical light intensities (E) at the sam-
ple position as a function of wavelength. The offset
at 620nm is due to the filter change.
Every set of SR measurements was preceded by a
measurements of the light intensity using a silicon
photodiode (HAMAMATSU 5106) with a known
spectral response. The light intensity can vary de-
pending on setup details but is generally constant
within ±20% for different sets of measurements (on
different days/weeks). Within the same set, we
4
A negative bias may be used to enhance the effect for
devices with very low EQE.
could not find a significant variation of the light
intensity.
A typical dependence of the light intensity at
the sample position on the wavelength is shown
in Fig.10.3. The offset around 620nm is due to a
580nm edge filter which is manually removed when
the response to wavelengths smaller than 620nm is
measured. The dark current was always assumed
to be wavelength independent and subtracted from
the photocurrent before the calculation of EQE.
The EQE was determined using Eq. (2.33) in
Chapter 2. The EQE spectrum of the silicon refer-
ence diode using this procedure is shown in Fig.10.4.
Both absolute values and the shape are as ex-
pected for commercial silicon photodiodes and con-
firm the validity of the applied mathematical pro-
cedure. The calculation of the EQE and irradiance
was performed using an Excel spreadsheet designed
by the author.
Figure 10.4: EQE spectrum and spectral response
of the silicon reference diode (Hamamatsu S5106).
We assumed that the efficiency of our cells does
not vary significantly with intensity in the used irra-
diance range. This assumption appears reasonable
since results on e.g. MEH-PPV gave similar results
using a completely different setup using a xenon
lamp. Hence the intensity distribution of the inci-
dent photons was assumed to have no effect on the
calculated EQE.
The IV curves were typically measured for wave-
lengths which gave the highest currents - which
is not necessarily the highest EQE - starting in
the negative voltage range. The voltage range was
roughly scanned manually avoiding high currents
which can destroy the device.
Depending on the open circuit voltage and the
chosen voltage range the step size was between 20
140 CHAPTER 10. APPENDICES
and 100mV. The dark characteristic was measured
before and (again) after the light characteristic has
been measured to confirm that the device behaviour
did not change i.e. the device was destroyed.
As observed by other researchers in the same
group the shape of the IV curves can vary more for
different pixels than the EQE spectra [114, 64, 150].
We note that we did not use any mathematical
smoothing procedure in more than 90% of the IV
or EQE curves in this thesis. When smoothing was
applied then to a very low extent (< 1 − 3%)only
to remove obvious and disturbing artifacts.
10.2.1 Note on the Dark Current at
0V
It can be assumed that the sometimes observed oc-
currence of a dark current (I
d
) at 0V is - at least
partly - due to the limited ability of the source mea-
sure unit (SMU) to apply exactly 0V±0V . The zero
point error of the SMU cannot be much smaller than
a few mV due to a finite resolution and non-ideal
ground levels. Thus, these few mV can drive the
current I
d
through the cell with the actual size de-
pending on R
sh
.
The sign of I
d
depends on the sign of the usually
non zero ground potential plus the zero point error
of the SMU itself. Note that devices with a very
low shunt (e.g. partly shorted or doped devices)
can show positive or negative
5
I
d
values of many
nA. Note that upon illumination such devices can
still generate a negative photocurrent and therefore
still work as solar cells. These devices show usually
no significant open circuit voltage.
Another possibility to explain the existence of
such a dark current is to consider the capacity (C)
of the device:
C =
0

r
A
d
(10.1)
with = 8.9 · 10
−12
J
−1
C
2
m
−1
, A ≈ 10
−6
m
2
,

r
≈ 2 we obtain capacities between 10 and 100pF
for a thickness d ranging between 10 and 100nm
respectively.
Together with the (series) resistor data from Ta-
ble 3.5 we obtain time constants (τ):
τ = RC (10.2)
5
In this thesis negative currents at >= 0V are defined
as currents which are generated by the device upon illumi-
nation. However,the dark current due to externally applied
voltage has always to be considered (subtracted) to obtain
the actually photogenerated current.
between ms and several 10s. Considering the
equation for discharging a capacitor:
U = U
0
· e

t
τ
(10.3)
this means that, once a device has seen light (i.e.
charged), it can take many seconds i.e. minutes to
discharge via the series resistor even if the device is
externally shorted (V=0V).
10.3 Absorption
Measurements
Optical absorption spectra of solutions and thin
films were measured with a Hewlett Packard 8453
UV-Vis spectrometer. The spectra were corrected
for the transmission of the substrate (ITO on
quartz) which was measured before a set of absorp-
tion spectra was taken. Most absorption spectra
were taken from the actual device by directing the
light beam through the space between the pixels
using a suitable (diameter=0.8mm) mask.
10.4 Thickness
Measurements
The thickness of the organic films was measured us-
ing a profilometer (Sloan Dektak IIa) by scratching
the film carefully with a sharp tweezer or a tooth-
pick avoiding to scratch into the substrate. We typ-
ically took 3 measurements at different positions on
the substrate and calculated the mean value. For
films thinner than 30nm or particularly soft mate-
rials like the liquid crystal HPc we also used the
absorption data to determine the film thickness. In
these cases the absorption coefficient was obtained
using thicker films were the profilometer results are
accurate enough.
In order to measure the thickness of very soft
films we sublimed a layer of Al on top of the film
with the scratch. For some materials like MEH-
PPV we found that drying in air at room tempera-
ture for about three days makes the surface harder
so that it can be readily measured using the pro-
filometer. In very doubtful cases AFM scans across
a scratch could be used.
10.6. LIST OF PUBLICATIONS 141
10.5 Thermal Polarisation
Microscopy
For the investigations of liquid crystalline phases
a standard microscope was equipped with a video
capturing system (PC+video camera+software)
and a hotstage (LINKAM). Moreover, the micro-
scope was fitted with a set of polarization filters
one of which could be rotated easily to any desired
angle with respect to the other.
The hotstage was also fitted with a cooling sys-
tem using liquid nitrogen that allowed us not only
to cool and heat the sample quickly to any temper-
ature between -180
o
and 300
o
but also to purge the
hotstage and sample environment with a constant
flux of nitrogen gas to avoid effects of air at those
temperatures.
The geometry of the hotstage requires a spe-
cial microscope objective (20x) with a long work-
ing distance when higher magnifications were de-
sired. We note that higher magnification could also
be achieved using stronger magnifying eye-pieces.
10.6 List of Publications
1. H. Ago, Th. Kugler , F. Cacialli, K. Petritsch,
R.H. Friend, W.R. Salaneck, Y. Ono, T.
Yamatabe and K. Tanaka ”Workfunction of
Purified and Oxidised Carbon Nanotubes.”
Synth.Met. 103, 2494-2495 (1998)
2. H. Ago, K. Petritsch, M.S.P. Shaffer,
A.H.Windle and R.H. Friend ”Composites of
carbon nanotubes and conjugated polymers for
photovoltaic devices” Adv.Mat. 11, 1281-1285
(1999)
3. A.C. Arias, K. Petritsch, M. Granstrom, J. J.
Dittmer, E. A. Marseglia and R. H. Friend ”Ef-
ficient Polymeric Photovoltaic Devices.” Proc.
CBECIMAT (1998)
4. A.C. Arias, M. Granstrom, D.S. Thomas, K.
Petritsch and R.H. Friend ”Doped conducting-
polymer-semiconducting-
polymer interfaces: their use in organic pho-
tovoltaic devices” Phys.Rev.B - Cond. Matt.
60, 1854-1860 (1999)
5. A.C. Arias, M. Granstrom, K. Petritsch and
R.H. Friend ”Organic photodiodes using poly-
meric anodes” Synth. Met. 102, 953-954
(1999)
6. J.J. Dittmer, K. Petritsch, E.A. Marseglia,
R.H. Friend, H. Rost and A.B. Holmes ”Pho-
tovoltaic properties of MEH-PPV/PPEI blend
devices” Synth.Met. 102, 879 (1999)
7. J.J. Dittmer, P. Lazaroni, P. Leclere, P.
Moretti, M. Granstrom K. Petritsch, E.A.
Marseglia, R.H. Friend, H.Rost and A.B.
Holmes ”Crystal Network Formation in Or-
ganic Solar Cells” Sol.En.Mat.Sol.Cells 61, 53-
61 (2000)
8. M. Granstrom, K. Petritsch, A.C. Arias, A.
Lux, M.R. Andersson and R. H. Friend ”Lam-
inated fabrication of polymeric photovoltaic
diodes.” Nature 395, 257-260 (1998)
9. M. Granstrom, K. Petritsch: World Patent
Nr.:WO9949525A1: MULTILAYER PHOTO-
VOLTAIC OR PHOTOCONDUCTIVE DE-
VICES, Filing date: Febr.02.1999
10. M. Granstrom, K. Petritsch: Australia Patent
Nr.: AU2433799A1: MULTILAYER PHOTO-
VOLTAIC OR PHOTOCONDUCTIVE DE-
VICES, Filing date: Febr.02.1999
11. M. Granstrom, K. Petritsch, A.C. Arias and
R. H. Friend ”High Efficiency Polymer Photo-
diodes.” Synth. Met. 102 957-958 (1999)
12. W. Graupner, K. Petritsch, G. Leising, G.
Lanzani, M. Nisoli, S. Desilvestri, U. Scherf,
”Femtosecond optical dynamics in a poly(para-
phenylene) type ladder polymer.”, MRS-
Proceedings 413 615 (1996)
13. W. Graupner, S. Eder , K. Petritsch, G. Leis-
ing and U. Scherf ”Origin and Stabilization
of Photoexcitations in Conjugated Polymers”
Synth.Met. 84, 507-510 (1997)
14. W. Graupner, T. Jost, K. Petritsch, S.
Tasch, F. Meghdadi, G. Leising, W. Graup-
ner and A. Hermetter ”Optoelectronic Proper-
ties of Polyphenyls.” ANTEC97-Plastics sav-
ing planet earth 1-3, 1339-1343 (1997)
15. E.J.W. List, J. Partee, J. Shinar, U. Scherf,
K. Mullen, E. Zojer, K. Petritsch, G. Leis-
ing and W. Graupner ”Localized triplet ex-
citations and the effect of photo-oxidation in
ladder-type poly(p-phenylene)s and oligo(p-
phenylene)s” Phys. Rev. B in print (2000)
16. A. Lux, G.G. Rozenberg, K. Petritsch,
S.C. Moratti, A.B. Holmes and R.H.Friend,
142 CHAPTER 10. APPENDICES
”A series of novel liquid crystalline octa-
kis(alkyl-thio)- substituted phthalocyanines”
Synth.Met. 102, 1527-1528 (1999)
17. K. Petritsch ”Optical Modulation Spec-
troscopy in Organic and Inorganic Semicon-
ductors” Diploma Thesis, Technical University
Graz (1996)
18. K. Petritsch, W. Graupner, G. Leising and
U. Scherf ”Photoinduced absorption in a
Poly(para-phenylene) ladder type polymer.”
Synth.Met. 84, 625-626 (1997)
19. K. Petritsch and R. H. Friend ”Ultrathin Or-
ganic Photovoltaic Devices.” Synth.Met. 102,
976 (1998)
20. K. Petritsch, M. Granstrom, A.C. Arias, R.H.
Friend, A. Lux, G.G. Rozenberg, S.C. Moratti,
A.B. Holmes and M.R. Andersson ”Towards in-
expensive solar cells” Proc. 2nd World Confer-
ence and exhibition on photovoltaic solar en-
ergy conversion Vienna, (1998)
21. K. Petritsch, R.H. Friend, A. Lux, G.G. Rozen-
berg, S.C. Moratti and A.B. Holmes ”Liq-
uid crystalline phthalocyanines in organic solar
cells” Synth.Met. 102, 1776 (1999)
22. K. Petritsch, J.J. Dittmer, E.A. Marseglia,
A. Lux, G.G. Rozenberg, S.C. Moratti, A.B.
Holmes and R.H. Friend ”Dye based Donor-
Acceptor Solar Cells” Sol.En.Mat.Sol.Cells
61/1, 63-72 (2000)
10.7 Conferences
I have presented results of my research at the fol-
lowing conferences:
• Spring 1999: European network meeting
on “Synthesis and Structure of Electroac-
tive Organic Architectures” (SELOA), Bologna
(Italy); 2 talks (5min. and 20min.)
• July 12-18, 1998: “International Conference
on Science and Synthetic Metals” (ISCM’98),
Montpellier (France); 7 poster presentations.
• August, 1998: European network meeting
“Synthesis and Structure of Electroactive Or-
ganic Architectures” (SELOA), Faro (Portu-
gal); 1 talk (20min.)
• Dec. 3-5, 1998: “European Conference on
Organic Solar Cells” (ECOS’98), Cadarache
(France); 1 talk (30min.)
• July 6-10, 1998: “2nd World Conference and
Exhibition on Photovoltaic Solar Energy Con-
version” (WCPEC), Vienna (Austria); 1 poster
presentation.
10.8 Curriculum Vitae of the
Author
1969 02. 02. Born in Graz (Austria)
1975 - 1979 Elementary School
1979 - 1983 Grammar School
1983 - 1988 Technical Grammar School of
Electronics and Telecomunications
1988 - 1996 Study of Applied Physics
at the Technical University Graz
Febr.,1996 Research project (2 months) at the
Max Planck Institut f¨ ur
Festk¨orperforschung in Stuttgart
(Germany) in the field of thin film
monocrystalline silicon solar cells.
Sept.,1996 Final exam with honours
1996-1999 Post as research assistant at the
University of Cambridge, UK, in
the opto-electronics group lead by
Prof. R.H. Friend. The project
was focused on the development of
efficient organic solar cells.
1999-2000 Completion of the PhD thesis using
the results from the work in
Cambridge
The author can be contacted under email:
klaus.petritsch@solarpassion.com

ii

For my dear mother

“If we did all the things we are capable of doing, we would astound ourselves.”
Thomas A. Edison

iii

iv .

Dr. T. Dr.without each of them this work would not have been possible. Dr. London. the o realisation and success of this project was only possible as a result of an excellent collaboration among the coauthors in Ref.Moons.C.Prof.N.Greenham.J.Ho.Dr.E.G.Sirringhaus.H. Dr. o o Lazzaroni. Dr. Thanks to the special relationship between the Cavendish Laboratory and its spin-off company Cambridge Display Technology (CDT) I had the opportunity to use the excellent AFM microscope and thereby gain a little insight into the industrial aspect of leading edge technology and research in this exciting new field .Chawdherry.A.J. Dr.G. de Mello.Weir.A.Denton.for which I am very grateful. Granstr¨m. M.[94] . Furthermore Dr. I am particularly indebted to Dr. A.Koch as well as Dr.N. P.Cacialli.Prof.A.Ramaprakash who spent many hours to read and comment on different parts of the manuscript. unlike all other results described in this thesis .Tessler. Paris. Graz. I would like to thank my mother Gertrude Petritsch and my uncle Arnfried Meyer for their generous financial support as well as the TMR-Network SELOA. Dr..A.Dr.N. However. Dr.. N.Rozenberg. S.H.M.Thomas. V.G. However. more than 95% of the actual experimental and theoretical work as well as 100% of the writing was performed by the author.Marseglia.R.E. I want to thank all fellow salseros and tangueros who made even my little free time highly enjoyable during these intense three years in Cambridge. I want to note that.Ago. Dittmer. v . Vienna. Bologna.Wan and B.the experimental part of the Section on the laminated polymer device in Chapter 6 was carried out by Dr. Faro. Dr. With the exception of the laminated polymer device.Lux for her patience when explaining organic chemistry to me and pointing out which of my “dream molecules” could be synthesised (and how!). Dr.K¨hler. G.Arias. Dr. materials and scientific input from others. This is also reflected in the long list of people mentioned in this acknowledgment and consistent with the fact that modern research relies on collaboration and teamwork.G¨nther Leising and Univ. Dr. Last but not least. Dr. A.Acknowledgement I want to thank both Univ. the term “we” is used throughout the text to underline the fact that every single result of the author’s work as presented in this thesis was only possible because of the provision of equipment.. F.Granstr¨m.Spearman.P. Richard Friend who together made u it possible for me to spend these precious three years in the opto-electronics group at the Cavendish Laboratory in Cambridge (UK) with all its benefits (traveling to conferences.M.) where I could learn so much not only about science but also life in general. excellent lectures and research facilities etc. For many interesting and exciting discussions of conjugated polymer science I want to thank J.Brown. Harrison.

vi .

Gorick 1994 Sunlight provides the energy for almost all life on Earth. small fish which are preyed on by herons. Algae are eaten by daphnia (water fleas) and these are caught by roach. when they were among the most advanced life forms on Earth. are arranged in clusters and absorb the energy of the photons. UK. From an oil painting by Glynn Gorick. c G. chlorophyll. That energy is used to turn carbon dioxide and water into food. The scale of sizes in the painting gradually changes from vii . Here the molecules of the green pigment.SUNLIGHT HARVESTING the magnified part of a single algal cell. The oxygen in our atmosphere was itself created by ancestral forms of algae. beginning millions of years ago. Photons of sunlight (represented as red dots) enter the food-producing regions within the algal cell which act like solar panels. Cambridge CB1 3BY. 113 Hemingford Road. Even today. making it possible for oxygenbreathing animals to evolve. They changed the ancient atmosphere of the Earth. The heron ultimately depends on a vast number of algal cells to supply it with food. releasing excess oxygen. covering the lower third of the painting. about 90% of all oxygen production and carbon dioxide use is carried out by algae living in the seas and freshwaters. to the Earths atmosphere at the top of the painting. including the microscopic algae living in water. It is harvested by plants.

viii .

. . . . . . .Outline 1. . . . . . . . Summary . . . . . . 2. . . . . . . . The Optical Filter Effect . . . . . . . . . . . . . . . . . . . . . . Organic Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Upper Limit of Voc . . . . . . . . . . . . . . . . . . . . . .4. . . .5 D/A Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 Space Charges & Traps . .2 Renewable Energy Sources . . . The Short Circuit Current Isc . . . . . . . . . 2. . . . . . . . . . . . . . . . . . . . . . . 2. . . . . . ix . . . . . . . . . . . . Photogeneration due to Oxygen Traps .2. . .1 Motivation .1. . . . . . . . .2. .2 General Properties of Organic Semiconductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2. . . . . . . . . . . . . Organic Solar Cells . . . . . . . . . . . . . . . . . . 2. .General 2. . . . . . . . . . . . . . . 1. . . . . . . . . . . . . .4. . . . . . . .2. . . . . . . . . . . . . . . .2 P. . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Photocurrent . . . . . .5 Interpretation of IV Characteristics . . . . . .3 Abbreviations and Synonyms . . . . . . . .4. . . . . . . . . . . . . .3 Electrode Materials . . . . . . . . . . . . . . The Effect of Photoluminescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. . . . . . . . . . . . . . . . Effect of Shorts . . . . . . . . . .6 Electrode/Semiconductor Interfaces . . . . . . . . . . . . . . 2. . . . . . . . . . . 2. . . . . . Fermi-level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4. . . . . . . . . . . . . . . . 2. . . . . .2 Outline . . . . . . . . . . . . 2.4 Photogeneration of Free Charges . . . . . . . . . . . . . . 2. . . . . . . . . . . . . . . . . . . . . . . . . . .2. . . . . . . . . . . . . . . .1 The Combustion of Fossil Fuels 1. . . . .4. . . . The Ideal Solar Cell . . . . . . . . . . . . . . . 1. . . . . . . . . . . . . . . . . Effect of Rsh . . . . . . . . . . . . . . . . . . . . .1. . . . . .2 The Equivalent Circuit Diagram (ECD) . . . . . . . . . . . .Contents 1 Motivation . . . . . . . . . . 1. . . . 2. . . . . . Doping . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Structural Properties of Organic Semiconductors 2. . . . . . . . . . . . 2. . . . . . . . . . . .type Organic Semiconductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inorganic Solar Cells . . . . . . . . . . . . . Inorganic Solar Cells . . . . .1 Power Conversion Efficiency . . . . . . . . . . . . . . . . . .3 Open Circuit Voltage . Current Through Barriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .and n. . . . . . . . . . . . . 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 1 1 2 2 2 3 5 5 8 8 10 10 10 11 12 12 12 14 15 16 16 18 18 19 22 22 22 22 23 23 25 25 25 26 27 27 28 30 2 Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Polarons and Polaron Excitons . .4 Photovoltaic Characterisation . . . . . . . . . . .1 Introduction . . . . . .2. . . . . . . The Donor/Acceptor Interface . . . . . . . . . . . .

. . . . . . .2 HPc+Per . . . . . . . . . . . . . . . . . . . . . . . . . . 96 . . . . . . . . . . . . 3. . . . . . . 31 31 31 33 33 34 36 36 44 52 55 57 60 62 64 67 67 70 73 75 75 78 82 84 86 89 91 4 Double Layer Devices 4. . . . . . . . . . . . . . . . . . . . . . . . . . . 8. . . . . . . . . 3. . . 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. . . . . . . . . .3 Single Layer Devices 3. . . . . .1 Introduction . . . . . . . . . . . 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 Per . . . . .2 The Simulator Setup . . . . . . . . . . . . .2 The Laminated Polymer Device PL Quenching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Doping of Organic Photovoltaic Materials . . . . . . . . . . 97 . . .5 PIF . . . 8 Determination of Solar Cell Efficiencies 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 LPPPT . . . . . . . . . . . . . . . . . . . . . . . . . .3 Simulation of Solar Radiation . . . . . . . . . . . . . . . . .7 PTV+PIF . . . 5. . . . . . . . . . . . . . . . . . . . .5 PTV+Ter . . . PL Quenching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . x . . . . . . . . . . . . . . . . . . . . . . . .5 Electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 MEH . . . . . . . . . . . . . . . . . . . . . . . . . .5 Summary . . . . . . . . . . 102 105 105 107 107 109 112 115 115 115 118 118 119 120 122 7 Discotic Liquid Crystals as Solar Cell Materials 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 The Dye/Pigment Double Layer Device (HPc/Per3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Introduction . . . . . . . . . . . . . . . . .4 Photovoltaic Properties of HPc . . . . . . . . . . . . . . . . . . . 5. .1. . . . . 5 Blend Layer Devices 5. . . . . . . . . . . . . . .1. . 8. . . . . . . . . . . . . .2 Solar Radiation . . . .4 PTV+Per . . . . . . . . . .5d Filter 8. . . . . . . . . . . . . .1. . . . . . .3. . . . 3. . . . . . . . 6. . . . . . . . . . . . . . . . . . . . . . . . . 8. .1 Introduction . . .1 Optical Properties of Organic Photovoltaic Materials 3. . . . . . . . . . . . . . . . . . . . . . . . .1 Introduction . . . . . . . . . . . . . 95 . . . . . . . . . . . . . 3. . . . . . . . . . . . 3. . . . . . . . . . 7. . . . . . . . . . . . . . . . . . .1. . . . . . . 5. 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1. . . . . . . . . . . . . . . . . . . . . . . . .4 Desired Properties of Organic Photovoltaic Materials 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 THPF . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 PTV . . . 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7. . .3. . . . . . . . . . . . . . . . . . .2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. . . . 4. . . . . . . . . . . . . . 5. . 7. . . . . . . . . . . . . . . . . . . . .4 “Numerical” Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. . . . . . 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8 Ter . . . . . . . . . . 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9 Single Layer Devices . . . . . . . . . . . . . . . . . . 95 . . . . . . .1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . .6 PTV+THPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 The Dye/Dye Double Layer Device (CuPc/Per) . . . . . . . . . . .5 Summary . . . . . . .Summary . . . . . . . . . . . . . . . 3. . . . 6 Laminated Layer Devices 6. . . . . . . . . . . . . . . . .3 The Laminated Dye Device . . . . . . . . . . . . . . . . . .3 HPc+PIF . . . . . . . . . . .3 Liquid Crystalline Properties . . . . . . . . . . . . . .1 Calculation of the Required AM1. . . . . . . 7. . . . . . . . . . . 8.2 Performance of Single Layer Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .7 Conferences . .4 Thickness Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Details of the Sample Preparation . . . . . . . . .6 List of Publications . . . 124 10 Appendices 10. . . . . . . . . 137 137 137 137 137 138 138 140 140 140 141 141 142 142 xi . . . . . . . . . . . . . . . . . . . . . . . . . .8 Curriculum Vitae of the Author . . . . . . . . . . . . . . . . . . . . . . . 10. . . .5 Thermal Polarisation Microscopy . . . . . . 10. . . . . . . . . . . . . . . .4 Sublimation of the Top Electrode . . . . . . . . . . . . . . . . . . .3 Film Deposition . . . . . . . . . . . . . . .1. .1 Note on the Dark Current at 0V . . . . . . . . . . . 10. . . .1 ITO Substrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10. . . . .1. . . . . . 10. . . . . . . . . . . . . . . . . . . . . . . . . . 123 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10. . . . . . . . . . . . . .1. . . . . . . . . . . . . . . . . . . . . . 10. . . . . . . . . . . . . . . . . . . . . . . . . .2 Preparation of the Dye and Polymer Solutions 10. . . . . . . . . . . . . . . . . . . . . . . . . 124 9. . . . . . . . . . . . . . . .2 Details of the EQE and IV Measurements . . . . . . . . . . . . 10. . . . . . . . . . . . . . . . . . . . . .2. . . . . .1 Summary . . . . .1. . . . . . . . . . . . . . .3 Absorption Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . 10. . . 10. . . . . . . . . .2 Bibliography .1 Suggestions for Future Investigations . . . . . . .1. 10. . . . . . . . . . . . . . . . .9 Summary and Bibliography 123 9. . . . . . . . . . . . . . . . . . . . . . . . . . .

xii .

. . . IV characteristics of the 117nm LPPPT device. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16 2. . .4 2. . . . . . . . . . . IV characteristics of the 185nm LPPPT device. . . . . . . . .12 3. . .5 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .inorganic cell. . . . . . . . . . . . . . . . .10 2.14 3. . Examples of structures for the different types of organic semiconductors . . . Effect of Rsh on the IV characteristics 107 ≤ Rsh ≤ 1010 Ω . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV characteristics of the 59nm LPPPT device. . . . . . . . . Absorption versus layer thickness (transparent rear contact) . . . . .organic cell. . . . . . . .1 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Band bending of D/A devices with electrode interfaces . . . . EQE of LPPPT films for different thickness when illuminated through AL. Chemical structure of LPPPT. . . . . . . . . Organic cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11 2. . . . . . . . . . .22 2. . . . . . . . . . . . . . . . . . . . .11 3. . . . . . . .18 2. . . . . . . . . . . . . . . IV characteristics of the 93nm LPPPT device. . Absorption versus layer thickness (reflective rear contact) . . . . . . . . . . . . Effect of Rsh on the IV characteristics 103 ≤ Rsh ≤ 106 Ω . . . . . . . The different D/A architectures used in organic solar cells. . . . .12 2. . . . . . . . . .13 2. . . . . . . . . . . . Polarons and singlet exciton . . . . . . . . . . . . . . . . . . . . . . . . . . . IV characteristics of the 67nm LPPPT device. . . . . . . . .4 3. . . . . . . . . . . .7 2. . . . . . . . . . . . . . .8 3. . . . . . . . . . Device structure of a typical organic solar cell. . . . . . . . . . . . . . Optical filter and active layer in device structure. . . . .17 Atmospheric CO2 and human population growth. . . . . . . . . . . . Organic semiconductor/semiconductor interfaces . . . . . . . . Thickness dependence of EQE (at 536nm) for LPPPT devices . . Categories of organic semiconductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17 2. .9 2. . Inorganic cell. . . . . . . . . . . . Transmission of the Al electrode. . . . . . . . IV characteristics related to band energy diagrams. . . . . . . . . Effect of light intensity on the IV curve in an organic cell. . . . . . . .15 3. .14 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16 3. . . . . . . . . . . . . . . . . . . .6 2. . . . . .8 2. . . .organic cell. . . . . . . . . . . . .10 3.9 3. . . . . . . . . . . Effect of Rsh on the IV characteristics 101 ≤ Rsh ≤ 104 Ω . . . . . . . . . . . . . . IV characteristics of the 32nm LPPPT device. . . . . . . . . . . . . .13 3. . EQE of LPPP versus thickness at 536nm . . . . . . . . . . . Extended ECD of a solar cell. . . . . . . . . . . . . . . . . . . . . . EQE and absorption for a thick and a thin LPPPT device. . . . . . . . IV characteristics of the 205nm LPPPT device. . . .15 2. . . 1 5 6 7 8 9 11 12 13 15 18 19 19 21 26 26 27 27 28 28 29 29 30 30 31 32 32 36 37 38 39 39 40 40 41 41 41 42 42 42 42 . . . . . .19 2. . . . . IV characteristics for space charge limited current with deep and shallow traps Device architecture of a single layer solar cell. . . . . . . . . . . . . . . . . Effect of Rs on the IV characteristics. . . Two scenarios for the formation of a thin active layer . . . . . . . . . . . . . . . . . xiii . . . . Solar cell efficiency limits as a function of the semiconductor bandgap.2 2. . . . . . .1 2. . . . . . . . .5 3. . . .List of Figures 1. .23 3. .1 3. . . . . IV characteristics and fillfactor. . . . . . . . Effect of Rsh on the IV characteristics 104 ≤ Rsh ≤ 107 Ω . .21 2. . . . . . . . . . . .3 2. . . . . . . . . . . . . . . . .7 3. .20 2. . . . . . . . . . . . .inorganic cell. . . . . . . . . . . . . .2 3. . . . . . . . . . .6 3. . . Effect of Rs on the IV characteristics. . . IV characteristics of the 122nm LPPPT device. ECD of a solar cell. . . . . . . . . . . .3 3. . . . . . . . . . . . . . . Energy conversion steps and loss mechanism in a solar cell. . . . . . .

. . . . . . . . . . . . . . . . . . . . IV characteristics of the PIF device (double logarithmic plot). . . . . . . . . . . . . . . . Fitted IV curves for MEH-PPV. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 4. . . . . . . . . . . . Chemical structure of PIF. . . . . . . . .51 3.28 3. . . . . . . . . . . . . .38 3. . . . . . . .37 3. . . . . . Chemical structure of MEH-PPV. . . . . . Ratio Rs /Rsh as a function of film thickness in LPPPT . . . . . . . . . .45 3. . . . . . . . . . . . . . . Chemical structure of Ter. . . . . . . . . . . . . . . . . . . .56 3.8 5. . . . . . . . . . . . . . . . . . . . . . . . .59 3. . .architecture and chemical structures . . IV characteristics for varying film thicknesses in MEH-PPV. . . . . . . .30 3. . . . . . . . . . . . . . . . . .3. . . . . . . IV Characteristics of the CuPc/Per double layer device . . . . . . . . . . . . . IV characteristics of the PTV device. . . . . . . . . . . . . . . . .32 3. . . . . . . . . . . . . . . . . IV characteristics of a 10nm thick MEH-PPV cell. . . . . . . . . . . . . . . . . . . .24 3. . .18 3. . Rs and Rsh (dark and light) versus film thickness in LPPPT devices. . . . . . . . . IV characteristics of the Per device. . . . . . EQE and absorption of the Ter-device (semi logarithmic plot). . . . . . . . . . . . Voc and Id in MEH-PPV versus film thickness. . . . . . . . . . . . . EQE and OD of the pristine and heat treated ITO/HPc/Per3/Al device . . . IV characteristics of various film thicknesses in LPPPT. . . . . . . . IV characteristics of a 250nm thick MEH PPV cell. . . . . . . .3 4. . . . . . . EQE and OD of the heated HPc/Per device . . . . . . . . . . . . . . . . . IV characteristics of a 210nm thick MEH PPV cell. . . . . . . . . . . . . EQE and absorption spectrum of the THPF device. . . IV characteristics of the Ter device. . . . .40 3. . . . . . . . .4 4. . . . . . . . . . . . . . . . . . . . . . . . . IV characteristics of a 35nm thick MEH-PPV cell.33 3. . . . . . . . . . . . . . . . . . . . . .43 3. . . . .5 4. . . . . . . . . .42 3.21 3. . . . . . . . . . . . .6 4. . . . . . . . Rsh and Rs versus film thickness in MEH-PPV. . . .1 4. . . . . . . . . . . . . EQE and absorption spectrum of the THPF device(semi logarithmic plot). . . . . . . EQE and absorption spectrum of a PIF device. . . . . .27 3. . . . . . . . . HPc/Per3 double layer . . . . . . . . . . . . . . . . EQE and absorption for the Per device. . . . .26 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . EQE of MEH-PPV versus film thicknesses. . . . . . . . Chemical structure of Per. . . . . . .54 3. . . . . . . . . . . . . .39 3. . . . . . . .52 3. .53 3. . . . . . . . . Chemical structure of CuPc and Per . . . . . . . . . . . . Rsh and Rs versus film thickness in MEH-PPV. . . . . . . . . . . . IV characteristics of the THPF cell. . . . . . . . . . .23 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19 3. . . . . . . . EQE and absorption spectrum of a 90nm thick PTV device. . . . . IV-characteristics of the pristine and heated HPc/Per double layer device. . . . . . IV characteristics of a 38nm thick MEH-PPV cell. 43 43 44 44 45 45 46 46 47 47 47 47 48 48 48 48 49 49 50 50 50 51 52 53 53 54 54 55 56 56 57 57 58 58 59 59 60 61 61 62 63 63 64 67 68 70 72 72 73 73 74 75 Solar cell architecture comprising a blend of two organic semiconductors . . . . . . . . . . . . . . .29 3. .46 3. EQE spectra for MEH-PPV thicknesses between 10 and 250nm . . . . . . . . . . . . .60 4. . . . . . . . .49 3. . . . . . . . . . . . . . . . . . . . . . . EQE and absorption for a thick and a thin MEH-PPV device. . . . . . . . . . . .36 3. . . . . . . . . .55 3. . . . . . . . . . . . . . . . . .57 3. . . . . EQE and absorption of PIF device (semi logarithmic plot) . . . . . Absorption and EQE of the CuPc/Per double layer device . . . . . . . . . . . . . . . . . EQE of a 90nm thick PTV device (semi logarithmic plot) . . . . . . . . . . . . . . .34 3. . . . . . . . . . . . . . . . . . . . .20 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Device architecture of a double layer solar cell. . Rs and Rsh (dark) versus film thickness in LPPPT devices. . . . . . Fitted IV curves for LPPPT. . . . . . . . .25 3. . . IV characteristics of the PIF device. . . . .41 3. . . xiv .35 3. . . . . . . . . . . . . . . . . . . . . . .22 3. . . . . . . . . . .47 3.58 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . EQE and absorption of the perylene device (semi logarithmic plot). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rsh and Rs versus film thickness in MEH-PPV. . . . . . .48 3. . .50 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 4. . . . IV characteristics of a 54nm thick MEH-PPV cell. . . . . .1 Voc and dark current versus film thickness in LPPPT devices. . . . . . . . . . Chemical structure of PTV. . . . . . . . . . . . IV characteristics of the PTV cell (double logarithmic). . . . . . . . . .31 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemical structure of THPF. . . . . . . .44 3. . . . . . . . . . . . . . . . . . . . . . .

5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13 5.14 5.15 5.16 5.17 5.18 5.19 5.20 5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 6.10 6.11 6.12 6.13 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9 7.10 7.11

Chemical structure of HPc and Per . . . . . . . . . . . . . . . . . . . . . . . . . . PL efficiency versus blend ratio (HPc+Per) . . . . . . . . . . . . . . . . . . . . . AFM phase contrast images of a HPc+Per blend . . . . . . . . . . . . . . . . . . Absorption and EQE of the HPc+Per blend device . . . . . . . . . . . . . . . . . EQE of the HPc+Per blend device and the single components . . . . . . . . . . . IV characteristics of the HPc+Per blend device . . . . . . . . . . . . . . . . . . . Chemical structure of HPc and PIF . . . . . . . . . . . . . . . . . . . . . . . . . Absorption and EQE of the HPc+PIF blend device . . . . . . . . . . . . . . . . . Absorption and EQE of the HPc+PIF blend device . . . . . . . . . . . . . . . . . IV characteristics of the HPC+PIF blend device . . . . . . . . . . . . . . . . . . Chemical structure of Per and PTV . . . . . . . . . . . . . . . . . . . . . . . . . EQE and absorption of the PTV+Per blend device . . . . . . . . . . . . . . . . . EQE of the PTV/PER blend device (semi logarithmic plot) . . . . . . . . . . . . IV characteristics of the PTV+Per blend device . . . . . . . . . . . . . . . . . . . Chemical structure of Ter and PTV . . . . . . . . . . . . . . . . . . . . . . . . . EQE and absorption of the PTV+Ter blend device . . . . . . . . . . . . . . . . . Absorption versus energy of the PTV+Ter device (vibronic modes) . . . . . . . . EQE of the PTV+Ter blend device (semi logarithmic plot) . . . . . . . . . . . . IV characteristics of the PTV+Ter blend device . . . . . . . . . . . . . . . . . . . Chemical structure of PTV and THPF . . . . . . . . . . . . . . . . . . . . . . . . Absorption and EQE of the PTV+THPF blend device . . . . . . . . . . . . . . . EQE of the PTV+THPF blend device (semi logarithmic plot) . . . . . . . . . . . IV characteristics of the PTV+THPF blend device . . . . . . . . . . . . . . . . . Chemical structure of PTV and PIF . . . . . . . . . . . . . . . . . . . . . . . . . Absorption and EQE of the PTV+PIF blend device . . . . . . . . . . . . . . . . EQE of the PTV+PIF blend device (semi logarithmic plot) . . . . . . . . . . . . IV characteristics of the PTV+PIF blend device . . . . . . . . . . . . . . . . . . IV characteristic (dark) of the PTV+PIF blend device (double logarithmic plot).

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

79 79 80 80 81 81 82 83 83 84 85 85 85 86 87 87 87 88 89 89 90 90 91 92 92 92 93 93 95 97 98 98 99 99 99 100 100 101 102 103 103

The lamination technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . PL-quenching of MCP+POPT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Absorption spectra of MCP and POPT . . . . . . . . . . . . . . . . . . . . . . . . . EQE and absorption of the laminated polymer device . . . . . . . . . . . . . . . . . AFM image of cross-section of the laminated polymer device. . . . . . . . . . . . . . AFM image of a MCP+5%POPT film. . . . . . . . . . . . . . . . . . . . . . . . . . . AFM image of a POPT+5%MCP film. . . . . . . . . . . . . . . . . . . . . . . . . . . IV characteristics of the laminated ITO/POPT/MCP+5%/Al device . . . . . . . . . IV characteristics of the laminated Au/PEDOT/POPT/MCP+5%POPT/Ca device Voc and Isc versus light intensity of the laminated PEDOT/Ca polymer device . . . PL-quenching HPc+Per . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . EQE and absorption spectrum of the laminated dye device . . . . . . . . . . . . . . EQE and absorption spectrum of the laminated dye device . . . . . . . . . . . . . .

Schematic illustration of the solid, liquid crystalline and liquid phase of discotic molecules. 105 Illustration of the different discotic phases. . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Chemical structure of some liquid crystalline phthalocyanines. . . . . . . . . . . . . . . . . 106 DSC trace of HPc. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Image of digitate star texture and birefringent defects. . . . . . . . . . . . . . . . . . . . . 108 Sketch of a bend mode as origin for birefringent defects in the mesophase of discotic molecules108 Image of a focal conic texture under the polarisation microscope. . . . . . . . . . . . . . . 109 TGA of a series of liquid crystalline phthalocyanines. . . . . . . . . . . . . . . . . . . . . . 109 TGA of CuPcSC12 under nitrogen and air. . . . . . . . . . . . . . . . . . . . . . . . . . . 110 Absorption and photoluminescence of HPc. . . . . . . . . . . . . . . . . . . . . . . . . . . 110 EQE of the thin HPc device, illumination through ITO. . . . . . . . . . . . . . . . . . . . 110 xv

7.12 7.13 7.14 7.15 7.16 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.11 8.8 8.9 8.10 8.12 10.1 10.2 10.3 10.4

EQE of the thick HPc cell, illuminated through AL. . . . . EQE of thin HPc device, illumination through Al. . . . . . EQE of the thick HPc device in vacuum and air illuminated IV characteristics of the thin HPc device. . . . . . . . . . . IV characteristics of the thick HPc device. . . . . . . . . . .

. . . . . . . . . . . . . . . . through ITO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

111 111 112 113 113 115 116 116 118 118 119 119 120 120 120 121 121 138 138 139 139

Definition of Air Mass. . . . . . . . . . . . . . . . . . . . . . . . . . . Global and direct solar spectrum . . . . . . . . . . . . . . . . . . . . Energy distribution in AM1.5d spectrum. . . . . . . . . . . . . . . . Emission of a QTH-lamp and filter transmission spectra. . . . . . . . Measured and modeled AM1.5d spectra . . . . . . . . . . . . . . . . Transmission of the required, modeled and measured AM1.5d filter. Measured, modeled and standard AM1.5d spectrum . . . . . . . . . Solar simulator setup. . . . . . . . . . . . . . . . . . . . . . . . . . . Deviation of actual filter absorption from the catalogue. . . . . . . . AM1.5d spectrum and filter calculated from measured filter curves . Calculated spectrum for new filter combination and AM1.5d . . . . . Intensity dependence of Voc and Isc . . . . . . . . . . . . . . . . . . .

