Professional Documents
Culture Documents
Holger Borchert
Solar Cells
Based on
Colloidal
Nanocrystals
Springer Series in Materials Science
Volume 196
Series editors
Robert Hull, Charlottesville, VA, USA
Chennupati Jagadish, Canberra, ACT, Australia
Richard M. Osgood, New York, USA
Jürgen Parisi, Oldenburg, Germany
Shin-ichi Uchida, Tokyo, Japan
Zhiming M. Wang, Chengdu, People’s Republic of China
123
Holger Borchert
Department of Physics
Carl-von-Ossietzky University
of Oldenburg
Oldenburg
Germany
v
vi Preface
Thereby, short introductions to the different methods are given, and their appli-
cation potential for exploring the properties of materials and solar cells is
discussed. The third part of the book describes different concepts for using colloidal
nanocrystals in solar cells and reviews the state of the art and recent developments
and tendencies in this research area.
As the author, I would like to express my gratitude to all who supported the
writing of the book, either by reading parts of the manuscript or helping me in the
planning of the book. Namely, I would like to mention here my wife, Dr. Yulia
Borchert, as well as my present, respectively, former colleagues Dr. Martin
Knipper, Dr. Marta Kruszynska, Dr. Florian Witt, and Prof. Dr. Elizabeth von
Hauff. I am also particularly grateful to Prof. Dr. Jürgen Parisi for his advice in the
planning and in whose working group I got the opportunity to perform active
research in the scientific field which the present book is focused on. I hope to
provide with this book a useful and appealing work and hope the readers will
enjoy it.
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Part I Materials
vii
viii Contents
5 X-ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.1 Basics of X-ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.2 Particle Size Determination . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.3 Rietveld Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.4 Small Angle X-ray Scattering (SAXS) . . . . . . . . . . . . . . . . . 91
5.5 X-ray Diffraction of Soft Matter. . . . . . . . . . . . . . . . . . . . . . 92
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Abbreviations
xi
xii Abbreviations
MO Molecular orbital
MPP Maximum power point
OFET Organic field effect transistor
OPV Organic photovoltaics
P3EBT Poly(3-(ethyl-4-butanoate)thiophene)
P3HT Poly(3-hexylthiophene)
P3OT Poly(3-octylthiophene)
PANI Polyaniline
PCBM Phenyl-C61-butyric acid methyl ester
PCE Power conversion efficiency
PCPDTBT Poly[2,6-(4,4-bis-(2-ethylhexyl)-4H-cyclopenta[2,1-b;3,4-b0 ]
dithiophene)-alt-4,7-(2,1,3-benzothiadiazole)]
PDI Polydispersity index
PDTPBT Poly(2,6-(N-(1-octylnonyl)dithieno[3,2-b:20,30-d]pyrrole)-alt-
4,7-(2,1,3-benzothiadiazole))
PEDOT:PSS Poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate)
PESA Photoelectron spectroscopy in air
photo-CELIV Photocharge extraction by linearly increasing voltage
PIA Photoinduced absorption
PL Photoluminescence
PPP Poly(para-phenylene)
PPV Poly(para-phenylene vinylene)
PV Photovoltaics
PVD Physical vapor deposition
PVP Polyvinylpyrrolidone
Q-DLTS Charge-based deep level transient spectroscopy
SAXS Small-angle X-ray scattering
SCLC Space charge limited current
SEM Scanning electron microscopy
SILAR Successive ionic layer adsorption and reaction
STEM Scanning transmission electron microscopy
TBP Tributylphosphine
TCO Transparent conducting oxide
TDPA Tetradecylphosphonic acid
TEM Transmission electron microscopy
TOP Trioctylphosphine
TOPO Trioctylphosphine oxide
UHV Ultra-high vacuum
UPS Ultraviolet photoelectron spectroscopy
XPS X-ray photoelectron spectroscopy
XRD X-ray diffraction
Chapter 1
Introduction
Abstract Photovoltaic devices absorb sun-light and enable the conversion of solar
radiation into useable electrical energy. In view of the world-wide growing energy
demand, limited resources of fossil fuels and the need for more eco-friendly ways
of energy production, photovoltaics is gaining more and more importance. Till
date, the most common solar cell technology is based on crystalline silicon as the
photoactive material. However, alternative concepts for solar cells have emerged
as well. A relatively new and innovative branch of photovoltaics are organic solar
cells, where the photoactive layer consists of organic materials which are able to
conduct charge carriers. Organic solar cells are considered to have a high potential
to become producible at low cost and have also other attractive properties. For
example, they can be realized on flexible substrates, which enable their imple-
mentation in curved or flexible surfaces. On the other hand, organic solar cells still
suffer from limited device efficiency and lifetime. An alternative to purely organic
solar cells are hybrid devices combining organic materials with inorganic colloidal
nanocrystals. Colloidal nanocrystals have interesting and partly even controllable
physical and chemical properties from where arises a high potential to bring
innovation to the photovoltaic technology. The present book gives an overview
over the relevant fundamentals and the state-of-the-art of photovoltaic devices
containing colloidal nanocrystals, and the present chapter introduces to the topic.
Currently, in 2014, approximately 7.2 billion people are living on the world, and
the population keeps on growing by about 80 million people per year. From the
growth of the world population, the ongoing growth of industry and the natural
desire of human beings to improve their living conditions results a world-wide
growing demand of energy [1]. Today, the by far largest part of the annually
consumed energy is taken from fossil energy sources: oil, coal and gas [1, 2].
Regarding the resources of fossil fuels that are known today, there will probably be
enough reserves for the next decades. However, thinking more provident, the
resources of fossil fuels are finite and their exploitation may become more difficult
in future, because not all of the reserves are equally easy accessible. Moreover, the
combustion of fossil fuels is accompanied by the release of carbon dioxide which
is believed today to have a significant impact on climate change [2]. From these
basic facts arises the need to supply the world with energy from another source
than fossil fuels. A certain alternative is nuclear power, but this technology has the
serious disadvantage of high safety risks and many problems related to the
treatment of nuclear waste. Furthermore, also the world-wide uranium reserves are
limited, at least as long as extracting uranium from sea water remains difficult.
Fortunately, there is a very powerful source of carbon-free and renewable energy
available on earth: the radiation coming from the sun. However, the author would
not need to write this book, if the sunlight were an easy solution to the world’s
energy problem. Unfortunately, it remains a difficult task to convert the sunlight in
efficient manner and at affordable costs into other forms.
Photovoltaics (PV) is the technology concerned with converting sunlight into
electricity. Very basically, in any type of solar cell, the conversion process can be
broken down into several important elementary steps. In the first step, light coming
from the sun needs to be absorbed. The energy provided by the absorbed photons
is used to promote electrons in the absorber material into higher energy levels.
Each electron excited to a higher energy level leaves behind in its original level a
hole, a positively charged quasiparticle which is nothing else than a missing
electron. In this sense, light absorption generates charge carriers in the absorber
material: electrons excited to higher energy levels and corresponding holes left
behind. To make these charge carriers usable in an outer electrical circuit, the
positive and negative charges need to be spatially separated and transported to two
different electrodes where they can be extracted from the solar cell. Thus, the
energy conversion process can be broken down into charge generation by light
absorption, charge separation, charge transport and charge extraction. Various
types of solar cells have been developed to put these processes into practice.
The most wide-spread PV technology till date dominating the market is based
on pn-junctions with p- and n-doped crystalline silicon (Si) as absorber material
[3]. Figure 1.1 reminds the energy scheme of a pn-junction in equilibrium and
illustrates the basic working principle of charge separation in a corresponding solar
cell. Si is an indirect semiconductor with a band gap of approximately 1.1 eV,
meaning that photons with a wavelength below *1,100 nm can be absorbed by
the material and can promote electrons from the valence into the conduction band
[4]. The driving force to spatially separate the electrons elevated into the con-
duction band and the holes remaining behind in the valence band is provided by
the energetic structure of the pn-junction [3]. Silicon solar cells can reach power
conversion efficiencies of up to *25 % [5, 6] and exhibit also a reasonable life-
time of approximately 20 years or longer [7] which makes them suitable for
installations in solar energy parks, on roof-tops of various types of buildings, and
so on. Despite the relatively high efficiency and good long-term stability, Si solar
cells still have difficulties to compete with electricity gained from fossil fuels or
nuclear power, because the fabrication costs of these photovoltaic cells and also
other costs related to the technology, e.g., the costs for the installation of photo-
voltaic modules, are relatively high. At least partly, this is due to the fact that
crystalline Si solar cells require silicon in the form of wavers which are cost-
intensive and also energy-intensive in their production [3]. Another disadvantage
1 Introduction 3
of classical crystalline Si solar cells is that the corresponding PV modules are rigid
and have a relatively high weight which in turn prevents their usage on part of the
surfaces that would in principle be available for PV installations.
The mentioned deficiencies of crystalline silicon solar cells gave rise to the
development of alternative PV technologies. Another type are for example thin
film solar cells based on Cu(InxGa1-x)(SySe1-y)2 compounds which can be pro-
duced by sputtering or evaporation processes [8–11]. These compound semicon-
ductors, often abbreviated as CIS (for pure CuInS2), CIGS (for Ga-containing
material), CISe (for Se-containing material) or CIGSe (for Ga and Se-containing
material) have usually chalcopyrite structure, and the band gap can be tuned in the
range from 1.04 to 2.4 eV by adjusting the elemental composition [12, 13]. Thin
film solar cells of this type are commercially produced, and power conversion
efficiencies approach now 20 % at the level of individual cells and about 19 % for
solar cell minimodules [5]. Their production avoids the need for wafer technology,
but still requires a relatively high technical effort due to the deposition of the
absorber material by sputtering or evaporation technology. Another issue critically
discussed is the dependence on indium which became a comparably expensive
element due to its limited occurrence on earth and usage at large scale in displays
and other technologies.
Other alternatives to wafer-based silicon are CdTe solar cells [9], solar cells
based on amorphous or microcrystalline silicon [8], and organic photovoltaics
(OPV) [14–17]. The probably best established type of photovoltaic devices with
organic materials is dye-sensitized solar cells (DSSCs) [18]. In a classical dye-
sensitized solar cell, an organic dye attached to porous titania is used to harvest the
sunlight, and a liquid electrolyte is necessary for regeneration of the dye after
electron transfer from the dye to the titania network [18]. Classical dye-sensitized
solar cells reach currently up to *12 % power conversion efficiency [5]. Diffi-
culties of this technology relate for example to the long-term stability of the
organic dye molecules and to the fact that the presence of a liquid electrolyte
4 1 Introduction
Fig. 1.3 Energy scheme of a donor/acceptor system (under open-circuit conditions). In a type II
heterojunction, the electron donor material has HOMO and LUMO levels which are both higher-
lying than the corresponding orbitals of the acceptor material. If a photon is absorbed in the
donor, an electron is raised into the LUMO level, and a hole remains in the HOMO level (step 1).
Due to the lower LUMO level of the acceptor, the excited electron can be transferred from the
donor to the acceptor (step 2). After charge separation, the electrons and holes need to be
transported to the cathode and anode, respectively (step 3)
Note, however, that this picture is a bit simplified, because it neglects the role of
Coulomb attraction between the electron and the hole. More precisely, the gain in
energy due to the transfer of the electron to the lower-lying LUMO level of the
acceptor must at least compensate the loss of Coulomb binding energy accom-
panying the charge transfer process [25]. The charge transfer across the donor/
acceptor interface is an important step towards separated positive and negative
charges. After charge separation, the holes need then to be transported through a
network of the conductive polymer to the ITO/PEDOT:PSS anode, whereas the
electrons need to be transported through the fullerene network to the metal cath-
ode. During operation, electrons are finally extracted at the cathode, can be used in
the outer electrical circuit, and are injected back into the solar cell at the anode
which corresponds to hole extraction at this electrode.
Regarding the mentioned processes of charge transfer at the donor/acceptor
interface and charge transport towards the electrodes, the detailed structure of the
bulk heterojunction, usually referred to as the morphology of the active layer, plays
an important role [15–17, 22, 26]. Light absorption in the polymer leads to the
creation of Coulomb bound electron–hole pairs, so-called excitons. Compared to
inorganic semiconductors, the binding energy of the excitons is relatively high in
organic semiconductors [27]. Therefore, splitting of the excitons into free charge
carriers requires the charge transfer process across the donor/acceptor interface as
discussed above. This means, however, that an exciton created by light absorption
has first of all to diffuse to the material interface. There, another property of
organic semiconductors comes into play: Typically used conductive polymers
have relatively short exciton diffusion lengths of the order of about 10 nm only
[28], meaning that the photo-generated electron hole-pairs will simply recombine
radiatively, if the donor/acceptor interface is too far away. The bulk heterojunction
concept was developed to realize an active layer which is on the one hand thick
enough to absorb a large fraction of the sun-light, and which at the same time
brings the two material components into close vicinity. On the other hand, the
rather arbitrary nature of the interpenetrating network of the two phases means that
the pathways for charge transport towards the electrodes will not be ideal in
general. Therefore, controlling and optimizing the morphology of the active layer
in bulk heterojunction solar cells is a crucial issue in the field of organic photo-
voltaics [15–17, 22, 26].
Polymer/fullerene BHJ solar cells are a promising PV technology. Many of the
involved organic semiconductor materials are not yet produced at large scale and
low cost. On the other hand, the materials are based on carbon chemistry and do
not contain rare elements. Thus, from this point of view, organic semiconductors
can be considered as materials where large scale production is at least not pre-
vented by limited occurrence of the elements on earth. An important feature of
BHJ solar cells is that the organic materials are usually soluble in selected sol-
vents. Therefore, the material layers can in principle be produced by relatively
simple deposition techniques such as printing technologies or spray coating [29,
30]. This promises to save costs when compared to the wafer-based Si technology
or other thin film PV technologies that are dependent on high-temperature or
1 Introduction 7
References
5. M.A. Green, K. Emery, Y. Hishikawa, W. Warta, E.D. Dunlop, Prog. Photovoltaics Res.
Appl. 22, 1 (2014)
6. S.W. Glunz, High-efficiency crystalline silicon solar cells. Adv. OptoElectron. (2007).
doi:10.1155/2007/97370
7. D. Heinemann, W. Jürgens, R. Knecht, J. Parisi, 30 years at the service of renewable
energies. Einblicke (Research Journal of the University of Oldenburg, Germany) 54, 6 (2011)
8. M.A. Green, J. Mater. Sci.: Mater. Electron. 18, S15 (2007)
9. M. Powalla, D. Bonnet, Thin-film solar cells based on the polycrystalline compound
semiconductors CIS and CdTe. Adv. OptoElectron. (2007). doi:10.1155/2007/97545
10. R. Knecht, M.S. Hammer, J. Parisi, I. Riedel, Phys. Status Solidi A 210, 1392 (2013)
11. J. Keller, R. Schlesiger, I. Riedel, J. Parisi, G. Schmitz, A. Avellan, T. Dalibor, Sol. Energy
Mater. Sol. Cells 117, 592 (2013)
12. T. Tinoco, C. Rincon, M. Quintero, G. Sanchez Perez, Phys. Status Solidi A 124, 427 (1991)
13. V.S. Saji, S.-M. Lee, C.W. Lee, J. Korean Electrochem. Soc. 14, 61 (2011)
14. S.E. Shaheen, D.S. Ginley, G.E. Jabbour, MRS Bull. 30, 10 (2005)
15. B.C. Thompson, J.M.J. Frechet, Angew. Chem. Int. Ed. 47, 58 (2008)
16. C. Deibel, V. Dyakonov, Rep. Prog. Phys. 73, 096401 (2010)
17. C.J. Brabec, S. Gowrisanker, J.J.M. Halls, D. Laird, S. Jia, S.P. Williams, Adv. Mater. 22,
3839 (2010)
18. M. Grätzel, J. Photochem. Photobiol., C 4, 145 (2003)
19. M.M. Lee, J. Teuscher, T. Miyasaka, T.N. Murakami, H.J. Snaith, Science 338, 643 (2012)
20. J.M. Ball, M.M. Lee, A. Hey, H.J. Snaith, Energy Environ. Sci. 6, 1739 (2013)
21. S.K. Hau, H.-L. Yip, J. Zou, A.K.-Y. Jen, Org. Electron. 10, 1401 (2009)
22. H. Hoppe, N.S. Sariciftci, J. Mater. Chem. 16, 45 (2006)
23. J.E. Slota, X. He, W.T.S. Huck, Nano Today 5, 231 (2010)
24. P.W.M. Blom, V.D. Mihailetchi, L.J.A. Koster, D.E. Markov, Adv. Mater. 19, 1551 (2007)
25. C. Deibel, T. Strobel, V. Dyakonov, Adv. Mater. 22, 4097 (2010)
26. L.-M. Chen, Z. Hong, G. Li, Y. Yang, Adv. Mater. 21, 1434 (2009)
27. M. Knupfer, Appl. Phys. A 77, 623 (2003)
28. P.E. Shaw, A. Ruseckas, I.D.W. Samuel, Adv. Mater. 20, 3516 (2008)
29. A.C. Hübler, H. Kempa, in Organic Photovoltaics, ed. by C. Brabec, V. Dyakonov, U. Scherf
(Wiley-VCH, Weinheim, 2008)
30. C. Girotto, B.P. Rand, J. Genoe, P. Heremans, Sol. Energy Mater. Sol. Cells 93, 454 (2009)
31. W. Ma, C. Yang, X. Gong, K. Lee, A.J. Heeger, Adv. Funct. Mater. 15, 1617 (2005)
32. S.H. Park, A. Roy, S. Beaupre, S. Cho, N. Coates, J.S. Moon, D. Moses, M. Leclerc, K. Lee,
A.J. Heeger, Nat. Photonics 3, 297 (2009)
33. H.-Y. Chen, J. Hou, S. Zhang, Y. Liang, G. Yang, Y. Yang, L. Yu, Y. Wu, G. Li, Nat.
Photonics 3, 649 (2009)
34. Y. Liang, Z. Xu, J. Xia, S.-T. Tsai, Y. Wu, G. Li, C. Ray, L. Yu, Adv. Mater. 22, E135 (2010)
35. T. Ameri, G. Dennler, C. Lungenschmied, C.J. Brabec, Energy Environ. Sci. 2, 347 (2009)
36. L. Dou, J. You, J. Yang, C.-C. Chen, Y. He, S. Murase, T. Moriarty, K. Emery, G. Li, Y.
Yang, Nat. Photonics 6, 180 (2012)
37. W. Li, A. Furlan, K.H. Hendriks, M.M. Wienk, R.A.J. Janssen, J. Am. Chem. Soc. 135, 5529
(2013)
38. R.F. Service, Science 332, 293 (2011)
39. W.E.J. Beek, R.A.J. Janssen, in Hybrid Nanocomposites for Nanotechnology, ed. by L.
Merhari (Springer Science+Business Media, New York, 2009)
40. Y. Zhou, M. Eck, M. Krüger, Energy Environ. Sci. 3, 1851 (2010)
41. H. Borchert, Energy Environ. Sci. 3, 1682 (2010)
42. T. Xu, Q. Qiao, Energy Environ. Sci. 4, 2700 (2011)
43. M. Wright, A. Uddin, Sol. Energy Mater. Sol. Cells 107, 87 (2012)
44. H. Weller, Angew. Chem. Int. Ed. 32, 41 (1993)
45. H. Weller, Adv. Mater. 5, 88 (1993)
46. A.P. Alivisatos, J. Phys. Chem. 100, 13226 (1996)
References 11
nanometer size regime. Examples are different types of lithography methods, such
as photolithography. Lithography methods are important and widely used in
industrial processes, but the resolution which can be obtained is usually still
limited to approximately *50 nm.
In the bottom-up approach, it is intended to build up nanostructured materials
from atomic or molecular precursors. Important methods within this category are
for example chemical vapor deposition (CVD) [8, 9], physical vapor deposition
(PVD) [10, 11], and colloidal synthesis. In particular, colloidal chemistry turned
out to be a method enabling a high degree of control over structural parameters
such as the particle size and shape which in turn means good control over physical
and chemical material properties. Therefore, colloidal nanocrystals attract high
attention for applications in various fields, one of them being optoelectronic
devices like light-emitting diodes [12] or solar cells [13, 14]. In the following,
some basics of colloidal synthesis will be briefly outlined here. More complete and
detailed introductions to the colloidal synthesis of nanocrystals can for example be
found in [6, 15–18].
Colloidal synthesis is a wet-chemical method with the general aim to synthesize
nanocrystals in the form of a stable dispersion in a solvent, which is called a
colloidal solution. In order to achieve this goal, usually a surfactant, alternatively
called ligand or stabilizer, needs to be used. The surfactants bind to the surface of
the inorganic nanocrystals in the solution and fulfill a variety of different functions.
An evident function is to act as spacer between the nanocrystals in the solution,
and thus to prevent and inhibit aggregation of the nanocrystals which would
otherwise result in the formation of bulk material. Another fundamental function is
to provide solubility. Without surfactants being present at the surface, in most
cases, the inorganic crystals would not be stable in solution, but precipitate. For
example, stabilization in a non-polar organic solvent can be provided, if organic
molecules are used as ligands which possess a functional group binding to the
nanocrystal surface and a long hydrocarbon chain enabling interaction with the
solvent molecules. Typical examples for molecules used as ligands are alkan-
ethiols, alkylamines or carboxylic acids. Figure 2.1 illustrates schematically the
stabilization of small nanocrystals in colloidal solution.
For the colloidal synthesis of nanocrystals it is necessary to choose appropriate
starting materials, so-called precursors. Typical examples for precursors are, e.g.,
various metal salts which can be dissolved in appropriate solvents to yield ionic
species. The dissolved ions are often called monomer and can be attached to the
surface of existing nanocrystals in the reaction solution.
The formation of nanocrystals in colloidal solution can usually be regarded to
comprise two phases—nucleation and growth [15]. First, during the nucleation
phase, small crystallization seeds of the desired compound need to be formed.
Therefore, the solubility product of the compound in question needs to be
exceeded. For example, in the case of a binary compound AB, this can be achieved
by mixing two solutions containing in sufficiently high concentration dissolved
precursors for the elements A and B, respectively. Once nuclei exist, monomer can
2.1 Basic Concepts of Colloidal Synthesis 17
Fig. 2.1 Schematic illustration of an inorganic nanocrystal with an organic ligand shell. In this
example, the crystal consists of InP, and the ligand shell is a mixture of trioctylphosphine (TOP)
and trioctylphosphine oxide (TOPO). Both of these ligands can bind to the nanocrystal surface
and have hydrocarbon chains providing solubility in non-polar organic solvents, such as toluene
or n-hexane
be attached to the surface and let the nanocrystals grow in solution. It is, however,
important to note that the growth of the nanocrystals is a dynamic process. On the
one hand, new monomer can be attached to existing crystals, but at the same time,
the nanocrystals can also dissolve, so that monomer is detached from the surface
and released back into the solution. From growth and dissolution finally results a
net growth rate [19, 20]. Figure 2.2 illustrates this situation.
The dynamic growth process of colloidal nanocrystals is complex in detail, and
many efforts were undertaken to provide theoretical descriptions of the growth
process [19]. One can further distinguish between the following two situations:
growth by attachment of monomer available in excess in the solution, and growth
by Ostwald ripening [15]. In the first case, monomer is still present in relatively
high concentration in the solution and therefore available to be attached to the
surface of existing nanocrystals. This situation can typically be encountered at
early stages of a reaction. Later, when most of the monomer is already attached to
the nanocrystals, the reaction solution gets depleted of free monomer. At this stage,
the ensemble of nanocrystals can still evolve by Ostwald ripening. Because of the
dynamics of dissolution and growth, a monomer can be detached from the surface
of one particle and then be attached to the surface of another particle. Experiment
and theory have shown that the net growth rate of the nanocrystals in a given
ensemble depends on the particle size [19]. In more detail, there is a critical radius.
Particles with larger size will grow, whereas particles below the critical size will
dissolve. This growth of large nanocrystals at the expense of small crystals is
known as Ostwald ripening [19]. The thermodynamic driving force for the
Ostwald ripening process is the surface energy. Smaller nanocrystals have a larger
surface-to-volume ratio. By consequence, Ostwald ripening lowers the total
surface energy of the ensemble of nanocrystals.
18 2 Physics and Chemistry of Colloidal Semiconductor Nanocrystals
The mechanisms briefly outlined here are not the only processes which can
occur. For example, in some cases nanocrystal growth can also occur by aggre-
gation. In that case, two individual crystals can merge to a larger crystal. This
process does not necessarily lead to grain boundaries in the nanocrystals. For
example in the case of ZnO, it was shown that single crystal nanorods can be
formed by an oriented attachment of quasi-spherical particles [21], i.e., the indi-
vidual crystals merge in a defined way without creation of a grain boundary.
One method, widely used to prepare nanocrystals of inorganic compounds is the
hot-injection method [22]. In this case, the precursor of one element is dissolved,
usually together with a stabilizer, in a high-boiling point solvent and heated to high
temperature (in many cases in the range of 200–300 C). The precursor for the
second element is separately dissolved and then rapidly injected into the first
solution, e.g., using a syringe. The injection is often accompanied by a rapid drop
of the reaction temperature by several tens of degrees. After injection, the solution
is then spontaneously oversaturated, so that a rapid nucleation phase takes place.
The consumption of monomer by the formation of nuclei rapidly lowers the
concentration of free monomer. By consequence, shortly after injection, the sol-
ubility product of the desired compound is no longer exceeded. Nucleation is then
terminated and the growth phase starts. The hot injection method opens a possi-
bility to separate the nucleation and growth phase in time [15]. This is in contrast
to so-called one pot synthesis methods, where both precursors are dissolved
simultaneously in the same solution. In that case, the formation of nanocrystals can
start before the precursors are completely dissolved. The first nanocrystals already
start to grow before the nucleation phase is terminated. In many cases the sepa-
ration of the nucleation and growth phase can result in a better degree of size and
shape control [23]. On the other hand, one pot methods can be advantageous, e.g.,
in view of easier upscaling [23].
Coming back to the stabilizing ligands, it is important to note that also the
ligands strongly influence the growth dynamics of colloidal nanocrystals. This can
qualitatively easily be understood. For example, the fraction of surface sites of a
given nanocrystal which is capped by ligand molecules will influence the
2.1 Basic Concepts of Colloidal Synthesis 19
One of the most prominent examples for size-dependent material properties is the
quantum size-effect mentioned already in the introduction. Figure 2.4 illustrates
the phenomenon of the size-dependent band gap of a semiconductor schematically.
As can be seen, small nanocrystals consisting of several hundreds to thousands
of atoms are an intermediate between extended solids and large molecules. If the
number of atoms is too small, the energy bands are not really quasi-continuous.
Instead, the energetic structure can be understood as consisting of ‘‘bands’’ which
comprise discrete energy levels. For simplicity, the terminology of energy bands as
known in solid state physics is applied in this book to semiconductor nanocrystals.
The reader should, however, be aware that this is a simplified description.
In the following, the physical origin of the quantum size effect will be described
in more detail. One aspect that needs to be considered is the binding energy of an
electron–hole pair created by light absorption. In a bulk semiconductor, an elec-
tron–hole pair, or an exciton, is a Coulomb-bound state of the electron in the
conduction band and the hole in the valence band. In the case of so-called Mott
Wannier excitons, the electron–hole pair can move through the crystal. The
quantum mechanical description of Mott Wannier excitons is very similar to the
2.3 Material Properties Depending on Particle Size 21
Fig. 2.4 Illustration of the quantum size effect in semiconductors. Part (a) shows molecular
orbitals which are formed by the linear combination of atomic orbitals (according to the so-called
LCAO approximation). In bulk solids (image c), the combination of atomic orbitals from a large
number of atoms yields quasi-continuous energy bands. In the case of a semiconductor, the
valence and conduction band are separated by a band gap of the energy EG. A small nanocrystal
(image b) is an intermediate between molecules and extended solids. The energy bands are then
no longer really continuous, but contain discrete levels. Furthermore, the gap between the lowest
unoccupied and the highest occupied state increases with decreasing particle size
problem of the hydrogen atom, and one can write the stationary Schrödinger
equation as follows:
^ ¼ EW
HW ð2:1Þ
with the Hamiltonian
_ h2 h2
H¼
De Dh þ V; ð2:2Þ
2me 2mh
and the potential given by
e2
V ¼ ð2:3Þ
re ~
4per e0 j~ rh j
Herein, me and mh are the effective masses of the electron and hole, respec-
re ~
tively, er is the relative dielectric constant, and j~ rh j is the distance between the
electron and the hole. The solution of this Schrödinger equation yields the
22 2 Physics and Chemistry of Colloidal Semiconductor Nanocrystals
Fig. 2.5 Schematic illustration of Mott-Wannier excitons. Part (a) shows an energetic scheme of
the electron and hole in the conduction and valence band, part (b) shows a spatial representation
of the exciton which can move through the crystal, and part (c) shows the excitonic levels
resulting from the solution of the Schrödinger equation
following, quantized energy levels for the excitonic states, with the origin of the
energy scale set to the valence band maximum [58]:
e2 1
En ¼ EG 2; n ¼ 1; 2; . . . ð2:4Þ
8per e0 aB n
Herein, EG is the band gap, n is a quantum number, and aB is the exciton Bohr
radius which can be considered as the mean distance of the electron and hole in the
lowest excited state (corresponding to n = 1). The exciton Bohr radius is given by
the following expression and resembles the Bohr radius of the hydrogen atom
because of the analogy of the problem:
4per e0 h2 1 1 4per e0 h2
aB ¼ þ ¼ ð2:5Þ
e2 me mh le2
The reduced mass l occurring in this equation needs to be calculated from the
effective masses of the electron and hole. As can be seen from the preceding
equations, the excitonic levels are located below the conduction band minimum.
Figure 2.5 illustrates the above discussions.
According to (2.5), the exciton Bohr radius depends only on the dielectric
constant and the effective masses. With these material parameters being known,
the mean distance of the electron and hole can easily be calculated. Typical values
resulting for semiconductors are of the order of *1–10 nm [59]. Furthermore,
with the exciton Bohr radius and the relative dielectric constant it is possible to
calculate the exciton binding energy which is the energy difference of the lowest
state with respect to the conduction band minimum. Typical values are in the range
of *5–500 meV [58]. Table 2.1 summarizes the dielectric constants, exciton Bohr
radii and exciton binding energies for a selection of semiconductors. The dielectric
constant is in general frequency-dependent. In many works on the quantum size
effect, values at optical frequencies are used [60–62]. When checking, if the data
collected in the table is in agreement with (2.4), good agreement is found only for
2.3 Material Properties Depending on Particle Size 23
Table 2.1 Relative dielectric constants er, exciton Bohr radii and the energy difference DE
between the lowest excitonic state and the conduction band minimum (exciton binding energy)
for a selection of semiconductors
Compound er Exciton Bohr radius/nm DE/meV
Si 11.9 (opt) [63] 4.3 [59] 14.7 [58]
CdS 5.2 (opt) [64] 2.8 [59] 29.0 [58]
8.3 (stat) [64]
CdSe 6.2 (opt) [64] 4.9 [59]–5.4 [65] 15.0 [58]
9.6 (stat) [64]
PbS 17.2 (opt) [64] 18.0 [66] 2.3 (calc)
161.0 (stat) [64]
PbSe 25.0 (opt) [64] 46.0 [67] 0.6 (calc)
227.0 (stat) [64]
ZnSe 8.1–9.1 [63] 3.8 [59] 22.0 (calc)
InP 10.6 (opt) [64] 9.6 [65] 4.0 [58]
15.0 (stat) [64]
InAs 12.3 (opt) [64] 36.8 [65] 1.6 (calc)
15.2 (stat) [64]
GaAs 10.9 (opt) [60] 11.3 [65]–12.5 [59] 4.2 [58]
12.5 (stat) [65]
AgBr 12.5 [63] 4.2 [59] 20.0 [58]
(stat) or (opt) specifies, if the values refer to the static or optical dielectric constant. The data
sources are indicated after each value. (calc) means that the value for DE was calculated
according to (2.4) from the entries in the two other columns, using the optical dielectric constant
some of the materials, e.g., for silicon. Thus, the reader should be aware that some
of the values reported may have considerable uncertainties.
So far, the description of Mott Wannier excitons refers to infinite solids. What
is different now in a small semiconductor nanocrystal? According to Table 2.1, the
exciton Bohr radius is only a few nanometers in many semiconductors and
therefore of the same order as the size of small nanocrystals. If the particle size is
smaller than the exciton Bohr radius for the bulk, the electron–hole pair cannot
adopt the same mean distance as it would have in the bulk. Instead, the exciton will
be confined in a smaller volume. This situation is called strong confinement
regime. In the following, we will concentrate on this case. Discussions for weak
and intermediate confinement can for example be found in [59, 68]. As a conse-
quence of strong confinement, the energies of the quantized states will be different
from those of the corresponding bulk material. Qualitatively, the exciton Bohr
radius occurring in (2.4) gets replaced by the particle diameter. In the strong
confinement regime, the Coulomb interaction is, however, not the main factor
determining the energy of the lowest excited state as will be discussed below. In a
quantum mechanical treatment of the problem, the Coulomb interaction can
therefore be treated by first order perturbation theory. In this case, further cor-
rection factors occur, so that the Coulomb term, denoted here ECoulomb, finally
becomes [59, 60, 69]:
24 2 Physics and Chemistry of Colloidal Semiconductor Nanocrystals
e2 1 RNano \ aB 1:8 e2 1
ECoulomb ¼ ! ECoulomb ¼ ; ð2:6Þ
8per e0 aB n2 4per e0 RNano n2
where RNano is the radius of the semiconductor nanocrystals. This term introduces
a size-dependence to the energy of the lowest excited state. However, this term
alone cannot explain the quantum size effect, because smaller particle size will
increase the Coulomb attraction of the electron and hole and therefore lower the
energy, whereas an increasing band gap with decreasing particle size needs to be
explained.
Another effect playing a role, actually the major role in the case of strong
confinement, is that the energy of the charge carriers gets quantized, because they
must be considered as particles confined in a box. The simplest model is to
describe the situation by a spherical potential with infinite walls. A detailed
treatment of the problem of the infinite spherical potential well can for example be
found in [70]. The Hamiltonian of the corresponding stationary Schrödinger
equation reads as follows:
_ h2
h2 0; r RNano
H ¼ De Dh þ V ðr Þ with V ðr Þ ¼ ð2:7Þ
2me 2mh 1; r [ RNano
The potential has radial symmetry, and the solutions are characterized by a set
of three quantum numbers: the main quantum number n, the quantum number l for
the angular momentum and the magnetic quantum number m. Since the electron
and hole are not correlated by the potential assumed to be zero inside the nano-
crystals, they can be considered as independent particles in this problem. Thus,
there are two independent sets of quantum numbers n, l, m characterizing the states
of the electron and the hole, respectively. In analogy to the hydrogen atom, states
with l = 0, 1, 2, … are named as S, P, D, … states and have the multiplicity
2l ? 1. The eigenvalues of the energy of a given charge carrier (electron or hole)
are given by the following expression:
h2 v2n; l
Enl ¼ ; ð2:8Þ
2m R2Nano
where vn; l is the n-th root of the spherical Bessel function of the order l, and m is
the effective mass. The energy is independent of the quantum number m, and the
lowest state is an S-state with l = 0 and n = 1. The corresponding root of the
spherical Bessel function, v1; 0 , equals p, and the energy becomes [59]:
h2 p2
E1S ¼ ; ð2:9Þ
2m R2Nano
The lowest excited state of the electron–hole pair is obtained, when both the
electron and the hole are in the 1S state (noted as 1Se1Sh). The energy due to the
confinement in the spherical potential with infinite walls becomes then:
2.3 Material Properties Depending on Particle Size 25
Fig. 2.6 Size-dependent change of the energy of the lowest excited state in spherical CdSe
nanocrystals. The total size-dependent correction (thick solid lines) is the quantity E1Se 1Sh Eg as
calculated according to (2.12) using aB = 5.15 nm and er = 9.6 (black lines) or er = 6.2 (red
lines). The individual contributions due to the infinite spherical potential (thin dashed lines) and
to the Coulomb interaction (thin solid lines) are shown as well. For comparison, a curve fitted to
experimentally observed shifts of the first excitonic absorption peak of CdSe nanocrystals is
included (dots). This curve was calculated according to a formula deduced by Jasieniak et al.
[71], and a value of 1.74 eV was subtracted for the band gap of bulk CdSe [58]
sph:potential p2 h2 e2
E1Se 1Sh ¼ Eg þ E1S e 1Sh
þ ECoulomb ¼ Eg þ 2
1:8
2lRNano 4per e0 RNano
ð2:11Þ
This is the simplest theoretical description of the quantum size effect in the
strong confinement regime (RNano aB). To remind, the formula was developed
assuming a spherical potential with infinite walls and considers only Coulomb
interaction of the electron and hole. More precise, the Coulomb term results from
treating the Coulomb interaction by first-order perturbation theory. With the help
of (2.5), the result for the energy of the lowest excited state (2.11) can also be
written as a function of the dielectric constant, the exciton Bohr radius and the
particle radius:
pe2 aB e2
E1Se 1Sh ¼ Eg þ 2
1:8 ð2:12Þ
8er e0 RNano 4per e0 RNano
26 2 Physics and Chemistry of Colloidal Semiconductor Nanocrystals
Figure 2.6 visualizes the contributions of the different terms in (2.12) to the
size-dependent band gap of CdSe. For small particle sizes, the quantization term
due to the spherical potential is clearly dominant over the Coulomb term.
Figure 2.6 also compares the relatively simple equation (2.12) to experimental
data obtained on CdSe nanocrystals. Jasieniak et al. [71] and previously already
Yu et al. [72] established so-called sizing curves which relate the energy of the first
excitonic peak in absorption spectra of semiconductor nanocrystals to the particle
size. Figure 2.7 shows these curves established by Yu et al. [72] for different Cd
chalcogenides.
A more recent analysis by Jasieniak et al. [71] is in good agreement with the
curves shown in Fig. 2.7, and the corresponding curve is included into Fig. 2.6,
where the value of the band gap of bulk CdSe had to be subtracted to enable
comparison with the theoretical curves. In such a comparison, it is assumed that
the energy of the first excitonic peak in absorption spectra directly corresponds to
the transition with the lowest possible energy, i.e., to a transition with the energy
of the size-dependent band gap. To be precise, this is not completely evident.
A method to determine the band gap of a direct semiconductor is to plot the
quantity ðOD hvÞ2 , where OD is the optical density, against the photon energy hv.
2.3 Material Properties Depending on Particle Size 27
Extrapolating the linear part of the curve near the absorption onset to the inter-
ception with the photon energy axis yields the band gap of a direct semiconductor.
This method was also applied to determine the band gap of semiconductor
nanocrystals in some cases [62, 73]. Figure 2.8 illustrates this procedure on the
example of absorption spectra of small CdSe nanocrystals. As can be seen, the
method yields a slightly lower value for the band gap than simply assuming that
the energy of the first excitonic absorption maximum corresponds to the band gap.
Another important point is that many semiconductor nanocrystals exhibit
Stokes shifts [75], i.e., energetic differences between the maxima in absorption and
fluorescence spectra. This leads to discussion whether it is more correct to use
absorption of photoluminescence spectra to experimentally determine the energy
of the lowest excited state. Despite this discussion, one can clearly see from the
comparison made in Fig. 2.6 that the theory described in this chapter obviously
overestimates the size-dependent shift of the band gap energy of CdSe, in par-
ticular in the case of small particles.
A main deficiency is that the presented theoretical model neglects the envi-
ronment of the nanocrystals. Already in 1984, Brus [60] showed that a surrounding
medium with a different dielectric constant leads to a dielectric solvation energy
which can be described by a polarization term as another correction term to the
formula in (2.11). Another simplification is the assumption of a spherical potential
with infinite walls. More complex models have been developed to achieve better
agreement with the experimental findings. For example, Pellegrini et al. [64] used
28 2 Physics and Chemistry of Colloidal Semiconductor Nanocrystals
a spherical potential with finite walls and could reproduce the experimentally
observed shift of the band gap energy for several semiconductors very well.
Figure 2.9 shows data for CdS and CdSe quantum dots from that work.
The tuneable band gap of semiconductor nanocrystals is probably the most
important size-dependent material property with respect to applications in opto-
electronics. This phenomenon enables tuning the absorption range which is of
course relevant for efficient light absorption in solar cells. In the case of lumi-
nescent materials, the wavelength of band edge photoluminescence is directly
related to the band gap energy. Thus, tuning the band gap can also be used to
control the color of emitted light, which is for example interesting for light-
emitting diodes. Furthermore, the absolute energetic position of the valence band
maximum and conduction band minimum with respect to the vacuum level can be
tuned by changing the particle size. This can be relevant for devices where the
band alignment between several material components plays a role.
Beyond the optical properties, there is a variety of other material properties that
can be dependent on the particle size as well. For example, thermodynamic
properties such as the melting temperature or transition temperatures and pressures
between different crystallographic phases of a solid can depend on the particle size
[79–81]. Moreover, in the case of metals, magnetic properties [82, 83] or catalytic
properties [84–86] can be size-dependent. A discussion of all these phenomena is,
however, beyond the scope of this book.
2.4 Material Properties Related to the Surface of Colloidal Nanocrystals 29
Fig. 2.11 Schematic representation of different ligands binding to the nanocrystal surface
(a octanethiol-capped Au nanoparticles, b octylamine-capped Pt nanoparticles, c thioglycolic
acid capped CdSe nanoparticles). The aliphatic hydrocarbon chains of the alkanethiol or
alkylamine ligands provide solubility in non-polar organic solvents (a, b), whereas ligands with
additional polar functional groups provide solubility in polar solvents such as water (c). Please
note that it is not a priori clear how the bond between the ligands and the particle surface looks
like in detail. In the case of thiols bonding to Au or Cd chalcogenides (a, c), evidence was found
that the H atom of the thiol group can be split off, so that covalent bonds between the S atom and
a surface atom result [88, 89]. In case of the amino group (b), the author considers a coordinative
bond by the electron lone pair of the N atom to a surface atom most likely, although references
clearly supporting this assumption are absent
level 0.64 eV above the valence band maximum [92]. The example shows that the
energy levels of defects such as dangling bonds are size-dependent, and that they
can be located quite deep inside the band gap.
An energy level in the band gap can act as a trap state for charge carriers. If
colloidal nanocrystals shall for example be used as emitter in light-emitting diodes,
the elimination of such trap states can be crucial. Let us consider the photolu-
minescence of semiconductor nanocrystals, i.e., the radiative recombination of
electron–hole pairs after their creation by excitation with light. The photolumi-
nescence spectra of InP quantum dots which were shown in Fig. 1.4 correspond to
so-called band edge photoluminescence. This means that electrons and holes
recombine from the edges of the conduction band and valence band so that the
energy of the emitted photons corresponds to the band gap. In the case of a
defect-free nanocrystal, the radiative recombination of electrons and holes by band
edge photoluminescence should be the dominant process. By consequence, such
nanocrystals should have high photoluminescence quantum efficiency. (The
quantum efficiency is defined as the ratio of emitted photons to absorbed photons).
