Electrical performance study of organic photovoltaics for indoor applications

(1)

Electrical performance study of organic photovoltaics for indoor applications

with potential in Internet of Things devices

Studie av elektriska egenskaper hos organiska solceller för inomhusbruk med potential för enheter inom Internet of Things

August Andersson

Faculty of Health, Science and Technology Master thesis in Engineering Physics 30 hp (ECTS)

Supervisor: Ellen Moons Examiner: Lars Johansson June, 2020

(2)

Abstract

The evolution of the internet of things (IoT) opens the market opportunity for organic photovoltaic cells, especially for indoor applications where the lifetime of the organic cells is longer than outdoor.

For example, IoT requires oﬀ-grid energy sources for many devices with low power consumption. In this work, new materials were tested as candidate components in the active layer of printed organic photovoltaics by fabrication of devices. The initial electrical performance of these devices and their stability over time were investigated by measurements of the current-voltage characteristics. Three selected active layers were further investigated with atomic force microscopy (AFM) measurements.

The current-voltage measurements showed that the addition of a solvent additive to the active layer ink aﬀects the initial electrical performance as well as the stability of the devices. The AFM measurements showed that the surface topography of the active layer was aﬀected by the sort of solvent additive that was used. Three new electron acceptor material and two solvent additives showed promising electrical performance and stability.

Keywords: Indoor light harvesting, Internet of things (IoT), Organic Solar cell, Organic photo- volaic

Sammanfattning

Framväxten av internet of things (IoT) öppnar en möjlig marknad för organiska solceller, särskilt inomhus där livslängden för organiska celler är längre än utomhus. Till exempel behöver IoT fristående energikällor för ett stort antal enheter med låg eﬀektförbrukning. I det här arbetet testades nya material som kandidatkomponenter i det aktiva lagret hos tryckta organiska solceller för inomhusbruk genom tillverkning av enheter. Den initiala elektriska prestandan för dessa enheter och deras stabilitet över tid undersöktes genom mätningar av ström-spänning egenskaperna. Tre utvalda aktiva lager studerades ytterligare med atomkraftsmikroskopi (AFM) mätningar. Ström- spänning mätningarna visade att lösningsmedelstillsatser påverkar den initiala elektriska prestandan samt stabiliteten hos enheterna. AFM mätningarna visade att yttopograﬁn av det aktiva lagret påverkades av vilken sorts lösningsmedelstillsats som användes. Tre nya elektronacceptorer och två lösningsmedelstillsatser visade lovande elektrisk prestanda och stabilitet.

(3)

Acknowledgments

This thesis was conducted in the spring semester 2020 at Karlstad University and Epishine AB.

There are several people I would like to thank for their support and advice, which made this thesis work possible. I want to thank my supervisor Thomas Österberg at Epishine AB for your time spent discussing the work and guiding my process throughout the project. All other people at Epishine for including and giving me all the help I could ask for. I wish you all the best, and it will be exciting to follow your future work. From Karlstad University I want to thank my supervisor Ellen Moons for all your feedback and guidance during the work and Leif Ericsson for your help with the AFM measurement.

My dear parents, for all the help and support you have given to me, you are worth your weight in gold. Finally, I would like to thank my friends at Karlstad University which has ﬁlled my time as a student with uncountably many fun and memorable moments.

August Andersson Borensberg, June 2020

(4)

Abbreviations

AFM Atomic force microscopy

FF Fill factor

HOMO Highest occupied molecular orbital IoT Internet of things

J_SC Short circuit current

LUMO Lowest unoccupied molecular orbital OPV Organic photovoltaics

PCE Power conversion eﬃciency Pmax Maximum generated power V_OC Open circuit voltage rpm Revolutions per minute

(5)

1 Introduction

Solar cells are an essential part in shifting the world energy systems away from fossil fuels and towards renewable technologies. The total energy from the sun is several thousand times higher than the energy need of the world population [1]. Besides solar energy, other renewable energy sources such as wind, wave, hydroelectricity and biomass are indirectly powered by the sun [2].

Photovoltaics is one technology that converts sunlight directly into electricity. There are several diﬀerent solar cell technologies.

Organic solar cells are an alternative to conventional silicon solar cells, as they can be produced on ﬂexible substrates from solutions and at low temperatures by using low-cost roll-to-roll processes [1]. They also have low weight and potential to be produced without toxic materials [3], [4], [5].

However, compared to silicon solar cells, organic solar cells have lower power conversion eﬃciency (PCE ) and shorter lifetime. Traditional silicon photovoltaic panels have a PCE about 17 % and a 25 year long lifetime, while organic solar cells have a PCE around 5 % and lifetimes around 5 years [6]. Further research is therefore needed in order to increase the PCE and lifetime for organic solar cells.

Not only the light from the sun can be converted into electricity. Photovoltaic cells can also be used to convert indoor light. This opens the opportunity for organic photovoltaic (OPV) usage in the internet of things (IoT). For example, IoT requires oﬀ-grid energy sources for many devices with low power consumption for indoor applications [7]. Under indoor lighting the PCE of OPV devices can be higher than silicon solar cells [7]. The problem with stability is not so severe compared to outdoor conditions [8].

The most central part of an OPV device is the active layer, which absorbs the light and converts it into charge carriers. To eﬃciently generate charge carriers, the active layer usually consists of a blend of two materials, an electron-acceptor and an electron-donor. Exploring new candidates for electron-acceptor and electron-donor materials will be the task this thesis contributes with.

This thesis was performed at Karlstad University and the Swedish company Epishine AB in Linköping.

Epishine develops fully printed OPV cells that can be produced in large scale. These cells are de- signed for indoor use, i.e. under low light conditions, to support low power devices e.g. sensors used in IoT. In this project, one electron-donor material was combined with seven different electron- acceptors materials and four different solvent additives. These new active layer materials were investigated by fabrication of devices, which were tested with current-voltage measurements to see which components that work well together in terms of electrical performance and stability. This project evaluates these different components and the results are intended to be a basis for Epishine in their work to further develop and create new products.