Electrode pattern of the substrate with 8 pixels. . . . . . . . . . . . . Electrode pattern of the substrate with 8 pixels. . . . . . . . . . . . . Typical light intensities versus wavelength in EQE/IV measurements. EQE spectrum and spectral response of the silicon reference diode. . .

xvi

List of Tables
1.1 1.2 2.1 2.2 3.1 3.2 3.3 3.4 3.5 4.1 5.1 6.1 7.1 7.2 9.1 9.2 9.3 List of synonyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Workfunction of the elements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Workfunction of molecular electrode materials . . . . . . . . . . . . . . . . . . . . . . . . . Survey Survey Survey Survey Survey of of of of of the absorption properties of various organic semiconductors . . . . . . . . . important performance parameters of various organic single layer solar cells solar cell parameters obtained from the IV curves . . . . . . . . . . . . . . . various solar cell parameters obtained from the IV characteristics . . . . . . photovoltaic parameters of the investigated organic semiconductors . . . . . . . . . . . . . . . 4 4 17 18 32 34 40 49 65 69

Survey of performance parameters of various organic double layer solar cells . . . . . . . .

Survey of performance parameters of various solar cells using organic semiconductor blends 77 Survey of performance parameters of laminated solar cells. . . . . . . . . . . . . . . . . . 96

Phase transition temperatures of a series of liquid crystalline phthalocyanine derivatives . 107 Survey of electron and hole mobilities of important organic and inorganic semiconductors 112 Overall assessment of the different solar cell architectures incl. pigments . . . . . . . . . . 123 Overall assessment of the different solar cell architectures for soluble materials only . . . . 123 Summary of important performance parameters of various organic solar cells including all types of architectures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

xvii

we have renewable energy sources which neither run out nor have any significant harmful effects on our environment. Thus. As a result the CO2 conwith renewable energy sources.1. the amount of power they can supply over a longer period3 often depends on geographical and weather conditions.1 Motivation 1. the inevitable permanent de.1 The Combustion of Fossil Fuels of nature on our planet.sooner or later Today’s plants are unable to absorb this huge to replace most of the currently used power plants amount of extra CO2 .1: The atmospheric concentration of greenhouse gases like CO2 have grown significantly since pre-industrial times. 136. power plants that use wind. about 20 · 1012 kg of carbon dioxide are put into the atmosphere every The limited supply of todays main energy sources year. Hence.0o C by the year 2100 [136]. 2 Projected 1 . the heat from the Earth’s crust or direct solar radiation rely on the local supply of their primary energy source.1. It is important to recognise that the installation of these power systems should always be preferred as long as they supply more energy throughout their sea level rise from the present to the year 2100 is 15-95cm [136]. 3 Using the existing energy storage technology. but also others like for example. by 0.1) adding considerably to the greenhouse efcline in the global oil production rate is expected to fect which will increase the global mean surface start within the next 10-20 years. references are listed in alphabetical order in Chapter 9. temperature . mostly fossil-fuel use [136].and wind-power systems. This can be largely attributed to human activities.1. a service to the reader. The consequences of this temperature change have already increased the frequency and severity of natural disasters [252] and are likely to have more devastating effects for humans and other life forms in all parts of Earth within the next decades. oil prices will then rise considerably ios and the actual climate sensitivity .3-0.(Fig. According to recent centration in the atmosphere continues to mount predictions [123. (oil.Outline However. Naturally. Today. 1. mainly by burning fossil fuel [17.6-7.0. coal.depending on future emission scenarWorldwide. most likely due to human activities [136].2 Renewable Energy Sources Figure 1. the combustion of fossil fuels in the past has already harmful effects on the delicate balance 1. Dashed lines are possible (optimistic) future scenarios [136. 252]. they do not appear in numerical order in the text. uranium) will force us . wave or tidal energy of water. the potential.6o C since the late 19th century and the hydroelectric. 1 As Fortunately. global sea level2 has risen by 10-25cm. 10]. 36]1 . ergy sources such as the direct conversion of solar Global mean surface temperature has increased energy (solar cells).by another favouring the introduction of various renewable en.Chapter 1 Motivation .

debit. 274. flexible substrates usrication. this cause of simpler processing at much lower temper.very specialised but also interdisciplinary field. illustrates that technologies which allow low fabrication costs . These can be processed from solution at Each of them begins with a survey of characteristic parameters of already reported devices . One approach Outline here would be to reduce the amount of silicon by 1.2 using thinner films on (cheap) glass substrates.mary of “organic solar cell relevant knowledge” that is consistent with the experience. which Inorganic Solar Cells generates problems with sealing against air.Chapter contains a unique compilation and sumatures (20-200 o C) than the above mentioned inor. blend (Chapter 5) and processing properties of polymeric i.and multi-crystalline sili. The advantage of polymeric photovoltaic cells when compared to electro-chemical cells is predominantly the absence of a liquid electrolyte. which The subsequent Chapters deal with the four combine the opto-electronic properties of conven. Today. references and are only summarized briefly in the first part. phone or other) cards which can display for vest up to as much as 24% [237] of the incoming example. double layer (Chapter 4). understanding ganic cells. “costs” of environmental hazards of the majority optically-pumped LASERs [249.5) have been confirmed ing the results of this thesis . Another interesting alternative to inorganic cells terpretation of current voltage characteristics. installation and maintenance. to photo-detectors solar energy which is already close to the theoret.e.in large area scanners or medical imaging and solar ically predicted upper limit of 30% [230] 4 . but At present. are not yet low enough.ing large area devices and the use of flexible subcon have found markets for small scale devices such strates. MOTIVATION . The ditions with numerous lithographic steps leading to more general properties of organic semiconductors can be found in many excellent text books and other relatively high manufacturing costs [162]. 250] and photovolof existing power sources in the planning of future taic diodes [112. 94]. 100]. the remaining amount.includ4 Note that using the more expensive GaAs even higher power conversion efficiencies (AM1. field effect transistors [232]. Organic Solar Cells The second part gives a more comprehensive insight into the important characteristic solar cell paConsiderably less effort and production energy is necessary if organic semiconductors are used be. solar cells comprising an inorganic semi. power plants.by drawing analogies to inorganic cells using the equivalent circuit diagram and the “traditional” intively low production costs. These conventional solar cells can har. pocket calculators and wa.rather than somewhat higher conversion efficiencies . Possible applications may range from small as solar panels on roofs. . and view of the author.OUTLINE lifetime than they have consumed during their fab. In fact.room-temperature onto e.2 CHAPTER 1. ter pumps. ”plastic” and the laminated device (Chapter 6).are now desired. For example. electro-chemical solar cells using tiHowever. This power applications on uneven surfaces. Since the discovery of electro-luminescence in Unless we can soon develop low cost technolo. this class of materigies for renewable energy sources we have to hope als has been used to build efficient light emitting that the world’s governments start to consider the diodes [248. materials.g.also the prospect of even cheaper production usconductor such as mono. the production of these solar cells still Chapter 2 is aimed to introduce researchers who requires many energy intensive processes at high are planning to work on organic solar cells into this temperatures (400-1400o C) and high vacuum con.pointing out specific [237]: 25% (monocrystalline GaAs)and 30% (GaInP/GaAs) advantages and encountered problems. is given by the semiconducting polymers.disposable solar cells to power smart plastic (credit. Although ing simple and therefore cheaper deposition meththis criterion is fulfilled for all of them their prices ods like spin or blade coating. since a full theoretical understanding tanium dioxide in conjunction with an organic dye and a liquid electrolyte [182] already exceeded 6% of organic solar cells is still not possible we have power conversion efficiencies [237] and are about to tried to improve understanding of device physics enter the commercial market thanks to their rela.conjugated polymers [34].known device architectures: the single layer (Chaptional semiconductors with the excellent mechanical ter 3). 278.rameters and links between them. 213.

see Section 5.3.2 and Section 3. • We have shown for the first time that dye/dye interfaces can be used for photogeneration of charges in solar cells .see Section 3. . Two devices even had a clear photo-response down to λ >1000nm . Chapter 9 concludes with a summary of the characteristic parameters comprising all four solar cell architecture. Section 5.2. The outstanding properties of discotic liquid crystals justify the discussion of this device in a Chapter separated from the single layer device Chapter.3 as well as publications Refs [200. However. [94.3 Abbreviations and Synonyms Table 1. A method to estimate the optimal thickness has been introduced . the following findings may be of particular interest for both experts and newcomers in the field: • We have introduced a new device architecture that combines advantages of double layer and blend devices and opens exciting new possibilities in device design such as selective doping. Note that the terms ”dye” and ”pigment” are frequently used as synonyms in literature . • All devices discussed in the Chapters on double layer.wavelength detector covering the entire visible range . However.1. 95.see Section 4. 201]. • The single layer device comprising PTV shows a very strong monoton dependency of the EQE on the wavelength so that the device can be used as simple colour . for example. blends and laminated structures show spectral responses covering at least the 3 wavelength range of visible light. can have two very different meanings: 1. It also comprises a survey of the interesting transport (charge carrier mobility) properties of liquid crystalline semiconductors as well as the mesogenic characterization of a series of discotic molecules from which one has been used to fabricate a single layer solar cell. 1.7 and Section 5. Incident photon to current conversion efficiency.2 and Section 6. • In Chapter 8 we describe how the standard solar spectrum can be simulated with a relatively simple setup that can be built in most laboratories for a fraction of the cost of commercial simulators. 2.3. Thus its value can be considerably larger than the external power conversion efficiency. This is the power conversion efficiency when losses due to reflection and transmission are considered.3.see Section 3.or even . some suggestions for future investigations and a comprehensive bibliography. How solar cell efficiencies can be determined in a reasonable yet practical way either by setting up a solar simulator or numerical simulation is discussed and described in Chapter 8.2. Note that.see Section 3. • In Chapter 7 we have shown that dyes with liquid crystalline properties can be used as active semiconducting components in solar cells. Two laminated devices are discussed in Chapter 6 and published in Refs. Our results together with the recent literature indicate that heating into the liquid crystalline phase is not necessary.3. ABBREVIATIONS AND SYNONYMS Chapter 7 concerns single layer devices comprising a liquid crystalline semiconductor. This is often the equivalent to EQE but could also be the IQE. an overall assessment.see also Section 2. 201]. which is near the estimated exciton diffusion length) give the highest currents .5 efficiency of solar cells.1 lists terms which are frequently used as synonyms in literature and throughout the text. Details of sample preparation and measurements as well as a list of publications by the author and a brief CV can be found in the Appendices in Chapter 10.4. the term IPCE which is sometimes used in publications on organic solar cells. Internal power conversion efficiency. we have also found that the shunt resistor grows 30 to 100 times faster than the series resistor with increasing film thickness favouring thicker films for larger fill factors. The object of this thesis was the investigation of various types of organic semiconductors (preferably with low bandgaps) in different solar cell architectures. We also discuss a method solely based on intensity dependent photocurrent/voltage measurements that can be used to estimate the AM1.while we differentiate clearly between these two in this thesis .3. • We have studied effects of film thickness and have found that those devices that have the thinnest films (10-30nm.

2. Many terms used in organic chemistry or to describe theoretical objects or concepts can grow very long and become unreadable.OUTLINE Table 1.5 efficiencies to allow meaningful comparisons of solar cell performances all over the world.1: List of expressions which are used as synonyms throughout the text or relevant literature. MOTIVATION . the big scientific community that has been researching on inorganic solar cells for many decades usually quotes the EQE and if required the IQE together with AM1.or dye) irradiance n-type laminated device material photocurrent p-type material shunt Shottky diode spectral response solar (power) conversion efficiency blend layer device semiconducting polymer photon to currentconversion efficiency exciton dissociation donor/acceptor interface bandgap HPcSC12 (dye) light intensity electron transport laminated layer device or acceptor material light current hole transport or electron donor material parallel resistor single layer device photo-response Table 1. These are listed in Table 1. both interpretations of IPCE could refer to monochromatic light (more common) but also to white light exposure or some other light spectrum.4 CHAPTER 1. . AM1. Hence we have had to use abbreviations most of which are common in the field. A CB CN-PPV CV D DSC EA ECD EQE FF FTO HOMO HPc IP ITO LED LPPP LPPPT LUMO MEH MCP MWNT OD PANI Pc PEDOT Per PIF PITN PL PR-TRMC PPP PPV PT PTV PVK P3HT SCE SMU SR TCNQ TGA TNF TO TOF VB electron acceptor conduction band cyano substituted PPV cyclic voltametry electron donor differential scanning calorimetry electron affinity equivalent circuit diagram external quantum efficiency fillfactor fluorinated tin oxide highest occupied molecular orbital metal free phthalocyanine ionisation potential indium tin oxide light emitting device poly(para-phenylene) type ladder polymer poly(phenylene-tienylene) type ladder polymer lowest unoccupied molecular orbital 2-methoxy.2: List of common abbreviations used throughout the text. In order to avoid these irritating ambiguities.5 efficiency blend device conjugated polymer external quantumefficiency exciton splitting heterojunction HOMO/LUMO gap HPc (pig. 5-(2-ethyl-hexyloxy) CN substituted MEH-PPV multiwall nanotubes optical density poly(aniline) phthalocyanine poly(ethylene dioxythiophene) perylene diimide derivative poly(indenofluorene) poly(iso-thionaphthene) photoluminescence pulse radiolysis time resolved microwave conductivity poly(para-phenylene) poly(para-phenylene vinylene) poly(thiophene) poly(tienylene vinylene) poly(vinyl carbazole) poly (3-hexylthiophene) saturated calomel electrode source measure unit spectral response tetra-cyano-quino-di-methane thermal gravimetry analysis trinitrofluorenone tin oxide time of flight valence band Moreover.

the highest possible continuous current of electrons flowing through the external circuit is then 100 electrons/s.2. The assumption of the absence of nonradiative recombination allows predictions for the upper limit for both the power conversion efficiency of a semiconductor with a given bandgap and the open circuit voltage. may then utilise the converted solar energy. much more time to reach the device electrodes before they recombine under the emission of light.g. These excitons carrying energy but no net charge may diffuse to dissociation sites where their charges can be separated. These issues are discussed in more detail in Section 2. Only if there are virtually no such defects.2 shows the conversion steps of photons into separated charges as it takes place in an organic solar cell. these leakage currents are simply realised by the recombination of the photoexcited charge carriers or using our analogy: the ”pumped” electrons drop back into their VB.2. In organic semiconductors. If the current flow through the external circuit is reduced by e. the remaining 20 electrons/s will drop back into the VB before they can leave the device. The organic film may comprise one [89] or more [245] semiconducting layers. a load resistor to .80 electrons/s.efficiency can be regarded as evidence for the absence of the faster and therefore more efficient non-radiative recombination channels. a blend [115] or a combination of these [95]. When connected to the external electric circuit any electrical device. such as a computer screen or the motor of a water pump. will remain as the only decay channel since it does not require the existence of mid-gap levels but can occur directly from band to band. Figure courtesy of Ref. The maximum current is determined by the pump rate. The separated charges then need to travel to the respective device electrodes. . As a consequence. Suppose the ”pump” can promote 100 electrons/s from the valence band (VB) into the conduction band (CB). a solar cell (Fig. In real semiconductors.1: Device structure of a typical organic solar cell.3. Fig.4.Chapter 2 Solar Cells .General 2.1 Introduction The conversion of solar light into electric power requires the generation of both negative and positive charges as well as a driving force that can push these charges through an external electric circuit. holes to the anode and electrodes to the cathode to provide voltage and be available for injection into an external circuit. It also shows the associated loss mechanism and the related electrical quantity used in the equivalent circuit diagram (ECD). [63]. which occurs on a much larger time scale.4.for example . absorption of photons leads to the creation of bound electron hole pairs (excitons) rather than free charges. high PL . The charge carriers will then have 5 Figure 2. radiative recombination.1) may be seen as a solar-light driven electron pump: The maximum height the electrons can be ”pumped” is equal to the highest voltage the solar cell can develop. The latter is described in more detail in Section 2.2. These leakage currents are typically caused by defects or other deviations from the ideal semiconductor structure which give rise to the appearance of allowed energy levels within the bandgap. In fact.

Semiconductor/metal interfaces are discussed in more detail in Section 2. Charge transport. SOLAR CELLS . A bandgap of 1. Al. Ca) the charges often have to overcome the potential barrier of a thin oxide layer.1eV (1100nm) is required to absorb 77% of the solar radiation on earth (see Table 8. Exciton diffusion. their diffusion length should be at least equal the required layer thickness (for sufficient absorption) . Since such a site may be at the other end of the semiconductor. See also Fig. Reflection losses are probably significant but little investigated in these 3. 113. In the latter one material can than act as electron acceptor (A) while the other keeps the positive charge and is referred to as electron donor (D) . 52.2. The transport of charges is affected by recombination during the journey to the electrodes . The typically low charge carrier and exciton mobilities require layer thickness in the order of 100nm. interaction with atoms or other charges may slow down the travel speed and thereby limit the current .2. oxygen) or between materials with sufficiently different electron affinities (EA) and ionisation potentials (IA).0eV (600nm) limiting the possible absorption to about 30%.since it did actually donate the electron to A.g.4. Absorption of photons. Charge collection.g.4. Fortunately the absorption coefficient of organic materials is generally much higher than in e. Anti-reflection coatings as used in inorganic devices may then prove useful once other losses such as recombination become less dominant. Figure 2. If the difference in IA and EA is not sufficient. all photoexcited excitons should reach a dissociation site. Also.6 CHAPTER 2. impurities (e. The symbols in brackets represent the quantities that allow for the specific loss mechanism in the ECD see Section 2. In the following we comment on the individual conversion steps with regard to the special situation in organic solar cells: 1. 2.3 whereas the majority of semiconducting polymers have bandgaps higher than 2. Systematic measurements of photovoltaic materials are desired to provide knowledge of their impact on absorption losses. • Reflection. Exciton diffusion ranges in polymers and pigments are typically around 10nm [111. Charge separation is known to occur at organic semiconductor/metal interfaces. Eventually it will recombine without contributing charges to the photocurrent. • The organic layer is too thin.2.2. 4. the metal may have formed a blocking contact with the semiconductor so that they can not immediately reach the metal. some pigments like perylenes are believed to have exciton diffusion lengths of several 100nm [154]. 5. Si so that only about 100nm are necessary to absorb between 60 and 90% if a reflective back contact is used.particularly if the same material serves as transport medium for both electrons and holes.2).g.6.otherwise they recombine and photons were wasted. For effects of doping see Section 2.GENERAL materials. . In addition. the exciton may just hop onto the material with the lower bandgap without splitting up its charges. Ideally.2: Survey of the specific conversion steps and loss mechanisms in an organic solar cell. 88]. In order to enter an electrode material with a relatively low workfunction (e. However. In most organic device only a small portion of the incident light is absorbed for the following reasons: • The semiconductor bandgap is too high.8.2. Charge separation.see also space charge limited current in Section 2.

This can usually only be partly achieved so the defects of the network structure .2 we discuss the special general properties associated with organic semiconducting ma- . lows only excitons of a thin layer to reach it The photoactive region is often very thin and and get dissociated. The structure is tle chance to recombine with its counterpart. physical/chemical conversion) before forming the blend layer and instant encapsulation between the two substrates. Fig. an electron donor (D) and an electron acceptor material (A). This structure benefits from Schottky diodes since charge separation occurs the separated charge transport layers that enat the rectifying (Schottky) junction with one sure connectivity with the correct electrode electrode.6 for more information on electrode materials.2. Thus. This relatively recent type combination losses are generally high. The reason why many aggregated materials can still be used to give excellent devices is not yet clear.1. Blend cell. Charge separation occurs at the interface between those two. 1 PL is often quenched considerably if close packing occurs indicating the presence of faster recombination (loss) channels.2.material should only be in contact with the electrode material with the higher workfunction (typically ITO) and the A-material with the lower workfunction electrode (typically Al)2 . INTRODUCTION 7 We note that both exciton and charge transport in organic materials usually require hopping from molecule to molecule. Charge separation occurs in the blend layer in the middle that is obtained after laminating the two separate layers together and charge transport can only occur via the correct transport layer. The blend structure ing the strengths and weak points of the four dif. the advantages of the two structures above. In Section 2.particularly the connectivity with the correct electrode . We note that dense packing also favours a higher absorption coefficient. Laminated device. The strong point of this type is the large interface area if the molecular mixing occurs on a scale that allows good contact between alike molecules (charge percolation) and most excitons to reach the D/A interface.3: Three different types of D/A architecIn the following we give a brief summary includ. In order to meet these specific requirement for efficient photon to charge conversion different device architectures have been developed in the past. The other electrode interface is supand give the separated charge carriers only litposed to be of ohmic nature. close packing of the molecules is assumed to decrease the width of the intermolecular barriers and a flat molecular structure should generally lead to better transport properties than bulky 3 dimensional molecules1 . The drawback is that certain mechanical properties of the organic semiconductors are required (low glass transition temperature) to form the intermixed layer.can be obtained by spincoating using a D/A mixferent architectures known today: ture whereas the double layer requires two separate deposition processes. The gain of improved charge transport may compensate the extra recombination channel here. This structure also features the useful options to treat each layer separately (e. Double layer cell. 2 See Section 2.tures for organic solar cells. simple but absorption covering the entire visiThe drawback is the small interface that alble range is rare using a single type of molecule.g. sist of only one semiconductor material and are often referred to as Schottky type devices or 3. Figure 2. Single Layer Cell.represents a technological challenge. Ideally the D.seen as a hybrid between these two.3 shows device designs used for cells comprising two components. doping. The laminated device can be 1. Single layer structures con. since both positive and negative photoexcited charges travel through the same material re4.2. represents the successful attempt to unify 2.

52]) compared to the film thickness necessary to absorb the major portion of light (typically >100nm) has made it difficult to reach practically interesting conversion efficiencies in organic solar cells. SOLAR CELLS . The strong electron-lattice coupling is responsible for the existence of the quasi-particles such as polarons and polaron excitons in conjugated systems. Moreover.4: Simple model that relates the IV characteristics qualitatively to energy level diagrams. This usually leads to a more rigid structure in the excited state the so called quinoid configuration. These quasi particles can be identified by the additional energy levels [70] associated with them which appear within the semiconductor bandgap5 see Fig. 197.8 CHAPTER 2. conjugated systems exhibit neither pure covalent nor pure polar bonds.using sensitive detection methods like photo-induced absorption spectroscopy [163. Voc . 2. Due to the dimerisation3 of the bonds.GENERAL ing interface (Schottky contact) in single layer devices or at the interface between an electron-donor and an electron-acceptor semiconductor material. the small diffusion range of the excitons (typically about 10nm [111. In fact. 2. the electron repulses adjacent electrons while attracting the nuclei resulting in a polarisation of the lattice in its closer vicinity. These excitons need to be split up (or dissociate) before the charges can be transported through the film and collected at the electrodes.2.2. terials whereas in Section 2. Tilted bands represent homogeneous electric fields that drive the charge carriers to the respective electrodes. 168]. they represent a separate class with regard to their bond nature exhibiting huge electronlattice coupling4 compared to the inorganic solids.4 we focus on the photovoltaic properties. For example. exciton dissociation can occur at a rectify- The origin of the term polaron for charged quasiparticles can be found in classical polar crystals where the associated charge e. this accompanying cloud of polarisation causes an increasing effective mass decreasing the mobility of these quasi-particles. Polarons in conjugated systems affect not merely the polarisation in their vicinity .g.2 General Properties of Organic Semiconductors Due to e. In addition.4eV [12. the strong electron-phonon interaction in organic materials the photo-physics in this class of materials is often different compared to the inorganic semiconductors and not yet fully understood. Note that photogeneration of charged excitations usually requires separation of charges between 3 The alteration of single and double bonds does not allow determination of which bond is double and which is single bond in degenerate semiconducting systems such as most semiconductors. EQE). 88. Thus. the concept of an equivalent circuit diagram (ECD) or the deduction of resistor parameters from the IV characteristics are the same.5 . 4 Excitation of the π electron used for the double bond has a strong effect on the ”lattice” structure. 113. Since solar energy conversion using solar cells is in principle the same for both organic and inorganic semiconductors ”general tools” like the definition of solar cell parameters (FF. .they even change the nature of bonds from σ to π and vice versa via excitation and while traveling.g. the π electron of the double bond can be seen as delocalised. One of the main differences is that photoexcitation in these materials does not automatically lead to the generation of free charge carriers. 198]. but to bound electron-hole pairs (excitons) with a binding energy of about 0.1 Polarons and Polaron Excitons Figure 2. For clarity it is assumed that bands are tilted rather than bent and exciton dissociation occurs throughout the entire device. The larger this interface area the more excitons can reach it and dissociate. 56. 5 upon photoexcitation or chemical doping (upon blending or intercalation with guest material (dopant)).

groups of the three main types. However. 5.95eV) [49] have been proposed. . Their energy levels can be described as hydrogen-like bound states. At high densities and low temperatures. Eb values or perylenes [205. Figure 2. 54]. an inorganic exciton is a Boson rather than a Fermion. the most effective generation is photoexcitation with a pump beam that is polarised perpendicular to the polymer backbone [76]. 79. GENERAL PROPERTIES OF ORGANIC SEMICONDUCTORS 9 Within the framework of this thesis we assume that binding energy is not larger than about 0. Hence. Frenkel exciton: The electron hole pair is confined to not more than one molecular unit [75. 3. As a consequence.4eV.and intrachain excitons may also be seen as sub ate values (around 0. the energy levels of the exciton are separated from the CB and VB by less than about 0. The binding en. with occupation statistics determined by the Bose-Einstein equation.consisting of two different molecules. In the following we give a brief description of different types of excitons6 : 1. 138]. a clear offset between the HOMO and LUMO levels of the D/A material is still required for the exciton to dissociate at room temperature. Since they are Bosons. 168.emission spectrum. 104]. It can be formed by interaction ergy Eb of excitons in organic semiconductors . Mott-Wannier exciton: The distance between electron and hole is much greater than the spacing between unit cell (here molecular units) [175. We note that there is another related excited species called “excimer” that can play an important role in organic semiconductors. the chains.4eV) [56. Charge transfer exciton: The exciton extends over only a few adjacent molecular units [138] 4. 12] to very high (up to 0. An exciplex is an excimer ranging from very small [158] over intermedi. Unlike its constituents particles.has been subject to in. Arrows with solid lines represent electrons with their spin orientation. 195]. calculated with a reduced effective mass.2.observed in organic molecules such as naphthalenes tense debates over the past decade.2. Both singlet and triplet excimers have been and its derivatives . It can be regarded as a charge transfer exciton. 6 Inter. The binding energy Eb of inorganic excitons is estimated to be around 16meV which means that these exciton become important predominantly at lower temperatures (where kT becomes small compared to Eb )[98].2eV which have been neglected in all band energy diagrams in this thesis since the bandgaps of the materials used are between 1 and 2eV. In the following we give a brief overview about differences between excitons in the much better understood inorganic systems and the organic ones: • Excitons in inorganic solar cells. 2. These excitons can reduce their energy further if they bind with impurities or defects. arrows with dashed lines show possible intra gap transitions that can be used to detect and identify the excited species. An excimer can be understood as a dimer that is only stable in the excited state [205. the free exciton gas can condense to give an electron-hole liquid phase with interesting properties [270]. 265]. all of them can occupy the lowest lying state (at once) giving sharp peaks in the light emitted at low temperatures due to radiative recombination of the constituent electron and hole. Inter-chain excitons: This term is used for polymeric semiconductors to indicate that the constituent charges are located on different polymer chains. It is believed that intrachain excitons represent the main species that is formed after photoexcitation in conjugated polymers [78.of an excited molecule with a molecule in its ground in particular for conjugated polymers like PPV state. it is only weakly absorbing and shows little vibronic structure in its • Excitons in organic solar cells.5: Energy levels of the positive and negative polarons as well as the polaron (singlet) exciton. Intrachain excitons: This term also refers to polymeric semiconductors to indicate that the constituent charges are located on the same polymer chain.

e. two separate Fermi levels . EF N and EF P are the quasi Fermi-levels that are associated with the balance of electrons and holes . I2 . illumination the increased concentration of electrons in the CB would shift Ef up while the higher concentration of holes in the VB would require to shift Ef down .both exist at the same time in one semiconductor.1) EF P = EV + kT · ln(NV /p) (2.2) EV and EC stand for the top edge of the VB and the bottom edge of the conduction band respectively7 . 43]. 8 These are also known as Imrefs (Fermi read backwards. H2 O. Since there is no net increase of mobile charge carriers the conductivity cannot increase (unless the mobility of charge carriers is affected). When voltage is applied and/or the semiconductor is illuminated.otherwise it is called p-type conducting. If the Fermi-level is closer to the CB the material is called n-type conductor since more electrons are available for conduction than holes . no applied voltage). NO2 .are then introduced to describe the situation under this non equilibrium condition [98. Other ptype dopants are TCNQ or also high electron affinity semiconductors like C60 or CN-PPV or perylene 9 Atoms are not used because they are too small. changes of the concentration of mobile charges are often achieved by trapping of e. Thus. cyclic voltametry [125. As discussed above. That way the Fermi level would move towards the VB and make the material a ‘doped” p-type conductor i.3) Fermi-level The energetic position of the Fermi-level in semiconductors is important for two reasons: 1. organic molecules like o-chloranil. Examples for dopants10 of p-type materials are: Cl2 . 162]: may be seen as the HOMO and EC as the LUMO level in a quasi molecular model. 233]. 126.2 P. whereas for p-type semiconductors the relation is: EF = EV − kT · ln(NV /NA ) (2. and 2.g. the position of the Fermi level can be written as a function of both the effective density of states in the conduction band NC and valence band NV as well as the concentration of donors ND and acceptors NA : For n-type semiconductors this gives [162]: EF = EC − kT · ln(NC /ND ) (2.g.2. Another possibility to achieve doping effects in conjugated systems is electrochemical oxidation or reduction which can be achieved and investigated by e. in practice we often find increased conductivity (particularly in light) upon exposure to e. O2 . 127. The relative position of the Fermi levels is a measure for the type of conductivity .at the same time.see discussion below. 2. However. Together with the workfunction of the metal the Fermi level determines whether a blocking or ohmic contact is formed at the semiconductor/metal interface. Doping In inorganic semiconductors doping occurs via the introduction of dopant atoms that provide additional free charge carriers at room temperature so that the extra charges increase the conductivity for this type of charge carrier.g. SOLAR CELLS . electrons in the CB so that the concentration of mobile holes in the VB (relative to the mobile electrons) increases. a better hole than electron transport material. the concept of a Fermi level can no longer be applied.GENERAL 2. the Fermi level would be moved towards the CB if extra electrons are introduced or towards the VB if the concentration of mobile holes has been increased.the so called quasi Fermi-levels8 .10 CHAPTER 2.) 7E V . Br2 .4. They would diffuse out of the bulk and eventually neutralize with other atomic dopants.whether the semiconductor conducts preferably holes in the VB (p-type) or electrons in the CB (n-type) The energetic position of the Fermi level in a semiconductor represents the balance between the concentration of holes and electrons that occupy allowed energy levels under equilibrium condition (dark. Moreover. Upon e.and n. 10 In decreasing order of effectiveness. This effect may be explained by the generation of charge carriers due to exciton dissociation .7-trinitrofluorenone.type Organic Semiconductors and EF N = EC + kT · ln(n/NC ) (2.g. As a consequence. oxygen [268.4) with n and p denoting the concentration of electrons in the CB and holes in the VB. In organic semiconductors “doping” is usually achieved by the introduction of foreign molecules rather than atoms9 .

The synthesis of the perylene dye is reported in [60] while more information about the other shown molecules can be found in Ref. Dopants for n-type materials11 are e. Side chains are usually attached to molecules to introduce or improve solubility.2. 43. Common solubilising side chains (R. 2.2. We also 11 Note that H . In addition the fabrication of certain device architectures such as blends or laminated devices requires much more work or are not even feasible if thermal sublimation techniques have to be used.R1 . These properties are crucial for the operation of devices such as solar cells or LEDs . 225. 200].2. In fact.so that extra mobile charges due to illumination or doping can make a big difference. Fig. 164. phenothiazines and also semiconductors with low ionisation potentials. they can sometimes be described better using models for photo-conducting insulators [138].7 shows that the “mechanical” properties depend mainly on the position and number of side chains.2. R2 ) are often linear or branched alkyl or alkoxy groups.e. Liquid crystalline materials have only recently been investigated for the use in LEDs and organic solar cells [151.3 Structural Properties of Organic Semiconductors We have seen that the optical and electronic properties of organic semiconductors can be tailored by attaching certain functionalities to the conjugated system. in a p/n junction device) where they can possibly neutralize with another (different) dopant to form a salt. 6]. organic semiconductors can have very few mobile charges available . alkali metals.however. processing properties are also very important.2. Such doping molecules are preferably large (polymers) so that they cannot diffuse and penetrate into another material (e. They 12 Note that the terms pigment and dye are sometimes regarded as synonyms among researchers who are not concerned about solubility properties. NH can be used to compensate radicals or 2 3 impurities and thereby somewhat enhance the photocurrent. For example the fabrication of devices via thermal sublimation requires high vacuum conditions and high temperatures whereas processing from solution can be done at room temperature at ambient pressure.7: The “mechanical” properties of a flat structure like perylene diimides depend to a large extent on the position.particularly in the dark . We note that compared to doped inorganic semiconductors. [217] Fig.g. Examples are doping of PANI with camphorsulfonic acid or PEDOT with polystyrene sulphonic acid [38. 11 distinguish between molecules that have only a few (oligomers) or no (monomer) repeat unit and molecules (polymers) which have more than about 10 repeat units. 268. . We will discuss these interesting molecules in Chapter 7 in more detail.g. Recently. 38]. Figure 2. For a certain temperature range these material exhibit a phase where the molecules move like in a liquid but are still able to sustain or attain a certain structural order among the molecules as in crystalline materials. More information on doping of molecular semiconductors can be found in Refs. 199.6 shows how organic semiconductors can be divided into different categories due to their processing and mechanical properties. Moreover liquid crystalline systems were not accessible for investigation if only sublimation was available. A different form of doping resulting in a truly increased concentration of mobile charge carriers on the semiconducting molecule and therefore considerably increased conductivity (in both the dark and under illumination) can be achieved by complexation with foreign molecules. the “mechanical” i. a method has been found to process many pigments like dyes using strong acids [283]. Such molecules have to have a strong asymmetry to show structural order and are therefore either rod-like (calamitic) or disc-like (discotic) molecules. GENERAL PROPERTIES OF ORGANIC SEMICONDUCTORS diimides. [49. Oligomers and monomers that absorb visible light are often called chromophores and are referred to as dyes if they are clearly soluble or pigments if they are not12 . number and properties of the side chains rather than the constituent atoms of the main structure.

The electron stems from a not yet dissociated exciton and it recombines with a hole that liquid crystalline materials solvent molecules are not even necessary since the cores of the molecules are ”dissolved” in their own side chains. Thus 3 scenarios may be distinguished: 1. oxygen traps can also act as exciton dissociation sites and therefore generate free charge carriers in the presence of excitons.12 CHAPTER 2. The Donor/Acceptor Interface We have seen that photogeneration of charge due to traps is not very efficient since there is always the recombination of two charge carriers involved. They may be divided further into monomers (dyes. Flat molecules in particular have a strong tendency to stick together due to π-π interaction and form clusters of solids in many solvents. None of the recombination partners belongs to an exciton.2. an electron in the oxygen trap may further decrease its energy by decaying to the VB where it recombines with either a free hole or the hole of an exciton. Only one recombination partner is associated with an exciton.GENERAL Figure 2. Here we describe a more efficient way to generate free charge carriers . Both excitons are dissociated . none is generated. Two charges are lost. dissolve them13 . 215]. However. 13 In . pigments) and polymers. smaller molecules are better soluble and/or have lower sublimation temperatures but larger molecules give better films upon spincoating.the formation of an electron donor 2. For example.6: Organic semiconductors may be divided into three different categories according to their mechanical i. processing properties: insoluble. 2.4 Photogeneration Charges of Free Photogeneration due to Oxygen Traps It is widely believed that oxygen can act as trap for electrons in the CB of organic materials and thereby lead to a higher number of free holes in the VB. The proposed mechanism is the following: After a certain period. photo-oxidised PPV is believed to contain a good deal of carbonyl-groups which can facilitate exciton splitting but may also decrease both electron & hole as well as exciton transport properties [114. In this sense oxygen can act as a dopant that favours p-type conductivity as discussed above. In any case.although other effects may become more important. 3. the electron in the oxygen trap still reduces predominantly the mobility of negative charges thus sustaining its p-dopant character. We note that basically any defect that leads to an allowed free energy level between the middle of the bandgap and the CB may have the same effect as the here discussed oxygen trap . SOLAR CELLS .e. Bulky side chains can separate theses molecules and make it easier for solvent molecules to surround the individual molecules i. are more successful in solubilising the molecule the better they can prevent aggregation between molecules.e. Note that both electrons and holes can be produced with the same probability according to the above list. soluble and liquid crystalline.but only two charges (the hole of the first exciton and the electron of the other) are produced since the other two recombined. This means its dissociation produces one free charge but still eliminates two. Typically. comes from a different exciton.

8 b) shows that an exciton that carries its (CV) measurements where the current through the excitation energy can be transferred from the ma. the physical process are often very different. [268]. Confirmation of exciton dissociation can be obtained by comparing the EQE of a blend solar cell with equivalent devices comprising only the pure components. However. LUMO levels are often taken from cyclic-voltametry Fig. 127].semiconductors can be done15 but there are very els and the direction in which band bending occurs. 13 which is accompanied with some energy loss is also known as Foerster transfer [72] and can be used to shift the emission band of the exciton towards lower energies in e. Extra confirmation may be obtained if the experimental and/or theoretical LUMO and HOMO data of the components also suggest a sufficient offset of the band edges to split the exciton. Foerster transfer also allows the design of a possibly helpful directionality in an organic solar cell since it is an one way mechanism. 15 Using e. If the EQE of the blend cell is clearly higher over the entire spectrum exciton splitting at the D/A interface is most likely. HOMO and the Fermi levels. If the PL of both pure components decreases upon blending Foerster transfer can be ruled out otherwise the PL of one component would enhance the other.8: The interface between two different or. the Kelvin probe method described in Ref. it is helpful to plot PL versus blend ratio. Although on a first glance.4eV.g. Otherwise charge transfer may occur but the excitons do not split into their constituent charges and recombine eventually at the D/A interface14 . The material with the lower IP can accept the hole from the VB of the contacting semiconductor may therefore be termed hole acceptor or electron donor since it also donates the electron to the contacting acceptor. Fig.2.since those have to equalize upon contact.g. For practical purposes The latter is determined by the relative position of Fermi-level are often assumed to lie around the middle of the semiconductor bandgap. Whether Foerster transfer or charge transfer occurs with a certain combination of organic semiconductor materials is often difficult to assess . 14 This is the situation that is desired for the design of light We note that this method appears very useful if emitting diodes (LEDs) to favour radiative recombination at the interface between an electron and a hole transporting material[100]. GENERAL PROPERTIES OF ORGANIC SEMICONDUCTORS (D)/electron acceptor (A) interface. LEDs [246]. The the hole and the electron can decrease their energy oxidation or reduction potential are then regarded thereby.2. Although. We note that the prediction of the Figure 2. the concept may be reminiscent of the p/n junction in inorganic materials.2. The determination of Fermi levels in organic citon splitting).particularly if one or both materials do not fluoresce.offset of the band edges actually requires the knowlganic semiconductors can either: a) facilitate charge edge of the accurate position of the Fermi levels transfer (D/A interface). . b) energy transfer (no ex. We note that placing the higher bandgap absorber in front of the lower bandgap material to avoid thermalisation losses is proposed in Ref.8a) shows that the required transfer of charges can be realised at the interface between two materials if one material has a higher electron affinity (EA) whereas the other has a lower ionisation potential (IP). The one with the larger EA can accept an electron from the CB of the other and is therefore called electron acceptor. if they do. carrying excitons from the the shorter wavelength absorber to the longer wavelength absorber. [172]. the exciton may still recombine at the D/A interface rather than split into its constituent charges there is a good chance for the latter. This can be helpful to concentrate delivered excitons in a spatially small high field region [267] to enhance dissociation. 126.2. few data in literature [172]. We note that the offset of IP and EA needs to be large enough so that the resulting field (the potential gradient at the junction) can overcome the exciton binding energy which is typically around 0. What kind of interface forms depends on the position of the HOMO and LUMO lev. This process of excitation energy transfer as HOMO and LUMO levels.semiconductor gives a peak at the voltage where oxterial with the higher bandgap to the other if both idation and reduction occurs [125.