If however, trap states are present, electrons and holes can be captured in these
defect states. By consequence, band edge photoluminescence will be suppressed,
and the photoluminescence quantum yield will be reduced. What is the fate of an
electron captured in a trap state inside the band gap? A trapped electron has in
principal different possibilities to relax to the ground state. One possibility is
radiative recombination from defect states [93, 94]. In this case, photons with
32 2 Physics and Chemistry of Colloidal Semiconductor Nanocrystals
energy smaller than the band gap energy would be emitted. However, there are
also non-radiative recombination processes where the energy can for example be
dissipated into the lattice due to electron–phonon interactions. In this case, the
trapped charge carriers do not lead to any photoluminescence. It depends on the
concrete material system, what type of recombination processes really occur.
Figure 2.12 illustrates the described processes of absorption and photolumines-
cence in semiconductor nanocrystals. A more detailed discussion of recombination
processes that can occur in semiconductor nanocrystals can for example be found
in [61].
Since defects such as surface dangling bonds reduce the quantum efficiency of
band edge photoluminescence, the removal of dangling bonds at the surface is a
crucial issue in order to obtain highly luminescent nanocrystals. A possibility to
avoid deep trap states due to dangling bonds at the surface is their saturation with
organic ligand molecules binding to the nanocrystal surface [26]. For example,
CdSe quantum dots with a ligand shell composed of mixtures of trioctylphosphine/
trioctyphosphine oxide (TOP/TOPO) and hexadecylamine were reported to reach
photoluminescence quantum efficiency of *10–25 % at room temperature [32]. It
has to be noted in this place, that usually not all of the surface atoms are really
capped with ligand molecules. For example, if branched molecules like TOP are
used as ligands, their sterical demands are simply too large as to enable a complete
coverage of the surface.
2.4 Material Properties Related to the Surface of Colloidal Nanocrystals 33
Fig. 2.13 Schematic representation of the band structure of single material (a) and core–shell
nanocrystals with a type I heterostructure (b). The conduction and valence band edges are shown
as a function of the radial position. A rectangular potential well structure is assumed in this
simplified representation
Fig. 2.15 Absorptivity and normalized photoluminescence spectra of a pure CdTe nanocrystals
(3.2 nm radius, grey lines) and CdTe/CdSe core–shell nanocrystals (3.2 nm core radius, 1.1 nm
shell thickness, black lines), and b pure CdSe nanocrystals (2.2 nm radius, grey lines) and CdSe/
ZnTe core–shell nanocrystals (2.2 nm core radius, 1.8 nm shell thickness, black lines) (Reprinted
with permission from [95]. Copyright 2003 American Chemical Society)
References
10. M.M. Frank, M. Bäumer, Phys. Chem. Chem. Phys. 2, 3723 (2000)
11. M. Bäumer, M.M. Frank, M. Heemeier, R. Kühnemuth, S. Stempel, H.-J. Freund, Surf. Sci.
454–456, 957 (2000)
12. A.L. Rogach, N. Gaponik, J.M. Lupton, C. Bertoni, D.E. Gallardo, S. Dunn, N.L. Pira, M.
Paderi, P. Repetto, S.G. Romanov, C. O’Dwyer, C.M. Sotomayor Torres, A. Eychmüller,
Angew. Chem. Int. Ed. 47, 6538 (2008)
13. W.E.J. Beek, R.A.J. Janssen, in Hybrid Nanocomposites for Nanotechnology, ed. by L.
Merhari (Springer Science+Business Media, New York, 2009)
14. Y. Zhou, M. Eck, M. Krüger, Energy Environ. Sci. 3, 1851 (2010)
15. C.B. Murray, C.R. Kagan, M.G. Bawendi, Ann. Rev. Mater. Sci. 30, 545 (2000)
16. H. Weller, Philos. Trans. R. Soc. London A 361, 229 (2003)
17. Y. Yin, A.P. Alivisatos, Nature 437, 664 (2005)
18. X. Peng, Nano Res. 2, 425 (2009)
19. D.V. Talapin, A.L. Rogach, M. Haase, H. Weller, J. Phys. Chem. B 105, 12278 (2001)
20. H. Borchert, D.V. Talapin, N. Gaponik, C. McGinley, S. Adam, A. Lobo, T. Möller, H.
Weller, J. Phys. Chem. B 107, 9662 (2003)
21. C. Pacholski, A. Kornowski, H. Weller, Angew. Chem. Int. Ed. 41, 1188 (2002)
22. C. de Mello Donega, P. Liljeroth, D. Vanmaekelbergh, Small 1, 1152 (2005)
23. M. Protière, N. Nerambourg, O. Renard, P. Reiss, Nanoscale Res. Lett. 6, 472 (2011)
24. I. Moreels, J.C. Martins, Z. Hens, Chem. Phys. Chem. 7, 1028 (2006)
25. X. Wang, P. Sonström, D. Arndt, J. Stöver, V. Zielasek, H. Borchert, K. Thiel, K. Al-
Shamery, M. Bäumer, J. Catal. 278, 143 (2011)
26. K.E. Knowles, D.B. Tice, E.A. McArthur, G.C. Solomon, E.A. Weiss, J. Am. Chem. Soc.
132, 1041 (2010)
27. R.P. Andres, J.D. Bielefeld, J.I. Henderson, D.B. Janes, V.R. Kolagunta, C.P. Kubiak, W.J.
Mahoney, R.G. Osifchin, Science 273, 1690 (1996)
28. M. Bruchez Jr, M. Moronne, P. Gin, S. Weiss, A.P. Alivisatos, Science 281, 2013 (1998)
29. R. Rossetti, J.L. Ellison, J.M. Gibson, L.E. Brus, J. Chem. Phys. 80, 4464 (1984)
30. C.B. Murray, D.J. Norris, M.G. Bawendi, J. Am. Chem. Soc. 115, 8706 (1993)
31. J.E.B. Katari, V.L. Colvin, A.P. Alivisatos, J. Phys. Chem. 98, 4109 (1994)
32. D.V. Talapin, A.L. Rogach, A. Kornowski, M. Haase, H. Weller, Nano Lett. 1, 207 (2001)
33. J. Kolny-Olesiak, V. Kloper, R. Osovsky, A. Sashchiuk, E. Lifshitz, Surf. Sci. 601, 2667
(2007)
34. X. Peng, L. Manna, W. Yang, J. Wickham, E. Scher, A. Kadavanich, A.P. Alivisatos, Nature
404, 59 (2000)
35. L. Manna, E.C. Scher, A.P. Alivisatos, J. Am. Chem. Soc. 122, 12700 (2000)
36. S.D. Bunge, K.M. Krueger, T.J. Boyle, M.A. Rodriguez, T.J. Headley, V.L. Colvin, J. Mater.
Chem. 13, 1705 (2003)
37. R. Kho, C.L. Torres-Martinez, R.K. Mehra, J. Colloid Interface Sci. 227, 561 (2000)
38. W.-S. Chae, R.J. Kershner, P.V. Braun, Bull. Korean Chem. Soc. 30, 129 (2009)
39. P.D. Cozzoli, L. Manna, M.L. Curri, S. Kudera, C. Giannini, M. Striccoli, A. Agostiano,
Chem. Mater. 17, 1296 (2005)
40. A.A. Guzelian, U. Banin, A.V. Kadavanich, X. Peng, A.P. Alivisatos, Appl. Phys. Lett. 69,
1432 (1996)
41. A.A. Guzelian, J.E.B. Katari, A.V. Kadavanich, U. Banin, K. Hamad, E. Juban, A.P.
Alivisatos, R.H. Wolters, C.C. Arnold, J.R. Heath, J. Phys. Chem. 100, 7212 (1996)
42. D.V. Talapin, N. Gaponik, H. Borchert, A.L. Rogach, M. Haase, H. Weller, J. Phys. Chem. B
106, 12659 (2002)
43. R. Xie, M. Rutherford, X. Peng, J. Am. Chem. Soc. 131, 5691 (2009)
44. M. Kruszynska, H. Borchert, J. Parisi, J. Kolny-Olesiak, J. Am. Chem. Soc. 132, 15976
(2010)
45. I. Mekis, D.V. Talapin, A. Kornowski, M. Haase, H. Weller, J. Phys. Chem. B 107, 7454
(2003)
46. S. Haubold, M. Haase, A. Kornowski, H. Weller, Chem. Phys. Chem. 2, 331 (2001)
References 37
47. Y.W. Cao, U. Banin, J. Am. Chem. Soc. 122, 9692 (2000)
48. Q. Dai, Y. Wang, X. Li, Y. Zhang, D.J. Pellegrino, M. Zhao, B. Zou, J. Seo, Y. Wang, W.W.
Yu, ACS Nano 3, 1518 (2009)
49. E. Witt, F. Witt, N. Trautwein, D. Fenske, J. Neumann, H. Borchert, J. Parisi, J. Kolny-
Olesiak, Phys. Chem. Chem. Phys. 14, 11706 (2012)
50. A. Petrella, M. Tamborra, M.L. Curri, P. Cosma, M. Striccoli, P.D. Cozzoli, A. Agostiano, J.
Phys. Chem. B 109, 1554 (2005)
51. Y. Li, M. Zhang, M. Guo, X. Wang, Rare Met. 29, 286 (2010)
52. F. Söderlind, M.A. Fortin, R.M. Petoral Jr, A. Klasson, T. Veres, M. Engström, K. Uvdal, P.-
O. Käll, Nanotechnology 19, 085608 (2008)
53. N.R. Jana, X. Peng, J. Am. Chem. Soc. 125, 14280 (2003)
54. T. Hyeon, Chem. Commun. 927 (2003)
55. K. Ahrenstorf, O. Albrecht, H. Heller, A. Kornowski, D. Görlitz, H. Weller, Small 3, 271
(2007)
56. X. Wang, J. Stöver, V. Zielasek, L. Altmann, K. Thiel, K. Al-Shamery, M. Bäumer, H.
Borchert, J. Parisi, J. Kolny-Olesiak, Langmuir 27, 11052 (2011)
57. J. Park, J. Joo, S.G. Kwon, Y. Jang, T. Hyeon, Angew. Chem. Int. Ed. 46, 4630 (2007)
58. C. Kittel, Introduction to Solid State Physics, 8th edn. (Wiley, New York, 2005)
59. S.V. Gaponenko, Optical Properties of Semiconductor Nanocrystals (Cambridge University
Press, Cambridge, 1998)
60. L.E. Brus, J. Chem. Phys. 80, 4403 (1984)
61. J.Z. Zhang, J. Phys. Chem. B 104, 7239 (2000)
62. Y. Wang, A. Suna, W. Mahler, R. Kasowski, J. Chem. Phys. 87, 7315 (1987)
63. K.F. Yang, H.P.R. Frederikse, J. Phys. Chem. Ref. Data 2, 313 (1973)
64. G. Pellegrini, G. Mattei, P. Mazzoldi, J. Appl. Phys. 97, 073706 (2005)
65. H. Fu, L.-W. Wang, A. Zunger, Phys. Rev. B 59, 5568 (1999)
66. M. Navaneethan, K.D. Nisha, S. Ponnusamy, C. Muthamizhchelvan, Rev. Adv. Mater. Sci.
21, 217 (2009)
67. H. Du, C. Chen, R. Krishnan, T.D. Krauss, J.M. Harbold, F.W. Wise, M.G. Thomas, J.
Silcox, Nano Lett. 2, 1321 (2002)
68. A.L. Efros, M. Rosen, Annu. Rev. Mater. Sci. 30, 475 (2000)
69. D.E. Gomez, M. Califano, P. Mulvaney, Phys. Chem. Chem. Phys. 8, 4989 (2006)
70. S.G. Prussin, Nuclear Physics for Applications (Wiley-VCH, Weinheim, 2007)
71. J. Jasieniak, L. Smith, J. van Embden, P. Mulvaney, M. Califano, J. Phys. Chem. C 113,
19468 (2009)
72. W.W. Yu, L. Qu, W. Guo, X. Peng, Chem. Mater. 15, 2854 (2003)
73. H.S. Zhou, I. Honma, H. Komiyama, J.W. Haus, J. Phys. Chem. 97, 895 (1993)
74. F. Zutz, I. Lokteva, N. Radychev, J. Kolny-Olesiak, I. Riedel, H. Borchert, J. Parisi, Phys.
Status Solidi A 206, 2700 (2009)
75. O.I. Micic, H.M. Cheong, H. Fu, A. Zunger, J.R. Sprague, A. Mascarenhas, A.J. Nozik, J.
Phys. Chem. B 101, 4904 (1997)
76. T. Vossmeyer, L. Katsikas, M. Giersig, I.G. Popovic, K. Diesner, A. Chemseddine, A.
Eychmüller, H. Weller, J. Phys. Chem. 98, 7665 (1994)
77. M.V. Rama Krishna, R.A. Friesner, Phys. Rev. Lett. 67, 629 (1991)
78. L.-W. Wang, A. Zunger, Phys. Rev. B 53, 9579 (1996)
79. J.Z. Jiang, L. Gerward, D. Frost, R. Secco, J. Peyronneau, J.S. Olsen, J. Appl. Phys. 86, 6608
(1999)
80. J.Z. Jiang, L. Gerward, R. Secco, D. Frost, J.S. Olsen, J. Truckenbrodt, J. Appl. Phys. 87,
2658 (2000)
81. C.-J. Lee, A. Mizel, U. Banin, M.L. Cohen, A.P. Alivisatos, J. Chem. Phys. 113, 2016 (2000)
82. E.V. Shevchenko, D.V. Talapin, A. Kornowski, F. Wiekhorst, J. Kötzler, M. Haase, A.L.
Rogach, H. Weller, Adv. Mater. 14, 287 (2002)
83. J.-I. Park, M.G. Kim, Y.-W. Jun, J.S. Lee, W.-R. Lee, J. Cheon, J. Am. Chem. Soc. 126, 9072
(2004)
38 2 Physics and Chemistry of Colloidal Semiconductor Nanocrystals
Abstract In 2000, the Nobel Prize in Chemistry was awarded to the researchers
Alan J. Heeger, Alan G. MacDiarmid and Hideki Shirakawa for the discovery and
development of conductive polymers, the original work going back to the late
nineteen seventies. Polymers are normally known as electrically insulating
materials. However, certain polymeric structures possess a special electronic
configuration which renders the materials electrically conductive. From the dis-
covery of conductive polymers arose a new area of research and engineering: the
field of organic electronics. Today, several types of conductive polymer are
known, and the field develops rapidly. In the present chapter, an overview on the
physical and chemical properties of this exciting class of materials is given.
Fundamentals enabling the conduction of electricity by organic materials are
summarized, a brief overview on different types of conductive polymer is given,
and selected physical and chemical properties with relevance for applications in
organic electronics are discussed.
3.1.1 Hybridization
Organic chemistry is funded on the chemistry of carbon. Other elements which can
occur in organic molecules are hydrogen, oxygen, nitrogen, sulfur, phosphorous,
and halides, i.e., fluorine, chlorine, bromine, and iodine. In order to understand the
phenomenon of electrical conductivity in organic materials, we have to regard the
electronic configuration of the molecules. The carbon atom has six electrons.
Obeying to Hund’s rules, the electrons occupy in the ground state the 1s and
2s orbitals completely, and two of the three degenerated 2p orbitals with a single
electron. The electronic configuration can thus be noted as (1s)2(2s)2(2px)1(2py)1 [1].
Fig. 3.1 Schematic representation of the electronic configuration of a carbon atom in the ground
state. a The 1s and 2s orbitals are filled with two electrons of opposite spin. The remaining two
electrons occupy two of the degenerate 2p orbitals, so that the total spin is maximized. Part
b shows an excited state where one of the 2s electrons is raised into the last 2p orbital
Figure 3.1a shows a schematic representation of the atomic orbitals and their
occupation in the ground state.
In the ground state configuration, the carbon atom would be able to establish two
covalent bonds to other atoms in a molecule, the angle between the bonds being 90
as imposed by the orientation of the perpendicular 2p orbitals. This is, however, not
going to occur in nature. Instead, hybridization takes place, meaning that the carbon
atom forms hybrid orbitals. The phenomenon of hybridization shall be briefly
summarized here. More detailed descriptions can for example be found in [1].
Briefly, hybrid orbitals of carbon are mixtures of the 2s and 2p orbitals.
Mathematically, the mixtures are obtained as linear combinations of the 2s and
2p wave functions. To reach the hybridized state, one can imagine that the carbon
atom is first excited by the promotion of one of the 2s electrons into the third
empty 2p orbital, as depicted in Fig. 3.1b. In such an excited state, all of the
orbitals belonging to the L-shell, i.e., to the main quantum number n = 2, are filled
with one electron. There are now three possibilities to form hybrid orbitals. In the
first case, the 2s and all three of the 2p orbitals participate in the hybridization.
From the linear combinations of the initial orbitals result then four equivalent sp3
hybrid orbitals which have shapes similar to asymmetric dumbbells and which
point into the corners of a regular tetrahedron. In the sp3 hybridized carbon atom,
each of the four hybrid orbitals is occupied with one electron. Figure 3.2a illus-
trates the shape and orientation of the sp3 hybrid orbitals.
The question arises, why the carbon atom should form such hybrid orbitals, if
reaching the hybridized state requires exciting one of the 2s electrons from the
ground state. To answer this question, the environment of the carbon atom in a
given molecule has to be considered. For atomic carbon, the sp3 hybridized state is
higher in energy than the ground state. So, hybridization will not take place in the
case of atomic carbon. However, the situation is different in a molecule. When
regarding for example methane (CH4) as one of the simplest hydrocarbons,
establishing four bonds between the C atom and the four H atoms would not be
possible with the ground state configuration of the carbon atom. As explained
3.1 Electrical Conductivity in Organic Materials 41
Fig. 3.2 a Schematic representation of the sp3 hybrid orbitals of carbon pointing into the corners
of a regular tetrahedron (which is equivalent to four corners of a cube). The shape of the hybrid
orbitals is asymmetric and is sketched here only schematically. The different colors, dark and
light green, represent regions where the wave function corresponding to the sketched orbitals has
opposite sign. b Illustration of a methane (CH4) molecule, where 4 covalent bonds are formed by
the overlap of the 1s orbitals of the H atoms with the sp3 orbitals of the C atom. The relative sizes
of the orbitals are arbitrary in the sketch
above, only two covalent bonds to neighboring atoms could be established with the
participation of the two electrons in the 2px and 2py orbitals, respectively. In
contrast, if the C atom is in the sp3 hybridized state, the four singly occupied
hybrid orbitals can overlap with the 1s orbitals of four H atoms, as illustrated in
Fig. 3.2b. In other words, sp3 hybridized carbon is able to establish covalent bonds
to four neighboring atoms. Since each covalent bond corresponds to binding
energy of the molecule, the expense of energy to excite the C atom will finally be
overcompensated by the establishment of more covalent bonds. This is the reason
why hybridization is favorable and takes place. sp3 hybridized carbon occurs not
only in methane, but in a large variety of organic compounds, wherever the C
atoms have four neighbors. For example, all alkanes (CnH2n+2) have their carbon
atoms in the sp3 configuration.
However, sp3 hybridization is not the only possibility. Considering for example
an ethene molecule (C2H4, common name: ethylene), it is fictively possible to
build the molecule with sp3 hybridized carbon atoms. In that case, a C–C bond
would result from the overlap of two hybrid orbitals belonging to the respective
atoms. Each C atom would then have three singly occupied sp3 orbitals left. Four
of these six orbitals could be used to establish bonds to the four H atoms, but two
singly occupied sp3 orbitals would be left and simply point into the space. This
fictive structure is not existent in nature. Instead, the ethene molecule has a con-
figuration where also the last two electrons contribute to the binding energy of the
molecule. This is possible with the C atoms in a sp2 hybridized state. In that case,
hybrid orbitals are formed by linear combination of the 2s and only two of the
three 2p orbitals. The last 2p orbital, usually selected to be the 2pz orbital, does not
take part in the hybridization process. The sp2 hybrid orbitals look in shape also
similar to asymmetric dumbbells, but differ in detail from the sp3 orbitals.
42 3 Physics and Chemistry of Conductive Polymers
Furthermore, the three sp2 orbitals are lying in a plane which is perpendicular to
the axis of the remaining 2p orbital. Figure 3.3a shows the orbitals of a C atom in
the sp2 configuration. With carbon in the sp2 hybridized state, the structure of
ethene is as follows: A C–C bond is formed by the overlap of two sp2 orbitals from
the respective atoms. The four H atoms are attached to the remaining four sp2
hybrid orbitals. The last two valence electrons are in the 2pz orbitals which are
perpendicular to the plane of the hybrid orbitals. In contrast to the fictive structure
discussed before, these electrons can now contribute to the bonding, because two
molecular orbitals are formed from the overlap of both 2pz orbitals: one bonding
molecular orbital and one anti-bonding molecular orbital. Since each of the
C atoms provides one electron, both electrons can be in the bonding molecular
orbital, and an additional bond results between the carbon atoms. Therefore, the
sp2 configuration is more favorable than the sp3 configuration in the case of this
molecule. The structure of ethene is schematically shown in Fig. 3.3b.
Thus, ethene has a double bond between the carbon atoms. The bond resulting
from the overlap of the sp2 hybrid orbitals is called a r-bond, because it has
rotational symmetry with respect to the C–C axis. The bond resulting from the
overlap of the 2pz orbitals is not rotationally symmetric and is called a p-bond.
For completeness, the last case of hybridization is the formation of sp orbitals.
In that case, only one of the three 2p orbitals takes part in the hybridization process
together with the 2s orbital [1]. The sp hybrid orbitals resemble asymmetric
dumbbells as well. An example for a molecule, where C is present in the sp
hybridized state is ethyne (C2H2, common name: acetylene). Ethyne is a linear
molecule, meaning that all atoms are arranged on one axis. Between the two
C atoms results a 3-fold bond: one r-bond originating from the overlap of the sp
hybrid orbitals, and two p-bonds originating from the overlap of the 2p orbitals not
participating in the hybridization.
Fig. 3.3 a Schematic representation of the sp2 hybrid orbitals of carbon. The rotational axes of
the three orbitals are in a plane. The 2pz orbital is not participating in the hybridization and is
oriented perpendicular to the plane. The shape of all orbitals is sketched only schematically, and
the different colors (dark or light color) represent regions where the wave function belonging to
the orbitals has opposite sign. b Illustration of an ethene (C2H4) molecule. 4 covalent bonds are
formed by the overlap of the 1s orbitals of the H atoms with the sp2 hybrid orbitals of the C
atoms. Between the carbon atoms results a double bond: one r-bond due to the overlap of the sp2
orbitals and one p-bond due to the interaction of the 2pz orbitals. The relative sizes of the orbitals
are arbitrary in the sketch
This depiction of the molecule suggests that the 2pz orbitals of the first and
second C atom in the chain interact to form a double bond. Similarly, a double
bond would be formed between the third and fourth C atom. In contrast, no
interaction would occur between the 2pz orbitals of the second and third C atom, so
that here, only a single bond is formed. However, the 2pz orbital of the second C
atom has a certain overlap with both, the 2pz orbitals of the first and the third
C atom. Thus, there is no physical reason why a double bond should be formed
with only one of the neighboring C atoms, whereas the corresponding interaction
should be suppressed in case of the other neighboring C atom. Therefore, in
reality, the bonds in the molecule are different from the simple situation suggested
by the structural formula. In fact, there is of course interaction between the 2pz
orbitals of the second and third C atoms. More precise, all of the four 2pz orbitals
interact, and four molecular orbitals (two bonding and two anti-bonding molecular
orbitals) are formed which are spatially extended over the whole carbon chain [2].
By consequence, the electrons in these orbitals, called p-electrons, are delocalized
over the whole carbon chain.
Despite the interaction of all the four 2pz orbitals, the bonds between the carbon
atoms in 1,3-butadiene are not all equal. The bond length between the carbon
atoms C1 and C2 as well as C3 and C4 is 134 pm [2], which is close to a normal
C=C double bond in a hydrocarbon chain. The bond length between the atoms C2
44 3 Physics and Chemistry of Conductive Polymers
and C3 is 147 pm [2]. This is longer than a normal double bond, but significantly
shorter than a normal C–C single bond like occurring in butane (154 pm [2]). The
difference in the bond lengths between the respective C atoms is probably the
reason, why the molecule is usually depicted by the simple structural formulas as
in Fig. 3.4, although the p-electrons are delocalized. From an electronic point of
view, the delocalization of the p-electrons over the whole carbon chain has
important consequences. As will be discussed later, delocalization of p-electrons is
the fundament of electrical conductivity in organic compounds.
Another important example for the delocalization of p-electrons meriting
special attention is the benzene molecule (C6H6). In this molecule, the sp2
hybridized carbon atoms form a planar ring, and each C atom has a 2pz orbital
oriented perpendicular to the plane of the carbon ring. The molecule might be
thought to have conjugated double bonds, as depicted in the skeletal formula in
Fig. 3.5a.
However, the configuration shown in Fig. 3.5b is obviously similar. Both
configurations can be considered as equivalent resonance structures of the benzene
molecule. Because of the symmetry of the molecule, it is obvious that the 2pz
orbital of a given C atom will interact similarly with the 2pz orbitals of both
adjacent C atoms. The interaction of all 6 of the 2pz orbitals results in molecular
orbitals which are delocalized over the whole carbon ring [2]. In the ground state,
the 6 p-electrons occupy the three binding molecular orbitals and are delocalized
over the ring. All C–C bonds have an identic length of 139 pm which is in between
of the length of normal single and double bonds [2]. The ground state with the
electrons delocalized over the ring is significantly lower in energy than the reso-
nance structures with localized double bonds shown in parts a and b of Fig. 3.5.
To depict the ground state of the molecule with the delocalized p-electrons, the
formula shown in Fig. 3.5c has been established.
3.1 Electrical Conductivity in Organic Materials 45
Fig. 3.5 a A resonance structure of benzene showing the molecule with alternating single and
double bonds. This situation does, however, not correspond to the ground state of the molecule.
b This configuration of the molecule is equivalent to the structure in (a). c In the ground state of
the benzene molecule, all of the p-electrons are delocalized over the carbon ring, and all C–C
bonds have the same length. To depict this situation, the electrons are symbolized by a circle in
the carbon ring
The first works on conductive polymers, which led later to the Nobel Prize in
Chemistry, concentrated on trans-polyacetylene [3]. This polymer is a hydrocar-
bon with a long chain of sp2 hybridized C atoms, and its structure is shown in
Fig. 3.6.
The question arises, what results now from the interaction of the large number
of 2pz orbitals aligned in the polymer chain of sp2 hybridized carbon atoms. In
analogy to the concept of energy bands known in solid state physics [4], we might
expect the following: Starting from an atom, the overlap of two atomic orbitals in a
2-atomic molecule leads to the formation of two molecular orbitals, one of them
bonding and one of them anti-bonding [1]. As we have seen on the example of 1,3-
butadiene, the linear combination of 4 atomic orbitals resulted in 4 molecular
orbitals, half of them bonding, and half of them anti-bonding. This concept can be
continued. Within the tight-binding model, where the wave function describing
electrons in a solid can be expressed as a linear combination of atomic orbitals, the
combination of N atomic orbitals leads then to the formation of a quasi-continuous
energy band comprising 50 % bonding and 50 % anti-bonding states [4]. This
situation is illustrated in Fig. 3.7.
According to this picture, the interaction of N 2pz orbitals in a chain of sp2
hybridized carbon atoms might result in the formation of a quasi-continuous
energy band. Because of the electron spin, each orbital can be occupied by two
electrons. Therefore, an energy band resulting from the linear combination of
N atomic orbitals can be filled with 2N electrons. Since every carbon atom has only
one electron in its 2pz orbital, the energy band would be half-filled. By conse-
quence, the polymer considered here should behave like a metal. This is, however,
not the case. Undoped trans-polyacetylene has a conductivity of approximately
10-5 S/cm and behaves like a semiconductor [5]. Thus, reality must be different
from the simple picture discussed in Fig. 3.7.
The reason for the semiconducting behavior is the so-called Peierls instability
[4, 5]. The corresponding theory was initially developed for one-dimensional
metals. For a one-dimensional metal with a half-filled conduction band, it can be
46 3 Physics and Chemistry of Conductive Polymers
Fig. 3.6 a The structure of trans-polyacetylene is a long hydrocarbon chain with conjugated
double bonds. The polymer chain has alternating bond length. However, precisely, the length of
the ‘‘double bonds’’ is longer than that of isolated double bonds, and the length of the ‘‘single
bonds’’ is shorter than that of normal C–C single bonds. b The polymeric structure can be noted
depicting only the repetition unit, with the index n standing for the number of units (monomers)
in the chain. c This configuration of trans-polyacetylene is energetically similar to the species
in (a)
Fig. 3.7 Schematic illustration of the formation of energy bands in a solid from the interaction of
atomic orbitals. In a 2-atomic molecule, the linear combination of 2 atomic orbitals leads to the
formation of 2 molecular orbitals. In a solid consisting of N atoms, the combination of N atomic
orbitals leads within the tight-binding approximation to an energy band comprising N energy
levels. If N 1, the levels in the band form a quasi-continuum of states
shown that a lattice distortion doubling the length of the unit cell, and thus halving
the extension of the first Brillouin zone, leads to an energy gap around the Fermi
energy [4, 5]. Transferred to the case of trans-polyacetylene, this means that a
distortion of the polymer chain leading to alternating bond length, and thus dou-
bling the length of the repetition unit, causes the appearance of an energy gap in
the middle of the half-filled energy band discussed before for the hypothetic case
of completely delocalized p-electrons [5–7]. Thus, the band is split into two bands
separated by an energy gap. Figure 3.8 illustrates this phenomenon.
3.1 Electrical Conductivity in Organic Materials 47
Fig. 3.8 Schematic representation of the energy bands originating from the interaction of the 2pz
orbitals of the carbon atoms in a long chain of sp2 hybridized C atoms, like occurring in trans-
polyacetylene. The dashed curve represents the band which would be expected for the hypothetic
case, where all C–C bonds would have an identical length a. The p electrons would then be
completely delocalized, and the band would be half-filled. The solid curve corresponds to a
distorted polymer chain, where the C–C bonds have alternating length. In this case, the spatial
period is doubled (2a), and the first Brillouin zone extends from -p/(2a) to +p/(2a). The most
important consequence is the appearance of an energy gap at the border of the Brillouin zone. The
band bending near the energy gap lowers the total energy of the filled states. This is the reason
why the configuration with identical bond lengths is unstable against the lattice distortion leading
to alternating bond lengths. This phenomenon is known as Peierls instability
Every C atom in the carbon chain supplies one p-electron to the bands. Thus, in
the case of alternating bond length, the band below the energy gap will be filled
completely, whereas the band above the gap will be empty (at T = 0 K). Due to
the band bending near the energy gap, the total energy of all electrons in the filled
band will be lower as compared to the case of completely delocalized p-electrons
with all C–C bonds being equivalent in length. This gain in energy is larger than
the difference in energy related to the different degree of delocalization. This is the
reason, why the structure with alternating bond length is preferred. It should be
noted, however, that the bonds cannot be considered as real single and double
bonds. The structural formulas used in Figs. 3.6 and 3.8 represent a simplified
picture. In reality, there is an alternation of the bond length in the chain, but with
all length being in between the length of normal single and double bonds.
The most important consequence of the Peierls instability is the appearance of
the energy gap. Since at T = 0 K the lower energy band would be completely
filled, whereas the higher band would be empty, the Fermi energy does no longer
fall into a partly filled band. Therefore, trans-polyacetylene does not behave like a
metal. The type of conductivity is governed by the width of the energy gap. For
trans-polyacetylene, the energy gap amounts to about 1.5 eV [5, 7]. By conse-
quence, this polymer behaves like a semiconductor.
48 3 Physics and Chemistry of Conductive Polymers
Fig. 3.9 a–c Representation of a neutral solution in trans-polyacetylene (a), a positively charged
soliton (b), and a negatively charged soliton (c), respectively. In reality, the delocalization
extends over more C atoms than in the sketch. d Energetically, a neutral soliton corresponds to a
singly occupied mid-gap state. In a positive soliton, the mid-gap state is empty, and in the case of
a negative soliton, it is occupied by two electrons of opposite spin
If the dopant atoms have less valence electrons, additional holes are provided, and
the semiconductor is said to be p-doped. In an organic semiconductor, the goal of
doping is similar. The intention is to provide additional p-electrons (n-doping) or to
remove p-electrons, which means to provide holes (p-doping). However, excess
charges are not introduced by replacing some atoms in the organic molecules.
Instead, other strategies can be pursued [5], one of them being chemical doping.
Chemical doping means that another compound is added to the organic semicon-
ductor, and a redox reaction alters its oxidation state [5]. For example, trans-
polyacetylene can be doped with iodine which acts as an oxidizing agent. In the
redox reaction, an electron is transferred from the polymer to the dopant which is
reduced from I2 to I3 . The polymer itself is oxidized in the process, meaning that
an electron is removed, or, in other words, a hole is injected. The addition of iodine
causes therefore p-doping of the polymer [3, 5]. In the case of trans-polyacetylene,
the electrons which are most easy to remove are the electrons in the soliton levels.
Thus, the addition of the oxidizing agent will oxidize the solitons. Figure 3.9b
shows schematically the structure of a positively charged soliton in p-doped trans-
polyactyene. Solitons possess certain mobility in the polymer, meaning that the
defect can move along the chain. Since positive solitons carry electrical charge,
p-doping significantly increases the conductivity of trans-polyacetylene.
In more detail, the increase in conductivity upon doping has further reasons
related to the charged solitons. First, it is important to note that a relatively high
concentration of charged solitons can exist on the polymer chain. If two neutral
solitons diffusing along the chain meet, they can annihilate [6], as one can easily
understand when reminding that a neutral soliton constitutes a boundary between
50 3 Physics and Chemistry of Conductive Polymers
Fig. 3.11 A selection of conductive polymers with relevance for organic electronics
The previous section focused on trans-polyacetylene. This was the first conjugated
polymer, where electrical conductivity was observed and studied. Today, a man-
ifold of conductive polymers is known. In this section, a brief overview is given
over a selection of polymers which are of interest for the field of organic elec-
tronics. Most of them contain aromatic ring systems. Figure 3.11 shows the
molecular structure of a few selected species.
A widely used polymer in the field of organic solar cells is polythiophene, or
more precisely, derivatives of it. Pure polythiophene is an organic semiconductor,
but it has poor solubility. To enable the preparation of thin films by deposition of
the polymer from solution, the solubility in appropriate organic solvents needs to
be improved. This is possible by attaching side chains to the thiophene rings [5].
52 3 Physics and Chemistry of Conductive Polymers
Typically, alkyl chains with 6–8 carbon atoms are used for this purpose. The prob-
ably most common derivative of polythiophene is poly(3-hexylthiophene) (P3HT).
The concept of using side chains without conjugated double bonds to provide the
polymer backbone with the aromatic rings solubility is also used in other types of
polymer. For example, poly(para-phenylene vinylene) (PPV) can be modified with
side groups, common derivatives being poly[2-methoxy-5-(20 -ethylhexyloxy)-para-
phenylene vinylene] (MEH-PPV) or poly[2-methoxy-5-(30 ,70 -dimethyloctyloxy)-
1,4-phenylene vinylene] (MDMO-PPV).
Polythiophene, PPV and their derivatives have a strong absorption in the visible
range of the spectrum. Therefore, such conductive polymers are relevant as light
absorbers in organic solar cells. However, electrical conductivity is of course not
always accompanied by absorption properties suitable for light harvesting. For
example, poly(3,4-ethylenedioxythiophene) doped with poly(styrenesulfonate)
(PEDOT:PSS), or also polyaniline (PANI) have only a low absorption coefficient
in the visible range. They are therefore not as interesting as light absorbers, but can
in contrast be used for charge transport layers with reasonable transparency [9].
The polymers shown in Fig. 3.11 possess one fundamental difference to trans-
polyacetylene: They do not have degenerate ground states. This can easily be seen,
when comparing the two structures of a given polymer which could be obtained by
flipping all double bounds, as illustrated in Fig. 3.12 for the case of PPP. The
structure with three double bonds in the aromatic ring is called the benzenoid
structure, the other one the quinoid structure. According to Hückel’s rule, the
benzenoid form is an aromatic system, whereas the quinoid form is not aromatic.
The benzeniod species has a lower energy than the quinoid one. This has conse-
quences for the charge transport properties, because in polymers without degen-
erate ground state configurations, solitons are not stable. However, there exists
another type of quasi-particles occurring instead, namely polarons.
In inorganic semiconductors, polarons are known as quasi-particles resulting
from the interaction of electrons with phonons. In a very simple picture, one can
imagine that the Coulomb attraction between an electron in the conduction band
and the ionic cores of the crystal lattice causes a slight deformation or distortion of
the lattice. The electron together with the lattice distortion can then be treated like
a quasi-particle. An important consequence of the interaction is that the effect
slightly increases the effective mass of the electrons, because the lattice distortion
has to follow the movement of the electron [4].
3.2 Different Types of Conductive Polymer 53
Electrical conductivity is of course not the only material property which is relevant
for the application of organic semiconductors in electronic devices. Other relevant
parameters are for example the optical properties, the stability of the material
against temperature and environmental influences like oxygen or moisture, and
also the mechanical properties. A selection of material properties shall be briefly
discussed in the following.
If all molecules of the ensemble had the same chain length, both average values
for the molecular weight would be identical. However, due to the distribution of
chain length in a real sample, the weight average value is higher than the number
average value. The ratio between both values is called the polydispersity index, as
defined by (3.3):
Mw
PDI ¼ ; ð3:3Þ
Mn
Thus, a PDI close to one indicates a narrow distribution of the molecular weight
of the individual molecules in the polymer sample.
Some polymers possess a special structural property which is known as regio-
regularity. The most prominent example is probably poly(3-hexylthiophene). To
provide the polythiophene solubility in various organic solvents, 3-hexylthiophene,
i.e., a thiophene with a hexyl side chain, is used as monomer for the polymerization
reaction. From a synthetic point of view, there arise several possibilities to couple
the monomers into a polymer chain, as far as the arrangement of the hexyl side
groups is concerned. Regarding the bond between two given thiophene rings, the
side groups can be located at the C atoms close to the bond or at the C atoms further
away from the bond. In regiorandom P3HT, illustrated in Fig. 3.15a, it is arbitrary
56 3 Physics and Chemistry of Conductive Polymers
to which of the C atoms the side chains are attached. However, it is also possible to
achieve a regular arrangement of the side groups. In this case, the polymer is said to
be regioregular (rr). There exist four types of regioregular P3HT, two of them being
represented in Fig. 3.15b, c.
Regioregularity has a significant impact on other properties of P3HT, because
regioregular P3HT can form a crystalline phase where the individual molecules are
arranged in an ordered manner. Due to the regular sequence of the alkyl side
chains, adjacent polymer chains can align with their alkyl side chains penetrating
each other. This leads to the formation of so-called lamellae. The lamellae, in turn,
can be stacked, so that the p orbital systems of adjacent lamellae are overlapping.
This is called p-p stacking. From this arrangement result crystalline domains with
three characteristic directions, corresponding to the conjugated polymer back-
bones, the p-p stacking, and the stacking of the alkyl side chains, respectively.
Figure 3.16 shows schematically the formation of such ordered domains. The
molecular order can experimentally be studied for example by X-Ray diffraction
[12] or with the help of optical spectroscopy methods with polarized light [13].
X-ray diffraction revealed that the distance between adjacent polymer chains is
approximately 3.8 Å in the p-p stacking direction, and 16.4 Å in the direction of
the interdigitated alkyl side chains, respectively [12]. With regiorandom P3HT, the
crystallization is inhibited, because the random sequence of the alkyl side chains
prevents the ordered alignment of the individual polymer chains. The establish-
ment of molecular order and crystalline domains can have a strong impact on the
optical and electrical properties of P3HT films.
3.3 Physical and Chemical Properties of Conductive Polymer 57
Many conjugated polymers have a strong absorption in the visible range which
makes them interesting as light absorbing material in organic solar cells.
Figure 3.17 shows as example UV-Vis absorption spectra of thin films of regio-
regular P3HT. Several characteristics are noteworthy, here. It can be seen that the
absorption maximum is located at about 550 nm, thus well in the visible range of
the spectrum. Also, the absorption coefficient is relatively high. It was reported to
be of the order of 3.5 9 105 cm-1 at the maximum [14]. Using Beer’s Law
(compare Sect. 8.1), one can estimate that a layer of only 100 nm thickness will
absorb around 97 % of the light at the wavelength corresponding to the absorption
maximum. Thus, very thin polymer layers are sufficient to achieve a strong light
absorption. On the other hand, the absorption is limited to a relatively narrow
spectral range. In the case of regioregular P3HT, the absorption range extends
approximately from 400 nm to 650 nm (see Fig. 3.17). In view of a solar cell this
means that all photons with wavelength above *650 nm cannot be absorbed. In
comparison, a classical silicon solar cell absorbs light up to *1,130 nm. The
narrow absorption range of P3HT is a disadvantage limiting the performance of
corresponding organic solar cells.
Regarding the absorption spectra of P3HT in more detail, different absorption
bands are visible in Fig. 3.17. The appearance of these bands is related to the
regioregularity which enables the formation of crystalline domains. Films prepared
with regiorandom P3HT show only one broad absorption band centered at about
450 nm [13]. The shift to higher wavelength and the appearance of the fine
structure with distinct peaks at *520 nm, *550 nm, and *610 nm, respec-
tively, are due to the establishment of molecular order [13–15]. Figure 3.17 shows
furthermore that the absorption properties can change upon thermal treatments of
the films. Annealing enables a rearrangement of the individual molecules in the
58 3 Physics and Chemistry of Conductive Polymers
Fig. 3.18 a Molecular structures of a low band gap polymer, named PTB1, and the fullerene
derivatives PC61BM and PC71BM, respectively. b Normalized absorption spectra of the polymer
PTB1 dissolved in dichlorobenzene (squares), of a PTB1 film (circles), and of a film prepared
from a PTB1:PC61BM blend (triangles) (Reprinted with permission from [22]. Copyright 2009
American Chemical Society)
film and thus facilitates the establishment of crystalline domains in the films of
regioregular P3HT. By consequence, the fine structure becomes more pronounced
in the annealed films.
As mentioned above, the relatively narrow absorption range of P3HT, which is
caused by the relatively large HOMO-LUMO gap of approximately 1.9 eV [16],
constitutes a limitation for the performance of organic solar cells with this polymer.
Based on several assumptions, Scharber et al. [17] performed calculations to pre-
dict the possible performance of organic bulk heterojunction solar cells with PCBM
and conductive polymer. According to their investigations, the efficiency is largely
determined by the HOMO-LUMO gap of the donor polymer, and by the absolute
position of the polymer’s LUMO level with respect to that of PCBM [17]. Whereas
solar cells with P3HT/PCBM as donor/acceptor system were reported to achieve at
the maximum about 5 % efficiency [12], up to *11 % efficiency were predicted to
be possible with a polymer having a smaller HOMO-LUMO gap of about 1.5 eV
and the LUMO level being located 0.3 eV above that of PCBM [17, 18].
