1. Technische Universit¨at M¨unchen Physik Department
Realizing flexibility and artificial
structures for organic solar cells
Bachelor thesis
Jochen Wolf
July 25, 2013
Lehrstuhl f¨ur Funktionelle Materialien E13
Advisor: Dipl.-Phys. Claudia M. Palumbiny
Supervisor: Prof. Dr. Peter M¨uller-Buschbaum
5. Abstract
In this thesis, organic solar cells based on the polymer poly(3-hexylthiophen-2,5-diyl)
(P3HT) are examined. Four different device architectures are manufactured via spin
coating to realize indium tin oxide (ITO)-free organic solar cells. They are analysed us-
ing current-voltage measurements and UV/Vis spectroscopy. As a replacement for ITO,
highly conductive poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS)
is analysed. As a replacement for the widely used glass substrates, poly(ethylene tereph-
thalate) (PET) foil is chosen and the manufacturing process adjusted. Scanning electron
microscopy (SEM) is used to obtain information about spin coating on a PET foil. A
compact disc (CD) structure is transferred into a PEDOT:PSS layer on a PET foil using
plasticizer assisted soft embossing and then analysed using atomic force microscopy.
ITO-free organic solar cells are realized on glass with efficiencies of 1.75 % compared to
2.75 % with ITO. Some problems with the manufacturing process for the substrate PET
are solved, but no solar cell is produced of a performance comparable to the substrate
glass. SEM reveals defects in the used ITO-PET foil which could short circuit the solar
cells. Measured by AFM, the depth of an artificial CD structure transferred into glycerol
doped PEDOT:PSS is (8 ± 1) nm.
v
9. 1 Introduction
Renewable energies such as water, wind and solar energy have been used for centuries
(Pierce 2011). The first proof of concept of a solar cell was discovered in 1887 by
Alexander Stoletov. However, it took a century until the price of commercially available
solar modules dropped to a level at which wide deployment was put into action. Most
of these lower cost solar cells are based on the semiconductor silicon. Prices for silicon
based solar cells are expected to drop in the next years, in accordance with Swanson’s
law stating that the price drops by 20 % for every doubling of global production capacity
(Carr 2013).
However, many other materials and technologies are being researched in order to open
new markets for solar cells. Organic solar cells for example offer flexibility and a po-
tentially much lower cost due to the possibility to print the devices. They are also
able to make more use of diffuse light conditions, which silicon based solar cells do not
(D. Cheyns 2008, p. 243310-1). For widespread usage of solar cells, this makes organic
solar cells more reliable because they can make more usage out of an for example a
cloudy day. This property could lead to a reduced need for increasing the electric grid
capacity.
In any case, for the actual deployment of organic solar cells the photo current efficiency
(PCE) and the lifetime have to be improved. As shown in figure 1.4, the record in
organic solar cells is an efficiency PCE = 11.1 %, while inorganic solar cells have a
record efficiency of PCE = 44.0 %. On the other hand, organic solar cells have been
researched for a relatively short period of time, compared to inorganic ones and still
have untapped potential.
In this thesis, organic solar cells based on the polymer poly(3-hexylthiophen-2,5-diyl)
(P3HT) are examined. Despite being commercially unsound, a lot of research into
organic solar cells use an indium tin oxide (ITO) transparent electrode. ITO has a
number of disadvantages, like being expensive and brittle. As a replacement for ITO,
1
10. 1 Introduction
the polymer poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) is
investigated, which puts ITO-free organic solar cells within reach.
One aspect of this thesis examines the replacement of glass as a substrate underlying
most organic solar cells. The possibility of poly(ethylene terephthalate) (PET) as a
substrate is investigated, and manufacturing processes adjusted. In order to improve
the PCE of organic solar cells, the technique of plasticizer assisted soft embossing is
used. A structure taken from a compact disc (CD) is transferred onto an organic solar
cell on PET. All of these aspects are steps towards the deployment of organic solar cells
which could contribute to future power supply.
1.1 Basic physics of organic solar cells
1.1.1 Solar spectrum
The sun was formed about 4.6 billion years ago and plays a crucial role in the energy
supply for most lifeforms on earth. It will continue burning for another 5 billion years
(Hufnagel 1997, p. 2) although the earth will only be habitable for another one billion
years (Carrington 2000). All of today’s fossil fuels are indirectly supplied by the sun-
light, stored via photosynthesis in plants whose remainder is converted to fossil fuels.
The intensity of sunlight at the average distance between earth and sun is called the
solar constant which is about 1366 W
m2 (NASA 07.07.2013). Despite its name, it does
change over time - solar activity has multiple so-called cycles, the shortest one being
the (11 ± 2) year discovered by Heinrich Schwabe (Schwabe 1843). During this cycle,
the solar constant varies from 1365 W
m2 to 1367 W
m2 for an 81 day average, as shown in
figure 1.1.
However, more recent measurements established the solar constant at (1360.8±0.5) W
m2 ,
with the Schwabe cycle amplitude from minimum to maximum at 1.6 W
m2 and daily to
weekly changes at 4.6 W
m2 (Greg Kopp 2011, p. 1). The solar constant is defined for
the average distance between earth and sun, but due to the earth’s elliptical orbit, the
solar irradiance on top of the atmosphere changes. Its amplitude is 6.9 %, thus making
it four times as large as the Schwabe cycle (Dahlback 2002).
2
11. 1.1 Basic physics of organic solar cells
Figure 1.1: Variations of the solar constant from 1874 to 2007 for an 81 day average (red) and
a daily average (yellow) (after A. D. Crouch & Paquin-Ricard 2008, p. 737)
In conclusion, the sun is a highly reliable source of energy, although some changes in
the solar irradiance have to be taken into account. The sun’s radiation spectrum closely
resembles a black body at about 5800 K (Demtr¨oder 2004, p. 303). Due to absorption
of gases both on the sun’s and the earth’s surface in addition to the Rayleigh and Mie
scattering in the earth atmosphere, this black body spectrum is modified. A comparison
of the spectrum of a black body at 5250 K, the sunlight at the top of the atmosphere
and the radiation at sea level can be seen in figure 1.2 (after Rohde 2007). However, the
solar irradiance on ground level is changed with weather conditions, local atmospheric
composition, and the angle of incidence. To account for these factors, an average is
taken over one year for every land mass on earth, see figure 1.3 (after Steiner 2008).
1.1.2 Polymers and their band structure
In order to harvest the solar radiation shown in figure 1.3, solar cells can be used. The
solar cells used in this work use the band gap that some polymers show, and fall under
the category of organic solar cells. There are many different types of solar cells, and it
is a popular research field (NREL 2013a). Figure 1.4 shows an overview of the highest
efficiencies of different solar cells, organic solar cells are plotted in solid orange circles,
3
12. 1 Introduction
Figure 1.2: A comparison of the spectrum of a black body at 5250 K (grey), the sunlight at
the top of the atmosphere (yellow) and radiation at sea level (red) together with
the corresponding absorbing molecules (purple) (after Rohde 2007)
Figure 1.3: Annual ground solar irradiance for every land mass on earth averaged over 1983
until 2005 in [ W
m2 ] (after Steiner 2008)
4
13. 1.1 Basic physics of organic solar cells
for which the record is at 11.1 % efficiency. In this chapter the band structure of semi
conductors and polymers is described.
Figure 1.4: Best research-cell efficiencies of different types of solar cells (after NREL
10.07.2013b)
Electrons inside a semiconductor can only occupy discrete energy states which in their
simplest form can be derived via the Schr¨odinger equation. Polymers are similar in
their mathematical description to semiconductors. As a consequence of the Schr¨odinger
equation, combined with the theory of the linear combination of atomic orbitals, there
are molecular orbitals that electrons occupy. An example of molecular orbitals is shown
on ethene in figure 1.5, two overlapping π atomic orbitals form a molecular orbital. The
interaction between overlapping atomic orbitals can lower the overall energy, which are
called bonding molecular orbitals. If they increase the overall energy, they are called
antibonding molecular orbitals. Not all molecular orbitals are occupied, and the so-called
band gap is the energy difference between the highest occupied states and the lowest
unoccupied states. These are called the highest occupied molecular orbital (HOMO)
and the lowest unoccupied molecular orbital (LUMO).
In solar cells, an incoming photon can excite an electron from the occupied states to the
5
14. 1 Introduction
Figure 1.5: Orbitals of ethene: a) individual atomic orbitals, b) one molecular orbital and c)
energy of bonding- and antibonding orbitals (Palumbiny 2011, p. 7)
unoccupied ones. This leaves a hole, an empty state, surrounded by occupied ones. In
order for energy to be conserved, this can only happen if the incoming photon has an
energy EPh sufficient to let the electron jump into a state above the band gap.
EPh =
h · c
λ
> Eg (1.1)
Eg denotes the band gap, h the Planck constant, c the speed of light and λ the wavelength
of the photon. The energy surplus of the photon ∆E = EPh − Eg given to the electron-
hole pair is used to excite either the electron to a state above the LUMO, or to excite an
electron of lower energy than the HOMO, or both. The energy surplus is then dissipated
by creating phonons and photons. Thus, the energy is transformed through this process
called thermalization.
Therefore the band gap Eg becomes a very important choice for solar cells in order to
maximize both the amount of potential photons that can be absorbed and their energy.
In the polymers used in this work to build organic solar cells, the electron-hole pair is not
free, in the sense that it can move independently, but forms a bound pair, an exciton.
This stands in contrast for example to silicon based solar cells, where the electron-hole
pairs can move independently.
For a given band gap, the Shockley-Queisser limit describes the highest reachable effi-
ciency. It assumes a black body spectrum of the sun at TSun = 6000 K, the solar cell
6
15. 1.1 Basic physics of organic solar cells
at TSC = 300 K and a single band gap Eg (William Shockley 1961, p. 1). According to
this limit, the optimal band gap is Eg = 1.1 eV, which gives a maximum efficiency of
30 %. Using an AM 1.5 solar spectrum, which will be explained in chapter 3.6, instead
of a calculated spectrum, the Shockley-Queisser limit is modified, as shown in figure 1.6.
The modified Shockley-Queisser limit increases the maximum efficiency to 33.7 % at a
band gap of 1.34 eV. One definition of a semiconductor is a band gap of Eg < 2.5 eV,
while for example silicon has a band gap of Eg = 1.11 eV (Streetman 2000, p. 524).
There are a number of ways through which this limit can be exceeded, among them the
use of lenses to concentrate incoming light, or the use of multiple p-n junctions. For
example, a triple junction solar cell by Scanlon (2012) under 942x concentrated light
reached 44.0% efficiency. The benefit of light concentration is an increase in the open
circuit voltage, which will be explained in chapter 1.1.4.