(7)

2 Theory

2.1 Organic semiconductors

Organic materials consist of carbon based compounds in which at least one carbon atom is covalent linked to an atom of another type, usually hydrogen, nitrogen, oxygen, phosphor or sulfur [1].

Carbon is element number six in the periodic table and with the electron conﬁguration 1s²2s²2p², where the four valence electrons are located in the second shell 2s²2p². When C forms chemical bonds, all orbitals of the second shell combine and forms new, so-called hybridized orbitals [1].

Figure 1 illustrates how the sp²hybridized orbitals have lower total energy compared to the original system.

Figure 1: Energy diagram of the electron conﬁguration for carbon atomic and sp² hybridized orbitals.

During the sp² hybridization of a carbon atom the 2s orbital is mixed with only two 2p orbitals to form three equivalent sp² hybrid orbitals. This process leaves one electron in the unhybridized 2p_z orbital. These three sp²orbitals lie in the same plane and the angle between any two of them is 120^◦, while the unhybridized 2pz orbital is perpendicular to that plane. The process of sp² hybridization is illustrated in Figure 2.

Figure 2: The process of sp² hybridization and the three hybridized orbital together with the perpendicular unhybridized 2p orbital.

The concept of hybridization is illustrated in Figure 3 for the ethene molecule, C₂H₄. Bonds via electrons in s or sp-hybrid orbitals are usually referred to as σ-bonds, while electrons in p orbitals

(8)

form π-bonds. Typically a single bond is a σ-bond, while a double bond is composed of one σ- and one π-bond. If the carbon atom forms two single bonds and one double bond, it is sp²-hybridized [1].

Figure 3: The Lewis structure of ethene, C₂H₄, and the formation of σ- and π-bonds from the sp-hybrid and p orbitals [9].

Organic semiconductors can be a polymer made of carbon compounds [1]. A polymer is a large molecule, composed of many repeated subunits, known as monomers. Figure 4 illustrates how the polymer backbone in polyethyne is built up by the repeating monomers (C₂H₂)_n. If the polymer is semiconducting or non-conducting depends on the chemical bonds through the molecule backbone [10]. The electrons in a σ-bond, are localized and unable to move along the molecule backbone. A polymer with only single bonds along the backbone will be an electrical insulator. In conjugated polymers, with alternating single and double bonds along its carbon backbone, each carbon atom will form three sp₂-hybridized orbitals and one unhybridized p orbital. Overlapping sp₂ orbitals forms σ-bonds while overlapping p-orbitals form π-bonds. The electrons participating in the π- bonds are delocalized, in contrast to σ-bonds, and can contribute to the electric conductivity along the molecule backbone [10].

Figure 4: The polymer polyethyne with its repeating units (C₂H₂)_n. The alternating single and double bonds contribute to the electric conductivity along its carbon backbone.

The bandgap in organic semiconductors is called HOMO-LUMO gap and it is determined by the energy diﬀerence between the highest occupied molecular orbital (HOMO) and the lowest unoc-

(9)

cupied molecular orbital (LUMO), it is usually in the range of 1-4 eV, similar as the bandgap of inorganic semiconductors [10].

Exciton

An exciton is a bound state of an excited electron and a hole which are attracted to each other and bounded by Coulomb forces. It can be described as an electrically neutral quasiparticle that can transport energy without transporting electric charge. The Coulomb force can be expressed as

F = 1

4πεε₀ q1q2

r² (1)

where q1 and q2 are the electric charges, r is the distance between the charges, ε0 the vacuum permittivity and ε is the relative permittivity. There are mainly two diﬀerent types of excitons, which are depicted in Figure 5, Mott-Wannier excitons and Frenkel excitons [11]. The Mott-Wannier excitons are widely delocalized and have low binding energies. They are typically found in inorganic semiconductors, where the high relative permittivity leads to binding energies less than the thermal energy at room temperature of about 25 meV [10]. In contrast, Frenkel excitons are found in organic semiconductors and are highly localized, typically on one molecule [11]. They have high binding energies in the order of 0.5-1 eV [10] due to the low relative permittivity of organic semiconductors.

At room temperature, the thermal energy is not suﬃcient to separate the exciton into free charge carriers [10].

Figure 5: The Mott-Wannier exciton with low binding energies to the left, and the Frenkel exciton with high binding energy to the right. The ﬁgure is inspired from [11].

The low relative permittivity of organic semiconductors will affect the performance of organic solar cells. The current from a solar cell flows only when the exciton is separated into free charge carriers. The exciton in organic semiconductors can’t be dissociated by thermal excitation alone to work efficiently. This can be solved by using two materials with different energy levels and will be discussed in the next part.

2.2 Working principle of organic solar cells

The photoactive layer in modern organic solar cells usually consists of two materials, electron-donor and electron-acceptor, in order to successfully dissociate the tightly bound exciton that occurs in

(10)

organic semiconductors. The exciton is dissociated by the electric ﬁeld that arises at the donor- acceptor interface, which occurs because of the potential diﬀerence between the materials. The electron-donor is usually a polymer and the electron-acceptor a fullerene, small molecules or another polymer[12].

The photovoltaic mechanisms in an organic solar cell can be described by four fundamental steps [13]:

Exciton generation, Exciton diﬀusion, Exciton dissociation and Charge carrier trans- port, which are illustrated in Figure 6.

Figure 6: The fundamental steps to generating current in an organic solar cell.

1. Exciton generation: Upon illumination of the active material, an electron is excited to the LUMO by absorption of a photon with energy larger than the HOMO-LUMO gap. The light absorption will generate electron-hole pairs, so-called excitons, where the electron and the hole are attracted to each other and bounded by Coulomb forces.

2. Exciton diffusion: The generated exciton can diffuse through the material to the donor- acceptor interface. Due to the short lifetime of the exciton in organic materials, the exciton diffusion length is in order of 10 nm.