The realization that smaller energy gaps should be more suitable for the
application of conductive polymer in organic photovoltaics gave rise to intensive
research on the development of low band gap polymers [16]. Indeed, progress was
made by using new polymers with smaller energy gaps [19–21]. Figure 3.18 shows
an example for the absorption by a low band gap polymer that turned out to be
suitable for organic solar cells. In combination with PC71BM (see Fig. 3.18a),
5.6 % power conversion efficiency were achieved under standard test conditions in
this case [22]. So far, the highest efficiency reported in a peer-reviewed scientific
journal for solar cells with a single BHJ layer is 7.4 % [21]. The corresponding
polymer had a structure basically similar to that in Fig. 3.18a, but with other side
3.3 Physical and Chemical Properties of Conductive Polymer 59
References
1. W. Demtröder, Atoms, Molecules and Photons, 2nd edn. (Springer, Berlin, 2010)
2. K.P.C. Vollhardt, N.E. Shore, Organic Chemistry, 6th edn. (W. H. Freeman and Company,
New York, 2011)
3. H. Shirakawa, E.J. Louis, A.G. MacDiarmid, C.K. Chiang, A.J. Heeger, J, Chem. Soc. Chem.
Commun. 16:578 (1977)
4. C. Kittel, Introduction to Solid State Physics, 8th edn. (Wiley, New York, 2005)
5. A.J. Heeger, Angew. Chem. Int. Ed. 40, 2591 (2001)
6. M. Rehahn, Chem. unserer Zeit 37, 18 (2003). (in German)
7. A. Moliton, R.C. Hiorns, Polym. Int. 53, 1397 (2004)
8. K. Michielsen, H. De Raedt, Europhys. Lett. 34, 435 (1996)
9. B. Ecker, J.C. Nolasco, J. Pallares, L.F. Marsal, J. Posdorfer, J. Parisi, E. von Hauff, Adv.
Funct. Mater. 21, 2705 (2011)
10. J.L. Bredas, G.B. Street, Acc. Chem. Res. 18, 309 (1985)
11. G. Harbeke, D. Baeriswyl, H. Kiess, W. Kobel, Phys. Scr. T13, 302 (1986)
12. W. Ma, C. Yang, X. Gong, K. Lee, A.J. Heeger, Adv. Funct. Mater. 15, 1617 (2005)
13. M.C. Gurau, D.M. Delongchamp, B.M. Vogel, E.K. Lin, D.A. Fischer, S. Sambasivan, L.J.
Richter, Langmuir 23, 834 (2007)
14. P.D. Cunningham, L.M. Hayden, J. Phys. Chem. C 112, 7928 (2008)
15. L. Li, G. Lu, X. Yang, J. Mater. Chem. 18, 1984 (2008)
16. R. Kroon, M. Lenes, J.C. Hummelen, P.W.M. Blom, B. de Boer, Polym. Rev. 48, 531 (2008)
17. M.C. Scharber, D. Mühlbacher, M. Koppe, P. Denk, C. Waldlauf, A.J. Heeger, C.J. Brabec,
Adv. Mater. 18, 789 (2006)
18. T. Ameri, G. Dennler, C. Lungenschmied, C.J. Brabec, Energy Environ. Sci. 2, 347 (2009)
19. S.H. Park, A. Roy, S. Beaupre, S. Cho, N. Coates, J.S. Moon, D. Moses, M. Leclerc, K. Lee,
A.J. Heeger, Nat. Photonics 3, 297 (2009)
20. H.-Y. Chen, J. Hou, S. Zhang, Y. Liang, G. Yang, Y. Yang, L. Yu, Y. Wu, G. Li, Nat.
Photonics 3, 649 (2009)
21. Y. Liang, Z. Xu, J. Xia, S.-T. Tsai, Y. Wu, G. Li, C. Ray, L. Yu, Adv. Mater. 22, E135 (2010)
60 3 Physics and Chemistry of Conductive Polymers
22. Y. Liang, Y. Wu, D. Feng, S.-T. Tsai, H.-J. Son, G. Li, L. Yu, J. Am. Chem. Soc. 131, 56
(2009)
23. M. Knupfer, Appl. Phys. A 77, 623 (2003)
24. B.C. Thompson, J.M.J. Frechet, Angew. Chem. Int. Ed. 47, 58 (2008)
25. C. Deibel, V. Dyakonov, Rep. Prog. Phys. 73, 096401 (2010)
26. P.E. Shaw, A. Ruseckas, I.D.W. Samuel, Adv. Mater. 20, 3516 (2008)
Part II
Characterization of Colloidal
Nanocrystals and Thin Polymer Films
Chapter 4
Electron Microscopy
ð4:1Þ
Therein, h is Planck’s constant, c is the speed of light, m0,e is the rest mass of the
electron, p is the momentum of the electrons and Ekin their kinetic energy. As one
can see, the de Broglie wavelength is typically in the picometer size regime.
Therefore, in contrast to classical optical microscopy, atomic resolution is no
longer prevented by the wavelength of the incident beam.
When the electron beam hits the sample, a variety of physical processes are
possible, as illustrated by Fig. 4.1. Part of the electrons will be scattered back. If
the sample is thin enough, part of the electrons will be transmitted, either with or
without scattering. Scattering processes can either be elastic, i.e., without loss of
energy, or inelastic, i.e., with a loss of energy. Elastic scattering is likely when the
incident electrons are scattered by the heavy cores of the atoms in the sample. Due
to the large difference in mass, nearly no momentum is transferred from the
electrons to the atomic cores in that case. In the interaction of the incident elec-
trons with the electron shells of atoms in the sample, both elastic and inelastic
scattering processes occur. As an example for an inelastic process, the incident
electron beam can lead to the emission of secondary electrons by ionization of
atoms in the sample. After ionization, further processes can follow. For example,
X-rays can be emitted, if a vacancy in the electron shell is filled by an electron
from a higher shell. Furthermore, Auger electrons can be emitted.
In scanning electron microscopy, the backscattered electrons are detected, and
the surface of the sample is probed with the electron beam scanning with high
spatial resolution over the sample surface. In transmission electron microscopy,
the transmitted electrons are detected. This technique requires sufficiently thin
samples. The probability of transmission depends on the nature of the sample and
also on the energy of the incident beam, but typically, samples have to be not
thicker than several hundreds of nanometers.
In principle, a transmission electron microscope is largely analogue to a normal
optical microscope, but with electromagnetic lenses instead of glass lenses [1].
Figure 4.2a shows a simplified scheme of the components of a TEM. A condenser
lens directs the incident beam on the sample. The heart of the electron microscope
is an objective lens which creates an image of the transmitted wave at the exit face
of the object. Further lenses can follow to magnify the image, which is finally
recorded by a CCD camera. Optionally, a TEM can be equipped with additional
instrumentation, e.g., with a detector for X-rays or with an energy filter which can
be passed only by electrons with a specific kinetic energy. Energy dispersive X-ray
analysis (EDX) is usually implemented in electron microscopes and enables
4.1 Basics of Electron Microscopy 65
elemental analysis of the sample based on the emitted X-rays which have char-
acteristic wavelengths for each element.
Figure 4.2b shows the ray path for imaging by the objective lens. Parallel rays
passing the sample without scattering are focused by an ideal lens to a point on the
optical axis in the backfocal plane before the image is formed in the image plane.
Rays which are scattered into a specific direction are also focused to a point in the
backfocal plane, but this point is no longer located on the optical axis (see
Fig. 4.2).
If an aperture blend is used in the backfocal plane, this enables selecting
whether only scattered or unscattered electrons or both can reach the image plane.
If the aperture blend has a small diameter and is centered on the optical axis, only
unscattered electrons will reach the detector. This case is called bright field
imaging [1]. In this case, the detected intensity will be proportional to the prob-
ability of transmission without scattering. This relationship leads to contrast in the
observable TEM image. For amorphous samples, the probability of unscattered
transmission is usually directly related to the sample thickness. Thus, thick areas of
the sample will appear dark in the image, whereas thinner areas will appear bright.
In the case of crystalline samples, the situation is more complicated, however,
because diffraction of the electron beam by the lattice planes (Bragg scattering)
will lead to a probability of transmission without scattering that depends on the
orientation of the crystal. Therefore, the contrast obtained in bright-field images of
crystalline samples can no longer be regarded as a direct measure for the sample
thickness. Figure 4.3 (left panel) shows as an example bright-field TEM images of
CoPt3 nanocrystals. The different contrast of the individual nanocrystals can be
attributed to different orientations of the crystals. Statistical evaluation of TEM
images enables for example the determination of average particle sizes and the
66 4 Electron Microscopy
characterization of the size distribution (see right panel in Fig. 4.3). Therefore,
TEM is a very important method for the structural characterization of nanoparti-
cles in general.
It is also possible to move the aperture blend away from the optical axis, so that
only scattered electrons will be detected. This condition is called dark field
imaging [1]. Especially in the case of crystalline samples, detecting electrons
scattered to high angles can sometimes be advantageous. At high angles, inco-
herent scattering of the electrons (Rutherford scattering) dominates over Bragg
scattering [3]. By consequence, the scattering probability depends on the atomic
number. This dependence can be used for elemental mapping in scanning trans-
mission electron microscopy (STEM). Furthermore, at high angles the intensity
becomes nearly proportional to the thickness of the sample. Thus, the contrast in
the image does no longer depend on the orientation of crystalline samples. This is
crucial in some cases, for example for studies by electron tomography where a
linear dependence of the intensity on sample thickness is required (see Sect. 4.7).
In order to collect enough electrons at high angles, special detectors have been
developed which comprise annular rings around the optical axis. So, instead
of detecting only electrons passing through a small aperture placed at high angle,
4.1 Basics of Electron Microscopy 67
Fig. 4.3 (Left panel) Bright-field TEM image of colloidally prepared CoPt3 nanocrystals. (Right
panel) Histogram for the size distribution as obtained by computer-assisted evaluation of *2,800
nanoparticles (Reprinted with permission from [2]. Copyright 2005 American Chemical Society)
all electrons scattered to this angle with respect to the optical axis can be collected.
Such detectors are called ‘‘high angle annular dark-field’’ detectors (HAADF
detectors).
If the aperture in the backfocal plane of the objective lens is opened to a high
diameter, both scattered and unscattered electrons will reach the detector. In this
case, the interference of the scattered and unscattered wave functions leads to an
interference pattern in the image plane. These conditions are used to acquire high-
resolution TEM images [1]. If Bragg scattering is the dominant scattering process,
it is comprehensible that the created interference pattern is directly related to the
crystal structure. Typically, high-resolution TEM images of crystalline samples
enable the visualization of lattice fringes which can be considered as the images of
lattice planes. Figure 4.4 shows a typical example for the observation of lattice
fringes in HRTEM images of crystalline samples.
With very good transmission electron microscopes, it is even possible to
visualize projections of vertical columns of atoms. Figure 4.5 shows as example
HRTEM images of Sb-doped SnO2 nanocrystals where atomic resolution was
achieved [5]. To corroborate the interpretation of HRTEM images, simulation
methods can be applied (see example in Fig. 4.5).
As discussed before, the resolution is not limited by the wavelength in electron
microscopy. The factors finally determining the possible resolution of the micro-
scope are mainly the aberrations of the electromagnetic lenses and other issues like
beam stability, etc. High-end transmission electron microscopes possess complex
lens systems to minimize the aberrations. With aberration-corrected microscopes,
68 4 Electron Microscopy
it becomes possible to get even insight into the molecular structure of organic
compounds. Figure 4.6 shows as example HRTEM images of functionalized ful-
lerenes attached to carbon nanotubes as studied by Liu et al. [6]. Whereas HRTEM
4.2 High-Resolution Transmission Electron Microscopy 69
Referring back to the ray path illustrated in Fig. 4.2, the objective lens focuses the
scattered and unscattered incident rays into points in the backfocal plane. Thereby,
a diffraction pattern is obtained in the backfocal plane, before the image is finally
formed in the image plane. The complex amplitude of the wave transmitted by the
sample, the diffraction pattern in the backfocal plane and the image in the image
plane are related by Fourier transformation. Precisely, the intensity distribution
obtained in the backfocal plane of an ideal lens is proportional to the square of the
absolute value of the Fourier transform of the complex amplitude in the plane at
the exit face of the sample. (If the lens is non-ideal, the intensity distribution in the
backfocal plane will be modified by a function depending on the characteristics of
the lens.) Similarly, the intensity distribution in the image plane is related to the
70 4 Electron Microscopy
The relationship between the real space image and the diffraction pattern in
reciprocal space can also be used for image filtering. In that case, usually the
diffraction pattern is calculated by Fourier transformation from the real space
image. In a second step, the obtained image in reciprocal space is manipulated, and
finally the manipulated image is transformed back to real space by inverse Fourier
transformation.
Image manipulation in Fourier space can for example mean to identify Bragg
reflections and to remove diffuse intensity between the bright spots corresponding
to Bragg scattering. In the back-transformed real space image, features like lattice
fringes should then be visible more clearly. So, this type of image manipulation
would mean to reduce noise in the image.
It is also possible to apply filters in Fourier space which remove the Bragg
reflections belonging to specific crystallographic phases. This type of analysis was
for example used by Haubold et al. [8] to study InP/ZnS core–shell nanocrystals.
In that study, HRTEM images of individual nanocrystals were transformed into
Fourier space, and Bragg reflections of the InP and ZnS lattices became visible in
the diffraction pattern. With appropriate masks, the Bragg reflections belonging to
InP or ZnS were then selectively removed before back-transformation. The back-
transformation of a diffraction pattern with the ZnS reflections suppressed yields
an image where only the InP component is visible, and vice versa. This procedure
allowed identifying which parts of the original image corresponded to the InP and
ZnS phase, respectively [8].
Fig. 4.8 Illustration of different approaches to determine a radius for not perfectly spherical
particles. a The smallest circle is drawn around the 2D projection of the particle. b The largest
circle fitting into the particle is drawn. c The area A of the particle is measured and the radius is
defined as the radius of a sphere with equivalent cross-sectional area A0 = A
the volume of the crystalline domain and yield a volume-weighted average value
for the particle size. In contrast, non-weighted values (number distributions) are
obtained by TEM analysis, if the average size R and standard deviation r of
N particles are simply calculated by (4.2) and (4.3), respectively:
XN
¼1
R Ri ð4:2Þ
N i¼1
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
PN
2
i¼1 ðRi RÞ
and r¼ ð4:3Þ
N1
A number distribution function f ðRÞ can, fortunately, easily be converted into a
volume-weighted distribution function fV ðRÞ by (4.4) as follows [2]:
R3 f ð R Þ
f ð RÞ ! fV ðRÞ ¼ R 1
3
ð4:4Þ
0 R f ðRÞ dR
In the case of a non-weighted histogram, where h Rj denotes the frequency of
particles with radii in the jth class of the number distribution, the transformation is
given by (4.5) [2]:
R3j h Rj
h Rj ! hV R j ¼ N P 3 ð4:5Þ
j Rj h Rj
X-ray scattering are in good agreement for well-defined particles (with a rather
spherical shape and less than 10 % standard deviation), if care is taken of a correct
comparison by the usage of weighting factors [2].
A few words are appropriate in this book concerning sample preparation for
electron microscopy and also concerning stability of samples in the electron beam.
A requirement for TEM is that the sample is thin enough to be transmitted by part
of the electrons. The limit for the sample thickness depends on the material studied
and the energy of the electron beam, but typically, the sample should be not thicker
than about 100 nm.
In some cases sample preparation is quite easy. For example, colloidal nano-
crystals can usually be directly deposited from a colloidal solution with suitably
adjusted concentration on a TEM grid. Afterwards, the solvent is evaporated and a
thin layer of nanocrystals is left on the grid. In other cases, sample preparation can
be more difficult. In the field of polymer-based optoelectronics, a frequently
occurring task is to study by TEM polymer-based thin films which occur as a
component in a given optoelectronic device, e.g., pure polymer films or polymer
films mixed with another material such as fullerenes or inorganic nanocrystals.
One strategy might be to prepare the polymer-based film directly on the TEM grid,
in analogy to the film preparation in the device. However, it is usually not possible
to ensure similar film formation conditions on a TEM grid as compared to the film
which is incorporated in the real optoelectronic device. On the other hand, the
complete devices will usually not be suitable samples for investigation in a
transmission electron microscope. Therefore, it becomes necessary to extract the
polymer-based film from a real device. Sometimes this is possible, if the device
contains layers with materials of different solubility. For example, in organic and
hybrid solar cells a hydrophobic polymer-based film is often located on top of a
water-soluble PEDOT:PSS film. In that case, it is possible to immerse the device
into pure water. The PEDOT:PSS film will dissolve, so that the polymer-based
film is detached from the underlying substrate and floats on the water surface. The
film can then be picked up with a TEM grid. This method is successfully applied to
study active layers of organic and hybrid solar cells [9]. A similar approach is to
prepare polymer films under analogous conditions as in the real device on other
water-soluble substrates, such as NaCl windows for IR spectroscopy [10].
If such solution-based methods are not applicable, a powerful strategy is the
preparation of TEM lamellae with the help of a focused ion beam (FIB). In that
case, a precisely controllable ion beam is used to cut a thin lamella out of the
sample which can then be studied in a TEM. This method is technically compli-
cated and also rather time-consuming, but it enables studying samples which can
otherwise not be investigated by TEM and is therefore also applied in some studies
[11]. As an advantage of FIB preparation should be mentioned that this approach
74 4 Electron Microscopy
allows extracting the TEM sample from a real device without the need to bring it
in contact with a solvent.
Sample preparation for scanning electron microscopy is easier in the sense that
sample thickness is not crucial. A requirement for SEM is, however, that the
sample must be sufficiently conductive. Otherwise charging of the sample will
occur and severely limit the resolution.
Another general point demanding attention is the sample stability in the elec-
tron beam. In particular soft matter such as organic molecules, polymer or bio-
logical tissue can be destroyed or modified by prolonged exposure to an electron
beam in the microscope [12, 13]. This is even relevant for inorganic material such
as semiconductor nanocrystals, where electron beam-induced changes of the
structural properties were observed as well in some cases [1, 14]. Therefore, care
must be taken in electron microscopy, if the images are stable and really reflect the
sample in its initial state.
Fig. 4.9 SEM images (left side top view; right side side view) of carbon nanotubes grown by
chemical vapor deposition on indium tin oxide. The relatively short length of the tubes was
achieved by applying a very short growth time (Reprinted with permission from [15]. Copyright
2012 American Chemical Society)
Fig. 4.10 Illustration of the working principle of electron tomography. The sample (in this
example 4 spheres distributed in a cubic volume) is rotated around a tilt axis, so that two-
dimensional projections (normal TEM images) can be recorded from different directions. From
the series of projections, it is possible with computer-assisted methods to reconstruct the three-
dimensional object
76 4 Electron Microscopy
Fig. 4.11 Reconstructed volumes of three differently prepared P3HT/PCBM bulk heterojunction
films. The dimensions of the volumes are 1,700 nm 9 1,700 nm 9 100 nm. The film at the left
side was prepared by spin-coating without subsequent annealing. The film in the middle was
thermally annealed for 20 min at 130 C. The film at the right side was prepared by a so-called
‘solvent-assisted annealing’ method. The needle-like structures are attributed to crystalline P3HT
domains (Reprinted with permission from [19]. Copyright 2009 American Chemical Society)
usually satisfied in the bright-field imaging mode. However, in the case of crys-
talline samples, Bragg scattering leads to contrast variations that depend on the
orientation of the sample (see Sect. 4.1), so that the projection condition is in
general not fulfilled in the bright-field imaging mode. As pointed out earlier, at
high scattering angles Bragg scattering becomes less important. Therefore, crys-
talline samples are studied by electron tomography preferentially in the dark-field
imaging mode of an STEM with a HAADF detector [18].
Furthermore, reliable object reconstruction requires recording a sufficiently
large series of projections that cover an angular range which is as large as possible.
As explained in Sect. 4.3, a TEM image is related to a diffraction pattern in
Fourrier space. Every 2D-projection of the 3D-object corresponds to a 2D-slice in
the three-dimensional Fourrier space. If the complete Fourrier space is covered, the
full information for the reconstruction of the 3D-object from the projections is
available. However, the full angular range is usually not accessible, because at
high tilt angles a normal sample holder will block the beam. The angular range
which is not covered leads to a so-called ‘missing wedge’ of information [12, 18].
Typically, projections are recorded every 1–2 in a tilt range of ±70 [17].
However, with a tilt range of ±70, only *78 % of the volume in Fourrier space
is covered [12]. A way to reduce the missing wedge is to use two perpendicular tilt
axes [12, 18]. The effect of the missing wedge on reliable image reconstruction and
the improvement by using two tilt axes was for example demonstrated in a study of
CdTe tetrapods, where in some unfavorable cases not all the four branches of the
tetrapods appeared in the reconstruction [18]. This should make us sensible con-
cerning the reliability of the method. Electron tomography is certainly a very
powerful method to get an impression of the three-dimensional structure of an
object, but the reconstruction is usually not free of uncertainty.
With respect to polymer-based solar cells, electron tomography offers unique
possibilities to study the three-dimensional interpenetrating network of donor and
acceptor materials in bulk heterojunction layers. The method was already
4.7 Electron Tomography 77
References
Abstract One of the most classical techniques for structure analysis is X-ray
diffraction (XRD). Many variants of diffraction experiments have been developed.
They can for example be classified into techniques for the examination of X-ray
scattering in the wide angle or small angle range. Furthermore, different X-ray
sources can be used, with synchrotron radiation opening an own complex field of
research within the diffraction experiments. Rather than to give a summary on the
fundamentals of X-ray diffraction and the variety of different existing methods, the
aim of this chapter is to provide a review on what kind of information common
X-ray diffraction methods, initially developed for extended crystalline solids, can
reveal in the case of nanocrystalline materials and also soft matter such as poly-
mer. Apart from the identification of crystalline phases, X-ray diffraction is in
particular used to determine the size of nanoparticles, and special interest will be
devoted to that task here. The main part of this chapter will be devoted to X-ray
diffraction in the wide angle range. A brief overview on small angle X-ray scat-
tering (SAXS) will be given as well.
where d is the distance between neighbored lattice planes and h is the angle between
the incident radiation and the planes (see Fig. 5.1). The difference in length
generally causes a phase shift which leads to extinction, if there are a large number of
parallel planes. The interference of the diffracted rays is constructive only, if the
phase shift equals zero. This is the case, if the difference in length is an integer
multiple of the wavelength, i.e., if Dg = n k. In combination with the geometric
determination of Dg results the well-known Bragg equation (5.1) as condition for
constructive interference:
2d sin h ¼ n k; n2N ð5:1Þ
The difference between the wave vectors k and k0 of the diffracted and incident
wave (see Fig. 5.1) defines the so-called scattering vector q as by (2.2):
q ¼~
~ k ~
k0 ð5:2Þ
Since the spacing of lattice planes is characteristic for each substance, X-ray
diffraction can be used to identify crystalline phases. While powder X-ray dif-
fraction patterns of bulk material usually show up sharp reflections, diffractograms
of nanocrystalline substances present rather broad reflections. Figure 5.2 shows as
an example the diffractograms of PbSe nanocrystals deposited as thin films on
sapphire substrates in a study by Law et al. [5]. The samples were annealed to
different temperatures, and annealing caused growth of the nanoparticles [5]. One
can clearly see that the Bragg reflections become sharper with increasing particle
size.
The enhanced broadening in the case of nanocrystalline material can be
understood, when the diffraction process is considered in a microscopic picture
where the distribution of the diffracted intensity results from the superposition of
elementary waves emerging from the individual atoms of the crystal. The
understanding requires some basics of crystallography which shall briefly be
outlined here. A crystal structure is in general described by a lattice and a basis.
The lattice describes the periodicity of the structure in space. At every lattice point,
an identical arrangement of constituent atoms or ions is found, which is described
5.1 Basics of X-ray Diffraction 81
by the basis. One can also define a small unit cell containing the complete
information on the crystal structure. The crystal can then be considered to be
composed of a large number of unit cells patterning the volume of the crystal. One
can distinguish between primitive and non-primitive unit cells. A primitive unit
cell contains only one single lattice point and is therefore the smallest volume
containing already the complete structural information. Furthermore, the crystal
lattice can be described by three lattice translation vectors a, b and c. If the lattice
is a primitive one, i.e., if lattice points are situated only on the corners of the unit
cell, every vector pointing from the origin to a lattice point can be expressed by
(5.3):
~ a þ v ~
ruvw ¼ u ~ b þ w ~
c; ð5:3Þ
where u, v and w are integer. In analogue manner, a lattice in reciprocal space, the
so-called reciprocal lattice, is based on three vectors a*, b* and c* defined by the
following equations:
82 5 X-ray Diffraction
2p ~ ~ 2p 2p
a ¼
~ b ~
c; b ¼ ~
c ~
a; c ¼
~ a ~
~ b; ð5:4Þ
V V V
where V ¼ ~ a ð~
b ~ cÞ is the volume of the unit cell. Every vector pointing from
the origin to a lattice point of the reciprocal lattice can be written as follows:
~
rnh;nk;nl ¼ n ~
rhkl a þ n k ~
¼ n h ~ b þ n l ~
c ; ð5:5Þ
where n is integer and h, k and l are so-called Miller indices and also integer. The
vector ~
rnh;nk;nl of the reciprocal lattice is perpendicular to the lattice planes with the
Miller indices h, k and l, and its norm is inverse proportional to the distance dhkl
between neighbored planes, as described by (5.6):
2p
rnh;nk;nl ¼ n
~ ð5:6Þ
dhkl
Herein, n is the same whole number as in (5.5) and has the meaning of referring
to the nth (hkl) plane from the origin. If the lattice is a primitive one, the Miller
indices h, k and l have no common divisor. If the lattice is non-primitive, there
exist additional restrictions for h, k and l. For example, in a face centered cubic
(fcc) lattice the Miller indices have to be all even or all odd, and in a body centered
cubic (bcc) lattice the sum of the Miller indices has to be even.
From the definition of the reciprocal lattice immediately results that the scalar
product of two vectors ~ ruvw of the direct lattice and ~
rnh;nk;nl of the reciprocal lattice
is always an integer multiple of 2p:
ruvw ~
~ rnh;nk;nl ¼ m 2p; with m integer: ð5:7Þ
Coming back to the description of diffraction now, one can treat the problem by
considering elementary waves that start to propagate from all constituent atoms of
the irradiated crystal. The superposition of all the elementary waves at a given
time and point in space (far away from the crystal) leads to an expression for the
diffracted intensity I. This expression contains two factors [2], as can be seen from
(5.8) to (5.10):
qÞj2 jSð~
I / jF ð~ qÞj2 ; ð5:8Þ
with
X
F ð~
qÞ ¼ ei~q~ruvw ð5:9Þ
u;v;w
and
X
Sð~
qÞ ¼ aj ei~q~rj ð5:10Þ
all atoms
of the basis
5.1 Basics of X-ray Diffraction 83
sin2 ðN p xÞ
Lð xÞ ¼ ; ð5:11Þ
sin2 ðp xÞ
where x is the scalar product of the scattering vector q divided by 2p with one of
the three lattice translation vectors of the crystal lattice, and where N is the number
of unit cells existing in the crystal along the direction of the corresponding lattice
translation vector. If u represents one of the three lattice translation vectors a,
b and c of the crystal lattice, the quantity x can thus be written by (5.12):
q ~
~ u
x¼ ð5:12Þ
2p
Figure 5.3 shows a plot of the Laue function for N = 5 and N = 20. Maxima
are observed around integer values of x which corresponds to the situation where
(5.13) is fulfilled:
q ~
~ u ¼ n 2p; with n integer ð5:13Þ
As a condition for the observation of interference maxima, (5.13) must be
simultaneously satisfied for all three lattice translation vectors u = a, b, c. This is
identical to the condition that the scattering vector q must verify (5.14) for any
choice of ~ruvw with integer coefficients u, v and w:
q ~
~ ruvw ¼ n 2p; with n; u; v; w integer ð5:14Þ
This equation reminds the former statement that the scalar product of two
vectors ~ruvw and ~ rnh;nk;nl of the direct and the reciprocal lattice is always an integer
multiple of 2p. In case of a primitive lattice where all lattice points are reached by
~
ruvw vectors with integer coefficients, the above result means that interference
maxima can only occur, when the scattering vector q is a vector of the reciprocal
lattice, i.e., if (5.15) is fulfilled:
q ¼~
~ rnh;nk;nl ð5:15Þ
For simplicity, Fig. 5.4 shows the Ewald construction for a two-dimensional lattice.
Around the basis of the incident wave vector ~ k0 is drawn a circle of radius k~k0 k, and
one lattice point of the reciprocal lattice is placed on the end of ~ k0 . Diffraction then
only occurs into those directions where a reciprocal lattice point is situated on the
circumference of the Ewald circle, i.e., if the scattering vector ~ q ¼~ k ~k0 is a
reciprocal lattice vector. In the three-dimensional case, the construction is similar
with a sphere instead of a circle.
If the scattering vector is a reciprocal lattice vector, its norm is given by the
following expression according to (5.6):
2p
qk ¼ ~
k~ rnh;nk;nl ¼ n ð5:16Þ
dhkl
On the other hand, knowing that the angle between the incident and diffracted
wave vector is 2h (see Fig. 5.1), one easily calculates the norm of the scattering
vector as follows by (5.17):
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4p
qk ¼ ~
k~ k ~
k0 ¼ ~ k ~ k0 ~ k ~ k0 ¼ ¼ 2~ k sin h ¼ sin h
k
ð5:17Þ
5.1 Basics of X-ray Diffraction 85
Fig. 5.8 a The crystallite shape resulting from refinement corresponding to the red fitting curve
in Fig. 5.7. b Illustrations for the understanding of pole figures. (step 1) The pole P of the (hkl)
planes is obtained as the intersection of the prolonged vector normal to the planes with a sphere
used for projection. (step 2) The stereographic projection projects the pole P onto P0 in an
equatorial plane. (step 3) In the equatorial plane the pole figures are obtained. c Reconstructed
pole distribution figures (color coding: red high frequency, black low frequency) corresponding
to the red fitting curve in Fig. 5.7. While the (100) planes have a high probability to be oriented
parallel to the sample holder, the (002) planes are preferentially oriented perpendicular to the
sample holder [panels (a) and (c) Adapted with permission from [18]. Copyright 2010 American
Chemical Society]
anisotropic size model in combination with a model for texture effects resulted in a
reasonable fit of the experimental data (see red curve in Fig. 5.7) [18]. Figure 5.8a
shows the shape resulting from the simulation method. An average length of
56 nm and a thickness of 21 nm were found for the CuInS2 nanorods. Those values
are in reasonable agreement with the results obtained by TEM, the agreement
being better for the short axis of the nanorods than for the long axis [18].
Simulation of diffractograms of nanorods requires also the usage of texture
models, because it is not reasonable to assume that the rods are randomly orientated
on the substrate. The texture model used in the presented example is based on a
development of the so-called pole distribution function into a series of spherical
harmonics [20]. For the understanding of this model, the stereographic projection
and pole figures have to be explained: For a given crystal, the origin of the reci-
procal lattice is placed in the center of an imaginary sphere. As illustrated by
Fig. 5.8b, the prolongation of a reciprocal lattice vector ~ rhkl which is perpendicular
to the (hkl) planes intersects the sphere in the point P. P is called the pole of the (hkl)
planes. The pole can be projected into an equatorial plane. In the plane of projection
5.3 Rietveld Analysis 91
one obtains the pole figures. For a single crystal, distinct points corresponding to the
different lattice planes are obtained. For a powder, one can proceed slightly dif-
ferent. Instead of projecting the poles from all lattice planes, only one (hkl) plane is
considered and the corresponding poles from all crystallites are projected into one
equatorial plane which is chosen parallel to the substrate. In the projection plane,
the frequency distribution can then be visualized. Figure 5.8c shows the distribu-
tion of the projected poles of the (100) and (002) lattice planes for the CuInS2
nanorods. The color scale is expressed in multiples of random distribution (mrd).
Red color indicates an enhanced probability, and blue/black color indicates a lower
probability as compared to random orientation. For the (002) planes, the projections
of the poles have enhanced probability (red color in the figure) to be at the cir-
cumference of the projection plane. This means that the ~ r002 vectors are preferen-
tially parallel to the substrate. In other words, the nanorods have a preference to lie
flat on the substrate. This result appears reasonable, in particular, because the
sample was not prepared as a powder, but the nanocrystals were simply deposited
from colloidal solution on the sample holder in this example [18].
The presented XRD study of CuInS2 nanoparticles gives an example how Ri-
etveld analysis can successfully be used to investigate even rod-shaped nano-
crystals. It has to be mentioned, however, that the residuum is not as smooth as it
can usually be in the case of bulk studies. This is a certain limitation and shows
that Rietveld refinement of diffractograms from nanocrystalline material is
accompanied by some restrictions. Nevertheless, the analysis is suitable to obtain
information on the microstructure, and Rietveld refinement is a frequently used
technique in nanoscience [18, 19, 21–25].
X-ray diffraction can not only be used to study inorganic compounds such as
semiconductor nanocrystals, but also to analyze molecular order in soft matter.
Organic compounds can assemble into crystals with the molecules as constituent
‘‘atoms’’. Then, the periodical arrangement of molecules in space gives rise to the
diffraction of radiation.
5.5 X-ray Diffraction of Soft Matter 93
References
1. H.P. Klug, L.E. Alexander, X-ray Diffraction Procedures for Polycrystalline and Amorphous
Materials (Wiley, New York, 1974)
2. J.-P. Lauriat, Introduction à la cristallographie et à la diffraction Rayons X – Neutrons, Paris
Onze édition N K 150 (Université de Paris-Sud, Orsay, 1998). (in French)
3. C. Kittel, Introduction to Solid State Physics, 8th edn. (Wiley, New York, 2005)
4. Y. Waseda, E. Matsubara, K. Shinoda, X-ray Diffraction Crystallography (Springer,
Heidelberg, 2011)
5. M. Law, J.M. Luther, Q. Song, B.K. Hughes, C.L. Perkins, A.J. Nozik, J. Am. Chem. Soc.
130, 5974 (2008)
6. P. Scherrer, Nachr. Ges. Wiss. Göttingen 1918, 98 (1918). (in German)
7. A.A. Guzelian, U. Banin, A.V. Kadavanich, X. Peng, A.P. Alivisatos, Appl. Phys. Lett. 69,
1432 (1996)
8. J.I. Langford, A.J.C. Wilson, J. Appl. Cryst. 11, 102 (1978)
9. C.E. Krill, R. Birringer, Philos. Mag. A 77, 621 (1998)
10. H. Natter, M. Schmelzer, M.-S. Löffler, C.E. Krill, A. Fitch, R. Hempelmann, J. Phys. Chem.
B 104, 2467 (2000)
11. H. Borchert, E.V. Shevchenko, A. Robert, I. Mekis, A. Kornowski, G. Grübel, H. Weller,
Langmuir 21, 1931 (2005)
12. T.H. de Keijser, E.J. Mittemeijer, H.C.F. Rozendaal, J. Appl. Cryst. 16, 309 (1983)
13. N.C. Popa, J. Appl. Cryst. 31, 176 (1998)
14. H. Natter, R. Hempelmann, T. Krajewski, Ber. Bunsen-Ges. Phys. Chem. 100, 55 (1996)
15. B.E. Warren, X-ray Diffraction (Addison-Wesley, Reading, 1968)
16. B.E. Warren, B.L. Averbach, J. Appl. Phys. 21, 595 (1950)
17. B.E. Warren, B.L. Averbach, J. Appl. Phys. 23, 497 (1952)
18. M. Kruszynska, H. Borchert, J. Parisi, J. Kolny-Olesiak, J. Am. Chem. Soc. 132, 15976
(2010)
19. L. Lutterotti, D. Chateigner, S. Ferrari, J. Ricote, Thin Solid Films 450, 34 (2004)
20. N.C. Popa, J. Appl. Cryst. 25, 611 (1992)
21. J.W. Stouwdam, M. Raudsepp, F.C.J.M. van Veggel, Langmuir 21, 7003 (2005)
22. V. Petkov, M. Gateshki, J. Choi, E.G. Gillan, Y. Ren, J. Mater. Chem. 15, 4654 (2005)
23. H. Schäfer, P. Ptacek, H. Eickmeier, M. Haase, Synthesis and characterization of
upconversion fluorescent Yb3+, Er3+ doped CsY2F7 nano- and microcrystals.
J. Nanomaterials (2009). doi:10.1155/2009/685624
24. S. Wilken, D. Scheunemann, V. Wilkens, J. Parisi, H. Borchert, Org. Electron. 13, 2386
(2012)
25. L. Kumar, P. Kumar, A. Narayan, M. Kar, Int. Nano Lett. 3, 8 (2013)
26. J. Wagner, W. Härtel, R. Hempelmann, Langmuir 16, 4080 (2000)
27. O. Glatter, O. Kratky, Small Angle X-ray Scattering (Academic Press, New York, 1982)
28. L. A. Feigin, D.I. Svergun, in Structure Analysis by Small Angle X-ray Scattering and
Neutron Scattering, ed. by G.W. Taylor (Plenum Press, New York, 1987)
29. C.B. Murray, C.R. Kagan, M.G. Bawendi, Science 270, 1335 (1995)
30. C.B. Murray, S. Sun, W. Gaschler, H. Doyle, T.A. Betley, C.R. Kagan, IBM J. Res. Dev. 45,
47 (2001)
31. D.E. Motaung, G.F. Malgas, C.J. Arendse, S.E. Mavundla, C.J. Oliphant, D. Knoesen, The
influence of thermal annealing on the morphology and structural properties of a conjugated
polymer in blends with an organic acceptor material. J. Mater. Sci. 44, 3192 (2009)
Chapter 6
Photoelectron Spectroscopy
side of a photoemission peak. These fundamentals are illustrated by Fig. 6.1 which
shows a typical survey spectrum of a gold surface.
Since the binding energies of core-levels are characteristic for each element,
each photoemission peak can be attributed to some element present in the sample,
and XPS can be used for elemental analysis. However, core-level binding energies
are not only characteristic for each element, but also depend on the chemical
environment of the atoms. The last dependence leads to the chemical shift which
can reach several eV [1]. Provided the experimental resolution is sufficient, core-
level photoemission spectra may be deconvoluted into components corresponding
to atoms in different chemical environments. This is the basic concept of high-
resolution photoelectron spectroscopy.
Figure 6.2 shows a detail spectrum of the Au 4f level recorded from a gold
surface and illustrates some further general characteristics of core-level spectra.
Because of the spin–orbit splitting all p, d and f levels are composed of two
sublevels with the quantum numbers jþ ¼ l þ 1=2 and j ¼ l 1=2. The two cor-
responding peaks can be separated by several eV and their relative intensity is
determined by the number of electrons in the two sublevels. The branching ratio is
defined as the intensity ratio of the peak with the quantum number j to that one
with the quantum number jþ and should be 0.5, 0.67 and 0.75 for p, d and f levels
respectively. Since p, d and f core-level spectra always comprise two peaks with a
well-defined spin–orbit splitting and branching ratio, those two peaks are fre-
quently referred to as ‘‘one component’’ in the sense that the doublet corresponds
to one chemical environment for the atoms in question.
As mentioned above, spectra may be composed of several components corre-
sponding to atoms of the same element, but in different chemical environments.
Figure 6.3 shows as an example an In 4d spectrum of an InP(110) surface. The
spectrum comprises two components corresponding to In atoms in the interior and
at the surface of the sample, respectively. Each of the components consists of two
6.1 Fundamentals of X-ray Photoelectron Spectroscopy 97
peaks due to the spin–orbit splitting of the In 4d level. The surface core-level shift
is about 0.3 eV in the given example.
Apart from the binding energy, core-level spectra are characterized by their
width. Several factors lead to broadening. Usually, the observed line broadening
can be interpreted as follows: First, there is a natural line width due to the life-time
of the electrons in the core-levels. The corresponding broadening is often
described by a Lorentzian line profile [2, 3]. Furthermore, every setup has a limited
experimental resolution. The corresponding broadening is often considered as
Gaussian. Finally, inhomogeneities of the sample may also lead to broadening of
the spectra. The inhomogeneous broadening due to sample properties can usually
be described by a Gaussian profile. Thus, the line profile of a photoemission peak
results from the convolution of the Lorentzian and the two Gaussian contributions.
A so-called Voigt profile is obtained which is characterized by its Lorentzian and
its Gaussian width. While the Lorentzian width is directly related to the natural
life-time, the Gaussian width results from the convolution of the two distributions
taking care of the experimental setup and sample inhomogeneities, respectively.
98 6 Photoelectron Spectroscopy
Thus, the Gaussian width of a photoemission peak is the geometric mean of the
widths related to those two contributions and can be described by (6.1):
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
DEGaussian;total ¼ DEGaussian;setup þ DEGaussian;sample ð6:1Þ
As indicated in the above sections, the main idea of high resolution photoelectron
spectroscopy is to analyze photoemission spectra by fitting them to the minimum
number of Voigt functions required to reproduce the original spectra. The
deconvolution of the spectra into the contributing components allows revealing
information about atoms in different chemical environments. This makes high
resolution photoelectron spectroscopy a powerful method to probe the surface
structure of semiconductor nanocrystals as well as internal interfaces in core–shell
nanocrystals [17].
With a conventional Al Ka or Mg Ka X-ray gun the experimental resolution is
typically of the order of 0.5 eV. This may be insufficient to study small chemical
shifts. The use of synchrotron radiation usually allows achieving better resolution.
Depending on the excitation energy and the spectrometer, experimental resolutions
of typically 0.2–0.3 eV may be achieved. Note as another advantage of using
synchrotron radiation in particular the possibility to adjust the excitation energy in
a manner to obtain maximum surface sensitivity.
In this section a brief overview will be given what kind of high resolution
photoemission work has recently been performed on semiconductor nanocrystals
and what kind of information can be revealed. A large variety of materials was
studied by XPS, among them II-VI and III-V semiconductors and transition metal
oxides. For example, CdS nanocrystals have been studied by Winkler et al. [18, 19]
6.3 High-Resolution Photoelectron Spectroscopy of Semiconductor Nanocrystals 101
and Nanda et al. [20]. CdSe [21] and CdSe/ZnS core–shell nanocrystals [21] as
well as highly luminescent CdTe nanocrystals [22] have been investigated by
Möller and co-workers. High-resolution XPS was also applied to study ZnS [23]
and PbS nanocrystals [24] and quantum dot quantum well structures of CdS and
HgS [25]. Concerning III-V compounds, InAs [26] and InAs/CdSe core–shell [27]
as well as InP nanocrystals etched with hydrogen fluoride [28, 29] were studied.
Examples for studies of oxides are investigations of Sb-doped SnO2 nanocrystals
[30, 31].
A very basic question in colloidal synthesis of nanocrystals is the stabilization
of the nanoparticles in solution and the passivation of the surface by organic
ligands. The success of a synthesis usually depends on an appropriate choice of the
kind and amount of ligand molecules. Thus, it is an important task to investigate
the bonding of organic ligands to the surface of the nanocrystals. Such kind of
information may be provided by high resolution photoemission spectra.