Much effort. 2.e. 154]. π electron arrangement. a D/A interface or a metal which can also act as an electron or hole acceptor or simply provide an electric field that can split the exciton. Otherwise care has to be taken if absolute numbers are compared.GENERAL mesomeric effect. since the scan rate. The latter considers that many conjugated structures can equally be drawn with a different double/single bond i. 140.as discussed above. 43]. 37] and perylenes [154.g.e. charge transfer salt) or only little extra (light) energy is required to complete it. CV values for films and solution may also differ. the mesomeric effect takes into account the influence of free electron pairs. CF3 substituted PPV and perylene diimides [167. Most organic semiconductors behave more like electron donors. Such structures can accomplish charge transfer sometimes already in the ground state (e. 267]. to form aggregates i. 2.5 D/A Materials For the design of D/A solar cells it is important to know whether a specific material combination is likely to lead to charge transfer + exciton splitting.14 all components can be measured at the same setup under comparable conditions. 154]. As a rule of thumb. the bandgap is lower the larger the conjugated π-system is. Otherwise they would stick together.g. Bandgap. One way to find out is to build blend devices and look for PL quenching of both components and enhanced EQE with respect to the single components .2. photogeneration of charge is always assisted by an external factor such as the presence of traps. The general rule is: Molecules dissolve in solvents with similar structure. Solubility.e. CF3 . In the following we provide some background knowledge that allows guessing not only the D/A properties but also some other relevant properties for organic photovoltaic materials: 1. for example PPV. alcohol or iso propanol dissolve polar molecules while less polar solvents (e. SOLAR CELLS . and the assessment of the peak (it is not clear if the peak or the onset should be taken) can vary and affect the result considerably. time and money can be saved if the chemical structure of the chosen molecules is considered. Polar solvents like water. Given that photo excitation results in the formation of excitons with a binding energy considerably higher than kT (26meV).3. particles which do not contain solvent. The average of these possible π electron locations can be regarded as the overall mesomeric effect of a specific molecular structure. It is essentially a result of both the inductive and the . F . Donor/Acceptor properties. PPP and phthalocyanines. Examples for large π-systems are phthalocyanines [117. Solubility is good if the planar parts of the conjugated π-systems of the molecules cannot get too close to each other. = O (keto-groups) or diimides (all perylene derivatives in this thesis are perylene diimides).particularly if the structure is asymmetric. Examples for functionalities that favour electron acceptor properties are: CN. 3. Examples for electron acceptors are CN-PPV. 119. 173. 118. CHAPTER 2. Flexible and bulky side chains or atoms that stick out of the molecular plane can prevent molecules from getting to close . The bandgap can also be small if a molecule consists of a donor and an electron acceptor that are connected via a conjugated structure to create a “push/pull system” as in e. Note that their ability to enhance the electron affinity (EA) of the entire molecule is not purely connected to the electron withdrawing properties of these functionalities. driven by their π −π interaction.g. PT. These groups are known as electron withdrawing groups if they are attached to an unsaturated (=conjugated) system. 155.these two terms can not really be used separately. naphthalocyanines [171. Sulfur in a conjugated systems also tends to reduce Eg . Thus photogeneration of charge in these organic semiconductors can be regarded as extrinsic as opposed to inorganic semiconductors where separation occurs mainly together with photo excitation.2. xylene) dissolve non-polar molecules. While the inductive effect is determined by the electronegativity of the substituents. Examples for charge transfer salts are PVK-TNF [154] and for polymethines: merocyanines [204. Since a donor can act as an acceptor for an even stronger donor . toluene. As a rule of thumb. it can be said that a solvent containing other atoms than carbon and hydrogen has a good chance to be more polar . larger clusters i. There are certain functionalities that are likely to make a material an electron acceptor with respect to most other materials.see also discussion in Section 2. poly-methines.

205] to which we refer the interested reader. band bending occurs as depicted in Fig. 15 2. it is not clear to what extent the developed models can be applied to organic semiconductors we only describe the basic concepts as they are often used to understand and predict the behaviour of organic semiconductor/ electrode interfaces. The theory of contacts between doped semiconductors and metals was developed in 1938 by Mott and Schottky [176. Note that band bending at the D/A interface would be the other way round (blocking) if both Fermi-levels are assumed to be close to the middle. 225]. both the electron affinity and ionisation potential of the semiconductors does not need to change .2 . the presence of impurities or structural defects usually leads to energy levels within the bandgap. 16 Systematic intentional interruption of the conjugation has been shown to increase the PL in PPV considerably [] Figure 2.9 we have drawn the band energy diagram as it is expected for the situation before contact and after contact. However. The energetic decay of an excited electron can occur faster and therefore more efficiently if smaller energy jumps are required. Since then charge injection from metallic contacts into semiconductors has been studied extensively and is explained in detail in the semiconductor literature [220. The same is true for holes from the VB near the ITO contact. In Fig. These types of contacts are referred to as blocking contacts. Interestingly.2.6 Electrode/Semiconductor terfaces In- The direction the band bending occurs depends on whether the Wf of the electrode is above or below the Fermi-level of the semiconductor. Materials with higher bandgaps tend to show stronger PL16 .since the corresponding vacuum level follows both the band curvature and vertical shifts. a) the situation before contact.2.2.2. c) the formation of a non-blocking (ohmic) contact for holes (Au/D) and electrons (A/Ca). The reason why this 17 p type conductivity is assumed for D and n-type for A see also discussion in Section 2.not even where band bending occurs .2. The workfunction of the electrodes are drawn as equal here to show the situation in the short circuit mode.2. molecules in solution fluoresce much stronger. Also conjugated systems containing certain atoms like sulfur tend to give lower PL than conjugated structures containing only carbon and hydrogen. Photoluminescence (PL).9: Energy band diagrams of D/A devices with the electrode interfaces. Practice shows that PL is quenched if the molecules can aggregate. After contact the Fermi levels (dashed lines)and Wf equalize and band bending occurs: b) the formation of a blocking contact for holes (ITO/D) and electrons (A/Al). Note that Al forms a blocking contact only with with the CB of the n-type semiconductor17 but an ohmic contact with the VB.after contact . Photoexcited electrons from the CB of the electron acceptor (A) encounter a barrier before they can reach the Al electrode. Consequently. 138. This may be the case and desired for LEDs [100]. The depicted band bending can be qualitatively predicted by assuming that the edges of the LUMOs and HOMOs are “pinned” to their absolute energy values as they were before contact. It can be seen that . The bulk of the semiconductor sustains the same distance to the equalising Fermi-level. In case the workfunctions of both electrodes are deeper inside the bandgap than the Fermi-levels of the materials they are contacting. Since these contacts can become nonblocking upon changing the potential of the electrode they are also called rectifying or Schottky contacts. Moreover.band bending occurs at all interfaces and that the Fermi levels and workfunctions equalize. 241. GENERAL PROPERTIES OF ORGANIC SEMICONDUCTORS 4.9 b).

Ca is more prone to oxidation than Al. For a Schottky barrier the current is given by Φ3/2 Φb + qV K (Φ+qV )1/2 (2. the current density (J) is given by erably. it is necessary to take into account the diffusion of carriers within the barrier region back towards the contact.3 Electrode Materials In organic devices the workfunction of the electrode materials is very important since it deterThere are two principal mechanisms for the charge mines together with the LUMO/HOMO and Fermicarriers to overcome barriers: Thermionic emission level of the semiconductor whether the electrode and quantum mechanical tunneling (field emission).5) J = A T e kT the elements with its values of Wf .7) vacuum. However.7) e Φ where V is the applied voltage and K a material constant. Ca. forms an ohmic or a blocking contact for the reFor thermionic emission over a triangular barrier spective charge carrier (holes in VB. can become important. both charge carriers find barriers if they want to get out of the semiconductor. band bending occurs the other way round so that electrons in the excited state can reach the Ca electrode and holes the Au electrode without encountering a barrier. T the absolute temperature If an electric field is applied across the barrier the many numbers for single crystals which depend on its crystallographic orientation have also been rebarrier height decreases by ported in Ref. However. [170].2. The Wf of both electrodes is further away from the middle of the bandgap than the semiconductors Fermi-levels. 94]. [170]. The analysis is complicated and depends on the type of barrier assumed. (2. At sufficiently low temperatures . Current Through Barriers CHAPTER 2.CB).8) where J0 is essentially a constant for low doping levels and n ≈ 1 at high temperatures [101]. Common electrode materials for the electron colq3 E lecting contact (low Wf required) of organic solar (2. In. Thus currents through this device are decreased by the contact barriers. The situation is analog for the semiconductor/ITO interface. The essential characteristic of field emission is that the current is insensitive to temperature.9 illustrates that the situation is improved if Au and Ca are used as contact materials. ∗ 2 −Φ In Table 2. Gold electrodes are expensive and have to be very thin to become transparent and are more difficult to pattern via etching.as far as it was where A∗ is the effective Richardson constant and known in Ref. For a Schottky barrier the current can then be written as J∝ J = J0 eqV /nkT (2. rectifying. electrons in of height Φ from a metal into a high-mobility semi.1 are valid only for poly-crystalline materials. thermal excitation can increase the tunneling current. after contact. Since charge separation at the D/A interface leads to an excess of electrons in the CB of A (n-type) and holes in the VB of D (p-type).1 we have listed the periodic table of (2. . The values in Table 2. SOLAR CELLS . Fig. Ag whereas for the hole collecting contact high workfunction materials like Au where E is the applied electric field. Thus.16 particular interface situation is called rectifying is because the n-type conductor is supposed to transport electrons in its CB .6) ∆Φ = 4π 0 cells are Al.and for electrons it is blocking i. q the elare preferred. For both solar cells and LEDs one ementary charge and 0 the dielectric constant of 18 This is not considered in Eq.e. 2. but strongly dependent on the applied field (voltage).GENERAL In a semiconductor with a low mobility. emission due to quantum mechanical tunneling through the barrier (field emission). because of the lower Wf of Ca compared to Al. A full treatment is given by Kao and Hwang [138]. Since the tunneling rate depends also strongly on the width of the barrier18 . or for large barriers at high fields. Practice confirms that organic devices using Au and Ca instead of ITO and Al show indeed considerably higher open circuit voltages and often also higher photocurrents and EQE values [114. For the above reasons Al and ITO electrodes are actually more suitable for research devices since they can be reproduced easier. Moreover a large difference in workfunction of the electrode materials can increase the Voc considconductor.

19 Considering the typical thickness of organic devices (30150nm) the substrate surface should ideally be smooth on a nm scale. Moreover.22 Hg 4.12 Pt 5. 191]. The transmission properties of the thicker ITO substrates do not change much since the material does not absorb in the visible but interference effects cause significant spectral dependence of the transmission. IA Li 2.22 VIA O S Se 5.96 Fe 4.7 Eu 2. The material can be highly conducting if there is excess of In due to a lack of oxygen .55 U 3.3 Ta 4.1 Tb 3.0 Rh 4.49 Pm - Sm 2.59 Ba 2.1 La 3. the thicker the ITO layer is. The surface can be modified via plasma etching[271.9eV is widely used.14 Fr IIA Be 4.2. 20 Thicker layers of PEDOT can be obtained by repeated spincoating since it seems poorly soluble in water once the film is formed.84 IVA C 5.3. Other conducting glasses are Tin oxide and Indium oxide. ITO covered quartz substrates are commercially available since they are widely used as conducting glasses in the liquid crystal screen industry.9 V 4. Also the optical properties change considerably in the thickness range from 10-20nm so that devices with slightly different thickness may not be immediately comparable. Alternatively.5 and 4.65 Cu 4.5 Gd 3.95 Po - Sc 3.16 Cs 2. The numbers refer to the workfunction in eV of the elements in poly-crystalline form.30 Rb 2.26 Au 5.9 Th 3.75 K 2. other water.45 Al 4.insoluble materials may be spun on top of it20 .27 Ni 5.0 Si 4.7 Ra IIIB IVB VB VIB VIIB VIII IB IIB IIIA B 4. For these reasons so called conducting glasses are often used.microns) are problematic because of the increased absolute surface roughness19 which can cause shorts in the thin organic films. Semi transparency can be obtained if the sublimed metal e.71 Os 4. 82] or the semiconducting molecules can even be chemically attached to it[178].5 Y 3. conjugated polymers with absorption in the whole visible range can be used if they are doped so that the allowed energy levels move deep into the bandgap and create an absorption minimum in the visible region. 55] and PITN [272.0 Sn 4. Au is not much thicker than about 15-20nm compared to about 50-100nm which are typical values for non transparent contact.3 Nb 4.1 Zn 4. Typical thickness of ITO in our devices is 100nm.4 Ce 2. .5 Ru 4. The large bandgap allows no absorption of wavelengths longer than about 350nm.so that In acts as ntype dopant leading to very low sheet resistances for already only 100nm thick layers of ITO.28 Ga 4. Particularly Indium Tin Oxide (ITO) which is a degenerated semiconductor comprising a mixture of In2 O3 (90%)and SnO2 (10%) with a bandgap of 3.25 Pa Pr - Cr 4.2 Mn 4.66 Ca 2. 263. An additional advantage of PEDOT is that it can be processed from (water) solution and therefore easily be spincoated.85 Ge 5.75 Sb 4.2.33 Cd 4..55 Bi 4.98 Ir 5.33 Zr 4.2 In 4.42 Pb 4.25 VA N P As 3.98 Mg 3. Example for such polymers are PEDOT [193. We note that ITO itself can be used as antireflection layer [51] as active semiconductor [122]. We note that the sheet resistance of such a thin layer can be considerably increased compared to a 50 to 100nm thick layer and possibly add to the series resistor of the cell significantly.6 W 4.9 Te 4.g.5 Mo 4.65 Ag 4.83 Co 5. ELECTRODE MATERIALS 17 Table 2. Some important soluble or transparent conductors are listed with their workfunction and references in Table 2. 222.15 Pd 5.63 Nd 3. Values smaller than 5 Ω/ square are available.1: Workfunction of the elements.5 Ac - Ti 4. The sheet resistance of these ITO substrates is lower.9 Na 2.12 Ti 3.1 Tc Re 4.87 Sr 2.7eV and a Fermi-level between 4.0 Dy - Ho - Er - Tm - Yb - contact has to be at least partly transparent. Very thick ITO layers (several 100nm .05 Hf 3.

[97].3 and 1. The solid lines are semi empirical limits assuming AM0 (=space) and AM1.e. place increasingly and the extracted external current decreases. The higher the bandgap the higher the voltage can be. the higher the photocurrent. 39] transp.4 2.5eV allows reaching power conversion efficiencies around 30%. Thus considering the voltage dependence of the IV curve .3 and 1. the photocurrent.the LUMO. The ratio between these two areas represents a measure of the quality of the shape of the IV characteristics: FF ≡ thus: Pmax = (IV )max = Voc · Isc · F F (2.9) In practice the photoexcited electron can decrease its potential energy by losing energy to phonons until it reaches the lowest lying level in the CB . polymer[38] sol.5 is the standard spectrum at sea level .5 condition a semiconductor bandgap between 1. polymer transp.2.g. SOLAR CELLS . Thus. a low bandgap material can absorb more photons and thus increase the number of photogenerated charge carriers i. transp.5eV gives the highest power. The abbreviations are listed in Chapter 1.4-4.5 (=terrestrial) solar radiation.2-5.1 Photovoltaic Characterisation Power Conversion Efficiency Figure 2. Since the phonon energy dissipates into heat this processed is known as thermalisation.3eV [33] 4.10: Power conversion efficiency limits as a function of the semiconductor bandgap. 144]21 4.84eV [172] [273] comment sol. Shockley and Queisser were the first who calculated the maximum power conversion efficiency for a semiconductor with a given bandgap assuming only radiative recombination and the solar radiation o earth.12eV like Si. Ip )combination that gives the largest power rectangle is called the maximum power point. The semi empirical limits take into account realistic loss mechanism e.11.GENERAL Table 2.18 CHAPTER 2. The voltage/current (Vp . The graph shows that for both AM0 and AM1.[55. material PANI PEDOT ITO FTO TO Wf (eV) 5.10) (IV )max Isc Voc (2. AM0 is the solar spectrum above earth’s atmosphere while AM1. This means that recombination can take The higher FF the more the IV characteristics resembles a constant current source with a maximum voltage and the higher is the electric power that can be extracted.2. Figure courtesy of Ref. The lower the bandgap. by assuming realistic values for the fillfactor. The graphs show that for both the terrestrial and the space solar spectrum a semiconductor bandgap between 1. This behaviour can be seen in the fourth Quadrant of the IV characteristic in Fig. How the maximum conversion efficiency varies with the bandgap can be seen in Fig.9[188.4. there must be an optimal bandgap for a given illumination spectrum. This maximum area is larger the more the IV curve resembles a rectangle with the area Voc x Isc .2: Molecular electrode materials for organic solar cells. 2. the dashed line is based purely on thermodynamic considerations for blackbody solar cells under AM0 radiation. On the other hand. transp. any appliance connected to a solar cell can utilize . As a consequence of thermalisation the semiconductor bandgap is often regarded as a measure for the achievable voltage. Since a high load resistor reduces the current flux the charges need more time to get out of the semiconductor.see Chapter 8.the maximum power is the maximum product of I and V that can be found amongst the data points in the fourth quadrant.10. Hence. They obtained a value of 30% for a semiconductor with a bandgap of 1.

2 The Equivalent Circuit Diagram (ECD) ECDs are frequently used to describe the electric behaviour of more complex semiconductor devices with a network of ideal23 electrical components such as diodes. (2. the cell should be measured by one of the internationally recognised institutions that offer solar cell efficiency measurements such as the National Renewable Energy Laboratories (NREL) in Golden (USA) or the Fraunhofer Institut for Solar Cell research in Freiburg (Germany). a diode that accounts for the nonlinear voltage dependence and a shunt as well as a series resistor.5 spectrum which can be approached by commercial solar simulators. In order to compare efficiencies of solar cells. The latter is neglected in Eq. The extracted current is negative.otherwise it is positive.according to the standard in electronics. In order to describe the power conversion efficiency η of a solar cell the maximum output power Pmax has to be related to the power of the incident light Plight . The most common standard at present is the AM1. Using Eq. The fourth quadrant represents the voltage and current that is generated by the cell. 19 recognition of a solar power conversion efficiency number of a cell is desired.12: ECD of a solar cell. Otherwise power would be wasted in heating the series resistor (V < Vp ) or via increased current losses through the ideal diode and shunt resistor (V > Vp ). Also shown is a load resistor R and its voltage drop and current.11). 2.11: Current versus applied voltage of a solar cell. solar radiation standards have been defined in the past. 23 A real current source for example always has losses due to a shunt and series resistor.4.2. We note that in Chapter 8 we describe how an inexpensive solar simulator can be built or . This can be the solar spectrum or the spectrum of the artificial illumination for indoor applications (for e. PHOTOVOLTAIC CHARACTERISATION the maximum output power only if it’s supply voltage is around Vp or in other words: The load resistor Rl = Vp /Ip . In case its not feasible to send a cell to such an institute.η can be estimated from intensity dependence measurements of the spectral response. current or voltage sources and resistors. it would still be very helpful to estimate the solar power efficiency of a solar cell rather than quoting monochromatic power efficiencies (or other) to allow at least rough comparisons among different research groups. 22 the Figure 2. The current I is negative if V > Voc and it flows “into” the device . pocket calculators). The circuit consists of the following ideal components: A current source IL that considers the light-generated current. We note that.10) to express Pmax considering the wavelength dependence of the parameters involved we can write: η(λ) ≡ Isc (λ) · Voc (λ) · F F (λ) Plight (2. .alternatively . (2.11) Because of the wavelength and intensity dependence22 power conversion efficiencies are only meaningful for a given spectral distribution and intensity.g. An externally applied voltage is necessary to obtain data points in the first and third quadrant. Figure 2.4. if international current usually scales linearly with the light intensity but Voc also increases slightly. The current arrows point into the direction the holes flow .

• solar cell voltage V. The mobility can be affected by space charges and traps or other barriers (hopping).g for holes at the ITO electrode) can affect the IV curve in the third quadrant or even lead to FF values < 0.see Fig. RS is also increased with a longer traveling distance of the charges in e. thicker transport layers. mobility of the specific charge carrier in the respective transport medium.e.g D/A interface). a more comprehensive ECD for organic devices . SOLAR CELLS .g. Provided the series resistor Rs is at least one order of magnitude lower than Rsh . the principal loss mechanisms are the same and we may therefore apply the same circuit. • the capacitor C. We note that also the voltage drop across a load resistor .12) CHAPTER 2. the current increases with applied voltage.GENERAL • ideal diode D = voltage dependent resistor that takes into account the asymmetry of conductivity due to the built in field in D/A cell (difference between the acceptor LUMO and the donor HOMO) or the nature of the semiconductor electrode interface (blocking contact) in single layer cells.2.the range between 0 and Voc .see Fig. Rs can be estimated from the (inverse) slope at a positive voltage > Voc where the IV curves becomes linear: RS =≈ ( I −1 ) V (2. In order to obtain IV curve data in other voltage ranges (V < 0 and Voc < V ) in the IV curve an external voltage source is required. This diode is always present and not considered in any of the ECDs drawn here. Rsh can be derived by taking the inverse slope around 0V: Rsh ≈ ( I −1 ) V (2. • The series resistor Rs considers conductivity i.2. The external voltage source can then act as a current amplifier to boost the photo sensitivity24 but the actual photon to current conversion efficiency (EQE) of the solar cell or photodiode can not really be increased.25. Otherwise an extra shunt resistor Rsh2 has to be considered .20 Fig. The capacitor that takes into account charging/discharging and other time 24 Provided This is because at very small voltages the diode D is not conducting and the current driven by the external voltage (positive or negative) is only determined by Rsh + Rs with Rsh (typically) being much larger. • The shunt resistor Rsh is due to recombination of charge carriers near the dissociation site (e. The cell can generate a voltage between 0 and Voc depending on the size of the load resistor.13. The formation of an extra blocking contact (e. However. We note that the normal blocking contact for electrons at ITO and holes at Al allows charge injection into the electrodes if a sufficiently high negative voltage is applied.before any recombination can take place.13) This is because at high positive external voltages V the diode D becomes much more conducting than Rsh so that RS can dominate the shape of the IV curve. The diode characteristic is not necessarily Shockley type. near the electrode. This diode is responsible for the nonlinear shape of the IV curves.can be simulated by the same voltage source so that the entire range can be scanned by applying an external voltage.13 may comprise the following extra components: • another diode D2 .g. IL = number of dissociated excitons /s = number of free electron/hole pairs immediately after generation . . Although the specific physical processes in organic semiconductors may be different and therefore lead to other parameters. Note that the IV characteristic of the ideal diode D is only equal to the IV characteristic of the entire cell (circuit) if Rs = 0Ω and Rsh = ∞Ω. We suggest to associate the following (more general) processes with the single components in the ECD: • The current source generates current IL upon illumination.2.12 shows the ECD that is typically used for inorganic solar cells. These are the components of a ECD with which we can associate the most important effects in solar cells of all types. For example mobility of holes in a p-type conductor or electron donor material. it may also include recombination further away from the dissociation site e. Note that the current for V > Voc and the extra current for V < 0V is delivered from the external voltage source.

It considers shorts due to pinholes or significant conductivity of the bulk material.relatively small Rsh values.17) If we now use Figure 2. we can formulate the following relation: Ish Rsh + Rs Rs =1+ Rsh Rsh we can write Eq. 26 Related to 1cm2 . (2. It may also account for recombination losses near the electrodes. Organic cells very often suffer from high series resistors and .12 and Kirchoff’s laws for current knots and voltage loops. As a result. The effect of Rsh2 is considered in Rsh if Rs is considerably smaller than any of the two shunts. Using the simple ECD in Fig. storage effects of charges and may be of particular interest for AC measurements i. the shape of the IV curves and the absolute currents are more affected by these resistor values in organic than in inorganic cells.19).19) even though 1 + Rs /Rsh an be neglected. As a consequence not only the losses due to both resistors are higher than in silicon cells but also the term 1 + Rs /Rsh can become significantly different from 1 and has to be considered in Eq.16) and replace Id in Eq.13: Extended ECD of a solar cell.4.2.in addition .18) I= IL − 1+ V Rsh Rs Rsh − V −IRs I0 · (e nkT /q − 1) Rs 1 + Rsh (IL − Id − I) Rsh = V + IRs which can be transformed into 25 Within (2.3Ω and a few Ohms and Rsh larger than 1000Ω so that the ratio Rs /Rsh is very small and the term 1 + Rs /Rsh is virtually 1 and can therefore be neglected in Eq. The formation of a counter diode D2 may sometimes need to be considered as well as another shunt Rsh that connects the two electrodes directly. We note that current and resistor values depend on the illuminated area and need to be related to it if comparisons between cells with different illumination areas are desired25 . Even though the current I in Eq.19) This equation will be used later when we discuss what can be learned from the specific shapes of IV characteristics.e.e. However. PHOTOVOLTAIC CHARACTERISATION 21 RS V + 1) = IL − Id − (2.17) as: (2. Note that both Rs and Rsh still appear in Eq. (2.2.14) this thesis the illuminated area was typically around 4mm2 (only slightly smaller than the active area) which allows immediate comparison of the absolute values. The capacitor accounts for accumulation i. we can already see the following: Inorganic solar cells have Rs values26 between 0.19) appears also in the exponent which requires numerical methods to find solutions. (2. (2. dependent effects that can be significant since the contact area A can be large and the distance between the electrodes d is very small: C= A d • an extra shunt resistor Rsh2 .15) Rsh Rsh if we now assume that the Shockley diode equation describes the voltage dependence of the current Id through the ideal diode D I( Id = I0 · (e nkT /q − 1) V −IRs (2. (2. However. even in silicon solar cells the voltage drop across Rs represents an important loss factor and has to be considered as well as the loss current through the shunt. Note that the dark current depends on the actual device area . photocurrent measurements with chopped light. (2.19). (2. the situation can be very different for organic cells.regardless of the size of the light spot.15) we obtain I = (IL − V −IRs V Rsh Rsh ) −I0 (e nkT /q −1) Rsh Rsh + Rs Rsh + Rs (2.

GENERAL 2. If we apThe current through such a cells or photodiode ply a positive voltage across the cell electrodes we is determined only by the current through the ideal create a voltage drop across Rsh that is equal to D . given a constant IL the Voc is higher the smaller the Note that n which determines the shape of the dark current is . . Typical Voc of silicon Effect of Shorts cells under solar conditions are around 550mV. In the preceding Section we have deduced a formula that allows calculating the output current of a solar cell.22) is not much smaller than IL . However. Organic solar cells have often values for series resistance which are more than 3 orders of magnitude Effect of Rsh A formula for Voc that considers the influence of Rsh can be deduced from Eq. SOLAR CELLS .note that I0 determines the height IV curve stands outside the logarithm in Eq. Both the ideal diode D and Rsh are now the comqV /nkT ponents that determine Voc : Suppose Rsh is not − 1) (2.here represented by the Shockley equation. also the formula for Voc becomes transcendent and requires numerical modeling to find solution.22 CHAPTER 2. tion of Rsh or I0 . (2.21) by setting I = 0. small. We will also discuss how PL efficiency can positively affect the solar cell performance and show how the upper limit for Voc can be estimated.19) after setting keep the ideality factor in this equation to consider different shapes. The value for Voc still depends on The Ideal Solar Cell the ratio IL /I0 but now IL is decreased by the presFirst we consider the case of an ideal solar cell that ence of a finite Rsh whereas Rs is unimportant since comprises only the ideal 27 diode D in the dark so there is no current flowing through it that can crethat Rs = 0 and Rsh = ∞ in Eq.4. The sum of current IL that is simply superimposed (added) the currents through D and Rsh yields the current upon the normal rectifying IV characteristics of the through the electrodes of the solar cell for a given diode D: applied voltage. The highest voltage measured with a voltmeter with high internal resisin this quadrant develops at the electrodes when tor across the device electrodes and is then termed IL just manages to cancel the dark current. The curpositive voltages I increases exponentially.21) the current IL some of which passes through the diode where a voltage drop is generated that is big The addition of IL results in a region of the fourth enough to allow the rest of IL to go through Rsh quadrant where electrical current and voltage can if the electrodes are open. We note that I0 can q I0 be related to PL efficiency as discussed later in this Note that the effect of the IL /I0 ratio is relatively thesis.20) Id = I0 (e very high and the device is in the dark. The same voltage can be be extracted from the solar cell.3 Open Circuit Voltage the output current to zero (I ≡ 0) to give28 : Voc = nkT IL − Voc /Rsh ln( + 1) q I0 (2. The latter controls the ”height” Voc can than be derived quantitatively using: of the IV curve of D. since Rsh can vary considerably it can seriously decrease Voc if Voc /Rsh nkT IL Voc = ln( + 1) (2. Thus. Now. Canceling of Id by means of and has therefore a stronger influence than a variaIL can be considered in Eq.23) of the characteristics. rent that can pass through the diode D at Vd is Upon illumination the light generates a photo determined by its IV characteristic29 . For the voltage Vd across the ideal diode D. Reduction of I0 by a factor 10x increases Voc only by 25mV ·ln(10) (58mV). open circuit voltage Voc . (2. In this Section we derive an expression for the open circuit voltage Voc and consider the effects of the single components of the ECD.23) This equation relates the maximum output voltage not only to the light generated current IL but also to the reverse saturation current and the shunt resistor. the diode D is still ideal in the sense that there are no Ohmic losses.19) turns into: ate a voltage drop (loss).while for organic cells it depends on many material and device parameters. (2. 29 Note that the IV characteristics of the diode D is not the IV curve one obtains when measuring the entire device (with Rsh and Rs )! For inorganic solar cells it is assumed to be the Schottky diode characteristics . Upon illumination. 27 We 28 There is no influence of R since R can only affect the s s voltage when there is a current flowing through it. However. (2. the current source generates I = I0 (eqV /nkT − 1) − IL (2.

25) ΦP L Note that the Shottky equation remains unchanged if the photoluminescence efficiency ΦP L = 1 i. still give reasonable voltages as long as their optical and charge transport properties are very good and the PL is not too low. 100%. Rsh and in particular I0 can affect the value for Voc .6V [234].2. In fact.3V) in contrast to poorly fluorescent materials like PIF (Voc < 0. polymers and dyes with higher PL yield appear to give high Voc values like PPV (1. So called non-fluorescent materials may. MEH-PPV (1.4. For example. 234]. Within this thesis we call I01 the PL independent reverse saturation current. (2. In this Section we discuss whether there is an upper limit for Voc and what might determine it. (2. We have seen that according to Eq. Analogously.24) Rs can actually reduce Voc if Rs reaches values in the order of Rsh2 . Thus ”shorts” do not need to have very low resistances to smaller Voc considerably .23) to derive Voc can be written as Id = Voc = nkT Il − Voc /Rsh ln( + 1) q I01 /ΦP L (2.25).they just need to approach the value of Rs which can be as big as 107 Ω in some small area organic devices! Vocs = Voc · Rsh2 Rs + Rsh2 (2. since a short across the electrodes represented by Rsh2 in Fig.g Si (10−3 to 10−4 ) is still high enough to result in a IL /I0 allowing Voc to reach 0. The Upper Limit of Voc In the previous Section we have discussed how Rs . Thus the expression Eq.13 represents a voltage divider in conjunction with Rs according to Eq. (2. The Effect of Photoluminescence Any type of recombination30 represents a less mechanism for the photoexcited charge carriers in a solar cell. respectively [208. Possible upper limits with regard to excitation intensity may only be passing through the external circuit. The higher PL (10−1 ) in GaAs predicts Voc values of 1V under 1 sun which is also in good agreement with the observations [234].g. Indeed the more fluorescent porous silicon has been reported to show lower I0 values being consistent within Eq. Similar considerations hold for the very good photovoltaic performance of certain organic material with relatively low PL efficiencies.24) 23 I0 V I01 ·(e nkT /q − 1) (2. .e. [234] radiative recombination ΦP L can be considered as a correction factor for the reverse saturation current I0 in the diode characteristics 30 Before The reverse saturation current I0 = I01 /ΦP L can be obtained experimentally by modeling of the IV curve and gives indeed about 103 larger values for e. This means that currents measured in short circuit mode still have to get through a large resistor which can seriously limit the output current particularly if also Rsh is low.2V). If there is no external circuit (open circuit mode) the charges accumulate at the electrodes to build up Voc before they decay showing their maximum PL efficiency. it has been shown that the PL of e.represented by Rsh may be more effective in decreasing Voc than recombination in the middle of the device if Rs is so high. PHOTOVOLTAIC CHARACTERISATION larger. and 30% if the chlorophyll molecules are separated from the rest of the electron transport chain. non-radiative recombination affects the photoexcited charge carriers statistically much faster than the unavoidable radiative decay that does not require any semiconductor defects.g.26) the voltage is higher the lower I0 and the higher IL is. (2.26) This also makes voltage measurements across the device electrodes more critical. silicon cells (ΦP L ≈ 10− 4) than for cells using organic semiconductors (ΦP L ≈ 10− 2). which corresponds to operation under load and open circuit mode. In other words. We note that non-radiative recombination losses affect both I0 and the shunt resistor Rsh . if there is only radiative recombination the charge carriers have much better chances to reach the electrodes and circumvent the radiative decay by entering an external circuit. For very fluorescent materials the I0 is small and Rsh should be high. According to Ref. Voltmeters need to have internal resistors larger than 107 Ω to exclude effects on the circuit that lead to lower readings. chlorophyll is about 3% in a living plant. the low PL of e. In virtually all inorganic cells and many organic ones IL scales linearly with the light intensity over many orders of magnitude.2. a highly fluorescent material can have relatively low charge carrier mobilities and still generate a high Voc and good photocurrents.2V). however. recombination that takes place near the electrodes . Moreover. Indeed. Hence. However.02V) and HPc (Voc <0.

On the other hand shorts may be avoided more efficiently which also helps to develop higher voltages at the electrodes. field. However.e.27) if only positive numbers are used.whichever is higher .28) Thus. it can be expected that the bandgaps of the semiconducting components determine the (theoretical) upper limit of D/A cells provided the ∆Wf is large enough.GENERAL low Wf for the negative and high Wf for the positive hole collecting electrode often helps to increase Voc . This is often explained by assuming that ∆Wf introduces a “built in field” across the thin semiconducting layer that helps to extract the photoexcited charge carriers. 32 In case of D/A devices the voltage relevant bandgap is the difference between the LUMO of A and the HOMO of D. ∆Wf > Eg may result in Voc larger that the “combined bandgap” EgDA since the internal field at the D/A interface can assist the extraction of the charges through the “high” lying electrode workfunction. We note that in single layer devices ∆Wf > Eg is not expected to increase Voc above Eg since charge injection from the LUMO and HOMO into the electrodes is limited. For organic solar cells. as it can be seen from Fig. SOLAR CELLS .4.8V [234]. We note that the consideration of the unavoidable radiative recombination losses already reduces the upper limit for Voc in Si from the bandgap value (1. What determines in practice the absolute value of I0 in inorganic devices has been reported: majority charge carrier and dopant concentration. gradient of a potential in space can be treated as a .e. Here we focus on a more general approach to find the Voc upper limit so that it may be applicable to the organic solar cell structures described in this thesis. In practice. In other words. As a result of these considerations the upper limit for Voc in single layer devices is expected to be the bandgap whereas in D/A cells ∆Wf of the electrodes or the bandgap . Thus the maximum photo-voltage that a semiconductor can develop should increase linearly with the the excitation energy.determines the maximum photogenerated voltage. As a consequence the difference between the HOMO and LUMO levels (the bandgap) can be seen as a theoretical upper limit for Voc . 31 A CHAPTER 2. heat this process is called thermalisation. The drawback is that the use of low Wf materials such as Ca often requires the sublimation of a protection layer on top because of the increased tendency to form an oxide layer when exposed to air that represents an extra obstacle for charge carrier extraction. Qualitatively. a 2eV photon should be able to supply a voltage of 2V. Voc larger than Eg has been observed in a laminated device as discussed in Chapter 6. the maximum Voc can be written as: Voc ≤ |LU M OA −HOM OD |+∆LU M O+∆HOM O (2. In D/A cells32 . it has been observed for many organic solar cells that a higher workfunction difference of the electrodes ∆Wf can also increase Voc . in both organic and inorganic semiconductors photoexcited electrons can decrease their potential energy very quickly using the high density of allowed states in the CB until they reach the lowest unoccupied molecular orbital (LUMO). we also find higher Voc for higher bandgap materials. however. Note that ∆LU M O+∆HOM O has to be greater than the exciton binding energy to split the excitons in D/A devices. As discussed earlier a high Voc may be favored by a lower reverse saturation current I0 due to the frequently low nonradiative recombination in these more fluorescent materials. charge carrier diffusion lengths and recombination velocities [97]. In case of ∆Wf > EgDA the built in field due to the offset of LUMOs (∆LU M O) and HOMOs (∆HOM O) between the two components has to be large enough to provide the extra driving force. Since the lost potential energy is usually converted into phonons i. Thus. Hence. maximum voltage. Practice confirms that indeed inorganic solar cells made of higher bandgap materials develop higher Voc and the maximum achieved Voc has never exceeded the width of the bandgap. a larger ∆Wf results in either a stronger tilting of the semiconductor band for voltages < Voc or a stronger bend bending near the electrode semiconductor interface if there are enough charges to cancel the field31 within a fraction of the bulk thickness. The above formula can be written as Voc ≤ EgD + EgA − |LU M OA − HOM OD | (2.1V) to about 0. Ideally the potential difference between the ground state and the excited state of an electron after photo-excitation can be regarded as a driving force i.24 reached if the charge carrier density approaches critical values which would typically only occur at unrealistic high excitation intensities.2.

Rsh is clearly larger than the DC resistor value of the ideal diode D.2. The EQE can be derived from the spectral response considering that the energy of a photon Ep = hc/λ with h being Planck’s constant.any current higher than Isc current originates in the external voltage source. However.33) 2. 33 Even Note that the EQE gives higher values for shorter wavelengths when compared with the SR spectrum. both Voc and Isc are high if Rs is small and Rsh is high. SR(λ) = 25 Jsc (λ) Φ(λ) (2. 36 Note that hc/q = 1240W nmA−1 . the bandgap is high and the conductivity is low. intensity measurements are an important application for photodiodes. the most interesting parameter of a solar cell.34) with R(λ) denoting the fraction of reflected light and T (λ) the fraction of the transmitted light. This is because the spectral response relates to the energy of photons whereas the EQE refers only to the number of particles. together with the power conversion efficiency η. The light current increases linearly with the illumination intensity. a large Isc is favored if the bandgap is small and the conductivity is high. Note that not all requirements to obtain a high Voc coincide with the requirements to obtain a high short circuit current. PHOTOVOLTAIC CHARACTERISATION Summary According to the described model a high Voc at a given light intensity is expected if: 1.2. Id : Isc = IL − Ish − Id (2.13) can be derived from Eq. This leads to the definition of the spectral response: if an external voltage is applied.4. 35 Light 34 The . (2. For example. I0 is low. the EQE represents a true measure for the photon to current conversion efficiency in contrast to SR. For the characterisation of solar cells a quantity that considers the actual fraction of the incident photons that can be converted into electrons in an external circuit is of higher interest: number of electrons in external circuit number of incident photons (2. hardly used but accurate term is: irradiance. Since the output current is determined by the number of electrons that can be “pumped” into the CB regardless of their energy.31) with Φ being the light intensity per illuminated area34 (W/m2 ) and Jsc representing the current density (A/m2 ) in short circuit mode.29) From the ECD in Fig. The shape of the IV curve is not too steep. The EQE can be converted into the internal quantum efficiency IQE if only the fraction of the actually absorbed photons are considered: IQE(λ) = EQE(λ) 1 − R(λ) − T (λ) (2. This quantity is frequently used to characterise the light sensitivity of photodiodes since the product SR · Jsc gives immediately the light intensity35 . 3. The reverse saturation current is often low if the PL efficiency is high. If we also consider the influence of the PL as discussed above we can write: IL −Ish Id Isc IR I0 /ΦP L nkTs IL = − (e /q − 1) Rs Rs 1 + Rsh 1 + Rsh (2.see Fig. c the speed of light and q the elementary charge using the following formula36 : EQE ≡ EQE(λ) = SR(λ) · hc qλ (2.4 Photocurrent The Short Circuit Current Isc The current Isc under short circuit condition (RL = 0 . number of split excitons/s) reduced by the current through the shunt Ish and the diode D. Isc has to be related to light intensity.2. only the photosensitivity can be amplified .19) by setting V ≡ 0. The spectral response (SR) thus gives the current in the external circuit per watt incident light (A/W ).12 can be seen that the short circuit current is simply the light current IL (photogenerated charge carriers/s i. In order to extract a quantity that describes how efficiently photons are converted into charges in the external circuit.32) EQE stands for external quantum efficiency and represents.4. The dark current can then only be canceled by the photocurrent at higher (Voc ) voltages.e. 2.30) This is the highest current that can be extracted from a solar cell33 .

free electrons have to be generated close to the negative electrode for charge collection. 2. Thus the width of the active layer is determined by the distance electrons can travel Le in the specific material i.15 a). In addition. SOLAR CELLS .15 b). Only if the active region extends across the entire bulk i.14. Thus. Often the mobility of the hole is considerably higher than the mobility of the electron. CHAPTER 2.see also Section 2.2.26 The IQE is very helpful and frequently used in inorganic devices to investigate the physical processes in the semiconductor material. The layer that absorbs light before it reaches the active region37 is the actual optical filter.GENERAL 1.2. only light absorbed in the active region can contribute to the photocurrent. Thus.2.14: In organic devices very often only a thin layer (active layer) near back electrode is electronically active (contributes to the photocurrent). The Optical Filter Effect The region in the semiconductor that allows photoexcited excitons to dissociate and the separated charges to reach their respective electrons is often called the ”active region” . Figure 2. b) exciton dissociation occurs only near the (here negative. the bulk is thin enough. excitons that are generated outside the band bending area have to diffuse into this region to supply charges. Figure from [63].see Fig. If exciton splitting occurs due to a strong local field (band bending). the electron mobility . Exciton dissociation occurs only near one electrode interface. a considerable fraction of the photogenerated excitons would dissociate too far away from the negative electrode. Depending on the thickness and absorption coefficient a significant fraction of photons can be absorbed in the semiconductor bulk (optical filter) before this region is reached.e.15: Two scenarios that can lead to the formation of an optical filter: a) exciton dissociation occurs throughout the entire bulk but one charge carrier (here the electron) can only travel a very short distance before it recombines. This makes the interpretation of IQE spectra in organic cells only more speculative rather than simpler and therefore generally less useful at the present time.see Fig.see Fig. oxygen molecules) can act as such .e. Exciton dissociation throughout the entire bulk but small mobility of one charge carrier. Al) electrode due to a local high field. Figure 2. interference effects would have to be considered because of the very thin films compared to inorganic cells. organic devices have typically large regions that act only as optical filter.15 shows two scenarios for the formation of an active region smaller than the bulk which can lead to an optical filter effect: 37 Electrode/polymer interfaces often act as dissociation site but also polymer/polymer interfaces or impurities (dopants. In contrast to inorganic solar cells where the complete bulk absorbs light and contributes to the photocurrent.2. Even if the excitons have large diffusion lengths.2. Thus a small exciton diffusion length determines the width of the active layer . Although small hole and electron mobilities can still become limiting factors it is usually the exciton diffusion length that determines the width of the active layer in this case . Fig. The EQE is therefore the preferred quantity in the devices discussed in this thesis.2.provided the charge carrier traveling distance is sufficiently high. most of the disso- .