3. Exciton dissociation: If the exciton reaches the donor-acceptor interface within its lifetime, it dissociates into a free electron and hole due to the local electric ﬁelds caused by an energy level oﬀset between the two materials.

4. Charge carrier transport: Free charge carriers are transported to the electrodes through the donor and acceptor material by an internal electric ﬁeld caused by electrodes with diﬀerent work functions. The electrons are collected at the cathode and the holes at the anode.

2.3 Morphology

Since the exciton only can be disassociated at the donor-acceptor interface, and its diffusion length lD is only in the order of 10 nm, the morphology of the active layer is important for how the solar cell will perform [4]. During the formation of the morphology, the donor and acceptor are phase-separated into domains that are rich in either of the components [10]. Figure 7 shows a cross-section of four different morphologies. A very fine mixture of donor and acceptor materials as in Figure 7a) will lead to efficient charge generation but poor charge transport. By arranging

(11)

the donor and acceptor material in a bilayer stack, as in Figure 7b), ideal charge transport could be achieved. But the charge generation will be reduced since it only can take place at the donor- acceptor interface. According to [4], the ideal performance is reached by the morphology in Figure 7c). There highly ordered donor and acceptor domains will give a good charge transport. At the same time, domain widths of two times the exciton diﬀusion length contributes to eﬃcient charge generation. In reality, it is hard to reach a morphology like that one in Figure 7c) and a more realistic morphology is illustrated in Figure 7d). It should also be mentioned that in reality, the phase-separated domains are not completely pure. These domains are generally rich in donor or rich in acceptor.

Figure 7: Different types of morphologies. In a) there is a fine mixture of donor and acceptor which gives a high charge generation but poor charge transport. The bilayer- stack in b) gives a good charge transport but with a poor charge generation. c) is the ideal morphology with domain widths of two times the exciton diffusion length lD and d) a more realistic morphology. The figure is inspired from [4].

The morphology can be manipulated during fabrication by thermal annealing of the active layer [14], or by adding solvent additives to the active layer ink, to control the ﬁlm drying to yield a favorable morphology [15]. Additives can aﬀect the nanoscale phase separation and crystallinity of the active layer because they usually have a higher boiling point than the main solvent and can show selective solubility of one of the two blend components [14].

Device structure

The device geometry of typical organic solar cells is illustrated in Figure 8. When designing the solar cells, it is preferable to choose materials for substrates, electrodes and electron transporting layer/hole transporting layer (ETL/HTL) with as large bandgap as possible, so all absorption happens in the active layer. Depending on if the solar cell is illuminated through the anode or the cathode the performance of the device can be diﬀerent [16]. That may depend on parasitic absorption from one of the materials before the active layer. Parasitic absorption refers to an

(12)

optical absorption process that does not generate an electron/hole pair [17].

Figure 8: Device structure for typical organic solar cells.

2.4 Solar cell characteristics

The electrical performance of a solar cell can be characterized by the short circuit current density (J_SC), the open circuit voltage (V_OC), the ﬁll factor (FF ) and the power conversion eﬃciency (PCE ). These parameters can be found by current-voltage measurement of the solar cell. The current density and power density as a function of voltage for a solar cell is shown in Figure 9, together with the characterization parameters.

(13)

Figure 9: Current density and power density as a function of voltage for a solar cell under illumination, together with the characteristic points short circuit current density (J_SC), maximum power point (mpp), max generated power (P_max) and open circuit voltage (V_OC).

Short circuit current density

JSC is the current density that flows through the solar cell when no voltage is applied. JSC depends on several factors, such as the efficiency of the exciton dissociation, charge transport, and charge extraction, but also the intensity and spectrum of the incoming light, and the absorption coefficients of the materials in the active layer [10].

Open circuit voltage

In organic solar cells, V_OC is determined by the diﬀerence of the HOMO level of the donor and LUMO level of the acceptor [18].

Fill factor

FF characterizes how ”square” the JV curve is and it represents how efficient the photo-generated carriers can be extracted out of a photovoltaic device [19]. The fill factor is defined as

FF = VmppJmpp

V_OCJ_SC (2)

Generated power

The maximum generated power is deﬁned as

(14)

Pmax= VmppJmpp= FF VOCJSC (3) Power conversion eﬃciency

PCE of a solar cell is deﬁned as the ratio between the maximal generated power and the power of the incident light P_in

PCE = P_max Pin

= FFV_OCJ_SC Pin

(4)

Equivalent circuit of a solar cell

The current-voltage characteristics of a solar cell can be modeled by an equivalent circuit consisting of a current generator in parallel with a diode. An ideal solar cell can be depicted as the equivalent circuit shown in Figure 10a). J_phstands for the photo-generated current density, which only depends on the illumination light intensity, here as a constant current source. Jdstands for the diode current density, which is a function of the external applied voltage V . The total current density J is a combination of J_ph and J_d, given by

J (V ) = J_d− Jph= J0

[ exp

( eV k_BT

)

− 1 ]

− Jph (5)

where J₀ is the diode saturation current density, e is the elementary charge, T the temperature and k_B is the Boltzmann constant.

Figure 10: Equivalent circuit for a) an ideal solar cell and b) a real solar cell.

(15)

In the real case, the J -V characteristic curve of the solar cell deviates from the ideal one because parasitic resistances are unavoidable factors. Therefore a series resistance Rs and a shunt resistance R_p parallel to the diode, are introduced into the ideal solar cell model from Figure 10a) to account for these energy losses. The equivalent circuit of a real solar cell is shown in Figure 10b). The contributions to Rs are internal resistances in the active layer and the contact resistances between the active layer and electrodes. R_p comes from current leakage induced by pinholes in the cell or current leakage from the edges of the device. Ideally is R_s zero and R_p inﬁnite to avoid power losses. The inﬂuence of Rs and Rp on the J -V curve is depicted in Figure 11.