One of the earliest synchrotron XPS studies of the bonding between organic
ligands and the surface of semiconductor nanocrystals was carried out by Winkler
et al. [18, 19] in 1999. The authors studied CdS nanocrystals capped with mer-
captopropionic acid and analyzed high-resolution XP spectra of the S 2p and Cd
3d5/2 core-levels at different excitation energies. S 2p spectra were deconvoluted
into three components that could be assigned to S atoms inside the volume of the
nanocrystals, S atoms at the surface of the nanocrystals, and to the formation of
Cd–S–CH2–R bonds between surface Cd atoms and the organic ligands, respec-
tively [18, 19]. The establishment of Cd–S–CH2–R bonds was additionally con-
firmed by the observation of a corresponding component in the Cd 3d spectra
[18, 19]. This example shows how bonds between organic ligands and the surface
of the nanocrystals may be detected by high resolution photoelectron spectroscopy
with the use of tuneable synchrotron radiation.
As another example, a study of colloidally prepared CdTe nanocrystals capped
with thioglycolic acid shall be presented here [22]. That study compared two
samples of CdTe quantum dots of the same particle size which were prepared
under different growth conditions. As described in Chap. 2, semiconductor
nanocrystals are formed in a dynamic growth process where at any stage of the
colloidal synthesis on the one hand new material is attached to the surface of
existing nanocrystals and on the other hand some atoms leave the surface and are
released back into the solution. From the rates of growth and dissolution results an
effective net growth rate. A careful study of the complex growth dynamics of
semiconductor nanocrystals has shown that nanocrystals formed at low net growth
rate possess the highest fluorescence quantum yields and the best photostability
[32]. This behavior was assumed to be due to the formation of different surface
structures under different growth conditions. High resolution photoelectron spec-
troscopy allowed to give a proof for those differences in the surface structure and
even to relate the observable differences to the optical properties of the nano-
crystals prepared under different growth conditions [22].
Figure 6.6 (left side) shows Te 4d spectra of a sample with poor photolumi-
nescence properties. Equilibrium between growth and dissolution was not realized
102 6 Photoelectron Spectroscopy
Fig. 6.6 Te 4d spectra of poorly (left side) and highly (right side) luminescent CdTe
nanocrystals capped with thioglycolic acid. For the poorly luminescent sample spectra at low
excitation energy (highest surface sensitivity, spectrum in the upper left corner) and higher
excitation energy (lower surface sensitivity, spectrum in the bottom left corner) are shown. Two
surface components are observed and can be assigned to a Te terminated (111) surface
(component S1) and oxidized surface Te atoms (component S2), respectively. In the case of the
highly luminescent CdTe nanocrystals (spectrum in the upper right corner, at highest surface
sensitivity), the surface component S1 is reduced with respect to the poorly luminescent sample,
and the component S2 is below the detection limit. The photographs in the bottom right corner
show colloidal solutions of both samples under illumination with UV light (Adapted with
permission from [22]. Copyright 2003 American Chemical Society)
for that sample. The spectra obviously consist of three spin-orbit split doublets.
Comparison of the relative intensities at surface sensitive (upper panel) and more
volume sensitive energy (lower panel) allows to assign the components labeled
‘S1’ and ‘S2’ to surface environments for Te whereas the component ‘V’ corre-
sponds to Te atoms in the interior of the quantum dots. Comparison of the
chemical shifts with literature values allows to assign the component ‘S1’ to Te
atoms on a Te terminated (111) surface, whereas the component ‘S2’ corresponds
to oxidized Te atoms at the surface [22]. On the right side of Fig. 6.6 one can see a
surface sensitive Te 4d spectrum recorded from a sample of highly luminescent
CdTe nanocrystals which have been prepared at a low net growth rate, i.e., where
growth and dissolution were in equilibrium. Compared to the spectra of the lowly
luminescent sample, the highly luminescent CdTe nanocrystals present much less
Te atoms at the surface. The component ‘S1’ is considerably reduced and the
component ‘S2’ is below the detection limit. The reduced number of Te atoms at
the surface could be explained by the formation of a Cd–S–R surface layer where
some of the surface Te sites are not occupied by Te atoms but by the organic
stabilizer molecules [22]. Such a structure can explain the improved photolumi-
nescence quantum yield, because surface oxidation and the number of dangling
6.3 High-Resolution Photoelectron Spectroscopy of Semiconductor Nanocrystals 103
bonds at the surface are reduced and because a CdS like surface layer gives rise to
a core–shell like structure with a potential wall at the surface. Furthermore, there
are reasons to assume that such a structure has a high thermodynamic stability. The
last point provides at least one possible explanation why the highest fluorescence
quantum yields are obtained only at low net growth rates [22]. This example shows
how powerful high resolution photoelectron spectroscopy can be used to corrob-
orate even studies of complex processes like the growth dynamics of semicon-
ductor nanocrystals.
Another example is a study of InAs nanocrystals (of 4.3 nm in diameter)
capped with trioctylphosphine (TOP) [26]. Fitting of As 3d spectra required three
doublets of Voigt functions corresponding to three distinct chemical environments
for the As atoms. One of them could be assigned to bonds of the TOP ligands to
surface As atoms [26]. Note the interesting information provided by that study. In
principle, one would expect that the TOP molecules providing electron density as a
Lewis base should preferentially bind to In atoms at the surface of the InAs
nanocrystals. However, high resolution photoelectron spectroscopy provided evi-
dence that the TOP ligands bind also to surface As atoms. In a follow-up study,
InAs nanocrystals coated with a CdSe shell were investigated [27]. As to be
expected the surface components observed for the uncoated InAs nanocrystals
were absent in the spectra of the core–shell nanocrystals. Instead, a new compo-
nent was observed which could be assigned to As–Se bonds at the interface [27].
Thus chemical bonds may be studied not only at the surface of semiconductor
nanocrystals, but also at the interface in core–shell nanocrystals.
Compared to high resolution studies of bulk material, spectra of nanocrystalline
structures present a considerably larger broadening. This was for example dem-
onstrated by Hamad et al. [33] who have investigated In 3d spectra of a series of
InAs nanocrystals of different particle size. One source of inhomogeneous
broadening is slight variations of bond lengths and angles which are supposed to
be larger for small nanocrystals and to occur especially near the surface of the
nanoparticles. Different factors leading to inhomogeneous broadening have also
been discussed in a study of CdSe and CdSe/ZnS core–shell nanocrystals [21].
Facetting and bonding of organic ligands to the surface have to be mentioned here.
Surface core-level shifts depend in general on the lattice planes [34]. Since
semiconductor nanocrystals may possess a variety of different surface planes,
small energetic shifts between the associated photoemission signals may remain
unresolved, so that inhomogeneous broadening is observed instead. In a similar
manner slightly different possibilities for bonds between organic ligands and the
nanocrystal surface may lead to small energetic shifts which cannot be resolved.
Furthermore, it has to be noted that in principal charging can also be a source of
broadening and some tests should be performed to check for charging effects.
Systematic tests may be performed by comparison of spectra recorded at the
beginning and at the end of a photoemission experiment and by application of
different illumination conditions [19, 23].
104 6 Photoelectron Spectroscopy
Apart from high resolution studies, photoelectron spectroscopy may also be used
for quantitative analysis of the chemical composition. In many cases, quantifica-
tion is done by simply dividing measured peak intensities by relative sensitivity
factors which are tabulated for all core-levels at the most common excitation
energies (Mg Ka and Al Ka radiation) [1]. However, the tabulated sensitivity
factors cannot always be applied, because the peak intensity depends on many
parameters and experimental conditions. In fact, the intensity of a given photo-
emission peak is in general determined by the following expression [23, 35–37]:
Z
I / Ibeam rðhmÞ AðhmÞ SDet ðEkin Þ r Þ edð~rÞ=kðEkin Þ
dV qð~ ð6:3Þ
sample
Therein,
Ibeam is the intensity of the incident radiation,
r is the photoionization cross-section for the element and core-level in
question,
A is the asymmetry term for the element and core-level in question,
SDet is the energy-dependent sensitivity of the detector,
qð~rÞ is the number of atoms per volume for the element in question at the
position ~
r,
dð~rÞ is the distance which photoelectrons generated at the position ~ r have to
pass through the sample, and
k is the mean free path length of the photoelectrons.
The factors in front of the integral expression in (6.3) may in principle be
determined. The intensity of the incident radiation has to be measured. Values for
the photoionization cross-section have been calculated theoretically for core-levels
in free atoms and may serve as an approximation for quantitative analysis in
photoemission [38, 39]. The asymmetry term takes into account the anisotropy of
the photoemission process and has to be evaluated for the geometry of the
experimental setup using literature resources [38–40]. The energy-dependent
sensitivity function of the detector has also to be evaluated for the used spec-
trometer. By division of measured intensities by the mentioned factors one can
obtain normalized intensities which are then directly determined by the sample
structure:
Z
Inorm / rÞ edð~rÞ=kðEkin Þ
dV qð~ ð6:4Þ
sample
This normalized peak intensity depends on the spatial distribution of the atoms
from which photoelectrons can be emitted and on the attenuation of the signal due
to scattering. Since the mean free path length depends on the kinetic energy of the
6.4 Quantitative Photoelectron Spectroscopy 105
photoelectrons, the exponential attenuation factor changes with energy and the
normalized peak intensities become a function of the excitation energy. The
evaluation of peak intensities as a function of energy is the main idea for revealing
information about spatial variations of the chemical composition.
For flat samples, spatially constant concentrations of the elements in the sample
and geometrical conditions with the axis of the direction of detection perpendic-
ular to the sample surface, the integral in (6.4) can easily be solved. Under these
conditions, the normalized intensity becomes simply proportional to the mean free
path length and the concentration of the element in question:
Inorm / q kðEkin Þ ð6:5Þ
By consequence, the intensity can then be written as:
I / s q; ð6:6Þ
where s is the relative sensitivity factor containing the mean free path length and
the parameters discussed before. From these considerations it becomes clear that
quantification of the elemental composition by simply dividing the measured peak
intensity by tabulated sensitivity factors is based on several important
assumptions:
1. The sample surface is flat.
2. The elements are homogeneously distributed in the sample. In other words,
spatial variations of the composition are neglected in this approach.
3. The sensitivity factors taken from tables are valid for the experimental con-
ditions (excitation energy, energy-dependence of the detector, angle between
the X-ray source and axis of detection, etc.).
If these conditions are not fulfilled, or if one intends to study spatial variations
of the elemental composition, quantitative analysis should be based on the eval-
uation of measured peak intensities with the help of (6.3) and (6.4).
Quantitative photoelectron spectroscopy, especially with the use of tuneable
synchrotron radiation, can for example be used to characterize core–shell nano-
crystals [17]. For example, InP/ZnS [37], CdSe/ZnS [21], InAs/CdSe [27] and
CePO4:Tb/LaPO4 core–shell nanocrystals [41] were studied by this method. In all
cases could successfully be verified that the core is surrounded by a shell of the
second material and computer simulation allowed to determine the average shell
thickness.
The method for shell thickness determination shall be briefly outlined here. The
basic idea is to record photoemission peaks of elements associated with the core
and the shell of the composite nanocrystals at a series of excitation energies.
Normalized peak intensities corresponding to the core element are stronger
attenuated than those corresponding to the shell element, simply because photo-
electrons emitted in the core of core–shell nanocrystals have to pass additionally
through the shell. To obtain quantitative information, a model needs to be assumed
for the sample structure. The simplest approach is a spherical model with only
106 6 Photoelectron Spectroscopy
Fig. 6.7 a Illustration of the spherical model used for simulation; b Intensity ratio of Ce 3d and
La 3d peaks recorded from CePO4:Tb/LaPO4 core–hell nanocrystals as a function of excitation
energy. Experimental data points are plotted together with a fitting curve (solid line). Also shown
are simulations for alloy formation (dashed line) and the ideal structure expected from the
amount of precursor used in the synthesis (dotted line). The results confirmed the formation of a
core–shell structure and enabled extracting an average value for the shell thickness (part
b Reproduced with permission from [41], Copyright (2003) Wiley–VCH Verlag GmbH & Co.
KGaA. The dotted curve was added to the original graphics)
three parameters for the radius of the core (rc), the thickness of the shell (ds) and
the thickness of the ligand shell (dlig), as depicted in Fig. 6.7a. For a given set of
thickness parameters, theoretical values for the normalized intensity at different
energies can then be numerically calculated. A note is necessary here concerning
the mean free path length. Within the approximation of inelastic scattering the
TPP-2M formula [11] may serve to determine the mean free path length in order to
enable calculation of the integral expressions for the normalized peak intensities.
To take care of slightly different mean free path lengths in the different regions of
the composite nanocrystals, the exponential attenuation factor in (6.4) should be
replaced by a product of exponential functions describing the attenuation in the
core, in the shell and in the ligand shell, respectively:
ed=kðEkin Þ ! esc =kc ðEkin Þ ess =ks ðEkin Þ eslig =klig ðEkin Þ ð6:7Þ
Therein, sc, ss and slig are the parts of the distance d which the photoelectrons
pass through the core, the shell and the outer ligand shell, respectively. kc, ks and
klig are the corresponding mean free path lengths. By comparison of the calculated
with the experimentally obtained normalized intensities, one can extract in a
computer simulation (least squares fit) the thickness parameters of the model
which lead to the best agreement. This procedure enables extracting the average
thickness of the overgrown shell from a series of XP spectra recorded at different
excitation energies [17, 42].
As a concrete example, Fig. 6.7b shows the measured and simulated energy
dependence of normalized peak intensities for CePO4:Tb/LaPO4 core–shell
nanocrystals [41]. Comparison of the data with a simulated curve for the case of
6.4 Quantitative Photoelectron Spectroscopy 107
homogeneous alloy formation, also plotted in Fig. 6.7, clearly confirmed the
formation of a core–shell structure and allowed determining the average shell
thickness [41].
Thus, comparison of experimental and simulated peak intensities at different
excitation energies is a suitable method to verify the core–shell structure of the
composite nanocrystals and allows determining the thickness of the shell grown on
the core material. However, it has to be noted that such a fitting procedure can not
result in values of high precision. Usually already the normalization of measured
peak intensities is accompanied by uncertainties which are only difficult to esti-
mate. The simple assumption of spherical geometry and the approximation of
inelastic scattering, i.e., the usage of an exponential attenuation factor and inelastic
mean free path lengths, are rather strong restrictions from the theoretical side.
Furthermore, structural inhomogeneities like the finite size distribution of the
sample, local variations of the shell thickness, etc. cannot be taken into account by
these calculations. Therefore, results from the described fitting procedure are best
understood as rough average values.
Although the use of tuneable synchrotron radiation is of great advantage,
characterization of core–shell nanocrystals can in principle also be performed with
a conventional X-ray gun. For example, Cao and Banin have studied InAs/CdSe
and InAs/InP core–shell nanocrystals [36]. Systematic series of core–shell nano-
crystals with varying thickness of the overgrown shell were investigated. In quite
similar numerical calculations the authors used the fact that the intensity ratio of
two different photoemission peaks associated with the core depends on the
thickness of the overgrown shell. Experimental intensity ratios as a function of the
shell thickness assumed from the synthesis were compared to calculated intensity
ratios. The good agreement confirmed the core–shell structure of the nanocrystals
[36].
Nanda et al. have studied CdS [20] and ZnS [23] nanocrystals capped with
1-thioglycerol. Quite similar to the results by Winkler et al. [18, 19] discussed in
Sect. 6.3, S 2p spectra showed three components which could be assigned to S in
the volume of the nanocrystals, S atoms at the surface and S atoms of the
1-thioglycerol ligands. Intensities of the different components were measured at
the excitation energies of Al and Mg Ka radiation. Quantitative calculations, rather
similar to the described method for analyzing core–shell nanocrystals, were per-
formed to determine the radius of the core region, the thickness of the surface layer
and the thickness of the ligand shell [20, 23]. Thus, although there are restrictions
for the achievable surface sensitivity and the number of data points, quantitative
analysis of semiconductor nanocrystals can also be performed with photoelectron
spectrometers using conventional X-ray sources.
As a last example for quantitative photoemission work, investigations of
Sb-doped SnO2 nanocrystals shall be mentioned here [31]. Measurement and sim-
ulation of normalized intensity ratios between Sb 3d3/2 and Sn 3d3/2 peaks enabled
determining the radial distribution of the dopant atoms in the host lattice [31].
108 6 Photoelectron Spectroscopy
References
1. C.D. Wagner, W.M. Riggs, L.E. Davis, J.F. Moulder, G.E. Muilenberg (eds.), Handbook of
X-ray Photoelectron Spectroscopy (Perkin-Elmer Corporation, Eden Prairie, 1979)
2. N. Mårtensson, R. Nyholm, Phys. Rev. B 24, 7121 (1981)
3. G.K. Wertheim, S.B. Dicenzo, J. Electron Spectrosc. Relat. Phenom. 37, 57 (1985)
4. J.N. Andersen, U.O. Karlsson, Phys. Rev. B 41, 3844 (1990)
5. W.G. Wilke, V. Hinkel, W. Theis, K. Horn, Phys. Rev. B 40, 9824 (1989)
6. H.P. Wampler, D.Y. Zemlyanov, K. Lee, D.B. Janes, A. Ivanisevic, Langmuir 24, 3164
(2008)
7. M.P. Seah, W.A. Dench, Surf. Interface Anal. 1, 2 (1979)
8. J. Szajman, J. Liesegang, J.G. Jenkin, R.C.G. Leckey, J. Electron Spectrosc. Relat. Phenom.
23, 97 (1981)
9. D.R. Penn, Phys. Rev. B 35, 482 (1987)
10. S. Tanuma, C.J. Powell, D.R. Penn, Surf. Interface Anal. 20, 77 (1993)
11. S. Tanuma, C.J. Powell, D.R. Penn, Surf. Interface Anal. 21, 165 (1994)
12. C.J. Powell, A. Jablonski, S. Tanuma, D.R. Penn, J. Electron Spectrosc. Relat. Phenom. 68,
605 (1994)
13. C.J. Powell, A. Jablonski, I.S. Tilinin, S. Tanuma, D.R. Penn, J. Electron Spectrosc. Relat.
Phenom. 98–99, 1 (1999)
14. P.J. Cumpson, Surf. Interface Anal. 31, 23 (2001)
15. D.E. Eastman, T.-C. Chiang, P. Heimann, F.J. Himpsel, Phys. Rev. Lett. 45, 656 (1980)
16. S. Wiklund, K.O. Magnusson, S.A. Flodström, Surf. Sci. 238, 187 (1990)
17. D.D. Sarma, P.K. Santra, S. Mukherjee, A. Nag, Chem. Mater. 25, 1222 (2013)
18. U. Winkler, D. Eich, Z.H. Chen, R. Fink, S.K. Kulkarni, E. Umbach, Phys. Status Solidi A
173, 253 (1999)
19. U. Winkler, D. Eich, Z.H. Chen, R. Fink, S.K. Kulkarni, E. Umbach, Chem. Phys. Lett. 306,
95 (1999)
20. J. Nanda, B.A. Kuruvilla, D.D. Sarma, Phys. Rev. B 59, 7473 (1999)
21. H. Borchert, D.V. Talapin, C. McGinley, S. Adam, A. Lobo, A.R.B. de Castro, T. Möller, H.
Weller, J. Chem. Phys. 119, 1800 (2003)
22. H. Borchert, D.V. Talapin, N. Gaponik, C. McGinley, S. Adam, A. Lobo, T. Möller, H.
Weller, J. Phys. Chem. B 107, 9662 (2003)
23. J. Nanda, D.D. Sarma, J. Appl. Phys. 90, 2504 (2001)
24. A. Lobo, T. Möller, M. Nagel, H. Borchert, S.G. Hickey, H. Weller, J. Phys. Chem. B 109,
17422 (2005)
25. H. Borchert, D. Dorfs, C. McGinley, S. Adam, T. Möller, H. Weller, A. Eychmüller, J. Phys.
Chem. B 107, 7486 (2003)
26. C. McGinley, M. Riedler, T. Möller, H. Borchert, S. Haubold, M. Haase, H. Weller, Phys.
Rev. B 65, 245308 (2002)
27. C. McGinley, H. Borchert, D.V. Talapin, S. Adam, A. Lobo, A.R.B. de Castro, M. Haase, H.
Weller, T. Möller, Phys. Rev. B 69, 045301 (2004)
28. S. Adam, C. McGinley, T. Möller, D.V. Talapin, H. Borchert, M. Haase, H. Weller, Eur.
Phys. J. D 24, 373 (2003)
29. S. Adam, D.V. Talapin, H. Borchert, A. Lobo, C. McGinley, A.R.B. de Castro, M. Haase, H.
Weller, T. Möller, J. Chem. Phys. 123, 084706 (2005)
30. C. McGinley, S. Al Moussalami, M. Riedler, M. Pflughoefft, H. Borchert, M. Haase, A.R.B.
de Castro, H. Weller, T. Möller, Eur. Phys. J. D 16, 225 (2001)
31. C. McGinley, H. Borchert, M. Pflughoefft, S. Al Moussalami, A.R.B. de Castro, M. Haase, H.
Weller, T. Möller, Phys. Rev. B 64, 245312 (2001)
32. D.V. Talapin, A.L. Rogach, E.V. Shevchenko, A. Kornowski, M. Haase, H. Weller, J. Am.
Chem. Soc. 124, 5782 (2002)
References 109
33. K.S. Hamad, R. Roth, J. Rockenberger, T. van Buuren, A.P. Alivisatos, Phys. Rev. Lett. 83,
3474 (1999)
34. N. Mårtensson, A. Nilsson, in Applications of Synchrotron Radiation, ed. by W. Eberhardt
(Springer Series in Surface Sciences 35, Springer, Berlin, 1995)
35. J.E.B. Katari, V.L. Colvin, A.P. Alivisatos, J. Phys. Chem. 98, 4109 (1994)
36. Y. Cao, U. Banin, J. Am. Chem. Soc. 122, 9692 (2000)
37. H. Borchert, S. Haubold, M. Haase, H. Weller, C. McGinley, M. Riedler, T. Möller, Nano
Lett. 2, 151 (2002)
38. J.-J. Yeh, Atomic Calculation of Photoionization Cross-Sections and Asymmetry Parameters
(Gordon & Breach Science Publishers, Langhorne, 1993)
39. I.M. Band, Y.I. Kharitonov, M.B. Trzhaskovskaya, At. Data Nuc. Data Tab. 23, 443 (1979)
40. V.I. Nefedov, I.S. Nefedova, J. Electron Spectrosc. Relat. Phenom. 107, 131 (2000)
41. K. Kömpe, H. Borchert, J. Storz, A. Lobo, S. Adam, T. Möller, M. Haase, Green-emitting
CePO4:Tb/LaPO4 core-shell nanoparticles with 70 % photoluminescence quantum yield.
Angew. Chem. Int. Ed. 42, 5513–5516 (2003)
42. H. Borchert, Untersuchungen von Halbleiter-Nanokristallen mit Hilfe von Photoelektro-
nenspektroskopie (Ph.D thesis, University of Hamburg, Hamburg, 2003), available online at
http://ediss.sub.uni-hamburg.de/volltexte/2003/1050/ (in German)
Chapter 7
Cyclic Voltammetry
The basic idea of cyclic voltammetry is to apply a periodic potential that will
alternately oxidize and reduce the material to be studied by extracting and
injecting electrons, respectively. Practically, this is typically realized in an elec-
trochemical cell with three electrodes: a working electrode, a reference electrode
and a counter electrode (or auxiliary electrode) [1]. Figure 7.1 illustrates a typical
setup for cyclic voltammetry. A periodic potential (usually a triangular voltage) is
applied between the working and the reference electrode. The substance to be
analyzed is then oxidized and reduced at the working electrode. If there were only
two electrodes, the working and the reference electrode, the redox reaction at the
working electrode would cause a current flow between both electrodes. This in
turn would change the potential of the working electrode with respect to the
Fig. 7.1 Schematic illustration of an electrochemical cell for cyclic voltammetry in three-
electrode configuration. A periodic potential E is applied with the help of a potentiostat between
the working electrode and the reference electrode. To maintain the potential at its desired value, a
current I flows over the counter electrode. Since that current compensates the current originating
from the electrochemical reactions taking place at the working electrode, it directly reflects the
signal which is measured
reference electrode. In order to maintain the potential always at the desired applied
value, a third electrode is required. This is the counter electrode. A current flow
over the counter electrode is used for compensation, so that the periodic potential
between working and reference electrodes can be precisely controlled. This reg-
ulation circuit is usually managed by a potentiostat.
Different choices can be made for the electrodes, electrolyte and conducting
salt. A detailed discussion would be beyond the scope of this book and can be
found elsewhere [1, 2]. Frequently, organic semiconductors are studied by cyclic
voltammetry using organic solvents as electrolyte, e.g. acetonitrile. The working
electrode is typically made of glassy carbon, gold or platinum. As reference
electrode, for example electrodes based on the redox couple Ag/Ag+ may be used,
and a platinum electrode is typically used as counter electrode. Concerning the
sample, there are essentially two possibilities: Either the analyte is dissolved in the
electrolyte, or it is deposited as a solid film on the working electrode.
The theory of cyclic voltammetry is rather complex. It is well understood for
simple cases, for example, if the analyte is dissolved in water and if the redox
reactions at the electrodes are uninhibited [1, 2]. However, organic semiconductors
7.1 Fundamentals of Cyclic Voltammetry 113
After having passed the turning point, the sample is reduced, meaning that elec-
trons are filled back into the HOMO level (sketched reduction peak). The LUMO
level, in contrast, would give rise to another pair of oxidation and reduction peaks
located in another potential region.
To conclude from the observed oxidation and reduction peaks on the absolute
position of the HOMO and LUMO levels with respect to vacuum, one needs to
know the potential of the reference electrode with respect to vacuum. This is
illustrated in the energy scheme in Fig. 7.3. One can infer from the scheme that the
ionization potential Ip and electron affinity Ea can be obtained by (7.1) and (7.2),
respectively:
vac
jIP j ¼ DEHOMO þ Eref ; ð7:1Þ
vac
and jEa j ¼ DELUMO þ Eref ð7:2Þ
vac
Therein, Eref denotes the potential of the reference electrode with respect to
vacuum, and DEHOMO ; DELUMO are the potentials with respect to the reference
electrode to be derived from the cyclic voltammogram. Note that absolute values
vac
are used for Ip, Ea and Eref in the above equations to avoid confusion with the
definitions of the signs of these quantities. For organic semiconductors measured
as thin films on the electrodes, DEHOMO and DELUMO are usually identified as the
onset potentials of the oxidation and reduction peaks corresponding to the HOMO
and LUMO level, respectively [3–6].
In contrast, in studies of colloidal semiconductor nanocrystals, there is some
inconsistency, whether onset or peak potentials should be used [7, 8]. There is still
a lack of theory providing clear instructions how CV data measured under different
circumstances and for different types of samples should be correctly analyzed.
This introduces some uncertainty into the accurate determination of the absolute
7.1 Fundamentals of Cyclic Voltammetry 115
In the case of PCBM, there is rather good agreement between the values
reported for the HOMO levels as measured by CV by different authors. Values
range from -5.9 to -6.1 eV with respect to vacuum [11–13]. The LUMO level
was reported to be in the range from -3.7 to -3.9 eV [9, 12, 13]. Such good
agreement is, however, not always obtained. For example, in the case of P3HT
values reported for the HOMO level range from -4.76 eV [6, 14] to -5.24 eV
[11, 15]. Reported values for the LUMO level of P3HT scatter even within a range
from -2.5 eV [13, 14] to -3.5 eV [11]. As large variations appear unlikely to be
real variations of the material property (e.g. due to different molecular weight of
the polymer). More likely, the scatter of the values reflects the uncertainties
associated with CV measurements and the corresponding data analysis and
referencing.
For many optoelectronic devices, the position of the energy levels is quite
crucial. Therefore, one should carefully consider how reported values for energy
levels were measured and determined by CV. If several organic materials are
involved, it is also preferable to measure all materials under similar conditions. In
a comparative study of a series of materials with one setup and a defined analysis
method, at least some uncertainties can be eliminated.
Apart from determining the absolute energetic positions of the HOMO and LUMO
levels, respectively the band edges, of semiconducting materials, cyclic voltam-
metry can also provide information on defect states present in a sample. As dis-
cussed for example in Sect. 2.4, semiconductor nanocrystals can contain defect
states energetically located within the energy gap. Removal of electrons from or
injection of electrons into defect levels will show up in the cyclic voltammogram
as well. For example, Kucur et al. [16, 17] investigated colloidal CdSe nano-
crystals by cyclic voltammetry and were able to detect a variety of electronic
defect states originating from vacancies and other structural defects.
References
7. E. Kucur, J. Riegler, G.A. Urban, T. Nann, J. Chem. Phys. 119, 2333 (2003)
8. H. Zhong, S.S. Lo, T. Mirkovic, Y. Li, Y. Ding, Y. Li, G.D. Scholes, ACS Nano 4, 5253
(2010)
9. Y. He, H.-Y. Chen, J. Hou, Y. Li, J. Am. Chem. Soc. 132, 1377 (2010)
10. Q. Xie, E. Perez-Cordero, L. Echegoyen, J. Am. Chem. Soc. 114, 3978 (1992)
11. M. Al-Ibrahim, H.-K. Roth, M. Schroedner, A. Konkin, U. Zhokhavets, G. Gobsch, P.
Scharff, S. Sensfuss, Org. Electron. 6, 65 (2005)
12. Q. Wei, T. Nishizawa, K. Tajima, K. Hashimoto, Adv. Mater. 20, 2211 (2008)
13. S. Wilken, D. Scheunemann, V. Wilkens, J. Parisi, H. Borchert, Org. Electron. 13, 2386
(2012)
14. T.V. Richter, C.H. Braun, S. Link, M. Scheuble, E.J.W. Crossland, F. Stelzl, U. Würfel, S.
Ludwigs, Macromolecules 45, 5782 (2012)
15. W.S. Shin, S.C. Kim, S.-J. Lee, H.-S. Jeon, M.-K. Kim, B.V.K. Naidu, S.-H. Jin, J.-K. Lee,
J.W. Lee, Y.-S. Gal, J. Polym. Sci. Part A: Polym. Chem. 45, 1394 (2007)
16. E. Kucur, W. Bücking, R. Giernoth, T. Nann, J. Phys. Chem. B 109, 20355 (2005)
17. E. Kucur, W. Bücking, T. Nann, Microchim. Acta 160, 299 (2008)
Chapter 8
Absorption and Photoluminescence
Spectroscopy
When light passes through matter, it can be absorbed which results in an attenu-
ation of the intensity. The probability of absorption is basically dependent on the
material, on the wavelength and on the distance which the photons travel through
the material. In the simplest approach, absorption can be described by the Beer-
Lambert law, given in (8.1):
I ðd Þ ¼ I0 ead ð8:1Þ
Therein, I0 and I are the intensities of the incident and attenuated light beams,
d is the thickness of the sample (i.e., the distance which the photons have to travel
in the material), and a is a coefficient which depends on the material and on
I ðd Þ ¼ I0 eedc ; ð8:2Þ
where c is the concentration, and e is the molar absorptivity, alternatively also
called the molar extinction coefficient. The coefficients a or e in (8.1) and (8.2),
respectively, should not be confused with the extinction coefficient j as the
imaginary part of the complex refractive index. The quantities are related to each
other, but not identical. Their relationship is given by (8.3):
4p jðkÞ
að kÞ ¼ e ð kÞ c ¼ ; ð8:3Þ
k
where k is the wavelength. The product a d is usually called the absorbance or
optical density. Furthermore, using the basis 10 instead of Euler’s number, one can
in analogue manner define the decadic absorption coefficient a10 by (8.4):
Fig. 8.1 Optical characterization of colloidal ZnO nanocrystals NCs dissolved in a chloro-
form:ethanol mixture. a UV-Vis absorbance spectrum. The inset shows a Tauc plot, where the
optical band gap was determined by plotting (A hm)2 against photon energy (hm) and
extrapolating the straight portion of the curve to the interception with the energy axis. A denotes
the measured absorbance. b Photoluminescence spectrum for an excitation wavelength of
340 nm. The two peaks are attributed to band edge and defect luminescence, respectively. The
inset shows an excitation spectrum, i.e., the PL intensity at a fixed detection wavelength of
550 nm, while the excitation wavelength was varied. (Reprinted from Wilken et al. [1],
Improvement of ITO-free inverted polymer-based solar cells by using colloidal zinc oxide
nanocrystals as electron-selective buffer layer, Copyright (2012), with permission from Elsevier)
obviously no longer recombine radiatively from the LUMO level of the donor
polymer. Thus, PL quenching is a useful tool to probe charge transfer processes in
donor/acceptor solar cells. However, care must be taken in the interpretation,
because under certain conditions, the PL intensity can also be reduced by processes
concurring with charge transfer. If the acceptor material has a lower energy gap than
the donor, it is also possible that the entire electron–hole pair is transferred to the
acceptor by Förster resonance energy transfer (FRET). If this is possible, it becomes
difficult to distinguish between both processes in simple PL quenching experiments.
8.3 Photoinduced Absorption Spectroscopy 123
Fig. 8.2 Photoluminescence spectra of thin films of pure P3HT, a P3HT/CdSe blend, and a
P3HT/PCBM blend. A 532 nm laser was used for excitation. All spectra were recorded at 80 K,
and all samples contained the same amount of polymer. (Reproduced with permission from
Heinemann et al. [3]: Copyright (2009) Wiley-VCH Verlag GmbH & Co. KGaA)
Fig. 8.3 The working principle of PIA spectroscopy of donor/acceptor blends. A laser pulse
creates electron-hole pairs in the donor polymer (step 1). If an electron acceptor is present, the
excited electron can be transferred to the acceptor (step 2). The hole remaining on the polymer
forms a polaron having energy levels located inside the HOMO–LUMO gap of the neutral
molecule (step 3). The lower polaron level is occupied by one electron. If now, the sample is
illuminated with a white light source, absorption occurs at energies corresponding to transitions
involving the polaron levels, labeled P1 and P2 in the figure (step 4). Thus, the transmission of the
sample after the laser excitation differs from the transmission in the ground state
light source, but without laser excitation. Let us denote the corresponding signal as
TWL. Next, the photoluminescence caused by the laser needs to be measured with
the laser switched on, but without the white light source (PLLaser). The third
measurement is done with both light sources, with chopped laser excitation and
Lock-In technology to detect the laser-induced changes of the transmission. Let us
denote the corresponding signal as DTWL,Laser. From these measurements, the PIA
signal can finally be calculated according to (8.7):
DT DTWL;Laser PLLaser
¼ ð8:7Þ
T TWL
Thus, the measured changes of the transmission are corrected for the PL caused
by the laser excitation, and the signal is divided by the transmission measured
without laser excitation. Figure 8.4 shows as an example for steady state PIA
spectra results from a study of blends of P3HT and PbSe nanocrystals [5]. In this
PIA spectrum, mainly three features are visible at about 0.4, 1.05, and 1.25 eV,
respectively. The signals at 0.4 and 1.25 eV can be assigned to the polaronic
transitions labeled P1 and P2 in Fig. 8.3. Their appearance provides therefore
evidence that charge transfer has occurred between P3HT and the PbSe nano-
crystals in the given experiment [5]. The last peak at about 1.05 eV is most likely
due to triplet–triplet absorption in the polymer [5, 6]. In fact, light absorption
creates first a singlet exciton, but the sample may undergo intersystem crossing, so
that a triplet state is reached. If part of the polymer molecules is still in this state
when the transmission is probed, the still excited molecule can be excited further
into a higher triplet state. This process will show up in the PIA spectrum as well.
Alternatively, according to a study of P3HT by Österbacka et al. [7], the feature at
8.3 Photoinduced Absorption Spectroscopy 125
1.05 eV might also be assigned to interchain singlet excitons, i.e., to singlet ex-
citons with the electron and hole on different polymer chains.
It is noteworthy in this place, that the modulation frequency is an important
parameter in PIA spectroscopy. Contributions to the PIA signal are only obtained
from excited states having a lifetime that is shorter than the inverse of the mod-
ulation frequency. In other words, if the modulation frequency is more and more
increased, signals from long-living species get more and more suppressed.
Therefore, analyzing the frequency-dependency of the PIA signals can reveal
information on the lifetime of the excited states. A few examples for such type of
analysis will be given later in Sect. 12.3.3.
Transitions between electronic states and also charge transfer can be very fast.
Many relevant processes occur on time scales in the nanosecond, picosecond or
even femtosecond range. Thus, if one intends to analyze the kinetics of an elec-
tronic process, time-resolved spectroscopy methods are required.
A relatively widely used method is time-resolved photoluminescence spec-
troscopy. Thereby, the sample is excited with a short laser pulse, and afterwards,
the decay of the PL signal is monitored as a function of time, the temporal
resolution depending on the setup. Figure 8.5 shows exemplarily PL decay curves
measured for thin films of P3HT and quasi-bilayer structures of P3HT and PCBM
[8]. In this context, a quasi-bilayer means that PCBM is deposited on top of a
P3HT film using a solvent for the second layer that was initially believed not to
dissolve the P3HT layer. However, closer inspection revealed that PCBM can
partly penetrate into the P3HT layer. Thus, the structures obtained do not corre-
spond to an ideal bilayer system [8]. The decay of the PL signal from the pure
P3HT film takes place on a picosecond time scale. According to the Figure,
126 8 Absorption and Photoluminescence Spectroscopy
roughly 50 % of the initial PL intensity is lost after 300 ps. If a PCBM overlayer is
present, the decay becomes much faster, because charge transfer takes place.
Time-dependent PL decay curves can be analyzed quantitatively to determine
decay constants. Therefore, usually exponential decay functions are assumed. The
evaluation is simple, if only a single process is involved and the decay follows an
exponential behavior. The data can then be fitted according to (8.8):
1
lnðPLðtÞÞ ¼ lnðPLðt0 ÞÞ t ð8:8Þ
s
Therein, PL(t) and Pl(t0) are the PL signal intensities at the time t and t0,
respectively, and s is the time constant of the decay process as fitting parameter.
However, sometimes a mono-exponential decay is not suitable to obtain a rea-
sonable fit of the experimental decay data. In that case, more complicated models
with more than one free parameter are required.
8.4 Time-Resolved Optical Spectroscopy 127
References
Electron spin resonance (ESR) spectroscopy, often also called electron paramag-
netic resonance (EPR) spectroscopy, is a method suitable to study electronic states
with a spin non-zero. Detailed introductions to ESR spectroscopy can for example
be found in [1]. Here, only a short summary of the most important basics will be
given. Let us first consider a free electron. The spin angular momentum and its
z-component are given by (9.1) and (9.2), respectively:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
~
S ¼ h s ðs þ 1Þ ð9:1Þ
and
Sz ¼ h ms ð9:2Þ
Therein, s is the spin quantum number which is s = 1/2 for a free electron, and
ms is the magnetic quantum number which can be ms = +1/2 (‘‘spin up’’) or
ms = -1/2 (‘‘spin down’’). Associated with the spin angular momentum, the
electron has a magnetic spin moment, given by (9.3):
ls ¼ ce ~
~ S
e ~
¼ ge S ð9:3Þ
2me
l
¼ ge B ~
S
h
The constant of proportionality ce between the spin angular momentum and the
magnetic moment is called the gyromagnetic ratio, lB is the Bohr magneton, and
ge is the g-factor. For a free electron, the g-factor equals ge = 2.0023. If an
external magnetic field is applied, the magnetic spin moment has a potential
energy, given by the following expression:
ls ~
E ¼ ~ B ð9:4Þ
If the B field is parallel to the z-axis, (9.4) becomes with the help of (9.3) and
(9.2):
E ¼ ge lB ms B ð9:5Þ
By consequence, the two states with spin up and spin down will split up in an
external magnetic field parallel to the z-axis, as illustrated in Fig. 9.1.
In thermal equilibrium the occupancy of both states is determined by the
Boltzmann factor. For the ratio of the probabilities of occupancy follows:
Pðms ¼ þ1=2Þ Pþ DE ge l B B
¼ ¼ ekB T ¼ e kB T ð9:6Þ
Pðms ¼ 1=2Þ P
To give a numeric example, at room temperature and for a magnetic field of
B = 0.5 T, the ratio becomes 0.998. In other words, both states are almost equally
occupied. The equilibrium can be perturbed by microwave irradiation, because
resonant microwave absorption occurs, if the energy of the radiation matches the
splitting of the quantum states. The condition for resonance can be expressed by
(9.7):
hmL ¼ DE ¼ ge lB B ð9:7Þ
The resonance frequency mL is called the Larmor frequency. Coming now to the
spectroscopy, in an ESR spectrometer, the sample to be studied is placed into a
magnetic field, usually generated by a pair of Helmholtz coils. Additionally, there
is a microwave cavity, so that the sample can be excited with microwave radiation.
The microwave radiation will be absorbed according to the resonance condition.
9.1 Fundamentals of Electron Spin Resonance Spectroscopy 131
hmL ¼ ge lB Blocal
¼ ge lB ð1 rÞB ð9:9Þ
¼ glB B with g ¼ ð1 rÞge
So, the resonance condition can be expressed with the external magnetic field, if
a factor g is introduced which differs from the g-factor ge of the free electron. In
fact, the g-factor introduced here depends on the chemical environment of the spin.
Therefore, the measurement of the resonance frequency and thus the g-factor can
provide important information about the studied sample. In general, the g-factor
can be anisotropic.
A note is appropriate concerning typical measurement conditions in ESR
spectroscopy. Usually, measurements are done with a fixed microwave frequency,
and the magnetic field is varied to study the resonance behavior. Resonance
absorption should result in a peak, if the microwave absorption is plotted against
the magnetic field. However, most spectrometers use lock-in technology and
directly measure the change of the absorption when the field is varied. Therefore,
132 9 Electron Spin Resonance
the ESR signal usually looks like the first derivative of an absorption peak.
Figure 9.2 shows schematically the ESR signal to be expected for a free electron.
In matter, ESR spectra look not always as simple as in Fig. 9.2, because the
spin of an electron can additionally couple with nuclear spins in the local sur-
rounding. This process is called hyperfine interaction and can lead to a splitting of
the signal into several components. Further information on the hyperfine interac-
tion can be found elsewhere [1].
(TOP/TOPO) and hexadecylamine (HDA). Optionally, this rather thick and dense
shell was replaced with pyridine by ligand exchange. The authors could show that
charge separation at the donor/acceptor interface is more efficient, if the initial
ligand shell surrounding the nanocrystals is replaced by pyridine [10]. CdSe
nanocrystals were studied with respect to charge transfer as well in combination
with P3HT [2, 11, 12]. Dietmueller et al. [13] investigated blends of Si nano-
crystals and P3HT or PCBM by L-ESR spectroscopy. As an interesting result, they
could show that both, P3HT/Si and Si/PCBM, form functional donor/acceptor
systems, meaning that the Si nanocrystals could serve either as electron acceptor or
as electron donor in solar cells [13].
Probing the success of the charge separation process at donor/acceptor inter-
faces is not the only valuable information that can be retrieved from L-ESR
studies. Additional information can be obtained by studying the decay of the ESR
signal, when the light source is switched off. Figure 9.4 shows for example the
time dependence of the signal intensity for the hole polaron in organic P3HT/
PCBM and hybrid P3HT/CdSe blends as studied by Heinemann et al. [2].