However.3.17: Effect of Rsh (ranging from 101 to 104 Ω) on the shape of the IV characteristics. Voc approaches zero and the fillfactor reaches its theoretical minimum of 0.Voc ) of the IV curves remains virtually unchanged and the current shows no significant field dependence for negative bias.if it is sufficiently transparent.4.4. Note that a symbatic spectrum can be obtained if the device is illuminated through the other contact . Figure 2.2. Fig. A large Rsh can reduce both the fillfactor and Voc but not Isc .2. I0 .2 we describe a method to estimate how thick a semiconducting layer can be before it shows the conversion losses due to the optical filter effect.16: Effect of Rsh (ranging from 104 to 107 Ω) on the shape of the IV characteristic. (2. (2. Voc is virtually zero and Rsh extends its “linearisation effect” into the first quadrant.25 (Ohmic .16 shows how the IV characteristic changes if the shunt resistor varies between 104 and 107 Ω assuming the shown38 values for Rs .probably due to the higher PL efficiency in the latter .19) and Eq. The values for Rsh and Rs have been estimated from the IV curve data as described in Section 2. clearly smaller Rsh values have enormous bad effects on the IV curve. 27 the electrodes before they undergo intense recombination .29) but also makes it much easier to distinguish the effects of the two resistors in the IV characteristics.2. IL . The slope (field dependence) in the third quadrant increases considerably. The dashed curve is the dark characteristic at 106 Ω Figure 2. In addition Rsh is typically more than 3 orders of magnitude larger than Rs .4.even though the devices are about 1000 times thicker. Such a spectrum is termed antibatic spectrum in contrast to the symbatic spectrum that shows no offset between absorption and EQE features [120].linear)39 38 The currents given in this Section are related to an area of 1mm2 . The photocurrent (Isc − I0 ) is too small to be visible on this scale. In Section 3. 39 Lower values than the one for ohmic behaviour may be 2. In addition the reverse saturation current I0 is typically larger than in organic devices .as discussed in Section 2. If Rsh > 106 Ω the shape (FF.4.5 Interpretation of IV Characteristics Inorganic Solar Cells The higher charge carrier mobility in inorganic semiconductors leads to sufficiently long diffusion lengths which allow most charge carriers to reach .which is sufficiently close to our experimental conditions (A≈ 4mm2 ) to get a “feel” for the absolute numbers. This Rs not only leads to a negligible term 1 + Rsh in Eq. PHOTOVOLTAIC CHARACTERISATION ciated excitons can generate a photocurrent so that the EQE essentially follows the absorption features. Due to short exciton diffusion lengths most organic cells have a bulk thickness that is considerably larger than the active layer (to avoid shorts). Thus these EQE spectra show a strong filter effect resulting in EQE maxima at wavelengths where the optical density is low and EQE minima where the OD is too high to allow many photons to reach the active region.

3 V in the first quadrant.18: Effect of Rs on the shape of the IV characteristics assuming the same values for an inorganic solar cell as above with Rsh = 106 Ω. current (I = R−1 U ).Organic Solar Cells fect the IV curve. if the two resistors have similar values the IV curve is dominated by their ohmic characteristics (the inverse slope in the 1 quadrant is equal to Rs ) and the ideality factor or voltage dependence of D may then have little influence. Note that in contrast to the effect of Rsh .Fig.GENERAL thereby decreasing the fillfactor to its minimum.12. tration lead to considerably smaller currents in most 40 Note that Voc is virtually zero. Figure 2. In fact.22) quadrant remains unchanged the slope in the first a similar Voc is expected.2. Figure 2. the short circuit current is not affected since the current through the shunt can be neglected if Rs << Rsh .28 very quickly. Fig.18 shows what happens if the series resistor devices is about as big as the dark current in silincreases assuming the same parameters as above icon devices while the dark current is still about and Rsh = 106 Ω: While the slope in the third 1000 times smaller so that according to Eq. both the dark and light for Rsh > 104 Ω that have been shown in Fig.2.16 currents are about 1000 times smaller (I0 ≈ 1pA and Isc ≈ 1nA) so that the light current in these do look exactly the same on this scale.17 shows how even smaller Rsh values af. SOLAR CELLS . We note that the dark current may be related to quadrant starts to decrease considerably when Rs the fraction of radiative recombination so that high reaches the same order of magnitude as Rsh . CHAPTER 2. Rs reduces FF and the slope in the first quadrant when it reaches the same order of magnitude as Rsh . Fig. Under the same as for the inorganic device in Fig.as observed in many organic semiconductors .the short circuit current can decrease but the open circuit voltage can not be affected at all . (2.since there is no current through Rs at Voc . Rs can reduce the short circuit current but not Voc .2. obtained if another diode (with reverse orientation) is formed (=counterdiode).is expected to give a low dark current. This ganic solar cell.2.19: Effect of Rsh (ranging from 107 to 1010 Ω) on the shape of the IV characteristics asNote that the reverse slope around 0V or in the suming typical dark and light currents for an orthird quadrant is equal to the value of Rsh . The effect of the large Rs even extends into the 3 quadrant PL-efficiency . 40 The dark current Id scales with I0 . The slope in the third quadrant increases even further and the “linearisation effect” Lower mobilities and smaller charge carrier concenof Rsh extends to about 0.2. The slope in the third quadrant is still dominated by Rsh . However. The effect of Rsh is qualitatively can easily be seen from the ECD . Since Rs is much smaller than Rsh it is only Rsh that determines the slope R−1 .16 but reverse bias the diode D is blocking and R = Rsh + 1000 times larger resistor values are required to give Rs only determine the voltage dependence of the the same shape. Note that the values organic solar cells. Moreover.

Again. As a guideline it may be said that the light current should be at least 3 orders of magnitude higher than the dark current before the fillfactor can be regarded as light intensity independent and used for direct comparisons.can vary significantly from pixel to pixel or even over time41 .5) conditions to allow direct comparisons.like e.2.20: Effect of Rsh (ranging from 103 to 106 Ω) on the shape of the IV characteristics assuming typical dark and light currents for an organic solar cell. However.17 but 1000 times larger resistor values are required to give the same shape. . Thus. Rs decreases FF and the slope in the first quadrant if it reaches the same order of magnitude as Rsh . the output current is 1000 times smaller and the resistor value 1000 times higher .2. PHOTOVOLTAIC CHARACTERISATION This is discussed in more detail in the Section about Voc (Section 2. IV curves should only be compared if light intensities are similar and FF and Voc values (mW/cm2 ) should be quoted together with light intensity numbers. Note that the IV curve for values larger than 106 Ω .21 shows the effect of Rs with currents typical for organic cells and assuming Rsh = 109 Ω.4.2. Even lower values for Rsh are considered in Fig.compare with Figure 2. IV curves of inorganic solar cells are usually measured under standard (AM1.20. In order to show the influence of only Rsh we assumed an unrealistic low value (1Ω) for Rs .21: Effect of Rs on the shape of the IV characteristics assuming the same values for an organic solar cell as above with Rsh = 109 Ω.2. Rs can decrease the short circuit current but not Voc .22 shows that .g.2. device type and measurement condition. Both figures show qualitatively the same effect of a decreasing Rsh : a steeper slope in the third quadrant and eventually reduction of Voc .19 look exactly the same on this scale. Fig.2. an Rsh as low as 105 Ω in the inorganic cell (Fig. although the Isc /Id ratio is exactly the same the Rsh value has to be 1000 times higher to cause the same effect.apart from high Rs and/or Rsh values .3). Fig.16) is necessary to decrease Voc by 50% whereas a value of 108 Ω can show the same effect in the organic device (Fig. IV curves .2.18. In fact. Otherwise. Fig. The slope in the third quadrant is still dominated by Rsh . all curves shown in Fig. 29 Figure 2. The problem with many organic solar cells is that they may not be stable enough under such high light intensities and that depending on the material.19 shows the effect of a decreasing Rsh on the IV curve using current values typical for organic solar cells. The effect of Rsh is qualitatively the same as for the inorganic device in Fig. 41 We note that unlike many IV curves the EQE spectra are usually very well reproducible. For example.2. we observe the same effects as in for the inorganic device upon increasing: decrease of the slope in the first quadrant and reduction of Isc . However.2. Fig.a low light current can cause both a low FF and a low Voc . the potential fill factor and Voc for realistic conditions may be underestimated considerably.4.2.particularly of very inefficient devices .19).

Provided there are no traps. the dielectric constant of the semiconductor. d is the semiconductor thickness.see Fig. Figure 2. values are below 10−4 cm2 V −1 s−1 . This second term is known as 42 Typical Once the applied electrical field is strong enough to release the charge carriers from the traps. The SCLC depends then on both the density of traps and their distribution inside the bandgap. SOLAR CELLS .see Fig.2.23: Schematic log-log plot showing the shape of the current-voltage characteristics for space charge limited current with deep and shallow traps.22: Effect of Isc on the IV characteristics of an organic solar cell. the total current (I) through a semiconductor considering space charge formation as a function of voltage (V) can be expressed as [138]: I = qp0 µp V V2 9 + µp 3 d 8 d SCLC (2. The transition from the ohmic to the space charge conduction here is very smooth.GENERAL Figure 2.30 CHAPTER 2.23. It can be seen that low Isc values can result in a considerably smaller fillfactor and also lower Voc values compared to higher Isc numbers. The space charge is then composed of the mobile free carriers and those carriers that are thermally freed from the shallow traps. The trap free Mott and Gurney equation will still be applicable.e.4. This phenomenon is termed space charge limited current (SCLC).23. As long as the density of the the thermally generated free carriers p0 is dominant the current follows Ohm’s law.2.36) 2.6 Space Charges & Traps The comparatively low charge carrier mobilities in most organic semiconductors42 in conjunction with the need to drive high currents through the device can lead to accumulation of charges in the semiconductor bulk. if the mobility µ is multiplied by a factor F defined as the proportion of free charge carriers. giving an effective mobility: µeff = F µ (2.35). increases the transit time of charges between the device electrodes. the current will return to obey the trap free Mott and Gurney equation . If the material contains shallow traps the residence time inside the traps is shorter than the charge carrier transit time. The transition from the ohmic conduction to the trap free condition occurs sharper than for shallow traps . If the material contains deep traps the residence time inside the traps is longer than the charge carrier transit time. (2. These excess charges build up a space charge that creates a field which reduces the throughput i.35) ohmic p0 is the density of the thermally generated free carriers. The dashed curve is the dark characteristic. the Mott and Gurney equation for trap free space charge limited currents. µ is the mobility. Otherwise the current increases with the square of the voltage according to the second term in Eq.

The thickness of the organic layer is typically between 40 and 200nm. is somewhat stronger absorbing when compared with the soluble derivative (Per(dye)).e. In Chapter 2 we have discussed this in context with the exciton diffusion range and the importance to reach dissociation sites.1.1.1 Introduction We have also calculated the percentage of absorbed photons α10 using α1/2 for a 10nm thin1 film.1: Device architecture of a single layer solar cell as it is used throughout this thesis. 31 . exciton diffusion length . 3.see Section 2. On the other hand. In any case. 2 For example: 50% of α 1/2 would have to be added if the absorption spectrum is approximated as being triangular. Interestingly the absorption coefficient of MEH-PPV is still twice as large as the other soluble polymers indicating it can arrange its aromatic system in a very compact fashion.1 gives insight into the absorption properties of some important organic semiconductors. it appears as if the differences between absorption properties of dyes and pigments do not differ tremendously. The (half peak) absorption coefficient α1/2 are listed together with the wavelength range that exceeds this value to give a “feel” for the spectral dependence of the absorption.3. however. Solar photons can enter the cell through the transparent contact (ITO) and create excitons upon absorption in the organic film. Figure 3.1.1 Optical Properties of Organic Photovoltaic Materials Table 3. In single layer devices 1 10nm were chosen since this is the typical thickness of the active layer i. pigment. The ITO and Al electrodes are sometimes replaced by other materials in the cited literature. Note that the percentage of absorbed photons in a 10nm film in the cited range would require integration of the absorption over the given range. In its simplest form.Chapter 3 Single Layer Devices 3. an organic solar cell consists of a single polymer layer sandwiched between two different electrode materials as depicted in Fig. polymer) is sandwiched between the transparent ITO covered quartz substrate and a sublimed Al layer. Not all photons will be absorbed in the thin film . PPV being the only precursor polymer in the list has clearly the highest absorption coefficient followed by its well known soluble derivative MEHPPV that has to accommodate also the bulky and optically non active side-chains in the film. The quoted number for α10 is therefore somewhat underestimated2 if it wants to be interpreted as absorption in the given range. Pigments also tend to have slightly broader absorption bands. The organic film (dye. This is actually expected because the absence of bulky side chains in the pigment should favour denser packing. the introduction of a metal like Cu in the phthalocyanine core can lead to an increase of the absorption. The metal-free phthalocyanine has virtually the same α1/2 regardless if eight bulky thio-alkyl chains are attached (HPc(dye)) or nothing at all (HPc(pig)) as in the insoluble version.some will be reflected by the back contact and have another chance to be absorbed in the film before they can leave the device again through the ITO layer. The perylene diimide pigments (Per(pig)).

placed in an optimal distance from the electrode using a second layer as optical spacer material [213] or the active layer can be made thick enough to allow a significant fraction of the visible light-wave to enter the device.004 0.3.1: Survey of the absorption properties of various organic semiconductors. material – PPV MEH-PPV CuPc(pig) [43] Per(pig) [133] Per(dye) CN-PPV MCP HPc(dye) HPc(pig) [68] α1/2 nm−1 0. such a dissociation site is typically the low workfunction electrode (Al)/semiconductor interface3 . the high workfunction electrode (ITO)/semiconductor interface can become active.013 0. The range parameter refers to the part of the absorption band that exceeds this value.006 0.see Chapter 2 and end of this Section.3 have been derived using Lambert Beer’s law to express the fraction of transmitted light T yielding: T A=1−e −αdr (3.particularly in multi-layer structures (considering coatings like the SiO2 . α10 is the percentage of absorbed photons using α1/2 in a 10nm film without and with (brackets) a fully reflective rear contact. interference with the reflected waves is not expected to cause dramatic changes of the exciton density in a 50nm single layer structure. We note that interference effects may be more important to consider in double layer. Ta2 O5 or MgF2 layer in silicon solar cells [162] may also be employed for organic cells.32 CHAPTER 3.3. In these structures the active layer can either be oxygen molecules that can also act as dissociation site do not seem to increase the active layer significantly . 4 Anti-reflection . α1/2 is half the value of the absorption peak in the visible.2: Percentage of absorbed photons in a film for various absorption coefficients 0.1) with α being the absorption coefficient (here α1/2 ) and r=1 for the transparent rear contact or r=2 for the fully reflective one. SINGLE LAYER DEVICES Table 3. interference and multiple reflections have been neglected above. However.005 < α < 0.006 0.2 and Fig.010 0. The layer can be half as thin if a reflective rear contact is used . Since the active layer is usually located next to the reflective electrode interface and not much more than 10nm thick with a refractive index of about 2. 3 Interestingly Figure 3.04nm−1 as a function of film thickness assuming a reflective rear contact.compare with Fig.005 < α < 0.017 0. More information on interference effects that may be used to optimise reflection4 and absorption properties in solar cells . blend or laminated device structures. It is helpful to know that the reciprocal of α gives the layer thickness that is necessary to absorb 63% (1-1/e) light. The absorbed fraction A as a function of layer thickness d in Fig.3: Percentage of absorbed photons in a film for various absorption coefficients 0.004 range nm 320-480 440-560 570-740 450-590 450-560 410-550 370-510 610-750 590-??? α10 % 16(29) 12(23) 11(21) 10(18) 7(13) 6(11) 6(11) 4(8) 4(8) Figure 3.04nm−1 as a function of film thickness.3.3.3.2 and Fig.007 0. Only if the organic material can be regarded as good electron acceptor material.012 0.

NO2 .even with a reflective rear contact.00001nm−1 at 1000nm to 0.that often exposure to ambient air of only a few seconds is required6 to obtain a photoresponse in hole transport materials i.45V many years ago. Dopants with even greater electronegativity than oxygen.e. 282].0004nm−1 at 600nm 5 . [281.8% of light at 600nm . PTV.1. PEDOT in conjunction with PPV here is heavily doped and therefore treated as a conductive transparent electrode film rather than a second semiconductive layer. these cells have still some way to go to become practically interesting since e.2 and Fig. The inability to absorb more photons in the 10nm layer represents a main loss mechanism in single and double layer organic solar cells. α Si varies over 3 orders of magnitude ranging from 0. organic semiconductors have very high absorption coefficients compared to their inorganic counterparts so that the extremely thin active layer of 10nm can still absorb more than 20-30% of the light in the absorption range.3. the peak absorption coefficient of pure silicon at room temperature is 0.they could not quite conquer the field of commercial solar cells. MEH-PPV. The recently published doped pentacene cell in Ref.independently of the thickness of their active layer i. 70-80% of the photons in the absorption range are lost for the photovoltaic conversion process if the active layer is only 10nm. 7 Good electron donors have a high lying LUMO = small electron affinity. According to Fig. Blends with a non conjugated polymer binder molecules are treated as single layer here. Thus.2 Performance of Single Layer Devices A survey of the photovoltaic performance of various single layer materials is given in Table 3. Indeed. [224] shows a solar power conversion efficiency of 2. However.02nm−1 . It has been found . phthalocyanines. INTRODUCTION also the about 100nm ITO film).5% [164]. . 6 Typically there are marked concentration effects: Exposure over many hours might be required before the photocurrent reaches its maximum which is then followed by slow decrease.4.[29] and in Ref. We note that 80% of the solar radiation on earth is distributed over wavelengths ranging from 400 to 1200nm .7trinitrofluorenone result in an even greater enhancement of the photoresponse[43] in such hole transporters. o-chloranil or 2.2) In other words: the easier they bind oxygen the higher the photocurrent.with some effort from (acidic) solution [283] . It was found that the photovoltaic efficiency of metal containing phthalocyanines follows the following order [231]: P cM g > P cBe > P cF e > P cCu > P cCo (3. 33 As can be seen from Table 3. Voc and fill factors.1. polymers or liquid crystals which are soluble in common and less harmful solvents.especially for pigments can be found on page 295 in Ref. for example. 5 In fact. most currently available organic semiconductors loose more than 60% of the (mainly) low energy solar photons because their bandgap is too large .see Table 8. 231]. illumination through the poorly transparent indium electrode was required or sub-linear intensity dependence of the photocurrent decreases the actually achieved solar power conversion efficiency down to 2. More information on coating techniques .2 cells using pigments have reached the highest EQE.g.3 even with the best absorbing materials (α1/2 ≈ 0. However. 3. Even though some can now be processed . [268] as well as references therein.since oxygen acts here as dopant.1 and 0. although they have been investigated intensely in the past three decades and they have replaced many inorganic photo-receptors in xerograpy . good electron donors7 . Note that in phthalocyanines. cheaper and less harmful using dyes. Thus. the development of new low bandgap materials is highly desirable.3.1.can be found in Ref. It can be seen that EQE values of 62 and 73% have been achieved with voltages of 1.see Table 3. This relates to materials like PPV. For example.01nm−1 at 400nm [98]. it is found that the influence of the central metal ions on the photovoltaic performance strictly follows their oxygen binding ability . film formation is still considerably simpler.1) and reflective rear contacts. merocyanines and porphyrins.and we can confirm this from our own experience .or dispersed in binder materials [164] and thereby circumvent thermal sublimation.e. such as halogens.3.3. a 10nm Si layer would only absorb 0. exciton diffusion range. It is well known that molecular substances can have a ”doping” effect on organic semiconductors [268.4%.2. Fortunately.3 Doping of Organic Photovoltaic Materials 3.

7 0.32 1. EQE % 73 62 ?? 33 14 3 1 0. layer are often found. Hence.g.our single layer device samples to air for at least 3 days. SINGLE LAYER DEVICES Table 3. Although exciton dissociation should there.02 FF % 33 33 ?? 25 39 34 20 25 23 25 25 25 27 25 25 25 25 25 25 25 range nm 400-800(?) 400-800(?) 400-800(?) 400-800(?) 400-800(?) 400-800(?) 400-500 400-700(?) 400-500 400-800(?) 400-750 400-560 400-570 400-600 400-570 400-750 400-650 400-650+ 400-820 400-1000 device – In/HPc+polymer/NESA Al/HPc+polymer/SnO2 doped pentacene Al/merocyanine/Ag Al/CuPc/Au Al/merocyanine/Au PPV Cr/chlorophyll-a/Hg Al/PPV+PEDOT/ITO Al/MgPc/Au Al/THPF/ITO Al/LPPPT/ITO Al/CN-MEH-PPV/ITO Al/Per/ITO Al/MEH-PPV/ITO Al/PTV/ITO Al/P3HT/ITO In/HPc/Au Al/HPc/ITO Al/PIF/ITO comments – pigment in polymer [43] pigment in polymer [43] pigment [224] pigment [43] pigment [43] pigment [43] polymer [168] dye [43] polymer [8] pigment [43] polymer polymer polymer [114] dye polymer [196] polymer polymer [63] pigment [68] liquid crystal polymer year 1981 1979 2000 1978 1981 1978 1994 1975 1998 1974 2000 2000 1997 2000 2000 2000 2001(?) 1978 2000 2000 On the other hand.5 0.our results. In the following we VB.007 0. different concentrations of water & iazine.2: Survey of important performance parameters of various organic single layer solar cells.5 0.015 0. These devices A high photocurrent requires both the generation were structured like Al8 /dye/Ag and reached quanof a free electron in the CB and a free hole in the tum efficiencies of up to 16%.05 0.6 1. molecas they have been identified in Ref.4 Desired Properties of Organic Photovoltaic Materials A very interesting review of photovoltaic. The list ular doping can be seen as essentially different from is followed and complemented by comments found D/A blend (or laminated) structures since the latin other studies. We note that the adsorption of water fore occur throughout the whole film.2 0.1 0. [204].due to different air exposure times of our In general.while the other one is trapped in the dopant within the inactive layer. NH3 or phenoth. 3.07 0.0 0.should not have a significant effect on rent.and not only a layer sentially planar and free form steric crowding as thin as the exciton diffusion length (≈ 10nm) 8 A AlO O layer on top of Al was considered.1 0. Consequently. ter allow both charge carriers to be free to travel to the electrodes through their percolation paths.7 0. organic electron transporters the EQE in single layer devices we have exposed all can be doped effectively with H2 .85 0.15 < 0.2 0. optical filter molecules also affects the photovoltaic performance effects that indicate a thick electronically passive [50]. We consider therefore that excitons can dissociate at e. The abbreviation for materials are listed in Chapter 1.15 0.1 ?? 1. the entire layer . oxygen .34 CHAPTER 3. fluorescence and redox properties of 60 semiconducting molecules .7 0.03 0.1 0. Best parameters parameters are printed bold while italic print marks devices discussed later in this Chapter.005 Voc V 0.7 0. In that way they can both meet after passing list the desired properties for photovoltaic molecules through the external circuit.02 0. . 2 3 can become active. For more information on oxygen 9 Chromophoric core = the pi conjugated part of the doping see Chapter 2.19 0.45 1.09 0.most of which are (soluble) merocyanine derivatives .samples . Since the presence of oxygen appears to enhance molecule is responsible for the optical properties in the visible and near IR range.is given in Ref. This would locally improve the charge transport properties but does not add to the photocurrent. The chromophoric9 core of the dye must be esOnly then. the photocurrent is much more affected by the presence of dopants than the dark cur.08 0.37 < 0.1.1 0. [204]. oxygen molecules and increase the concentration of one type of free charge carrier . 1.

importance in e. complemented and further elaborated as can be seen from the following list of comments: ad 1): The finding that a flat. SCE for efficient want to emphasise that exciton mobility looses its electron injection through the Al2 O3 barrier. 245]. further insight into a material’s photovoltaic properties. we also more negative than -1. 12 Rigid molecules are preferred to flexible structures since they have lower probability for those internal conversions that dissipate the energy carrying species before transport can occur. For included a claim in his patent regarding the use of example. blend structures . High PL of about the isolated molecule and its ability to dissithe isolated molecules in e. 154. Dyes should not be subject to protonation or deprotonation by ambient materials in the PH. a low lying LUMO level. tween photo-conductive molecules to achieve good performance. Dyes having central conjugated chains and the ruthenium dye in electrochemical photovoltaic steric crowding are to be avoided since such cells to the TiO2 surface [182].g.reduce the time an exciton has to reach a dissociasorption band but up to 100%11 on the short tion site.g. predisposing it to photo-isomerisation or internal conversion. The performance of photovoltaic dyes can be correlated with their chemical parameters. Me.for excitons are present. 7. INTRODUCTION effects which tend to twist the dye. rigid12 . this rocyanines on the other hand pack in deck of would indicate that other .3V vs. In the configuration of Al/Al2 O3 /dye/Ag. The effect of PL efficiency in the solid state on the reverse saturation current and the theoretically 10 Chromophore = coloured molecule 11 Photovoltaic quantum efficiency numbers in Ref.tron acceptor and transport materials. fluid solutions pate excitation energy through e. [204] consider transmission and reflection losses and are more likely identical with the internal quantum efficiency (IQE) .1.inter-planar distance should be short (≈ 3. 3. PL studies of organic films can provide therefore preferable. Low bandgap materials seem to have difficulties to maintain a large negative reduction potential i.g.in particular high exciton mobility . 2.which gives higher numbers . aromatic molecule favors close packing and good photovoltaic properties . The quadrupolar charge distribution of e. Small groups projecting out of the plane of the chromophore10 play a critical role in “indexing” the packing of adjacent molecules to obtain the desired aggregate structure. These fast decay channels tovoltaic quantum efficiency in their main ab. We note that close packing also in4.has been recognized in other publications [268. instead of sensitisation by electron injection. We are not aware that any of these early findings have been proven wrong but many have been confirmed.g. intra-molecular signal the effect of less torsional stress and is vibrations. However. synthesise. the adhesion of the dye to the Al surface is critical and can be analyzed by studying the chromatography of the dyes on the Al substrates. In Ref.e.over longer distances. ad 3): We note that COOH groups play indeed an range from 3 to 10. This is discussed in Chapter 2.Tang actually should be chemisorbed at the Al surface. The reduction potentials of the dues should be creases the absorption coefficient. The importance of this point may become obviThe electron rich end of the dye molecule ous by referring to the fact that C. materials can readily degrade excitons to heat ad 6): While PL in solution provides information by torsional internal conversion.decay channels cards or stacks showing only up to 15% pho. J-aggregate structure which seems tons is relatively long and allows excitons to diffuse to favour mobility of low energy excitons.3.faster .5˚) beA tum yields. 35 8.rials in solar cells. The dye must be a stable material and a strong absorber.rather than the usually used EQE.e. This would mean that low bandgap electron acceptors are harder to find i. important role in bonding sensitiser molecules like 6. 267. High PL in the solid state has been found to be squaric acid dyes may pack into a “brick. .e. [154] it is stated that the most always in much higher photovoltaic quan. wavelength edge of the absorption band. dyes with COOH groups have a much flat large aromatic molecules as photo-active matestronger bond to the Al surface and result al.desirable since it means that the life time of excistone” i.so that even spheric molecules like C60 can act as excellent elec5. If there is little or no PL.

The unique performance of this polymer is likely related to a very low concentration of electronically active traps such as structural defects or impurities [96] and the high charge carrier mobility (up to 10−3 cm2 /V s for holes at room-temperature) [128].36 achievable open circuit voltage is discussed in more detail in Chapter 2 and Ref. Indeed by replacing every other phenylene unit with a thiophene ring the structure of LPPPT . Hence.g. planar structure which allows only minimal changes of the polymer geometry during the transition from the ground to the excited state i. For example the bandgap of PPV could be lowered by about 0. higher open circuit voltages and sometimes higher photocurrents can be obtained using electrodes like e.5 thienylene units. [234].e. the bandgap of this polymer with 2.as discussed in Chapter 2. This results in a reduced π − π ∗ transition energy between the benzoid ground state and the quinoid excited state.the low bandgap version of LPPP could be obtained. 3.4-C6H4-n-C10H21 and n=20-25. photoexcitation [160]. SINGLE LAYER DEVICES proven to give more reliable and reproducible results than most other materials under laboratory conditions. then they are more suitable as electron donor (D) materials rather than electron acceptors (A) . 137]. R=-1.5eV by replacing every benzene unit with a 2. CHAPTER 3. A well established method to shift the absorption onset towards lower energies is the replacement of benzene by electron rich hetero-arylene moieties such as 2.see Fig.g.when used in combination with another semiconducting component to form a D/A system.4 . both electrodes have been It has been found that a poly(para-phenylene)-type ladder polymer exhibits many interesting properties such as high PL not only in solution (90%) but also in the film (50%) enabling the fabrication of efficient organic multi-colour LEDs [246] and even optically pumped blue solid state lasers [279].45eV compared to LPPP . The bandgap in solution was lowered by 0. In the following Sections of this Chapter we have always used the same combination of electrodes (ITO and Al) to decrease the number of changing parameters. The other possibility using a low band gap electron acceptor with high bandgap electron donor gives indeed only very low photocurrents (e.5 thienylene unit to give poly-thienylene vinylene (PTV) [275.3. However. It is ladder poly(para-phenylene thienylene) or short: LPPPT. In fact the combination phthalocyanine & perylene representing a system with a low bandgap electron donor and a high bandgap electron acceptor works very well. We note that planarity of aromatic systems is also suspected to favour charge and energy (exci- . the conclusion is that a low bandgap material in a D/A solar cell is likely to act as electron donor and thus require a partner with high electron affinity to enable charge transfer i. The reason for this effect is the lower aromatic resonance energy of the thiophene building block when compared to that of benzene.1.4: Chemical structure of the polymer used in this Chapter. ad 8): If low bandgap materials tend to have higher lying LUMO levels. Moreover. 3.7eV is too high to absorb a large fraction of the solar spectrum. Both polymers exhibit a dominating 0-0 transition and an extraordinary small Stoke’s shift of 35meV (LPPP) and 17meV (LPPPT) [74].5 Electrodes The workfunction and adsorption properties of electrodes have a clear effect on the performance of organic solar cells. Au and Ca instead of ITO and Al.which means a redshift by about 85nm. HPc+PIF in Chapter 5 or MEH-PPV+PIF [202]). The bridge between the phenylene-rings enforce a planar structure despite the proximity of the thiophene and phenylene units.e. The latter (small stokes shift) is a consequence of the geometrically fixed. The latter has to do with band bending due to equilibration of the workfunctions with the Fermi-levels of the semiconductors upon contact . However. exciton splitting at their interface.2 LPPPT Figure 3.

Results and Discussion Absorption and EQE Spectra Fig.3. 15. The suppression of the rotational degree of freedom between adjacent benzene rings in the latter results in a particularly small Stoke’s shift. Only if the bulk is thin enough.2.which we were exploiting by studying thickness effects in this special material.see Chapter 2.can 13 but Experimental The chemical structure of the polymer used in this Section is shown in Fig. More details on the preparation of substrates and the EQE/IV measurements can be found in Chapter 10. most of the dissociated excitons can generate a photocurrent so that the EQE essentially follows the absorption features. LPPPT films were prepared by spin coating from chloroform solution onto ITO covered quartz substrates (12x12mm) after filtering (0. • The EQE of the thinner device reaches values that are about 5 times larger than those for the thick one.51eV) not the shape of the EQE peak . The peak shift13 can be quantitatively explained with the so called optical filter effect[120]. 20 and 30 mg/ml have been prepared and spincoated at 1600 and 3200rpm. strong antibatic behaviour. The energetic spacing of 190eV is constant and can be attributed to the dominant vibronic (C-C) mode that has been observed in the absorption of films of other higher ordered polymers such as stretch oriented PPV [108] or the methyl substituted ladder type polymer comprising only benzene rings [197. The peaks in the photoresponse of the thick cell occur not exactly at wavelengths with absorption maxima.see also Chapter 2. LPPPT has been provided by and synthesised at the MPI-Mainz.32eV) and 495nm (2.5 shows the EQE spectrum of a thick and a very thin LPPPT device when illuminated through the ITO contact as well as the linear absorption curve. however.3. 37 and a shoulder at 460nm (2. 197]. there are some differences: • The EQE of the thinner device is almost “in phase” with the absorption while the thick one shows a clear red shift by 20nm. The very thick (205nm) device shows not only smaller absolute numbers but also a distinct offset to the absorption i. This has been observed in organic solar cells before and is sometimes termed as antibatic behaviour [120] . 239. 198]. In order to obtain a wide range of different film thicknesses.5: Features of the EQE spectrum of a thin (33nm) LPPPT device do almost coincide with the absorption features. high photo . • The EQE of the thicker device increases while The EQE of the thin device decreases in the shorter wavelength region.4. The well resolved vibronic structure allows better attribution of EQE features to absorption features . solutions with concentrations of 10.70eV). Figure 3.e. The EQE spectra of both the thick and the thin device show roughly the same features (the vibronic peaks 0-0 and 0-1). LPPPT ton) transport as discussed in Chapter 2 and below.and electro-luminescence quantum yields but also in a clearly resolved vibronic structure in both PL and absorption [238.3. Details of the synthesis can be found in Ref [74]. • The second EQE peak (at 500nm) is virtually as high as the first which is not the case for the thick device. In the thick device only light which is absorbed less for example in the onset or the minimum between the vibronic peaks in the absorption spectrum .45µm). In fact the second photoresponse maximum coincides with the absorption minimum between the two dominant vibronic peaks. This seems to be approached for the 30nm thin device since there is only a small peak offset left. The absorption spectrum shows two dominant peaks at 535nm (2.

4.3. that can be used in single layer devices without losses due to the optical filter effect. The optical density (OD) at this position together with the OD at the absorption peak (Fig.8) cells.7) 105 V cm−2 ).5) and the film thickness d0 of the measured EQE spectrum can then be taken to get and estimate for the optimal thickness dopt. we cannot be sure that the exThe large error of 30nm or 45% for OD(λe ) is citon diffusion length is as long as 30nm but at least due to the steep slope in the absorption onset and we know that the absorbed light is most efficiently the relatively broad peak in the EQE spectrum of used in a device with about 30nm thickness. Hence. using: OD(λe ) · d0 OD(λ0 ) CHAPTER 3. OD(λ0 ) = We also need to consider the high electric field 0.g.3) This equation is solely based on Lambert Beer’s Law: I(λ) (3. OD(λ0 ) = port.168 and 205nm. 2 Generally. The ∆d0 2 (3. On the other hand.59.0696±45%.93.5V/30nm≈ 2 · dopt = 65 ± 30nm (3. In any case. the large value for dopt.6) actual value might be smaller if e.127. = (3.9) dopt. the error of the input variables is: ∆dopt dopt 2 (3. This would yield in our case: dopt = (36 + 24)/2 = 30nm (3. We obtain a considerably smaller error and value length in LPPPT is not surprising knowing that plafor dopt using the data for the thinnest investigated nar structures are expected to favour exciton transdevice (10nm): OD(λe ) = 0.4 15 oxygen can act as such in electron donors .163.5) Figure 3. SINGLE LAYER DEVICES This is a more reasonable result considering that we know already that dopt has to be (slightly) smaller than 33nm.201 ± 12%.117 (dashed line). Assuming the filter effect14 is the main factor that determines the offset between absorption and EQE peaks we can estimate the maximum thickness of the organic film dopt. may be regarded as an upper limit of the exciton diffusion range for this material. dopt. compared 0. there is an + d0 effect of other dissociation sites such as impurities or dopants15 in addition to the electrode-polymer Using the data for the 205nm thick device and esinterface. Given that we are trying to avoid the optical filter effect we can justify to use the lowest number within the error bars and calculate the mean in case of several results.67. 14 See = ∆OD(λe ) OD(λe ) 2 + ∆OD(λ0 ) OD(λ0 ) + Section 2. First we need to know the wavelength λe of the EQE peak in the onset of the antibatic absorption spectrum.4) = e−α(λ)d = 10−OD(λ) I0 (λ) (with Io being the incident and I the transmitted intensity) and the resulting simple relation: d ∝ OD The error of the optimal thickness ∆dopt.38 create excitons in the active region so that absorption maxima actually turn out as EQE minima and vice versa. a relatively large diffusion the thick device. timated errors (OD(λe ) = 0.6: The EQE of LPPPT at 536nm when illuminated through ITO decreases according to due to the layer thickness of 33. Here we obtain: to about 10nm for MEH-PPV devices makes this material an interesting candidate for use in D/A dopt = 30 ± 6nm(20%) (3.22 ± 2% and d0 = 205nm ± 5%) we obtain: across the thin layer due to the difference in workfunctions of the electrodes (0.22 ± 2% and d0 = 33nm ± 15%.

8). Even though the open circuit voltage is relatively small (106mV) this device shows a good rectification i.see Fig.7. Id here is the actual current measured at smaller series resistant. The figure clearly shows the trend to higher EQE for thinner devices as well as the dip at 117nm. The figure also shows the curve for EQE at 536nm obtained upon illumination through the semitransparent Al contact taking into account the transmission losses through the Al layer (Fig. We obtain then even higher values than with illumination through ITO.e. The dashed line illustrates a possible trend.e. IV Characteristics The IV characteristics in the dark and under monochromatic illumination (550nm) of the thickest device (205nm) is shown in Fig. Optical interference effects between the incident and the reflected beam from the Al rear contact may be super-imposed here. lie on top of each other at this scale.2.3. we can expect that the strong increase of EQE with thinner devices when illuminated through ITO will slow down considerably when reaching the trend line from the Al illumination curve. In fact. slight increase of the EQE for thinner devices is probably due to improved electric properties such as a 16 The Figure 3. This is expected since there should be no losses due to optical filtering16 .8: Optical transmission spectrum of the 15nm thick Al layer i. .3. Thus. The thickness at this point can be assumed to coincide with optimal layer thickness dopt .6. The antibatic behaviour of the 205nm thick device gradually changes into a more and more symbatic response by going to thinner cells.10.3. How the shape of the EQE spectra varies with layer thickness is shown in more detail in Fig. the shape.4V is about 15 times (=rectification ratio) higher than at -0.e.3. LPPPT 39 Figure 3. We plotted the experimental data of the EQE at 536nm versus thickness in Fig. The solid line in the circles is a fit with a polynom of 3rd order whereas the solid line in the squares is a linear fit. This also holds true for the short wavelength range (400-500nm). peak positions and absolute value of the EQE do not vary considerably when illuminated through Al17 . numerical simulation using the described model and taking into account interference effects showed that there is indeed a dip around this wavelength in the thickness dependence curve of the EQE for such a device [266]. The parameter Id should not be confused with the reverse saturation current Io which can only be obtained via numerical modeling of the IV curves.7: EQE versus film thickness when illuminated (λ=536nm) through ITO (circles) and Al contact (squares). Dark and light characteristics have virtually the same shape i. 17 The small variation seen in this figure may well be due to random errors in the measurement. electrode of the LPPPT devices discussed in this Chapter.4V. As expected in the described picture.3. Hence. diode behaviour: The current that can pass through the device at 0.3. Only the 117nm thick device seems to behave differently.9. this curve displays the maximum EQE that is achievable for a single layer device of this material using ITO/Al electrodes regardless of film thickness or which side of the cell is illuminated.

5(11) ber of recombinations should go up if the charges 67 77 126 31 150(530) 6. However.7-full circles).5) travel through less material.3: Survey of solar cell parameters obtained ducting material to travel through to reach the elecfrom the IV curves and the EQE (at 536nm) for trodes.20 shows that the shunt also increases for 205 15 106 10 – – thicker films when measured in the dark .see Fig.8(5. it is not obvious why the absolute num59 85 105 31 83(200) 7. the cuit voltage.3(9.0) However. . The conductivity in the see Fig. Figure 3.19.although there seems to be some consolidation for d>60nm. dark) in comparison to the thick device.5(11) Fig.3. Figure 3.3(8. While the increase of Rs can be qualitatively explained by the fact that charge carriers have simply more semiconTable 3. Although there might be a trend towards lower dark currents with increasing thickness and increasing Voc it is not completely conclusive.chosen scale . 32 132 28 33 16(17) 4.3. However.17 resembles more the characteristic of an dark seems to be high enough to reduce the open ohmic resistor (constant slope) with virtually no circuit voltage. In Chapter 10 we have stated that recombination Rsh Rs d EQE Voc Id losses somewhere between the dissociation site and nm 10−3 % mV pA MΩ MΩ the electrodes can be associated with a small Rs . Larger open circuit voltages appear to require a rectification and only a small (≈ 30mV) open cirshunt higher than about 80MΩ here. the rectification ratio also decreases un. correlation between dark current and Voc seems to With decreasing film thickness .3.emerge though if the data are plotted on a suitably til the IV characteristics of the thinnest device .3.5(10) accept that the number of recombinations per sec117 36 176 18 250(1400) – ond increased since the charges need more time to 122 52 120 38 200(500) 6.3.40 CHAPTER 3.see Fig.3. The highest EQE was obtained with a 93nm thick film.11 to Fig. Shape and ratio of the features do not vary significantly with film thickness.18.3. The steep slope reveals a significantly higher conductivity (under illumination and in the table also shows that Voc does not necessarily increase for even higher shunts. a 18 0V . SINGLE LAYER DEVICES curves with different film thickness is given in Table 3.9: EQE of the LPPPT films when illuminated through the semitransparent Al contact (not corrected for the Al transmission) for different thickness (same data as in Fig.10: IV characteristics of the 205nm thick LPPPT device.8) get out of the thicker semiconductor.see Fig.3. the distinct growth of the shunt is harder to understand. It would be easier to 93 64 141 26 170(700) 7. A very interesting result is that Rs does increase A survey of parameters obtained from the IV relatively slowly (11kΩ/nm) with film thickness 18 For the suspected origin of Id see Chapter 10 whereas the shunt increases more than 100 times faster (1. 185 26 99 27 250(600) 7.5MΩ/nm) .16. LPPPT films of varying thickness.