Figure 11: The inﬂuence of the series resistance Rs and the shunt resistance Rp on the J -V curve. The ﬁgure is taken from [20]

Taking the parasitic resistances into consideration, the equivalent circuit equation can be expressed by

J (V ) = J₀ [

exp

(e(V − JRs) nkBT

)

− 1 ]

+V − JRs

Rp − Jph (6)

Where n is the diodes ideality factor, which depends on recombination mechanisms.

2.5 Degradation

Organic solar cells usually have a shorter lifetime in ambient compared to inorganic solar cells due to degradation [21]. Degradation can be described as a decrease in the electrical performance of the device over time and can be divided into two categories, extrinsic and intrinsic degradation [22].

Extrinsic degradation is related to external factors such as oxygen and water from the atmosphere.

Oxygen and water that enter into the device cause chemical degradation of the organic layers and interfaces, this is more severe under illumination than in the dark. This can change the absorption, energy levels and charge carrier mobilities of the active layer [22]. Small amounts of oxygen and

(16)

water can be introduced into the device during manufacturing, but a bigger problem is that they may diﬀuse into a ﬁnished device [23]. Proper encapsulation is used to reduce the extrinsic degradation that causes rapid device failure [21]. Intrinsic degradation arises even with perfect encapsulation and is related to temperature and light [21]. For instance, heat can rearrange small molecules at material interfaces, which reduces charge extraction. Over time, the two materials of the active layer can phase separate over a large distance and decrease the ability to create free carriers from absorbed photons [21]. Light can cause photochemical reactions in the absorbing layer, mostly photo-oxidation [21]. Photons with high energies, typically in the UV regime, may break the σ- bonds leading to degradation of the polymer [18].

2.6 Organic photovoltaics for indoor light harvesting

The standard conditions for outdoor solar cell measurements are: solar spectrum at sea level (AM 1.5G), light intensity of 100 mW/cm²(1 sun), and cell temperature at 25^◦C [4]. During these conditions, organic solar cells usually have a lower power conversion efficiency compared to conventional silicon solar cells [8]. But for indoor light harvesting, when the illumination comes from artificial light sources such as fluorescent lamps and light emitting diodes (LEDs), OPVs can have a higher PCE [7]. That’s because the indoor light intensities are two to three orders of magnitude lower than 1 sun intensity [24], and the emission spectrum of today’s modern artificial light sources are very different compared to daylight or earlier techniques, such as incandescent or halogen lamps. The emission spectrum of these light sources is illustrated in Figure 12. Fluorescent lamps or LEDs only emit light in the visible range, while the emission spectrum for daylight, incandescent, and halogen lamps continues into the infrared regime. According to [25], an optimal band gap for harvesting the energy from fluorescent lamps or visible LEDs is 1.9 eV, while the optimum bandgap for 1 sun illumination (AM 1.5 G, 100 mW/cm²) is smaller, about 1.4 eV.

Figure 12: Light emission spectrum for daylight and common artiﬁcial light sources.

The y-axis has normalized light intensity. The ﬁgure is taken from [26]

The degradations mechanisms described earlier pointed out that the OPVs are sensitive to oxygen

(17)

and water, and especially in combination with strong light. But it’s not only the light intensities and emission spectrum that diﬀers between outdoor and indoor conditions. The outdoor conditions usually have bigger variations in temperature, humidity and illumination, while the indoor conditions are more stable. Therefore, the stability of OPVs for indoor applications is longer compared to outdoor organic solar cells [8].

Light can be measured in radiometric units, which characterize light in units of power, or photometric units which characterizing the action of light upon a human eye [27]. In photometry, the radiometric power is scaled by the spectral response of the human eye. Table 1 shows the radiometric and photometric units for two light quantities.

Table 1: The radiometric and photometric terminology and units for two light quantities.

Lux is the unit of illuminance (luminous flux per unit area incident on a surface). The illuminance produced by a luminous flux of 1 lumen uniformly distributed over a surface of area 1 square metre is 1 lux [27]. The indoor illuminance is usually measured in lux. For office and rooms, the indoor illuminance usually ranges between 200-1000 lux [24]. Light intensity corresponds to irradiance in radiometric units and illuminance in photometric units [27].

(18)

3 Experimental

This work can be divided into three experiments with the purpose to investigate the inﬂuence of several parameters on a selection of new active layers by fabrication of photovoltaic devices, and ﬁnd trends in their electrical performance.

One electron-donor was tested with seven diﬀerent electron-acceptors, with or without additives.

Devices were fabricated and studied with J -V measurements. Three selected compositions were further investigated with AFM.

Due to conﬁdentiality the donor, acceptors or additives will not be named by their real names.

Instead, they are referred to with placeholders, i.e. the donor is referred to as D, the seven acceptors as A1, A2, ..., A7 and the four additives as additive1, additive2, ..., additive4. Devices fabricated with Epishines standard active layer are denoted as ”Reference”.

In the ﬁrst experiment the reproducibility of making devices for this work was improved (by training and improved skills), and the stability of the three active layers D:A1, D:A2 and Reference was investigated. Due to conﬁdentiality the results are excluded from this report.

3.1 Ink preparation

The donor and acceptor powders were weighted in ambient, while the solvent was added to the bottle inside a glovebox. The solution was stirred on a hotplate at 100^◦C in ambient overnight. In the experiments where solvent additives were used, the stock solution was divided into small bottles inside the glovebox, and the additives were added. The solution was stirred on a hotplate at 100^◦C in ambient for some hours before spin-coating the ink.

Second experiment

• Donor: D

• Acceptor: A3

• Additives: additive1, additive2, and additive3

The donor and acceptor were mixed with 1:1.5 D:A ratio, at 40 mg/ml in the solvent. The additives were added to the active layer ink with the amounts 0.5 %, 2 %, 5 % or 10 % of the ink volume.