In the case of the organic blend, the signal decays relatively fast back to the
initial level, if the excitation source is switched off. Obviously, separated charge
carriers can easily diffuse back to the interface and recombine [2]. In contrast, a
fast and a very slow decay process are observable in the hybrid blend. The slow
decay process (‘persistent signal’) was interpreted to be due to trapping of charge
carriers in defect states, so that they cannot easily diffuse back to the interface and
recombine [2]. This example shows that L-ESR is also suitable to reveal infor-
mation about the presence of trap states for charge carriers. Slow decay kinetics of
part of the light-induced signal was previously observed also for purely organic
MDMO-PPV/PCBM blends at T = 90 K [4, 14]. In detail, important differences
occur between the organic and hybrid systems. In the case of the MDMO-PPV/
PCBM blends, the persistent signal was removable by annealing the samples to
9.2 Light-Induced Electron Spin Resonance (L-ESR) Spectroscopy 135
Fig. 9.5 Schematic illustration of charge transfer (CT) complexes in a donor/acceptor system.
The scheme shows the HOMO and LUMO levels of the electron donor and acceptor. The
effective band gap EGeff is the energy difference between the acceptor’s LUMO and the donor’s
HOMO level. The CT states are located within the effective band gap and determine the upper
max
limit of the open-circuit voltage VOC of organic solar cells
Fig. 9.6 ESR spectra of a P3HT/PCBM blend at T = 80 K. The microwave power and
frequency were 2 mW and 9.44 GHz, respectively. The black curve shows the ESR signal in the
dark. The green curve shows the signal after excitation with green laser light (2.33 eV photon
energy), and the red curve shows the signal after excitation with infrared laser light (1.58 eV
photon energy). (Reprinted with permission from [17]. Copyright 2010 American Chemical
Society)
low energy photons [17]. More recently, Behrends et al. [22] showed that also the
dissociation of charge transfer complexes can be studied with ESR spectroscopy
when applying time-resolved measurements where a pulsed laser is used to excite
the sample.
References 137
References
1. J.E. Wertz, J.R. Bolton, Electron Spin Resonance: Elementary Theory and Practical
Applications (Chapman and Hall, New York, 1986)
2. M.D. Heinemann, K. von Maydell, F. Zutz, J. Kolny-Olesiak, H. Borchert, I. Riedel, J. Parisi,
Photo-induced charge transfer and relaxation of persistent charge carriers in polymer/
nanocrystal composites for applications in hybrid solar cells. Adv. Funct. Mater. 19,
3788–3795 (2009)
3. N.S. Sariciftci, L. Smilowitz, A.J. Heeger, F. Wudl, Science 258, 1474 (1992)
4. V. Dyakonov, G. Zoriniants, M.C. Scharber, C.J. Brabec, R.A.J. Janssen, J.C. Hummelen,
N.S. Sariciftci, Phys. Rev. B 59, 8019 (1999)
5. J. De Ceuster, E. Goovaerts, A. Bouwen, J.C. Hummelen, V. Dyakonov, Phys. Rev. B 64,
195206 (2001)
6. S. Sensfuss, M. Al-Ibrahim, A. Konkin, G. Nazmutdinova, U. Zhokhavets, G. Gobsch,
D.A.M. Egbe, E. Klemm, H.-K. Roth, Proc. SPIE 5215, 129 (2004)
7. M. Al-Ibrahim, H.-K. Roth, M. Schroedner, A. Konkin, U. Zhokhavets, G. Gobsch, P.
Scharff, S. Sensfuss, Org. Electron. 6, 65 (2005)
8. V.I. Krinichnyi, E.I. Yudanova, N.N. Denisov, J. Chem. Phys. 131, 044515 (2009)
9. P.A. van Hal, M.P.T. Christiaans, M.M. Wienk, J.M. Kroon, R.A.J. Janssen, J. Phys. Chem. B
103, 4352 (1999)
10. M. Pientka, V. Dyakonov, D. Meissner, A.L. Rogach, D.V. Talapin, H. Weller, L. Lutsen, D.
Vanderzande, Nanotechnology 15, 163 (2004)
11. I. Lokteva, N. Radychev, F. Witt, H. Borchert, J. Parisi, J. Kolny-Olesiak, J. Phys. Chem. C
114, 12784 (2010)
12. N. Radychev, I. Lokteva, F. Witt, J. Kolny-Olesiak, H. Borchert, J. Parisi, J. Phys. Chem. C
115, 14111 (2011)
13. R. Dietmueller, A.R. Stegner, R. Lechner, S. Niesar, R.N. Pereira, M.S. Brandt, A. Ebbers,
M. Trocha, H. Wiggers, M. Stutzmann, Appl. Phys. Lett. 94, 113301 (2009)
14. N.A. Schultz, M.C. Scharber, C.J. Brabec, N.S. Sariciftci, Phys. Rev. B 64, 245210 (2001)
15. K. Marumoto, M. Kato, H. Kondo, S. Kuroda, N.C. Greenham, R.H. Friend, Y. Shimoi, S.
Abe, Phys. Rev. B 79, 245204 (2009)
16. C. Carati, L. Bonoldi, R. Po, Phys. Rev. B 84, 245205 (2011)
17. F. Witt, M. Kruszynska, H. Borchert, J. Parisi, J. Phys. Chem. Lett. 1, 2999 (2010)
18. T. Drori, C.-X. Sheng, A. Ndobe, S. Singh, J. Holt, Z.V. Vardeny, Phys. Rev. Lett. 101,
037401 (2008)
19. K. Vandewal, K. Tvingstedt, A. Gadisa, O. Inganas, J.V. Manca, Nat. Mater. 8, 904 (2009)
20. C. Deibel, T. Strobel, V. Dyakonov, Adv. Mater. 22, 4097 (2010)
21. J.E. Brandenburg, X. Jin, M. Kruszynska, J. Ohland, J. Kolny-Olesiak, I. Riedel, H. Borchert,
J. Parisi, J. Appl. Phys. 110, 064509 (2011)
22. J. Behrends, A. Sperlich, A. Schnegg, T. Biskup, C. Teutloff, K. Lips, V. Dyakonov, R. Bittl,
Phys. Rev. B 85, 125206 (2012)
Chapter 10
Electrical Characterization of Solar Cells
Abstract Photovoltaic cells convert the energy of light emitted by the sun into
electricity. A basic question is to judge how efficient this energy conversion
process is for a given cell. Therefore, the performance of the solar cells must be
characterized by electrical measurements. This chapter gives an introduction to a
number of basic methods for the electrical characterization. The probably most
fundamental technique is the measurement of current–voltage curves. Although
the measurement itself is relatively simple, the conditions how measurements
should be carried out are not as simple and will also be discussed here. Another
important technique is the determination of the external or internal quantum
efficiency. These methods provide information how efficient the conversion of
light into electrical current is for photons of a specific wavelength. Spectrally
resolved quantum efficiency measurements can therefore provide information on
the contribution of different materials in the solar cell to the energy conversion
process.
10.1.1 Fundamentals
Fig. 10.1 Current density–voltage (J-V) curves for a bulk heterojunction solar cell with P3HT/
CdSe as donor/acceptor system. The black curve was measured in the dark. The red curve was
measured under illumination with simulated sunlight (AM 1.5G, 100 mW/cm2). The blue curve
(referring to the axis on the right) represents the power density as obtained by multiplying the
voltage with the current density
In the dark, a solar cell should behave like a diode. For an ideal pn-junction
solar cell, the dependency of the current density on voltage would be given the
Shockley equation [1]:
ffi
eV
J ðV Þ ¼ J0 exp 1 ð10:1Þ
kB T
Therein, e is the elementary charge, V is the applied voltage, kB is the Boltz-
mann constant, T is the temperature, and J0 is the so-called reverse saturation
current density, i.e., the current density flowing through the diode, if a (relatively
high) reverse bias is applied. Under illumination, a photocurrent is generated. The
photocurrent, treated as a positive quantity and denoted JPh (as a current density),
flows in the opposite direction as the current due to the injection of charge carriers
by a positive applied voltage. Accordingly, (10.1) must be modified as follows
under illumination:
ffi
eV
J ðV Þ ¼ J0 exp 1 JPh ð10:2Þ
kB T
Equation (10.2) describes the J-V curve for an ideal pn-junction solar cell under
illumination. The J-V curve under illumination contains several characteristic
points. At zero voltage, no current is injected by an external voltage source. The
current is then purely due to current generation by the conversion of solar energy.
At this point, the current density is called the short-circuit current density, JSC.
At a certain positive voltage, the injection current compensates the photocurrent,
10.1 Current-Voltage Measurements 141
so that no net current is flowing. The voltage at this point is called the open-circuit
voltage, VOC. Concerning (10.1) and (10.2), at least two important questions arise:
Is this model also suitable to describe organic or hybrid solar cells with donor/
acceptor heterojunctions? And how should non-ideal behavior be treated? The first
question is quite fundamental. A donor/acceptor blend is of course not the same as
a classical pn-junction for which the Shockley equation can be theoretically
derived. Nevertheless, the concept turned out to be suitable to describe also J-V
curves of organic and hybrid solar cells. However, deviations from the behavior of
an ideal diode should be taken into account. Often, this can be done by assuming
an equivalent circuit model, where a series and a parallel resistance are introduced
(see Fig. 10.2).
The series resistance, RS, represents resistive losses, e.g., due to barriers at
interfaces and contacts. For an ideal solar cell, the series resistance should be zero.
The parallel resistance, RP, takes into account that the current may find an
alternative path in a real solar cell, e.g., if contact materials penetrating into the
organic layers lead locally to shorts (also called shunts). The parallel resistance,
sometimes also called shunt resistance, should be infinite in an ideal solar cell.
From the equivalent circuit model in Fig. 10.2, the following relationship between
the current density and voltage can be deduced [2–4]:
ffi ffi
1 e ðV JRS AÞ V
J ðV Þ ¼ J0 exp 1 JPh ð10:3Þ
1 þ RS =RP nkB T RP A
Equation (10.3) is sometimes referred to as the enhanced Shockley equation.
Therein, A denotes the area of the device, and n is the so-called ideality factor of the
diode. For an ideal diode, n should equal 1. Note that (10.3) cannot be analytically
solved for the current density, because J occurs also in the exponential function.
However, (10.3) can be used to fit experimental J-V curves to this theoretical
expression. Thereby, RS, RP, J0, and n can be treated as fitting parameters [2–4].
Frequently, the parallel and series resistance are estimated from J-V curves
without fitting the whole curves. This can be done by evaluating the slope of the
142 10 Electrical Characterization of Solar Cells
J-V curves at high forward bias and at 0 V. In forward direction, the slope is
dominated by the series resistance, so that RS may be calculated according to
(10.4) [5]:
1 dV
RS ¼ lim V!1 ð10:4Þ
A dJ
At 0 V, the slope is determined by the sum of the series and the parallel
resistance, so that RP can be obtained by (10.5) [5]:
1 dV
RP ¼ RS ð10:5Þ
A dJ V¼0V
Another important quantity is of course the power delivered by the solar cell.
Power is the product of voltage and current, and it depends on the voltage applied,
how much power is obtained. The blue curve in Fig. 10.1 shows the power for
voltages between 0 V and VOC. At a certain voltage, a maximum is obtained. This
point of operation is called the maximum power point (MPP). The dashed area in
Fig. 10.1 represents the power delivered at the MPP. With JMPP and VMPP
denoting the current density and voltage at the maximum power point, respec-
tively, one can furthermore define the fill factor (FF):
JMPP VMPP
FF ¼ ð10:6Þ
JSC VOC
The fill factor describes how ‘rectangular’ the J-V curve is. For solar cells with
high fill factor, JMPP and VMPP have to approach JSC and VOC, respectively. The
shape of the current–voltage curve will then become more and more rectangular.
To calculate the power conversion efficiency (PCE, g), the power delivered by the
device has to be divided by the incident power from the sunlight (Plight):
JMPP VMPP
PCE ¼ 100 % ð10:7Þ
Plight
Using the definition of the fill factor, the efficiency can also be expressed by
(10.8):
FF JSC VOC
PCE ¼ 100 % ð10:8Þ
Plight
incident light intensity and also the spectral intensity distribution of the incident
light. Every solar cell is expected to convert the energy stored in light coming from
the sun. But although we have just one light source in mind, the sunlight available
to illuminate a photovoltaic cell strongly depends on the place on earth. This is due
to the fact that part of the sunlight is absorbed and scattered on its way from the
sun through the atmosphere of the earth. The spectral intensity distribution of the
sunlight on the surface of the earth depends on the distance which the photons
have to travel through the atmosphere. This in turn depends on the angle between
the incident light and the surface normal of the earth sphere. Thus, the solar
spectrum irradiating the surface of the earth depends on the place on earth. To
enable nevertheless comparison between performance data, scientists and engi-
neers have agreed to measure J-V curves under defined illumination conditions.
Outside the atmosphere of the earth, the solar spectrum has a defined spectral
intensity distribution. This extraterrestrial spectrum is called the AM 0 spectrum.
The abbreviation AM 0 stands for ‘air mass zero’ and means that the light does not
pass through any air in the atmosphere. If the sun is at the zenith, the distance
which the light has to pass through the air is minimal. The corresponding spectrum
at the surface of the earth is called the AM 1.0 spectrum. Thereby, several
assumptions have to be made concerning the absorbing species in the atmosphere.
However, as the most-widely accepted standard to measure solar cell performance
data, the so-called AM 1.5G spectrum has been chosen. This spectrum corresponds
to specific angular orientations, where the sun is located at an angle of about 48.2
with respect to the zenith, and where the surface normal of the solar cell forms an
angle of 37 with respect to the zenith [6, 7]. Furthermore, there are again a
number of conditions concerning the attenuation of the incident light on its way
through the atmosphere [6, 7]. To realize corresponding illumination conditions in
a laboratory, solar simulators are used, where complex filter systems modify the
emission spectrum of a light source in order to approximate the AM 1.5G spec-
trum. The degree to which the simulated spectrum matches the real AM 1.5G
spectrum depends on the instrument, and solar simulators are classified into dif-
ferent categories accordingly.
Apart from the spectral distribution, also the total (integrated) intensity is of
importance, because the performance of many solar cells depends on the illumi-
nation power. As a standard, solar cells should be measured at a total illumination
power of 1,000 W/m2 (=100 mW/cm2). Furthermore, the temperature should be
25 C during the measurements.
Concerning again the spectral distribution of the radiation, another relevant
issue for the correct characterization of solar cells is spectral mismatch. In this
book, only the problem shall be outlined. In order to expose the solar cells to a
defined incident light intensity (1,000 W/m2), the illumination power of the solar
simulator must be adjusted. This is usually done in practice by adjusting the
distance between the light source and the sample. This adjustment procedure
requires measuring the light intensity. Typically, a reference solar cell (made of
crystalline Si) with a known PCE is placed under the solar simulator, and the
distance is adjusted until this reference solar cell yields the correct current
144 10 Electrical Characterization of Solar Cells
corresponding to the desired illumination power. Afterwards, the sample solar cell
can be placed at the same distance under the solar simulator and measured.
This procedure neglects, however, that the reference solar cell and the sample to
be investigated can have a different spectral response. For example, a typical
organic solar cell will absorb in a narrower spectral range than a Si reference solar
cell. If the simulated spectrum differs from the real AM 1.5G spectrum, calibrating
the illumination power with a Si solar cell integrates over the spectral range
absorbed by silicon (up to *1,100 nm). Let us hypothetically suppose that a non-
ideal solar simulator provides too much photons in the range between 400 and
500 nm, but less photons in the range between 900 nm and 1,000 nm (with respect
to the AM 1.5G standard spectrum). During measurement of the reference solar
cell, the too high photon flux in the low wavelength range will be partly com-
pensated by the low photon flux in the long wavelength range. However, if
afterwards a sample absorbing only up to *600 nm is placed under the solar
simulator, the low photon flux at long wavelength cannot result in any compen-
sation of the too high photon flux at lower wavelength. By consequence, within the
absorption range of the sample, the illumination power provided by the solar
simulator would be too high in this example, the reason being the spectral mis-
match between the sample and the reference solar cell. In principle, it is possible to
correct measured performance data for spectral mismatch. Procedures how to take
this phenomenon into account are for example described in [8, 9]. It should be
critically noted that spectral mismatch is unfortunately still neglected in many
works.
Another aspect meriting special attention is the active area. In the electrical
measurements, a current is measured. In order to determine the current density,
one needs to know the active area of the solar cell. Unfortunately, measuring the
active area is not as simple [9]. In many studies, the active area is assumed to be
given by the geometric overlap of the electrodes, as illustrated in Fig. 10.3.
However, electron–hole pairs generated by light absorption outside the as-defined
active area have also a certain chance to contribute to the photocurrent, because
the electrodes will not only collect charge carriers generated directly below them.
The border region surrounding the active area will also give a certain contribution
to the current. Thus, defining the active area as the geometric overlap of the
electrodes slightly underestimates the area in which light absorption is relevant for
the energy conversion process. By consequence, the current density will be
overestimated, if the current is divided by a too small area.
Gupta et al. [10] investigated the contribution of the border region to the
photocurrent in detail. By exciting organic solar cells locally with a small laser
spot, they could clearly show that charge carriers generated outside the area
defined by the geometric overlap of the electrodes contribute to the photocurrent as
well [10]. According to these experiments with the local laser excitation, the width
of the contributing border region can be estimated to be of the order of *50 lm.
Furthermore, solar cells with different active area in the range of 0.01–0.5 cm2
were prepared in the same study [10]. Whereas the open-circuit voltage was not
10.1 Current-Voltage Measurements 145
Fig. 10.3 3D view a and top view b of a typical bulk heterojunction solar cell architecture. The
geometric overlap of the electrodes, marked in red in part b, defines the active area of the device.
However, some charge carriers generated in the surroundings (border region marked in green) of
the as-defined active area may be collected at the electrodes as well
affected by the area, the calculated current density strongly depended on the area.
For the devices with the smallest geometric overlap of the electrodes (0.01 cm2),
the calculated current density was about 85 % higher than for devices with 0.5 cm2
active area [10]. This can partly be attributed to the edge effects which are more
important in the case of small areas. Assuming a width of 50 lm for the con-
tributing border region, one should expect to overestimate the current density of a
device with a geometrically defined active area of 1 mm 9 1 mm by
about *10 %. In contrast, the influence of a 50 lm large border region would be
only *1 % for a device with an active area of 1 cm2.
Still referring to the work by Gupta et al. [10], for devices with 0.09 cm2 active
area, the calculated current density was still about 20 % higher than for the largest
devices with an active area 0.5 cm2. Only in the range of 0.25–0.5 cm2, the
differences became less significant [10].
These considerations show that the current density and thus the device effi-
ciency deduced from the electrical characterization of solar cells are unfortunately
not independent of the active area. In particular in the case of small devices, a
rather strong dependence can be observed. Partly, this can be attributed to edge
effects which are non-negligible in small devices. Nevertheless, using active areas
of only about 0.1 cm2 is quite common in research at lab scale. A way to cir-
cumvent the contribution of border regions to the photocurrent is to use shadow
masks for the illumination of the solar cells. However, also in that case care must
be taken to work properly. A recent work by Snaith [9], mainly focused on dye-
sensitized solar cells, but also considering organic BHJ devices, reports on dif-
ferent methods to realize precise and reliable measurement of current–voltage
curves.
Finally, it should be noted here that many scientific studies compare series of
samples prepared and measured under comparable conditions. To allow for con-
clusions within a given series of experiments, taking care of spectral mismatch,
146 10 Electrical Characterization of Solar Cells
edge effects, etc., is certainly not indispensable. However, neglecting such effects
makes it more difficult to compare results reported in different studies. From this
point of view, it would be desirable to achieve improved standards in research.
Zkmax
EQE e
JSC ¼ PðkÞ k EQEðkÞ dk ð10:9Þ
hc
0
Fig. 10.4 The external quantum efficiency of a depleted heterojunction solar cell with PbS
nanocrystals as absorber material. The layer sequence of the materials in the solar cell was
fluorine-doped tin oxide (FTO)/porous TiO2/PbS nanocrystals/Au. The absorption spectrum of
the PbS quantum dots is included for comparison. (Reprinted with permission from [11].
Copyright 2010 American Chemical Society)
References
1. C. Kittel, Introduction to Solid State Physics, 8th edn. (Wiley, New York, 2005)
2. S. Yoo, B. Domercq, B. Kippelen, J. Appl. Phys. 97, 103706 (2005)
3. W.J. Potscavage Jr, A. Sharma, B. Kippelen, Acc. Chem. Res. 42, 1758 (2009)
148 10 Electrical Characterization of Solar Cells
4. J. Huang, J. Yu, H. Lin, Y. Jiang, Chin. Sci. Bull. 55, 1317 (2010)
5. D. Chirvase, J. Parisi, J.C. Hummelen, V. Dyakonov, Nanotechnology 15, 1317 (2004)
6. C.A. Gueymard, D. Myers, K. Emery, Sol. Energy 73, 443 (2002)
7. National Laboratory of Renewable Energy (NREL), http://rredc.nrel.gov/solar/spectra/.
Accessed 5th Aug 2013
8. V. Shrotriya, G. Li, Y. Yao, T. Moriarty, K. Emery, Y. Yang, Adv. Funct. Mater. 16, 2016
(2006)
9. H. Snaith, Energy Environ. Sci. 5, 6513 (2012)
10. D. Gupta, M. Bag, K.S. Narayan, Appl. Phys. Lett. 93, 163301 (2008)
11. A.G. Pattantyus-Abraham, I.J. Kramer, A.R. Barkhouse, X. Wang, G. Konstantatos, R.
Debnath, L. Levina, I. Raabe, M.K. Nazeeruddin, M. Grätzel, E.H. Sargent, ACS Nano 4,
3374 (2010)
12. F. Zutz, I. Lokteva, N. Radychev, J. Kolny-Olesiak, I. Riedel, H. Borchert, J. Parisi, Phys.
Status Solidi A 206, 2700 (2009)
Chapter 11
Charge Carrier Mobility Measurements
mobility for electrons, le, as in (11.4), and the mobility for holes, lh, according to
(11.5), p denoting therein the concentration of holes:
rh ~ rh rh
vD ¼
~ E ¼ þlh E ; with lh ¼ ¼ ; q ¼ þe
~ ð11:5Þ
pq pq pe
For a semiconductor, the conductivity in total is then obtained from the con-
tributions of both the electrons and holes, as expressed in (11.6):
r ¼ re þ rh ¼ nele þ pelh ð11:6Þ
It should be noted that the above equations describe a simple case. In reality,
the mobility is usually not constant for electrons and holes. The mobility can
depend on the temperature and the electric field [1, 5]. Furthermore, it is not
necessarily isotropic.
nanoparticle surface, up to 0.9 and 0.2 cm2 V-1 s-1 were obtained for the electron
and hole mobility, respectively [14]. The possibility to tune of the conductivity in a
wide range from n-type to p-type behavior was also shown for PbS nanocrystals
[15].
Due to the design of OFETs, charge transport occurs in those devices always in a
narrow channel parallel to the substrate. However, charge transport is not always
isotropic. In particular, some organic semiconductors, among them also the widely
used polymer P3HT, tend to crystallize during the preparation of thin films. If the
orientation of crystalline domains on the substrate is not arbitrary, charge transport
is expected to become strongly anisotropic [16]. Therefore, it is useful to study
charge transport also along the direction perpendicular to the semiconductor film.
This is possible with the help of single carrier devices [17, 18].
In a single carrier diode, the sample to be studied, e.g., a polymer film is placed
between two electrodes, similar as in a solar cell. However, the electrodes are not
chosen to have high and low work function, respectively. In hole-only devices,
both electrodes must have a high work function. The electrodes can then form
Ohmic contacts to the HOMO level of the organic semiconductor, so that holes can
easily be injected and transported through the device. In contrast, a large barrier
results for the injection of electrons into the LUMO level of the semiconductor.
Therefore, electron conduction is suppressed. In electron-only devices, both
electrodes must have a low work function to enable electron injection into the
LUMO level, whereas hole injection into the HOMO level is suppressed [17, 18].
Figure 11.2 visualizes the concept of such single carrier diodes.
In such single carrier diodes with ideal Ohmic contacts, the conductivity will be
limited by the low charge carrier mobility of the organic semiconductor. As a
consequence, so-called space charges will accumulate in the diode, when carriers
are injected at the contacts by an applied voltage. The space charges, in turn, lower
the electric field imposed by the applied voltage and therefore limit the current.
Finally a space-charge limited current (SCLC) density is obtained, which can be
described by (11.10), which is known as the Mott-Gurney law [2, 19]:
9 V2
jSCLC ¼ ee0 l 3 ; ð11:10Þ
8 d
where e is the relative dielectric constant, V is the applied voltage, and d is the
thickness of the semiconductor layer. The equation is valid for current density–
voltage measurements under dark conditions and unipolar transport, meaning that
only holes or only electrons contribute to the transport of electric charge. Fur-
thermore, Ohmic contacts are assumed, and any other effects like trap states in the
semiconductor are neglected. Provided these conditions are fulfilled, (11.10) can
154 11 Charge Carrier Mobility Measurements
References
15. O. Voznyy, D. Zhitomirsky, P. Stadler, Z. Ning, S. Hoogland, E.H. Sargent, ACS Nano 6,
8448 (2012)
16. E.J.W. Crossland, K. Rahimi, G. Reiter, U. Steiner, S. Ludwigs, Adv. Funct. Mater. 21, 518
(2011)
17. V.D. Mihailetchi, L.J.A. Koster, P.W.M. Blom, C. Melzer, B. de Boer, J.K.J. van Duren,
R.A.J. Janssen, Adv. Funct. Mater. 15, 795 (2005)
18. V.D. Mihailetchi, H.X. Xie, B. de Boer, L.J.A. Koster, P.W.M. Blom, Adv. Funct. Mater. 16,
699 (2006)
19. C. Melzer, E.J. Koop, V.D. Mihailetchi, P.W.M. Blom, Adv. Funct. Mater. 14, 865 (2004)
20. M.M. Mandoc, B. de Boer, P.W.M. Blom, Phys. Rev. B 73, 155205 (2006)
21. T. Kirchartz, Beilstein J. Nanotechnol. 4, 180 (2013)
Part III
Solar Cells with Colloidal Nanocrystals
Chapter 12
Hybrid Polymer/Nanocrystal Solar Cells
Abstract Part III of the book discusses concepts to realize solar cells involving
colloidal nanocrystals. The first concept, being the subject of this chapter, is hybrid
polymer/nanocrystal solar cells. Hybrid solar cells are basically similar to organic
polymer/fullerene bulk heterojunction solar cells, but with the difference that the
fullerene acceptor is replaced by inorganic semiconductor nanocrystals. The active
layer is then a composite of an organic component, the conductive polymer, and an
inorganic component, the semiconductor nanocrystals. This organic–inorganic
material combination gave rise to the expression ‘‘hybrid solar cells’’. This chapter
discusses general, potential advantages of hybrid solar cells with respect to organic
polymer/fullerene devices. An up-to-date overview over the success achieved with
different material systems is provided as well as insight into the difficulties
encountered. In general, the performance of hybrid solar cells still lacks behind
that of organic polymer/fullerene devices. Therefore, special emphasis is dedicated
to the specific differences between hybrid and organic solar cells.
Fig. 12.1 Upper part The solar spectrum (standard AM 1.5G spectrum) is shown together with
arbitrarily scaled absorption spectra of a typical P3HT film and a colloidal solution of CdSe
nanocrystals. Lower part The spectral range accessible due to the quantum size effect is visualized
for selected semiconductors. All data is based on optical measurements. The gray shaded range
corresponds to the size-dependent position of the first excitonic absorption maximum or to the
maximum of photoluminescence spectra or to the absorption onset. The particle diameters given
in brackets specify to which size range the grey shaded range corresponds. Data was taken for
CdSe and CdTe from [5], for InP from [7], for CuInS2 from [8, 9], for PbSe from [6], and for PbS
from [10]. (The data given in this figure does not necessarily mean that the size range accessible
by synthesis methods is restricted to the values mentioned here)
that only light absorption in the donor polymer plays a role. At first sight, one
might think that the poor absorption of PCBM can simply be compensated by
making thicker active layers of the donor/acceptor blend. However, increasing
layer thickness will lead to higher losses related to charge transport. Therefore, it
should be more suitable, if the acceptor material in the active layer gave also a
significant contribution to the absorption of sun-light which can be converted into
electrical current. Possible candidates for electron acceptors which can themselves
efficiently harvest sun-light are inorganic semiconductor nanocrystals. Due to the
quantum size effect discussed in Sect. 2.3, the absorption range can even be tuned
within certain limits. For example, tuning the diameter of CdSe nanocrystals from
1.5 to 8 nm results in a shift of the first excitonic absorption maximum from *400
to *680 nm [5], and in the case of PbSe, the first absorption maximum shifts from
*1,050 to *1,650 nm, if the particle diameter is varied from 3 to 5.5 nm [6].
Figure 12.1 gives a graphical overview over the accessible spectral range with
selected inorganic semiconductors in comparison to the solar emission spectrum.
As one can see, in some materials, e.g., PbS, the energy gap can be tuned over a
wide range by controlling the particle size.
12.1 Potential Advantages of Using Inorganic Nanocrystals 161
In order to efficiently harvest sun-light, not only the energy gap is of impor-
tance, but also the absorption coefficient. From this point of view, inorganic
semiconductor nanocrystals are also suitable candidates as absorber material in
solar cells. For example, CdSe nanocrystals have an absorption coefficient of the
order of 105 cm-1 at 350 nm [11]. This is comparable to the absorption coefficient
of typical conjugated polymers. Therefore, such semiconductor nanocrystals give a
significant contribution to the absorption of sunlight, if incorporated into the active
layer of hybrid solar cells. This was for example shown by Zutz et al. [12] who
presented absorption spectra measured through the active layer of P3HT/CdSe
solar cells. Furthermore, measurements of the external quantum efficiency showed
that light absorption in the nanocrystals can also lead to a significant contribution
to the photocurrent generated by P3HT/CdSe solar cells [13, 14]. In more detail, it
depends on the design of the nanoparticle surface, how efficient the conversion of
photo-generated excitons in the nanoparticles into extractable photocurrent really
is [14]. This issue will be discussed in detail in Sect. 12.3.3. In this place should
only be mentioned that most synthesis routes of colloidal CdSe nanocrystals yield
particles surrounded by a relatively thick organic ligand shell. In a pioneering
work, Greenham et al. [15] examined in 1996 the charge transfer in blends of
MEH-PPV and TOPO-capped CdS and CdSe nanocrystals. They could show that
the compact TOPO ligand shell was unsuitable for charge separation. However,
after exchange of the original ligands with pyridine, evidence for successful charge
separation was found [15]. Ligand exchange with pyridine remained the standard
procedure to treat the surface of CdSe nanocrystals for application in hybrid solar
cells for quite a long time. Solar cells based on quasi-spherical, pyridine-capped
CdSe quantum dots in conjunction with typical polymers such as P3HT or PPV
derivatives were reported to reach efficiencies of approximately 1 % [12, 16].
Improved light absorption is not the only potential advantage of semiconductor
nanocrystals with respect to typical fullerene derivatives such as PCBM. Another
important issue concerns charge transport. In a polymer/fullerene solar cell,
electrons must be transported by thermally assisted hopping through a chain of
individual fullerene molecules. In analogous manner, the electrons must be
transferred from nanoparticle to nanoparticle in a hybrid solar cell. Here it is
possible, however, to use elongated nanocrystals such as nanorods or tetrapods. If
the nanocrystals are oriented with a long axis parallel to the direction of desired
charge transport, i.e., perpendicular to the active layer, the number of required
hopping steps can be strongly reduced. In 2002, Huynh et al. [17] could con-
vincingly demonstrate the beneficial effect of elongated nanocrystals by comparing
hybrid solar cells with quasi-spherical CdSe nanoparticles and CdSe nanorods of
different length. The external quantum efficiency more than doubled with 60 nm
long nanorods instead of quantum dots with 7 nm diameter, and power conversion
efficiencies of 1.7 % were obtained under AM 1.5 conditions with the nanorod-
based solar cells [17]. In a disordered donor/acceptor blend, nanorods still have the
disadvantage that facile charge transport along the growth axis does not help, if the
nanorods are not oriented perpendicular to the active layer. Therefore, the bene-
ficial effect is partly suppressed in the case of random orientation. In contrast, ideal
162 12 Hybrid Polymer/Nanocrystal Solar Cells
tetrapods will always have one of their 4 branches oriented more or less perpen-
dicular to the active layer. Indeed, it was shown later by Sun et al. [18], that using
blends of conductive polymer (MDMO-PPV) and CdSe tetrapods the efficiency
can be improved to almost 3 %.
A third potential advantage of colloidal semiconductor nanocrystals originates
from the quantum size effect. The change of the energy gap as a function of
particle size is accompanied by a shift of the conduction and valence band edges.
For CdSe, pseudopotential calculations by Wang and Zunger [19] revealed that the
conduction band edge shifts up in energy while the valence band edge shifts down,
if the particle size is reduced. Referring to Fig. 9.5 in the chapter about ESR
spectroscopy (Sect. 9.2), an up-shift of the acceptor’s conduction band edge means
an increase of the effective bang gap of the donor/acceptor heterojunction. Since
the effective band gap is related to the maximum possible open-circuit voltage,
tuning the particle size of semiconductor nanocrystals should enable adjusting the
energy levels in order to improve the open-circuit voltage. Indeed, Brandenburg
et al. [20] could demonstrate that P3HT/CdSe solar cells show a size-dependent
open-circuit voltage, although the effect remained less pronounced than expected
according to the theoretically predicted shift of the conduction band edge.
The potential advantages depicted here show that colloidal semiconductor
nanocrystals are exciting candidates as electron acceptor components in polymer-
based solar cells. Nevertheless, it was not possible until now to really benefit from
these potential advantages. So far, hybrid solar cells still lack behind their organic
counterparts. In the following sections, on overview over the achievements with
different material systems will be given, elementary processes and the specificities
of hybrid solar cells as compared to organic solar cells will be discussed.
between MEH-PPV and CdSe or also CdS nanoparticles. In the case of MEH-PPV
blended with TOPO-capped CdS, no quenching of the PL of the polymer was
observed at all. With TOPO-capped CdSe, the PL was quenched, but this was
attributed to Förster resonance energy transfer (FRET), i.e., transfer of the exciton
instead of the electron only. In fact, the CdSe particles had an energy gap which
was smaller than the HOMO–LUMO gap of the donor polymer, so that energy
transfer was possible in this system. After ligand exchange with pyridine,
enhanced quenching of the PL was observed, and this was interpreted as a proof
for charge transfer between MEH-PPV and pyridine-capped CdSe [15]. Thus, the
work provided evidence that ligand exchange procedures have to be applied prior
to using colloidal nanocrystals in hybrid solar cells. The authors prepared also test
solar cells in the same work and studied their performance with monochromatic
laser excitation at 514 nm. A photovoltaic effect was observed, but the quantum
efficiency reached only 12 % at this wavelength. It was estimated that the effi-
ciency of these first hybrid solar cells would be approximately 0.1 % under
standard test conditions (one-sun AM 1.5 conditions) [15].
Poor charge transport was assumed to be one of the main difficulties, and as
already outlined in Sect. 12.1, using nanorods instead of quasi-spherical particles
was examined as a strategy for improvement [17, 21]. Hybrid solar cells with
P3HT as polymer and pyridine-capped CdSe nanorods were shown to reach 1.7 %
power conversion efficiency under standard test conditions in 2002 [17]. Due to
this success, CdSe tetrapods attracted attention in the following years, because
these three-dimensional objects will even in the case of random orientation pro-
vide a better pathway with less hopping steps for electrons towards the electrode
[18, 22]. Still using ligand exchange with pyridine, hybrid solar cells with CdSe
tetrapods and MDMO-PPV reached 2.8 % efficiency under standard test condi-
tions in 2005 [18].
The improvement from 0.1 to about 3 % efficiency was certainly not solely due
to the usage of elongated nanocrystals instead of quasi-spherical quantum dots. In
parallel to the studies focusing on the crystallite shape, many efforts were
undertaken to optimize relevant preparation parameters, such as the choice of
solvents, blending ratios and annealing temperatures. In particular, the choice of
the solvent for processing the hybrid blends from solution is not simple. Typical
solvents used in organic photovoltaics are chloroform, chlorobenzene or dichlo-
robenzene. Unfortunately, colloidal CdSe nanocrystals that have been subjected to
ligand exchange with pyridine are not soluble in these types of solvents in suffi-
ciently high concentration. They are instead soluble in pyridine, but pyridine is not
a suitable solvent for conductive polymers such as P3HT or the commonly used
PPV derivatives. As a compromise, one has to use binary solvent mixtures, e.g., of
chloroform and pyridine to get both the donor and acceptor component dissolved
in the desired concentration in one solution. Huynh et al. [23] addressed the
question to optimize the solvent ratio in such binary mixtures. They studied films
of pyridine-capped CdSe nanorods and P3HT processed from chloroform-pyridine
mixtures and could show that the surface roughness as measured by atomic force
microscopy (AFM) is small (with a root mean square of about 5 nm) only in a
164 12 Hybrid Polymer/Nanocrystal Solar Cells
Fig. 12.2 a TEM image of CdSe tetrapods. The length of the scale bar is 50 nm. b Absorption
spectra of films of pure CdSe tetrapods, pure PCPDTBT and a CdSe/PCPDTBT blend. c NREL-
certified current–voltage curve of a hybrid solar cell with CdSe tetrapods and PCPDTBT. d The
structure of PCPDTBT (parts a–c Reprinted with permission from [26]. Copyright 2010
American Chemical Society)
the hybrid layers with ethanedithiol as an additional preparative step. Therefore, the
hybrid films deposited on ITO/PEDOT:PSS substrates were immersed into a
solution of ethanedithiol in acetonitrile and then rinsed with pure acetonitrile,
before the active layer was annealed and the device finished by the thermal evap-
oration of an aluminum contact [30]. This treatment was found to remove residual
TOPO and TDPA ligands to a large extent, and the performance of the photovoltaic
devices was considerably enhanced due to higher photocurrent densities. In the case
of P3HT, the power conversion efficiency increased from 2.2 to 2.9 %, and in the
166 12 Hybrid Polymer/Nanocrystal Solar Cells
case of PCPDTBT, from 3.3 to 4.7 % under standard test conditions [30]. Till date,
this remained the highest efficiency reported for CdSe-based hybrid solar cells.
Comparing PCPDTBT to P3HT, advantages besides the lower band gap were
discussed to be higher hole mobility and an energetically lower-lying HOMO level
[29]. The latter point merits special attention. The HOMO level of PCPDTBT was
determined by cyclic voltammetry by several groups. Reported values vary in the
range of -5.0 eV [31] to -5.3 eV [32] with respect to vacuum. This is in average
indeed slightly lower than the values reported in average for the HOMO level of
P3HT (see discussion in Sect. 7.2). Zhou et al. [33] made a direct comparison of
hybrid solar cells with CdSe quantum dots combined with either PCPDTBT or
P3HT. In that work, the choice of the polymer had only little impact on the
open-circuit voltage, which was even slightly higher in the case of P3HT [33].
Thus, benefiting from the lower-lying HOMO level of PCPDTBT in the regard to
increase the effective band gap of the donor/acceptor system and therefore the
open-circuit voltage was not achieved, so far. The main improvement of the device
performance achieved with PCPDTBT in comparison to P3HT can clearly be
attributed to higher photocurrent densities [27].
Of interest is also the alignment of the LUMO level of PCPDTBT and the
conduction band edge of CdSe. LUMO levels of PCPDTBT were measured to be
in the range of -3.4 eV [31] to -3.57 eV [32]. For CdSe, the valence and con-
duction band edges depend on the particle size due to the quantum size effect. The
size-dependency of the absolute position of the band edges was studied in detail by
Jasieniak et al. [34]. Besides the particle size, also the surface chemistry was
shown to have an impact on the position of the energy levels. Using pyridine
ligands instead of alkyl amines was found to cause a down-shift of the valence
band by as much as 0.35 eV [34]. If we consider now pyridine-capped, spherical
particles with about 4.7 nm diameter, as used for example in the above-mentioned
work on CdSe/PCPDTBT solar cells by Jeltsch et al. [29], the valence and con-
duction band edges of CdSe are expected at about -5.75 and -3.55 eV, respec-
tively (taking already into account the shift induced by pyridine) [34].
Accordingly, the conduction band edge of CdSe seems to have only a very small
offset with respect to the LUMO level of PCPDTBT. Nevertheless, the material
combination forms a working donor/acceptor system as evidenced by the perfor-
mance of corresponding hybrid solar cells.
Another important tendency in current research on polymer/CdSe solar cells is
the elaboration of alternative strategies to the ligand exchange with pyridine. With
the usage of short alkylamines instead of pyridine, the performance of hybrid solar
cells based on P3HT and quasi-spherical CdSe nanocrystals was successfully
increased from 1 to 2 % power conversion efficiency [14, 35]. Similar success was
achieved by treating the surface of CdSe nanocrystals with hexanoic acid [36].
Quasi-spherical CdSe nanoparticles treated with hexanoic acid reached 2.1 %
efficiency in hybrid solar cells with P3HT [33, 36], and even 2.7 % in cells with
PCPDTBT [33]. The reasons for the improvements caused by other surface
treatments will be discussed later in Sect. 12.3.
12.2 Material Combinations for Hybrid Solar Cells 167
Other cadmium chalcogenides, CdS and CdTe, were also considered for hybrid
solar cells. CdS has a larger band gap (*2.4 eV for bulk material). Therefore, it is
less interesting as efficient absorber of sun-light than CdSe and remained, possibly
for this reason, less intensively studied for a long time, although CdS was shown to
be suitable as electron acceptor material already in 1996 [15]. In 2007, Wang et al.
[37] prepared hybrid solar cells with CdS nanocrystals and MEH-PPV. TEM
images of the nanocrystals revealed a shape which other authors might have called
tetrapods, but in this work the nanocrystals were named multiarmed nanorods,
because not all of the particles had exactly four arms [37]. In fact, deviations from
the ideal tetrapod geometry can also be observed in many studies of CdSe where
so-called ‘‘tetrapods’’ were used. The ligand shell originating from the synthesis
consisted of hexadecylamine and was replaced with pyridine by ligand exchange,
and the MEH-PPV/CdS hybrid solar cells reached efficiencies of about 1.2 %
under 100 mW/cm2 illumination with simulated AM 1.5 radiation [37].