According to Eq.13: IV characteristics of the 117nm thick LPPPT device. A considerable decrease of Rs is required to allow both high voltages as well as high photocurrents and thus high power conversion efficiencies [206.3.12: IV characteristics of the 122nm thick LPPPT device. Such a large difference cannot be explained with the effect of the Rs /Rsh .21 we have plotted β versus film thickness.04 (60-200nm) to at least 1. In Chapter 2 we have seen that the ratio Rs /Rsh not only affects the fillfactor strongly but also the output current in a solar cell. According to Eq. In fact. The figure shows that the factor β is close to 1 for film thicknesses larger than about 60nm. In the preceding Section we have already seen that with decreasing thickness the EQE is indeed rising.3. The thickness dependence of Rs appears unchanged in the dark with absolute numbers about 30% larger than under illumination.3.24) the value for Rs /Rsh = 0. The increasing Rsh can be recognized immediately from the decreasing slope in the 3 quadrant (negative voltage range here) although there are exceptions in the expected order.18 before the Voc reaches a maximum value for the 117nm thin layer and not for the thickest device as it may be expected from the flat slope in Fig. For thinner films the reduction factor can increase from 1.3 for this device can only account for a reduction of Voc by about 30%. The graph illustrates clearly the distinct changes of slopes and open circuit voltages. Voc of the 32nm device is as small as 28mV whereas the highest voltage in LPPPT is 176mV (117nm). apart from a low shunt. . there In Fig. As we have seen in Fig.10) Figure 3.see Fig.3.19) the current through a solar cell is reduced by the factor β: β = 1 + Rs /Rsh (3. Figure 3.3. In order to see the effect of film thickness on the shape of the IV curves.which essentially works as a voltage divider. 214]. Rs is expected to decrease significantly with increasing light intensity simply because more charge carriers become available for the conduction process. (2.2.3 (30nm) in light. we have plotted the light IV curves in a single graph .11: IV characteristics of the 185nm thick LPPPT device.22. Although this does not affect the photocurrent dramatically it indicates that there is an additional effect for devices thinner than 60nm in LPPPT. LPPPT 41 Figure 3. (2. Hence.22.

Rs = the one for Rsh .14: IV characteristics of the 93nm thick LPPPT device.3.10 higher voltages19 .5V must be another not yet known reason to cause the (Rs ) whereas in the former the formalism considlow open circuit voltage in this device.42 CHAPTER 3. Note that the resistor values shown in this list are somewhat different from the values shown in Figure 3.1V to 0. This is because of both the steeper 2.4V using the fol.derived from the slope around 0.8 · 107 Ω and IL = voltages the (inverse) slope can only represent a upper limit of Rs .19) that could be We consider the numbers for Rs values which are applied for the region -0.17: IV characteristics of the 32nm thick LPPPT device. Figure 3. 1610pA .5V to be more aclowing fit parameters: curate than the value obtained from fitting the entire curve since Rs really starts to dominate only at 1. Figure 3.see Chapter 2. ered effects of n.3. resistors under illumination (0.15: IV characteristics of the 67nm thick Table 3. The results for the shunt Fig. times higher in the fit.3 · 107 Ω slope at higher voltages which results in stronger 4.8 · 107 Ω. Rs = 1. resistors in the dark: Rsh = 18 · 10 Ω. In the latter the numbers were derived LPPPT device. IL . (2. the comparison of these two methods indicates that the error for Rs is much higher than 7 3. SINGLE LAYER DEVICES Figure 3.16: IV characteristics of the 59nm thick LPPPT device. It also shows the fit with Eq. ideality factor n=1.1mW/cm2 ): 19 Unless the IV curve becomes clearly linear for higher Rsh = 8. directly by taking the slopes of the characteristics at 0V (Rsh ) and the highest voltages around 0. I0 . 2. reverse saturation current I0 = 10pA In any case.23 shows again the dark and light characterare fairly similar and the values for Rs are about 4 istics of the ITO/LPPPT/Al device (59nm thick).

18: Open circuit voltage and dark current seem to be “out of phase” when plotted versus film thickness on the shown scales.the internal quantum efficiency (IQE) was calculated.029% for the thickest (205)devices to 0.091pA obtained for the later discussed MEH-PPV. • The highest EQEs were achieved for the thinnest cells where the optical filter effect has virtually disappeared.3.2% [35. • The highest EQE is almost twice as high as the best i. Thus.012 = 0.135% for the thinnest (32nm) device. • Since LPPPT is a very large flat molecule but without any other obvious outstanding structural properties the high EQE in this material and the lack of an optical filter effect supports the suspicion that planar rigid structures favour exciton diffusion. • The EQE increases from 0.11) The value for I01 is close to the 0.if the polymer/Al interface has to be reached. it can not be ruled out here that I0 depends only or at least predominantly on the PL efficiency.25): Figure 3. • Thickness is an important parameter. dependence and voltage fluctuations of the source measure unit in the low mV range.e.19: Series and shunt resistor as obtained from the slopes in the IV curves under illumination for varying film thickness. 63] to obtain a value for the PL independent component of the reverse saturation current according to Eq.12pA (3. The optimal thickness can be estimated from the absorption spectrum and an antibatic EQE spectrum.the exciton diffusion length could be as high as 30nm in this material . (2. LPPPT 43 Figure 3.in other words . According to the discussion on the photoluminescence effects we can now take the reverse saturation current together with the PL efficiency of LPPPT of about 1. This would then indicate that the maximum performance of both materials LPPPT and MEH-PPV showing the used PL efficiency has been closely approached in these devices. Summary In this Section we have found for LPPPT that: . thinnest (10nm) MEH-PPV device. Since the entire 30nm layer contributes to the photocurrent . A more detailed study comprising a higher number of samples and systematic IV analysis is necessary to confirm this possibility and explain the observed variation of Voc with thickness. The higher EQE numbers of thinner devices would possibly disappear if the optical losses like the filter effect were considered or .2. • Both shunt and series resistor increase with increasing film thickness but Rsh grows more I01 = I0 · ΦP L = 12pA · 0. • Although the EQE increases with thinner devices without reaching a maximum in the observed range (30-200nm) the Voc appears to reach a maximum value at 117nm.

3. a fast decreasing Rsh in the dark. Single layer MEH-PPV cells have already been subject to photocurrent spectroscopy [120]. Shortly afterwards. This was the birth of MEH-PPV. Figure 3.3 MEH After the discovery of EL in conjugated polymers [34] the next step towards inexpensive device manufacturing was the design of a molecule with similar semi-conducting properties but even easier processibility. Hence. 10hrs20 .12pA) as for MEH-PPV (0. At that time PPV films could only be obtained via spincoating a precursor solution and a subsequent heat conversion step at about 200C for 5 to . 21 This technique could be used to fabricate not only computer or video screens but also photosensitive CCD arrays.even inkjet printing21 has already been reported [19] . Already in 1991 researchers in Santa Barbara (CA. • The PL efficiency multiplied with the reverse saturation current (12pA) gives approximately the same number for LPPPT (0.but also made it possible to investigate blends with other semiconductors such as C60 while maintaining the advantages of solution processing.1W/cm2 ) for an area of 4mm2 . [234]. This is consistent with a strong correlation between the IV curve i. The availability of a soluble semiconductor not only opened the way towards cheaper processing via numerous coating methods .20: Series and shunt resistor as obtained from the slopes in the IV curves in the dark and under illumination for varying film thickness.21: Current reduction factor β as a function of film thickness in LPPPT . 0. more similar Rs and Rsh values. Photo-induced electron transfer from MEH-PPV onto the fullerene C60 has been discovered and exploited in solar cells reaching EQE as high as 29% for monochromatic light [278]. These bulky side chains allow solvent molecules to penetrate into the space between the polymer chains thus allowing device fabrication using simple spin-casting techniques ([32] and References therein). blends of MEH-PPV with its more electro-negative derivative CN-PPV that also benefit from photo20 Recently.2%).calculated using the numbers under illumination. than 100 times faster (1. impedance and photo-voltage spectroscopy [121] as well as deep level transient spectroscopy [92]. SINGLE LAYER DEVICES Figure 3. probably the worlds first soluble semiconducting polymer. a fast decreasing Voc . • Films of LPPPT which are thinner than about 60nm show a strong increase of EQE. another deposition method which can be applied for PPV such as laser-ablation has been found and investigated [240].5MΩ/nm) than Rs under illumination (λ = 536nm.091pA).e the photovoltaic performance and the PL as discussed in Chapter 2 and Ref. • The PL efficiency is rather low (1. the physics in these thin devices seems to be dominated by other effects which are not yet understood.44 CHAPTER 3. USA) were successful at obtaining a PPV derivative with branched saturated side chains [124]. These side chains are also not conjugated so they do not alter the semiconducting or electronic properties significantly.

at the Melville Laboratory for Polymer Synthesis in Cambridge. The onset of absorption is not very sharp.3. Together with the absorption data we calculated the absorption coefficient which we used in order to confirm the thickness of the thinner films. The MEH-PPV films were spincoated in air after filtering (5µm) from chloroform solution (1 . Figure 3.46eV). Experimental We fabricated ITO/MEH-PPV/Al photodiodes with different thickness of the MEH-PPV layer ranging between 10 and 250nm. The numbers represent the film thickness. LPPPT where the 30nm device keeps following the absorption. More details on sample preparation and the used setup can be found in Chapter 10. MEH-PPV was synthesised by H. information on the synthesis can be found in Ref.22: 3rd and 4th quadrant IV characteristics of LPPPT devices under light.19). For the EQE spectra of these two devices we found that: • The thickest (250nm) device shows an EQE peak in the onset of the absorption spectrum while the EQE peak coincides with the absorption peak for the thinnest (10nm) cell. Here we investigate to what extent the thickness (between 10 and 250nm) of the MEH-PPV layer effects the photovoltaic properties. MEH 45 Figure 3. Rost.3.6 mg/ml) onto ITO coated glass substrates (Balzers).23: IV characteristics (solid lines) in the dark and under illumination of a ITO/LPPPT/Al device (59nm thick) and calculated curves (dashed lines) using Eq. (2. [181. 124] and references therein. According to Table 8. We measured the thickness of the thicker films (50 to 250nm) with a profilometer (Dektak) after 2 days drying in air in darkness which is necessary to obtain films hard enough to avoid the profilometer needle to scratch the polymer considerably.3. Such behaviour is unexpected for an otherwise symbatic EQE spectrum and indeed in contrast to the results in e. • The EQE of the thin cell increases again for shorter wavelengths after reaching its maximum at the absorption peak whereas the optical density continues to fall.07eV. starting at 600nm which corresponds to a bandgap of 2. . The absorption spectrum reveals essentially one broad feature with the peak at 505nm (2. induced charge transfer with 6% monochromatic EQE have been reported[112]. Results and Discussion Absorption and EQE Spectra Fig.25 shows the linear absorption of MEH-PPV together with the EQE spectra for two different thicknesses.g. Unless stated otherwise all devices where measured when illuminated through the ITO contact.3 a sufficiently thick MEHPPV film can absorb about 28% of the solar radiation. Note that the bandgap has been significantly reduced with respect to the insoluble PPV.

and electroluminescence quenching has been found in organic films near (<20nm) metal electrodes [14. effective photo. (3. However. all spectra show very similar features and trends as observed in the LPPPT film . A plot of the EQE versus film thickness taken at λ =505nm shows how the efficiency increases by nearly a factor 50 from 0.0014% for a thickness of 250nm to 0.referred to as MEH-PPV.064% for 10nm . 103]. 54nm) but higher again for the 10nm film. More recently. 111]. the high EQE in our thin devices may be affected by the generation of surface plasmons but the gain of a smaller filter effect and higher field due to the small distance between the two electrodes seems to overcompensate those quenching effects. The energy transfer onto these plasmons represents an effective decay channel particularly if the excitons are close (≈20nm) to the metal film [40]. All films show an increasing EQE in the short wavelength range .3) to estimate the thickness of the active layer from the antibatic EQE of the 35nm cell. Also shown is the absorption of a 38nm MEH-PPV film. • The EQE peak of the 10nm device is almost 5 times as high as for the 250nm cell. Both these earlier findings would let us expect lower EQEs in such thin films since the lost excitons are believed to be transferred onto the metal where their energy dissipates into heat rather than being split into their constituent charges. We consider this mainly to be due to the low resolution of the EQE scan (10nm) but also random errors related to the photocurrent measuring process.3. which is 17nm higher than the value found for PPV [113. The fact that the efficiency is highest in a layer as thin as 10nm is particularly interesting in the . SINGLE LAYER DEVICES Figure 3.see previous Section. It is a poly(2-methoxy-5-(2ethyl-hexyloxy)-1.in other words .46 CHAPTER 3.although the slope is flatter for medium thickness (35.see Fig.24: Chemical structure of the polymer used in this Chapter. it has been shown that non radiative energy transfer .3.4-phenylene vinylene .12) Figure 3.the transfer of an exciton from an excited molecule to the metal electrode can generate surface plasmons. The EQE values for thicknesses ranging from 10 to 250nm with some intermediate values are plotted in Fig. 15. if we apply Eq. we obtain: dopt = 26 ± 7nm (3.26. view of other discoveries: For example. Hence. Apart from the raising EQE of the 10nm film. The large error stems from the uncertainty of the peak position of the 35nm film in conjunction with the steep slope of OD which results in a relative error of λe = 22%. Note that the 54nm film actually has its peak closer to the absorption peak than the 35nm film. The EQE increases clearly with thinner films.25: EQE spectrum of a thick (250nm) and very thin (10nm) MEH-PPV cell.27.

28: IV light (solid line) and dark (dashed line)characteristics of a 250nm thick MEH-PPV device.27: EQE versus film thicknesses for the MEH-PPV devices ranging from 10 to 250nm.29: IV light (solid line) and dark (dashed line)characteristics of a 210nm thick MEH-PPV device.26: EQE spectra for MEH-PPV devices with thicknesses ranging from 10 to 250nm. . The dashed line is a guide to the eye.3. MEH 47 Figure 3. Figure 3. Figure 3. Figure 3.3.

SINGLE LAYER DEVICES Figure 3. Figure 3.48 CHAPTER 3.31: IV light (solid line) and dark (dashed line)characteristics of a 38nm thick MEH-PPV device. . Figure 3.32: IV light (solid line) and dark (dashed line)characteristics of a 35nm thick MEH-PPV device.30: IV light (solid line) and dark (dashed line)characteristics of a 54nm thick MEH-PPV device.33: IV light (solid line) and dark (dashed line)characteristics of a 10nm thick MEH-PPV device. Figure 3.

A rough comparison with the table for LPPPT reveals that both the shunts (Rsh ) and Rs in MEH-PPV are generally about an order of magnitude higher and Voc by about a factor 6. For more information on the dark current which is an experimental artifact see Chapter 10.36.039 0.3. The series and shunt resistors are plotted versus film thickness in Fig.039 0.5(8.4: Survey of various solar cell parameters obtained from the IV curves and the EQE (at 536nm) for MEH-PPV films of varying thickness.34: IV characteristics under illumination for different film thicknesses in MEH-PPV.004 0. Table 3. The slightly smaller dark currents Id are also consistent with the higher Voc in MEH-PPV. The changes of the shape of the IV curves under illumination can be assessed more easily in Fig.3. MEH 49 from the thinnest device the IV curves for positive voltages (1 quadrant)under illumination reveal two different slopes.29 to Fig. In Fig.3.3. The numbers represent the thickness in nm.35 we have plotted Voc and dark current versus film thickness.064 0.4 gives a survey of important cell parameters obtained from these IV curves plus the EQE values at λ =505nm. The dark currents Id are only slightly smaller than for LPPPT. In contrast to the findings in LPPPT films the IV characteristics in the dark for MEH-PPV do not change their shape much for different thicknesses in the negative voltage range (3rd quadrant). The lines are guides to the eye. the EQE is reduced by a factor of 2-3 with respect to the ladder type polymer.0) 40(74) 14(23) 40(270) 100(170) 54(220) IV Characteristics Figure 3.047 0.33 show the IV characteristics for various film thicknesses in darkness and under illumination (through ITO) with λ =550nm at about 0. However. The dark current Id is the actually measured current at 0V in darkness and should not be confused with the reverse saturation current which can only be obtained by fitting the IV curve.34.as discussed in Chapter 2 qualitatively: Higher Rs results in lower photocurrents and a large shunt favours higher Voc . Fig. As in LPPPT there seems to be an opposite trend between Voc and Id : For similar film thickness the Voc is smaller if Id is higher and vice versa.3. The latter also decreases with thickness while the other remains more or less constant. Apart .001 Voc mV 650 730 730 700 500 420 Id pA 31 14 16 9 6 7 Rsh MΩ 200(2400) 500(11000) 400(12000) 380(18000) 1400(21000) 2400(24000) Rs MΩ 7.2mWcm−2 .3.3. The Rsh and Rs values were taken from the inverse slopes around 0V and the steeper slope in the 1st quadrant. As in the LPPPT films we find that the shunt increases much faster Figure 3. The slope at lower voltages is always steeper than the other.35: Voc and Id in MEH-PPV versus film thickness. These results can be understood if we consider the equivalent circuit diagram ECD . Table 3. d nm 10 35 38 54 210 250 EQE % 0.

β in MEH-PPV is not much different from 1 for thicknesses ranging from 10 to 250nm. reasonable fill factors should be possible for higher light intensities.3.39 shows again the dark and light characteristics of the ITO/LPPPT/Al device (59nm thick).54 .g. (2. Considering the average size of Rs ≈ 40MΩ in MEH-PPV the increase for a device with e. SINGLE LAYER DEVICES Figure 3. According to Eq. polymer film thickness of 100nm is 25MΩ which is about 60% is relatively small.3. (8MΩ/nm) than the series resistor (250kΩ/nm).3. Very interesting and somewhat surprising is that Rsh increases very quickly (350MΩ/nm) and virtually linearly up to a thickness of at least 54nm.50 CHAPTER 3. Thus the shunts for these two thick films do contain a large relative error and can therefore only give and indication22 . The shunts in light and darkness are plotted in Fig.36: Series and shunt resistor versus film thickness in MEH-PPV cells. The figure shows that . In Fig. In Chapter 2 we have seen that the ratio Rs /Rsh does not only affect the fill factor but also the output current of a solar cell.19) that could be applied for the region -0.2V using the following fit parameters: 1. Hence. Figure 3. The lines are linear fits.38: Shunt resistor of MEH-PPV devices in the dark and under illumination versus film thickness. the Rs /Rsh ratio is small enough so that 22 Similar considerations could apply for LPPPT and thickness values larger than 120nm.13) Figure 3.37: Reduction factor β versus film thickness in MEH-PPV under illumination (λ = 550nm) and in the dark.unlike the situation in LPPPT where we found a strong increase for d< 60nm .4V to 1.38. We note that the differences in dark currents around 0V for the thicker films become less than 1pA which is close to the resolution limit of the source measure unit. (2.19) the current through a solar cell is reduced by the factor β β = 1 + Rs /Rsh (3.37 we have plotted β versus film thickness. However. It also shows the fit with Eq. ideality factor: n=4. Fig. The solid lines are linear fits while the dotted line represents a possible saturation. for larger thicknesses we find consolidation at around 23000MΩ.

MEH 51 ficiency of MEH-PPV of about 13% [63]) we obtain a value for the PL independent component of the reverse saturation current according to Eq.9V to be more accurate since Rs really starts to dominate only at higher voltages23 . • Both resistors increase with increasing film thickness.064% for the thinnest (10nm) device.1pA (dark) 3.12pA obtained for LPPPT.4. (2.14) Figure 3. The results for the shunt are very similar but the values for Rs are about 10 times larger in the fit.39: Experimental IV curves (solid lines) in the dark and under illumination of a ITO/MEHPPV/Al device (38nm thick) and fits (dashed lines) using Eq.19). Rs = 2. Thus.2W/cm2 ) for an area of 4mm2 . but Rsh grows more than 30times faster (8. The maximum active layer was estimated to be around 26±7nm which is an estimate for the upper limit for the exciton diffusion length in MEH-PPV. Summary In this Section we have found for MEH-PPV that: • Thickness is an important parameter.3MΩ 4. • The highest EQE efficiencies were achieved for the thinnest cells where the optical filter effect has virtually disappeared. The optimal thickness can be estimated roughly from the absorption spectrum and an antibatic EQE spectrum. (2.3.7pA (light). We consider the numbers for Rs values which are derived from the slope around 0. This would then indicate that the maximum performance of both materials LPPPT and MEH-PPV showing the used PL efficiency has been closely approached in these devices. resistors under illumination: (0.13 = 0. 0. I0 . A more detailed study comprising a higher number of samples and systematic IV analysis is necessary to confirm this possibility and explain the observed variation of Voc with thickness.0MΩ and IL = 1720pA Note that the resistor values shown in this list are somewhat different from the values shown in Table 3.0MΩ/nm) than Rs (250kΩ/nm) under illumination (λ = 550nm. Note that in contrast to the fit for the LPPPT device we had to choose two different values for I0 in MEH-PPV to obtain a reasonable fit. 23 Unless The value for I01 is close to the 0. I0 = 0.3.0014% for the thickest (250nm) devices to 0.see Chapter 2. According to the discussion on the photoluminescence effects in Chapter 2 we can now take the reverse saturation current I0 together with the PL efthe IV curve becomes clearly linear for higher voltages the (inverse) slope can only represent a upper limit of Rs .25) of: I01 = I0 · ΦP L = 0. reverse saturation current: I0 = 0.2mW/cm2 ): Rsh = 400MΩ. Rs = 1. it can not be ruled out here that I0 depends only or at least predominantly on the PL efficiency. • For similar thicknesses devices with lower dark current Id shows higher open circuit voltages. • The PL efficiency multiplied with the reverse saturation current (0. In the latter the numbers were derived directly by taking the slopes of the characteristics at 0V (Rsh ) and the highest voltages around 0. 2. Since the entire 10nm layer contributes to the photocurrent the exciton diffusion length can be at least as high as 10nm in this material . resistors in the dark: Rsh = 1130MΩ. IL .if the polymer/Al interface has to be reached.9V (Rs ) whereas in the former the formalism considered effects of n.091pA (3.7pA · 0. • The EQE increases from 0.7pA) gives approximately .

The molecular weight of the polymer is high but also shows strong polydispersity (Mn = 38600.3. Absorption and photocurrent measurements were performed in air.see Table 8.091pA) as for LPPPT (0. only about 15% of the solar light on earth could be absorbed in a linked aromatic rings tend to twist . 24 Directly Figure 3. Hence the bandgap should be even lower than in many PTs. If we compare PTV with poly-thiophene derivatives. we find that the introduction of ethylene linkages between the thiophene rings should lead to a more rigid structure with less rotational disorder and lower aromatic character24 and therefore higher electron delocalisation.40. The polymer was spincoated from chloroform solution (15mg/ml)onto pre-etched ITO coated quartz substrates with 2000rpm in a laminar flow box. devices and the EQE/IV measurements can be found in Chapter 10. 3.12pA).40: Chemical structure of Poly[3. More details on the preparation of substrates. Note that replacement of carbon atoms with sulfur in a conjugated systems generally decreases the bandgap.see Fig. 91]. • Rs is on average about 15 times smaller than Rsh so that the output current reduction factor β is not very different from 1 for all thicknesses.52 the same number for MEH-PPV (0. CHAPTER 3. One of these derivatives (MEH-PPV) has been investigated in the previous Section and showed broader wavelength sensitivity but otherwise poorer photovoltaic properties than PPV. Only the 10nm device shows clear dependence for (negative) bias larger than 0.e the photovoltaic performance and the PL as discussed in Chapter 2 and Ref. • The dark current in MEH-PPV is virtually voltage independent for negative voltages and the thickness range 30 to 250nm. • The EQE of the thinnest MEH-PPV device follows the main absorption band but increases again in the blue.4-bis(S)-(2-methylbutylthio) thienylenevinylene]. Experimental PTV has been synthesised via the McMurry reductive polymerisation by Francesca Goldoni. solvatochromic effects and circular dichroism (a bisignated sigal has been observed for this polymer) can be found in Ref. while the absorption still decreases. Here n ≈ 120 (Mn) With a bandgap of 2.[90. The Netherlands).6V. In this Section we investigate a molecule where the benzene ring of MEH-PPV is replaced by a thiophene ring and the alkoxy by the stronger electron donating thioalkyl chain. PPV has shown some of the highest solar cell parameters for polymers in single layer devices. SINGLE LAYER DEVICES very thick film of PTV . This is not only an obstacle for the application as photodetector material but also for the use in solar cells. the high bandgap (≈ 2. This is consistent with a strong correlation between the IV curve i. The following observations are in contrast to the findings for LPPPT: • Voc in MEH-PPV is generally about 6 times higher and Voc also tends to increase with thinner films. The introduction of alkoxy substituents could lower the bandgap by about 0. Moreover.2V. The number of repeat units can therefore be calculated as 120 (Mn) and 470 (Mw). [234]. This has already been observed for other semiconducting polymers [156].3. University of Technology (Eindhoven.4eV and lead to good solubility in many solvents.4eV ) does not allow sufficient absorption of photons in the visible range.4eV. the side-chains are chiral and can therefore induce interesting optical properties such as the ability to turn the plane of electromagnetic waves (optical activity) by means of aggregation of the main chains into chiral superstructures. Mw = 146680).4 PTV With an EQE of about 1% and Voc of 1. Details of the synthesis. Thus we obtain a polythiophene (PT) derivative with ethylene linkages PTV . Unfortunately. PTV.

42: The semi logarithmic plot of the EQE spectrum of the same device as in Fig.3.42.g.73eV. The photoresponse covers the full visible range with virtually unequivocally correlated EQE values distributed over 4 orders of magnitude.4. The photocurrent increases by about an order of magnitude when illuminated with e.43.41 shows the EQE spectrum of an ITO/PTV (70nm)/Al device when illuminated through the ITO contact as well as the linear absorption curve. The low bandgap suggests well ordered regions with extended conjugation along the polymeric backbone.41: The EQE of a 90nm thick PTV device is very low in the main absorption region but reaches values similar to MEH-PPV in the blue range.41reveals that the photoresponse starts in fact with the onset of light absorption and increases nearly by a factor of 10 every 100nm. Results and Discussion Absorption and EQE Spectra Fig.3 a sufficient thick layer of a material with such a bandgap can absorb about 42% of the solar light (ranging from 300 to 720nm).8. PTV 53 combination of two extrinsic charge generation processes such as a photoactive Al/polymer interface and the presence of exciton dissociation sites such as oxygen distributed within the bulk.07% around 400nm. For comparison: Tada et al.41 does not reveal any response in the region of the main absorption band (not even at the absorption onset) but reaches its maximum EQE at around 0. Such a behaviour. had to use the photoresponse of a thick and a thin layer of a poly thiophene derivative (Poly(3-hexylthiophene) to obtain a colour sensor with similar sensitivity but much narrower spectral range (550 and 630nm) [243].3. The absorption spectrum shows features of the vibronic structure . As a consequence. A possible explanation may have to consider the special optical properties (optical activity) due to a chiral superstructure25 or simply be the result of a 25 A bisignated circular dichroism spectrum was found for aggregates in solution [90] .3. The onset of absorption is at around 720nm which corresponds to a bandgap of 1.[196]) or other soluble polythiophenes [202] but clearly lower than the 1% found in PPV [168]. Another interesting result is that despite the rea- Figure 3.similar to the ones observed in solution[90]. neither a symbatic response nor a clear sign of an optical filter effect is rather unusual. Even under the low illumination intensity (0. This number is very similar to peak EQEs of good single layer materials such as MEHPPV (see Ref. The EQE spectrum as shown in Fig.3.3. this device can be used as a colour or wavelength detector for a wavelength range as wide as 400 to 800nm (the full visible range). Figure 3.reveals a nearly single exponential increase of the EQE starting from the absorption onset at 720nm. IV Characteristics The IV characteristics in the dark and under monochromatic illumination (420nm) is shown in Fig. 500nm instead of 600nm. According to Fig.3. We found that EQE ∝ λ−9 fits very well for λ > 480nm The EQE plotted on a logarithmic scale Fig.05mW/cm2 ) the device develops an open circuit voltage of 370mV with a fill factor of ≈ 30%.

We note that the Rs values taken from the first (around 0.76eV can be used to built a ITO/polymer/Al cell. SINGLE LAYER DEVICES Figure 3.54 CHAPTER 3.possibly because of the lower light intensity here. the presence of deep and/or shallow traps can lead to an even stronger dependence in a limited range of V before returning to obey the trap free equations above. sonable Voc and relatively good fill factor. From the slopes in Fig.44 may display the trap dominated region.3. .3. this device shows no strong rectification ratio since there is no clear off state current.43: IV characteristics of the PTV device in the dark (dashed line) and when illuminated with λ =420nm (solid line).25V) are virtually identical. A study of Rs as a function of light intensity would be desirable as part of a future project. Summary In this Section we have found that: • A semiconducting polymer with a bandgap as small as 1. Ohm’s law seems to fit for lower voltages while the I ∝ V 4 relation may be due to charges released from deep and/or shallow traps. The smaller difference between the shunt in dark and light compared to LPPPT and MEH-PPV may also be due to the lower illumination intensity here.2V.3. 231] for trap free space charge limited current the current through the diode obeys: I∝ for low voltages and I∝ V2 d3 (3.44: A double logarithmic plot of the dark characteristics.2 to 1 V in Fig. The range from 0.43. 0.45V) or the third quadrant (counterdiode. For voltages larger than 0.16) V d (3.44 shows a double logarithmic plot of the same IV characteristics as in Fig. This has probably to do with the low bandgap that is more likely to lead to low interface barriers for both forward and reverse bias. It looks more as if a counter-diode has been formed with a even lower breakthrough voltage than in forward bias. around -0.43 we have extracted the following resistor values: • light: Rs = 200MΩ and Rsh =3000MΩ • dark: Rs = 200MΩ and Rsh =6400MΩ The series resistor is about 10 times higher than in MEH-PPV or LPPPT . current and voltage obey Ohm’s law of direct proportionality.2V up to 1V we find an I ∝ V 4 relationship.15) for high voltages However. There are two different slopes fitting the dark characteristics in the positive bias voltage range: For voltages up to Figure 3. Fig.3. According to the MottGurney equation [138.

n = 20-25.3.3. 3. Results and Discussion Absorption and EQE Spectra Fig. Germany. • The EQE reaches 0.see Fig. high lying HOMO level. • The cell shows a relatively good fill factor for a single layer polymer cell of about 30% with an open circuit voltage of 370mV • The device shows low breakthrough voltages for both negative and positive voltages. all polymers that have been discussed in this Chapter can be used as electron donors because of their small ionisation energy i. devices and the EQE/IV measurements can be found in Chapter 10.3.45 shows the chemical structure of a polymer (PIF) that has the potential to be both a low bandgap absorber as well as an electron acceptor with respect to many other polymers. referred to as PIF. In this Section we will discuss the absorption and photovoltaic properties of ITO/PIF/Al devices.07% at 400nm. [210]. The thickness of the organic layer was about 30nm. Thus.3. Thus it can be used to amplify positive and negative voltages. The polymer was spincoated from chloroform solution (8mg/ml) onto pre-etched ITO coated quartz substrates at 2000rpm in a laminar flow box. Absorption and photocurrent measurements were performed in air.5.8. • The EQE spectrum is neither antibatic nor symbatic . this device can be used as colour sensor covering the full visible range. The quinoid character was expected to lead to relatively high electron affinity which would classify this material as an electron acceptor with regard to the majority of other conjugated polymers. The short distance between the fairly large repeat units causes a twisting angle of 33o [210].13eV. PIF • The EQE is very low at long wavelengths but very sensitive to the illumination wavelength and obeys EQE ∝ λ1/9 between 480 to 720nm. Wiesler at the MPI in Mainz. this might be the . The distinct twist between the monomer units and the short but bulky side chains allow for plenty of space between the molecules so that it dissolves readily in many common solvents. A sufficiently thick device of a semiconductor with this material can absorb 77% of the direct solar radiation on earth . The quinoid character.possibly due to the optical activity in conjunction with a chiral superstructure • Resistor values derived from the IV curves at 0. the large π . Thus. Details about the synthesis can be found in Ref.system and the bulky side chains make this material a candidate for a low bandgap electron acceptor.45: Chemical structure of poly indeno fluorene. Note that the bandgap is virtually identical to the one of crystalline silicon.46 shows the absorption spectrum of PIF together with the EQE of an ITO/PIF/Al device (20nm thick). More details on the preparation of substrates.5 PIF Up to now. This enables direct determination of the wavelength of incident light from the photocurrent using the spectral response curve.e. The absorption peak is at 800nm whereas the onset of absorption is around 1100nm which corresponds to a semiconductor bandgap of 1. Reisch and U. Experimental PIF has been synthesised by H. Despite the distortion of the conjugated backbone this molecule can still have a very low bandgap thanks to the large π system of the single repeat units. Fig.05mW/cm2 are: Rs = 200MΩ and Rsh =3000MΩ and in the dark: Rs = 200MΩ and Rsh =6400MΩ 55 Figure 3.

MEH-PPV which may indicate either partial shorts due to pinholes or high conductivity due to dopants or rather low barriers in the pathway of the charges.Rs = 5. Although these intensities are still relatively low.confirms that the photoresponse coincides very well not only with the peaks but also with the onset of absorption.7MΩ.3. Organic semiconductors with very low bandgaps often lack photo-stability. with this material we did not observe any signs of photo-degradation before or during the measurements where the polymer was exposed to light intensities of up to 0.3. The lack of a reasonable Voc is most likely related to the low shunt resistor.46 shows that absorption and EQE are very much “in phase” even at the onset at 1100nm IV Characteristics The dark and light characteristics of the same PIF device are shown in Fig. MEH-PPV. first photodiode based on a soluble semiconductor that covers the same spectral range as commercial silicon photodiodes.g. the material is very conductive already without applying a voltage. The values are considerably smaller than in e.g.003% at 800nm). The whole bulk is active since we observe a symbatic behaviour. We want to note here that the error of the EQE spectrum between 400 and 500nm can be as high as ±50%.47 . The EQE is again higher for the shorter wavelengths.56 CHAPTER 3.46: EQE and absorption spectrum of a PIF device when illuminated through ITO and measured in air. Note that the current through this device is about 10 times higher than in MEH-PPV so that the slope around 0V is actually very steep i. are about an order of magnitude lower than in e. there is neither a significant Voc nor a reasonable rectification ratio.7MΩ (negative voltage range) and Rsh =66MΩ • dark: Rs = 1. However. however.47: The semi logarithmic plot of the same data as in Fig. Both the dark and the light characteristics appear almost symmetrical and on top of each other at the scale shown.3. From the slopes in Fig. Despite a significant Isc .3. for example.2mW/cm2 for several minutes in air. The absolute EQE numbers (0. A semi logarithmic plot .7MΩ (negative voltage range) and Rsh =83MΩ Note that according to the discussion in Chapter 2 all Rs values obtained from the slope in the IV curve can only represent upper limits. The EQE spectrum closely follows the absorption indicating that the entire bulk contributes to the photocurrent. .e. This is because the signal to noise ratio is very low due to the low emission intensity of the QTH-lamp and the low EQE of PIF in this region.49 we have extracted the following resistor values: • light: Rs = 1.see Fig. SINGLE LAYER DEVICES Figure 3.48. Thus.7MΩ. as photo-detector at such intensities. The photoresponse covers the same range as commercial silicon photodetectors (400-1100nm) Figure 3.Rs = 5. the device was working and could serve. the relatively few photogenerated charge carriers cannot change the shape of the dark characteristic significantly.

64]. extended conjugation has been found in both organic and organo-metallic poly-ynes [142].7MΩ (negative voltage range) and Rsh =83MΩ • Child’s law for trap free materials fits the IV characteristics between 0.e. Such a steep slope with a low Voc seems to occur more often in electron transport materials particularly if they have a low bandgap [150. Moreover.48: IV characteristics of the PIF device (32nm thick). There are also no clear signs for deep or shallow traps. Figure 3. The steep slope i.003% at its absorption peak (800nm) • There is only little rectification and a negligible Voc (smaller 20meV) .7MΩ. strong voltage dependence of the current around 0V seems to occur more often in electron transport materials.6.49 we plotted the dark curve of Fig.7MΩ (negative voltage range) and Rsh =66MΩ and in the dark: Rs = 1.49: The IV curve (filled circles) fits the Mott and Gurney equation (Child s law) for trap free materials (I ∝ V 2 . The large and electron rich π system should be reasonably flat and therefore result in a particular low bandgap . The IV curve follows the Mott-Gurney equation (Child’s Law) for trap free materials (I ∝ V 2 ) between 0.Rs = 5. in fact.1 and 1V.3. solid line) and shows no ohmic behaviour as expected for such low voltages (I ∝ V . • Resistor values derived from the IV curves at 0. two fluoro substituted phenylene rings extend the the conjugated system further.3. • The EQE of this material reaches 0.1 and 1V Summary In this Section we have found that: • The conjugated polymer PIF with a bandgap as low as 1.6 THPF In this Section we are going to investigate a polymer with a molecular structure which has a number of interesting peculiarities .most likely due to the low shunt resistor. The triple bond does not break the conjugation in these poly-ynes26 . In Fig.2mW/cm2 and λ = 800nm are: Rs = 3.50: A dialkoxy substituted phenylene ring is coupled via a triple bond to a thiophene with a pyrazine fused to it. . Dark and light curve coincide on this scale.48 on a double logarithmic scale. dashed line).3. THPF 57 Figure 3.13eV can be used to manufacture a photovoltaic cell with a photoresponse covering the same range as a commercial crystalline silicon based solar cells.7MΩ.see below. 1.Rs = 5.3.13eV) may well lead to low electrode/polymer . The device was illuminated at λ=800nm The extraordinary low bandgap among today’s polymers (1.interface barriers for both injection and extraction of charges and therefore lead to the formation of a counter-diode as in the low bandgap polymer PTV.see Fig. 26 The term -yne denotes the triple bond.

27 Cyano Experimental THPF has been provided and synthesised by M.8.see Fig. The shape as well as the absolute numbers with the considerably high peak EQE around 400nm resemble the EQE spectrum found in the PTV device. The plateau around 600nm makes an application as colour sensor clearly more difficult.72eV). n ≈ 22 In addition. From the comparison with a derivative that lacks the fluor atoms. Figure 3.1% at 400nm which is more than twice as much as typical MEH-PPV devices.an interesting candidate as electron acceptor in D/A solar cells. 147.3. However. the bandgap is negligible . The IV characteristics of the same device is shown in Fig. the shoulder in the EQE curve at 700nm might indicate the presence of the optical filter effect.shows no exponential dependence on the wavelength for THPF. Khan. More details on substrate. The IV curves show ohmic . The onset of absorption is at 720nm (1. A sufficiently thick layer with this bandgap can absorb about 48% of the direct solar radiation on earth . Unlike the results in PTV devices.51 . More information on organic and organo-metallic poly-ynes can be found in Refs. The EQE exceeds values of 0.51: The EQE of an ITO/THPF/Al device reaches 0. device preparation and the EQE/IV measurements can be found in Chapter 10. SINGLE LAYER DEVICES Results and Discussion Absorption and EQE Spectra Fig. However. low bandgap acceptor materials are particularly promising for the use in D/A solar cells but still very hard to find among polymeric semiconductors.which is consistent with the findings for MEH-PPV and its cyano substituted analogue27 . Absorption and photocurrent measurements were performed in air. the introduction of the fluoro substituted phenylene could increase the electron affinity significantly and make it . Figure 3. The absorption spectrum shows a dominant peak at 620nm and a shoulder at 680nm.53.58 CHAPTER 3. the semi logarithmic plot . 184]. the spatial separation from the thieno-pyrazine via the phenylene could weaken their effect to some extent. In any case. In addition. The polymer was spincoated from chloroform solution (16mg/ml) onto pre-etched ITO coated quartz substrates with 2000rpm in a laminar flow box. [141. University Chemical Laboratory (Cambridge. [143].3.3.3. we know that their effect on e.see Fig. 146. Note that in both materials the EQE in the main absorption band is considerably lower than near 400nm. IV Characteristics (CN) groups are like fluor atoms strong electron withdrawing and therefore often used to increase the electron affinity of an organic compound.g.together with the low bandgap . in the main absorption range (500-700nm) the efficiency is only 0.003% which is an order of magnitude lower than in analogous MEH-PPV cells. Details of the synthesis can be found in Ref.50: Chemical structure of the polymer THPF used in this Chapter.52 shows the linear absorption spectrum together with the EQE spectrum of a 110nm thick ITO/THPF/Al cell.15% around 400nm. UK).