Figure 1.6: Shockley-Queisser limit calculated using an AM 1.5 solar spectrum instead of a
black body spectrum (after Byrnes 2011)
1.1.3 The diffusion length of excitons
The excitons described in the previous section have a typical lifetime τ = 200 ps before
they recombine, sending out phonons and photons (Jorge Piris 2009, p. 14502 & 14505).
Due to their relatively short life expectancy, they fall in the category of Frenkel excitons.
In this chapter the diffusion length of excitons and consequences for polymer based solar
cells are discussed.
7
16. 1 Introduction
The lifetime limits the distance which an exciton can travel within the solar cell. Accord-
ing to (Paul E. Shaw 2008, p. 3519), this can be described by the following equation:
LD =
√
D · τ (1.2)
With a typical lifetime τ = 200 ps and a diffusion coefficient D = 5 · 10−4 cm2
s
(according to L. L¨uuer 2004), the excitons can travel about LD = 3.2 nm in the polymer
poly(3-hexylthiophen-2,5-diyl) (P3HT). A newer measurement corrects this distance to
LD = (8.5±0.7) nm (Paul E. Shaw 2008, p. 3519). A description of P3HT can be found
in chapter 2.1. Within this distance, an interface between the polymer acting as a donor
and the polymer acting as an acceptor has to be located, else the exciton will decay.
Figure 1.7: The basic process inside an organic solar cell exciton generation (a), transport (b),
dissociation (c) and charge carrier transport (d) (Meier 2012, p. 6)
At the interface, the band structure changes rapidly, as can be seen in figure 1.8, resulting
in high enough electrical fields to separate the exciton. The electron and the hole move
independently towards the electrodes. Therefore, the theoretically ideal morphology
between these two polymers is an interdigitated donor-acceptor interface as shown in
figure 1.9. The width of the individual fingers is in the same range as the distance
LD = (8.5 ± 0.7) nm.
However, for the purpose of this work the so-called bulk heterojunction, see figure 1.9,
is chosen in which the polymers are mixed together, forming disordered regions in which
one polymer dominates, as can be seen in figures 1.7 and 1.9. The actual structure
can, to some extend, be influenced by the choice of spin coat parameters and solvent
(C.Y. Kwong 2004).
8
17. 1.1 Basic physics of organic solar cells
Figure 1.8: Band structure of the polymers PEDOT:PSS and P3HT:PCBM, ITO and alu-
minium. The top line of each column represents the HOMO, the bottom line the
LUMO. (after Seong Kyu Janga 2012, p. 427)
Figure 1.9: Bilayer junction (left), bulk heterojunction (middle) and interdigitated junction
(right) between the donor and acceptor material (after Palumbiny 2011, p. 21)
1.1.4 Working principle of organic solar cells
In order to gain an insight into the solar cell, the generalized Shockley equation is used,
based on an equivalent circuit shown in figure 1.10. It describes the resulting current
through a solar cell I(V ) dependent on the external voltage V :
I(V ) = IS exp
e (V − IRshunt)
nkBT
− 1 −
V − IRshunt
RSerial
− IPh (1.3)
IS is the saturation current, IPh the photo current, Rshunt the shunt resistance, Rseries the
series resistance, n the ideality factor, kB the Boltzmann constant, T the temperature
and e the elementary charge (Giebink et al. 2010). These characteristic parameters are
9
18. 1 Introduction
Figure 1.10: Equivalent circuit model of organic solar cells with the shunt resistance Rshunt,
serial resistance RSerial, saturation current IS, photo current IPh and ideality
factor n (Palumbiny 2011, p. 37)
now explained in more detail.
The photo current IPh is the current generated by the photoelectric effect. It is therefore
independent of the external voltage, only dependent on the incoming radiation. Further-
more, the photo current Iph is equal to the short circuit current ISC, under the so-called
short circuit condition V = 0.
IPh = I(V = 0) = ISC (1.4)
The open circuit voltage VOC is the voltage at which no current flows I(VOC) = 0
(Andre Moliton 2006, p. 594). It is dictated by the band structure, as can be seen
in figure 1.8. The point at which the external voltage V is opposite and equal to the
internally generated voltage of the band structure is the open circuit voltage VOC.
The shunt resistance Rshunt is defined as the slope of the IV-curve at V = 0 (Andre Moli-
ton 2006, p. 594). For decreasing shunt resistances Rshunt, the current through the solar
cell I(V ) decreases. Consequently, the open circuit voltage VOC decreases as well.
The series resistance Rseries is defined as the slope of the IV-curve at the open circuit
voltage VOC (Andre Moliton 2006, p. 594). For high Rseries, it decreases the current
I(V ) around the open circuit voltage VOC. Also for large values of Rseries, it suppresses
the diode behaviour, which then looks like a resistor. It does not modify the open circuit
voltage VOC, but can decrease the short circuit current ISC for high series resistances
Rseries.
For an ideal solar cell, the ideality factor is n = 1. It describes how well the theory
matches the actual behaviour of the solar cell. Furthermore, the shunt resistance is
ideally Rshunt = ∞ and the serial resistance is Rseries = 0.
10
19. 1.2 Benefits of structured organic solar cells
1.2 Benefits of structured organic solar cells
Figure 1.11: Structured organic solar cell. The incoming ray of light is diffracted at the
structured PEDOT:PSS-P3HT:PCBM interface, and consequently trapped (after
Lang 2012, p. 15)
Organic solar cells built in this work only absorb about 50 %-75 % of the light inserted
into the active layer, as can be seen in figure 4.4. A higher absorption in the active layer
would be favourable, as it implies an increased exciton generation rate which would lead
to a higher photo generated current and a higher efficiency. Organic solar cells can be
structured in order to increase the absorption in the blend. In artificially structured
organic solar cells, there are three major ways in which the efficiency is influenced.
1.2.1 Diffraction gratings
The CD structure used in this work has a periodicity of 1.5 µm. Since the wavelengths
of visible light are in the same order of magnitude, the periodic CD structure acts as
a diffraction grating. This can be seen on a structured organic solar cell in figure 1.11.
Some of the incoming light gets diffracted and consequently enters the solar cell at an
angle α = 0◦
measured to the surface normal. This increases the optical path through
the solar cell, which increases absorption.
11
20. 1 Introduction
1.2.2 Light trapping
As described for diffraction gratings, structured solar cells lets light enter at an angle
α neq 0◦
. Using a reflective back coating, for example an aluminium layer, the vast ma-
jority of the light gets reflected back inside the solar cell. Through tuning the structure
and the refractive indexes, it is possible to reflect part of the light back. As can be seen
in figure 1.11, the light can be reflected back and forth until it is absorbed, which is
called light trapping.
1.2.3 Morphology change
PEDOT:PSS shows best conductivities when the charge is transported along the back-
bone of the polymer. Therefore, alignment of the backbones has a major impact on the
final conductivity of the PEDOT:PSS layer. The technique of plasticizer assisted soft
embossing influences that alignment (Jin Young Park & Carter 2011, p. 11251).
12
21. 2 Sample preparation
2.1 Materials used
Substrates
Two kinds of substrates are used in order to realize all the different device structures as
detailed in section 2.3 on both glass and PET. For the glass substrates, this is 1.1 mm
thick glass bought from Solems L’´energie lumi`ere. They have a layer of indium tin oxide
(ITO), a transparent metal oxide electrode, evaporated on top. For samples on glass
without an ITO layer, it is removed via a catalytic reaction with hydrochloric acid and
zinc.
PET substrates are PXM739 foils bought from HiFi industrial film. They have a
thickness of d = 175 µm. The ITO-PET substrates are V 150A − OFS bought from
Nitto Denko. They consist of six layers, ITO being the top one, as shown in figure 2.1.
Once the release liner is removed, it has a thickness of d = 45 µm. The ITO-PET foils
are annealed for 1.5 hours at 140 ◦
C to bind the glue. This annealing step must be
performed before the four step cleaning process, in order to avoid the dissolving of the
glue, which would result in the splitting of the foil.
PEDOT:PSS
Poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) is a mixture of
two polymers. Its chemical structure is shown in figure 2.4. Dissolved in water, it has a
high transparency and can be treated with various solvents to increase the conductivity
from 1-10 S
cm
for pristine PEDOT:PSS by two to three orders of magnitude (Yong
Hyun Kim 2011, p. 1076). One of the most important properties of PEDOT:PSS is its
good hole conducting and electron blocking ability. In this work, it is used as an electron
13
22. 2 Sample preparation
Figure 2.1: The six layers of the used ITO-PET substrates: ITO (white), under coating (or-
ange), PET (blue), back coating (pink), adhesive (olive), and release liner (red).
The total thickness is 120µm, the thickness from ITO to adhesive 45µm
Figure 2.2: Carbon (grey), hydrogen (black), sulfur (yellow) and oxygen (red) atoms form the
chemical structures of PEDOT (left) and PSS (right) which forms PEDOT:PSS
if solved in water. Hydrogen atoms bound to oxygen (pink) is highly polar (after
Meier 2012, p. 66)
blocking layer, as a transparent electrode, or both. It is soluble in water, in which PSS
acts as an acid (Yong Hyun Kim 2011, p. 1078).
P3HT:PCBM
Poly(3-hexylthiophene) (P3HT) is a polymer and an electron donor. Phenyl-C61-butyric
acid methyl ester (PCBM) is a small molecule and an electron acceptor (Wanli Ma 2005,
p. 1618). Both are soluble in chlorobenzene. Their chemical structure can be seen in
figure 2.3. The band gap of P3HT lies at about 1.9 eV (Renee Kroon 2008, p. 531).
Solved in CD, P3HT:PCBM is called blend.
14
23. 2.2 Solar cell production
Figure 2.3: The chemical structures of P3HT (left) and PCBM (right) solved in CB form
P3HT:PCBM, the so-called blend. Coloring is analogous to figure 2.4. (after
Meier 2012, p. 62 & 68)
Aluminium
Aluminium granules are bought from ChemPur. They have a size of 2-10 mm with a
purity of 99.99%.
2.2 Solar cell production
2.2.1 Substrate cleaning
As previously mentioned, the two substrates used in this work are PET-foil and glass.
The PET-foil is cut into 2.2 x 2.2 cm2
squares, while the glass already comes in the
required size. The substrates are cleaned using ten-minute long ultrasonic baths in the
following sequence:
• Alconox high precision aqueous cleaning solution, 24 g
l
(Alconox Inc.)
• Ethanol, purity ≥ 99.5% (Carl Roth)
• Acetone, purity ≥ 99.5% (Carl Roth)
• Isopropanol, purity ≥ 99.5% (Carl Roth)
15
24. 2 Sample preparation
The substrates are initially rinsed off with distilled water, and at each change of solvent,
with the current and subsequently the next solvent. Finally, they are taken out of the
isopropanol one by one to be blown dry with nitrogen. Care is taken to ensure the
samples do not dry in between the solvent baths.