Recently, considerable progress was made with CdS nanocrystals for hybrid
solar cells. Ren et al. [38] examined CdS quantum dots stabilized by butylamine in
combination with P3HT and could reach about 0.6 % power conversion efficiency
in usual bulk heterojunction solar cells. In the same work, the performance was
strongly enhanced by preforming nanowires of the conjugated polymer and
chemically grafting the CdS quantum dots onto the P3HT nanowires. The short
circuit current density increased approximately from 2 to 6 mA/cm2 [38]. Addi-
tionally treating the surface of the nanoparticles by ethanedithiol resulted in further
improvement of the solar cells. The short circuit current density was raised to
*10 mA/cm2 and the open-circuit voltage reached even *1.0 V due to the large
energy level offset between the HOMO of P3HT and the conduction band edge of
the CdS quantum dots. These characteristics would make the system able to
compete with typical polymer/fullerene solar cells. Unfortunately, the fill factor
(*0.32) remained quite low, so that the average device performance under
standard test conditions was ‘‘only’’ *3.2 %. The best device was reported to
reach even 4.1 % efficiency [38]. Thus, despite the large band gap which may
appear less favorable for efficient light absorption, CdS nanocrystals were recently
demonstrated to be comparably useful for hybrid solar cells as CdSe nanoparticles.
It will be interesting to see in future, if the grafting procedure applied in the cited
work to CdS can also be transferred to other semiconductor nanocrystals.
CdTe has a lower bulk band gap of *1.45 eV, so that the absorption extends into
the near infra-red (NIR) region. Therefore, CdTe nanocrystals might be expected to
be more suitable for hybrid solar cells than CdSe or CdS. Zhou et al. [24] investi-
gated this question in a systematic series of experiments with pyridine-treated
CdSexTe1-x tetrapods. Bulk heterojunction solar cells were fabricated with blends
of these nanocrystals in combination with MEH-PPV as conductive polymer.
Devices with pure CdSe showed 1.1 % efficiency at 80 mW/cm2 illumination with
simulated AM 1.5 radiation, but the efficiency decreased monotonously with
increasing tellurium content. The hybrid solar cells with MEH-PPV and pure CdTe
tetrapods exhibited only 0.003 % power conversion efficiency [24]. To elucidate the
origin of this behavior, the authors studied the absolute position of the energy levels
168 12 Hybrid Polymer/Nanocrystal Solar Cells
of their materials by cyclic voltammetry. The valence band edge of CdTe tetrapods
was determined to be at -5.0 eV with respect to the vacuum level, while -5.1 eV
were found for the HOMO level of MEH-PPV. Therefore, it was concluded that
energy transfer occurs instead of charge separation, meaning that the MEH-PPV/
CdTe system does not form a working donor/acceptor combination [24]. Attempts
to realize P3HT/CdTe solar cells resulted in efficiencies below 1 % as well [39, 40].
Although the explanation given above for the low efficiency of CdTe-based
hybrid solar cells appears convincing, there is some doubt about the exact positions
of the energy levels in the material. Recently, Jasieniak et al. [34] addressed the
question of precise measurements of absolute energy levels and studied CdSe, CdTe,
PbS and PbSe nanocrystals with great care by photoelectron spectroscopy in air
(PESA). The experimental data was used among others to establish a formula for the
size-dependent valence band edge of CdTe quantum dots. According to these results,
the valence band edge of CdTe nanocrystals depends only slightly on the particle
diameter in the range of 2–6 nm and is situated close to -5.0 eV with respect to the
vacuum level [34]. This appears to be in good agreement with the values mentioned
before. However, there are also other reports, mentioning significantly differing
values. For example, -5.5 eV were reported in [41] and even -6.1 eV in [42].
Interestingly, there are also a few studies where CdTe nanoparticles seem to
work suitably as electron acceptors. Kang et al. [43] reported on hybrid solar cells
where vertically aligned CdTe nanorods fabricated by an electrodeposition method
were infiltrated with poly(3-octylthiophene) (P3OT), and the devices reached
*1 % power conversion efficiency. Furthermore, in 2011 appeared a report on
PPV/CdTe hybrid solar cells, where 2.1 % efficiency was measured under standard
conditions [42]. This work is interesting not only because of the relatively good
performance of the CdTe-based hybrid solar cells, but also because the active layer
was deposited from water as solvent. For this purpose, 2-mercaptoethylamine was
used as stabilizer for the CdTe quantum dots which had an average diameter of
2.8 nm [42]. Despite the success it should be noted that the authors state in their
article that the valence band edge of the CdTe nanocrystals was measured by
ultraviolet photoelectron spectroscopy (UPS) to be located at -6.1 eV with respect
to the vacuum level [42]. This value shows a discrepancy of *1 eV as compared
to the results provided by Jasieniak et al. [34]. Moreover, unusually high annealing
temperatures of 320 C were used to fabricate the PPV/CdTe solar cells [42]. This
leaves some doubt, if the organic components (PPV and PEDOT:PSS) of these
hybrid solar cells will remain intact. In a follow-up work, the efficiency of hybrid
solar cells involving a BHJ layer of PPV and water-soluble CdTe nanocrystals was
reported to be increased up to 4.7 %, although also in this case relatively high
annealing temperatures (250 C) were used [129].
In another study, up to 3.2 % power conversion efficiency were reported for
hybrid solar cells with a BHJ of CdTe tetrapods and a low band-gap polymer
having its HOMO and LUMO levels at -5.1 and -3.65 eV, respectively [44]. The
solar cells were constructed in an inverted architecture involving also an interface
between the CdTe/polymer layer with an interlayer made of C60 [44]. Values
mentioned for the valence and conduction band edges of the CdTe nanoparticles
12.2 Material Combinations for Hybrid Solar Cells 169
were -5.78 and -4.07 eV, respectively, thus again significantly lower than values
found by other authors.
In conclusion, values given for the absolute position of the energy levels in
CdTe nanocrystals present large scatter and remain under controversial discussion.
However, there is meanwhile experimental evidence that also CdTe nanocrystals
may serve as electron acceptors in polymer/nanoparticle hybrid solar cells.
The lead chalcogenides PbSe and PbS have small band gaps as bulk material, and
due to the large exciton Bohr radius, the absorption edge can be tuned in a large
spectral range by controlling the particle size (see Fig. 12.1). This makes these
compounds very attractive as efficient light absorbing materials with adjustable
optical properties. A number of studies were devoted to polymer/nanocrystal
hybrid solar cells with this type of colloidal semiconductors [45–51]. For a long
time, the performance remained inferior to 1 % power conversion efficiency, and
thus inferior than for hybrid solar cells involving colloidal Cd chalcogenide
nanocrystals. Only recently, this has changed.
In the case of PbSe-based hybrid solar cells, the relative position of energy
levels at the donor/acceptor interface seems to play a crucial role. Jiang et al. [49]
measured the energy levels in MEH-PPV, P3HT and colloidal PbSe nanocrystals
of different size by cyclic voltammetry. In the combination with MEH-PPV, the
HOMO level of the polymer was found to lie below the valence band edge of the
PbSe nanocrystals for all particle sizes studied (4–10 nm diameter). In contrast,
with P3HT as donor polymer, the valence band edge of the nanocrystals was found
to move below the polymer’s HOMO level, if the particle size is reduced below a
critical value of *5–8 nm diameter [49]. However, also with a particle diameter
of 4 nm, the offset between the filled energy levels in the donor and acceptor was
found to be only 0.17 eV, and the best solar cells prepared within that work did not
exceed 0.1 % power conversion efficiency [49].
The questions whether charge transfer is possible or not at polymer/PbSe
nanocrystal interfaces was later also addressed by measurements with photo-
induced absorption spectroscopy. Noone et al. [52] performed a PIA study of
blends of PPV or P3HT with small PbSe nanocrystals (3.5 nm diameter), where
the original ligand shell consisting of oleic acid was exchanged to a butylamine
ligand shell. Upon light excitation, no indications for the formation of long-lived
polarons in the polymer were found, whereas control experiments with CdSe
nanocrystals as electron acceptors showed well pronounced spectral signatures of
optical transitions involving polaron levels. Thus, it was concluded, that charge
transfer leading to long-lived polarons does not take place in the studied polymer/
PbSe blends [52]. However, the same article contains a note that other researchers
observed polaron transitions in PIA spectra of P3HT/PbSe blends, where the initial
ligand shell was exchanged by pyridine [52]. Thus, it remains still an open
170 12 Hybrid Polymer/Nanocrystal Solar Cells
Fig. 12.3 Schematic illustration of thermalization losses (a) and the MEG process (b)
In order to benefit from MEG in a solar cell, the big challenge is to extract the
photo-generated charge carriers before they recombine. This is a difficult task,
because multiple excitons in a single quantum dot can recombine relatively fast by
Auger recombination. Klimov calculated the lifetimes for two electron–hole pairs
(called a biexciton) in several II–VI semiconductor nanocrystals as a function of
particle size [54]. Biexciton lifetimes were found to be of the order of *20–50 ps
for PbSe quantum dots [54]. This means that charge extraction from the quantum
dots has to take place on a faster time scale, if one wants to benefit from the MEG
process. So far, there are only a few reports claiming that multiple excitons were
successfully extracted. In a study from 2005 about MEH-PPV/PbSe hybrid solar
cells, measurements of the external quantum efficiency yielded values up to 150 %,
which was attributed to the successful extraction of multiple excitons [45]. How-
ever, this report was controversially discussed by other authors later [57]. Recently
appeared a more convincing article by Sambur et al. [58], where PbS nanocrystals
were chemically bound to TiO2 in quantum dot-sensitized solar cells. The authors
determined the incident photon-to-current efficiency (IPCE) and the absorbed
photon-to-current efficiency (APCE). In simple words, the APCE corrects the IPCE
for losses due to incomplete absorption of light by the solar cell. It was shown that
above a threshold for the photon energy, slightly larger than the double of the
energy gap, the APCE increased well above 100 %, meaning that multiple excitons
were successfully extracted in those devices [58]. Although the IPCE was not high
in that work, the study provides promising evidence that it may become possible in
future to really exploit the phenomenon of multiple exciton generation for the
fabrication of efficient solar cells based on semiconductor quantum dots.
Although this chapter concentrates on hybrid polymer/nanocrystal solar cells, it
is appropriate to anticipate here that lead chalcogenide nanocrystals turned out to
be very promising materials for polymer-free solar cells [59]. Those device con-
cepts will be treated in detail in Chap. 13.
Despite of the success achieved with cadmium and lead chalcogenide semicon-
ductor nanocrystals outlined in the preceding sections, those materials have of
course the inherent disadvantage that they contain highly toxic elements (Cd or Pb).
Therefore, there is a strong interest in finding alternative semiconductors that are
less toxic, but have also suitable absorption properties, energy levels and electrical
properties for solar cell applications. Semiconductors of the I–III–VI family, such
as CuInS2 or CuInSe2, are regarded as promising candidates in this view.
First research in this field was made in 2003 by Arici et al. [60] who prepared
CuInS2 nanocrystals by a colloidal route using triphenyl phosphite as capping
ligand. Powder X-ray diffraction provided evidence that the nanocrystals had
chalcopyrite structure and revealed an average particle size of 3–4 nm, but evi-
dence for a narrow size distribution was not given [60]. The nanocrystals were
12.2 Material Combinations for Hybrid Solar Cells 173
tested in different types of solar cells. Devices with a bilayer of CuInS2 and PCBM
embedded between ITO/PEDOT:PSS and LiF/Al as anode and cathode, respec-
tively, yielded under white light illumination with 80 mW/cm2 incident intensity a
promising open-circuit voltage of 0.7–0.8 V, a short-circuit current density of
0.26 mA/cm2, a fill factor of 0.44 and a power conversion efficiency of *0.1 %
[60]. In another configuration, the CuInS2 phase was additionally blended with
PEDOT:PSS, so that the active layer was a bilayer of PCBM and a CuInS2/
PEDOT:PSS blend. The addition of PEDOT:PSS improved the short-circuit cur-
rent density to 0.84 mA/cm2, but the voltage and fill factor were reduced, so that
the efficiency did not improve [60].
In another work from the same group, CuInSe2 nanocrystals were synthesized
using trioctylphosphine oxide as ligand [61]. Hybrid solar cells with a bulk het-
erojunction of the nanocrystals and P3HT were prepared. The open-circuit voltage
and fill factor reached 1 V and *0.5, respectively, but the short-circuit current
density was limited to *0.3 mA/cm2 which resulted in an efficiency of nearly 0.2 %
at 80 mW/cm2 illumination power [61]. Although these initial works did not result
in highly efficient solar cells, it was successfully demonstrated that CuIn(S,Se)2
nanocrystals can lead to operating devices and might indeed be suitable candidates
to replace the toxic cadmium or lead chalcogenides in future. In particular, the
mentioned works left largely space for further optimization. For example, no ligand
exchange procedures were reported to be applied in those works, although it is
known from previous reports on CdSe and CdS that a thick ligand shell, e.g.,
consisting of TOPO, is not suitable for efficient charge transfer [15].
Only in recent years, colloidal synthesis methods for CuInS2 enabling a high
degree of structural control in terms of narrow size distributions and defined
particle shapes were developed [62–69]. This gave a new impact to research on
polymer/CuInS2 hybrid solar cells. Yue et al. [68] synthesized CuInS2 nanocrystals
with zincblende structure and around 3 nm particle diameter. Cyclic voltammetry
was used to measure the nanocrystals’ energy levels and yielded -5.8 and -
4.0 eV with respect to vacuum for the valence and conduction band edges,
respectively, so that the nanocrystals should form a type II heterojunction with
common polymer such as P3HT or PPV [68]. The authors realized BHJ hybrid
solar cells with MEH-PPV and tested the devices under monochromatic illumi-
nation (470 nm, 16 mW/cm2). Although the devices suffered from low current
densities and poor fill factors, the study revealed that the surface of the nano-
crystals plays an important role. Initially, thiophenol ligands were used to stabilize
the colloidal particles. It was demonstrated that a surface treatment with 4-tert-
butylpyridine significantly enhances the open-circuit voltage and the short-circuit
current density [68].
Radychev et al. [70] prepared CuInS2 nanocrystals with wurtzite crystal
structure and studied them in BHJ solar cells with P3HT as polymer. In this case, a
surface treatment with pyridine was applied to the nanoparticles which had after
synthesis a ligand shell composed of oleylamine, dodecanethiol and TOPO. Solar
cells with pyridine-treated CuInS2 nanocrystals of about 10 nm size reached an
open-circuit-voltage of 0.4 V, a short-circuit current density of 0.28 mA/cm2 and a
174 12 Hybrid Polymer/Nanocrystal Solar Cells
Fig. 12.4 Current density–voltage (J–V) curves for CuInS2/P3HT hybrid solar cells in the dark
and under illumination with simulated sun-light (100 mW/cm2, AM1.5G radiation). The
nanocrystals used were treated with pyridine and had a rod-like (a) or pyramidal shape (b). Note
the difference in the order of magnitude on the current density axis of the two plots. The insets
show TEM images of the pure nanocrystals before treatment with pyridine (Reprinted from
Organic Electronics, vol. 13, Radychev et al. [70], Investigation of the morphology and electrical
characteristics of hybrid blends based on poly(3-hexylthiophene) and colloidal CuInS2
nanocrystals of different shapes, pp. 3154–3164, Copyright (2012), with permission from
Elsevier)
fill factor of 0.24, which resulted in a still low power conversion efficiency of
0.03 %; Fig. 12.4 shows the corresponding J–V curves [70]. Additionally, CuInS2
nanorods were synthesized and applied in the solar cells, with the intention to
benefit from the particle shape. However, the performance was even worse in the
case of rods (see Fig. 12.4). As a reason, electron microscopy analysis of the
morphology of the active layer revealed that the pyridine-treated nanocrystals had
a strong tendency to form large aggregates [70]. The morphology could be con-
siderably improved by applying a surface treatment with hexanethiol instead of
pyridine. This resulted also in a better rectifying behavior of the diodes, but the
photovoltaic performance was not improved. Finally, analysis of the energy levels
indicated an unfavorable alignment between CuInS2 and P3HT. The valence band
12.2 Material Combinations for Hybrid Solar Cells 175
reaching up to 1.1 % power conversion efficiency [74]. The weight ratio between
the nanoparticles and P3HT was 1:1 in that study, which corresponds to an
unusually small volume fraction of the nanoparticles. Furthermore, the synthesis
involved relatively long-chained ligands, namely oleic acid, oleylamine, and
1-dodecanethiol. These were not reported to be exchanged before device fabrica-
tion [74]. Thus, the promising device performance reached appears a bit surprising.
Future research will have to reveal more on the application potential of AgInS2
nanocrystals as well as mixed systems of AgInS2 and CuInS2 in hybrid solar cells.
Several III–V semiconductors possess absorption properties which are suitable for
solar cell applications as well. However, so far, there are only a few research
reports on hybrid solar cells with III–V semiconductor materials. Pientka et al. [75]
studied charge transfer at the interface of colloidal InP nanocrystals and MDMO-
PPV by photo-induced absorption spectroscopy and light-induced electron spin
resonance. The nanoparticles had a ligand shell consisting of TOP/TOPO after
synthesis, and no indications for charge transfer were found in corresponding
blends with the polymer. However, after ligand exchange with pyridine, evidence
for successful charge transfer at the donor/acceptor interface was found [75].
Corresponding hybrid solar cells were not prepared in that study. Later, Novotny
et al. [76] fabricated hybrid solar cells, where an array of InP nanowires deposited
by chemical vapor deposition was infiltrated with P3HT. A photoresponse was
observed, but the efficiency of the devices was very low, because the short-circuit
current density was only of the order of a few lA/cm2, and the open-circuit voltage
was limited to about 0.2 V [76]. In principle, there is no obvious reason, why
hybrid solar cells based on III–V semiconductors were much less intensively
studied so far than those based on II–VI compounds.
Another class of materials that attracted attention for application in hybrid solar
cells is transition metal oxides, in particular ZnO and TiO2 [77]. These wide band
gap semiconductors will certainly not be able to harvest much photons emitted by
the sun, but nevertheless the materials can be used as efficient electron acceptors in
combination with conductive polymer. In comparison to fullerene acceptors, the
transition metal oxides ZnO and TiO2 have the potential advantage that they can
be easier produced at low cost and large scale. This advantage holds at least as
long as the cost-efficient large-scale production of fullerenes will remain a difficult
task.
12.2 Material Combinations for Hybrid Solar Cells 177
Beek et al. [57, 78, 79] synthesized colloidal ZnO nanocrystals and developed
hybrid solar cells in combination with MDMO-PPV or P3HT as polymer. BHJ
devices with MDMO-PPV/ZnO reached efficiencies up to 1.6 % [78], and devices
with P3HT/ZnO reached about 0.9 % efficiency [79]. In both cases, if compared to
typical polymer/fullerene solar cells, the efficiency seemed to be mainly limited by
the smaller photocurrent. Recently, a BHJ of P3HT and colloidal ZnO nanocrystals
was incorporated into solar cells with different electrode configurations [80].
Devices with a ‘‘normal’’ architecture (layer sequence ITO/PEDOT:PSS/
P3HT:ZnO/Al) were compared to solar cells with inverted device architecture
(layer sequence ITO/ZnO/P3HT:ZnO/PEDOT:PSS/Au). The inverted devices
reached about 1.0 % power conversion efficiency, and, probably as the more
interesting result, exhibited promising stability in aging tests conducted over
2,000 h [80].
Alternatively to the colloidal synthesis approach, P3HT/ZnO bulk heterojunction
layers were also fabricated by in situ synthesis methods, where the ZnO phase is
formed in the presence of the polymer. An impressive work from Janssen’s group
[81] demonstrated that about 2 % efficiency can be reached by this approach, and
electron tomography studies revealed that the in situ formed ZnO phase forms a
highly connected three-dimensional network that should facilitate electron trans-
port. A shortcoming of the P3HT/ZnO system is that both materials have relatively
large energy gaps. Bulk ZnO has a band gap of *3.3 eV, and P3HT has a HOMO–
LUMO gap of *1.9 eV. So, a large fraction of the sunlight cannot be absorbed in
P3HT/ZnO solar cells. A strategy for improvement could be combining ZnO with
conductive polymer with lower band gap. In this view, Oosterhout et al. [82]
recently transferred the in situ synthesis method for interpenetrating networks of
ZnO and P3HT to BHJ layers of ZnO and poly(3-hexylselenophene) which has a
lower HOMO–LUMO gap of *1.7 eV. The corresponding devices reached
*0.4 % power conversion efficiency under standard test conditions, whereas ref-
erence solar cells with the poly(3-hexylseleophene) and PCBM reached *1.5 %
efficiency [82]. Future research on combinations with other low-band gap polymers
might in principle bring improvement to polymer/ZnO solar cells.
In the case of TiO2, Lin et al. [83] prepared hybrid solar cells with a BHJ of
colloidal TiO2 nanorods and P3HT. The nanorods used in that work had a ligand
shell consisting of oleic acid after synthesis, which was removed by ligand exchange
with pyridine. Optionally, the surface of the nanorods was afterwards modified
further by attaching other organic molecules [83]. BHJ solar cells with pyridine-
capped TiO2 nanorods reached *1.1 % efficiency, but by modifying the surface
with an organic dye, the efficiency was successfully doubled to *2.2 % [83].
ZnO and TiO2 are not only of interest for BHJ solar cells. These transition metal
oxides can be grown on different substrates in the form of arrays of elongated
nanostructures with vertical orientation [77, 84–88]. Various processes have been
successfully used to reach such structures, among them for example electro-
chemical methods or thermal evaporation routes. Figure 12.5 shows exemplarily
SEM images and X-ray diffraction data for arrays of vertically aligned ZnO
nanorods electrochemically grown on ITO [89].
178 12 Hybrid Polymer/Nanocrystal Solar Cells
The availability of preparation methods for such aligned nanorod arrays opens
the perspective to fabricate solar cells with an ordered donor/acceptor hetero-
junction, where vertically oriented arrays of the donor and acceptor material would
be interdigitating. In contrast to the disordered BHJ architecture, active layers with
an ordered heterojunction should be favorable for efficient transport of separated
charge carriers to the electrodes. However, it is a big challenge to make the lateral
dimensions of the nanostructures small enough to take account of the short exciton
diffusion length in conductive polymer. The nanostructures presented in Fig. 12.5
were infiltrated with MEH-PPV, followed by the deposition of a Au film as anode.
By this approach, ordered hybrid solar cells with efficiencies up to 0.34 % were
obtained [89]. A good overview on hybrid solar cells with vertically aligned ZnO
nanorod arrays was recently given by Huang et al. [77]. So far, efficiencies of this
type of solar cells did not exceed 1 %. On the other hand, there is still progress to
expect, if the methods for the preparation of such nanostructured arrays and also
the techniques for infiltration of the inorganic structures with conductive polymer
will further improve in future.
Since it remains to date a problem to make the lateral dimensions small enough,
it was also tried to infiltrate transition metal oxide nanoarrays not just with a p-type
conductive polymer, but with a polymer/fullerene blend. In this case, the vertically
12.2 Material Combinations for Hybrid Solar Cells 179
aligned nanorods would still have the role to improve electron transport towards the
cathode. With P3HT/PCBM as donor/acceptor blend, the infiltration of vertically
ordered ZnO or TiO2 nanostructures led to efficiencies up to about 4 % [86, 90].
Although the working principle of organic and hybrid BHJ solar cells is the same,
there are specific differences when looking in more detail at the elementary steps
involved in the light conversion process. The differences basically arise from the
complexity of semiconductor nanocrystal systems as alternative electron accep-
tors. In Sect. 12.1 it was discussed that the most widely used fullerene derivative
PCBM gives a rather poor contribution to the absorption of sun-light. Neverthe-
less, this compound evolved to a well-established standard acceptor material in
organic photovoltaics. So, one may ask for the reason why PCBM is so successful
in OPV despite its poor absorption properties, or, in other words, why alternative
materials with more suitable absorption did finally not lead to an improved overall
efficiency of BHJ solar cells, so far. In the following, the knowledge on elementary
processes in hybrid solar cells will be reviewed and comparison to polymer/
fullerene systems will be made.
The probably most widely accepted model for charge transport in organic semi-
conductor materials is the thermally activated hopping model [107], briefly out-
lined in Sect. 11.1. According to this model, the energy levels of organic
semiconductors present a certain fluctuation, e.g., induced by structural disorder.
More quantitatively this means that the density of states for a given energy level is
broadened and can in the simplest approximation be described by a Gaussian
distribution. The standard deviation, which is also called disorder parameter in this
context, can typically be as large as *100 meV [107]. In a spatial picture this
means that there are localized states with varying energy, and charge transport is
supposed to occur by hopping of the charge carriers between these localized states.
Detailed models have been developed to describe the probability of hopping
between two given sites. Briefly, the hopping rate depends on the spatial distance
and the energy difference between the sites, and also on the temperature [107]. The
latter is obvious, because the probability to hop from a given site to another site
with higher energy will be higher at elevated temperature.
184 12 Hybrid Polymer/Nanocrystal Solar Cells
By the analogy of organic and hybrid solar cells it is tentative to transfer the
charge transport models developed for organic semiconductors directly to hybrid
systems. On the other hand, it is a priori not justified, that a hybrid blend must
behave similar. Here, Ginger and Greenham [108] made a fundamental study of
charge transport in thin films of CdSe nanocrystals deposited between two elec-
trodes. As an important finding, it was observed that the conductivity exhibits an
Arrhenius-like dependency on the temperature in the range of *180–300 K, and it
was concluded that charge transport obeys also in these systems to a thermally
activated hopping mechanism [108]. This suggests that models derived for charge
transport in organic optoelectronic devices may also be transferred to hybrid
devices. Nevertheless, care should be taken, and the applicability of the respective
model should be critically regarded.
In hybrid polymer/nanocrystal blends, it is usually assumed that charge trans-
port within a crystalline nanoparticle is relatively facile, because the semicon-
ductor nanocrystals form energy bands which will be more or less quasi-
continuous, if the particle size is not too small. In contrast, transport from one
nanocrystal to another is more critical, and this is the step supposed to occur by
thermally activated hopping. As opposed to typical organic heterojunctions, e.g.,
polymer/fullerene blends, hybrid systems possess additional parameters which can
affect the hopping probability. Similar as for charge separation, the organic ligand
shell plays also an important role for charge transport, because the thickness of the
ligand shell will have an impact on the interparticle distance. In other words, a
thick insulating ligand shell is supposed to reduce the hopping probability for
charges from one nanocrystal to another. Therefore, the design of the ligand shell
is also of importance for the charge transport step.
As another consequence, charge transport through the nanoparticle phase is
supposed to be dependent on the number of hopping steps which a charge carrier
has to pass from the point of its generation to its extraction at the electrode. In
contrast to organic compounds, the number of steps can be influenced in hybrid
systems by choosing the size and the shape of the nanocrystals. It was already
discussed in Sect. 12.2.1 that using elongated nanocrystals instead of quasi-
spherical quantum dots turned out to be a successful strategy to improve the
performance of hybrid solar cells, the gain being due to improved charge transport
by a reduced number of hopping steps.
In hybrid polymer/nanocrystal solar cells, the inorganic semiconductor nano-
crystals usually act as electron acceptors. Therefore, it is in particular important to
study the electron transport in the nanoparticle phase as well as the impact of the
nanocrystals on hole transport through the polymer phase. An impact on the hole
mobility might for example be expected, if the nanocrystals perturb the molecular
order in the polymer phase. To address such questions, for example, Ginger and
Greenham [108] prepared devices with a layer of TOPO-capped CdSe quantum
dots deposited between ITO and a metal electrode to study charge transport through
the nanoparticle film. The current–voltage characteristics was analyzed with a
space-charge limited current (SCLC) model including deep trap states, and values
for the electron and hole mobility were estimated. Electron mobility was found to
12.3 Elementary Processes in Hybrid Solar Cells and Strategies 185
be in the range of (10-4–10-6) cm2 V-1 s-1, whereas the estimated hole mobility
was very low (*10-12 cm2 V-1 s-1) [108]. This shows that CdSe quantum dots
are a suitable material for electron transport. Later, Kumari et al. [109, 110]
investigated hole transport in hole-only devices with a blend of P3HT and pyridine-
capped CdSe nanocrystals placed between ITO/PEDOT:PSS and Au electrodes.
Again, an SCLC model with traps was used to model the current–voltage charac-
teristics, and the extracted hole mobility was reported to increase from
*3 9 10-6 cm2 V-1 s-1 for pristine P3HT films to *8 9 10-6 cm2 V-1 s-1 for
hybrid P3HT/CdSe blends [110]. Thus, the addition of nanocrystals to the polymer
film did not have a detrimental effect on the hole mobility, but resulted even in a
slight improvement. A similar trend was also observed in a study of hole transport
in organic field effect transistors (OFETs): Here, the hole mobility of pure P3HT
was measured to be 1.4 9 10-4 cm2 V-1 s-1, whereas a value of
2.3 9 10-4 cm2 V-1 s-1 was obtained for P3HT/CdSe blends [111]. The order of
the absolute values mentioned here is in reasonable agreement with mobility
measurements in OFETs performed for P3HT/PCBM blends [112].
Talapin and Murray [113] studied electron and hole transport in thin films of
PbSe nanocrystals in field effect transistors. The usage of oleic acid as capping
ligand resulted in an interparticle distance of (1.1–1.5) nm which in turn led to
very poor conductivity. However, a chemical treatment of the samples with
hydrazine removed the bulky oleic acid ligands to a large extent and reduced the
interparticle distance. By consequence, good n-type conductivity was achieved
with the electron mobility as high as 0.9 cm2 V-1 s-1 [113]. Furthermore, by
applying vacuum or heat treatments, it was possible to switch the conductivity to
p-type, with the hole mobility reaching values of 0.2 cm2 V-1 s-1 [113]. This
example shows how critical the ligand shell can be for charge transport through a
network of colloidal nanocrystals, and that appropriate procedures to treat the
surface of the nanocrystals are a powerful strategy to reach thin films with good
conductivity. Moreover, the example demonstrates that colloidal nanocrystals can
in some cases be tuned to show either n-type or p-type behavior.
Thin polymer films and also organic polymer/fullerene films can possess a variety
of defects [114], e.g., induced by structural disorder or by impurities. In hybrid
blends, the inorganic semiconductor nanocrystals are an additional source of
defects. In fact, many different types of defects can occur in small nanocrystals.
For example, structural defects like vacancies, interstitial atoms or impurities can
be present in the volume of the nanoparticles. Additionally, the surface of semi-
conductor nanocrystals is a source of defects. As already mentioned in Sect. 2.4,
from an energetic point of view, surface atoms with dangling bonds are in many
186 12 Hybrid Polymer/Nanocrystal Solar Cells
cases associated with energy levels located rather deep inside the energy gap
[115–117]. For example, the energy levels of In and P dangling bonds at the
surface of InP nanocrystals were investigated in detail by a theoretical approach
[115]. For In dangling bonds, a strong size-dependence was observed. The cor-
responding defect level was found to be located below the conduction band
minimum, if the particle diameter gets smaller than 5.7 nm. The defect level
becomes deeper with decreasing particle size and can be as deep as *0.7 eV for
small InP nanocrystals with a diameter of *1.5 nm [115]. For the P dangling
bonds, the size dependence turned out to be less pronounced, but also in this case
rather deep defect states were predicted, with the states being located *0.7 eV
above the valence band maximum for small InP quantum dots [115].
Defect levels can act as traps for charge carriers in the system and will therefore
be of relevance for the fundamental process of charge transport in the material. For
example, Schafferhans et al. [118] investigated oxygen-induced trap states in P3HT
and could show that the charge carrier mobility as deduced from measurements by
photocharge extraction by linearly increasing voltage (photo-CELIV) strongly
decreased with increasing density of deep trap states. Furthermore, defect levels can
act as recombination centers for charge carriers. In general, a high density of deep
defect levels will be detrimental for the performance of optoelectronic devices.
Therefore, it is crucial to develop strategies to avoid deep levels in hybrid blends. In
the following, the knowledge recently obtained on the influence of defects in hybrid
solar cells will be reviewed, and current strategies to avoid high defect concentra-
tions will be discussed.
Useful information about defects in polymer-based films can be obtained from
recombination studies, e.g., in frequency-dependent PIA measurements (see Sect. 8.
3) or time-dependent L-ESR measurements (see Sect. 9.2). In Fig. 9.4, the recom-
bination of photo-generated polarons was shown after switching off the laser
excitation source in an L-ESR experiment, and strongly different decay kinetics was
observed for organic P3HT/PCBM and hybrid P3HT/CdSe blends. The observation
of an extremely slow recombination process in the case of the hybrid blends sug-
gested that more and deeper trap states must be present in this system (see also
discussion in Sect. 9.2) [13]. The finding that the hybrid blend contains deeper traps
for charge carriers constitutes a quite fundamental difference between organic and
hybrid solar cells which is likely to contribute to the inferior device performance
despite the potential advantages discussed before. It is intuitive that deep trap levels
will for example hinder efficient charge transport, because the energetic barriers
encountered in the hopping mechanism will become considerably larger. Such an
effect can also be observed experimentally. For example, Kumari et al. [109, 110]
investigated hole-only devices based on P3HT and P3HT/CdSe. The current–volt-
age characteristics was evaluated with an SCLC model including an exponential
distribution of traps in energy and space, and more and deeper trap states were
shown to lower the charge carrier mobility [109, 110].
The presence of deep traps in blends for hybrid solar cells was also evidenced in
PIA studies [13, 103]. For example, Ginger and Greenham [103] studied blends of
MEH-PPV and pyridine-capped CdSe quantum dots by PIA spectroscopy and
12.3 Elementary Processes in Hybrid Solar Cells and Strategies 187
found a broad distribution for the lifetime of positive polarons. From the analysis,
it was concluded that part of the polarons had a lifetime below 100 ls, whereas
another fraction was very long-lived with the lifetimes extending to several mil-
liseconds. Furthermore, some polarons were reported to persist even at room
temperature [103]. This is well consistent with the observation that P3HT/CdSe
blends need to be annealed above room temperature to remove the persistent signal
observable in L-ESR studies [13] (compare also discussion in Sect. 9.2). PIA
measurements of hybrid P3HT/CdSe blends were done by Heinemann et al. [13]
and directly compared to an organic P3HT/PCBM blend as reference system.
Figure 12.6 shows the obtained frequency-dependent data [13].
For the organic blend, the signal saturates at low modulation frequency, indi-
cating that there are no polarons with lifetimes exceeding the inverse of the lowest
applied modulation frequency. For the hybrid blend, saturation was not observed
in the low frequency regime. This already indicates that species with longer
lifetime are present in the hybrid system. To reach more quantitative conclusions,
the experimental data was fitted with the help of established recombination models
[13]. The used bimolecular recombination model, well described for example in
[104], assumes that positive and negative polarons are both mobile, so that the
recombination rate will be proportional to the concentrations of both types of
charge carriers, similar as in the kinetics of second order for a bimolecular
chemical reaction. For the P3HT/PCBM blends, the fit is not perfect, but can
reasonably approximate the experimentally observed behavior (see Fig. 12.6).
This indicates that both electron and hole polarons are relatively mobile and can
easily recombine [13].
In contrast, the bimolecular model completely fails to explain the recombina-
tion behavior in the hybrid blend. This suggests that charge carrier trapping plays
an important role in this system. The presence of deep trap states can lead to
capturing of part of the charge carriers. In that case, their deliberation from the
traps can become the rate-limiting step for the recombination. Such a situation can
for example be described by a so-called dispersive recombination model. The
frequency dependency of the PIA signal should then obey to the following
equation [13, 75]:
DT ðDT=T Þ0
¼ ð12:1Þ
T 1 þ ðxsÞc
Therein, (DT/T)0 is the intensity of the PIA signal in the limit of zero frequency,
x is the modulation frequency, s is the mean lifetime, and c is a parameter between
0 and 1. Values of c close to 1 correspond to a situation where trapping plays nearly
no role, whereas decreasing values of this parameter indicate the increasing
influence of traps states. Concerning the lifetime, it must be mentioned, that the
fitting parameter s reflects just an average value, although the ensemble can have a
broad distribution of lifetimes. The model described here turned out suitable to fit
the frequency dependence observed for the hybrid P3HT/CdSe system, and also for
the organic blend (see Fig. 12.6). The parameter c was fitted to 0.55 for P3HT/CdSe
188 12 Hybrid Polymer/Nanocrystal Solar Cells
and 0.98 for P3HT/PCBM. This supports the conclusion that traps play a minor role
in the organic blend as compared to the hybrid system. The values obtained for the
mean lifetimes were 1.25 ms for the organic blend, and 60.7 ms for the hybrid
blend at T = 80 K, respectively [13]. The lifetime measured for the hybrid system
appears rather large, but is in reasonable agreement with the observation of a large
distribution of lifetimes in the MEH-PPV/CdSe system mentioned above [103].
In conclusion, there is evidence from long recombination times observed in PIA
and L-ESR measurements that blends of conductive polymer and pyridine-capped
CdSe quantum dots contain deep trap states which can capture part of the charge
carriers. The questions arise what the physical origin of the trap states is, how deep
they are, and if they can be reduced.
Recent studies provided evidence that the surface coverage by the organic
ligands plays an important role with respect to trap states. As explained in Sect. 2.2
, typical ligands that are used in colloidal synthesis are molecules with a functional
group and relatively long hydrocarbon chains (typically *6–20 carbon atoms in
the chain). The resulting, relatively thick ligand shells are favorable to reach a
good stabilization of the nanocrystals in colloidal solution and to prevent aggre-
gation of the particles. However, such compact ligand shells are not suitable for
charge transfer, and ligand exchange with pyridine was established as a suitable
way to replace the thick insulating shells originating from the synthesis with a thin
12.3 Elementary Processes in Hybrid Solar Cells and Strategies 189
Fig. 12.7 TEM images of P3HT/CdSe films as they were used in hybrid BHJ solar cells. The
film preparation parameters (blending ratio, annealing temperature, etc.) was similar for all
samples, but the CdSe nanocrystals initially stabilized with oleic acid were treated one time (a),
two times (b) and three times (c) with pyridine (Reprinted with permission from [121]. Copyright
2010 American Chemical Society)
ligand shell, still providing solubility, but also enabling the charge transfer
process.
A question which is neglected in most photovoltaic studies concerns the effi-
ciency of this ligand exchange procedure. In the 1990s, a few fundamental studies
addressed this question and revealed that even after repeated ligand exchange with
pyridine typically *10–15 % of the surface sites remain capped by the original
ligands in the case of CdSe quantum dots initially stabilized with tributylphosphine
and trioctylphoshine oxide (TBP/TOPO) [119] and TOP/TOPO [120], respec-
tively. However, the impact of incomplete ligand exchange on the performance of
CdSe-based hybrid solar cells remained uninvestigated until 2010, when Lokteva
et al. [121] studied the influence of multiple pyridine treatments of CdSe nano-
crystals initially stabilized with oleic acid. After a single treatment with pyridine,
*26 % of the Cd surface sites were estimated to be still capped by species from
the original ligand shell, and *54 % of the Cd sites were capped with pyridine.
Repetition of the ligand exchange procedure helped to lower the rests of the
original ligand shell, but even after 3 cycles, *12 % of the Cd sites were still
capped with original ligands, while pyridine occupied up to *80 % of the sites
[121]. This shows that it is not evident to exchange ligand shells completely.
Nevertheless, an improvement of the corresponding solar cells might be expected
with quantum dots, where the ligand exchange is more complete. However, the
cited study revealed that the hybrid solar cells based on particles subjected to only
one ligand exchange step showed the best performance [121]. As the main reason,
it was found that the number ligand exchange cycles increases the tendency of the
nanocrystals to form large aggregates in the polymer matrix. Figure 12.7 shows
TEM images of P3HT/CdSe films as used in hybrid solar cells based on CdSe
quantum dots subjected one, two and three times to ligand exchange with pyridine,
respectively [121].
The example shows that the ligand shell can have a strong impact on mor-
phology. This issue was also addressed in a study by Olson et al. [35] who
190 12 Hybrid Polymer/Nanocrystal Solar Cells
Fig. 12.8 Current density–voltage characteristics of a typical BHJ hybrid solar cell based on
P3HT and colloidal CdSe nanocrystals stabilized with butylamine under illumination with
simulated sun-light (AM 1.5G radiation) at 100 mW/cm2 illumination intensity. The insets show
a schematic drawing of the device architecture as well as a photograph of three solar cells
prepared on one substrate. In the scheme, the layers are: Al cathode 1, active layer 2,
PEDOT:PSS layer 3, ITO 4, and glass 5 (Reprinted with permission from [14]. Copyright 2011
American Chemical Society)
prepared hybrid solar cells with P3HT and CdSe quantum dots capped by a series
of different ligands. In detail, the nanocrystals were initially stabilized by TOP and
treated once with pyridine. In a second step, ligand exchange was done with other
compounds, namely butylamine, tributylamine, stearic acid, and oleic acid. The
morphology of the blends as studied by AFM measurements and the electrical
characteristics of the solar cells exhibited significant differences. The best results
were obtained with butylamine ligands where efficiencies up to *1.8 % were
reached [35]. This was at that time a new record for hybrid solar cells with quasi-
spherical CdSe nanocrystals, and the improvement was attributed to improved
morphology as compared to the case of using pyridine as the final ligand [35].
Later, Radychev et al. [14] investigated the impact of using butylamine instead
of pyridine and the physical origin of the strong improvement in efficiency in more
detail. Figure 12.8 shows a current–voltage curve for a hybrid solar cells based on
P3HT and CdSe quantum dots capped with butylamine, and Fig. 12.9 compares
the external quantum efficiency for typical solar cells based on ligand exchange
with pyridine and butylamine, respectively [14].
Comparing the electrical characteristics shown in Fig. 12.8 to results published
for hybrid solar cells with pyridine-capped, quasi-spherical CdSe quantum dots
[12, 16], the improved performance can be attributed mainly to a higher current
density. The analysis of EQE spectra shown in Fig. 12.9 helps to understand where
the additional current originates from. Part (b) of the figure shows the difference in
the EQE for the pyridine- and butylamine-based systems. The shape of this dif-
ference curve does not resemble the absorption spectrum of P3HT. So, more
efficient current generation, e.g., by a more fine morphology reducing recombi-
nation losses due to the short exciton diffusion length in the polymer is an unlikely
explanation for the improved performance [14]. In contrast, the difference plot
12.3 Elementary Processes in Hybrid Solar Cells and Strategies 191
Fig. 12.9 Part a shows EQE spectra of BHJ P3HT/CdSe solar cells based on CdSe quantum dots
with pyridine and butylamine as ligands, respectively. Part b shows absorption spectra (left axis)
of a pure P3HT film (solid black line) and colloidal solutions of pyridine-capped CdSe
nanocrystals (blue line) and butylamine-capped CdSe nanocrystals (red line). Additionally, the
difference of the EQE spectra (obtained by simply subtracting the curves in (a) from each other
point by point) is shown (dashed black line, right axis) (Reprinted with permission from [14].
Copyright 2011 American Chemical Society)
exhibits a shape which is more similar to the absorption spectra of the nano-
crystals, but with a slight red-shift. Since the solar cells were annealed at 180 C, it
is likely that the particle size slightly increased due to sintering processes, so that
the red shift with respect to the absorption spectra of pure nanocrystals in colloidal
solution is not surprising. Thus, the analysis of the EQE data suggests that the
additional photocurrent is mainly due to more efficient conversion of light that is
absorbed in the nanocrystals [14].