The values are somewhat higher than in e. THPF 59 • light: Rs < 6300MΩ (λ =552nm). However.2mW/cm2 and λ = 552nm are: Rs < 6300MΩ and Rsh =6300MΩ and in the dark: Rs < 17000MΩ and Rsh =17000MΩ . From the slopes in Fig. the unusually big difference between the EQE in the main absorption and around 400nm suggests that there may be other wavelength dependent effects dominating. Summary In this Section we have found for ITO/THPF/Al devices: • The EQE is relatively low (0. 0.g.06mW/cm2 ).52: A semi logarithmic plot of the EQE spectrum in Fig. • Resistor values derived from the IV curves at 0.3.Rs < 1400MΩ (λ =440nm) and Rsh =6300MΩ (λ =552nm).3.003%) in the main absorption band (500-700nm) but reaches more than 0. This would require further investigation preferably involving light intensity dependent EQE/IV for different wavelengths in future studies. However.so that both the shunt and the series resistors are equal here.although some signs of an internal filter effect are visible.shown here for two excitation wavelengths) of the THPF device. Figure 3. MEH-PPV which indicates the absence of significant shorts through pinholes and thus account for the good open circuit voltage.53 we have extracted the following resistor values: • The EQE can be doubled by applying a reverse bias voltage of 1V. This is particularly important here since the voltage range is probably not large enough to exceed the onset of the diode breakthrough voltage . Note that the shunt (and the upper limit for Rs ) at the wavelength with the lower light intensity is almost 5 times smaller as is the short circuit current. 0. We can also see that the EQE can be doubled by increasing the reverse bias voltage by 1V. The behaviour is ohmic. Rs could be considerably smaller. Considering that the OD for the two wavelengths is roughly the same (0. • The photo-responsive range covers a broad wavelength range (400 to 800nm).3. behaviour between -1 and +1V but develop a reasonable open circuit voltage of 600mV (at λ =552nm. • A relatively high open circuit voltage of up to 700mV.6.52 reveals that the EQE follows the absorption .37 at 552nm) the higher conductivity at lower light intensity is rather surprising. Figure 3.1% around 400nm.51 at 440nm and 0.53: IV dark (dashed line) and light characteristics (solid lines . Rsh =1400MΩ (λ =440nm) • dark: Rs < 17000MΩ and Rsh =17000MΩ Note that according to the discussion in Chapter 2 all Rs values obtained from the slope in the IV curve can only represent upper limits.2mW/cm2 ) and 700mV (at λ =440nm.

Ise = 3.54: Chemical structure of the soluble perylene diimide (Per) used in this Section. Experimental The perylene dye has been purchased from Synthec (Germany). In this Section we investigate a photovoltaic cell using a perylene derivative (a dye) that allows for simple processing via spincoating from solution.” In this basic patent he also mentions that electron-acceptor compounds should have at least 7 aromatic rings and 8 for elec28 The actual perylene structure (4 aromatic rings fused ˚ together around a central ring) measures about 7x6A tron donors to avoid shorts due to pinholes in the about 40 to 50nm thin films. Only very recently (1998). the more planar the molecule is [154]. it has been found that the photosensitivity of perylenes is higher. Furthermore. In this Lewis-acid pigment solubilisation (LAPS) process. Their motivation to investigate these molecules were likely driven by the fact that many perylene diimides are strongly absorbing in nearly the entire visible range and that they were available.. Also photo-conducting properties of perylenes have been recognized as early as 1972 when Regensburger and Jakubowski studied the xerographic behaviour of a number of perylenes in bi-layer devices[218].0% under AM2 conditions. We note that in the meanwhile also soluble perylene containing polymers are available [217]. energy intensive and expensive thermal sublimation has to be employed since their solubility in common solvents is very low. FF = 0. We note that perylene bis-imidazole absorb at longer wavelengths relative to those of perylene diimide due to the extended π-conjugation[154]. most of which also exhibit photoluminescence efficiencies around 100% in solution [69]. chemists at Xerox have found a method to process many hetero cyclic pigment classes from solution to facilitate cheaper manufacturing [283].0V. Figure 3.9.4.7 Per The most frequently used perylene pigments are diimides and bis imidazoles of perylene-3. [244]: A NESATRON29 / diimide-perylene/Ag device with Voc =440mV. in order to fabricate thin film solar cells with pigments. The following formulation can be found in claim 17 of Chin Tang’s Patent on a multi-layer organic photovoltaic device which underlines the importance of size28 and planarity of photovoltaic molecules in general . .not only perylene derivatives [245]:“.. he managed to fabricate one of the best organic solar cells until today using a double layer comprising two large and flat aromatic molecules such as a perylene bis-imidazole together with a copper phthalocyanine [244]. The polymer was spin-coated from chloroform solution (8mg/ml)30 onto pre-etched ITO 29 NESATRON is a brandmark for a material with similar properties as ITO .10tetracarboxylic acid. Later it was discovered that charge separation in perylenes primarily takes place at an interface with a hole accepting molecule unless the exciton decays non-radiatively to the ground state [154]. The solubilisation involves coordination of the acid with the pigment molecules’ hetero atoms to form a soluble electron donor-acceptor complex which is later removed by washing.6 and a power conversion efficiency as high as 1. Unfortunately this method does not work for all pigments and the acid might attack the pigment and/or the electrode material causing other complications. Details of the synthesis can be found in Ref. Interestingly at least one device in the patent was even better than in Ref. However. The particular diimide perylene used in this cell had a phenyl ring coupled to both nitrogen atoms. Other soluble perylenes have been reported. the pigments are mixed with a solution of a Lewis-acid in nitromethane. SINGLE LAYER DEVICES 3. One has already been employed successfully in LED structures to produce white light by means of excitation energy and charge transfer from a blue emitting ladder type polymer in a blend [246].materials comprising a compound containing a generally planar polycyclic nucleus having a ˚ surface area of at least about 40 square A and a ˚ width of at least about 5 A. Following these guidelines.60 CHAPTER 3.[60].mainly used in the 1970s 30 According to Ref [60] up to 70mg(!) can be dissolved in 1ml. [244]. They show good light and water fastnesses and are best known for their use as colourants for automobile paints and plastics.

3.56 reveals that the onset of the EQE spectrum does not exactly coincide with the onset in the absorption at 620nm. In the same reference a very similar perylene diimide has been classified as n-type semiconductor by means of the symbatic response31 and field-effect-transistor IV curves. PER coated quartz substrates with 2000rpm in a laminar flow box. active interface with ITO rather than Al which is expected to result in a symbatic EQE spectrum when illuminated through ITO . strong inter molecular interaction seems to favour efficient charge transport and/or exciton generation. The lack of an optical filter effect upon illumination through ITO in this 40nm thick device is consistent with the assumption that perylene can be regarded as n-type semiconductor: N-type semiconductors form an rectifying i. This is still between one and two orders of magnitude smaller than the values for perylene diimide pigments published in Ref. Absorption and photocurrent measurements were performed in air.55 reveals an onset of the photoresponse even before the onset of absorption (as determined by our absorption setup). Significant photoresponse can already be found at about 700nm.56: The semi logarithmic plot of the data in Fig.3.3. In addition.earlier in this Chapter. Photo-thermal deflection (PDS) measurements may be useful to investigate this further. Photoresponse before the absorption onset has also been found in THPF . More information on sample and device preparation can be found in Chapter 10. We obtained very tractile films of reasonable thickness (about 100nm) . the same side chains that facilitate solubility prevent the molecules from close packing as in pigment films.e. We find virtually no antibatic behaviour so that the two main features can be found in both spectra: A dominant peak at 495nm and another peak at 540nm. MEH-PPV films.55 shows the EQE spectrum of a 40nm thick ITO/Per/Al cell when illuminated through the ITO as well as the linear absorption curve.55: The EQE of this Per device follows the absorption spectrum closely reaching numbers comparable to MEH-PPV but covering a larger spectral range.see also Chapter 2.3. the EQE reveals another local maximum at 370nm. [134].08% which is about twice as high as MEH-PPV cells with the same thickness.regardless of the actual exciton diffusion length. The EQE reaches about 0. The logarithmic plot in Fig. Further confirmation of the n-type nature of this particular organic dye could be obtained from field 31 when illuminated through SnO2 . Results and Discussion Absorption and EQE Spectra Fig.with no spin marks but the surface appeared rough in contrast to the shiny surface of e. Hence. Both materials have electron withdrawing groups that are expected to lower the LUMO to make them stronger electron acceptors . 61 effect measurements following the argumentation in Ref. [245].g. Figure 3. Thus. Figure 3.7.

compact packing of molecules (as with pigments or using the LAPS method) seems to be required to obtain the best efficiencies. there is some evidence that large polycyclic aromatic hydrocarbons .3. • The EQE onset appears to be significantly before the absorption onset which requires further clarification in future studies. In addition.which might be .9MΩ and Rsh =50MΩ – dark: Rs = 0.since the slope there is smaller than the one in forward bias representing the “lower” upper limit for Rs . Summary In this Section we have found that: • A soluble perylene (Per) derivative can be used to fabricate an ITO/Per/Al solar cell from solution avoiding the usual high temperature sublimation steps hitherto used for perylene derivatives.9MΩ and Rsh =60MΩ Both the series and the shunt resistors are more than an order of magnitude smaller than in e. the dark and the light characteristics have similar shapes even in the negative bias region.57 we have extracted the following resistor values: • light: Rs = 0.responsible for the high conductivity here. We note that the formation of crystals has been observed in Per/P3HT blends but also to a smaller extent in single layers [62. 63]. MEH-PPV or LPPPT.8 Ter As we have mentioned earlier. Note that photocurrent multiplication as large as 10 000 times has been observed in other perylene diimides sandwiched between Au electrodes [133]. • A counter-diode has been formed similar to PTV and PIF devices. The value for the series resistor has been taken from the slope in the negative voltage range . From the slopes in Fig. solid line) characteristics of the Per device. CHAPTER 3. • The EQE reaches up to 0. This would probably account for the relatively small Voc and the high positive dark current.57: IV dark (dashed line) and light (λ =550nm.57. A clear rectification for voltages above 1V is observed.3.9MΩ and Rsh =60MΩ 3. This means that the EQE cannot be increased by increasing the negative bias voltage as in MEH-PPV or THPF. Note that similar to the previously described PTV and PIF devices there is a relatively small negative breakthrough voltage indicating the presence of a counter-diode. Figure 3. Hence. • The Per/ITO interface is the “active” interface which is consistent with the n-type semiconducting properties of other known perylene derivatives.62 IV Characteristics The light and dark characteristics of this device is shown in Fig. Crystal formation is a known problem with dye films and can lead to pinholes and partial shorts . • The Voc (190mV) of the dye device is somewhat smaller than in pigment devices (280-530mV) although different illumination conditions and electrode material have to be considered. SINGLE LAYER DEVICES least partly .at • Resistor values obtained from the slopes in the IV curves are: – light: Rs = 0.08% which is twice as much as comparable MEH-PPV devices but 10 to 100 times less than the best pigment devices. We found very large currents indicating high conductivity and a rectification ratio of about 20 at ±1V.g.9MΩ and Rsh =50MΩ • dark: Rs = 0.

e. • The higher concentration of π . large rylene should have a vanishing bandgap indicating the possibility of intrinsic electrical conductivity (metallic behaviour) according to theoretical predictions . Well-known examples for large polycyclic aromatic hydro-carbons are phthalocyanines and perylenes.see e.58: Chemical structure of the soluble terrylene derivative.e. are likely to give good photovoltaic conversion efficiencies . The “pure” perylene structure can be regarded as two peri-condensed naphthalene compounds. Chapter 10.58eV) and pentarylene (≈ 950nm or 1. shorts more unlikely.59: The EQE of a Ter-device shows a very broad spectral response (UV to near IR) reaching high numbers in the short wavelengths range (around 1%). We note that such a terrylene derivative has played an important role in nano-science where single molecule spectroscopy was possible even at room Figure 3. We note that the solubility (< 1mg/ml in Chloroform) of the quaterrylene of the same series (same sidechains) was not sufficient to produce continuous films via spincoating.3. Theoretically three or more naphthalene compounds can be condensed together to form a series of oligo-rylenes leading to a rylene-polymer. the application of.3. Experimental The terrylene has been synthesised by Stefan Becker at the Max Planck Institut (Mainz. • The higher molecular density makes the occurrence of pinholes i. These oligo-rylenes were also amenable to spectroscopic measurements in solution (1. TER 63 Figure 3.see Ref. investigated in this Section. quaterrylene (≈ 800nm or 1. In fact a decreasing bandgap has been found for a series of increasing rylene units: terrylene (≈ 620nm or 2. More details about the sample i. In the foreseeable future.02eV). Such a polymer i. Ter. for example.59 shows the EQE spectrum of an ITO/Ter (≈ 15nm thick) /Al cell when illuminated through .8. Ref.e. Details of the synthesis can be found in Ref. device preparation and the IV/EQE measurements can be found in Ref. heat to a single molecule and thereby the initiation of an optically detectable chemical reaction affecting the PL of a single molecule may lead to optical data storage on a molecular scale.electrons per unit volume increases the absorption coefficient.4 dioxane) thanks to substitution with four tert-butyl side chains [145]. temperature [153. Absorption and photocurrent measurements were performed in air. Results and Discussion Absorption and EQE Spectra Fig. [245]. The cited bandgaps have been obtained for thin films after thermal sublimation. [16]. [145] and references therein. Germany).g. In this Section we investigate the photovoltaic properties of this well soluble terrylene diimide derivative as active material between ITO and Al. The polymer was spin-coated from chloroform solution (6mg/ml) onto pre-etched ITO coated quartz substrates with 1600rpm in a laminar flow box.32eV) [145]. 71]. The reason for that may be found in the possibility of a more compact arrangement of the molecules in the thin films that leads to the following list of advantages: • The relatively undisturbed π system of planar molecules favours exciton and charge transport as well as a low bandgap.

97pA. we have chosen a device with a “typical” characteristic. since we could not obtain a detectable Voc we believe the device performance may be affected by local shorts due to the formation of crystals causing a resistance of only about 900kΩ. However. since the IV curve suggests a very conducting and/or strongly pinhole dominated and therefore partly shorted film.60.64 CHAPTER 3.5 we have listed the most important solar cell parameters for the different organic semiconductors investigated in this Chapter. The partial shorts would also explain the lack of a detectable open circuit voltage 10mV. The absorption spectrum of the Ter film is very broad (410 to 700nm) and featureless apart from a little dip at 630nm. However.3. The EQE value of the Ter device has to be treated with caution since the device showed non-ideal IV and EQE characteristics. IV curves with very high conductivities and small open circuit voltages tend to be observed more often in electron conductors like PIF and others [150. IV Characteristics The light and dark characteristics of this device is shown in Fig. SINGLE LAYER DEVICES gests that the device is considerably shorted via pinholes in the only 15nm thin film. We also have to consider that the values for Rs are only upper limits and that the actual values could be considerably smaller. The very steep slope that should represent the shunt resistor (here only 330kΩ) sugIn this Section we have found that: • A soluble terrylene derivative can be used to produce a photocurrent. Generally.e. However. both the HOMO and the LUMO is expected to become lower with increasing π system i.e. the terrylene should be a better electron acceptor than the perylene. Many of these materials will be investigated further in blend and/or double layer devices in the subsequent Chapters. Where possible. In fact the EQE is surprisingly high for a single layer cell with values above 0. Summary Figure 3. . For example.60: The IV dark (dashed line) and light (solid line. the ITO as well as the linear absorption curve. The positive dark current is most likely an artifact related to a slight ground potential of the source measure unit. 3.28mW/cm2 ) characteristics of this Ter device could only be recorded for a very small voltage range due to the steep slope i.g MEH-PPV reaching higher values of up to 1% around 400nm. high conductivity (330kΩ). a voltage change of 1mV can lead to an increase of current of 3nA equivalent to change in EQE of nearly 1% here. The EQE spectrum resembles the spectrum observed in PTV with no clear antibatic peak in the absorption onset but reaching the highest EQE in the UV range with the local absorption minimum. we have to take into account that the EQE might be affected here by small fluctuations in the voltage applied by the source measure unit. The list shows that only devices with shunt resistors above 100MΩ and dark currents under 30pA do develop a Voc larger than about 200mV. the IV and EQE characteristics of this particular device have therefore to be interpreted with caution.01% throughout the visible spectrum reaching about 4% in the UV. λ = 650nm. 0.9 Single Layer Devices Summary In Table 3. upon illumination the current at 0V changes from +1000pA to . Note that an offset of only 1mV would change the current by about 3nA here. This is not surprising considering the observations of crystal formation even to a larger extent than in previously investigated related molecule Per [64]. The EQE around 500nm is even in the same order of magnitude as in e. 64]. As a consequence.

9(0.7) comment – dye polymer polymer dye polymer polymer polymer Hence.9.7(1.9(0. The list also reveals that both polymers and dyes give peak EQE values of approximately the same size ranging from 0.8 14 116 Rsh MΩ 0. EQE values were taken at peak wavelength.SUMMARY 65 Table 3.with MEH-PPV being near the lower end in terms of the photocurrent but on the high end with respect to the open circuit voltage.9(0.5(11) 0. .3. d stands for the thickness of the organic layer.9) 200(200) 40(74) 1.9) 6300(17000) 7. a suspected correlation of the EQE with the series resistor or the ratios between Rs and Rsh cannot be ruled out here. material – Ter THPF LPPPT Per PTV MEH-PPV PIF d nm 15 110 59 40 90 35 30 EQE 10−3 % 4000 100 85 80 70 45 3 Voc mV 0 600 105 190 370 730 0 Id pA 1000 6 31 825 0. The number in brackets are obtained from the dark characteristics.10% . SINGLE LAYER DEVICES .5: Survey of photovoltaic parameters of the investigated organic semiconductors.04 to 0.9) 6300(17000) 83(200) 50(60) 3000(6400) 500(11000) 66(83) Rs MΩ 0.

SINGLE LAYER DEVICES .66 CHAPTER 3.

in both cases equalisation of the Fermi-levels occurs at the interfaces between semiconductors and/or metals. The separation of the charge carriers resulted in a higher shunt and low series resistor such that the fillfactor was comparable to commercial Si photodiodes.5 condition (75mW/cm2 ) was 450mV. In both organic and inorganic solar cells the equalisation of Fermi levels leads to a significant potential gradient that can drive the photoexcited charge carriers (minority carriers) through the rest of the bulk towards the electrodes.4. Figure 4. care has to be taken using such comparison. Although a rough similarity to the P/N junction in inorganic solar cells can often be drawn. The active region is now extended to the exciton diffusion length of both materials. The electron acceptor A and electron donor D are sandwiched between the transparent ITO contact and an Al layer. The open circuit voltage at AM1. With the introduction of an electron acceptor layer between the active material and the negative electrode (Al). both the exciton diffusion range and the poor shunt resistor can be improved . absorption of photons from 300 to 1000nm is required to utilise 70% of the solar radiation on earth. Although. Moreover. Since the absorption of the perylene derivative and the Cu-phthalocyanine complement each other very well. However.g. the Fermi-levels are still important the offset of the valence band (VB) and conduction 67 In the previous Chapter we have found that the main losses in single layer structures are due to short exciton diffusion lengths and recombination of the excited charge carriers.Chapter 4 Double Layer Devices 4.1 Introduction tors allows tuning of the absorption spectrum for e.1: Device architecture of a double layer photovoltaic cell. Provided the offset between the HOMO and LUMO levels of both materials upon contact are sufficient to split excitons. The latter is not only desired for optical photodetectors but also essential if high solar cell efficiencies are required.1. maximum coverage of the spectrum of the light source. However. a copper phthalocyanine and a perylene derivative on top of each other. and obtained one of the hitherto most efficient organic photovoltaic devices. sandwiched between an ITO and Ag contact. One of the first organic double layer solar cells is the one reported by Tang in 1985 [244]. as discussed in Chapter 2 the nature of doping in organic semiconductors can be very different from the usual doping in inorganic materials. In fact. The short exciton diffusion length causes a very thin active region of the device whereas recombination losses before and after charge separation can reduce the shunt resistor considerably. He sublimed two pigment films. the device showed a good EQE over the full visible range. Only about 22% of the solar radiation can be exploited with a single layer material like MEH-PPV.see Fig. the use of two different semiconduc- . the separated charges can travel to the respective electrode with very small chances of meeting a recombination partner.

4. Right: For higher charge carrier concen. As a consequence.2: Band energy diagrams for a D/A dou. which creates a constant electronic properties of liquid crystalline (LC) mafield across the bulk. The table also shows that technically pigments can already be replaced by polymers. 260.(see Chapter 6). CHAPTER 4. Although the successful synthesis of perylene The figure on the right in Fig.g.g. the laminated device structure Figure 4. Note that the term heterojunction is also found in the literature but we prefer D/A junction in this thesis since a junction between two different semiconductors does not necessarily lead to the required splitting of the excitons. 3 In 1993 more than 90% of the xerographic photoreceptors were made of organic photo-conductors [154] 4 Recently a method was found to dissolve certain pigments in acids [283]. 226.bility. 267]. spin or blade coating from non-acidic solution4 . this simplifies film deposition. Thus. a device with sufficient VB/CB offset can split excitons and separate charges efficiently if both materials are e. 43.68 band (CB) at the interface after equalisation of the Fermi-levels is a crucial factor for the function of these devices.2 shows the band On the other hand.4. 1. In Table 4. As a consequence. liq1 Here we have also neglected the band bending at the D/A junction.2 shows the situa. Although.driven by the fast progress of polyof workfunctions of the electrodes ∆Wf can cause mer based devices and the discovery of outstanding the bands to tilt.1 we have listed D/A double layer devices from the literature. This is largely because there are excitons in organic devices that need to be split at the first place. insignificant in such devices1 . to tilted bands if only few free charge carriers are Both Pcs and Per pigments show very high stapresent.5mW/cm2 . The figure on the left in Fig. ohmic contacts are more likely to form using Au and Ca instead of ITO and Al. ically modified so that they can be used as dyes. p-type conductors. tion for a device with a concentration of free charge phthalocyanine dyes and polymers of both have carriers that is high enough to compensate the built only been synthesised in the more recent past . 184. 254.015 mW/cm2 so that significantly higher values for Voc may be achieved if measured under intensities in the order of 0. However. the easier processible soluble energy diagram for a double layer device with a low derivatives have attracted the interest of researchers concentration of free charge carriers. allowed semiconducting polymers ble layer device in short circuit mode. the use of acids introduces other problems. 253. 154.been proven suitable as photo-active pigments in sotrodes can cause distinct band bending2 . 255. fraction of the layer thickness. as shown. 231. makes soluble molecules technologically more interesting. 2]. the prospect of easier processing by e. Pigments have been successfully used as photoreceptors and charge transport media in colour copying machines and related devices (Xerography) replacing many inorganic materials3 . The difference more recently . the electrodes can still occurs. Left: the to exceed in performance values for the best pigbuilt-in field due to ∆Wf of the electrodes leads ment devices [94]. Although band bending at terials [228. dyes and even liquid crystals but at the expense of lower efficiency. it may considered 262.see in field due the electrode workfunctions within a Refs above. The device with ∗ has been measured at only 0. 2 We have drawn the formation of blocking contacts. good absorption and good photo-conductive tration the bands can remain flat in the bulk but properties which makes them the probably most inblocking contacts may be formed vestigated photovoltaic materials in the past [268. 259.dyes has been reported already 18 years ago [209]. We note that many other molecules that have accumulation and depletion of charges near the elec. may also be of utmost interest to be chemAccording to this picture the driving force i. . DOUBLE LAYER DEVICES Voc is mainly determined by the offset at the D/A junction although the increased built in field (tilted bands) can help to transport charges through the bulk [223]. The table shows that the highest efficiencies have been achieved with insoluble molecules (pigments).e. lamination or even inkjet printing. We note that only recently. a modification of the double layer device. lar cells. 29. it is more appropriate to speak about a D/A -rather than P/N junction in case of organic solar cells. 227.

note that in Ref. Among those should be mentioned: squaraines. but also charge transfer complexes like PVK:TNF [154] and certainly the large number of different merocyanines and azo pigments [154.g.g. [187] Foerster energy transfer facilitating exciton diffusion to the active junction is given as explanation for better performance in these devices. 29].0 0.i.g. 282]. Polymers and dyes in this table were processed from solution whereas pigment films have been formed by thermal sublimation.4 Voc V 0. The following we list some ideas that have been published to improve the performance of double layer photovoltaic structures: 1. The EQE for devices marked with + was estimated from the information given in the respective papers. Singh have published an optical analysis method to optimise the structure of multi-layer thin film solar cells with respect to the thickness of the layers and the electrode materials [281.5 0. INTRODUCTION 69 Table 4. Their model takes into account multiple reflections of multi-junction structures composed of absorbing and non absorbing materials. and n stands for the doped and intrinsinc layer of amorphous silicon. a phthalocyanine) behind.4. EQE % 30+ 23 23 9 6 ≈ 3+ 2 0. Choosing suitable electrodes and thicknesses of the involved layers in such a device allows to optimize the optical field taking into account interference effects at the active interface and thus enhance the exciton concentration within a layer with thickness comparable to the short exciton diffusion length. a higher workfunction material than Al might then be required to avoid the formation of blocking contacts. such structure can show better efficiencies than the reverse order of layers [268. azulenium. It has been shown that the device efficiency of a double layer (PEOPT/C60) is strongly connected to the geometry and optical field distribution [213]. 3. The abbreviation for materials are listed in Chapter 1. quinacridone. As possible reasons5 for such an improvement may also be considered: longer exciton diffusion lengths in the high bandgap material or simply lower thermalisation losses of the photoexcited carriers in the first layer. Best parameters are printed bold while results discussed in this Chapter are printed italic.1: Survey of performance parameters of various organic double layer solar cells. glass/ITO/p/i/n/rear-contact where p. Zhu and J. 187].0 1. F. They have applied their analysis on e.2 and 3 refer to different perylene pigments.12 0.5 0. a perylene derivative) next to the transparent (ITO) contact and the lower bandgap material (e.8 1.6 0. Although.1. 2. 5 We . Another possibility to optimise organic double layer cells is to place the material with the higher bandgap (e.3∗ 0.35 FF % 65 ≈30 40∗ 48 60 23 51 25 20 range nm 400-800 400-700+ 400-600 400-560 400-620 400-620 400-590 400-800 400-850 materials D/A CuPc/Per1 HPc1/Per2 PEDOT/PEOPT/C60 PPV/C60 PPV/Per3 PPyV/P3HT PEDOT/PPV/Per CuPc2/Per HPc/Per3 comments – pigment/pigment [244] pigment/pigment [131] polymer/pigment [213] polymer/pigment [111] polymer/pigment [113] polymer/polymer [242] polymer/dye [9] dye/dye [200] liquid crystal(dye)/pigment [199] year 1985 1990 1998 1996 1996 1997 1999 2000 1999 uid crystals or polymers in solar cells.0 1. Many dyes are commercially available from suppliers for the copy industry (Xerography) like Synthec/Synthone in Germany but also laser dye manufacturers like Kodak. Per denotes the soluble perylene derivative while Per1. Their method may be of interest for application on organic multi-layer structures to maximise the effects of the incident radiation.

the mesophase (372K) [21].1-0. In practice liquid crystalline molecules had plenty of opportunities to rearrange themselves during purification steps (e. Morelength in most organic semiconductors represents a over. recrystallisation) or simply during the last syntheAs mentioned above the small exciton diffusion sis steps which typically involves solutions. The characteristics (IV. the solid phase for charges in discotic liquid crystalline materials allows even closer packing and even less disorder [1. One device has been heated and slowly cooled in order to obtain the alignment of the discotic molecules of HPc. Charge separation typically occurs at the HPc/Per3 interface so that electrons and holes can travel individually without a high chance to meet a recombination partner. PR-TRMC studies have revealed that is interpreted as increased mobility of the excitons both solid (crystalline) and mesophases of many in the mesophase so that they can now reach more discotic molecules allow have higher charge carrier nonradiative decay sites 6 . This quenching effect Indeed. acetylene units has also increased the PL[] 4. but their use. DOUBLE LAYER DEVICES Figure 4. More 40o may have seen sufficiently high temperatures recently researchers have discovered not only long during their storage or transport period prior to diffusion lengths (several 100nm) for excitons. Thus. It is the strong intermolecular interaction of the Although the PR-TRMC studies seem very conflat disc-like molecules which acts as driving force clusive there are also signs for poorer transport to induce a higher degree of order.see Chapter 3.3: The figure shows the chemical structure of the two organic semiconductors (HPc.g. 21]. For example. column chromatography. In other words it is a priori not clear whether The same study revealed also that the crystalline or not the effort to orient the columns perpendicphase of the same liquid crystalline materials shows ular to the electrodes pays off in terms of photoeven higher numbers than the mesophase of a given material.5cm2 Vs−1 ) molecules at lower temperatures i. as shown here. Our own studies of HPc in single layer de6 We note that breaking the conjugation(reducing the exvices are consistent with these improved transport citon diffusion length) of PPV chains by the introduction of properties at room temperature.2K) which can be found in the liquid crystalline phase which is considerably quenched at the transition to (mesophase). EQE) of this device were compared with the figures of a non-heated reference cell. Per3) and the architecture of the double layer device structure. tion of such a highly ordered phase requires a ceralkoxy substituted metal free phthalocyanine shows tain degree of mobility of the molecules themselves very high PL (50%) at low temperatures (4. The formaproperties in the crystalline phase. materials with transition temperatures <≈ serious limitation for double layer structures. once the molecules have arranged themselves.e. 57.70 CHAPTER 4. mobilities than in most pigments or polymers [57]. which is expected to affect the charge and exciton transport properties. The decreased motion (vibrations) of the also particularly high mobilities (0.2 The Dye/Pigment Double Layer Device (HPc/Per3) . only the exciton diffusion range determines the width of the active region in such a structure.

voltage curves and photocurrents were measured using a Keithley 237 source measure unit. this device may in fact be regarded as a hybrid between a single layer device (ITO/Per3/Al) and a double layer device (ITO/HPc/Per3/Al) with varying thickness of both organic layers. Hence.2 the difference of charge carrier mobilities in HPc at the solid to mesophase transition is only very small. One advantage of liquid crystalline phases is that they are self-repairing and can thereby minimise defects that can act as recombination sites. This suggests that a considerable part of the HPc layer absorbs light but does not contribute to the photocurrent (optical filter effect).for the pristine device . HPc. The Al top contact was obtained through thermal evaporation.see Chapter 7 we found that only one compound. much thicker droplets which are surrounded by areas of no or very little material are created. Quantum efficiencies were determined by normalisation with a calibrated silicon photodiode at the sample position. The transition into the liquid phase as well as the formation of a few linear defects . After that we sublimed a thin layer of Per3 as electron acceptor and transport layer. However. Exposure to the illumination source (tungsten) has been minimised to avoid photo-degradation effects. then 20o C/min) cooled down to room temperature using a hot stage (Linkam). we heated the phthalocyanine until it forms an isotropic liquid (clearing point =292o C) and cooled it slowly down to room temperature . The particularly high mobility of the molecules in the liquid phase allows the formation of the perpendicular arrangement at the transition to the discotic phase as observed for HPc in Chapter 7 or for a similar derivative in Ref.4. As a consequence. already the high solubility that allows simple spin-coating like normal dyes . HPc was dissolved in chloroform (30mg/ml) and filtered with 0. Reflection losses were neglected for all measurements. Current. (3. such a device has not been investigated yet and it should at least be possible to see to what extent the two materials can contribute to the photocurrent8 in such a hybrid structure. The thickness of the active layer can be estimated using Eq. The sample was kept under vacuum (< 10−5 mbar) during the measurements. instead of a thin film. After spin-coating HPc onto ITO.1mm PTFE-filter.3) and gives about 180 ± 90nm. Monochromatic illumination was provided by the output of a tungsten lamp dispersed by a Bentham M300 single-grating monochromator. In order to avoid these shorts the electron acceptor Per was sublimed prior to the deposition of the Al top contact. Absorption spectra were taken with a UV-Vis spectrometer (HP 8453). Hence. 71 by the sublimation of an Al layer as top contact. This value is in good agreement with . HPc. it is not possible to probe the effect of alignment via comparison with a pristine device directly7 . is also soluble enough to allow spin coating from solution and sufficiently viscous to form excellent films on the ITO substrates despite the relatively low molecular mass compared to polymers.illumination through ITO results in an EQE spectrum with maxima where the absorption is low and EQE minima at absorption peaks. Results and Discussion Absorption and EQE Spectra Fig.see Chapter 3 (visible with crossed polarisers) has been observed simultaneously under the microscope (Vickers) which has been adapted for temperaturedependent polarisation microscopy. One HPc film on ITO was heated to 292o C (20o C /min) under a constant nitrogen flow and slowly (5o C/min until 270o C.see Chapter 3 and Ref. Also the exciton diffusion length is likely much longer than the layer thickness . After characterisation of a number of mesogenic phthalocyanines . Thus.is technologically very interesting.2. The double layer device was then completed 7 Note that according to Ref. forms a liquid phase while all the others started to decompose before the isotropic liquid can be formed. The thickness of the HPc film after spincoating was 450nm and the Per3 film was about 20nm. THE DYE/PIGMENT DOUBLE LAYER DEVICE (HPC/PER3) voltaic performance.[152]. In any case.4. subsequent sublimation of the Al top contact inevitably results in a shorted device.4 shows that . During the transition from the solid to the liquid phase the surface tension leads to the destruction of the continuous thin film. Experimental Films of HPc have been fabricated in air by spin coating from solution onto indium tin oxide (ITO) coated glass substrates. Then we compare the external quantum efficiency (EQE) of this device with the same device architecture which has not been heat treated.[57] see Table 7. Thus a dramatic change in performance is not expected even if a non hybrid double layer could be fabricated. 8 Such a distinction is possible since Per3 does not absorb around 650-850nm but HPc does.[21].according to the procedure described in Chapter 3.

72 CHAPTER 4.5: IV-characteristics in light at 570nm and 0.4. This is expected in this device since the Per3 layer is rather thin (about 20nm) so that a filter affect due to a passive Per3 layer can not be very strong.19mW/cm2 of the heated device and the pristine device. Increased conductivity of the thick HPc droplets may here increase the contribution of HPc. • Heat treatment to align the discotic columns perpendicular to the electrodes could not improve the efficiency. The heated device shows higher currents for small voltages and also little rectification. the 200 ± 80nm obtained from the measurements of the single layer HPc . Figure 4.4. The latter dominate the absorption spectrum leading to the shown low OD numbers. The Voc is 350mV which is slightly smaller than in devices made of insoluble molecules. the parts of the device which contain only Per3 seem to be responsible for the low Voc . The heated device using the same materials and structure was heated until it turned into a liquid at 292o C and slowly cooled down to room temperature to obtain the perpendicular orientation of the discotic molecules.5 and 1% which is about 30 times smaller than comparable devices using only pigments.similar to typical pigment devices. DOUBLE LAYER DEVICES contact the EQE reaches up to about 1%. Illumination through ITO. Interestingly the absorption spectrum of the heated device is broader in the range where HPc absorbs only (650-850nm). . The photo-response spectrum of this device is very broad. The pristine device shows only little rectification but an Voc of 350mV. If we consider the transmission losses of the Al The IV curves in vacuum for both the pristine and the heated device are shown in Fig. (7. This may be explained by the formation of a “counter”-diode in conjunction with a vanishing HPc layer for parts of the device. Interestingly the photocurrent of the heated cell reaches a maximum after about 4 hours exposure to air [199]. We note that the shape of the IV curves can be changed considerably upon exposure to air approaching the shape of the pristine device [199].1. The low OD of the heated device is expected since droplets have been formed leading to very thick but also very thin layers. Summary In this Section we have shown that: • A layer of discotic liquid crystals can be used to manufacture a working D/A double layer photovoltaic device.4: EQE and optical density of the pristine (solid lines) and the heat treated (dashed lines) device. • The efficiency (EQE) of the pristine device is between 0. The Voc is only about 50mV.see Table 4. The redistribution of oscillator strength into the low energy region may be due to the high ordered phase of HPc.5.2). However.4 shows also that the heat treated device only reaches about 1/10 of the EQE of the pristine device. Fig. see [199]). In addition the heated device reveals a stronger voltage dependence in both reverse and forward bias. The Voc was 350mV . IV Characteristics Figure 4.see Eq. This number is not very different from the EQE obtained in double layer structures containing at least one polymer layer . Upon illumination through the semitransparent Al contact (transmission about 2%) the EQE of the pristine device follows the features of the absorption spectrum (not shown here.

The D/A double layer cell reported by Tang who sublimed insoluble CuPc and perylene derivatives onto ITO with a silver top electrode is still one of the most efficient organic solar cells [244]. we have found that the absorption coefficient at the absorption maximum of HPc is about 20% smaller than in typical phthalocyanine pigment films.3. The device was prepared in air but measured after 1 hour in vacuum. they are also likely to change the intermolecular arrangement in the solid (film) which can affect both charge and exciton transport and the optical properties of the films. Per/chloroform solution (30mg/ml) was spin coated on top of the CuPc layer with subsequent sublimation of the Al electrode.4.6.6: Chemical structure of the two dyes used in this section: A perylene tetracarboxylic acid-bis imide with aliphatic side chains (Per) and a copper phthalocyanine with polar side chains (CuPc).7: Linear optical absorption and EQE of the double layer device. even though only about 10% of the incident photons around 700nm are absorbed the EQE spectrum is almost constant between 450nm and 720nm. That way we could circumvent the thermal sublimation process and simply manufacture the device at room temperature via spincoating.3 The Dye/Dye Double Layer Device (CuPc/Per) double layer device (Fig. Since these side chains are not electronically active. Indeed. In this Section we investigate a double layer solar cell using soluble derivatives (see Fig. Experimental The chemical structures of the two dyes used in this section are shown in Fig. The CuPc solution is not very viscous despite the high concentration and did not only give very thin films with moderate quality. Results and Discussion Absorption and EQE Spectra Figure 4.4. However.4. However. the thickness of the active region. Per and CuPc are commercially available (Synthec. This indicates that unlike the 15nm thin CuPc layer the Per layer can not facilitate dissociation of all absorbed photons. Fig. which is the exciton diffusion range of both the Per and the CuPc layer. The Per film was about 60nm thick and the CuPc film 15nm.4. we might expect that these molecules would have very similar electronic properties to those in Ref. Aldrich). [244].4. The EQE around 700nm seems to be enhanced compared to the rest of the spectrum. The EQE roughly follows the absorption spectrum suggesting the absence of significant filter effect. The Figure 4. These side chains do alter the mechanical properties since they prevent aggregation in solution although they do not induce liquid crystalline phases here. THE DYE/DYE DOUBLE LAYER DEVICE (CUPC/PER) 73 4. is likely shorter than the 60nm thick Per film.1 at wavelengths where Per does not absorb (650-800nm). Because of the very thin CuPc layer (15nm) the optical density is less than 0.6) of Tang’s pigments.1) was fabricated by spin coating onto an ITO/quartz substrate from a solution containing 30mg/ml CuPc in distilled water followed by 10min annealing at 150 o C in air. Hence. A more efficient device may be obtained by re- .7 shows the EQE spectrum together with the optical absorption spectrum of the dye-based double layer cell. The essential differences to Tangs pigments are the polar side chains attached to the phthalocyanine structure and the short but bulky aliphatic side chains of Per.

doping are necessary to estimate the potential of these molecules.74 ducing the thickness of Per and increasing the thickness of CuPc. We note that Per tends to form crystals that may easily penetrate the thin CuPc film and thereby favour a lower10 Rsh . CHAPTER 4. Moreover the hyper-linear increase of current for negative bias suggests the presence of a diode with reverse polarity (=“counter”-diode). This can be used if the device is used as a photo-diode with a broad spectral response .8: IV characteristics in the dark and under illumination at 610nm and an intensity of 0. judged from the difference between dark and light characteristics. The device shows a very broad spectral response for a solution processed cell. An analysis of the slopes at -0. Both Rs and Rsh appear to play an important role as limiting factors in these devices.25mW/cm2 . This may be exploited if used as a photodiode . In Chapter 2 we have seen that the effect of the series resistor becomes important if the shunt is not at least an order of magnitude 9 We 4. Figure 4. Although the Rs is not unusually high9 the shunt is relatively low. Thus. 2. The thickness of the CuPc layer was about 15nm and the Per layer was about 60nm thick. The surprisingly high Rs for a double layer structure may be due to larger intra-molecular barriers (hopping distance is increased by the insulating side chains) that limit both the charge and exciton transport in the investigated material. We note that according to the IV curve the EQE. Ca instead of Al electrodes. 3. Although there is only little rectification for all shown curves certain features are interesting: The linear IV characteristics for small positive voltages indicates that the series resistant affects the shape and FF in this region. The exciton diffusion length of Per is smaller than 60nm whereas the number for CuPc is likely larger than 15nm.4. DOUBLE LAYER DEVICES higher.7 · 107 Ω. The small shunt compared to Rs may also be responsible for the low Voc . Note that the fairly symbatic response spectrum indicates that the exciton diffusion length of CuPc is possibly higher than 15nm. .g. can be more than doubled if a small negative bias voltage is applied.can be used to built double layer photovoltaic cells. A whole new class of organic semiconductors small soluble organic molecules (dyes) . polymer devices and about 30 times lower than the pigment double layer structures. 5. IV Characteristics The (IV) characteristics in the dark and under illumination are shown in Fig. The efficiency of dyes in these first studies is about an order of magnitude lower than e. Voc for other double layer devices at comparable intensities is typically a few hundred mV. found values around 0. It even extends into the negative voltage range so that the EQE at 0V is directly affected and reduced by resistive losses.5V gives values of Rsh =1.4 · 107 Ω and Rs =0. This device develops a relatively low open circuit voltage of 120mV.8.3V and +0.2107 Ω in MEH-PPV devices 10 Also the Na+ ions may contribute to the higher conductivity.where the sensitivity can also enhanced by at least a factor 2 by applying a small bias voltage. The reason for a low shunt resistor can be recombination at the interface or local shorts. We also note that the CuPc film also contains Na+ ions which may affect the electronic properties. Summary In this Section we have found that: 1. further studies that focus on reducing Rs by e.g.or as solar cell with e.g. The highest EQE would then be expected when both layers are as thick as their respective exciton diffusion lengths.