Inks without additives were also fabricated.

Third experiment

• Donor: D

• Acceptor: A3, A4, A5, A6, and A7

• Additives: additive1, additive2, additive3, and additive4

The donor and acceptor were mixed with 1:1.5 D:A ratio, at 40 mg/ml in the solvent. The additives were added to the active layer ink with an amount of 5 % of the ink volume. Inks without additives were also fabricated.

(19)

3.2 Device fabrication

The device was made in two pieces, where one piece is referred to as cathode and the other as anode. These were fabricated separately and then laminated together with the active layer as glue.

A conceptual illustration of the lamination is depicted in Figure 13.

Figure 13: Conceptual illustration of the lamination process. The ﬁgure is taken from [29].

Both the anode and cathode was made from polyethylene terephthalate (PET) substrates with roll-

to-roll slot died electrodes made of PEDOT:PSS (Poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate)).

These were cut out in rectangles with sizes of 3 times 5 cm, and three cells were cut into the anode, with sizes between 0.15-0.4 cm², by cutting through the PEDOT:PSS layer with a scalpel. All solutions were spin-coated inside a fume hood in ambient atmosphere. Due to conﬁdentiality, the structure of the device will not be speciﬁed any closer. The anode and cathode were laminated at 110 ^◦C, and then hold together by two glass sheets and clamps to get protection from mechanical stress which could lead to delamination. To increase the conductivity at the contacts during the J -V measurements, a small amount of silver paint was applied on the layers at the cathode and the anode. An image of a real device is shown in Figure 14.

Figure 14: Real device.

This thesis standard procedure was spin-coating the active layer with warm ink at 80 ^◦C for 30 seconds at coating speed 2000 rpm, followed by 2 minutes annealing at 80 ^◦C.

All devices containing additive were fabricated according to the standard procedure, while the devices without additives were fabricated both according to the standard procedure and without annealing the active layer. J -V and stability measurements were performed on the fabricated devices.

(20)

3.3 Equipment and measurements

In this section, the equipment and the measurements in this work are described.

Spin-coating

Spin-coating is a process technique that is used to create thin films in small scale. An ink-solution is deposited on a substrate, which then spins at a given rotation speed and time until a dry film has formed. The process is illustrated in Figure 15, where the excess amount of solution first is dis- carded due to rotational acceleration of the substrate, then flows off while the viscosity increases by evaporation, until a critical viscosity has been reached. Finally, a dry film is formed by evaporation [10].

.

Figure 15: The spin-coating process illustrated in four steps. The ﬁgure is inspired from [10]

Glovebox

A glovebox is a sealed container ﬁlled with inert gas to keep the amount of O2 and H2O vapor low in order to, for example, prevent degradation of devices or chemical reactions with other sensitive materials. Devices and materials can be transferred in and out of the glovebox through a load lock, and sealed gloves attached to the side of the container can be used to handle equipment inside the box. In these experiments, a glovebox with N2 as inert gas was used, and the O2 and H2O levels were < 10 ppm.

J -V measurement

The current-voltage characteristics (J -V curve) of a solar cell reﬂect the overall performance and the electrical processes in the device. The J -V measurements are done under deﬁned illumination conditions and by varying the voltage over the device and measure the current in every point.

This allows directly measuring the most important parameters of the device, like J_SC, V_OC, FF and PCE . J -V measurements in dark conditions are usually performed as well since it gives information about the diode quality of the solar cell. This can give insights into the presence of recombination

(21)

losses [1]. Figure 16 shows the J -V curve of such measurements for both dark and illuminated conditions.

Figure 16: Current density as a function of voltage for an ideal solar cell in dark and under illumination.

In the J -V measurements, a warm white LED lamp (3000 K) was placed inside a box at room temperature in ambient conditions, and the light intensity was set to 500 lux. The device was placed on a white plate, 25 cm under the lamp. A source measure unit (KEITHLEY 2401 SourceMeter ) was used to measure the current through the device as a function of voltage. The voltage was varied from -1 to 1 V (in steps of 0.05 V), and the source measure unit was used together with an in-house software that applied constant voltage steps and measured the current. One measurement was done per device. Figure 17 shows the setup for the J -V measurements. The current density of each device was calculated by dividing the current with the illuminated cell area. The cell area was measured for each cell individually with a caliper.

An additional J -V measurement was done for each device with the lamp turned oﬀ just for control of the diode behavior. This was done before the illuminated J -V measurement.

(22)

Figure 17: Setup J -V measurement.

Stability measurement

After the initial J -V measurements of the fabricated devices they were stored in the glovebox, in an oven at 55^◦C. The warm oven is intended to accelarate the intrinsic thermal degradation, while the inert atmosphere is intended to prevent the devices from extrinsic degradation. Further J -V measurements were done in room temperature and ambient conditions with some days intervals in order to check the stability of the devices.

Atomic force microscope

Atomic force microscope (AFM) is one type of scanning probe microscope that is used to characterize surface topography and properties. The instrumental setup of a typical AFM system is illustrated in Figure 18, and it works by scanning the sample surface with a very sharp tip, which is attached to a cantilever. The probe and the sample move relative to each other in x, y and z-direction

(23)

by piezoelectric crystals. When the probe approaching the surface it is affected by forces, which will deflect the cantilever. By reflecting a laser beam from the top surface of the cantilever to a photodetector, these deflections can be measured. The signals go through a feedback control system and at each point along the scan, the cantilever’s vertical position is recorded in order to generate a pixel-by-pixel image of the surface topography. An AFM generates three-dimensional images of the sample surface with a resolution within the nanometre range [28].

Figure 18: Setup for a typical AFM.

The AFM can be operated in diﬀerent modes, where the three primary modes are: contact mode, non-contact mode and tapping mode. In contact mode, the tip is brought into contact with the sample surface and the applied force on the sample is kept constant by the feedback loop that adjusts the height of the sample relative to the probe. This mode suits hard samples, while soft materials can be damaged due to the shear forces between the tip and surface.