2.2.2 Oxygen plasma treatment
After the sample cleaning, the substrates are treated with an O2-plasma in order to
achieve a surface functionalization which, in turn, leads to homogeneous thin films after
spin coating. This is due to an increase in the hydrophilic character of the PEDOT:PSS.
To this end, the samples are placed inside a vacuum chamber, which is pumped down
to the pressure of P0 = 0.16 mbar. While further pumping down, oxygen is released
into the chamber at a rate to reach a pressure equilibrium at Pequ = 0.40 mbar.
Then, a voltage high enough to ignite the oxygen is applied for 15 min at a power of
P = 200 W. The PET samples have to be placed away from all air vents to avoid the
pressure wave from opening the vents after the oxygen plasma treatment. The surface
functionilization effect gradually wears off, therefore the samples must be spin coated
within the next hour.
2.2.3 Spin coating and post-treatment of PEDOT:PSS
For the mixing of polymer solutions, welted glass with an polyethylene lid bought from
Carl Roth is cleaned with the appropriate solvent of the polymer solution and blown-
dry using nitrogen. Liquids are measured using micro litre pipettes (Carl Roth), while
solids are weighted using a high precision balance (Sartorius BP 210, error 0.01 mg).
In order to achieve maximum reproducibility, PEDOT:PSS is taken out of refrigerated
storage 30 minutes early for it to achieve room temperature. It is then given a 15 min
ultrasonic bath to break up larger clusters that PEDOTS:PSS tends to form in aqueous
solution over time. Afterwards, it is filtered using a 0.45 µm PTFE filter to remove any
remaining larger clusters.
For the next step, the different layers are applied onto the substrates. For this purpose,
the widely used technique of spin coating is chosen, which creates polymer films of
reproducible thickness by spinning the sample, thus distributing the solution evenly
16
25. 2.2 Solar cell production
and evaporating the solvent. The substrates are held in place on a S¨uss MicroTec
Lithography Delta 6Rs TT spin coater using a vacuum. The highest acceleration setting
of 9 is used for all the samples, which accelerates the sample within 6 seconds to its speed
of 2500 rpm.
For the PET substrates the use of a vacuum is not directly possible, since the PET foils
would bend, resulting in an inhomogeneous layer thickness. Furthermore, the PET foils
would also wrinkle while annealing, and therefore the annealing temperature would not
be homogeneous over the sample. To solve this issue, 10 µl of glycerol, 99.5% purity
(Carl Roth), is used on top of a microscopy slide.
By placing the PET foil on the glycerol, it is secured in place strongly enough to with-
stand the spin coating, while the glycerol remains on the lower side of the substrate and
therefore is not interacting with the sample, flattening it. After being placed on the
glass, excess glycerol is pushed out on the side by a stream of nitrogen, originating from
directly above the sample. This method also flattens the PET foil for the spin coating
as well as preventing the PET foil from wrinkling during annealing, because of its higher
boiling point of 290◦
C.
The first layer to be spin coated on top of the cleaned and plasma-treated substrates is
the filtered PEDOT:PSS. The exact parameters of the PEDOT:PSS spin coating differ,
as the different device architectures come into play. These will be described in more
detail in section 2.3, along with the device architecture dependant areas covered with
PEDOT:PSS. To achieve high conductivity in the PEDOT:PSS, three out of four device
architectures are placed in an ethylene glycol (EG) bath for three minutes. Afterwards,
they are placed on the spin coater for 30 s at 1500 rpm in order to remove excess EG,
followed by an annealing step of 15 min at 140 ◦
C. After two minutes, the substrates
have to be moved once in order to let the evaporated EG from the bottom out, otherwise
it will leave white drying rings. This is especially critical for the PET foils, because of
the capillary attraction between the substrate and the surface, due to the intermediary
EG. To minimize scratches on the PET foils, the surface should be as smooth as possible,
in this work, microscopy glass slides are used.
17
26. 2 Sample preparation
2.2.4 Spin coating of P3HT:PCBM
After the PEDOT:PSS layers, the active layer must now be administered. The polymers
P3HT and PCBM are both soluble in chlorobenzene (CB). According to a survey by
Minh Trung Dang (2011), most authors use a mixture weight ratio of 1:1, although the
best ratio would be somewhere in range 1:0.8 to 1:1. In order to achieve maximum
comparability, a 1:1 weight mixture of 24 mg
ml
of each polymer in CB is chosen. To avoid
getting dirt into the polymers, the welted glasses, pipette tips and tweezers were cleaned
with CB.
Each polymer is first weighted, and mixed with the appropriate amount of CB sepa-
rately. Since the polymers do not dissolve instantly, an hour of shaking is required, after
which the better dissolving polymer, PCBM, is transferred to the P3HT. To minimize
evaporation of the CB, the lids of the glasses are further sealed with parafilm by Bemis.
Due to the light sensitivity of the blend, the solution is also wrapped in aluminum foil.
The resulting mixture is then weighed and shaken for another (20±2) hours, after which
it is weighted again. The change in mass ∆m is used to calculate the amount of CB VCB
needed to mitigate the evaporated CB, using the density of CB ρCB = 1.1 g
ml
:
VCB =
∆m
ρCB
(2.1)
These steps are timed, so that the P3HT:PCBM is done vibrating at the time the
PEDOT:PSS layers are created. The P3HT:PCBM is then spin coated on the samples,
at 2000 rpm for 30 s. The lights in the room are kept to the minimum necessary because
of the light sensitivity of P3HT:PCBM. Also, after the spin coating, care is taken to
ensure that the samples are exposed to as little light as possible.
2.2.5 Aluminium evaporation
After these steps, the samples are placed in an evaporation chamber. In order to avoid
aluminium penetration into the P3HT:PCBM, potentially creating a short circuit, the
heating is adjusted to limit the growth of the aluminium layer to 0.1 nm
s
(Jonas Weickert
2010, p. 2372). After the first 10 nm, the incoming aluminium atoms are not able to
penetrate the already grown layer of aluminium, and therefore the rate of evaporation
is increased to yield 100 nm of aluminium at the end.
18
27. 2.3 Different device architectures
PEDOT:PSS annealing EG spinning annealing PEDOT:PSS annealing
A 2500 rpm, 60 s 140 ◦
C, 10 min - - - - -
B 2000 rpm, 60 s 140 ◦
C, 10 min 3 min 1500 rpm, 30 s 140 ◦
C, 10 min - -
C 1500 rpm, 60 s 140 ◦
C, 10 min 3 min 1500 rpm, 30 s 140 ◦
C, 10 min - -
D 1500 rpm, 60 s 140 ◦
C, 10 min 3 min 1500 rpm, 30 s 140 ◦
C, 10 min 2500 rpm, 60 s 140 ◦
C, 10 min
Table 2.1: The steps necessary for different device architectures, which are explained in the
text
2.3 Different device architectures
As mentioned before, samples are treated differently in order to compare their per-
formance. Details of the treatments are listed in table 2.1. The annealing steps are
performed in order to evaporate all water out of the PEDOT:PSS as well as all the
CB out of the P3HT:PCBM. Therefore, a temperature above the boiling point of both
water, TH2O = 100◦
C and CB TCB = 131◦
C is chosen. To ensure full evaporation,
140◦
C for 10 minutes is applied in the annealing steps. As the solvent evaporates, the
polymer crystallizes, which improves efficiency.
The amount of PEDOT:PSS used for spin coating depends on the underlying surface.
On glass both with and without ITO, 200 µl is used, while for PET the amount has to
be raised to 300 µl due to comparatively bad surface wetting. For the second layer of
PEDOT:PSS on device architecture D, the higher amount of 300 µl is used as well. For
the P3HT:PCBM layer, an amount of 200 µl remains constant among all the devices.
Finally, the samples are taken into a nitrogen glove box by MBRAUN for a last annealing
step. The nitrogen atmosphere is necessary to keep the P3HT:PCBM from reacting with
ambient oxygen, which would happen very quickly due to the annealing temperature.
The samples are annealed at 140 ◦
C for 10 minutes. Afterwards they are taken directly
to be measured.
2.3.1 Partial spin coating
In order to contact the electrode and the aluminium of the organic solar cells, some of
the layers are not administered on the complete sample. An overview can be seen in
table 2.2. Etching is done with the tip of a cotton bud, soaked in hydrochloric acid
and zinc dust and prior to the four step cleaning process. For etching half the ITO off
glass, the other side is sealed with scotch tape. On ITO-PET foil, this method has to
be adapted due to the tendency of the ITO-PET foil to split. A razor blade is pressed
19
28. 2 Sample preparation
etching PEDOT:PSS PEDOT:PSS blend aluminium
A half etched partial spin coating - partial spin coating masked evaporation
B half etched partial spin coating - partial spin coating masked evaporation
C fully etched full spin coating, half removed - partial spin coating masked evaporation
D fully etched full spin coating, half removed partial spin coating partial spin coating masked evaporation
Table 2.2: The steps necessary to contact device architectures A-D
on the ITO-PET foil instead to prevent the acid from spilling into the other half of the
foil. A razor blade is also used for removing half of the PEDOT:PSS after spin coating
with the help of distilled H2O and a cotton bud.
Partial spin coating is done by leaving out 5 mm from the edge that contains the electrode
when spilling the solved polymer onto the substrate. Despite the use of an O2-plasma,
the solved polymer does not distribute itself over the whole foil due to residual surface
tension. After spinning the sample, most of the previously uncovered area is still un-
covered, therefore a contact to the layer below can be made. During the aluminium
evaporation, a hole mask covers the samples, and creates aluminium contacts for both
the top and the bottom electrode.
2.3.2 Imprinting of PEDOT:PSS
The second PEDOT:PSS layer on device architecture D can be mixed with glycerol
prior to spin coating. The aim is to soften the PEDOT:PSS layer. After spin coating
the second layer of PEDOT:PSS, it can be imprinted using a stamp made out of the
polymer PDMS.
In order to give the stamp, and in turn the PEDOT:PSS layer, an artificial structure a
compact disc (CD) is reverse engineered. The used CD is a blank polycarbonate master
whose structure has a periodicity of d = 740 nm. The CD is cut into four equal pieces,
then cleaned using the same four step process described in chapter 2.2.1.
Afterwards, the PDMS is mixed using a glass cleaned with isopropanol and a mixture
of elastomer:curing agent of 10:1. To remove residual air bubbles from the mixture, a
vacuum is created. After 5 min, a vent is opened and ambient air inserted. The process
of creating and lifting the vacuum is repeated three to five times, until all of the visible
bubbles are removed. Then, it is poured onto a cleaned CD piece. To prevent the PDMS
from flowing beyond the CD piece, a border square measuring 3 cm is used. In order
to remove newly created bubbles due to pouring the mixture, the still liquid stamp and
20
29. 2.3 Different device architectures
(a) Device architecture A (b) Device architecture B
(c) Device architecture C (d) Device architecture D
Figure 2.4: Schematic representation of the different device architectures A-D, which are
prepared with the help of tables 2.1 and 2.2. The substrate (blue), ITO
(white), PEDOT:PSS (light green), highly conductive PEDOT:PSS (dark green),
P3HT:PCBM (red) and aluminium (grey) are not to scale. The change in apparent
color of the P3HT:PCBM is due to the underlying layer.