Chapter 5 Blend Layer Devices 5. the main advantage of the blend device over double layers is that the active layer can be considerably thicker than the sum of the exciton diffusion lengths in the D and A material as long as the blend comprises an interpenetrating network that not everything could be improved.1). for example. In such double layer structures exciton dissociation takes place at the D/A interface. Thicknesses of the blends vary between 20 and 120nm. with domains not larger than twice the exciton diffusion length. In addition. a single layer can only use about 7% and a double layer about 14% of the incident light .e.e. Even with a very low bandgap. If processing from solution is desired.1: Solar cell architecture comprising a blend of two organic semiconductors as used in this Chapter. Consequently. the layer thickness has to be at least 150nm of the pure material to approach 100% absorption over a broader wavelength range. the spectral range of the photoresponse can be greatly increased if the absorption spectra of the components (D and A) complement each other. they have the potential to utilise 15 times more photons then typical single layer structures and 7 times more than double layer devices. At the same time the acceptor material transports the electrons to the negative electrode (Al) while the hole acceptor provides a pathway for the holes to the positive electrode (ITO). However. mixtures of donor and acceptor materials (Fig. the thickness of the active layer can be increased compared to single layer structures since excitons created within their diffusion length from both sides of the D/A interface can now contribute to the photocurrent. the open circuit voltage is generally higher in single layer devices than in double layers. 1 Note Figure 5. The latter is typically only about 20nm (2x10nm).1 Introduction In the previous Chapter we have seen that the introduction of an electron acceptor (A) i.assuming an exciton diffusion length of 10nm in the constituent material(s). However. Both are spatially well separated so that mutual recombination of the photoexcited charges during their journey out of the device is reduced. Since both blend and double layer structures use two materials that can form a charge separating D/A interface they share many advantages over single layer devices . electron transport layer between the negative electrode (Al) and the electron donor material (D) can increase the EQE with respect to single layer architectures1 . blend devices have also drawbacks as can be seen from the following list: 1. Furthermore. If we consider both the spectral dependence of the absorption as well as the absorption coefficient of the majority of presently available organic semiconductors. since blends could have active regions as thick as necessary to absorb most light.for example: separate charge transport and a thicker active layer. In this Chapter we discuss another D/A-device architecture: Blends i.5. blends require high solubility of both components2 that double layer have similar restrictions: they require at least one component to be insoluble in the other components solvent or high thermal stability in order to use 2 Note 75 .

blends generally show stronger field (voltage) dependence of the photocurrent which often decreases the fill factor and may cause lower Voc .001% (CN-PPV) and 0.only little more (0. 3. [278] which is readily soluble in more common solvents and therefore rather a dye than a pigment. Photoexcited charge transfer from the polymer donor to the C60 is about 1000 times faster5 than radiative and nonradiative recombination[83]. “internal resistive” losses6 do not allow the use of thicker devices to circumvent the 40% transmission and reflection losses at the absorption maximum in these devices.in contrast to the C60 derivative in Ref. The random network of a blend device benefits strongly from a driving force that directs the separated charges from the D/A interface to the respective electrodes. most blends tend to develop lower open circuit voltages than single layer structures .2 dichlorobenzene).for reasons which are not well understood to date. In Table 5. However. Another problem here is the higher polarity of the electron transporting CN-PPV that favours it’s accumulation on the positive ITO electrode4 during the spincoating process [114]. Thus the D/A interface between MEH-PPV and C60 serves as an efficient exciton dissociation site. The difference between the workfunctions of the electrodes and/or an externally applied electric field (bias voltage) can represent such a driving force. • The highest EQE values (29%) among blend devices have been achieved with a blend between MEH-PPV and a soluble C60 (Fullerene) derivative as electron acceptor sandwiched between ITO and Al [278]. Despite the mentioned obstacles for further improvement. on the other hand. The table also illustrates that there are problems to realise all of these performance parameters in the same device. open circuit voltage) reach i. the achieved performance with the 4 Molecular dipoles on the ITO surface attract the polar CN groups. phase separation on a scale near or smaller than the double exciton diffusion length in the respective material. . The blend of MEH-PPV with the hardly soluble C60 could only be accomplished in a very strong solvent (1. the shunt must be large enough to allow the device to develop the relatively high open circuit voltage (0.6V with fill-factors of 25% were obtained. even surpass the numbers of their silicon counterparts using the blend architecture. Note that already two of these parameters (sensitivity range. BLEND LAYER DEVICES • The first report on D/A solar cells with conjugated polymer blends has been reported by Halls et. In the following list we will comment on many of these structures in order to point out specific novelties and strengths3 : thermal sublimation for film deposition. The low fill-factor is most likely due to the a relatively small shunt that also causes the field dependence of the light current. Since C60 is virtually insoluble in most known solvents we regard it as pigment within this thesis . The MEH-PPV/C60 blends also show the field dependent photocurrent that leads to low fill-factors (here still only about 30%) typically encountered in blend devices. Despite the larger active region. 6 Probably related to charge and exciton transport problems through the C60 network. al.68V using ITO and Al electrodes but . For maximum performance.04% (MEH-PPV) respectively. Under high monochromatic light intensities (20mW/cm2 ) this device develops a Voc of 0. The EQE of the single layer structures is 0.e.e.76 2.1 we give a survey of the four key parameters for solar cell efficiency in various blend structures extracted from our own results and from literature.6V). However. Phase separation on the scale of 10-100nm was demonstrated and a Voc of 0.interestingly .82V) with Ca instead of Al. [112]. As a consequence. the EQE of the best organic cell (29%) is yet only a third and the best fillfactor in blends (44%) about 50% of typical commercial silicon solar cells. 3 Here we follow the historical order CHAPTER 5. Blends comprising an electron accepting (CN-PPV) and an electron donating (MEH-PPV) polymer have shown EQE values of up to 6% which boosted the EQE with respect to single layer cells by about two orders of magnitude. Double layer structures. Unfortunately. are less dependent on external fields and do therefore not always benefit from an extra driving force. blends require a defect-free interpenetrating network for efficient percolation of charges to the electrodes i. 4. 5 Radiative and nonradiative decay of singlet excitons occurs typically between 100 and 1000ps [102].

7 0.01 <0.2 0. INTRODUCTION 77 Table 5. Best parameters are printed bold while results of this work are in italic.0 0.02 0. The EQE however.5 0.9V) instead of the expected shorts in the only about 200nm thin films.e. The surprising fact that the formation of such large crystals does still allow the device to perform reasonably well.consistent with the results and conclusions in Ref.1: Survey of performance parameters of various solar cells comprising at least one organic component in a blend. Apart from the small shunt resistant that causes the field dependence. have also been used to fabricate solar cells. The abbreviations for materials are listed in Chapter 1. sandwiched between ITO and Al.0007 Voc V 0. for example. Phase sensitive 7 Both AFM images revealed the formation of a large electron conducting network consisting of very long (µm) Per crystals in addition to phase separation on a scale of 50nm upon annealing.4V and an EQE of about 1% [62].6 <0. dispersed in chloroform . upon changing their diameter from 6 to 2nm.35 0.4 0.6eV. We note that a further breakthrough towards economically attractive organic devices has been achieved with the introduction of the laminated device structure . blends of MEH-PPV and a perylene dye (Per) as electron acceptor have been investigated. Although the MWNTs were as long as 2µm (and only 10nm in diameter) and show relatively high conductivity (8S/cm) the devices developed the highest open circuit voltage in our list of blend devices (0.6 0. EQE % 29 23 12 11 6 1.particularly with respect to the use of non polymeric semiconductors such as dyes which are prone to form crystals in the solid state.5 0.26 [102]. • A blend between a polymer (PPV) and multiwall carbon nanotubes (MWNT’s) has been investigated with respect to its photovoltaic properties in Ref.0 to 2.2 0.8 1.see below. charge transport between the nanocrystals seems to be the limiting factor in these cells.01 FF % ≈25 ≈33 26 41 25 23 25 44 25 25 25 25 25 range nm 400-550 400-550 400-640 400-620 400-600 400-500 400-800 400-600 400-850 400-900 400-800 400-1000 400-1100 materials – MEH-PPV+C60 MEH-PPV/C60 MEH-PPV+CdSe P3HT+Per MEH-PPV+CN-PPV PPV+MWNT HPc+Per MEH-PPV+Per PTV+Per PTV+Ter PTV+THPF PTV+PIF HPc+PIF comments – polymer+dye [278] polymer+pigment [83] polymer+nanocrystals [102] polymer+dye [65] polymer+polymer [112] polymer+nanotubes [3] liquid crystal+dye [200] polymer+dye [62] polymer+dye polymer+dye polymer+polymer polymer+polymer polymer+liquid crystal year 1995 1998 1996 2000 1995 1999 2000 1999 2000 2000 2000 2000 2000 C60 cells was sufficient to stimulate commercial interest in organic solar cells. This small but significant improvement is considered to be due to the high workfunction of the nan- can be dissolved i. • Blends of MEH-PPV with CdSe nanocrystals7 .see Chapter 6. Voc of 0. opens new possibilities .4 0. The devices show EQE values up to 12%.4 0. These nanocrystals turn out to be good electron acceptors with the extra advantage that their absorption properties can be tailored by varying their size during the synthesis.1 0. This confirms the feasibility of fabrication of working solar cells using materials containing or forming large crystals . The IV characteristics show that a fill-factor as high as (44%) can be achieved in blends despite the relatively low Voc of 0. is only about twice as high as in the pure PPV device. the energy gap can be increased from 2.12 0.5.4 0. Indeed blends with 2µm long nanotubes have been made . [3].1. • More recently.9 0.5V and again the strong field dependence leading to a fillfactor no larger than 0. [62] as mentioned above.

. we are not aware of any reports of photovoltaic D/A blend structures using solely dyes before this work. 251]. 8 Polymers CHAPTER 5. efficient recombination representing “deep traps”. ranging from a factor of 5 for MEHPPV+CN-PPV blends to a factor of 104 for MEHPPV+C60 blends. “dead ends” of aggregates in blends) • formation of blocking contacts at the electrodes • charge carrier traps that can lead to recombination or built up space charges that limit the current Large flat aromatic molecules like perylenes.78 otubes (5. In this Chapter we investigate the photovoltaic properties of blends comprising the materials discussed in Chapter 3.g. strong photoluminescence quenching has been found.see Fig.2 HPc+Per In this Section we explore the possibilities to form a D/A blend network in a photovoltaic device using the two small soluble molecules (dyes) Per and HPc .sometimes even more . Hence. Although there are reports of dye doped9 polymers [276] and and mixtures of D/A pigments [264.g. 5. the dominant loss mechanism has to be on the way of the charges to their respective electrodes. This indicates that the vast majority (over 80%) of photoexcited excitons can reach a dissociation site (D/A interface) within their diffusion length and dissociate before they can recombine under the emission of light (PL). trap levels in the bandgap of a low bandgap polymer are likely relatively close to both the HOMO and the LUMO so that they can facilitate fast i. We note that HPc also exhibits liquid crystalline properties as discussed in Chapter 7. • The first D/A blend between two dyes with one showing liquid crystalline properties at roomtemperature has been published [200] and will be discussed in detail in this Chapter.1eV.77eV (700nm) to 1. The EQE however is yet only modest (0.02%) and in one case (EQE=7 · 10−4 %) even smaller than the single layer device. distortion of conjugation along the polymer backbone by e.g. that are planarised through bridges between i. Both molecules have been investigated in both single and double layer devices earlier in this thesis. Syntec/Synthone in Germany.12eV (1100nm)and most of them are flat aromatic molecules. aromatic units are said to resemble a “ladder” along their backbone. Both devices will be discussed later in this Chapter. various soluble derivatives of pigments (dyes) with promising photovoltaic properties can be purchased from suppliers that specialise in chemicals for xerography e.e. • Blends with the low bandgap semiconductor PIF allowed for the first time to extend the spectral response of an organic D/A blend beyond the 1000nm range.2. Presently. For all these blend structures discussed above.g. On the other hand. The charge transport to the electrodes can be affected by: • inter-molecular barriers (e. twisting of aromatic units) • super-molecular structural defects (e.1eV) and/or the increased interface area due to the formation of an interpenetrating network. long distances for the hopping process) • intra-molecular barriers (e.5. 600nm) all components have low bandgaps ranging from 1. Two well soluble small molecules like Per and HPc are likely to mix on a molecular level to form a solid solution rather than showing phase 9 Note that mixtures containing less than about 10% by weight . The specific combinations of semiconducting materials has been chosen with the aim to accomplish both charge transfer which would indicate a D/A interface and a extended photoresponse covering at least the visible range (400-800nm).of the guest material are often regarded as “doped” materials in the field of organic semiconductor devices. With the exception of Per (2.g. BLEND LAYER DEVICES It is also noteworthy that low bandgap materials are more likely to form non-blocking (ohmic) contacts with metals simply because the workfunction of the electrode material is less likely to lie within the bandgap. phthalocyanines or ladder-type structures8 like LPPPT have a good chance to overcome the problems described in the first 3 points in the list above. The reason for the lack of published work on this interesting subject in literature is presumably related to difficulties in synthesising of soluble semiconducting molecules with the desired optical and electronic properties.e.

Our hope was that these specific dyes would allow sufficient percolation of charges through molecular clusters (small aggregates) since both show strong aggregation as individual components. Details of the procedure can be found in Ref.3: The absolute photoluminescence efficiency of the Per+HPc blend film on a quartz substrate plotted as a function of composition. Images were recorded using an atomic force microscope (NanoScope IIIa Dimension 3100) operated in the tapping mode which is located at and provided by Cambridge Display Technology (CDT). [169]. it is not clear a priori to what extent percolation paths are present in such solid solutions. Such an intimate mixture may then be regarded as a “new” material rather than a blend with a large D/A interface.2. The 1:1 (weight) blend device was fabricated by spin coating onto an indium tin oxide (ITO) coated quartz substrate from chloroform solution containing 30mg/ml HPc and 30mg/ml Per followed by sublimation of the Al electrode. We know that Per has a tendency to form crystals during spincoating from solution or upon heat treatment [62] while HPc is liquid crystalline at room temperature and forms even aggregates in solutions that quench the PL . We note that a double layer device with centrations resulting in the formation of crystals.see Chapter 7. However. In the following we try to shed light on these questions. Figure 5. Per was used as received from Synthec (Germany) while HPc has been synthesised and purified by G.2. Figure 5. The line shown represents a guide to the eye. segregation12 as usually observed between polymers. The excitation wavelength was 488nm (1.0mW light intensity). However. HPC+PER 79 Experimental PL efficiency measurements were performed using an integrating sphere coupled to a CCD spectrometer (Oriel) via a liquid light guide. we did not know if the dyes still prefer to form aggregates with like molecules13 and if the size of the aggregates would be on a suitable scale. 13 They might aggregate with the other material and/or a sublimed perylene derivative is discussed in Chapter 4. Too large aggregates for example would quench excitons before they reach the D/A interface whereas too small aggregates may not provide percolation paths.2: Chemical structure of the two dyes investigated in this Section: A perylene tetracarboxylic diimide with aliphatic11 side chains (Per) and a metal free phthalocyanine with aliphatic side chains (HPc). HPc does not fluoresce in the film due to aggregation (π-stacking). The chemical structure of the dyes investigated in this Section are shown in Fig. The liquid crystalline properties of HPc are discussed in Chapter 7 and have been published in 12 Phase segregation in dyes often happens at higher conRef. Lux as reported elsewhere [165]. Rozenberg and A. form exciplexes. .G. Charge carriers might be able to find their way to the respective electrodes via intermolecular hopping between like molecules even if they are not always in direct contact with each other. IV characteristics and photocurrent measurements were done in vacuum (< 10−5 mbar).5. [199].5.

. if the dyes do not form aggregates at all.3 for Per+HPc is more than one order of magnitude higher than in the 1:1 (by weight) blend between MEH-PPV and CN-PPV [112].5. Fig. Right: Scan size is 200x200nm. 267]. BLEND LAYER DEVICES Figure 5. For example.5: Linear absorption and EQE of the blend device. a 50% Per blend shows virtually no PL indicating that most excitons are affected by the presence of HPc. Results and Discussion PL-Quenching Photoexcited excitons can decay either radiatively showing photoluminescence (PL) or dissipate their energy via direct conversion into heat.80 CHAPTER 5. Thus possible aggregates can only be smaller than about 20nm.3 shows that in a film of 100% Per about 7% of the excitons decay under emission of light. Figure 5. We note that the PL quenching shown in Fig. the strong PL quenching in our dye blend is consistent with the formation of aggregates on a scale <20nm. the blend may be regarded as a solid solution (= blend on a molecular scale) rather than a phase segregated mixture. Since the aggregates in the latter are between 10 and 100nm. Left: Scan size is 2x2 microns.5. The lack of contrast suggests that there are no crystals or aggregates between about 20-1000nm. Moreover. The photoexcited excitons can undergo several processes: • recombination at the interface • energy transfer to the material with the lower bandgap followed by radiative or nonradiative recombination • charge transfer (exciton splitting) at the interface We consider the last process (charge transfer) to occur in these devices since we found enhanced photocurrents in the blend compared to devices made of the single components Fig.6.5. the interface of similar phthalocyanine and perylene derivatives has been recognized as the active site for exciton dissociation elsewhere [244.4: AFM images (phase contrast) of a 1:1 (by weight) blend of HPc and Per. In fact. This ratio can be considerably decreased upon the addition of HPc.

5.25mW/cm2 . There is significant recombination of the separated charges during their journey through the fine interpenetrating network Since we did not find any pinholes under the AFM and the device is not particularly thin (70nm) we may exclude the presence of shorts due to pinhole formation. The device shows little rectification in the depicted voltage range. Both the dark and the light current depend highly on the applied field. at the D/A interface) and the built in field drives the separated charges through percolation paths to the different electrode materials. Indeed. the formation of a very fine network may account for the observed strong PL quenching. the experiment reveals a strong linear voltage dependence under negative bias. A voltage and light dependent device like a photodiode can be formed if exciton splitting occurs (here Figure 5. However. IV Characteristics The (IV) characteristics in the dark and under illumination are plotted in Fig.5.2. Hence. both the light and the dark IV curves in Fig. From the fact that the EQE spectrum follows the absorption features we can infer that most of the bulk of the cell contributes to the generation of the external current.5. Thus we conclude that the shunt is too small which may be the case for the following reasons: 1.7 show a weak but significant diode behaviour under forward bias. The EQE spectrum reaches its peak (≈ 1 %) at around 500nm which exceeds the EQE of a double layer cell made with very similar materials (Chapter 4) by a factor of 2. The onset of the EQE is around 900nm as in the single layer and double layer devices comprising these components. Absorption and EQE Spectra The EQE and optical absorption spectrum of the 1:1 (by weight) HPc+Per blend cell are plotted in Fig. The absorption band between 600nm and 850nm corresponds to HPc whereas Per starts to absorb at around 650nm reaching its peak at 500nm. The blend show clearly enhanced EQE over the entire range indicating the presence of a D/A interface and percolation paths to the electrodes. 2. the separated charge carriers have many opportunities to recombine at this large but presumably not very conducting D/A interface. Instead.5.5 showing that the cell is able to utilise light from the entire visible spectrum.7. The open circuit voltage is 110mV and the fill factor 25 percent under these conditions. 14 Which is caused by the “fine” percolation paths. A short has been formed through a pinhole that results in a shunt resistor approaching the value of the series resistor. the enhanced photocurrents and the high series resistor14 .7: IV characteristics of the HPc+Per blend in the dark and under illumination with light at 650nm and an intensity of 0. .6: EQE of the HPc+Per blend device and the single components. HPC+PER 81 Figure 5. The thickness of this device was about 70nm.

82 This loss mechanism could be assisted by the presence of trap levels in the bandgap.

CHAPTER 5. BLEND LAYER DEVICES

Summary
In this Section we have found that: • Photovoltaic devices with a photoresponse ranging from 400 to 800nm can be built using a blend between two soluble derivatives of a perylene diimide and a phthalocyanine. • The high solubility of both molecules appears to facilitate the formation of a large D/A interface - possibly even a solid solution - that enables efficient exciton splitting. • Such a very fine interpenetrating network (<20nm )is probably also responsible for the small shunt resistor since it offers increased possibilities for recombination during the journey of the separated charges. • The small shunt may be the main reason for the relatively low Voc of 110mV and fill-factor (0.25).

Figure 5.8: Chemical structure of the two organic semiconductors investigated in this Section: A poly-indenofluorene with short aliphatic side chains (PIF) and a metal free (M=H2 ) phthalocyanine (HPc) with thio-alkyl side chains (R = C12 H25 ). While HPc is very flat, the short distance between the repeat units of PIF causes a 33o distortion between adjacent indenofluorene moieties [210].

Results and Discussion
Absorption and EQE Spectra The EQE and optical absorption spectrum of the HPc+PIF blend cell are plotted in Fig.5.9. The features around 440, 680 and 720nm are due to the HPc absorption whereas the peak at 800nm can be found in the absorption of PIF - see also Chapter 3. Both components have a small absorption coefficient around 550nm. However, since the absorption onset is indeed around 850nm for HPc and at 1040nm for PIF, the broad peak in the infrared (1040nm) is probably due to enhanced aggregation of PIF in the presence of HPc rather than poor background correction or interference effects in the absorption measurement. Another reason why the latter can be ruled out is that we observe a significant photocurrent in the absorption onset. We point out that the absorption range of this device is larger than the range of the spectrometer using a silicon based CCD array. The EQE spectrum follows the absorption features very well indicating that the entire bulk - involving both components - contributes to the generation of the external current. However, the EQE does not reach more than 6.5 · 10−4 % which is lower than in devices comprising the pure components only: 3 · 10−3 % in HPc as well as in PIF - see

5.3

HPc+PIF

Experimental
Fig.5.8 shows the chemical structure of the organic semiconductors used for the device fabrication in this Section. The liquid crystalline properties of HPc are discussed in Chapter 7 and have been published in Ref. [165]. PIF has been synthesised by H. Reisch and U. Wiesler at the MPI in Mainz, Germany [210]. HPc has been synthesised by G.G. Rozenberg and A. Lux, Melville Laboratory for Polymer Synthesis, Cambridge (UK) as reported elsewhere [165]. We note that a double layer device with a sublimed perylene derivative is discussed in Chapter 4 of this thesis and published in Ref. [199]. The ITO/HPc+PIF(1:1 by weight)/Al devices were fabricated by spin coating onto an ITO coated quartz substrate from chloroform solution containing 30mg/ml HPc and 10mg/ml Per followed by sublimation of the Al electrode. More details about preparation of substrates and the setup can be found in Chapter 10.

5.3. HPC+PIF

83

Figure 5.9: Linear absorption and EQE of the HPc+PIF blend device. Fig.5.10. The following reasons may be considered to explain such low external quantum efficiencies: 1. Pinholes in the device that can cause shorts when filled with e.g. Al during the sublimation of the electrode. 2. A defect-rich charge transport network in the blend. Such defects could be “dead ends”, isolated areas, accumulation of a homolayer on the wrong electrode. In addition, the blend ratio of 1:1 (weight) may be far from the percolation threshold15 . 3. The offset between the LUMO and HOMO levels between PIF and HPc is not larger than the exciton binding energy16 which is required to separate the charges i.e. to enable photoexcited charge transfer at the D/A interface. 4. The formation of new recombination channels for the photoexcited species upon mixing the two components. ad 1): Shorts caused by e.g. pinholes in the organic film can be ruled out since the IV characteristics shows very low currents in the dark - even under bias voltage (Fig.5.11). ad 2): In MEH-PPV+CN-PPV blends the EQE values are at least two orders of magnitude higher than the devices made of the single components 15 which 16 ≈

Figure 5.10: EQE of the HPc+PIF blend device and the single components. The blend shows a significantly lower EQE than each component - indicating that there is no charge transfer between HPc and PIF. despite serious network defects such as the formation of a homolayer of CN-PPV17 near the wrong electrode (ITO) [114]. In addition, network defects such as “dead ends” cannot account for EQE numbers lower than the single components because extra exciton splitting should always lead to enhanced photocurrents compared to the single layer. This is like introducing oxygen in a single layer device using a hole transport material - see Chapter 2. Molecular oxygen would be the electron acceptor with probably no percolation path at all - and is still better than the oxygen free material. Hence insufficient network formation cannot easily be blamed alone for the poor performance of the HPc+PIF blend. ad 3): The quinoid structure of PIF and the known donor properties of phthalocyanine derivatives suggest that the difference between the HOMO and LUMO levels of HPc and PIF should be sufficient to allow charge transfer. However, since both materials do not show enough PL we cannot test for PL quenching and have therefore no indication that excitons are split at the interface between these two molecules. ad 4): In fact, the combination of low bandgaps in both components and the (probably) different electron affinities may bring the LUMO of PIF rel17 The more polar CN-PPV is attracted by the polar ITO surface.

is difficult to estimate here 0.4eV according to Ref. [56, 168, 12]

84

CHAPTER 5. BLEND LAYER DEVICES (<10mV). The device also shows only little rectification in the depicted voltage range.

Summary
In this Section we have found that: • A working photovoltaic device comprising a 1:1 blend with low bandgap dye and a low bandgap polymer has been built. • The device shows a photoresponse in a wavelength range as wide as 400 to 1100nm which is virtually identical to commercial photodetector systems based on silicon. Figure 5.11: IV characteristics in the dark (dashed line) and under illumination with light at 700nm and an intensity of 0.35mW/cm2 . The thickness of this device was about 40nm. atively wide into the middle of the HPc bandgap. PIF with its low LUMO may then capture an electron to act as a “deep trap” and allow recombination with a hole in the HOMO of HPc (electron back transfer). The mirror process, photoexcitation of an electron from the HPc directly into the LUMO of PIF may then account for the new absorption band around 1100nm. Note that a pure PIF film does not absorb wavelengths longer than about 1000nm - see Chapter 3 and [221]. Recombination via this “deep trap”, PIF, could then even outnumber the extra photoexcited charge carriers if exciton splitting actually occurs at the PIF/HPc. • The blend between such low bandgap materials with different electron affinities may open new decay channels for the photoexcited species that lead to EQEs that are lower than in devices with only the single component. This may be seen as a first hint for the lower limits of the size of the semiconductor bandgap in organic solar cells. • The EQE can be tremendously increased by applying -2V bias voltage.

5.4

PTV+Per

Experimental

The chemical structure of the organic semiconductors discussed in this Section are shown in Fig.5.12. Per was purchased from Synthec (Germany) while PTV has been synthesised by F. Goldoni, UniverIV Characteristics sity of Technology (Eindhoven, The Netherlands). The (IV) characteristics in the dark and under illu- Details of the synthesis, solvatochromic effects and mination are plotted in Fig.5.11. The dark current circular dichroism can be found in Ref. [90]. is small and relatively independent of the voltage We note that single layer devices with both comin both forward and reverse bias in the depicted ponents are discussed in Chapter 3 of this thesis. range18 . Upon illumination the device generates a The 1:1 (weight) blend device was fabricated by small but significant photocurrent that can be in- spin coating onto an indium tin oxide (ITO) coated creased considerably (up to 15 times) by applying quartz substrate from chloroform solution containan electric field. We can identify a threshold (≈ ing 15mg/ml PTV and 8mg/ml Per followed by sub0.5V) under negative bias where the photocurrent limation of the Al electrode. More details about enhancement seems to begin. The device, however, preparation of substrates and the EQE/IV setup does not develop a significant open circuit voltage can be found in Chapter 10. The thickness of the or18 This rules out the presence of a short - as discussed ganic layer was measured with a Dektak profilomeabove. ter and found to be 80nm.

We consider that the presence of the rigid and planar Per molecules may favour a more planar arrangement of the PTV molecules which are ex- Figure 5.-1. pected to form a chiral superstructure19 in the pure film. since we find a significant photocurrent down to about 900nm.4. The EQE reaches 0. We note. Results and Discussion Absorption and EQE Spectra The EQE and optical absorption spectrum of the 1:1 (by weight) PTV+Per blend cell are plotted in Fig. However. 19 because of the chiral side chains [90] . In addition enhanced aggregation or even crystallisation may take place due to the flatter alignment which has been suggested to be responsible for enhanced exciton splitting leading to enhanced photocurrents in the absorption onset in MEH-PPV [120].13: EQE of the PTV+Per blend device with 0 and -3V bias and its absorption spectrum.and probably real feature with a fine structure in the absorption onset of the PTV+Ter device discussed in a later Section of this Chapter. we can assume that charge separation takes place at the PTV/Per interface.14: EQE of the PTV+Per blend device with 0. Interestingly. Such a more planar molecular arrangement may then lead to a lower bandgap due to increased conjugation compared to the pure PTV molecule.5 and -3V bias and the single layer device efficiencies at 0V. The two peaks around 500nm correspond to the absorption of a single Per layer whereas the shoulder at 650nm is due to an absorption maximum of PTV. there must be some absorption in that region.14 are clearly lower. the EQE spectrum of the blend extends further into the infrared than the EQE of the single components.5.44% in short circuit mode and 8% under 3V negative bias. The relatively strong and broad band between 800 and 1000nm is most likely an artifact and probably related to the low signal to noise ratio that affects the background correction. that we find a more pronounced .5.see Fig. PTV+PER 85 Figure 5. Since the EQE values for single layer devices made of the pure components PTV .13.5. The EQE spectrum follows the main absorption bands. This indicates that there is a long absorption tail in the blend film. Figure 5.12: Chemical structure of the two organic semiconductors discussed in this Section: A polythienylene vinylene with chiral aliphatic side chains (PTV) and a perylene diimide with aliphatic side chains (Per).

The light current.The chemical structure of the organic semiconducterface. quaterrylene-. The Netherlands). From the observation that photoexcitation reduces predominantly the shunt rather than Rs we consider that a new recombination channel near the D/A interface opens under illumination. light IV curve represents the sum of both Rs and Rsh as Single layer devices comprising the pure comdiscussed in Chapter 2 so that Rsh is possibly even smaller pounds have been investigated in Chapter 2. Summary In this Section we have found that: • The external quantum efficiency of a photovoltaic device comprising a 1:1 blend between a small soluble perylene derivative (Per) and a poly-thiophene derivative (PTV) is considerably higher than the EQE of the pure single layer devices. The linear dependence of the photocurrent tors used in this Section are shown in Fig. What is remarkable here is that the shunt of the dark IV curve decreases by at least20 3 orders of magnitude from 1. Becker at the MPI (Mainz. [90]. 83]. solar cell but also as light dependent resistor (photo-resistor).3 · 1010 Ω to 1.e. is highly field dependent following Ohm’s law between +3 and -3V.3 · 107 Ω upon illumination with light while there is no indication for a significant decrease of Rs . .synthesised by S.15: IV characteristics of the PTV+Per blend in the dark and under illumination with light at 550nm and an intensity of 0. Details not allow any rectification. The was synthesised by F. PTV shows that the current is not trap limited. • In short circuit mode the solar cell reaches an EQE up to 0. 102.15.16. • Thanks to the low bandgap of PTV the photovoltaic cell shows a very broad photoresponse surpassing the visible wavelength range.4% with Voc of 150mV • The shunt resistor decreases by at least 3 orders of magnitude upon illumination with monochromatic (550nm) light of 0. The low 5.5 PTV+Ter shunt can also explain the moderate open circuit voltage (150mV). Experimental The relatively high EQE (despite the low shunt) suggests efficient exciton splitting at the D/A in. The dark current is very low and virtually field independent over a wide voltage range as observed in many D/A blend devices. [217]. however. The thickness of the organic film is 80nm.3 · 107 Ω. the strong field dependence of the photocurrent can be used to increase the photosensitivity i.5.of the synthesis can be found in Ref. Figure 5. We note that the recti. than 1. BLEND LAYER DEVICES MEH/CdSe and MEH/C60 mixtures with low C60 ratios [112. deriva20 Note that the ohmic characteristics as obtained for the tives including polymers can be found in Ref.15mW/cm2 . EQE by a factor of 20 (reaching about 8%) by applying either -3V (photodiode mode) or +3V (photo-resistor mode). terrylene-. In any case. Goldoni at The University of ohmic IV characteristics of the light IV curve does Technology (Eindhoven. • The EQE can be increased by a factor of 20 (reaching 8%) by applying a negative or positive voltage of 3V.25. CHAPTER 5.86 IV Characteristics The IV characteristics in the dark and under illumination are plotted in Fig. The open circuit voltage is 150mV and the fill factor 0.15mW/cm2 . More information about the synthesis of various perylene-. • The distinct difference between the dark and the light IV characteristics makes it possible to operate this device not only as a photo-diode. many). Ter was fication ratios under light are also low for other or.5. Gerganic semiconductor blends MEH-PPV/CN-PPV.

5. 1.649. Also interesting is the occurrence of a broad absorption band between 780 and 1100nm where none of the single components absorbs at all. Two vibronic modes can be found: 165meV and 82meV.which is 83 times 2.18: Linear absorption of the PTV+Ter blend versus energy.5V bias and its absorption spectrum. The EQE at 0V is multiplied by 20 for clarity.16: Chemical structure of the two organic semiconductors discussed in this Section: A polythienylene vinylene with with chiral aliphatic side chains (PTV) and a terrylene diimide (Ter) with aliphatic side chains. Figure 5. In fact.320.878 and 2.see Chapter 3. Both PTV and Ter have very similar absorption spectra with features at 600 and 660nm and the onset at 750nm.237. We consider that these two peaks reflect the 0-0 and 0-1 vibronic transition with the 0-0 peak being the dominant feature.402. 1. 1.5. The spectrum of the blend in Fig.567. The band here is different from the one in PTV+Per in that the band here is stronger and structured. The two main peaks are at 1.5.5.17 shows qualitatively the same features but they are much more pronounced than in the single compounds . More details about preparation of substrates and the setup can be found on Chapter 10.485.5V of the 1:1 PTV+Ter blend cell are plotted in Fig. A similar vibronic mode of 50meV in addition to the dominant C-C mode of the polymer backbone (180meV) has been found in the absorption .18)reveal a constant energetic spacing typical for vibronic modes. The thickness of the organic film is 32nm. Figure 5.5. the same data plotted on an energy scale (Fig.17: EQE of the PTV+Ter (32nm thick) blend device with 0 and -1.17. We note that we have observed a similar absorption band in the PTV+Per blend as discussed in the previous Section in this Chapter.732 which correspond to an energetic spacing of 83meV. Results and Discussion Absorption and EQE Spectra The optical absorption spectrum and EQE at 0V and -1. We found peaks at 1. 1. 1. 1. PTV+TER 87 Figure 5. The 1:1 (by weight) blend device was fabricated by spin coating onto an ITO coated quartz substrate from chloroform solution containing 15mg/ml PTV and 6mg/ml Ter followed by thermal sublimation of the Al electrode.043eV with a spacing of 165meV .

5.: 1. IV Characteristics The IV characteristics in the dark and under illumination are plotted in Fig.see Chapter CHAPTER 5.20.19).see Chapter 3. the presence of impurities.88 spectrum of the well defined poly-phenylene derivative (m-LPPP). We believe it is more likely that the π − π interaction of the large planar Ter molecules favours a more planar (less disturbed) molecular arrangement of the PTV molecules thus leading to longer effective conjugation lengths and enhanced aggregation and the observed vibronic structure in the polymer. In fact. 1.g.e. The introduction of sub-bandgap levels via unintentional doping by e. The dark current is low and stays virtually constant under negative bias whereas the light current depends strongly on the applied field. 197.19: EQE of the PTV+Ter blend device with 0 and -1.unless they are the PTV+Per device discussed earlier. The latter had to be especially synthesised [109. Although this could explain the observation of both the well structured absorption spectrum and the broad absorption band around 900nm further structural investigation . 2. although we cannot identify the exact nature of this effect. EQE data for the pure Ter film may not be representative due to the formation of shorts . Both the occurrence of such well resolved vibronic structures and a dominant 0-0 transition have only been found either in small molecules or well ordered conjugated polymers like m-LPPP and PPV [198. the higher their photon energy (Fig.such as the one mentioned above . 3. in the range between 400 and 500nm the EQE with 0V increases despite the local decrease of absorption in this range. In other words. Since the band is very strong and does usually not appear in films of comparable optical properties we consider the first two possibilities (doping. Interference effects in the thin film. regardless of the vibronic structure such a band in the absorption onset may be caused by e. Hence. Another interesting observation is that photons absorbed in the low energy absorption band contribute to the photocurrent the more.55eV (800nm) photons are almost 100 times more efficient in contributing to the photocurrent than photons with only 1. We note that we found the same effect in the PTV+Per blend discussed in a previous Section. photons carrying higher energies contribute more efficiently to the photocurrent already in the pure PTV (containing oxygen and water) than photons with lower energy. However. 163]. 110] or stretch oriented [108] to obtain the required degree of order i. The device develops an open circuit voltage of 120mV upon illumination with light at 600nm. Under negative bias voltage the sensitivity to photon energy decreases considerably in both the high energy (400 to 600nm) and the low energy range (800 to 900nm). BLEND LAYER DEVICES Figure 5. Aggregation effects. we consider that the PTV component is responsible for this effect. interference) to be very unlikely. the photoall longer than the effective conjugation length. From the fact that the EQE spectrum follows essentially the absorption features we can infer that most of the bulk of the cell contributes to the generation of the external current. maximum conjugation length21 .5. In fact the pure PTV single layer shows a monoton increase of EQE with shorter wavelengths despite a local absorption minimum at 420nm . However. However.5 bias voltage and the single layer device efficiency of PTV at 0V. 3.g. sensitivity can be enhanced by a factor of 20 by .38eV (900nm). Similar to 21 Different lengths of conjugated segments on a polymer chain smear out the absorption spectrum .are desirable to confirm such a possibility. they could be ruled out by performing photo thermal reflection (PDS) measurements as part of a more detailed study of these types of blends.

Khan.2mW cm2 ).21: Chemical structure of the two organic semiconductors discussed in this Section: A polythienylene vinylene with chiral aliphatic side chains (PTV) and a fluorinated poly-yne. Details of the synthesis can be found in Ref. Figure 5.e. Figure 5. The chemical structure of the organic semiconductors discussed in this Section are shown in Fig.4% at 670nm by applying a negative voltage of 1. PTV+THPF 89 • The device can operate as a photodiode.8eV) with a vibronic mode (83meV).especially if the open circuit voltage is relatively low (< 300mV at typical light intensities around 0. The thickness of this device was about 32nm. • The absorption of the blend shows a sharp vibronic mode (165meV) with a dominant 0-0 transition indicating high molecular order i.20: IV characteristics of the PTV+Ter blend in the dark and under illumination with light at 600nm. THPF.8V) in the dark but virtually none under illumination. PTV was synthesised by F. This has probably to do with a tendency towards low shunt resistors in blends since separated charges have more opportunities to recombine within the large D/A network. The Netherlands).21. • The device shows a photoresponse exceeding the visible wavelength range (400-900nm) • The photoresponse can be increased by a factor of 20 reaching 2. 5. The synthesis has not been published yet but some information about D/A interaction in fluoro-substituted phenylene containing organic and organo-metallic poly-ynes can be found in .5. • The EQE reaches up to 0.6 PTV+THPF Experimental Summary In this Section we have found that: • The external quantum efficiency of a device comprising a 1:1 blend between a soluble terrylene derivative (Per) and a soluble polythienylene vinylene derivative (PTV) is considerably higher than in the pure PTV device. solar cell and a photo-resistor. In contrast to the PTV+Per cell the PTV+Ter device shows some rectification in the dark but virtually none under illumination. THPF was synthesised by M. applying a voltage of -1.2 to 1. [90]. Imperial College London.5V. Goldoni at the University of Technology (Eindhoven.5. • We find little rectification (5 at ± 1. long effective conjugation of the PTV in the presence of Ter.5V. Note that strong electric field dependence dependence of the photocurrent under negative bias is often observed in blend devices . We also found an extra absorption band at low energies (1.6.4% with an open circuit voltage of 120mV.