In tapping mode, the cantilever oscillates at or near its resonance frequency by a piezoelectric actuator. The tip oscillates over the surface and will come into contact with the surface once per oscillation and ”taps” it. As the oscillating probe starts to intermittently contact the surface, the cantilever oscillation amplitude is reduced due to energy loss caused by the tip contacting the surface. Thus, when operating in tapping mode the feedback loop works to maintain constant amplitude. Since the applied force is just vertical and there are no shear forces, this mode suits well for investigating soft materials as polymers [28].

Tapping mode can also produce phase images that provide additional information about the surface properties than topography. By comparing the phase diﬀerence between the cantilever oscillation and the signal sent to the cantilever’s piezoelectric actuator, variations in surface properties such as composition, adhesive and viscoelasticity can be determined [28].

The AFM measurements at Karlstad University were performed in tapping mode with an atomic force microscope (Veeco Innova), along with a silicon cantilever (Olympus, OMCL-AC160TS-W2).

It has a resonance frequency of 300 kHz and a spring constant of 42 N/m.

(24)

3.4 Experiments

Three devices were fabricated and measured for each active layer in order to see the distribution of the electrical performance.

In every experiment three devices were fabricated with Epishines standard active layer according to the standard procedure, and they are referred to as Reference. It was important to include these devices in the experiments since the characteristics of devices with Epishines standard active layer are well known. They could be used as a control to exclude variations by other eﬀects and to ensure the quality of the other components in the device, such as electrodes. The Reference devices were also used as a reference to compare the results of the new materials with. Table 2 gives an overview of what was varied in the experiments.

Table 2: An overview of the experiment variables. Yes or No refers to if the variable was varied or not.

The two diﬀerent experiments can be explained as follows:

• The purpose of the second experiment was to get an understanding of how the electrical performance of the device was affected by additives, and to find a baseline of the additive amount that could be used in the third experiment. The baseline was extracted from the results of the electrical performance and was decided to be 5 %, since apparent differences in the electrical performance begin to be observed. The experiment investigated how the electrical performance of the device was affected by additives in various amounts for one donor-acceptor blend.

• The purpose of the third experiment was to ﬁnd ink compositions with high electrical performance from a set of acceptors and solvent additives.

The donor-acceptor blend D:A3 was investigated together with the three solvent additives: addi- tive1, additive2 and additive3, in various amounts. Devices without additives were also fabricated.

All devices containing additive were fabricated according to the standard procedure, while the devices

(25)

without additives were fabricated both according to the standard procedure and without annealing the active layer. J -V and stability measurements were performed on the fabricated devices.

An overview of the diﬀerent active layers used in this experiment is included in the appendix, page 34 Table A1.

AFM measurement

The three active layers from the second experiment that showed most variation in the electrical performance were further investigated with AFM measurements. The AFM measurements were performed in tapping mode for the three active layers:

– D:A3 no additive, annealed – D:A3 + 10 % additive1 – D:A3 + 10 % additive2

The measurements were performed on samples that were fabricated in the same way as the devices, except the lamination process.

The donor D was blended with each of the ﬁve acceptors: A3, A4, A5, A6, and A7, and the solvent was mixed with each of the four additives: additive1, additive2, additive3 and additive4. Devices without additives were also fabricated. All devices containing additive were fabricated according to the standard procedure, while the devices without additives were fabricated both according to the standard procedure and without annealing the active layer. J -V and stability measurements were performed on the fabricated devices.

An overview of the diﬀerent active layers used in this experiment is included in the appendix, page 35 Table A2.

(26)

4 Results

This chapter is divided into three sections, where the ﬁrst show the results from the second experiment. In the second section, the results of a selection of the ”best” active layers from the second and third experiment are present. In this work, Pmax was used to compare the performance of the devices and the best active layers are considered to be the ones that give devices with highest P_max. The last section shows the eﬀect on the device performance when storage in the ambient atmosphere.

In the experiments, the investigated active layers are compared to Epishines standard active layer, the Reference. The illumination conditions for the J -V measurements were a warm white LED (3000 K) spectrum at 500 lux illuminance.

4.1 Eﬀect of additives

In the second experiment, one donor-acceptor blend, D:A3, was tested together with three diﬀerent solvent additives: additive1, additive2 and additive3, in diﬀerent amounts. Active layers containing solvent additive were annealed, while the active layers without solvent additives were fabricated both with and without annealing.

Initial J -V measurements of D:A3

Figure 19 - 22 illustrate the average electrical performance of the three devices in the respective active layer.

Figure 19: Average and standard deviation of the maximum generated power for the respective active layer.

(27)

Figure 19 shows that P_max from the Reference devices is higher than the devices made of D:A3 active layer. Without solvent additives, Pmaxdecreases in average 8 % without annealing compared to annealed devices.

Compared to ”No additive, annealed”, Pmax for additive1 decreases around 15 % and 22 % at amounts of 5 % and 10 %, respectively. For additive2, Pmax increases around 13 % and 20 % at amounts of 5 % and 10 %, respectively. No major eﬀects on P_max are seen for additive3.

Figure 20: Average and standard deviation of the ﬁll factor for the respective active layer.

(28)

Figure 21: Average and standard deviation of the open circuit voltage for the respective active layer.

Figure 22: Average and standard deviation of the short circuit current for the respective active layer.

The results observed in Figure 20 - 22 can be related to the eﬀects on Pmax observed in Figure

(29)

19.

additive1: Pmax decreases due to a strong decrease in JSC. VOC increases at the amounts 5 % and 10 %. FF increases slightly at the amount of 10 %.

additive2: Pmaxincreases due to strong increase in FF , while VOCand JSCremained approximately independent of the amount of additive.

additive3: P_maxremains approximately independent of the amount of additive. The small decrease in VOC at the amounts 0.5 %, 2 % and 5 % is compensated by a small increase in FF .