This leads to the question how the ligand shell can influence on the fate of
excitons created by light absorption in the nanoparticles. A piece of evidence was
provided by L-ESR spectroscopy. Figure 12.10 compares for different samples and
temperatures the time-dependent decay of light-induced ESR signals corre-
sponding to polarons in the polymer after switching off the excitation source. It can
be seen that a fast decay indicative for mobile charge carriers which can easily
recombine is more pronounced in the butylamine-based system (compare the
decay in the first few seconds after switching off the laser source). This provides
already qualitatively a hint for charge carrier trapping to be less pronounced in the
butylamine-based system as compared to the use of pyridine as ligand.
192 12 Hybrid Polymer/Nanocrystal Solar Cells
Fig. 12.10 Decay of the light-induced ESR signal corresponding to positive polarons in P3HT
after switching off (t = 0 s) the excitation source (a 532 nm laser) for thin films of P3HT blended
with butylamine-capped CdSe quantum dots (a) and pyridine-capped quantum dots (b). The black
decay curves were obtained at T = 237 K, whereas the red curves were measured at T = 276 K.
The insets show ESR spectra after 30 min of continuous laser illumination (red curves) and
25 min after switching off the excitation source (black curves). The green curves represent an
attempt to fit the curves as described in [14]. (Reprinted with permission from [14]. Copyright
2011 American Chemical Society)
Fig. 12.11 Qualitative molecular orbital diagrams describing bonding interactions between
electron-poor Cd2+ surface sites on quantum dots (QDs) and r-donating ligands. a An empty
Cd2+ surface site results in a state within the energy gap that can trap electrons. b A strong
r-donating ligand eliminates the electron-trapping midgap state by binding to the Cd2+ site and
forming an antibonding orbital that is higher in energy than the LUMO (or conduction band
minimum) of the QD. The corresponding bonding orbital does not form a hole-trapping midgap
state because it is lower in energy than the HOMO (or valence band maximum) of the QD. c A
weak r-donating ligand increases the energy of the electron-trapping midgap state over that of
bare Cd2+, but does not eliminate the trap because it forms an antibonding orbital that is lower in
energy than the LUMO of the QD (Reprinted with permission from [117]. Copyright 2010
American Chemical Society)
anilines. They described the effect of ligands on the surface states as qualitatively
illustrated in Fig. 12.11: In the absence of ligands, Cd and Se dangling bonds will
lead to energy levels located inside the energy gap [part (a) of the figure]. If a
ligand is added, the unoccupied surface dangling bond orbitals can overlap with
filled orbitals of the ligand (e.g., the lone electron pair located at the nitrogen atom
of an amine). According to the molecular orbital (MO) theory, two molecular
orbitals will result [parts (b) and (c) of the figure], with the antibonding MO
remaining unoccupied. The unoccupied MO can still be an electron trap, if it is still
located inside the energy gap of the semiconductor. So, the removal of the trap
states by passivating it with a surfactant depends on the nature of the ligand. A
strong r-donating ligand will raise the antibonding MO above the conduction band
minimum and thus remove the trap [part (b) of the figure], whereas a shallow trap
can persist in the case of a weaker r-donating ligand [part (c) of the figure] [117].
According to this qualitative consideration, it appears possible that the different
activation energies found for the traps in the butylamine and pyridine-based sys-
tems (see discussion above) are related to different capability of both ligands to
raise the trap level towards the conduction band minimum. However, to verify this
interpretation, it would be necessary to have theoretical calculations predicting the
energetic positions of the antibonding MOs with respect to the conduction band
minimum for the materials in question. Such calculations are absent, so far.
According to [117], another alkylamine, namely hexadecylamine, should raise the
level above the conduction band minimum, i.e., remove the trap associated with
Cd dangling bonds.
Apart from electron traps related to the passivation of Cd dangling bonds, it is
of high importance to note that the butylamine and pyridine-based systems are also
supposed to possess different trap states for holes. In fact, it is known for more than
ten years that pyridine can accept holes, because the positive charge can be
194 12 Hybrid Polymer/Nanocrystal Solar Cells
stabilized on the aromatic ring [14, 116, 123, 124]. This means that in addition to
hole traps present for example because of Se dangling bonds, the pyridine ligand
itself is likely to induce additional hole traps into the system. This should have
consequences for the fate of excitons created by light absorption in the nanocrystal
phase. If the hole of such an exciton gets trapped at a localized state at the
nanocrystal, and if the corresponding state is not in close contact with the polymer
phase (for example, because the state is located at the inner surface of a small
nanocrystal aggregate), then it will be unlikely for the hole to be transferred to the
polymer [14]. Thus, the additional hole traps introduced by pyridine as ligand
reduce the probability for excitons created in the nanoparticles to be split at the
donor/acceptor interface. Removal of these traps by using another ligand than
pyridine is probably the main reason, why light absorption by the nanocrystals
contributes more efficiently to the photocurrent in P3HT/CdSe solar cells when
butylamine is used as ligand (compare discussion of the EQE data above) [14].
In conclusion, the discussion in the preceding paragraphs shows that the usage
of different capping ligands is a promising strategy to improve the performance of
hybrid solar cells. The surface of the inorganic semiconductor nanocrystals is a
source of additional trap states which do not exist in the organic polymer/fullerene
counterparts. Therefore, the efficient passivation of these traps is crucial, if one
wants to benefit from the potential advantages of inorganic nanocrystals as
acceptors in future. Different ligands have a different capability to reach a high
degree of surface coverage and to eliminate the traps or at least to make them
shallow. Some ligand molecules, on the other hand, can also introduce a new type
of traps into the system. In addition, different ligands will lead to different solu-
bility properties of the nanocrystals which in turn can have a strong impact on the
morphology of the polymer/nanocrystal active layers. Thus, the ligand shell plays
a crucial role in view of many aspects of the hybrid solar cells, and future research
will have to concentrate more on understanding the impact of different ligands and
optimization strategies.
This section has focused so far mainly on ESR and PIA spectroscopy as useful
methods to study recombination processes, which in turn enables conclusions on
charge carrier trapping effects. There are, of course, more methods that can pro-
vide information on trap states. For example, charge-based deep level transient
spectroscopy (Q-DLTS) was applied to study traps in light-emitting diodes based
on hybrid blends of MEH-PPV and CdSe/ZnS core–shell nanocrystals [125]. In
that study, five different trap states were observed, and it was possible to determine
apart from the trap depth also the capture cross-section and the concentration of
the traps [125]. Since the nature of the hybrid blends in polymer/nanocrystal LEDs
and solar cells is similar, valuable information might be expected from similar
studies of hybrid solar cells. However, Q-DLTS investigations of polymer/nano-
crystal solar cells were not reported, yet.
Hybrid solar cells based on CdSe nanocrystals are probably the most intensively
studied material system in the field, so far. But there are also valuable reports on
recombination processes and defects in other hybrid systems. For example, the
device physics of polymer/ZnO hybrid solar cells was investigated in much detail
12.3 Elementary Processes in Hybrid Solar Cells and Strategies 195
as well [57, 78, 79, 126]. ZnO nanocrystals as used in hybrid solar cells are often
prepared based on a method described in detail by Pacholski et al. [127] which
uses a reaction of zinc acetate dihydrate and KOH in methanol as solvent. In this
synthesis route no long-chained ligand molecules are present, so that a ligand
exchange is usually not applied. Beek et al. [78] investigated the effect of adding
propylamine as a ligand. The solar cell performance did not improve, but the
surfactant was reported to have a positive effect on the morphology of the thin
films in the sense that the films became more smooth [78]. Later, Park et al. [128]
modified the surface of colloidal ZnO nanocrystals with oleic acid or a newly
developed semiconducting dicarboxylic acid. Both molecules improved the dis-
persion of the ZnO nanoparticles in the hybrid blends with MDMO-PPV, but the
insulating oleic acid ligand shell decreased the photocurrent and power conversion
efficiency, whereas the semiconducting dicarboxylic acid significantly increased
the device performance [128]. Thus, further research on different ligands in
polymer/ZnO solar cells might also bring advances to this material system in
future.
The widely used approach to perform a ligand exchange in order to replace a thick
organic ligand shell resulting from the synthesis by a thinner shell enabling charge
transfer has also some disadvantages. First, ligand exchange is usually not com-
plete [119–121]. How effective a ligand exchange procedure really is, will depend
on the material system and has in principle to be studied in detail for each case.
Another inconvenience is that ligand exchange with relatively small molecules can
also cause a stronger tendency of the nanocrystals to form large aggregates [121].
Of course, this depends also on the material system, including the choice of the
ligand. Finally, not all ligands will provide solubility in solvents that are suitable to
dissolve the polymer. For example, the active layer of hybrid polymer/CdSe solar
cells is usually processed from binary solvent mixtures of chloroform/pyridine or
chlorobenzene/pyridine, if the ligand shell of the nanocrystals is replaced by
pyridine. Since the morphology of the films strongly depends on solvent ratio [23],
this introduces another parameter into the film preparation which has to be opti-
mized and therefore makes the process more and more complicated.
An alternative approach to classical ligand exchange was introduced for
polymer/CdSe solar cells by Zhou et al. [36]. In that work, quasi-spherical CdSe
nanocrystals were prepared by colloidal chemistry using hexadecylamine (HDA)
as stabilizer. Instead of performing a ligand exchange, e.g., with pyridine, the
authors treated the nanoparticles in solution with hexanoic acid. The acid was
supposed to react with part of the HDA ligands and to form an organic salt which
can easily be removed by centrifugation [36]. The acid-treated nanocrystals were
196 12 Hybrid Polymer/Nanocrystal Solar Cells
reported to have a reduced ligand shell, but still a high solubility in orthodichlo-
robenzene (oDCB) which was used as solvent to prepare blends with P3HT for
solar cells. The efficiency of the solar cells reached 2.0 % under standard test
conditions [36]. This is approximately twice the typical efficiency of hybrid solar
cells with pyridine-capped, quasispherical CdSe nanocrystals and also as high as
the efficiency achieved in the case of butylamine as ligand. The study shows that
such acid treatments are an alternative to classical ligand exchange procedures.
A completely different approach is so-called in situ synthesis methods. Instead
of synthesizing nanocrystals by colloidal chemistry which are then blended in a
second step with conductive polymer, it is also possible to synthesize inorganic
compounds directly in a polymer matrix. This strategy was for example investi-
gated in the case of polymer/ZnO hybrid solar cells [81, 82]. Oosterhout et al. [81]
prepared a solution containing P3HT and diethylzinc as the zinc precursor, which
was used to cover an ITO/PEDOT:PSS electrode by spin-coating under conditions
of controlled humidity. Zn(OH)2 was reported to form which can finally be con-
verted to ZnO by annealing to 100 C [81]. The interesting point is that the ZnO
phase forms as a connected inorganic network which is interpenetrating with the
P3HT phase. The three-dimensional structure of the obtained network was studied
in much detail by electron tomography [81]. The reconstructed three-dimensional
objects allowed obtaining a good impression of the 3D-morphology, and advanced
statistical methods to evaluate the images enabled even the extraction of detailed
quantitative information. For example, the probability to have a polymer domain at
a certain distance of a given ZnO domain was determined [81]. The solar cells
prepared by this in situ approach reached 2.0 % device efficiency [81] which is
superior to the performance of BHJ P3HT/ZnO solar cells prepared by the classical
route involving colloidal synthesis of ZnO nanocrystals as the first step. Thus, the
study shows that in situ synthesis of inorganic semiconductor networks is a
promising alternative. Furthermore, it must be emphasized that the cited work also
demonstrates that electron tomography is a method which can provide information
on the three-dimensional morphology of active layers in unique detail. In partic-
ular, this is possible, because in hybrid systems, there is good contrast in the TEM
images between the organic and inorganic materials, respectively.
In a later work, the in situ method was extended to the preparation of ZnO in
poly(3-hexylselenophene) with the intention to improve the absorption of sunlight,
because this polymer has a lower energy gap than poly(3-hexylthiophene) [82].
However, an improvement was finally not obtained [82]. In situ synthesis approa-
ches were also reported for other inorganic semiconductor materials. For example,
Maier et al. [72] synthesized CuInS2 in a matrix of poly(3-((ethyl-4-butano-
ate)thiophene) by spin-coating a solution containing Cu, In and S precursors as well
as the polymer on ITO substrates, and annealing the films to 180 C under vacuum.
Solar cells prepared within that work reached up to *0.4 % efficiency [72]. With
other precursors and choosing poly[(2,7-silafluorene)-alt-(4,7-di-2-thienyl-2,1,3-
benzothiadiazole)] as conductive polymer, hybrid polymer/CuInS2 solar cells
reached even 2.8 % power conversion efficiency [73]. This is at date the by far
12.3 Elementary Processes in Hybrid Solar Cells and Strategies 197
highest efficiency for polymer/CuInS2 solar cells and demonstrates again that in situ
synthesis is a promising alternative to the colloidal chemistry approach.
It should be mentioned, however, that the in situ methods can also have some
disadvantages. Proponents of the in situ approach consider it advantageous that no
additional surfactants are necessary in that type of synthesis, so that no barriers for
charge transfer are introduced. On the other hand, the discussion in Sect. 12.3.3
has shown that ligands are also of importance in the sense that they can eliminate
defect states related to the nanocrystal surface. So, there will probably be some
trade-off between positive and negative effects of ligands. A real advantage of
in situ formed networks is certainly the high degree of connectivity of the inor-
ganic domains, as it was evidenced by the electron tomography studies reported in
[81]. On the other hand, the in situ methods reported so far enable less control over
the structural properties of the inorganic semiconductor materials than the col-
loidal chemistry approach. Thus, it is not clear so far which approach will finally
offer better possibilities to control the morphology of the active layer in the most
suitable manner.
References
1. S. Cook, H. Ohkita, Y. Kim, J.J. Benson-Smith, D.D.C. Bradley, J.R. Durrant, Chem. Phys.
Lett. 445, 276 (2007)
2. S. Cook, R. Katoh, A. Furube, J. Phys. Chem. C 113, 2547 (2009)
3. J.W. Jeong, J.W. Huh, J.I. Lee, H.Y. Chu, J.J. Pak, B.K. Ju, Thin Solid Films 518, 6343
(2010)
4. M.C. Scharber, D. Mühlbacher, M. Koppe, P. Denk, C. Waldlauf, A.J. Heeger, C.J. Brabec,
Adv. Mater. 18, 789 (2006)
5. W.W. Yu, L. Qu, W. Guo, X. Peng, Chem. Mater. 15, 2854 (2003)
6. Q. Dai, Y. Wang, X. Li, Y. Zhang, D.J. Pellegrino, M. Zhao, B. Zou, J. Seo, Y. Wang, W.W.
Yu, ACS Nano 3, 1518 (2009)
7. S. Adam, D.V. Talapin, H. Borchert, A. Lobo, C. McGinley, A.R.B. de Castro, M. Haase, H.
Weller, T. Möller, J. Chem. Phys. 123, 084706 (2005)
8. H. Zhong, S.S. Lo, T. Mirkovic, Y. Li, Y. Ding, Y. Li, G.D. Scholes, ACS Nano 4, 5253
(2010)
9. S. Peng, Y. Liang, F. Cheng, J. Liang, Sci. China Chem. 55, 1236 (2012)
10. I. Moreels, K. Lambert, D. Smeets, D. De Muynck, T. Nollet, J.C. Martins, F. Vanhaecke,
A. Vantomme, C. Delerue, G. Allan, Z. Hens, ACS Nano 3, 3023 (2009)
11. J. Jasieniak, L. Smith, J. van Embden, P. Mulvaney, M. Califano, J. Phys. Chem. C 113,
19468 (2009)
12. F. Zutz, I. Lokteva, N. Radychev, J. Kolny-Olesiak, I. Riedel, H. Borchert, J. Parisi, Phys.
Status Solidi A 206, 2700 (2009)
13. M.D. Heinemann, K. von Maydell, F. Zutz, J. Kolny-Olesiak, H. Borchert, I. Riedel, J.
Parisi, Adv. Funct. Mater. 19, 3788 (2009)
14. N. Radychev, I. Lokteva, F. Witt, J. Kolny-Olesiak, H. Borchert, J. Parisi, J. Phys. Chem. C
115, 14111 (2011)
15. N.C. Greenham, X. Peng, A.P. Alivisatos, Phys. Rev. B 54, 17628 (1996)
16. L. Han, D. Qin, X. Jiang, Y. Liu, L. Wang, J. Chen, Y. Cao, Nanotechnology 17, 4736
(2006)
198 12 Hybrid Polymer/Nanocrystal Solar Cells
17. W.U. Huynh, J.J. Dittmer, A.P. Alivisatos, Science 295, 2425 (2002)
18. B. Sun, H.J. Snaith, A.S. Dhoot, S. Westenhoff, N.C. Greenham, J. Appl. Phys. 97, 014914
(2005)
19. L.-W. Wang, A. Zunger, Phys. Rev. B 53, 9579 (1996)
20. J.-E. Brandenburg, X. Jin, M. Kruszynska, J. Ohland, J. Kolny-Olesiak, I. Riedel, H.
Borchert, J. Parisi, J. Appl. Phys. 110, 064509 (2011)
21. W.U. Huynh, X. Peng, A.P. Alivisatos, Adv. Mater. 11, 923 (1999)
22. B. Sun, E. Marx, N.C. Greenham, Nano Lett. 3, 961 (2003)
23. W.U. Huynh, J.J. Dittmer, W.C. Libby, G.L. Whitting, A.P. Alivisatos, Adv. Funct. Mater.
13, 73 (2003)
24. Y. Zhou, Y. Li, H. Zhong, J. Hou, Y. Ding, C. Yang, Y. Li, Nanotechnology 17, 4041
(2006)
25. I. Gur, N.A. Fromer, C.-P. Chen, A.G. Kanaras, A.P. Alivisatos, Nano Lett. 7, 409 (2007)
26. S. Dayal, N. Kopidakis, D.C. Olson, D.S. Ginley, G. Rumbles, Nano Lett. 10, 239 (2010)
27. M. Wright, A. Uddin, Sol. Energy Mater. Sol. Cells 107, 87 (2012)
28. D. Celik, M. Krüger, C. Veit, H.F. Schleiermacher, B. Zimmermann, S. Allard, I. Dumsch,
U. Scherf, F. Rauscher, P. Niyamakom, Sol. Energy Mater. Sol. Cells 98, 433 (2012)
29. K. Jeltsch, M. Schädel, J.-B. Bonekamp, P. Niyamakom, F. Rauscher, H.W.A. Lademann, I.
Dumsch, S. Allard, U. Scherf, K. Meerholz, Adv. Funct. Mater. 22, 397 (2012)
30. R. Zhou, R. Stalder, D. Xie, W. Cao, Y. Zheng, Y. Yang, M. Plaisant, P.H. Holloway, K.S.
Schanze, J.R. Reynolds, J. Xue, ACS Nano 7, 4846 (2013)
31. J. Hou, T.L. Chen, S. Zhang, H.-Y. Chen, Y. Yang, J. Phys. Chem. C 113, 1601 (2009)
32. D. Mühlbacher, M.C. Scharber, M. Morana, Z. Zhu, D. Waller, R. Gaudiana, C.J. Brabec,
Adv. Mater. 18, 2884 (2006)
33. Y. Zhou, M. Eck, C. Veit, B. Zimmermann, F. Rauscher, P. Niyamakom, S. Yilmaz, I.
Dumsch, S. Allard, U. Scherf, M. Krüger, Sol. Energy Mater. Sol. Cells 95, 1232 (2011)
34. J. Jasieniak, M. Califano, S.E. Watkins, ACS Nano 5, 5888 (2011)
35. J.D. Olson, G.P. Gray, S.A. Carter, Sol. Energy Mater. Sol. Cells 93, 519 (2009)
36. Y. Zhou, F.S. Riehle, Y. Yuan, H.-F. Schleiermacher, M. Niggemann, G.A. Urban, M.
Krüger, Appl. Phys. Lett. 96, 013304 (2010)
37. L. Wang, Y. Liu, X. Jiang, D. Qin, Y. Cao, J. Phys. Chem. C 111, 9538 (2007)
38. S. Ren, L.-Y. Chang, S.-K. Lim, J. Zhao, M. Smith, N. Zhao, V. Bulovic, M. Bawendi, S.
Gradecak, Nano Lett. 11, 3998 (2011)
39. J.Y. Kim, I.J. Chung, Y.C. Kim, J.-W. Yu, J. Korean Phys. Soc. 45, 231 (2004)
40. I. Gur, N.A. Fromer, A.P. Alivisatos, J. Phys. Chem. B 110, 25543 (2006)
41. P.T.K. Chin, J.W. Stouwdam, S.S. van Bavel, R.A.J. Janssen, Nanotechnology 19, 205602
(2008)
42. W. Yu, H. Zhang, Z. Fan, J. Zhang, H. Wei, D. Zhou, B. Xu, F. Li, W. Tian, B. Yang,
Energy Environ. Sci. 4, 2831 (2011)
43. Y. Kang, N.-G. Park, D. Kim, Appl. Phys. Lett. 86, 113101 (2005)
44. H.-C. Chen, C.-W. Lai, I.-C. Wu, H.-R. Pan, I.-W.P. Chen, Y.-K. Peng, C.-L. Liu, C.-H.
Chen, P.-T. Chou, Adv. Mater. 23, 5451 (2011)
45. D. Qi, M. Fischbein, M. Drndic, S. Selmic, Appl. Phys. Lett. 86, 093103 (2005)
46. S.A. McDonald, G. Konstantatos, S. Zhang, P.W. Cyr, E.J.D. Klem, L. Levina, E.H.
Sargent, Nat. Mater. 4, 138 (2005)
47. S. Zhang, P.W. Cyr, S.A. McDonald, G. Konstantatos, E.H. Sargent, Appl. Phys. Lett. 87,
233101 (2005)
48. A.A.R. Watt, D. Blake, J.H. Warner, E.A. Thomsen, E.L. Tavenner, H. Rubinsztein-
Dunlop, P. Meredith, J. Phys. D Appl. Phys. 38, 2006 (2005)
49. X. Jiang, R.D. Schaller, S.B. Lee, J.M. Pietryga, V.I. Klimov, A.A. Zakhidov, J. Mater. Res.
22, 2204 (2007)
50. J. Seo, S.J. Kim, W.J. Kim, R. Singh, M. Samoc, A.N. Cartwright, P.N. Prasad,
Nanotechnology 20, 095202 (2009)
References 199
51. K.M. Noone, E. Strein, N.C. Anderson, P.-T. Wu, S.A. Jenekhe, D.S. Ginger, Nano Lett. 10,
2635 (2010)
52. K.M. Noone, N.C. Anderson, N.E. Horwitz, A.M. Munro, A.P. Kulkarni, D.S. Ginger, ACS
Nano 3, 1345 (2009)
53. J. Seo, M.J. Cho, D. Lee, A.N. Cartwright, P.N. Prasad, Adv. Mater. 23, 3984 (2011)
54. V.I. Klimov, J. Phys. Chem. B 110, 16827 (2006)
55. M.C. Beard, J. Phys. Chem. Lett. 2, 1282 (2011)
56. M.C. Beard, A.G. Midgett, M. Law, O.E. Semonin, R.J. Ellingson, A.J. Nozik, Nano Lett. 9,
836 (2009)
57. W.E.J. Beek, R.A.J. Janssen, in Hybrid Nanocomposites for Nanotechnology, ed. by L.
Merhari (Springer, New York, 2009)
58. J.B. Sambur, T. Novet, B.A. Parkinson, Science 330, 63 (2010)
59. E.H. Sargent, Nat. Photonics 6, 133 (2012)
60. E. Arici, N.S. Sariciftci, D. Meissner, Adv. Funct. Mater. 13, 165 (2003)
61. E. Arici, H. Hoppe, F. Schäffler, D. Meissner, M.A. Malik, N.S. Sariciftci, Appl. Phys. A 79,
59 (2004)
62. D. Pan, L. An, Z. Sun, W. Hou, Y. Yang, Z. Yang, Y. Lu, J. Am. Chem. Soc. 130, 5620
(2008)
63. B. Koo, R.N. Patel, B.A. Korgel, Chem. Mater. 21, 1962 (2009)
64. D. Pan, D. Weng, X. Wang, Q. Xiao, W. Chen, C. Xu, Z. Yang, Y. Lu, Chem. Commun.
4221 (2009)
65. M. Kruszynska, H. Borchert, J. Parisi, J. Kolny-Olesiak, J. Am. Chem. Soc. 132, 15976
(2010)
66. H. Zhong, S.S. Lo, T. Mirkovic, Y. Li, Y. Ding, Y. Li, G.D. Scholes, ACS Nano 4, 5253
(2010)
67. M. Kruszynska, H. Borchert, J. Parisi, J. Kolny-Olesiak, J. Nanopart. Res. 13, 5815 (2011)
68. W. Yue, S. Han, R. Peng, W. Shen, H. Geng, F. Wu, S. Tao, M. Wang, J. Mater. Chem. 20,
7570 (2010)
69. D. Aldakov, A. Lefrancois, P. Reiss, J. Mater. Chem. C 1, 3756 (2013)
70. N. Radychev, D. Scheunemann, M. Kruszynska, K. Frevert, R. Miranti, J. Kolny-Olesiak,
H. Borchert, J. Parisi, Org. Electron. 13, 3154 (2012)
71. Y. Yang, H. Zhong, Z. Bai, B. Zou, Y. Li, G.D. Scholes, J. Phys. Chem. C 116, 7280 (2012)
72. E. Maier, T. Rath, W. Haas, O. Werzer, R. Saf, F. Hofer, D. Meissner, O. Volobujeva, S.
Bereznev, E. Mellikov, H. Amenitsch, R. Resel, G. Trimmel, Sol. Energy Mater. Sol. Cells
95, 1354 (2011)
73. T. Rath, M. Edler, W. Haas, A. Fischereder, S. Moscher, A. Schenk, R. Trattnig, M. Sezen,
G. Mauthner, A. Pein, D. Meischler, K. Bartl, R. Saf, N. Bansal, S.A. Haque, F. Hofer,
E.J.W. List, G. Trimmel, Adv. Energy Mater. 1, 1046 (2011)
74. A. Guchhait, A.J. Pal, ACS Appl. Mater. Interfaces 5, 4181 (2013)
75. M. Pientka, V. Dyakonov, D. Meissner, A.L. Rogach, D.V. Talapin, H. Weller, L. Lutsen,
D. Vanderzande, Nanotechnology 15, 163 (2004)
76. C.J. Novotny, E.T. Yu, P.K.L. Yu, Nano Lett. 8, 775 (2008)
77. J. Huang, Z. Yin, Q. Zheng, Energy Environ. Sci. 4, 3861 (2011)
78. W.J.E. Beek, M.M. Wienk, M. Kemerink, X. Yang, R.A.J. Janssen, J. Phys. Chem. B 109,
9505 (2005)
79. W.J.E. Beek, M.M. Wienk, R.A.J. Janssen, Adv. Funct. Mater. 16, 1112 (2006)
80. S.V. Bhat, A. Govindaraj, C.N.R. Rao, Sol. Energy Mater. Sol. Cells 95, 2318 (2011)
81. S.D. Oosterhout, M.M. Wienk, S.S. van Bavel, R. Thiedmann, L.J.A. Koster, J. Gilot, J.
Loos, V. Schmidt, R.A.J. Janssen, Nat. Mater. 8, 818 (2009)
82. S.D. Oosterhout, M.M. Wienk, M. Al-Hashimi, M. Heeney, R.A.J. Janssen, J. Phys. Chem.
C 115, 18091 (2011)
83. Y.-Y. Lin, T.-H. Chu, S.-S. Li, C.-H. Chuang, C.-H. Chang, W.-F. Su, C.-P. Chang, M.-W.
Chu, C.-W. Chen, J. Am. Chem. Soc. 131, 3644 (2009)
84. B. Pradhan, S.K. Batabyal, A.J. Pal, Appl. Phys. Lett. 89, 233109 (2006)
200 12 Hybrid Polymer/Nanocrystal Solar Cells
85. B. Pradhan, S.K. Batabyal, A.J. Pal, Sol. Energy Mater. Sol. Cells 91, 769 (2007)
86. G.K. Mor, K. Shankar, M. Paulose, O.K. Varghese, C.A. Grimes, Appl. Phys. Lett. 91,
152111 (2007)
87. T. Goshal, S. Biswas, S. Kar, A. Dev, S. Chakrabarti, S. Chaudhuri, Nanotechnology 19,
065606 (2008)
88. Y. Hames, Z. Alpaslan, A. Kösemen, S.E. San, Y. Yerli, Sol. Energy 84, 426 (2010)
89. D. Bi, F. Wu, W. Yue, Y. Guo, W. Shen, R. Peng, H. Wu, X. Wang, M. Wang, J. Phys.
Chem. C 114, 13846 (2010)
90. K. Takanezawa, K. Tajima, K. Hashimoto, Appl. Phys. Lett. 93, 063308 (2008)
91. C.-Y. Liu, Z.C. Holman, U.R. Kortshagen, Nano Lett. 9, 449 (2009)
92. C.-Y. Liu, Z.C. Holman, U.R. Kortshagen, Adv. Funct. Mater. 20, 2157 (2010)
93. R. Dietmueller, A.R. Stegner, R. Lechner, S. Niesar, R.N. Pereira, M.S. Brandt, A. Ebbers,
M. Trocha, H. Wiggers, M. Stutzmann, Appl. Phys. Lett. 94, 113301 (2009)
94. J.-S. Huang, C.-Y. Hsiao, S.-J. Syu, J.-J. Chao, C.-F. Lin, Sol. Energy Mater. Sol. Cells 93,
621 (2009)
95. H.-J. Syu, S.-C. Shiu, C.-F. Lin, Sol. Energy Mater. Sol. Cells 98, 267 (2012)
96. L. He, C. Jiang, Rusli, D. Lai, H. Wang, Appl. Phys. Lett. 99, 021104 (2011)
97. M. Al-Ibrahim, H.-K. Roth, M. Schroedner, A. Konkin, U. Zhokhavets, G. Gobsch, P.
Scharff, S. Sensfuss, Org. Electron. 6, 65 (2005)
98. C.R. McNeill, A. Abrusci, J. Zaumseil, R. Wilson, M.J. McKiernan, J.H. Burroughes, J.J.M.
Halls, N.C. Greenham, R.H. Friend, Appl. Phys. Lett. 90, 193506 (2007)
99. S. Wilken, D. Scheunemann, V. Wilkens, J. Parisi, H. Borchert, Org. Electron. 13, 2386
(2012)
100. B.C. Thompson, J.M.J. Frechet, Angew. Chem. Int. Ed. 47, 58 (2008)
101. C. Deibel, V. Dyakonov, Rep. Prog. Phys. 73, 096401 (2010)
102. C.J. Brabec, S. Gowrisanker, J.J.M. Halls, D. Laird, S. Jia, S.P. Williams, Adv. Mater. 22,
3839 (2010)
103. D.S. Ginger, N.C. Greenham, Phys. Rev. B 59, 10622 (1999)
104. P.A. van Hal, M.P.T. Christiaans, M.M. Wienk, J.M. Kroon, R.A.J. Janssen, J. Phys. Chem.
B 103, 4352 (1999)
105. E. Witt, F. Witt, N. Trautwein, D. Fenske, J. Neumann, H. Borchert, J. Parisi, J. Kolny-
Olesiak, Phys. Chem. Chem. Phys. 14, 11706 (2012)
106. M. Kruszynska, M. Knipper, J. Kolny-Olesiak, H. Borchert, J. Parisi, Thin Solid Films 519,
7374 (2011)
107. D. Hertel, H. Bässler, Chem. Phys. Chem. 9, 666 (2008)
108. D.S. Ginger, N.C. Greenham, J. Appl. Phys. 87, 1361 (2000)
109. K. Kumari, S. Chand, P. Kumar, S.N. Sharma, V.D. Vankar, V. Kumar, Appl. Phys. Lett.
92, 263504 (2008)
110. K. Kumari, S. Chand, V.D. Vankar, V. Kumar, Appl. Phys. Lett. 94, 213503 (2009)
111. H. Borchert, F. Zutz, N. Radychev, I. Lokteva, J. Kolny-Olesiak, E. von Hauff, I. Riedel, J.
Parisi, in Proceedings of the 24th European Photovoltaic Solar Energy Conference
(EUPVSEC) (2009), p. 643
112. E. von Hauff, J. Parisi, V. Dyakonov, J. Appl. Phys. 100, 043702 (2006)
113. D.V. Talapin, C.B. Murray, Science 310, 86 (2005)
114. S. Neugebauer, J. Rauh, C. Deibel, V. Dyakonov, Appl. Phys. Lett. 100, 263304 (2012)
115. H. Fu, A. Zunger, Phys. Rev. B 56, 1496 (1997)
116. P. Guyot-Sionnest, M. Shim, C. Matranga, M. Hines, Phys. Rev. B 60, R2181 (1999)
117. K.E. Knowles, D.B. Tice, E.A. McArthur, G.C. Solomon, E.A. Weiss, J. Am. Chem. Soc.
132, 1041 (2010)
118. J. Schafferhans, A. Baumann, C. Deibel, V. Dyakonov, Appl. Phys. Lett. 93, 093303 (2008)
119. J.E.B. Katari, V.L. Colvin, A.P. Alivisatos, J. Phys. Chem. 98, 4109 (1994)
120. M. Kuno, J.K. Lee, B.O. Dabbousi, F.V. Mikulec, M.G. Bawendi, J. Chem. Phys. 107, 9869
(1997)
References 201
In Chap. 12, hybrid bulk heterojunction solar cells were discussed where colloidal
semiconductor nanocrystals are blended with conductive polymer to obtain a
solution which can be used for deposition of the materials from a liquid medium by
spin-coating, printing, spray-coating or other technologies. Thereby, the solution
contains both the inorganic and the organic material component. Solution-
processability of the colloidal nanocrystals is, however, not caused by the addition
of polymer, but an inherent property of the colloidal synthesis approach. In other
Fig. 13.1 a Sketch of the layer sequence in a typical Schottky solar cell. The Schottky contact is
formed at the interface of the semiconductor nanocrystal layer with an appropriately chosen
metal. In detail, to obtain a Schottky contact, there must be large offset between the Fermi levels
in the semiconductor and the metal before contact. b Energy scheme for the Schottky solar cell,
drawn for the case of a junction between a p-type semiconductor and a low work function metal
in equilibrium. The band bending in the depletion zone at the semiconductor–metal interface
results in a driving force for the separation of photogenerated electron–hole pairs
Fig. 13.2 a Sketch of the layer sequence in a typical depleted heterojunction solar cell. A
depletion zone is formed at the interface of the semiconductor nanocrystal layer with the TiO2. In
reality, the TiO2 is often nanoporous, so that the border to the semiconductor nanocrystals layer is
not as sharp as in this simplified sketch. b Energy scheme for a depleted heterojunction solar cell,
drawn for the case of a junction between a p-type semiconductor and n-type TiO2. The band
bending in the depletion zone at the semiconductor-TiO2 interface results in a driving force for
the separation of photogenerated electron–hole pairs
should be established between the cathode and the conduction band of TiO2, as
well as between the anode and the valence band of the p-type semiconductor.
Recent studies in the field of solar cells with inorganic nanocrystals as absorber
layers focused mainly on PbS and PbSe nanocrystals [1, 3, 6, 10–12]. In a work by
Koleilat et al. [13], colloidal p-type PbSe nanocrystals stabilized with octylamine
ligands were used to coat an ITO electrode. Afterwards, the PbSe film was treated
with 1,4-benzenedithiol which has two functional groups and acts as a cross-linker
because the molecule can bind to two neighbored nanocrystals in the film. The
treatment rendered the film insoluble in the solvent used previously for the
deposition of the nanocrystals. Therefore, after the treatment with the cross-linker,
a second layer of nanocrystals could be deposited to enhance the thickness of the
absorber layer. Treatment with benzenedithiol was applied after the second
nanocrystals deposition step as well, before finally, a Mg contact was deposited as
counter electrode [13]. Figure 13.3 shows a scheme of the energetic structure of
the devices as well as current–voltage curves measured under irradiation with
simulated sun-light. At the interface of the p-type PbSe nanocrystal layer and the
Mg electrode, a Schottky contact is formed which serves to separate electrons and
holes. The width of the depletion zone was estimated to be about 65 nm [13].
These quantum dot-based Schottky solar cells reached up to 1.1 % efficiency under
standard test conditions [13].
As can be seen from Fig. 13.3, this type of Schottky solar cells produces rea-
sonable current densities, but the open-circuit voltage remains rather low. Luther
et al. [14] addressed the question to optimize the open-circuit voltage in solar cells
with a PbSe/metal Schottky junction by systematically varying the size of the PbSe
nanocrystals (in the range from about 3 nm to about 7 nm diameter) as well as the
choice of the metal. Smaller particle size was found to be favorable for higher
open-circuit voltage, because lowering the valence band edge by the quantum size
effect increases the height of the Schottky barrier, which in turn influences the
208 13 Solar Cells with Inorganic Absorber Layers
open-circuit voltage [14]. Concerning the metal, a series of materials with the work
function varying in the range of approximately 3–5 eV was tested. Metals with
lower work function were found to enhance the open-circuit voltage, but the dif-
ference amounted only to about 150 meV for metals differing by about 2 eV in their
work function. From this observation, it was concluded that the Schottky barrier
height is only slightly dependent on the choice of the metal, due to Fermi level
pinning effects. Within the mentioned work, the open-circuit voltage remained
restricted to about 0.25 V, and the efficiency of the Schottky solar cells with PbSe
quantum dots reached 2.1 % under standard test conditions [14]. It is noteworthy,
that the fabrication of the absorber layer was done in this work by a layer-by-layer
deposition method, where the deposition of each nanocrystal layer from hexane as
solvent by dip-coating was followed by washing with a solution of ethanedithiol as
a cross-linker dissolved in acetonitrile [14]. In comparison to the earlier work by
Koleilat et al. [13], it seems possible that the improvement of device efficiency is
partly also due to the usage of 1,2-ethanedithiol instead of 1,4-benzenedithiol as
13.3 Solar Cells with Inorganic Absorber Layers 209
cross-linker. Later it was shown that the open-circuit voltage can be further
improved to 0.45 V by using ternary PbSxSe1-x nanocrystals as absorber in such
Schottky solar cells; with optimized composition, the efficiency was improved to
3.3 % [15].
In 2011, Ma et al. [16] readdressed the issue of the particle size and studied
Schottky solar cells with very small PbSe nanocrystals having diameters in the
range of 1–3 nm. Also in this work, layer-by-layer deposition was used to obtain
films of the colloidal nanocrystals. 1,4-benzenedithiol was used as cross-linker,
and, as a difference to the before-mentioned works, a PEDOT interlayer was
introduced between the ITO contact and the PbSe nanocrystal layer [16]. It was
found that the open-circuit voltage can reach up to *0.6 V for dots having the first
excitonic absorption peak at *1.6 eV photon energy. This corresponds to a par-
ticle diameter around 2.3 nm, and efficiencies reached *3.5 % on average, with
the best cells having even *4.6 % power conversion efficiency [16].
Schottky solar cells were not only realized with PbSe. Szendrei et al. [17]
studied corresponding solar cells with PbS nanocrystals. The absorber was placed
between ITO and LiF/Al as electrodes. Strictly speaking, it is not evident, if this
corresponds still to a normal Schottky contact, because the LiF interlayer is of
course not metallic. On the other hand, the LiF layer had only a very small
thickness of just 1 nm [17]. Layer-by-layer deposition with 1,4-benzenedithiol as
cross-linker was applied. The authors reported that each deposition step increased
the layer thickness by 6–7 nm, so that the overall thickness of the absorber layer
can be precisely controlled by the number of deposition cycles applied. Smooth
surfaces were obtained as confirmed by measurements with atomic force
microscopy [17]. Using quasi-spherical PbS nanocrystals with diameters of 3.5 or
4.3 nm, the solar cells reached power conversion efficiencies of 3.5 and 3.9 %,
respectively [17]. Moreover, the dependency of the short-circuit current density
and open-circuit voltage on the illumination intensity were examined. From the
measurements was concluded that build-up of space charge does not occur as a
factor limiting the devices, but trap states were found to play a role [17].
In conclusion, Schottky solar cells with absorber layers fabricated via layer-by-
layer deposition of colloidal lead chalcogenide nanocrystals reach currently effi-
ciencies which can compete with those of the best polymer/nanocrystal bulk
heterojunction systems discussed in Chap. 12. However, the Schottky solar cell
concept has some general limitations, a good discussion of which can for example
be found in [3]. One limiting factor is the relatively narrow depletion zone. For
example, in [13], the width of the depletion region in ITO/PbSe/Mg Schottky solar
cells was estimated to be only about 65 nm. Outside the depletion zone, band
bending is not pronounced. This is also illustrated in the energy scheme in Fig. 13.
3a, where we can find a quasi-neutral region of the thickness LQN. By consequence,
the band structure of the devices provides only in a narrow region a driving force
helping to separate photo-generated electron–hole pairs. On the other hand,
enabling more or less complete absorption of the sunlight in the spectral absorption
range of the nanocrystals requires an absorber layer thickness which is signifi-
cantly larger than the width of the depletion zone. If the layers are thicker,
210 13 Solar Cells with Inorganic Absorber Layers
electron–hole pairs generated outside the depletion zone, i.e., in the quasi-neutral
region, need first to diffuse into the depletion zone, before the band structure helps
splitting them [13]. Since recombination can occur during the diffusion process,
this limits the device performance [3].
An additional disadvantage is that the depletion zone is located on the backside
of the solar cells [3]. The light is coming in trough the ITO electrode, so that
absorption starts in the quasi-neutral region instead of the depletion region. By
consequence, the generation rate for electron–hole pairs is higher in the region
where their separation into free charge carriers is more difficult.
The depleted heterojunction concept outlined in Sect. 13.1 provides a solution
at least to the latter problem. Referring to Fig. 13.2, one can see that the depletion
zone is now located on the side where the incident light enters the absorber layer.
A step towards depleted heterojunction solar cells was made in 2009 by Choi et al.
[18], who prepared solar cells in the sequence ITO/PEDOT:PSS/PbSe-nanocrys-
tals/ZnO-nanocrystals/Al. Thus, a heterojunction between PbSe and ZnO was
used, but the PbSe/ZnO interface was still located at the backside of the absorber
layer with respect to the direction of the incident light. Functionality of the devices
was found to depend strongly on the particle size of the PbSe quantum dots,
because only for particles with a diameter below *3.8 nm, the conduction band
edge of the PbSe nanocrystals is higher in energy than the conduction band edge of
ZnO. Thus, only for small PbSe nanocrystals, the PbSe/ZnO system forms a type II
heterojunction [18]. Figure 13.4 shows current density–voltage curves for solar
cells with PbSe nanocrystals of systematically varied particle size. The depen-
dence of the open-circuit voltage on the particle diameter is shown as well. A
diameter of about 2 nm was found to give the best results, and corresponding solar
cells yielded up to 3.4 % power conversion efficiency [18].