Since there is no photocurrent beyond 800nm we consider the band around 900nm to be an artifact or due to interference effects. Figure 5.90 Ref.23) shows clearly that the photoresponse actually starts at around 800nm and increases over more than four orders of magnitude when reaching shorter wavelengths. [141]. More details about preparation of substrates and the setup can be found in Chapter 10. The low bandgap of THPF shows that the triple bond is not breaking the conjugation of the backbone . The average molecular weight is 15440 corresponding to n=22 repeat units [142]. The semi logarithmic plot of the EQE (Fig. They do not absorb light with wavelengths longer than 800nm. the small improvement of the EQE compared to the pure THPF device may be explained in that charge separation is not particularly effective at the interface between those two molecules.5. Results and Discussion Absorption and EQE Spectra The EQE and optical absorption spectrum of the 1:1 (by weight) PTV+THPF blend cell are plotted in Fig.they are separated not only by the thiophene but also the pyrazine and phenyl units . CHAPTER 5.5. As observed in the pure PTV device and other Figure 5.thus charge transport along the backbone should be efficient. . The blend shows about 10 times higher numbers than in the pure PTV device but only little (2x) more than in the pure THPF cell.7eV (730nm) in solution.22. A sufficiently thick layer of such a blend can absorb 54% of the solar radiation on earth . Single layer devices comprising the pure compounds have been investigated in Chapter 3 of this thesis. This would mean that THPF is not a very strong electron acceptor as expected because of the electron withdrawing fluor atoms that couple to the conjugated system.weakens the electronic effect. The thickness of the organic film is 50nm.22: Linear absorption and EQE of the PTV+THPF blend device. Such a weak effect of the fluor atoms is also consistent with the observation of an unchanged bandgap22 regardless whether the fluor atoms in this particular systems are present or not [142].1% around 400nm. The 1:1 (by weight) blend device was fabricated by spin coating onto an ITO coated quartz substrate from chloroform solution containing 15mg/ml PTV and 16mg/ml THPF followed by sublimation of the Al electrode. The EQE numbers between 800 and 900nm are not significant due to the low signal to noise ratio in this region so that the actual bandwidth of operation of this device is 400 to 800nm.see Table 8. Both components have very similar absorption spectra with features at 600 and 660nm see also Chapter 3.3. The organic film has been measured with a Dektak profilometer and found to be 50nm thick. 22 The bandgap in both systems was found to be around 1. However.23: EQE of the PTV+THPF blend device and the single layer device efficiencies at 0V. BLEND LAYER DEVICES PTV blends the EQE increases for shorter wavelengths and reaches values greater than 0. It may be that the rather long distance between the fluor atoms and the conjugated polymer backbone .

g. The IV characteristics resemble the curves obtained for blends of PTV with Per and Ter of the previous Section but with the exception that the (light) currents are lower and Voc is higher.5. More details about preparation of substrates and the setup can be found in Chapter 10. the open circuit voltage in blend devices appears mainly to be determined by the workfunction of the electrodes and increases only very little with light intensity [102]. This device develops a relatively high open circuit voltage (550mV) which is close to the difference in workfunctions of the electrodes. Single layer devices comprising the pure compounds have been investigated in Chapter 3 of this thesis.2% with Voc of 550mV which is close to the difference of workfunctions of the electrodes.7 PTV+PIF Experimental The chemical structure of the organic semiconductors used in this Section are shown in Fig. • The photoresponse can be increased by a factor of 4 by applying a negative voltage of 2V. Similar open circuit voltages have been observed for e.as opposed to the rather voltage dependent light current.calculated from the slope in the 3 quadrant . • The device shows a photoresponse covering the entire visible wavelength range (400-800nm).5.24: IV characteristics of the PTV+THPF blend in the dark and under illumination with light at 550nm and an intensity of 0.24. The broad absorption band at 600nm stems from PTV whereas the peak at 800nm with the band Summary In this Section we have found that: • The external quantum efficiency of a device comprising a 1:1 blend between THPF and a . [90]. There is no significant rectification in the investigated range (3V to +3V). Wiesler at the MPI in Mainz. PTV was synthesised by F. The 1:1 (weight) PTV+PIF blend device was fabricated by spin coating onto an indium tin oxide (ITO) coated quartz substrate from chloroform solution containing 15mg/ml PTV and 8mg/ml PIF followed by thermal sublimation of the Al electrode. Another result often found in blends is the low dark current even under higher negative bias voltage .5. Germany [210].24 equals Rs + Rsh . PTV+PIF IV Characteristics The IV characteristics in the dark and under illumination are plotted in Fig. Indeed. 91 polythiophene derivative (PTV) is about two times higher than the EQE of the more efficient single material device (THPF).7. Figure 5. The thickness of the organic film was determined with a profilometer (Dektak) and found to be 55nm. 5. • The dark current is very small and voltage independent in contrast to the photocurrent. • The EQE reaches up to 0.26. The Netherlands). Reisch and U. The shunt . Results and Discussion Absorption and EQE Spectra The EQE and optical absorption spectrum of the 1:1 PTV+PIF blend cell are plotted in Fig. Note that the (inverse) slope of the ohmic IV curve in Fig.decreases considerably (from 1010 Ω to 6 · 108 Ω). The thickness of this device was about 50nm.5.25.2mW/cm2 . • Thanks to a linear IV curve the device can operate as a photodiode. PIF has been synthesised by H. MEH-PPV+CN-PPV blend devices under comparable conditions [112].5. • We find some rectification (5) at ±1. Goldoni at the University of Technology (Eindhoven.8V in the dark but virtually none under illumination. Details of the synthesis can be found in Ref. solar cell and a photoresistor.

An alternative23 and more likely possibility to explain the partial enhancement of the blend compared to the single layer is the presence of partial shorts i.92 CHAPTER 5. As observed in .28.e. we can conclude that photo-induced charge transfer takes place at the D/A interface but there must be another decay channel present that causes the device to lower its performance . A sufficiently thick layer of such a blend can absorb 77% of the solar radiation on earth which is comparable to commercial silicon based photodetectors and solar cells .possibly wavelength independent.27: EQE of the PTV+PIF blend device and the single layer device efficiencies at 0V. this device can utilize light with wavelengths ranging from less than 400 to up to 1100nm. The figure also shows that the EQE of the blend is higher than both pure components only around 600nm. a low shunt resistor.26: Linear absorption and EQE of the PTV+PIF blend device. IV Characteristics The IV characteristics in the dark and under illumination are plotted in Fig. Since the absorption spectra of the low bandgap material PTV and the very low bandgap semiconductor PIF complement each other. Figure 5.014%).3.27) reveals that the photoresponse of the blend really starts at 1100nm. 23 Another Figure 5. between 900 and 1100nm is due to the absorption of PIF . Thus.see Chapter 3. BLEND LAYER DEVICES both the pure PTV and the pure PIF device as well as other PTV blends the EQE increases for shorter wavelengths reaching its maximum around 400nm (0. Figure 5.5. The onset of absorption of PTV is around 790nm.25: Chemical structure of the organic semiconductors discussed in this Section: A polythienylene vinylene with chiral aliphatic side chains (PTV) and a poly indenofluorene (PIF).see also 8.5. The semi logarithmic plot of the EQE (Fig. From the symbatic photoresponse we can exclude charge separation at the ITO/blend interface. This would be consistent with the results of the IV measurements see following Section. Although the device shows a more than significant photoresponse we believe that the device is possibility is that the organic layer of PIF device was close to its optimal thickness .whereas thickness effects of the double layer here are not investigated or optimised. The thickness of the organic layer is 50nm.

2.enhanced with respect to devices comprising only the pure component suggesting photoexcited charge transfer between the two components • The device shows a very wide photoresponse exceeding the visible wavelength range (4001000nm).2mW/cm2 . Figure 5. Also the current is relatively voltage dependent in both the dark and under illumination which results indeed in a low shunt resistor Rsh = 4 · 107 Ω.00V which is driven by potential fluctuation of the source measure unit which are in the mV range.see text. but is much lower here (≈10mV) Fortunately. The dark current24 is relatively high (more than 20 times higher near 0V than in the other blend devices discussed in this thesis).g. • The photoresponse is . The first slope is clearly Ohmic. the second is ∝ V 2 and the third ∝ V 3 . Summary In this Section we have found that: • The external quantum efficiency of a solar cell comprising a 1:1 blend between PIF and a polythiophene derivative (PTV) can reach at least 0.see also Chapter 2. pinholes for the following reasons: 1.015% but can most probably increased considerably if partial shorts can be avoided.29. • We found evidence for the presence of shallow traps from the analysis of the IV characteristics.5. Three different slopes fit the curve suggesting the presence of shallow traps .28: IV characteristics of the PTV+PIF blend in the dark and under illumination with light at 550nm and an intensity of about 0. partly shorted by e.29: Double logarithmic plot of the dark IV curve of the PTV+PIF device (forward bias only).5V (the workfunction difference between the electrodes). 24 This . The quadratic dependence of the current may be due to shallow traps while the third slope may represent the transition to the trap free condition . A double logarithmic plot of the dark IV curve reveals tree distinct slopes (regimes).see Fig. the shunt resistor is still high enough to avoid complete cancelling of the photogenerated current so that at least a qualitative evaluation of this blend is possible. We note that Rs = 106 Ω (derived from the slope in the first quadrant).at least partly . PTV+PIF 93 Figure 5.5. We note that the low shunt resistance requires higher fields across the device to release the trapped charges but should not affect this analysis otherwise .7. is the current at 0. The open circuit voltage is expected to reach about 0.

94 CHAPTER 5. BLEND LAYER DEVICES .

4.the (unwanted) connectivity with the wrong electrode.e. The so obtained network structure is less likely to result in both isolated areas and thanks to the separate charge transport layers . Relatively simple fabrication . Both types of D/A cells. In the following we list the strong points of the laminated structure: 1. The lamination process can lead to an improved D/A interface when compared to both a standard double layer and even a blend.while the situation is reversed for the double layer structure. Recently we have published a new device architecture i. have certain strengths but also disadvantages. This is because molecules in each layer have the opportunity to diffuse into the other layer during the lamination process. Automatic encapsulation of the two semiconducting layers between the substrates. during or after the fabrication. the fabrication of a blend device requires solubility of both components in the same solvent . applying pressure or exposing the layers to solution vapor before.from solution applying the technologically well known lamination technique. both types would benefit a lot if there was a possibility for separate treatment such as doping or heating of the specific D or A layer in order to modify the electrical and optical properties. 2.g. during and after the lamination process 95 . 5. double layer and blend structures. For instance. doping or heat treatment. Both of which represent serious problems in normal blend devices. 95].Chapter 6 Laminated Layer Devices 6. Moreover. charge-separation can occur at the interface of the hole and electron transport layers. The basic idea is to deposit both the D and A layer on separate electrodes (on substrates) before they are laminated together to obtain a D/A double layer device structure.1: Scheme of the manufacturing method and architecture used to obtain the laminated solar cell. The quality of the D/A interface can be controlled by heating. Unfavourable dimensions and/or defects of the interpenetrating network structure are the main problem in blends whereas short exciton diffusion lengths limit the performance of double layers severely. Here we investigate both a device comprising polymeric semiconductors and another using dyes. The quality of the D/A interface may be controlled by treating the layers before. Opportunity to treat each layer separately to improve the electronic and optical properties by e. After lamination. manufacturing method that allows to realise all these possibilities in a surprisingly simple way [94. 200. 3. Figure 6. Improved D/A network structure.1 Introduction In the previous chapters we have seen that single layer devices can be considerably improved by introducing a second semiconducting component resulting in a D/A type solar cell.

. The second device in this Chapter comprises the two soluble small molecules HPc and Per.2 Voc V 1. Reflection losses were neglected for all measurements. The bandgap of the spincoated film of POPT can be lowered by about 100nm upon moderate heat treatment.37 FF % 33 23 range nm 400-700 400-900 materials – POPT/MCP HPc/Per comments – polymer/polymer [94] liquid crystal(dye)/dye [200] year 1998 2000 Among the listed advantages of laminated devices we want to emphasize that the possibility of separate treatment of the D and the A layer represents a crucial point if commercially interesting efficiencies are to be achieved [277].despite some advantages of the used dyes over the polymers as discussed in Section 6. Absorption spectra were measured with a Hewlett Packard 8453 UV-Vis spectrometer.which represent the only reported structures of this type in literature to date .1: Survey of performance parameters of laminated solar cells. 213].similar to the requirements for a blend structure.3 0. the performance of the laminated dye device was surprisingly poor.) between the electrodes could deteriorate the quality of the interface. 4. 2.2 The Laminated Polymer Device In this section we investigate a laminated device structure using two polymers (Fig. Both polythiophenes and polyphenylenevinylenes have been showing good photovoltaic properties within the class of conjugated polymers [112. 6. preferably dissolve in the same solvent and form aggregates not larger than about twice the exciton diffusion length .1. see Chapter 5. 3. Experimental Current voltage curves and photocurrents were measured using a Keithley 237 source measure unit. Particularly flat ITO covered glass substrates or flexible substrates are commercially available and may be used to overcome such problems.2) that have been chosen because of the following properties: 1. Together with MCP virtually the entire visible range can be covered which increases efficiency. Liquid crystalline phases are particularly “soft”. The curvature of the substrates can become important since the films are only about 100nm thin and good contact over the entire area is desired.6. EQE % 30 0. 3. the following limitations using the lamination technique have to be considered: 1. . 2. Note that . it was even lower than the blend structure using the same molecules.is given in Table 6.e. In the following sections we investigate first how soluble polymers can be used to obtain a device whose performance exceeds even the best pigment based device in terms of solar efficiency. Quantum efficiencies were determined by normalisation with a calibrated silicon photodiode in the sample position. Many polymers can be heated above a certain temperature (glass transition) where they become soft and “rubberlike”.96 CHAPTER 6. In practice. Monochromatic illumination was provided by the output of a tungsten lamp dispersed by Bentham M300 single-grating monochromator. Both components have to be compatible i. Dust particles or others (crystallites. LAMINATED LAYER DEVICES Table 6. A survey of the performance parameters of both devices . PL quenching experiments indicate exciton splitting at the polymer/polymer interface .a crucial requirement for building a D/A solar cell. Both material need to be sufficiently“soft” to enable lamination. Both materials show suitable mechanical and film forming properties together with the required thermal stability. In fact. 4.3.

PL-efficiencies were measured using an integrating sphere and an argon ion laser as described previously[169]. In other words.not only in the first few ms . The device was fabricated in a nitrogen-filled glove box (oxygen<3ppm).5mm2 . Results and Discussion PL Quenching Fig.2 shows that the single components are fluorescent and the PL of both is quenched upon addition of a small amount of one polymer to the other. as much as 45% of all excitons generated in MCP recombine radiatively in the pure film. a similar “bath tub” shape may be obtained. Note that MCP with 45% PL efficiency is very fluorescent. Upon the addition of only 5% POPT the number of radiative recombination is reduced to only 15%. Note that PEDOT and Au have approximately the same workfunction. the PL of both materials is quenched in the blend. commercially available) film. 97 Figure 6. In order to sustain the photo-induced charge generation mechanism over time the trapped hole in POPT eventually has to recombine with an electron. Energy transfer between the blend partner as an alternative explanation has been ruled out1 since 1 Only in the unlikely case that energy transfer is followed . One half of the device comprises the electron acceptor material(MCP) spincoated from filtered (0.2.45µm) chloroform solution onto a conducting polymer (PEDOT. 278.g. MEH-CN-PPV or MCP is a fluorescent cyano substituted derivative of poly(p-phenylene vinylene) with a large electron affinity as a result of the electron withdrawing cyano-groups [174]. From the fact that we do observe “continuous PL quenching” . The half with the POPT layer was heated to 200 o C under vacuum before the device was laminated together by applying a gentle pressure while one half was still at elevated temperature. The other half of the device comprises the hole acceptor material (POPT) spincoated from filtered (0. We added 5% (by weight) POPT to the MCP solution which has improved the efficiency considerably. Santa Barbara).6. PEDOT was spincoated from water onto a transparent gold layer to decrease the sheet resistance of PEDOT. the electrons in the CB of adjacent MCP molecules. THE LAMINATED POLYMER DEVICE The chemical structures of the polymer semiconductors used in this work are shown in Fig. The AFM images were obtained using phase sensitive detection. This indicates efficient exciton dissociation in both polymers when blended. The gold layer was sublimed on a glass substrate.6.6. Such a ”bath tub” shape of the PL versus blend ratio plot may therefore be taken as indication for charge transfer at the polymer/polymer interface particularly if the electron affinity of one component (here MCP) is higher.45µm) chloroform solution (5mg/ml) on Ca which was sublimed on a glass substrate.2: Chemical structures of the polymers used in the polymer double-layer blend and PL efficiency as a function of the blend ratio. 62]. This means a good deal of the photoexcited excitons is now split into its constituent charges with the holes trapped on the POPT molecules and extra free electrons in the conduction band of MCP. by strong non-radiative recombination. Note that PL quenching ratios of 100 to about 10000 have been reported for blends involving relatively small molecules whereas quenching ratios in polymers are typically (here:20) about one order order of magnitude lower [112.2. The total thickness of the semiconducting layers was about 80nm and the active area 2. 4] which absorbs more in the red region of the spectrum 2) to exceed the glass transition temperatures so that both polymers become “soft”.we can infer that these trapped holes do find a recombination partner such as e. and were taken with a NanoScope IIIa Dimension 3100 (Digital Instruments Inc. Heating of the polymer device prior to lamination was required for two reasons: 1) to introduce a transition of POPT into the more crystalline phase [18.

pure POPT shows already only 5% differences are no more than 5nm. to reveal the cross section was made at low temperThe addition of 5% of MCP to the POPT layer ature. Indeed. 20%)of the guest molecule (Fig. it becomes more diffiFig. viscoelastic properties of the two polymers. The cut nation partner.7 show an in-plane image of the photoluminescence.as discussed in Chapter 2. which contribute to the photocurrent and the reduction of the series resistor. LAMINATED LAYER DEVICES found that the addition of 5% MCP to POPT could not improve the device performance [94]. The absorption of the POPT film can be considerably improved (broadened) upon thermal annealing. tively high OD can be achieved over a wide spectral range.not expected to be revealed in these images.6. These citons from 5 to 3%. Fig.98 Note that this charge generation mechanism is not as efficient as the one at D/A interfaces of double layer structures where both charges of an exciton can travel to the electrodes. In fact the generation mechanism here resembles more the situation as we find it with oxygen traps distributed throughout an organic semiconductor . Thus. on a length scale of 20-30nm.6 and Fig. Apart from the formation of the .3 shows the absorption spectra of MCP and POPT films.here electrons in MCP. Figure 6. [18] and is probably related to a temperature induced structural reorganisation of the regio-regular POPT molecules.procedure.5 shows the cross section of a laminated cult to get rid of the trapped charge since there are structure: There is inter-penetration between the more and more like molecules in the neighborhood two layers following the lamination and annealing carrying the same trapped charges and no recombi. we other component. Fig. Absorption and EQE Spectra Fig. POPT∗ .absorption (480nm).spectrum nicely so that the entire semiconducting served incomplete PL quenching for higher concen. We note that finer scale inter-penetration is reduces the fraction of radiatively recombining ex.4 shows the linear absorption as well as the EQE spectrum of the laminated device Figure 6. This is about the same PL (and the other) AFM images are shown in phase reduction factor as for the quenching of the PL of contrast mode. proaches a certain threshold.4: External quantum efficiency and absorption spectrum of the laminated polymer cell.2): generally high and reaches 28% near the peak of the When the concentration of the guest material ap.6. This has been observed before in the same molecule by Bergren et al. The EQE follows the shape of the absorption The described scenario is consistent with the ob. This results in the generation of extra charge carriers in the host system . The EQE is trations (e. The and POPT layer. Upon heat treatment the POPT absorption spectra of the heated POPT and the film can shift its absorption onset by about 100nm MCP film complement each other well so that relatowards the infrared. the relatively few extra charges due to exciton splitting in the blend are not two semiconducting polymers mixed with 5% of the expected to make much of a difference.bulk can be regarded as active layer. CHAPTER 6.6. most of which is due to differences in pure MCP.6. POPT is a semiconductor with a sufficiently low bandgap and can therefore also produce excitons.6.3: Absorption spectra of the pure MCP (ITO/PEDOT/POPT+5%MCP/MCP/Ca).6. height However.g. The main difference to oxygen induced charge generation is that unlike the oxygen atoms.

The larger spots are most likely islands of MCP clusters in the POPT matrix. THE LAMINATED POLYMER DEVICE 99 Figure 6.8.finer scale mixing is suspected. shown clusters or islands . We note that a comparable device that does not contain 5% POPT in the MCP film showed only EQE values around 5% which is comparable to “conventional” polymer blend devices [112]. Figure 6. An increased D/A interface (outlined on the right picture) on a 40nm scale can be seen.6.2. The open circuit voltage of 600mV is comparable to good double layer devices and close to the difference of workfunctions of the electrodes. .5: AFM phase contrast image of a cross-section of a laminated MCP/POPT+5%MCP polymer interface without metal electrodes sandwiched between polyester substrates.7: AFM phase contrast image of a POPT+5%MCP film.6: AFM phase contrast image of a MCP+5%POPT film. IV Characteristics The IV characteristics of the laminated polymer cell with ITO and Al contacts in both the dark and under illumination is shown in Fig. Figure 6. The larger spots are most likely islands of POPT clusters in the MCP matrix.6.

6. the increasing difference between light and dark curve (associated with a not negligible Rsh ) allows to double the EQE or spectral response if -2V are applied. IL = 114nA. illuminated area = 2. (6. Both resistor values are only little better than in the laminated polymer device with ITO and Al contacts.6.1V we can see that Eq.206V · ln(E) (6. Interestingly.2) with kT/q=26mV. The IV characteristics of the laminated polymer cell with Au/PEDOT and Ca contacts in both the dark and under illumination is shown in Fig.9. These resistor values are in the order of inorganic cells of comparable dimensions . However. Rsh = 2 · 107 Ω.8. The high value for n (and the low I0 ) was necessary to fit the high Voc .5V shows particularly little voltage dependence for which we have no conclusive explanation yet. the slope of Voc can be fitted well with V (E) = 2.9: IV-Characteristics of the laminated polymer cell with Au/PEDOT and Ca contacts in the dark (dashed line) and under monochromatic illumination with λ=480 nm. it is the shunt and not the series resistor that limits both the photocurrent and the fillfactor in this device. Using nkT /q = 0. The open circuit voltage reaches now 1. irradiance = 35mW/cm2 (solid line). Voc also increases with light intensity but slower and levels out around 2. From the slope in the first and third quadrant we obtain an upper limit for the series resistor of 7 · 104 Ω and a shunt resistor of approximately 107 Ω.5mm2 .5mm2 .2V. the open circuit voltage is considerably higher and the shape of the IV curve is now clearly dominated by the two resistor values (Rsh and Rs ).see Chapter 2. The large offset value of 2. Hence.10 we plotted the open circuit voltage and the photocurrent versus intensity (irradiance) of the incident light. Circles represent data points while the solid line shows the modeled curve using the Shockley equation . illuminated area = 2.1) if we ignore the offset value of 2.0 which is reasonably close to the n=5. LAMINATED LAYER DEVICES Figure 6.3V whereas the photocurrent and the fillfactor (0.is not surprising. I0 = 6pA and n=5.206V we obtain a diode ideality factor n=8.see Chapter 2.1V + 0. the both the high photocurrent and open circuit voltage with a relatively good FF . In fact.8: IV-Characteristics of the laminated polymer cell with ITO and Al contacts in the dark (dashed line) and under monochromatic illumination with λ=480 nm. However. In Fig. irradiance = 35 mW/cm2 (solid line).33) remains the same as in the ITO/Al device. The dark characteristics remains unchanged and Figure 6. The light characteristic can be fitted well with the Schockley equation (see Chapter 2) using Rs = 1 · 105 Ω.100 CHAPTER 6. With such bias voltage at least 58% can be achieved in the photodiode mode.1) resembles the expression as derived from the Schockley equation in Chapter 2: U =n IL kT · ln( + 1) q I0 (6. The figure shows that the photocurrent increases linearly with light intensity over at least 6 orders of magnitude. the current in the region between -1 and -1.8 obtained from the IV curve.1V may be caused by the driving force that can be associated with .

5d. This number is remarkable since it is almost twice stead of polymers.Au = 5. The electronic transport properties of the electron acceptor material could be improved considerably upon mixing a small amount (5%) of the donor material into it. The shunt resistor rather than Rs are responsible for remaining EQE and fillfactor losses in this device. We note that the method of lamination is already well established in (general) polymer & plastics industry and often realised as a roll to roll process Figure 6. THE LAMINATED DYE DEVICE 101 4. Wf.Ca = 2.11. Selective treatment of the two laminated layers. Both materials are relatively small molecules which dissolve readily in the same solvent.for details see In this section we investigate a laminated device structure using soluble small molecules (dyes) inChapter 8.10: Dependence of Voc and Isc on the light using self supporting sheets.1eV. 6. The workfunction difference of the electrode materials seems to determine the achievable Voc in this device. 2. 3. The lamination method allows the fabrication of organic solar cells that surpass the power efficiencies of earlier organic device architectures such as sublimed double-layers or blends.9 % . Under the assumption that Isc and Voc scale 6.3 The Laminated Dye Dewith intensity in the same way for all wavelengths vice we calculated a power-conversion efficiency for the polymer solar cell under solar conditions on earth (AM1. The laminated polymer device can utilise light over a broad spectral range (400-750nm). ticularly promising to achieve good performance values using the new technique for the following reaSummary sons: In this Section we have shown that: 1. further around 2. The liquid crystalline properties of HPc may favour the lamination process.2V. for example. featuring many technological advantages. . 2. such as large-area photosensitive arVoc ∝ ln(E) up to 20 mW/cm2 and saturates then rays for scanners and medical imaging etc. 76mW/cm2 ) of 1. Isc ∝ E 1.6. 1.02 while applications. Both molecules have high light-fastness and thermal stability.see Fig. 3. We have chosen the two soluas high as the hitherto best organic devices with ble derivatives of a metal-free phthalocyanine (HPc) reported solar efficiencies [244] and was achieved and a perylene diimide (Per) . intensity (E) at λ=488nm for the laminated PENevertheless. annealing or ”doping” of the organic semiconducting components in a D/A structure is now possible and can lead to high EQE and Voc numbers. Insoluble derivatives of these molecules have shown good performance in both single layer and double layer solar cells.9eV). efforts to improve charge transport and absorption are necessary and possible stability problems will have to be addressed. 5. Thus inter-diffusion after lamination may lead to a very fine and therefore efficient interpenetrating network sandwiched between to transport layers.. 4.3. A combination of these two materials looked pareven without sublimation of an active layer. before it comes to its first industrial DOT/Ca polymer solar cell.6. the difference of workfunctions of the electrodes (Wf.

1mm PTFE filters. Figure 6. However. The low temperature PL decreases with rising temperature until a sharp drop occurs at the solid to mesophase transition where the PL is virtually quenched completely [21]. PL efficiency of about 50% has been reported in a metal-free Pc with alkoxy substituents at low temperature (4K). The lamination was performed under gentle pressure in a laminar flow box after exposing both halves of the device for 15 min to chloroform vapor. 6. we cannot completely exclude that transfer of the exciton from the high bandgap material Per to the lower bandgap material HPc is taking place. Quantum efficiencies were determined by normalisation with a calibrated silicon photodiode in the sample position. Per is commercially available from Synthec/Synthone (Germany) while HPc has been synthesised as reported elsewhere[165]. Both solutions were filtered with 0. LAMINATED LAYER DEVICES Experimental Current voltage curves and photocurrents were measured using a Keithley 237 source measure unit.as discussed below increased the performance significantly.see Chapter 3 and Chapter 4. The electron-accepting dye (Per) was spincoated from chloroform solution (15 mg/ml) onto indium tin oxide (ITO) covered glass substrate. which . Because of the absence of PL in the solid state.6. Reflection losses were neglected for all measurements.11: Chemical structure of the dyes used (R=C12 H25 .6. The chemical structures of the dyes used in this work are shown in Fig. Radiative recombination of Per is strongly reduced for all blend ratios indicating charge-separation at the HPc/Per interface. PR-TRMC measurements indicate very high charge carrier mobilities in the crystalline and the mesophase of HPc. the investigation of the HPc+Per blends in Chapter 5 indicates that charge transfer Results and Discussion PL Quenching If exciton splitting occurs in a blend with at least one fluorescent material. This sharp drop has been associated with an increase of the exciton diffusion length in the mesophase that enables more excitons to reach quenching sites. As a consequence. The pure HPc layer does not show any detectable PL at room-temperature. HPc does not fluoresce in the solid state. Absorption spectra were measured with a Hewlett Packard 8453 UV-Vis spectrometer. However. M=H2 ) and photoluminescence efficiency as a function of the blend ratio. This means that virtually all excitons in the blend film are affected by the presence of HPc molecules suggesting a good . it is not possible to probe for quenching of the PL of HPc upon addition of Per. In addition. We also added 10 percent (by weight) Per to the HPc solution. Monochromatic illumination was provided by the output of a tungsten lamp dispersed by Bentham M300 single-grating monochromator. We note that our photovoltaic measurements of cells involving HPc in different architectures suggest exciton diffusion lengths of around 200nm at room-temperature .102 5. The hole accepting dye (HPc) was spincoated from chloroform solution (35 mg/ml) onto a 50nm thick Al layer which was thermally sublimed on glass. Efficient PL quenching has been observed suggesting efficient exciton quenching at the dye/dye interface. The photovoltaic properties of HPc as a single layer may be improved by introducing more charge free charge carriers by the addition of a few % Per. CHAPTER 6. Fig. solution and/or a sufficiently long diffusion range for excitons in Per. it is usually faster than its radiative decay and can therefore be observed as photoluminescence (PL) quenching.11.11 shows that the relative high PL of pure Per is almost completely quenched if the blend contains more than about 20% HPc.

3.42%) and only about a third when compared with the standard blend device (1.6.13. 103 Figure 6. This value is slightly smaller than in a comparable double layer device (0.0%). Absorption and EQE Spectra The EQE of this device is plotted together with the absorption spectrum in Fig. In the same Chapter we show AFM images that are consistent with the formation of a very fine network with these molecules. IV Characteristics The IV characteristics of the laminated dye cell is shown in Fig. The photocurrent to dark current ratio can be increased by a factor three upon applying a bias voltage of -1V. The cell is able to harvest light from the UV to the near infrared (900nm). The EQE spectrum is strongly antibatic in the wavelength range where only HPc absorbs (600-900nm) indicating the presence of a filter effect due to a too thick HPc layer. low shunt and high series resistor. Note that.12.6. We note that according to the considerations of IV curves in Chapter 2 a series resistor that is larger than the shunt would dominate the entire IV curve.3% at the short wavelength end of the visible range. Although the value for Rs can only represent an upper limit (since the slope could Summary In this Section we have shown that: • Small soluble molecules (dyes) can be used to manufacture a laminated solar cell.25. The thickness of HPc is about 400nm to avoid shorts in this soft material. • The efficiency could not be increased with respect to neither the double layer nor the blend cell. if the network is too fine.12: External quantum efficiency (EQE) and absorption spectrum of the laminated dye cell. Since this is not the case here we consider that there is an effect of a counter-diode that increases the slope in the negative voltage range. The slope in the first quadrant can be associated with a series resistor of 0. • The laminated dye cell can harvest photons ranging from 400 to about 900nm over the entire visible range with EQE comparable to good polymeric single layer devices like ITO/MEH-PPV/Al. increase for higher voltages) the low fillfactor and flat continuation of the curve after leaving the third quadrant suggests that Rs is truly large. THE LAMINATED DYE DEVICE takes place at the interface of HPc/Per mixtures. The IV curve would then become completely linear with a inverse slope according to the value Rs +Rsh .5 · 108 Ω whereas the inverse slope in the 3 quadrant gives a shunt of 0. The EQE approaches 0. This would also comply with the small fillfactor < 0.6.e. the connectivity between molecules may not be very good leading to a high recombination rate i.13: IV characteristic in the dark (dashed line) and under illumination (solid line) of the laminated dye device.1 · 108 Ω. Figure 6. .

LAMINATED LAYER DEVICES . CHAPTER 6.104 • The EQE can be increased by a factor 3 in the photodiode mode by applying -1V bias voltage.

In 1994. Lehmann who phase. molecules that have one or more atoms sticking out of the middle of the disc which limits the achievable degree of positional order.since the single molecules are free to move in the same fashion as in a liquid . With the exception of organic single crystals [139] 2 S. This can be the case for disc shaped (discotic) aromatic molecules since their strong π−π interaction acts as driving force that favours aggregation in stacks i.7. liquid crystal. and liquid phases that can be formed by disc-shaped molecules. Chandrasekhar reported the first observation of therthe reported mobility (0.g.2 shows that the degree of order within such a liquid crystalline phase can vary and mesophases with a different degree of positional order can form. The disc like shape also favours hexagonal arrangement of the columns.1 Introduction Molecules in the solid phase have virtually no opportunity to move and orient themselves with respect to one another. Reinitzer. The mobility in HTTP increased by a factor of 10 at the transition from the liquid to 1 The first man who discovered this special phase of matter the hexagonal discotic phase and gains further two was F.7.1 illustrates how both orientational and positional order may be improved since the molecules are free to move and to orient themselves. was the first to use the term liquid crystal [159]. However.Chapter 7 Discotic Liquid Crystals as Solar Cell Materials 7.orders of magnitude in a higher ordered discotic zoate [219]. The degree of order of the solid phase depends on its (thermal)history. reported high charge carrier (hole) mobility in the liquid crystalline phase of a discotic molecule (hexa-hexyl-thio triphenylene. HTTP) based on time of flight (TOF) measurements [2]. The situation is different when a solid melts to a liquid crystal1 : The positional order is then lost . to sustain a certain degree of orientational order since all discs can still arrange such that they face the common substrate plane . the molecules can now move and tumble randomly.which is sufficient to distinguish the phase clearly from a liquid. an Austrian botanist. He noted the double melting behaviour of cholesterol acetate and cholesteryl ben. if the same solid melts to a liquid both types of order are lost completely. Fig.e. Fig.1cm2 V−1 s−1 ) for photomotropic mesomorphism of a disc like molecule (benzeneinduced charge carriers of HTTP was higher than hexa-n-alkanoate) [44]. Reinitzer collaborated with O. 105 .1: Schematic illustration of the solid. The liquid crystalline phase allows self-organisation of the molecules in a limited temperature range (discotic phase). However. columns2 . al. They have a certain degree of positional and orientational order that cannot be changed in this phase. Adam et. The nematic phase for example can occur with e. Note that even these molecules are still able Figure 7.but at least some of the orientational order remains or even increases.

Thus a hotstage with a polarisation microscope and video camera to record images of the developed textures was set up and access to the equipment for essential thermal characterisation (TGA. CV data and aggregation effects have been reported for similar derivatives in Ref. 67].2: Schematic illustration of three different types of liquid crystalline discotic phases. The ordered columnar phase is expected to show the best transport properties. [180]). 267.6 ˚whereas the distance between A the columns was 21.e. cross-linked in the mesophase [262]. Although their monomers are well known in literature. DSC) of other research groups has been arranged. Such columns may be used to transport efficiently not only charge carriers but also excitons and ions [262]. Lux and G. Columnar discotic molecules (triphenylenes) have already been used in LEDs where they showed strong reduction of the onset voltage upon perpendicular alignment of the columns [53] Because of these interesting opportunities molecular wires made of discotic liquid crystals are considered a promising alternative to conjugated polymers which can also be seen as molecular wires since transport occurs most efficiently along their conjugated backbone. for any other organic system reported at that time. both hole and electron transporting discotic liquid crystals can already be prepared [23]. DISCOTIC LIQUID CRYSTALS AS SOLAR CELL MATERIALS have been intensely investigated over many decades [268. simply upon suitable cooling or become temperature independent if the molecules are polymerised i. 59.7 ˚. After a description of the experimental details in the next section we present and discuss the results of the liquid crystalline properties. In order to exploit the full potential of our molecules in photovoltaic devices it was important to know the ”mechanical” properties such as film forming properties. these soluble and probably liquid crystalline phthalocyanines have either not been synthesised before or not investigated in terms of their liquid crystalline and photovoltaic properties. Moreover. 154. 43.3: Chemical structure of the liquid crystalline phthalocyanines studied in this thesis. 231.7. A As a consequence. solubility but also the temperature range of the mesophases. Larger aromatic molecules like phthalocyanine pigments with their strong absorption and high stability despite the low bandgap have been proven very successful in the past as semiconducting layers in photovoltaic cells and therefore Figure 7. Only then can we orient the columns such that their axes points perpendicular to the electrodes.G. The obtained information is then used to fabricate two double . A. These particular derivatives are subject of our investigation in this Chapter. The distance between adjacent discs in a column was found to be 3. The HTTP molecule in which the high mobility was first measured is a relatively small conjugated system and absorbs not much light in the visible range. Rozenberg have synthesised phthalocyanine derivatives which carry 8 thio-alkyl side chains (Fig.3) to facilitate solubility and possibly liquid crystallinity [165]. Figure 7. The discotic phases may even be frozen in. charge (and exciton) transport is expected to occur preferably along the columns whereas the mobility in directions perpendicular to column axes can be more than 500 times smaller [23] since the excited species have to tunnel through a saturated hydro-carbon mantle (the insulating side chains).106 CHAPTER 7. 29.

1. Absorption spectra were taken using a Hewlett Packard 8453 UV-Vis spectrometer.3. Monochromatic illumination was provided by the output of a tungsten lamp dispersed by Bentham M300 single-grating monochromator. 105.7.3. 7. Reflection losses were neglected for all measurements. The peaks at 292 and 288o C indicating the transition to the liquid phase are clearly visible here .which is fairly rare for discotic phthalocyanines. These results are discussed in Chapter 4 which treats double layer devices structures. The Al top contact for the two devices was obtained via thermal sublimation. DSC and TGA measurements were performed with a Perkin Elmer DSC7 at a heating rate of 10K/min. [165. In this Chapter we investigate the photovoltaic properties of a simple device comprising a single layer of the liquid crystalline semiconductor sandwiched between ITO and AL.7. 269. both comprising a liquid crystalline phthalocyanine (HPc) with a perylene pigment as electron acceptor. NMR and mass spectroscopy. Brackets indicate transitions upon cooling. External quantum efficiencies were determined by normalisation with a calibrated silicon photo diode in the sample position.1: Phase transition temperatures (o C) of a series of liquid crystalline phthalocyanine derivatives. semicond. Since all mesophases were found to be very viscous at lower temperatures not all DSC peaks could be confirmed under the polarisation microscope. All temperatures have been determined using DSC scans and there was no contradiction with the results from polarisation microscopy. All materials were purified carefully by column chromatography and the structures established by IR.2 Experimental Current voltage curves and photocurrents were all measured in air using a Keithley 237 source measure unit.4: DSC trace of HPc. molecule H2 PcSC12 CoPcSC12 CuPcSC8 CuPcSC12 CuPcSC18 K • • • • • 20 (4) 52(20) 79(-) 63(9) 52(45) D • • • • • 292 – – – – I • • • • • 7. The oriented HPc film was obtained by heating to 292 o C (20K/min) under constant nitrogen flow and then slowly (5K/min until 270o C. we could determine the transition temperatures as summarised in Table 7. . A more detailed description of how we calculated the EQE can be found in Chapter 10. The solution was filtered with a 0. LIQUID CRYSTALLINE PROPERTIES layer devices. 73]. The chemical structure of the dyes used in this section are shown in Fig. I= isotropic (liquid). D=discotic=liquid crystalline. 107 Table 7.1µm disposable PTFE-filter. The spin-speed for the thick layer was 1600rpm and 2000rpm for the thin film. One device was heat treated to orient the columns perpendicular to the electrodes while the other one was kept at room temperature. K=crystalline. HPc was spincoated from chloroform solution (90 mg/ml for the thick (450nm) and 30mg/ml for the thin (80nm)layer) onto either an ITO covered quartz substrate to built a device or a spectrosil substrate for an optical measurements (polarisation microscopy).3 Liquid Crystalline Properties As a result of our investigations of these phthalocyanines in the hotstage under the polarisation microscope together with DSC analysis. In order to perform polarisation microscopy. Figure 7. Information on the synthesis of the different phthalocyanine’s can be found in Refs. then 20K/min) cooled down to room temperature using the hotstage (Linkam). an adapted Vickers microscope has been equipped with a hotstage (Linkam) and a set of polarisers. The transition temperatures of both the crystalline (K) to the discotic phase (D) and the discotic (=liquid crystalline) to the isotropic liquid (I)3 of 3 Which were in fact only found for HPc.

Image B shows the same section but with crossed posed in Ref. The origin of these linear defects has been proizers upon cooling from the liquid phase of HPc. The shown texture side-chains with alkyl or alkoxy chains with ethy.7. The birefringent lines can be fects that indicate perpendicular orientation of the the results of a deformation mode as illustrated in columns.was retained during cooling down to room temperature.5.nitrogen atmosphere occurs at about 320o C. 152]. DISCOTIC LIQUID CRYSTALS AS SOLAR CELL MATERIALS could be detected by DSC (see Fig.108 CHAPTER 7. characteristic texture of a columnar phase.substituted phthalocyanines investigated here are generally lower than those of their analogues with alkoxy side chains [73] and decrease even further with increasing chain length as reported for similar derivatives in Ref. see Fig.8.the clearing temperature . The observation of such defects may be taken as indication for perpendicular alignment of the columns with respect to the substrate plane. [253] whereas other defects are studpolarizers which reveals the linear birefringent de. For all other materials the mesophases are stable up to about 290-350C (see also the TGA results below) where they start to decompose before a liquid phase can be detected.4). the be achieved generally by replacing the thio alkyl focal conic texture. initial decomposition of CuPcSC12 under isotropic liquid phase . The thermal stability of the phthalocyanines was For HPc the K→D transition temperature is reduced to such an extent that the mesophase can investigated by thermogravimetric analysis (TGA) form already at room temperature (20o C). Figure 7. Both pictures were taken at 280 o C after Fig.6: Proposed molecular arrangement of discotic molecules to explain the formation of birefringent defects: The main deformation mode is a bend mode [253]. the octakis(alkylthio).7. As they grow they coalesce to a homeotropically aligned sample whereby some linear defects are formed. then .7.polarisers.7. Upon cooling HPc from the isotropic liquid at a rate of 10K/min. For exthe only material for which the transition to the ample. digitate stars appear under parallel polarizers (Fig.7. Fig. Heating of a sample of the cobalt derivative cooling from the liquid phase. [253].ied in Refs [31.7. 30].6. These appear as bright bireFigure 7. It is also at a heating rate of 10K/min.5: The image on the top (A) shows the fringent lines in the dark background under crossed digitate stars as they appear under parallel polar.7. more comNote that even lower transition temperatures can mon. Such a texture has been observed for alkoxy-substituted PCs [253. lene oxide units [152]. see Fig.5 A). (CoPcSC12) to 280 o C revealed another.

38eV.7. The onset of absorption is around 900nm which corresponds to a semiconductor bandgap of 1. the temperatures for initial and main decomposition are decreased by about 40o C. after cooling from the liquid phase. The somewhat pro- .11 follows the abso