In Figure 21, VOC for devices with D:A3 active layers is higher than the Reference devices.

The average of the standard deviation for the parameters P_max, FF , V_OC and J_SC in Figure 19 - 22 is around 7 %, 0.7 %, 0.4 %, and 7 %, respectively.

In Figure 23, the J -V curve of the active layer without any additive is compared to the J -V curves of the active layers with the lowest and highest Pmax in the experiment, i.e. 10 % additive1 and 10 % additive2, respectively. The corresponding values of J_SC, V_OC and FF are presented in Table 3.

Figure 23: J-V curve of solar cell with active layer D:A3, illustrating how electrical performance is aﬀected by additive1 and additive2.

(30)

Table 3: The characteristic values for devices with active layer D:A3 with 10 % additive1, 10 % additive2 and without additive.

The results in Figure 23 and Table 3 shows that Pmaxincreases with 31 % for additive2, while Pmax

decreases with 33 % compared to no additive.

additive1: the decrease in P_max is mainly due to a decrease in J_SC, while FF and V_OC are slightly increased.

additive2: the increase in P_max is mainly due to an increase in J_SC and FF , while V_OC is slightly decreased.

Surface topography of the active layers

These three active layers were further investigated with AFM in order to see if the electrical performance can be related to the surface topography. The result from the AFM measurement is shown in Figure 24.

Figure 24: Topographic measurement with atomic force microscope of three active layers. Note the diﬀerent scan areas and height scales between the three images.

The AFM images show that both the height diﬀerence and the domain size are larger for additive1 and additive2 compared to no additive. The surface roughness (RMS) for D:A3 active layer is slightly increased with 10 % additive1, and drastically increased (factor 20) with 10 % additive2.

Stability measurements of D:A3

The devices were stored inside a glovebox at 55 ^◦C after the initial J -V measurement in order to accelerate the intrinsic degradation. Further J -V measurements were done with some days intervals

(31)

in order to check the stability of the devices. Figure 25 shows how the maximum generated power changed over time, for the active layers with diﬀerent amount of additive3.

Figure 25: Stability measurement for D:A3 with additive3 over 15 days, where the lines are the average of the three devices in the respective active layer.

P_max decreases less the ﬁrst 5 days with an increasing amount of additive3. With 10 % additive3, Pmaxis increasing the ﬁrst 5 days. Without additive, Pmaxof the devices with annealed active layer decreases faster than the devices without annealing.

The corresponding result to Figure 25, but without taking the average of the performance for each active layer, is included in the appendix, page 37 Figure A3. The results observed in Figure A3, shows that the stability behavior of the devices is similar for each active layer. The lines for each active layer have similar shapes.

The results from the stability measurements of the maximum generated power for additive1 and additive2 are included in the appendix, page 36 Figure A1 and A2 respectively.

4.2 Selection of high performance active layers

In this section, the electrical performance of a selection of best performing active layers is presented.

The best active layers are considered to be the ones that give devices with highest P_max. Initial J -V measurements

In Figure 26, the J -V curves for some of the best devices (highest Pmax) are shown. The corre- sponding values of JSC, VOC and FF are presented in Table 4.

(32)

Figure 26: J -V curves for some of the best devices with corresponding active layers.

Table 4: The characteristic values for some of the best devices with corresponding active layers.

The results in Figure 26 and Table 4 shows that the Reference device has higher electrical perfor- mance than the devices with the new active layers. P_max of the Reference device is about 10 % higher than the second best device (D:A3 + 10 % additive2), and about 28 % higher than the fourth best device (D:A7, no additive, annealed).

D:A7 with 5 % additive4 show highest JSC but lowest VOC. D:A3 with 10 % additive2 has the highest VOC but, together with D:A7 without additive, the lowest JSC.

Stability measurements

The result from the stability measurements of the eight best devices are presented in Figure 27.

These consist of the acceptors A3, A6, A7, and the solvent additives additive1, additive2 and additive3.

(33)

Figure 27: Stability measurement of the maximum generated power for the eight best active layers over 15-20 days. The lines are the average of the three devices in the respective active layer.

Compared to reference, all have good power stability, over at least 15 days, except D:A3 + 10 % additive2.

Pmax increases the ﬁrst 5 days and then is stabilized for additive4, while Pmax decreases the ﬁrst 5 days and the slightly increases (except D:A3 + 10 % additive2). Additive3 is slightly decreasing over the 20 days.

More results of the stability measurements are available in the appendix, page 37-39.

4.3 Degradation in ambient

Figure 28 shows how the performance is aﬀected by storing a device (D:A3 + 2 % additive 1) in ambient for 10 days. The corresponding values of J_SC, V_OC and FF are presented in Table 5.

(34)

Figure 28: Illustrates how the performance decreases for a device when it has been stored in ambient for 10 days.

Table 5: Change in the characteristic values for a device with D:A3 + 2 % additive1 when stored in ambient for 10 days.

The results in Figure 28 and Table 5, shows that after stored in ambient for 10 days the electrical performance of the device is apparent decreased. P_max is decreased with more than 50 %, where the major loss comes from a decrease in JSC and FF . The shape of the J -V curve shows a clear eﬀect of increased series resistance Rs.

(35)

5 Discussion

The goal of this work is to ﬁnd material compositions and electrical performance patterns that can be interesting from the point of view of a large scale manufacturing process. Explanations to the underlying physics of the results are hard to state without further experiments, and is not the focus of this thesis. Possible explanations for the results are discussed.

5.1 Eﬀect of additives

Initial J -V measurements of D:A3

In this experiment, the donor-acceptor blend D:A3 was used together with three different additives in various amounts. The results show that additive variation affects the initial values of the parameters P_max, FF , V_OCand J_SCdifferently. With additive1 for example, the V_OCincreases with an increasing amount of additive, while the J_SC is decreasing for 10 %. The FF for additive2 increases with an increasing amount of additives, while no major effects are seen for additive3.