CD are put into a vacuum chamber, and the process of creating and lifting the vacuum
is repeated another three to five times.
For the glycerol doped PEDOT:PSS layer, 30 mg
ml
glycerol is mixed with filtered PEDOT:PSS
for 20 min using water cleaned stirring magnets. Afterwards, it is spin coated on top
of the other layer of PEDOT:PSS, using the spin coat parameters of table 2.1 of device
architecture D.
After spin coating, the annealing step is modified. The sample is placed on a heating
plate at ambient temperature while the stamp is given an oxygen plasma treatment with
the same parameters as described in chapter 2.2.2, but only for 30 s. Due to the surface
functionalization of the PDMS stamp, higher imprint depths are achievable.
21
30. 2 Sample preparation
Afterwards, the PDMS stamp is placed on top and cut to the size of the sample. The
cutting is done to get a flat surface of well known area. To imprint the CD structure, a
pressure of P = 30 kPa is chosen to be comparable to the work by (Lang 2012, p. 30).
Therefore, metal blocks with a combined mass of m
m =
P · A
g
=
30 kPa · (2.2 cm)2
9.81 N
kg
= 1.5 kg (2.2)
are placed on the sample, calculated with the pressure P, the sample area A and the
acceleration constant g.
The heating plate is adjusted to a temperature of T = 80◦
C for three hours. Afterwards,
the metal blocks are removed and the sample including stamp placed on a cold metal
block to facilitate the removal of the stamp. With the help of a knife, an edge of the
stamp is lifted, after which the rest is carefully lifted as well.
The sample is then annealed at 140◦
C for 10 minutes to remove rests of glycerol and
water, and crystallize the PEDOT:PSS layer. Afterwards, it is treated the same way as
non-structured samples of device architecture D.
22
31. 3 Device Characterization
3.1 Optical microscope
In this work, optical microscopy is used to determine the size of the pixels of the solar
cells, which is needed for the exact calculation of short-circuit current, photocurrent
efficiency, as well as series and shunt resistance, explained in chapter 3.6. Furthermore,
it is used to identify the imprint of a CD on a PDMS stamp, as can be seen in chapter
5. A Pixellink CCD camera in conjunction with an Axiolab C microscope by Carl Zeiss
is used to take pictures with a resolution of 1280 pixel x 1024 pixel then processed with
the free image analysis program ImageJ version 1.46r. The scale is dependent on the
objective used, measured with a high precision grid and shown in table 3.1.
3.2 Atomic force microscope
Atomic force microscopy (AFM) is used for topological measurements of the sample sur-
faces. An ULTRASHARP NSC35/ALBS cantilever from MikroMasch with a typical
length of l = 110 µm, resonance frequency of f = 210 kHz and conically shaped tip
with a radius of r = 10 nm is used in conjunction with a JEOL JSPM 5200 atomic
magnification scales µm
pixel
1.25x 6.250
2.5x 3.125
10x 0.811
50x 0.165
100x 0.081
Table 3.1: Microscopy scales at magnifications 1.25x - 100x measured with a high precision
grid
23
32. 3 Device Characterization
force microscope. Using a piezoelectric crystal, the cantilever is vibrated near to its reso-
nance frequency and brought close to the surface without making contact, resulting in an
attractive force. The vertical position is constantly adjusted, in order for the vibration
amplitude to remain constant. Through these adjustments, a topographical image is
generated, which is then processed by the free AFM software Gwyddion version 2.3.1.
3.3 UV/Vis spectrometer
The transmittance T and reflectance R of different layers of solar cells is measured from
260 nm to 800 nm. For this purpose, a Perkin Elmer LAMBDA 900 UV/V IS/NIR
spectrometer with an integrating sphere is used. The software UV WinLab is used to
process the data. The integrating sphere has a diameter of 150 mm and is made out
of Spectralon USRS − 99 − 020 from labsphere, a highly reflective material capable of
diffusely reflecting 97 % to 99 % of the light, depending on the wavelength (labsphere
2008).
As a result, the light from a sample is scattered until it reaches the detector. Although
the absorption of Spectralon is low, the light is weakened by two orders of magnitude
due to multiple reflections. In order to avoid over emphasizing the some angles of diffuse
reflection or transmittance over other angles, two baffles are inside the sphere, as shown
in figure 3.1. They block the direct line of sight between the sample and the detector.
Through the relation
R + A + T = 1 (3.1)
the absorption A can be extracted out of the transmittance T and reflectance R, which
can then be used to compare relative layer thicknesses. For calibration purposes, the
transmittance T is first measured without a sample, measuring the ambient air only. This
spectrum defines 100 % transmittance T. To check the calibration, another measurement
is done without a sample, yielding 100 % transmittance T plus background noise. The
transmittance measurement with a sample can be seen in figure 3.1a.
To calibrate the reflectance Rcal, a reference is needed. For this purpose, the certified
reflectance standard Spectralon USRS−99−020 from labsphere is used, which has a well
known reflectance Rref (labsphere 2008). It is also the coating of the integration sphere.
The reflectance reference is measured and the result defined as its reflectance Rcal =:
24
33. 3.4 Four-point measurement
Rref . Another measurement of this sample then yields its calibration curve of 97 % to
99 % reflectance Rref plus background noise.
(a) Transmittance measure-
ment with a sample
(b) Reflectance measurement
with a sample
Figure 3.1: Transmittance and reflectance measurements with a diffuse sample (red) using an
integrating sphere made out of Spectralon from labsphere
3.4 Four-point measurement
As shown in figure 3.2, for a four-point measurement four contacts are pushed onto the
surface of a sample via a spring. A Keithley 2400 sourcemeter is used to apply a current
through the outer two probes and detect the voltage through the inner two. Due to the
separation of current and voltage, the resistance of the wires cancels out. Therefore, a
more accurate result is gained.
Figure 3.2: Four-point measurement of a sample (Akira Shimamoto & Ike 2012, after)
25
34. 3 Device Characterization
3.5 Scanning electron miscroscopy
In a scanning electron microscope (SEM), electrons are accelerated by a high voltage
up to several kV. They are focused onto a sample to be measured. There, they are
decelerated and produce a variety of signals, including secondary electrons, backscattered
electrons and photons of various wavelengths (Swapp 17.07.2013). At the SEM used in
this work, the so-called InLens modus is used, which means that secondary electrons
are detected. Images are processed using the free image editing software Paint.NET to
enhance visibility of features.
Due to the used acceleration voltage V = 1.50 kV and V = 5.00 kV, the electron speed
exceeds 10 % of the speed of light. Therefore, relativistic effects have to be taken into
account. The relativistic impulse p, using the electron charge e and mass me, the total
electron energy E, the electron rest energy E0 and the speed of light c
p =
E2 − E2
0
c
=
(V · e − mec2)2
− (mec2)2
c
(3.2)
Using the De Broglie equation, the corresponding relativistic De Broglie wavelength λe
is
λe =
h
me · v
=
h · c
(V · e + mec2)2
− (mec2)2
= 17.3 pm (3.3)
using the planck constant h and an acceleration voltage V = 5.00 kV. Since the De
Broglie wavelength of these electrons is much lower than the lowest wavelength of visible
light λe = 17.3 pm << 400 nm = λvis, it is possible to resolve much smaller features of
the sample.
3.6 Current-voltage measurement
In order to characterize the solar cells, a Solar Constant 1200 from K.H. Steuernagel
Lichttechnik, which simulates sunlight after passing through 1.5 times the terrestrial
air mass (AM), the AM 1.5 standard. AM 0 is equal to the solar constant defined
by PAM0 = 1347.9 W
m2 from ASTM (2012) which was measured from a wavelength of
280 nm to 4000 nm. The AM 1.5 takes into account that the sunlight has to traverse the
26
35. 3.6 Current-voltage measurement
atmosphere at an angle α, measured from the surface normal. This leads to an increased
distance travelled through the atmosphere, see figure 3.3.
Figure 3.3: Visual representation of the air mass (AM) models AM 0, AM 1 and AM 1.5
(Energy 10.07.2013)
AM =
1
cos α
(3.4)
Therefore, the AM 1.5 standard implies an inclination angle of αAM1.5
αAM1.5 = cos−1 1
AM
= cos−1 1
1.5
= 48.2◦
(3.5)
As mentioned at the end of section 1.2, the actual spectrum is, among others, depen-
dant on factors like air composition, actual angle of incident, or weather. The AM 1.5
standard uses the reference spectrum ASTM G-173, of the American Society for Testing
and Materials (ASTM) International (ASTM 2012). Air pollution was then taken into
account through the Simple Model of the Atmospheric Radiative Transfer of Sunshine
(SMARTS) software version 2.9.2 (ASTM 2012, chapter 4.5).
For a solar cell facing towards the equator, under the AM 1.5 angle αAM1.5 = 48.2◦
, this
spectrum yields a total irradiance PAM1.5G = 1000.4 W
m2 of direct and diffuse light. If
the diffuse light is ignored, the total irradiance drops to PAM1.5D = 900.1 W
m2 . The solar
simulator is first pre-heated for an hour to reach the described spectrum, then calibrated
using a Fraunhofer WPVS-ID 3 solar cell to an intensity of P = 1000 W
m2 . The solar
27
36. 3 Device Characterization
cells analysed in this work are capable of using diffuse light, therefore the choice of solar
simulator and intensity is made to ensure the AM 1.5 G standard measuring conditions
for solar cells. The resulting spectrum can be seen in 3.5.
Afterwards, the samples are placed on top of the solar simulator, as seen in figure 3.4,
first under dark conditions accomplished by covering a duct. The contact between the
sample and a Keithley 2400 sourcemeter is facilitated by using silver contact paste
from Ferro GmbH. To check the contact, the current is measured with the sourcemeter
sweeping the voltage from -1 V to 1 V, using a step size of 0.01 V and measuring for
0.1 s each. Afterwards, the cover is removed and the sample measured again under light
conditions. This procedure is repeated for every one of the four pixels on each solar cell.