Depleted-heterojunction solar cells as depicted in Fig. 13.2 were introduced in
2010 by Pattantyus-Abraham et al. [3], who investigated PbS nanocrystals of
different size in solar cells with the device architecture FTO/TiO2/PbS-nano-
crystals/Au. Also here, a layer-by-layer technique was used for the deposition of
the PbS absorber layer: The nanocrystals were stabilized by oleate ligands and
formed a colloidal solution in a mixture of octane and decane. After deposition of
one layer by spin-coating, the film was treated with 3-mercaptopropionic acid, in
order to replace the oleate ligands and to make the film insoluble in the solvent
mixture used to process the nanocrystals. Applying around 10 deposition cycles,
absorber layers of *200 nm thickness were obtained [3]. With PbS nanocrystals
of 3.7 nm diameter, the solar cells reached an open-circuit voltage of 0.53 V, a
short-circuit current density of 15.3 mA/cm2, a fill factor of 57 %, and a power
conversion efficiency of approximately 5 % under standard test conditions [3].
Recently, the influence of different cross-linkers on the properties of the devices
was studied by Jeong et al. [19]. The authors studied PbS-based depleted het-
erojunction solar cells, where mercaptopropionic acid or ethanedithiol were used
as cross-linkers during the layer-by-layer deposition of the PbS absorber,
respectively. Usage of mercaptopropionic acid was found to result in more than
twice as large photocurrent density when compared to the devices prepared with
13.3 Solar Cells with Inorganic Absorber Layers 211
ethanedithiol [19]. The physical origin of this effect was elucidated to be the
so-called mobility-lifetime product, which is the product of the charge carrier
mobility and their lifetime. The square root of the mobility-lifetime product is
proportional to the diffusion length of the charge carriers, i.e., to the distance
which the carriers can diffuse on average before they would be subjected to
recombination. The two cross-linkers studied were found to result in significantly
differing mobility lifetime-products [19]. This clearly demonstrates that the mol-
ecules surrounding or linking the nanocrystals in the film are of great importance
for the physical properties of the colloidal quantum dot films—quite similar as
discussed before in Chap. 12 for hybrid bulk heterojunction solar cells.
A rather different approach to design the nanoparticle surface is the usage of
inorganic, ionic ligands [6, 20, 21]. In 2011, Tang et al. [6] synthesized PbS nano-
crystals capped initially with oleic acid ligands and treated the particles with a
212 13 Solar Cells with Inorganic Absorber Layers
The inorganic solar cells based on PbSe and PbS quantum dots made rapid and
impressive progress in just a few years. However, they retain the restriction that
the materials involved are highly toxic which might restrict the application
13.4 Solar Cells with Inorganic Absorber Layers 213
potential of the technology at large scale. Therefore there are also research
activities on similar solar cells with less toxic semiconductors.
Recently, Scheunemann et al. [25] reported on solar cells with a heterojunction
between two layers of colloidally synthesized CuInS2 and ZnO nanocrystals,
respectively. ITO/PEDOT:PSS was used as front contact on the CuInS2 side, and
an opaque Al layer was used as back electrode on the ZnO side. Figure 13.5 shows
TEM images of the colloidal nanocrytals used as well as a cross-sectional SEM
image of a solar cell, without the Al contact. It can be seen that nanoparticle layers
rather uniform in height could be realized. With respect to the incident light, these
devices had the depletion zone located at the back side of the absorbing CuInS2
layer. The devices showed a photovoltaic effect, but the performance remained
with a power conversion efficiency of 0.2 % rather low, so far [25]. One limiting
factor (but not the only one) was the relatively small absorber layer thickness
which amounted to about 70 nm and permitted absorbing only a quite small
fraction of the incident photons in the visible range [25].
Another interesting work was presented by Li et al. [26], who prepared solar
cells with solution-producible, inorganic absorber layers in the sequence ITO/
CuInS2/CdS/Al. The authors did not use the colloidal chemistry approach, but
coated ITO with a solution containing precursor materials for CuInS2. The crys-
talline absorber material was then formed in situ on the ITO substrate upon
annealing [26]. The CdS layer was deposited on top by a second step using other
precursors. The devices reached up to *4 % power conversion efficiency [26]. As
a disadvantage, this work relies again on toxic cadmium chalcogenides as one of
the material components. However, it appears an interesting alternative to the
colloidal approach, and should in principle not be limited to specific semicon-
ductor materials, although structural features like particle size and shape are
certainly very difficult to control by such in situ synthesis methods.
As a last example, a study by Jeong et al. [27] on CuInSe2 solar cells with a
solution-producible absorber layer is mentioned here. The strategy followed much
closer the concepts of inorganic solar cells with CuInSe2 where the absorber is
usually prepared by evaporation or sputtering techniques. Starting from precursors
214 13 Solar Cells with Inorganic Absorber Layers
References
Abstract Apart from the polymer/nanocrystal hybrid solar cells and solar cells
with nanocrystal-based inorganic absorber layers discussed in the previous chap-
ters, there exist a number of other approaches to incorporate colloidally prepared
nanocrystals into photovoltaic devices. This chapter discusses selected alternative
concepts, where the nanocrystals fulfill quite different functions. One field of
research focusses on bulk heterojunction solar cells where the absorber layer is
composed of three material components, e.g., conductive polymer, fullerene
derivatives, and colloidal nanocrystals. A second topic concerns the usage of
colloidal semiconductor nanoparticles with a wide band gap in order to realize
solution-producible interlayers in organic solar cells. A third important research
field is so-called quantum dot-sensitized solar cells, which are similar to dye-
sensitized solar cells with the organic dye replaced by inorganic nanocrystals.
Finally, the usage of metal nanoparticles for enhancing light absorption in organic
solar cells by the exploitation of plasmonic effects will be discussed.
max 1 eff
VOC E 0:3 V ð14:1Þ
e G
The effective band gap EGeff is the energetic difference between the HOMO level
of the donor polymer and the LUMO level of the fullerene acceptor, as illustrated
in Fig. 9.5. To estimate the achievable photocurrent, it was assumed that only light
absorption by the polymer contributes to current generation and that the external
quantum efficiency equals 65 % throughout the absorption range of the polymer,
i.e., for all photon energies above the energy gap of the donor polymer [1]. The
short-circuit current density can then be calculated from the EQE according to.
(10.9). Assuming furthermore a fill factor of 65 % [1], the power conversion
efficiency can finally be calculated according to (10.8). From these simulations, it
was concluded that the ideal polymer to be combined with PCBM should have a
HOMO–LUMO gap of about 1.5 eV, with the LUMO level located 0.3 eV above
the LUMO of PCBM. In that case, 10–11 % power conversion efficiency could be
expected [1].
One strategy to reach higher efficiency is tandem solar cells which use two BHJ
layers connected in series [2], as illustrated in Fig. 14.1a. The semiconducting
polymers in both absorber layers must have different energy gaps. The sun-light is
first passed through the layer with the polymer having the higher energy gap.
Thereby, high energy photons can be absorbed, whereas low energy photons are
transmitted through the first absorber layer. Part of the low energy photons can
then be absorbed in the second absorber layer with the polymer having the lower
energy gap. For tandem solar cells, possible efficiencies up to 14–15 % have been
predicted in simulations similar to those described above for single junction solar
cells [2]. Experimentally, tandem solar cells with polymer/fullerene absorber
layers reached recently 8.9 % power conversion efficiency, and extending the
concept to triple junction solar cells yielded even 9.6 % efficiency [3].
As an alternative to enhance light absorption in polymer/fullerene BHJ solar
cells, the usage of ternary blends has been introduced. In that case, three materials
are blended in a single BHJ layer [4], as schematically illustrated in Fig. 14.1b.
The third material component added to the polymer/fullerene absorber can have
different functions. It can for example be a low band gap polymer, with the
intention to harvest also sun-light which would not be absorbed by the binary
polymer/fullerene blend [4]. However, the third component can also bring other
benefits than just extending the spectral absorption range of the absorber layer. For
example, the third component can have an impact on the elementary processes of
charge separation and charge transport. A recent review article on organic ternary
solar cells by Ameri et al. [4] summarizes the advantages of and achievements
made with ternary BHJ solar cells.
One direction of research focusses in this context on the addition of colloidal
nanocrystals to polymer/fullerene absorber layers, so that ternary blends composed
of conductive polymer, fullerenes and inorganic nanocrystals are obtained [4].
Thereby, semiconductor nanocrystals have been used as well as metal nanoparti-
cles. In this section, the impact of adding semiconductor nanocrystals will be
discussed. Metal nanoparticles will be considered in Sect. 14.4.
14.1 Bulk Heterojunction Solar Cells 219
Fig. 14.1 a Schematic illustration of a possible architecture for tandem solar cells, where two
absorber layers containing different polymer/fullerene bulk heterojunctions are connected in
series, with a recombination layer separating the absorbers. The absorber layer containing the
donor polymer with the higher energy gap should be placed on the side of the incident light, in
order to minimize thermalization losses. b Scheme for bulk heterojunction solar cells containing a
ternary blend in a single absorber layer
In 2011, Peterson et al. [5] studied ternary solar cells by adding CdSe nano-
crystals functionalized with methyl viologen to P3HT/PCBM bulk heterojunction
layers. Concentration ratios of all three components were systematically varied.
An impact of the CdSe nanocrystals on photocurrent generation in the spectral
region around the excitonic absorption peak of the nanocrystals was reported as
well as an effect on the build-up of space charge [5]. Performance data in terms of
power conversion efficiency was, however, not presented in that study [5]. Fu et al.
[6] investigated ternary P3HT/PCBM/CdSe BHJ solar cells in an inverted device
architecture, namely with the layer sequence FTO/TiO2/ternary BHJ/PEDOT:PSS/
Ag. The CdSe nanocrystals were subjected to ligand exchange with pyridine in this
case. With weight ratios of 1:1:0.1 for P3HT:PCBM:CdSe, an improvement of the
device efficiency from about 2.1 % (without CdSe) to about 3.1 % (with CdSe)
was reported in that work [6]. The increased efficiency resulted mainly from an
enhanced photocurrent and improved fill factor, whereas the open-circuit voltage
was only slightly affected. Improved light absorption as well as electron transport
through percolation pathways established by the slightly aggregated pyridine-
220 14 Other Types of Solar Cells Containing Colloidally Prepared Nanocrystals
capped CdSe nanocrystals in the ternary absorber films were discussed as the
physical reasons for the improved performance [6].
In another study, the addition of CuInS2 nanocrystals capped with 1-dode-
canethiol to P3HT/PCBM solar cells was recently investigated [7]. Weight ratios
were varied in a large range from 1:1:0 to 1:1:1 for P3HT:PCBM:CuInS2. Also
here, a positive effect was reported. Devices with 20 wt% CuInS2 (weight fraction
with respect to P3HT) showed an enhanced short-circuit current density with
respect to the P3HT/PCBM reference system. With the fill factor and open-circuit
voltage remaining nearly constant, this resulted in a slight increase of the power
conversion efficiency from about 2.4 % to about 2.8 % [7]. From a critical point of
view, it should be noted, however, that the differences in the performance data
were relatively small, with information on the statistical scatter of the results from
sample to sample not being provided. Furthermore, optimized P3HT/PCBM solar
cells can in principle reach higher power conversion efficiency [8]. The latter
argument applies also to the study mentioned in the previous paragraph. These
circumstances make it a bit difficult to conclude how significant the improvements
attributed to the incorporation of the semiconductor nanocrystals into the blends
really are. Future research in the field may bring further insight.
Ternary absorber layers with inorganic nanocrystals are not necessarily always
composed of nanocrystals, polymer and fullerenes. Instead, Yu et al. [9] studied
ternary BHJ solar cells using P3HT in combination with two types of inorganic
nanocrystals, namely TiO2 and CuInSe2 nanoparticles. Optimized devices of this
type reached an open-circuit voltage of 0.34 V, a short-circuit current density of
8.1 mA/cm2, a fill factor of 53 %, and a power conversion efficiency of 1.4 % [9].
A detailed study in the field of ternary solar cells involving inorganic nano-
particles was presented in 2012 by Liao et al. [10], who investigated ternary BHJ
solar cells with P3HT, PCBM and colloidal Cu2S nanocrystals capped with oleic
acid as ligand. Contrary to the above-mentioned studies, the fraction of the
nanocrystals in the ternary blends was kept much smaller in this work. With
respect to P3HT, the weight ratio was varied from 0.1 to 5 %, which corresponds
to volume fractions of only 0.012–0.62 % [10]. Thus, here the nanocrystals cannot
fulfill the role to significantly increase themselves the absorption of sun-light.
Nevertheless, with the concentration properly chosen, a positive effect on the
device performance was observed. The power conversion efficiency was enhanced
from 3.5 % without Cu2S to 4.3 % with 0.06 vol.% of Cu2S nanocrystals added to
the blends [10]. The reason for this improvement was attributed to changes
induced in the morphology of the blends which was studied in detail by different
types of X-Ray scattering experiments. Figure 14.2 shows a schematic picture of
the morphology derived from those experiments [10].
In the pure P3HT/PCBM system, the absorber layer was suggested to contain
three types of regions [10]: (1) amorphous domains containing P3HT and PCBM
(2) PCBM clusters, and (3) crystalline domains of P3HT with PCBM molecules
14.1 Bulk Heterojunction Solar Cells 221
Fig. 14.2 Schematic representation of the different domains present in the phase-separated BHJ
absorber layers composed of a P3HT/PCBM and b P3HT/PCBM/Cu2S (with 0.06 vol.% of Cu2S
nanocrystals). The proposed structural model is based on an analysis by X-ray scattering
techniques (Reprinted with permission from [10]. Copyright 2012 American Chemical Society)
intercalated into the lamellar structure of the P3HT crystals. These three types of
domains are illustrated in Fig. 14.2a. With 0.06 vol. % of Cu2S nanocrystals added
to the blends, the morphology was suggested to change as follows: As indicated by
smaller distances between adjacent lamellae, PCBM intercalation into the P3HT
crystallites was suppressed. Simultaneously, the PCBM clusters were found to be
significantly larger and concluded to form aggregates together with the Cu2S
nanocrystals [10]. Thereby, percolation pathways for more efficient electron
transport are established, as illustrated in Fig. 14.2b. This more favorable mor-
phology was concluded to be responsible for the increased device performance
[10]. The study demonstrates that colloidal nanocrystals added in low concentra-
tion to polymer/fullerene blends can open possibilities to tune the morphology of
BHJ absorber layers.
222 14 Other Types of Solar Cells Containing Colloidally Prepared Nanocrystals
Transition metal oxides, in particular zinc oxide and titanium dioxide, find
applications in solar cells not only as electron acceptor component in the active
layer, but also for the fabrication of interlayers between the cathode and the active
layer [11–19]. Because of the absolute position of the conduction band edge with
respect to vacuum in combination with the relatively large band gap, ZnO and
TiO2 are suitable materials to transport electrons between the active layer and the
electron contact (cathode), whereas transport of holes to the cathode is blocked.
Therefore, introducing such an interlayer is discussed to help avoiding undesired
recombination of holes at the cathode [11, 17].
Transition metal oxide interlayers can be produced by numerous approaches,
among them many different sol–gel methods where precursor materials are
deposited from solution on the electrode and subsequently converted to the
crystalline oxide by an annealing step [11, 15, 16, 20]. As an alternative, ZnO
interlayers were also realized by the deposition of nanocrystals from colloidal
solution [13, 21, 22]. A particularity of using preformed ZnO nanocrystals is that
the interlayer can be deposited from an appropriate solvent either on top of the
active layer made of a polymer/fullerene blend [13, 21], or on the cathode in the
case of inverted device architectures, where the active layer is deposited after-
wards on top of the ZnO interlayer [22]. In contrast, sol–gel approaches are not
always suitable for both types of device architectures. In the case of procedures
where the conversion of the deposited precursors to a crystalline transition metal
oxide requires relatively high annealing temperatures, the applicability is restricted
to inverted device architectures only [11]. However, sol–gel methods have also
been developed with moderate annealing temperatures below *150 C, so that at
least amorphous transition metal oxide interlayers can also be fabricated on top an
organic polymer/fullerene layer [12].
Apart from preventing holes to reach the cathode, an electron-selective ZnO or
TiOx interlayer can simultaneously have the function to act as an optical spacer
[21]. Since organic solar cells involve layers which are thin as compared to the
wavelength of visible light, interference effects play an important role for the
spatial distribution of the light intensity within the devices. Considering an organic
solar cell as a sequence of layers made of different materials, it is possible to
calculate the spatial distribution of the electric field intensity within the solar cell
by means of the so-called transfer matrix formalism [23, 24]. For a given wave-
length, intensity maxima and minima are obtained at specific positions in the cell.
Figure 14.3a shows as an example for such simulations the optical electric field for
a selection of wavelengths as a function of position in P3HT/PCBM solar cells
where a 40 nm thick active BHJ layer was placed between ITO/PEDOT:PSS and
LiF/Al as contacts [13]. It can be seen that throughout the wavelength range from
400 to 600 nm, i.e., the range where P3HT has a strong absorption, the electric
field reaches a maximum in the PEDOT:PSS layer, but decays within the active
14.2 Wide Band Gap Semiconductor Nanocrystals 223
layer toward the metal cathode. This is obviously not ideal in view of current
generation in the active layer. Figure 14.3b shows corresponding optical simula-
tions for devices, where a 39 nm thick interlayer of ZnO nanocrystals was intro-
duced between the active layer and the cathode [13]. Now, the maxima of the
electric field are shifted to fall into the active layer.
From the spatial distribution of the optical electric field, it is possible to cal-
culate the current generation rate and finally the photocurrent to be expected under
illumination, if there were no losses, i.e., for an internal quantum efficiency of
unity. According to the result shown in Fig. 14.3, introducing the ZnO interlayer
into devices with a 40 nm thick active layer should improve the photocurrent
considerably. This was indeed found theoretically as well as confirmed experi-
mentally [13]. As another remarkable result from the same study, the authors
concluded that the improvement of the photocurrent could be explained by the
optical effects only. In other words, additional effects like the above-mentioned
suppression of hole transport towards the wrong electrode were suggested to be not
224 14 Other Types of Solar Cells Containing Colloidally Prepared Nanocrystals
the field of BHJ solar cells, where the electron-collecting electrode is considered as
cathode, and the hole-collecting electrode as anode.
Classical DSSCs as depicted above reach currently up to *12 % power con-
version efficiency [27]. A certain disadvantage of the technology is the usage of a
liquid electrolyte which causes for example problems related to the temperature
stability of the devices. Furthermore, long-term stability of the organic dye mol-
ecules is a critical issue. Concerning the liquid electrolyte, there are also concepts
to replace it by a hole conducting polymer [28]. In that case, the hole remaining
after light excitation in the HOMO level of the dye is transferred to the polymer
and then conducted to the counter electrode.
In the last years, considerable progress was made in the field of DSSCs by
moving from organic dye molecules to perovskites with suitable absorption
properties and simultaneously high electron conductivity. Perovskite-based solar
cells can meanwhile reach power conversion efficiencies exceeding 12 % [29], up
to 15 % efficiency having been reported in 2013 on scientific conferences in the
field.
Another approach to improve dye-sensitized solar cells is to use semiconductor
nanocrystals as alternative sensitizers [30, 31]. In that case, the nanoporous titania
structure (see Fig. 14.4) is decorated by semiconductor nanocrystals instead of
metal–organic dye molecules. Such devices are then called quantum dot-sensitized
226 14 Other Types of Solar Cells Containing Colloidally Prepared Nanocrystals
solar cells (QDSSCs). For example, Im et al. [32] prepared solar cells where PbS
quantum dots were deposited by a layer-by-layer method on mesoporous TiO2 on
FTO. The structures were then infiltrated with P3HT as a hole conductor, and
completed with a solution-processed interlayer of PEDOT:PSS and finally a
thermally evaporated Au film as counter electrode. Under standard test conditions,
up to about 3 % power conversion efficiency were reached in that work [32].
Santra and Kamat [33] investigated CdS-based quantum dot-sensitized solar cells.
The photoactive semiconductor was deposited on FTO/TiO2 by a method called
successive ionic layer adsorption and reaction (SILAR) [34], and a liquid elec-
trolyte using sulfide/polysulfide as redox couple was employed [33]. Solar cells
with pure CdS reached about 1.6 % power conversion efficiency. The fabrication
of structures where the CdS was coated with CdSe on top increased the efficiency
to *4.2 %, and doping the CdS in the CdS/CdSe system with Mn2+ ions brought
further improvement to *5.4 % power conversion efficiency [33].
There are also attempts to fabricate QDSSCs with less toxic semiconductors.
For example, Santra et al. [35] prepared colloidal CuInS2 nanocrystals and
incorporated them into QDSSCs, again using a liquid electrolyte with sulfide/
polysulfide as redox couple. The device performance reached about 1.1 % effi-
ciency. In the same work, an improvement to *3.9 % power conversion efficiency
was achieved, but only when the CuInS2 nanocrystals were coated with a CdS
layer [35], thus using again highly toxic compounds.
Another interesting material is antimony sulfide. Chang et al. [36] coated FTO/
TiO2 with Sb2S3 using a chemical bath deposition process. Note in this place that
this deposition method, similar like the before-mentioned SILAR method, does
usually not result in highly defined nanocrystals as accessible by colloidal syn-
thesis. In contrast, a broad distribution of particle sizes or even coatings in the
form of thin films can result. Although the photoactive semiconductors prepared
by chemical bath deposition or the SILAR technique do not always resemble well-
defined quantum dots, the term of quantum dot-sensitized solar cells is generally
also applied to devices with this type of inorganic nanostructures. The mentioned
Sb2S3-based QDSSCs were completed with P3HT as hole conducting polymer and
PEDOT:PSS/Au as counter electrode [36]. Promising power conversion efficien-
cies up to *5.1 % were obtained for QDSSCs with this material [36]. In a later
work, different polymers were investigated as hole conductors in similar QDSSCs
with Sb2S3 as sensitizer [37]. Using PCPDTBT instead of P3HT raised the effi-
ciency to *6.2 % under one sun illumination, the improvement being mainly due
to a higher photocurrent density, but also to a slightly higher open-circuit voltage
[37]. These examples demonstrate that quantum dot-sensitized solar cells are also
a promising concept to use inorganic semiconductor nanostructures for solar cell
applications, in particular since relatively high efficiencies were reached here with
Cd- and Pb-free materials.
14.4 Metal Nanoparticles for Enhanced Light Absorption 227
In the last section of this chapter, the usage of metal nanoparticles as components
in organic solar cells will be discussed. Metal nanoparticles are of interest for the
field mainly because of three effects. The first idea is to use metallic nanostructures
to improve charge transport. Another strategy is to incorporate metallic nano-
particles as light scattering centers into organic solar cells, with the intention to
increase the path length which the light travels through the photoactive layer [4].
The third concept relates to plasmonic effects. In fact, the interaction of light with
thin metal structures, e.g., metallic nanoparticles, can lead to the excitation of
plasmons [38]. These are the quantums of collective longitudinal oscillations of
the electrons in the conduction band. In the simplest picture we can imagine that
the electrons in the conduction band are excited by the incident electromagnetic
wave to move collectively, so that the barycenter of the electrons’ negative charge
does no longer coincide with the barycenter of the positive charge given by the
ionic cores of the crystal. The resulting Coulomb force causes a repelling force, so
that the electrons will finally make an oscillating movement, characterized by a
specific frequency. The excitation of plasmons results in a strong absorption of
light, if the frequency of the incident electromagnetic radiation is in resonance
with the frequency of the longitudinal oscillations. Figure 14.5 shows as example
absorption spectra of colloidal Au nanocrystals of different size [39]. A pro-
nounced absorption maximum is visible. In the field of nanoparticles, the term
localized surface plasmon resonance (LSPR) is frequently used for the depicted
collective oscillations of conduction electrons. The LSPR effect is not restricted to
the given example of gold, but also observable for nanocrystals made of other
metals, e.g., silver and copper [40–42].
Theory has been developed to describe the plasmon resonance in absorption
spectra of metal colloids. The basic theory behind the phenomenon is the so-called
Mie theory, but advanced models are necessary to describe important features like
the size-dependency of the spectral position of the absorption maximum [39, 41,
43]. In regard of solar cells, the LSPR effect is of high interest, because in the
vicinity of the metallic nanoparticles results a local enhancement of the electro-
magnetic field [44]. This in turn is expected to cause enhanced light absorption in
the active layer of organic solar cells [4].
Although the motivation is rather evident, the success of using colloidal metal
nanoparticles to enhance the efficiency of organic solar cells by increased light
absorption is still under debate. In an early work, Kim and Carroll [45] added
colloidal Au and Ag nanoparticles stabilized with dodecylamine to the active layer
of solar cells with a bulk heterojunction made of poly(3-octylthiophene) and C60.
Reference samples without metal nanoparticles yielded about 1.2 % power con-
version efficiency. A positive effect of adding Au or Ag nanoparticles was
observed, the power conversion efficiency being increased by 50–70 %, and
attributed to improved electron transport [45]. However, the amount of metal
228 14 Other Types of Solar Cells Containing Colloidally Prepared Nanocrystals
nanoparticles used in the blends was of the order of a few percent by weight.
Translating this into volumetric ratios, and taking into account the insulating
nature of the relatively thick ligand shell, the interpretation in terms of improved
electron transport left room for discussion.
Later works came partly to controversial results. There are some studies, where
the addition of metal nanoparticles to the active layer of organic solar cells reduced
the device performance [46, 47]. For example, Topp et al. [46] investigated the
addition of colloidal Au nanoparticles to P3HT/PCBM BHJ solar cells. Thereby,
different surface modifications of the metal nanoparticles were considered: Au
particles synthesized directly with P3HT as stabilizer, dodecylamine-capped Au
nanoparticles, and particles where the dodecylamine ligand shell was exchanged
with pyridine. However, in all cases, the addition of Au nanoparticles to the active
layer reduced the power conversion efficiency [46]. In another study [47], dode-
canethiol-capped Au nanoparticles were added to the active layer of P3HT/PCBM
solar cells, but again, a reduction of the performance was observed.
On the other hand, there are also reports where introducing Ag or Au nano-
particles to the active layer of organic solar cells was found to enhance the device
performance [4, 44, 48]. Wang et al. [44] studied the impact of adding Au
nanoparticles stabilized with poly(ethylene glycol) to BHJ solar cells with PCBM
and the conducting polymer poly[2,7-(9,9-dioctylfluorine)-alt-2-((4-(diphenyl-
amino)phenyl)thiophen-2-yl)malononitrile]. The content of Au nanoparticles was
systematically varied in the range of 0–6 wt%, and a strong dependence of the
device performance on the amount of Au used was observed. For small amounts of
nanoparticles (0.5 wt%), a positive influence was obtained, but for larger amounts
of the metal nanocrystals, the performance was reduced [44]. In the same study,
the absorbance enhancement due to the LSPR effect was investigated both,
experimentally and theoretically. Increasing amounts of Au nanoparticles were
found to enhance light absorption more and more. However, large amounts of Au
had a negative influence on the morphology of the BHJ layer which in turn
reduced the charge carrier mobility as measured in single carrier diodes, as well as
14.4 Metal Nanoparticles for Enhanced Light Absorption 229
Fig. 14.6 a Schematic illustration of the device structure and lamination process for fully
solution-processed organic solar cells with a Ag nanowire electrode. All layers below the
nanowires are deposited by spin-casting onto the Ag substrate. b Cross-sectional SEM image in
which the Ag film, the organic layers, and the top Ag nanowire mesh electrode are visible.
Nanowires sunk into and adhered to the organic layer can be seen at the interface. c Top-view
SEM image of the devices in which the nanowire mesh is shown to be a continuous network
(Reprinted with permission from [49]. Copyright 2010 American Chemical Society)
the probability for exciton dissociation. Thus, competition between positive and
negative effects was obtained which resulted in enhanced power conversion effi-
ciency only in a narrow range of the Au nanoparticle content in the blends [44].
This shows that device optimization is a crucial issue, if one intends to really
benefit from plasmonic effects in organic solar cells.
A different possibility to benefit from metallic nanoparticles in organic solar
cells refers to conductive electrodes and shall be outlined here as a last example in
this book. As briefly discussed in the introduction as well as many review articles
and books on organic photovoltaics, an advantage of OPV is the possibility to
process the organic material layers from solution. However, the organic layers
need to be contacted by appropriate electrodes. This is usually done by starting the
preparation with one of the electrodes, e.g., ITO-coated glass or plastic foil, and
depositing the second electrode, usually a metal, on top of the organic-based layers
by thermal evaporation. In view of low production costs and an energy-efficient
fabrication process, it is desirable to avoid the usage of ITO as well as the last
230 14 Other Types of Solar Cells Containing Colloidally Prepared Nanocrystals
thermal evaporation step. In this regard, Gaynor et al. [49] proposed a promising
concept by showing that Ag nanowires are suitable materials to obtain top elec-
trodes without the need for a thermal evaporation step. Figure 14.6 illustrates the
device architecture used in that work and shows SEM images showing the
nanowire electrodes. In detail, organic solar cells were prepared by subsequently
spin-casting layers of Cs2CO3, P3HT:PCBM and PEDOT:PSS onto a Ag substrate
[49]. The top electrode was realized by separately preparing a mesh of Ag
nanowires on a glass substrate. The nanowire mesh was then pressed onto the
organic solar cells. When the supporting glass was lifted off, the nanowires
remained on top of the solar cell and formed together with the PEDOT:PSS layer a
conductive electrode. As-prepared P3HT/PCBM solar cells reached a reasonable
performance with 2.5 % power conversion efficiency under standard test condi-
tions [49]. Economic considerations on the perspectives for Ag nanowires as
electrode materials can be found in [50]. The examples given in this last section
show that also metallic nanoparticles are of interest in different regards for the field
of organic-based solar cells.
References
1. M.C. Scharber, D. Mühlbacher, M. Koppe, P. Denk, C. Waldlauf, A.J. Heeger, C.J. Brabec,
Adv. Mater. 18, 789 (2006)
2. T. Ameri, G. Dennler, C. Lungenschmied, C.J. Brabec, Energy Environ. Sci. 2, 347 (2009)
3. W. Li, A. Furlan, K.H. Hendriks, M.M. Wienk, R.A.J. Janssen, J. Am. Chem. Soc. 135, 5529
(2013)
4. T. Ameri, P. Khoram, J. Min, C.J. Brabec, Adv. Mater. 25, 4245 (2013)
5. E.D. Peterson, G.M. Smith, M. Fu, R.D. Adams, R.C. Coffin, D.L. Carroll, Appl. Phys. Lett.
99, 073304 (2011)
6. H. Fu, M. Choi, W. Luan, Y.-S. Kim, S.-T. Tu, Solid State Electron. 69, 50 (2012)
7. M. Nam, S. Lee, J. Park, S.-W. Kim, K.-K. Lee, Jpn. J. Appl. Phys. 50, 06GF02 (2011)
8. W. Ma, C. Yang, X. Gong, K. Lee, A.J. Heeger, Adv. Funct. Mater. 15, 1617 (2005)
9. Y–.Y. Yu, W.-C. Chien, Y.-H. Ko, S.-H. Chen, Thin Solid Films 520, 1503 (2011)
10. H.-C. Liao, C.-S. Tsao, T.-H. Lin, M.-H. Jao, C.-M. Chuang, S.-Y. Chang, Y.-C. Huang, Y.-
T. Shao, C.-Y. Chen, C.-J. Su, U.-S. Jeng, Y.-F. Chen, W.-F. Su, ACS Nano 6, 1657 (2012)
11. M.S. White, D.C. Olson, S.E. Shaheen, N. Kopidakis, D.S. Ginley, Appl. Phys. Lett. 89,
143517 (2006)
12. J.Y. Kim, S.H. Kim, H–.H. Lee, K. Lee, W. Ma, X. Gong, A.J. Heeger, Adv. Mater. 18, 572
(2006)
13. J. Gilot, I. Barbu, M.M. Wienk, R.A.J. Janssen, Appl. Phys. Lett. 91, 113520 (2007)
14. H. Cheun, C. Fuentes-Hernandez, Y. Zhou, W.J. Potscavage Jr, S.-J. Kim, J. Shim, A. Dindar,
B. Kippelen, J. Phys. Chem. C 114, 20713 (2010)
15. C. Tao, G. Xie, F. Meng, S. Ruan, W. Chen, J. Phys. Chem. C 115, 12611 (2011)
16. J. Huang, Z. Yin, Q. Zheng, Energy Environ. Sci. 4, 3861 (2011)
17. E.L. Ratcliff, B. Zacher, N.R. Armstrong, J. Phys. Chem. Lett. 2, 1337 (2011)
18. A. Bauer, T. Wahl, J. Hanisch, E. Ahlswede, Appl. Phys. Lett. 100, 073307 (2012)
19. Y. Vaynzof, A.A. Bakulin, S. Gelinas, R.H. Friend, Phys. Rev. Lett. 108, 246605 (2012)
20. A.C. Arango, L.R. Johnson, V.N. Bliznyuk, Z. Schlesinger, S.A. Carter, H. Hörhold, Adv.
Mater. 12, 1689 (2000)
References 231
21. J. Gilot, M.M. Wienk, J.A. Jansen, Appl. Phys. Lett. 90, 143512 (2007)
22. S. Wilken, D. Scheunemann, V. Wilkens, J. Parisi, H. Borchert, Org. Electron. 13, 2386
(2012)
23. H. Hoppe, S. Shokhovets, G. Gobsch, Phys. Status Solidi RRL 1, R40 (2007)
24. R. Häusermann, E. Knapp, M. Moos, N.A. Reinke, T. Flatz, B. Ruhstaller, J. Appl. Phys. 106,
104507 (2009)
25. M. Grätzel, J. Photochem. Photobiol. C: Photochem. Rev. 4, 145 (2003)
26. G. Boschloo, A. Hagfeldt, Acc. Chem. Res. 42, 1819 (2009)
27. M.A. Green, K. Emery, Y. Hishikawa, W. Warta, E.D. Dunlop, Prog. Photovoltaics Res.
Appl. 21, 827 (2013)
28. G. Liang, Z. Zhong, S. Qu, S. Wang, K. Liu, J. Wang, J. Xu, J. Mater. Sci. 48, 6377 (2013)
29. J.M. Ball, M.M. Lee, A. Hey, H.J. Snaith, Energy Environ. Sci. 6, 1739 (2013)
30. P.V. Kamat, J. Phys. Chem. C 111, 2834 (2007)
31. I. Mora-Sero, S. Gimenez, F. Fabregat-Santiago, R. Gomez, Q. Shen, T. Toyoda, J. Bisquert,
Acc. Chem. Res. 42, 1848 (2009)
32. S.H. Im, H.-J. Kim, S.W. Kim, S.-W. Kim, S.I. Seok, Energy Environ. Sci. 4, 4181 (2011)
33. P.K. Santra, P.V. Kamat, J. Am. Chem. Soc. 134, 2508 (2012)
34. D.R. Baker, P.V. Kamat, Adv. Funct. Mater. 19, 805 (2009)
35. P.K. Santra, P.V. Nair, K.G. Thomas, P.V. Kamat, J. Phys. Chem. Lett. 4, 722 (2013)
36. J.A. Chang, J.H. Rhee, S.H. Im, Y.H. Lee, H.-J. Kim, S.I. Seok, M.K. Nazeeruddin, M.
Grätzel, Nano Lett. 10, 2609 (2010)
37. S.H. Im, C.-S. Lim, J.A. Chang, Y.H. Lee, N. Maiti, H.-J. Kim, M.K. Nazeeruddin, M.
Grätzel, S.I. Seok, Nano Lett. 11, 4789 (2011)
38. C. Kittel, Introduction to Solid State Physics, 8th edn. (Wiley, New York, 2005)
39. S. Link, M.A. El-Sayed, J. Phys. Chem. B 103, 4212 (1999)
40. J.A. Creighton, D.G. Eadon, J. Chem. Soc. Faraday Trans. 87, 3881 (1991)
41. S. Link, M.A. El-Sayed, J. Phys. Chem. B 103, 8410 (1999)
42. S.D. Solomon, M. Bahadory, A.V. Jeyarajasingam, S.A. Rutkowsky, C. Boritz, J. Chem.
Educ. 84, 322 (2007)
43. M.M. Alvarez, J.T. Khoury, T.G. Schaaff, M.N. Shafigullin, I. Vezmar, R.L. Whetten, J.
Phys. Chem. B 101, 3706 (1997)
44. C.C.D. Wang, W.C.H. Choy, C. Duan, D.D.S. Fung, W.E.I. Sha, F.-X. Xie, F. Huang, Y.
Cao, J. Mater. Chem. 22, 1206 (2012)
45. K. Kim, D.L. Carroll, Appl. Phys. Lett. 87, 203113 (2005)
46. K. Topp, H. Borchert, F. Johnen, A.V. Tunc, M. Knipper, E. von Hauff, J. Parisi, K. Al-
Shamery, J. Phys. Chem. A 114, 3981 (2010)
47. Y.-J. Huang, W.-C. Lo, S.-W. Liu, C.-H. Cheng, C.-T. Chen, J.-K. Wang, Sol. Energy Mater.
Sol. Cells 116, 153 (2013)
48. D.H. Wang, D.Y. Kim, K.W. Choi, J.H. Seo, S.H. Im, J.H. Park, O.O. Park, A.J. Heeger,
Angew. Chem. Int. Ed. 50, 5519 (2011)
49. W. Gaynor, J.-Y. Lee, P. Peumans, ACS Nano 4, 30 (2010)
50. C.J.M. Emmott, A. Urbina, J. Nelson, Sol. Energy Mater. Sol. Cells 97, 14 (2012)
Index
A C
Absorbance, 120 Cadmium selenide (CdSe), 77, 101, 103,
Absorption, 2, 5–8, 20, 25–28, 32, 34, 52, 116, 123, 133, 134, 140, 160–162,
57–59, 119–121, 123, 124, 129–133, 182, 184, 186, 187, 189, 190, 192,
144, 146, 147, 159–162, 164, 165, 167, 206, 219
171, 172, 176, 180, 182, 190, 191, 194, Cadmium selenide nanorods (CdSe nanorods),
196, 210, 217–220, 222, 224, 225, 227, 161, 163, 164
228 Cadmium selenide tetrapods (CdSe tetrapods),
Absorption coefficient, 120 162–164
AgInS2, 175 Cadmium sulfide (CdS), 99–101, 107, 161,
AM 1.5G spectrum, 143, 146 167, 181, 226
Angle-resolved photoelectron spectroscopy, Cadmium telluride CdTe, 3, 76, 101, 102, 167,
98 168, 182, 206
Antimony sulfide (Sb2S3), 226 Carbon nanotubes, 74, 75
Atomic form factor, 83 Chalcopyrite, 3
Attenuation coefficient, 120 Charge-based deep level transient spectros-
Auger recombination, 172 copy (Q-DLTS), 194, 212
Charge carrier mobility, 149, 150, 186
Charge transfer, 6, 132, 163, 169, 176, 181,
B 182
Band edge photoluminescence, 31, 32, 121 Charge transfer complex (CTC), 135, 136
Beer-Lambert law, 57, 119 Charge transfer (CT) state , 135
Benzene, 44 Charge transport, 6, 149, 161, 183, 184
Benzenedithiol, 207, 209 Chemical doping, 48
Benzenoid, 52, 53 Chemical shift, 96, 100, 102
Benzenoid structure, 52 Chemical vapor deposition (CVD), 16, 74, 176
Bimolecular recombination, 187 cis-polyacetylene, 48
Bottom-up methods, 15 Colloidal solution, 16
Bragg equation, 80, 85 Colloidal synthesis, 15, 16
Bragg scattering, 65–67, 70, 71, 76 Conjugated double bonds, 42
Branching ratio, 96 Copper indium diselenide (CuInSe2, CISe),
Bright field imaging, 65, 76 172, 173, 175, 213, 220
Bulk heterojunction (BHJ), 5 Copper indium disulfide (CuInS2, CIS), 3, 70,
Butadiene, 43 88, 89, 91, 172–175, 183, 196, 206,
Butylamine, 167, 169, 190, 191 213, 220, 226
Core-shell nanocrystals, 19, 33, 71, 100, 101, Form factor of the crytsal, 83
103, 105–107 Fourier analysis, 69
Cross-linker, 207–209
Cyclic voltammetry (CV), 111, 112, 116, 168,
169, 175 G
g-factor, 130, 131
Growth phase, 18
D Gyromagnetic ratio, 130
Dangling bonds, 30, 32, 103, 185, 193
Dark field imaging, 66, 76
de Broglie wavelength, 64 H
Delocalization, 44, 47 Hexadecylamine, 167, 182
Depleted heterojunction solar cells, 9, 146, High-resolution transmission electron
203, 205, 210, 214 microscopy (HRTEM), 67, 68
Depletion zone, 204, 205, 209 Hole-only devices, 153, 185, 186
Dispersive recombination, 187 Hopping model, 183
Donor/acceptor system, 5, 217 Hopping transport, 149
Drift velocity, 150 Hot injection method, 18
Dye-sensitized solar cell (DSSC), 3, 145, 224, Hückel’s rule, 52
225 Hund’s rules, 39
Dynamic growth process, 17 Hybridization, 39, 40, 42
Hybrid orbitals, 40
Hybrid solar cells, 7, 159
E
Effective band gap, 135, 136, 162, 166, 217
Effective mass, 21 I
Elastic scattering, 64 Image filtering, 69, 71
Electron affinity, 114 Ideality factor, 141
Electron-only devices, 153 Impact ionization, 171
Electron spin-resonance (ESR), 129 Indene-C60 bisadduct (ICBA), 115
Electron tomography, 66, 74, 76, 177, 196, Indium arsenide (InAs), 98, 101, 103
197 Indium phosphide (InP), 8, 71, 96, 98, 101,
Energy dispersive X-ray analysis (EDX), 64 176, 186
Equivalent circuit model, 141 Indium tin oxide (ITO), 4, 6
Ethanedithiol, 165, 167, 170, 208, 210 Infinite spherical potential, 24
Ethene, 41, 43 In situ synthesis, 177, 196, 213
Ethyne, 42 Internal quantum efficiency (IQE), 147
Ewald construction, 83, 85 Intersystem crossing, 124
Exciton, 6, 59 Ionic ligands, 211
Exciton Bohr radius, 22, 23 Ionization potential, 114
Exciton binding energy, 22, 23, 59
Exciton diffusion length, 6, 59, 127, 178, 190
External quantum efficiency (EQE), 146 L
Extinction coefficient, 120 Lamellae, 56
Larmor frequency, 130
Lattice fringes, 67
F Lattice planes, 82
Ferrocene, 115 Laue function, 83–85
Fill factor (FF), 142 Layer-by-layer deposition, 208, 209, 212
Fluorescence markers, 35 LCAO approximation, 21
Fluorescence quantum yield, 101 Lead selenide (PbSe), 80, 81, 125, 152, 160,
Focused ion beam (FIB), 73 169, 171, 182, 185, 207, 208, 210
Förster resonance energy transfer (FRET), Ligand exchange, 30, 163, 166, 167, 176, 181,
122, 163, 180, 181 189, 195
Index 235
P
Parallel resistance, 141 R
Passivation, 30, 35, 101, 212 Radiative recombination, 31
PbS, 101, 146, 153, 169, 170, 207, 209, 210, Reciprocal lattice, 81, 82
212 Reference electrode, 112, 114
236 Index