Figure 3.4: Visual representation of the solar simulator setup used in this work (after Meier
2012, p. 49)
Solar cell characteristics are calculated using a self-made Python program that can be
found in the appendix. It takes the data points from the measurement and first plots the
curves. Afterwards, it fits one linear polynomial each through the five points surrounding
the voltage V = 0 and the point of current I = 0. Short circuit current ISC =: I(V = 0)
and open circuit voltage VOC =: V (I = 0) are defined respectively. The inclines of the
28
37. 3.6 Current-voltage measurement
polynomials are defined as the shunt resistance
Rshunt =
1
dI
dV V =VOC
(3.6)
and the series resistance
Rseries =
1
dI
dV V =0
(3.7)
In order to find the maximum power point (MPP), all pairs of current and voltage are
multiplied. The resulting maximum defines the pair (VMPP , IMPP ) used to calculate the
fillfactor (FF).
FF =
VMPP · ISC
VOC · ISC
(3.8)
The efficiency η is calculated via
η =
Pout
Pin
=
VMPP · ISC
A · Pin
(3.9)
using the area of the pixel A and the intensity of incoming radiation Pin. The plots are
saved as a vector graphic in the portable document format .pdf and portable network
graphic .png, examples are figures 4.6 and 4.7. A more elaborate description on features
and usage can be found in the appendix in the form of a readme file as well as the source
code written in python 2.7.
Figure 3.5: Spectrum of a Solar Constant 1200 from K.H. Steuernagel Lichttechnik solar
simulator (yellow) compared to the global radiation (black) in arbitrary units (after
Lang 2012)
29
38.
39. 4 Towards ITO-free organic solar cells
In order to realize ITO-free organic solar cells, two device architectures based on highly
conductive PEDOT:PSS (H-PEDOT:PSS) as a replacement for ITO are chosen. For
comparison, two ITO-based device architectures are built as well.
4.1 Different device architectures on glass
The device architectures detailed in section 2.3 are prepared on glass.
4.1.1 UV/Vis spectroscopy
Figure 4.1: Visual representation of the splitting intensities of an incoming beam falling on a
sample (light red). It can either be absorbed, reflected or transmitted. If absorbed,
it can also create photoluminescence and emit light of a lower wavelength. (after
Miguel A. P´erez & Arias 2013)
For the UV/Vis spectroscopy of both glass and PET samples, partial spin coating is
changed to full spin coating, as well as leaving out the aluminium evaporation. That way,
the transmittance T and reflectance R can be measured in the whole sample, yielding
a more accurate result. However, it has to be kept in mind that the highly reflective
31
40. 4 Towards ITO-free organic solar cells
aluminium back layer would reflect most of the light. Therefore, the absorption in the
active layer, and in the other layers, as well as the reflectance would be higher while
transmittance through the solar cell would drop to near zero.
Overall, 10 samples are measured to reconstruct the transmittance T and reflectance R,
and therefore the absorption A, in all layers. The underlying equation
1 = R + A + T (4.1)
is based on the assumption that an incoming photon can either be reflected, absorbed or
transmitted, as shown in figure 4.1. However, photoluminescence, which is also shown
in figure 4.1 is counted towards both transmittance and reflectance, because the photon
can be emitted in either direction.
3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0
0
2 0
4 0
6 0
8 0
1 0 0
absorption[%]
w a v e l e n g t h [ n m ]
Colour
Devicearchitecture
Substrate
Electrode
Eletrconblockinglayer
Blend
── Glass
── Glass ITO
── A Glass ITO PEDOT:PSS
── A Glass ITO PEDOT:PSS P3HT:PCBM
── B Glass ITO H-PEDOT:PSS
── B Glass ITO H-PEDOT:PSS P3HT:PCBM
── C Glass H-PEDOT:PSS
── C Glass H-PEDOT:PSS P3HT:PCBM
── D Glass H-PEDOT:PSS PEDOT:PSS
── D Glass H-PEDOT:PSS PEDOT:PSS P3HT:PCBM
Figure 4.2: Absorption of different device structures on glass calculated via transmittance and
reflectance using UV/Vis spectroscopy
Figure 4.2 shows the calculated absorption spectra A(λ). Figures 1 and 2 in the appendix
show the measured reflectance R and transmittance T. Each device architecture A-D
is built once with and once without P3HT:PCBM, plus one ITO glass and an etched
glass. H-PEDOT:PSS is used to abbreviate highly conductive PEDOT:PSS. Two sam-
ples based on the same device architecture A-D are plotted in the same color. They are
distinguishable at 400 nm - the higher absorbing four curves belong to the samples with
blend.
The curves with blend show an increased absorption A by about 20 to 40 % compared
32
41. 4.1 Different device architectures on glass
to the corresponding samples without blend. For all device architectures A-D, the re-
flectance R increases for wavelengths bigger than 400 nm, while the transmittance T
decreases for wavelengths between 300 nm to 650 nm. Therefore, the increase in ab-
sorption A by adding blend is due to a decrease in transmittance T, while the increased
reflectance partly mitigates the increase in absorption A.
The ITO layer on glass contributes only very slightly to the absorption, only for wave-
lengths below 350 nm, there is a significant difference due to a much higher reflectance of
etched glass. The increase in reflectance by the blend layer might be due to the detector
of the used UV/Vis spectrometer not being able to discriminate different wavelengths.
Therefore, decaying excitons in the blend that send out a photon of lower energy, pho-
toluminescence are detected as well and counted towards the reflectance.
Another effect to contribute to the higher reflectance is a change in the refractive in-
dex n (λ) of the blend, see figure 4.3. Since the refraction index n (λ) was measured by
Florent Monestiera (2007), they have used different production parameters like annealing
temperature. Therefore, only general tendencies in the refraction index are valid.
The refractive index n (λ) rises from its minimum n (440 nm) = 1.6 to n (600 nm) = 2.1,
then falls to n (800 nm) = 1.9. These extrema correspond well to the measured re-
flectance R, which increases for device architecture C from RC (400 nm) = 12 %
to RC (640 nm) = 24 %, then decreases to RC (800 nm) = 18 %. Only device archi-
tecture D behaves somewhat differently as it has an overall decreased reflectance com-
pared to the other device architectures B-D but a higher reflectance for 750 to 800 nm.
The refraction index of PEDOT:PSS is at about n (400 nm) = 1.6 and decreases to
about n (800 nm) = 1.5 (Leif A.A Pettersson 2002, p. 146), but is dependent on the
mixture ratio of PEDOT:PSS, and treatment methods. Therefore, the refraction index of
blend is similar to that of PEDOT:PSS for smaller wavelengths λ = 400 . . . 500 nm, but
larger for bigger wavelengths. Lastly, due to the post treatment method, H-PEDOT:PSS
behaves differently than PEDOT:PSS.
Overall, the dominating effect that leads to higher reflectance R (λ) after a layer of blend
is administered, is not the photoluminescence, but the refractive index n (λ).
For wavelengths λ = 330 nm to λ = 610 nm, the absorption A is higher with blend than
without, see figure 4.2. However, for higher wavelengths, the absorption A of device
architectures A-C lies below the samples without blend. This is due to the increased
33
42. 4 Towards ITO-free organic solar cells
reflectance for higher wavelengths because of the increased refractive index n (λ). There-
fore, the blend is not able to absorb incoming photons, since they are reflected at the
PEDOT:PSS - P3HT:PCBM barrier due to the higher refraction index of P3HT:PCBM.
All device architectures A-D show higher absorption A for wavelengths λ = 330 . . . 610 nm
in samples with blend, although the reflectance R = 10 . . . 20 % is still substantial.
The fraction of intensity that arrives inside the blend and is absorbed can be calculated,
for which the following equation is used
PP3HT:PCBM =
TP3HT:PCBM
TPEDOT:PSS − (RP3HT:PCBM − RPEDOT:PSS)
(4.2)
with the transmittance TPEDOT:PSS and reflectance RPEDOT:PSS of the corresponding
sample without blend, and the transmittance TP3HT:PCBM and reflectance RP3HT:PCBM
with blend. It is based on the assumption that the fraction of light that is not reflected
is either absorbed or transmitted. However, photoluminescence is neglected since the
fraction of the reflectance RP3HT:PCBM and transmittance TP3HT:PCBM that is due to
photoluminescence is not determined.
Figure 4.3: Refractive index n (λ) (solid line) and extinction coefficient k (λ) (circles) of
P3HT:PCBM measured by spectrometric ellipsometry (after Florent Monestiera
2007)
Figure 4.4 shows the absorbed light in the blend layer as a fraction of the light intensity
that arrives inside the blend, calculated via equation (4.2). It also shows the intensity
of light that is transmitted into the blend layer as a fraction of the incoming beam
intensity. Corresponding curves are shown in the same color and can be distinguished
at the wavelength λ = 500 nm, at which the transmittance curves are on top. For
34
43. 4.1 Different device architectures on glass
4 0 0 5 0 0 6 0 0 7 0 0 8 0 0
0
2 0
4 0
6 0
8 0
1 0 0
1 2 0
1 4 0
0
2 0
4 0
6 0
8 0
1 0 0
1 2 0
1 4 0
transmittancetoblendlayer[%I0
]
absorptionofinjectedlightinblend[%]
w a v e l e n g t h [ n m ]
t r a n s m i t t a n c e
a b s o r p t i o n
Colour
Devicearchitecture
Substrate
Electrode
Eletrconblockinglayer
Blend
── A GLASS ITO PEDOT:PSS P3HT:PCBM
── B GLASS ITO H-PEDOT:PSS P3HT:PCBM
── C GLASS H-PEDOT:PSS P3HT:PCBM
── D GLASS H-PEDOT:PSS PEDOT:PSS P3HT:PCBM
Figure 4.4: Ratio of absorption of injected light into blend and intensity of light arriving
in blend of different device structures on glass calculated via transmittance and
reflectance using UV/Vis spectroscopy
wavelengths λ = 400 . . . 550 nm, a clear trend is visible; out of lower injected intensity
follows higher chance of absorption. For wavelengths λ = 650 . . . 800 nm, the chance of
absorption increases steadily, even beyond 100 %, while the injected intensity drops.
The chance of absorption beyond 100 % for wavelengths λ = 650 . . . 800 nm is due
to an increase in reflectance RP3HT:PCBM combined with a decrease in transmission
TP3HT:PCBM . The critical wavelengths of 4.4 are λ = 350 . . . 660 nm, because the band
gap is located in this range. As mentioned in chapter 2.1, the band gap of P3HT lies at
about 1.9 eV. Therefore the wavelength at which P3HT is capable of creating a photo
current is
λ =
h · c
EPh
=
6.6261 · 10−34 J
s
· 2.9979 · 108 m
s
1.9 eV
= 650 nm (4.3)
calculated with the Planck constant h, the speed of light c and the energy of the photon
EPh. For wavelengths λ = 350 . . . 500 nm, the chance of absorption might be modified by
the photoluminescence, therefore yielding a less accurate result. For wavelengths lower
than λ = 320 nm, equation (4.2) becomes impracticable, due to almost no transmittance
T and consequently high uncertainty.
Overall, the graphs of chance of absorption and injected intensity are inversely correlated
for wavelengths λ = 500 . . . 800 nm, but whether they are causally linked is uncertain.
For increasing intensities, the chance of absorption decreases. However, the data can be
explained by the polymer having a limited absolute capacity for absorbing photons.
35
44. 4 Towards ITO-free organic solar cells
4.1.2 Current-voltage measurement
(a) Camera picture of the solar cell with silver
paste, modified by increased brightness and
contrast. As mentioned in chapter 2.2.1, the
cell measures 2.2 x 2.2 cm
(b) Microscopy image under 1.25x magnification
of a pixel of an organic solar cell. The black
dots on the yellow aluminium are blisters that
could come from the blend annealing step.
Figure 4.5: Images of a finished organic solar cell
For this measurement, every device architecture is prepared twice on glass. Right after
the last annealing step, the samples are taken to the pre-heated solar simulator. The
measured current-voltage curves can be seen in figure 4.6. The order in which the
samples are measured is A-1, B-1, C-1, D-1, A-2, B-2, C-2 and then D-2. The evaluated
characteristics from figure 4.6 are plotted in figure 4.7. Every symbol stands for a pixel
of a solar cell, the white symbol for the average of all pixels of one solar cell.
To measure the size of each pixel, microscopy at 1.25x magnification is used. Figure
4.5 shows a camera picture (a) and a microscopy image (b) of a finished organic solar
cell. In the microscopy image, the yellow area is measured, up to the line at which the
underlying electrode ends and the material changes color.
Device architecture A shows the best diode behaviour of all device architectures. It
has the highest fillfactor FF, shunt resistance Rshunt, lowest serial resistance Rseries and
a short circuit current ISC nearly as high as device architecture B. It also shows the
highest open circuit voltage, albeit only by a small margin. All these characteristics lead
to the highest efficiency η of all the device structures.
36
45. 4.1 Different device architectures on glass
Curves of device architecture B show no diode behaviour in the lower right quadrant of
the plot. This could be due to a comparatively low shunt resistance, which tilts the left
side of curve down and the right side up. The curve does show an increase in its incline at
the open circuit voltage VOC of the other device architectures. The subsequently lower
fillfactor FF, and open circuit voltage VOC lead to the lower efficiency of the device
architecture. However, it has the highest short circuit current ISC.
The reason of the low shunt resistance Rshunt lies in the post treatment process of the
PEDOT:PSS layer. Due to the choice of spin coat parameters, the H-PEDOT:PSS layer
of device architecture B is thinner than the H-PEDOT:PSS layer of device architecture
C. The post treatment process might create a layer of PEDOT:PSS of uneven thickness.
Therefore, the layer might be too thin in some places, leading to a very low shunt
resistance Rshunt.
Device architecture C shows some diode behaviour, albeit not as good as device architec-
ture A. Especially the lower incline produces a higher series resistance Rshunt, compared
to device architecture A. Consequently, the fillfactor decreases, which combined with a
lower short circuit current ISC leads to a lower efficiency η. The higher series resistance
is due to the use of H-PEDOT:PSS as an electrode instead of ITO, since H-PEDOT:PSS
has about three times the resistivity of ITO.
The second ITO-free device architecture, D, shows nearly equal diode behaviour to
device architecture C. Sample D-2 is first measured without silver paste, then again
with silver paste. It is plausible to assume that it would have performed similar to
sample D-1 at the first time with silver paste. Even if it is assumed that sample D-2
would be equal to sample D − 1, the short circuit current ISC is the lowest, while the
series resistance Rseries the highest and the fillfactor FF rather low. Therefore, the
efficiency η of this device architecture D is lower than C.
For device architecture D, the high series resistance Rseries comes from the inclusion of
a second PEDOT:PSS layer that is not highly conductive. Also, the contact between
the PEDOT:PSS and the H-PEDOT:PSS layers on the one side, and the PEDPT:PSS
and P3HT:PCBM layers on the other side leads to a higher series resistance Rseries.
All in all, the best performing device architecture is A, while the best ITO-free is C.
Assuming sample D-2 would have performed similar to sample D-1, every solar cell
prepared on the basis of the same device architecture shows the same short circuit
current, open circuit voltage, fillfactor, resistances and efficiency within the variance of
37
46. 4 Towards ITO-free organic solar cells
the pixels. Therefore, the different device architectures have a reproducible effect on the
device performance.
To show the reproducibility of this result, another batch of six samples is prepared on
glass1
. The order in which the samples are measured is A-3, B-3, C-3, D-3, A-4, and
then D-4. The results can be seen in the appendix, figures 5 and 6.
However, one pixel of device architecture B shows a very good diode behaviour, and a
shunt resistance Rshunt comparable to device architecture D. Consequently, the efficiency
increases, and even reaches 3.3 %. As mentioned before, the reason for the low shunt
resistance Rshunt is due to the post treatment process. The uneven surface of device
architecture B might be thick enough for this individual pixel, resulting in the highest
measured efficiency.
1
two more samples based on device architectures A and D are prepared on PET. However, these show
efficiencies below 0.15 %
38
47. 4.1 Different device architectures on glass
0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
voltage (V)
12
10
8
6
4
2
0
2
currentdensity(mA/cm2)
1.0 0.5 0.0 0.5 1.0
voltage (V)
50
0
50
100
150
currentdensity(mA/cm2)
Figure 4.6: Current-voltage graphs of four different device architectures detailed in chapter
2.3 built on glass measured on a solar simulator. The samples A-1 (yellow), A-2
(orange), B-1 (red), B-2 (brown), C-1 (purple), C-2 (blue), D-1 (green) and D-2
(olive) are plotted from -1 V to 1 V (lower picture) and from -0.1 V to 0.7 V.
39
48. 4 Towards ITO-free organic solar cells
A-1
A-2
B-1
B-2
C-1
C-2
D-1
D-2
0.0
0.5
1.0
1.5
2.0
2.5
3.0
efficiency(%)
A-1
A-2
B-1
B-2
C-1
C-2
D-1
D-2
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
opencircuitvoltage(V)
A-1
A-2
B-1
B-2
C-1
C-2
D-1
D-2
10
9
8
7
6
5
4
shortcircuitcurrent(mA/cm2)
A-1
A-2
B-1
B-2
C-1
C-2
D-1
D-2
20
30
40
50
60
fillfactor(%)
A-1
A-2
B-1
B-2
C-1
C-2
D-1
D-2
0.0
0.1
0.2
0.3
0.4
0.5
shuntresistance(kΩ·cm2)
A-1
A-2
B-1
B-2
C-1
C-2
D-1
D-2
0.00
0.05
0.10
0.15
0.20
serialresistance(kΩ·cm2)
Figure 4.7: Characteristics of the four device architectures detailed in chapter 2.3 built on
glass calculated from the data of the graphs of figure 4.6 with a self made program
40
49. 4.2 Different device structures on PET
4.2 Different device structures on PET
The same device architectures prepared on glass are now prepared on PET as well.
Despite five attempts to build device architectures A-D on PET, using both ITO-PET
foil and PET foil, none of the samples reached efficiencies higher than 0.15 %. This
chapter investigates the causes of the different performance of device architectures A-D
on PET and glass.
4.2.1 Prerequisites for annealing and cleaning of ITO-PET foils
For device architectures B-D, H-PEDOT:PSS is used. In order to increase the conduc-
tivity a post treatment step with EG is performed. To make sure the PET foil is not
damaged by the EG, the post treatment step is done with a pristine PET foil. It is
heated up to 140◦
C for 20 minutes, bathed in EG for 6 minutes and then heated up
again for 20 minutes. The time for each of these steps is twice the length of normal post
treatment to make any ill effects visible. However, a visual inspection of the foil yields
no damage to the foils.
For the cleaning steps described in section 2.2.1, the foil needs to be resistent against
solvent. This is checked with the pristine ITO-PET foil, yielding two results. If the foil
is not thermally treated beforehand, it splits completely, even for an increased foil size of
4 cm squares. However, if the ITO-PET foil is first annealed for 1.5 hours, then cleaned,
there is only some splitting around the edges of the foil. This effect reaches about 2 mm
from the edge, which does not affect the rest of the production process.
4.2.2 Four point measurement
To measure the conductivity of the H-PEDOT:PSS layer, a four point measurement is
done. For this four point measurement, PEDOT:PSS is spin coated on top of glass and
PET, with 1500 rpm for 60 s. It is further treated with EG to produce H-PEDOT:PSS.
Contrary to the method of adding glycerine on a glass slide, 7.5 µl of H2O is used
here instead. The spin coating method is otherwise unaffected, since this amount of
distilled H2O holds the sample in place as well. The following annealing step, however,
is somewhat affected due to bending of the foils. The results of two batches can be
41
50. 4 Towards ITO-free organic solar cells
seen in figure 4.8. The resistivity can not be calculated, due to a lack of thickness
measurements. The used 4-point measurement setup is not calibrated, which is why
arbitrary units are chosen.
The sheet resistance of ITO on glass is measured in both batches and very reproducible,
as all 6 measurements have a spread of less than 3 %. Samples S-1 - S-3 are measured
in a first batch, S-4 - S-10 in a second. Assuming the same layer thickness, the samples
with the lowest sheet resistance have about 6 times the sheet resistance of ITO on glass,
as can be seen in figure 4.8.
According to Seok-In Na (2008, p. 4062), these values are reasonable for H-PEDOT:PSS
in comparison to ITO. Therefore, the post-treatment process is shown to work on PET
and H-PEDOT:PSS is produced with a conductivity of two to three orders of magnitude
higher than pristine PEDOT:PSS.
I T O S - 1 S - 2 S - 3 S - 4 S - 5 S - 6 S - 7 S - 8 S - 9 S - 1 0
1 , 0 6
1 , 0 8
6
8
1 0
1 2
1 4
1 6
1 8
sheetresistance[a.u.]
s a m p l e
Figure 4.8: Sheet resistances of H-PEDOT:PSS spin coated on PET, labelled samples S-1 -
S-10 and ITO on glass in arbitrary units
4.2.3 UV/Vis spectroscopy
Analogous to chapter 4.1.1, 10 samples are measured to reconstruct the transmittance T
and reflectance R, and therefore the absorption A, in all layers. The only difference is
42
51. 4.2 Different device structures on PET
the substrate, which is ITO-PET foil for device architectures A and B, and etched ITO-
PET foil for device architectures C and D. Figure 4.9 shows the calculated absorption
spectra A(λ). Figures 3 and 4 in the appendix show the measured reflectance R and
transmittance T.
3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0
0
2 0
4 0
6 0
8 0
1 0 0
absorption[%]
w a v e l e n g t h [ n m ]
Colour
Devicearchitecture
Substrate
Electrode
Eletrconblockinglayer
Blend
── PET
── PET ITO
── A PET ITO PEDOT:PSS
── A PET ITO PEDOT:PSS P3HT:PCBM
── B PET ITO H-PEDOT:PSS
── B PET ITO H-PEDOT:PSS P3HT:PCBM
── C PET H-PEDOT:PSS
── C PET H-PEDOT:PSS P3HT:PCBM
── D PET H-PEDOT:PSS PEDOT:PSS
── D PET H-PEDOT:PSS PEDOT:PSS P3HT:PCBM
Figure 4.9: Absorption of different device structures on PET measured calculated via trans-
mittance and reflectance using UV/Vis spectroscopy
In general, the samples using a PET substrate are performing similar to those on glass,
with some important differences.
On PET, device architecture D absorbs less for wavelengths λ = 620 . . . 800 nm with
blend than without blend, which is not the case on glass.
On glass, the highest absorbing device architecture without blend for wavelengths λ =
300 . . . 600 nm, A, is also the lowest absorbing one with blend. This trend also holds
for for the lowest absorbing one without blend, as it is the highest absorbing one with
blend. However, on PET, this trend is no longer visible.
For all PET samples, there is some amount of oscillation in the reflectance R and trans-
mittance T that steadily increases starting at about λ = 500 nm. This is be due to the
usage of a much thinner substrate. As mentioned in chapter 2.1, the glass substrates
have a thickness of d = 1.1 mm, while the ITO-PET substrate is only d = 45 µm thin.
For reflectance T, the light can be reflected on the surface of the substrate, and on the
first interface inside the substrate. The two beams interfere, creating the oscillation
pattern visible in 3.
43
52. 4 Towards ITO-free organic solar cells
Most importantly, the average absorption A is higher for samples built on PET than for
glass. This is due to both a decrease in transmittance T of about 10 % and an increase
in reflectance R of about 5 %. This is not due to a different transmittance T of the sub-
strates, because those are similar. The reflectance R, however, shows some differences -
the etched glass sample shows a high reflectivity for wavelengths λ = 260 . . . 350 nm.
The characteristic form of the curves of absorption of samples with blend on glass and
PET can be seen in figures 4.2 and 4.9. Both rates of absorption go up at a wavelength
of 640 nm, therefore the band gap of P3HT remains unchanged.
4 0 0 5 0 0 6 0 0 7 0 0 8 0 0
0
2 0
4 0
6 0
8 0
1 0 0
1 2 0
1 4 0
1 6 0
0
2 0
4 0
6 0
8 0
1 0 0
1 2 0
1 4 0
1 6 0
a b s o r p t i o n
t r a n s m i t t a n c e
transmittancetoblendlayer[%I0
]
absorptionofinjectedlightinblend[%]
w a v e l e n g t h [ n m ]
Colour
Devicearchitecture
Substrate
Electrode
Eletrconblockinglayer
Blend
── A PET ITO PEDOT:PSS P3HT:PCBM
── B PET ITO H-PEDOT:PSS P3HT:PCBM
── C PET H-PEDOT:PSS P3HT:PCBM
── D PET H-PEDOT:PSS PEDOT:PSS P3HT:PCBM
Figure 4.10: Ratio of absorption of injected light into blend and intensity of light arriving
in blend of different device structures on PET calculated via transmittance and
reflectance using UV/Vis spectroscopy
4.2.4 Usage of different blend without annealing step
With the blend P3HT:PCBM, no solar cells of are produced on PET with a perfor-
mance comparable to glass. One reason might be the necessary last annealing step of
P3HT:PCBM, at which the foil could still bend despite of capillary force of glycerol
described in chapter 2.2.3.
Therefore, a different blend, poly([4,8-bis[(2-ethylhexyl)oxy]benzo[1,2-b:4,5-b’]dithiophene-
2,6-diyl][3-fluoro-2-[(2-ethylhexyl)carbonyl]thieno[3,4-b]thiophenediyl]) PTB7:PC70BM is
used that does not need annealing. Three samples are prepared on basis of device ar-
chitecture C. Two use etched ITO-PET foil as a substrate, and the third one PET foil.
44
53. 4.2 Different device structures on PET
All three samples show efficiencies below 0.1 %, despite other samples on glass working
in the same batch. Consequently, the last annealing step is either not the reason, or at
least not the only reason for the comparatively low efficiencies of device architectures
A-D on PET.
4.2.5 Scanning electron microscopy
For scanning electron microscopy (SEM), 3 samples are measured. The SEM images
of H-PEDOT:PSS (H-PEDOT:PSS) spin coated on etched ITO-PET foil are shown in
4.11a-c. For comparison, an image of H-PEDOT:PSS spin coated on glass is shown in
4.11b. The images of H-PEDOT:PSS in figures 4.11d and d are taken in about the same
magnification.
On glass, there is a homogeneous layer of H-PEDOT:PSS visible, while on the etched
ITO-PET foil there are some droplets apparent. These droplets also appear on other
magnifications, as shown in figure 4.11a and b. Therefore, there dewetting of the H-
PEDOT:PSS is taking place. However, the surface around the droplets appears mostly
intact though there are some holes visible.
Figures 4.11e and f show the edge of the same spin coated sample as shown in fig-
ures 4.11a, c and d. Due to the spin coating, the edge is sometimes not covered with
PEDOT:PSS. The H-PEDOT:PSS reaches to the right of figure 4.11e. The surface of
these etched ITO-PET foils shows an uneven topography with some spikes spread out.
Therefore, the H-PEDOT:PSS layer might be pierced in some places, which could lead to
short circuits between the H-PEDOT:PSS and the aluminium electrode. This might be
the reason of the low performance of device architectures C and D on etched ITO-PET
foil.
Comparing the uncovered, etched ITO-PET foil shown in figures 4.11e and f, and the
surface surrounding the droplets in figures 4.11a-c, they look contrasting. Therefore,
the surface surrounding the droplets is not the underlying etched ITO-PET foil, but
H-PEDOT:PSS that has not dewetted. Consequently, there is a mostly intact H-
PEDOT:PSS layer on top of device architectures C and D. Due to the formation of
the droplets, the H-PEDOT:PSS layer becomes thinner and more uneven.
45
54. 4 Towards ITO-free organic solar cells
Figures 4.12a-c show the surface of an ITO-PET foil. The ITO shows a pattern similar to
the Voronoi polygons. This pattern might be due to a crystalline growth of the ITO on
top of the under coating of the PET foil. If there is a significant mismatch in the lattice
constants of the under coating and the ITO, the result would be a columnar growth
mode. A small tilt between crystallites of the ITO could produce the pattern shown
in figures 4.12a and b. This tilt would then increase the surface roughness, therefore
influencing the spin coating.
Figure 4.12b shows a black spot of a diameter d = 3 µm, which are also visible in figure
4.12c. Therefore, these spots are spread around the whole ITO-PET foil. In the top
right corner, figure 4.11f shows a spike with a diameter of d = 1 µm. It might be that
those spikes are not due to etching, but due to the manufacturing process of this ITO-
PET foil. If this is the case, they could interfere with the growth of the ITO layer in
the dark spots of figure 4.12c.
46
55. 4.2 Different device structures on PET
(a) H-PEDOT:PSS spin coated on an etched
ITO-PET foil under 5 000x magnification and
a working distance of 1.5 mm
(b) H-PEDOT:PSS spin coated on an etched
ITO-PET foil under 40 000x magnification
and a working distance of 1.5 mm
(c) H-PEDOT:PSS spin coated on an etched
ITO-PET foil under 40 000x magnification
and a working distance of 1.5 mm
(d) H-PEDOT:PSS spin coated on glass under 44
380x magnification and a working distance of
1.0 mm
(e) Etched ITO-PET foil with some H-PEDOT
on the right under 4 000x magnification and
a working distance of 1.5 mm
(f) Etched ITO-PET foil under 10 000x magnifi-
cation and a working distance of 1.9 mm
Figure 4.11: Scanning electron microscopy images of H-PEDOT:PSS and etched ITO-PET
with an acceleration voltage of 1.50 kV
47
56. 4 Towards ITO-free organic solar cells
(a) ITO-PET foil under 40 000x magnification, a
working distance of 3.6 mm and acceleration
voltage of 5.00 kV
(b) ITO-PET foil under 3 240x magnification, a
working distance of 3.6 mm and acceleration
voltage of 5.00 kV
(c) ITO-PET foil under 150x magnification, a
working distance of 3.6 mm and acceleration
voltage of 1.50 kV. The cuboids visible in this
picture are measuring artefacts due to previ-
ous measurements.
(d) ITO on glass
Figure 4.12: Scanning electron microscopy images of ITO on glass (Cao Xia 2010, p. 1872)
and ITO-PET foil
48
57. 5 Towards structured organic solar
cells
As can be seen in figures 2 and 4, 15 % to 25 % of the incoming light is not transmitted
through the substrate and PEDOT:PSS layer. To increase the transmittance, artificial
structuring can be used, as discussed in chapter 1.2. In order to realize structured
organic solar cells, a PDMS stamp based on a CD is used.
5.1 Structured devices on PET
The manufactured stamp is shown in figure 5.1a. The structure size of CDs is near
the optical wavelengths, therefore they refract light, making the individual wavelengths
visible. This can be seen in figure 5.1a, in which the stamp is bent both to show its
flexibility and to increase the visibility of the reflections. The PDMS stamp is used
to imprint its structure into glycerine doped PEDOT:PSS. Light microscopy at 100x
magnification is used to view the resulting structure, which can be seen in figure 5.1b.
For the imprinted PEDOT:PSS, the structure was not visible to the naked eye.
In order to measure the depth of these structures, AFM is used. The resulting topo-
graphical images can be seen in figure 5.2. The spots of locally highly increased height
might be due to dust. The image is recorded at an angle close to 45◦
in order to avoid
measuring artefacts. An average is taken over lines perpendicular to the structure. This
profile is shown in figures 5.2b and d, for figures to their respective left. Out of the
profiles, an average structure depth of (8 ± 1) nm is extracted.
Earlier works by Lang (2012) have shown that the average structure depth for an im-
printed CD structure can be up to (25.2 ± 1.9) (Lang 2012, p. 38). However, a glass
substrate was used in that work. The PET foil used in this work bends under pressure,
49
58. 5 Towards structured organic solar cells
(a) Light reflections of a stamp with a CD structure
made out of the polymer PDMS, modified by
increased contrast and decreased brightness
(b) Microscopy image of imprinted CD structure
on glycerine doped PEDOT:PSS on PET un-
der 100x magnification, modified by increased
contrast
Figure 5.1
and although a polished metal surface is used, there might still be a substantial amount
of roughness due to dust for example. This would lead to a bending of the PET foil
and an uneven pressure distribution. The pressure might be too low in most places to
achieve the structure depth reported by Lang (2012, p. 38). A solution would be to clean
and blow dry the metal surface right before a sample is placed on it. Also, the overall
pressure could be increased to achieve a high structure depth in the whole sample.
50
59. 5.1 Structured devices on PET
(a) 4 µm topography (b) profile cut of 4 µm topography
(c) 10 µm topography (d) profile cut of 10 µm topography
Figure 5.2: AFM images of imprinted CD structure on glycerine doped PEDOT:PSS on PET,
modified by subtracting a one and a two dimensional polynomial
51
60.
61. 6 Summary and outlook
In this thesis, organic solar cells based on the polymer P3HT are examined. Four different
device architectures are manufactured via spin coating to realize ITO-free organic solar
cells. As a replacement for ITO, highly conductive PEDOT:PSS is analysed. Two device
architectures use ITO, while two use H-PEDOT:PSS as an electrode.
They are analysed using current-voltage measurements and UV/Vis spectroscopy. ITO-
free organic solar cells are realized on glass with efficiencies of 1.75 % compared to
2.75 % with ITO. Device architecture A shows the best reproducible efficiency, while
device architecture C yields the highest efficiencies for ITO-free organic solar cells.
As a replacement for the widely used glass substrates, PET foil is chosen. Some problems
with the manufacturing process for the substrate PET are solved, but no solar cell is
produced of a performance comparable to the substrate glass. SEM reveals defects in
the used ITO-PET foil which could short circuit the solar cells.
A CD structure is transferred into a PEDOT:PSS layer on a PET foil using plasticizer
assisted soft embossing and then analysed using atomic force microscopy. The structure
depth of (8 ± 1) nm in glycerol doped is less than (25.2 ± 1.9) reported by Lang (2012,
p. 38), due to the use of a flexible PET substrate instead of rigid glass. By optimizing
the parameters of temperature, annealing time and pressure, the structure depth can be
increased.
Instead of using a CD as the artificial structure of choice, black silicon can be used,
which shows promising properties for structuring (J.S. Yoo 2006). It has already been
proven to work for glass substrates, as can be seen in figure 7 in the appendix (Heller
2013)
In order to improve the performance of device architectures B, C and D, the following
suggestion is made. Device architecture B suffers from a low shunt resistance Rshunt,
that might be due to a too thin H-PEDOT:PSS layer. Therefore, instead of spinning
53
62. 6 Summary and outlook
at 2000 rpm, 1500 rpm could be used. The increased layer thickness would lead to a
slightly decreased transmittance through the PEDOT:PSS layer, but it could increase
the shunt resistance Rshunt, which would contribute to a much higher efficiency of device
architecture B.
To further increase the performance of device architecture C, a thicker layer of H-
PEDOT:PSS can be used, but that would decrease the transmission to the blend layer.
Alternatively, a different post treatment method that achieves higher conductivities than
EG could be used.
As mentioned before, device architecture D has a high series resistance Rseries due to
the second PEDOT:PSS layer. The goal of this second PEDOT:PSS layer is to be able
to structure it, see chapters 2.3.2 and 5. The post treatment method used in this work
would destroy the structure, but a post-treatment method that preserves the structure
would increase the performance of device architecture D.
The UV/Vis spectrometer used in this work was not able to account for photolumines-
cence. To increase the accuracy of the results, the amount of photoluminescence can be
derived from using a filter that covers the detector. With proper calibration, the filter
does not interfere with the measurement. A comparison of the results with and without
the filter yields the amount of photoluminescence.
Another way of increasing the accuracy of the UV/Vis spectrometer is to measure re-
flectance and transmittance simultaneously in one measurement. To achieve this, the
sample is placed on a holder inside the integrating sphere of the UV/Vis spectrometer,
thus both the light of the reflectance and the transmittance is detected.
In order to circumvent the difficulties associated with the spin coating, etching, and
manufacturing defects, in the future a Flextrode provided by DTU Energy Conversion
will be used instead (DTU 24.07.2013). It consists of PEDOT:PSS on a PET foil,
available in different sheet thicknesses. It will also increase the comparability of research
based on PEDOT:PSS on PET, contributing to the goal of ITO-free, flexible organic
solar cells.
54
63. List of Figures
1.1 Variations of the solar constant from 1874 to 2007 for an 81 day average
(red) and a daily average (yellow) (after A. D. Crouch & Paquin-Ricard
2008, p. 737) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 A comparison of the spectrum of a black body at 5250 K (grey), the
sunlight at the top of the atmosphere (yellow) and radiation at sea level
(red) together with the corresponding absorbing molecules (purple) (after
Rohde 2007) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Annual ground solar irradiance for every land mass on earth averaged
over 1983 until 2005 in [ W
m2 ] (after Steiner 2008) . . . . . . . . . . . . . . 4
1.4 Best research-cell efficiencies of different types of solar cells (after NREL
10.07.2013b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5 Orbitals of ethene: a) individual atomic orbitals, b) one molecular orbital
and c) energy of bonding- and antibonding orbitals (Palumbiny 2011, p. 7) 6
1.6 Shockley-Queisser limit calculated using an AM 1.5 solar spectrum instead
of a black body spectrum (after Byrnes 2011) . . . . . . . . . . . . . . . 7
1.7 The basic process inside an organic solar cell exciton generation (a), trans-
port (b), dissociation (c) and charge carrier transport (d) (Meier 2012,
p. 6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.8 Band structure of the polymers PEDOT:PSS and P3HT:PCBM, ITO
and aluminium. The top line of each column represents the HOMO, the
bottom line the LUMO. (after Seong Kyu Janga 2012, p. 427) . . . . . . 9
1.9 Bilayer junction (left), bulk heterojunction (middle) and interdigitated
junction (right) between the donor and acceptor material (after Palumbiny
2011, p. 21) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.10 Equivalent circuit model of organic solar cells with the shunt resistance
Rshunt, serial resistance RSerial, saturation current IS, photo current IPh
and ideality factor n (Palumbiny 2011, p. 37) . . . . . . . . . . . . . . . 10
55
64. List of Figures
1.11 Structured organic solar cell. The incoming ray of light is diffracted
at the structured PEDOT:PSS-P3HT:PCBM interface, and consequently
trapped (after Lang 2012, p. 15) . . . . . . . . . . . . . . . . . . . . . . . 11
2.1 The six layers of the used ITO-PET substrates: ITO (white), under coat-
ing (orange), PET (blue), back coating (pink), adhesive (olive), and re-
lease liner (red). The total thickness is 120µm, the thickness from ITO
to adhesive 45µm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Carbon (grey), hydrogen (black), sulfur (yellow) and oxygen (red) atoms
form the chemical structures of PEDOT (left) and PSS (right) which
forms PEDOT:PSS if solved in water. Hydrogen atoms bound to oxygen
(pink) is highly polar (after Meier 2012, p. 66) . . . . . . . . . . . . . . . 14
2.3 The chemical structures of P3HT (left) and PCBM (right) solved in CB
form P3HT:PCBM, the so-called blend. Coloring is analogous to figure
2.4. (after Meier 2012, p. 62 & 68) . . . . . . . . . . . . . . . . . . . . . 15
2.4 Schematic representation of the different device architectures A-D, which
are prepared with the help of tables 2.1 and 2.2. The substrate (blue),
ITO (white), PEDOT:PSS (light green), highly conductive PEDOT:PSS
(dark green), P3HT:PCBM (red) and aluminium (grey) are not to scale.
The change in apparent color of the P3HT:PCBM is due to the underlying
layer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
(a) Device architecture A . . . . . . . . . . . . . . . . . . . . . . . . . 21
(b) Device architecture B . . . . . . . . . . . . . . . . . . . . . . . . . 21
(c) Device architecture C . . . . . . . . . . . . . . . . . . . . . . . . . 21
(d) Device architecture D . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1 Transmittance and reflectance measurements with a diffuse sample (red)
using an integrating sphere made out of Spectralon from labsphere . . . 25
(a) Transmittance measurement with a sample . . . . . . . . . . . . . . 25
(b) Reflectance measurement with a sample . . . . . . . . . . . . . . . 25
3.2 Four-point measurement of a sample (Akira Shimamoto & Ike 2012, after) 25
3.3 Visual representation of the air mass (AM) models AM 0, AM 1 and
AM 1.5 (Energy 10.07.2013) . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4 Visual representation of the solar simulator setup used in this work (after
Meier 2012, p. 49) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
56
65. List of Figures
3.5 Spectrum of a Solar Constant 1200 from K.H. Steuernagel Lichttech-
nik solar simulator (yellow) compared to the global radiation (black) in
arbitrary units (after Lang 2012) . . . . . . . . . . . . . . . . . . . . . . 29
4.1 Visual representation of the splitting intensities of an incoming beam
falling on a sample (light red). It can either be absorbed, reflected or
transmitted. If absorbed, it can also create photoluminescence and emit
light of a lower wavelength. (after Miguel A. P´erez & Arias 2013) . . . . 31
4.2 Absorption of different device structures on glass calculated via transmit-
tance and reflectance using UV/Vis spectroscopy . . . . . . . . . . . . . 32
4.3 Refractive index n (λ) (solid line) and extinction coefficient k (λ) (cir-
cles) of P3HT:PCBM measured by spectrometric ellipsometry (after Flo-
rent Monestiera 2007) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.4 Ratio of absorption of injected light into blend and intensity of light ar-
riving in blend of different device structures on glass calculated via trans-
mittance and reflectance using UV/Vis spectroscopy . . . . . . . . . . . . 35
4.5 Images of a finished organic solar cell . . . . . . . . . . . . . . . . . . . . 36
(a) Camera picture of the solar cell with silver paste, modified by in-
creased brightness and contrast. As mentioned in chapter 2.2.1, the
cell measures 2.2 x 2.2 cm . . . . . . . . . . . . . . . . . . . . . . . 36
(b) Microscopy image under 1.25x magnification of a pixel of an organic
solar cell. The black dots on the yellow aluminium are blisters that
could come from the blend annealing step. . . . . . . . . . . . . . . 36
4.6 Current-voltage graphs of four different device architectures detailed in
chapter 2.3 built on glass measured on a solar simulator. The samples A-1
(yellow), A-2 (orange), B-1 (red), B-2 (brown), C-1 (purple), C-2 (blue),
D-1 (green) and D-2 (olive) are plotted from -1 V to 1 V (lower picture)
and from -0.1 V to 0.7 V. . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.7 Characteristics of the four device architectures detailed in chapter 2.3
built on glass calculated from the data of the graphs of figure 4.6 with a
self made program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.8 Sheet resistances of H-PEDOT:PSS spin coated on PET, labelled samples
S-1 - S-10 and ITO on glass in arbitrary units . . . . . . . . . . . . . . . 42
4.9 Absorption of different device structures on PET measured calculated via
transmittance and reflectance using UV/Vis spectroscopy . . . . . . . . . 43
57