The J_SC is related to the whole photoelectric conversion process, namely exciton generation, diffusion and dissociation, charge transport and collection. Domain size, phase purity, and the inter- penetrating network within the active layer are crucial for eﬃcient charge generation and carrier transport [30]. V_OC is mainly determined by the energy levels of the donor and acceptor materials, but it is also related to the morphology of the active layer, e.g. D/A interface area, microstructure and crystallinity [14]. FF represents how eﬃcient the photogenerated carriers can be extracted out of a photovoltaic device and it is related to charge transport and recombination in the active layer.

Fast charge transport and low charge recombination depend on the morphology, e.g. crystallinity, molecular orientation, domain purity and vertical phase separation [14].

For instance, an increasing amount of additive1 seems to promote V_OC but inhibit J_SC. Higher hole and electron mobilities due to an increase in phase purity can be one reason for the increase in FF for additive2. No major eﬀects are seen for additive3, it may be because additive3 has similar properties as the solvent, e.g similar boiling point.

Stability measurements of D:A3

The result of the stability measurements of D:A3 with different amounts of additive3 show that Pmaxdecreases less the first 5 days with an increasing amount of additive3. At an amount of 10 %, P_max was increased the first 5 days instead. The result also shows that the shape of the stability curve of P_max is very similar for the devices with the same active layer.

For some devices in the stability measurements, Pmax increases the ﬁrst days before it stabilizes.

One explanation may be that O₂ and H₂O molecules that enter the device during the fabrication, can leave the device when it is stored in the glovebox. As mentioned in the theory chapter, the active layers is sensitive to O2 and H2O molecules, and a decrease of these molecules may increase P_max. During the stability measurments the devices are exposed to ambient conditions for a shorter time compared to the fabrication process.

(36)

Surface topography of the active layers

The change in performance is assumed to relate to the change in morphology due to the additives.

The AFM height images of the surface of the active layers confirm that the roughness of the active layer is affected by the additives, i.e. 10 % of additive1 enhances the roughness of the D:A3 active layer slightly, while 10 % additive2 makes the surface of the D:A3 film drastically rougher by a factor 20.

It is somewhat unexpected that the device with highest P_max has the roughest surface topography, and it is not clear whether this rough surface topography could be correlated to the increased ﬁll factor. The AFM measurements were performed on active layers which have not been laminated.

During the lamination process, the device is exposed to heat and pressure, which may aﬀect the roughness.

Ideally, the morphology should provide domain widths of two times the exciton diffusion length (5-20 nm [10]) for efficient charge separation. Therefore it might seem strange that 10 % additive2 shows the best electrical performance since the AFM pictures show much larger structures compared to no additive and 10 % additive1. But Figure 24 shows the topographic images for the three active layers, which only indicates the height difference. The height differences may be originating from phase separated domains and can be used to estimate domain sizes, but they do not say anything about composition.

General reﬂection

For the D:A3 blend, the best solvent additive was additive2 where the amounts 5 % and 10 % showed the highest initial P_max and a small change in P_max over time. The initial P_max of additive1 and additive3 was lower than additive2. Over time, Pmax decreased fast for additive1, while additive3 decreased moderate, except D:A3 + 10 % additive3 which increased the ﬁrst 5 days.

The results from the second experiment show that both the type and the amount of additive aﬀects the initial performance as well the stability. The combination of amount and type of additive that gives the highest initial performance is necessarily not the combination that provides the best stability. This can be important to take into account in exploring a combination that meets the device requirements of electrical performance.

5.2 Selection of high performance active layers Initial J -V measurements

The J -V curves of D:A3 + 10 % additive2, D:A7 without additive, D:A7 + 5 % additive4, and the reference shows good J -V shape of the measured devices, even though the new compositions have lower electrical performance than the reference.

Stability measurement

The stability of the eight best active layers shows that regardless of the acceptor, the same additives give a similar shape of the P_max stability curve. P_maxfor additive2 first decreases and then flattens out, while Pmax for additive4 first increases and then flattens out. D:A6 + 5 % additive3 shows fairly flat but slightly decreasing behavior.

(37)

General reﬂection

The reference devices with Epishines standard active layer show the highest P_max in both experiments. However, the devices with active layers of A3, A6 and A7 together with additive 2 and 4 show potential to reach the same electrical performance. They could probably be further optimized by testing other amounts of additives, donor-acceptor concentrations, spin-coating speeds and annealing conditions.

For commercial products, the active layer that gives the highest electrical performance of a device is not necessarily the best choice. The material components of that active layer may be more expensive, harder to produce on a large scale and so on. Other active layers can therefore be more suited for commercial products even if the electrical performance is lower. Of course, this also depends on the application and requirements of the product.

5.3 Degradation in ambient

The degradation of D:A3 + 2 % additive 1 when stored in ambient conditions for 10 days shows how fast the electrical performance of OPV devices decreases when they are exposed to ambient atmosphere. Pmax decreased with about 50 % compared to inert conditions. The devices in the experiments were not encapsulated, just laminated. Proper encapsulation is therefore required for commercial products.

5.4 Error sources

The area of each cell was measured by hand with a caliper. This introduces uncertainty into the current density (which is the current divided by the area) and further to the power density (which is the current density times the voltage). A clear example can be seen in Figures 19-22, where the standard deviation usually is larger for J_SC and P_max compared to V_OC and FF . An exception is 0.5 % additive3 where the standard deviation for J_SC is small, while the standard deviation of V_OC and FF are similar to the other additives, and thereby the standard deviation of Pmax is also small since it depends on the other parameters according to Equation 3.

There are of course several other causes that can affect the variations in results. For example local defects as scratches in the PEDOT:PSS layer, dust, inks that are some days old, scratching interlayers with the pipette when applying the inks before spin-coating. But it should be mentioned that in this work the absolute values are not that important. It is more interesting to find trends and find active layers that can be further optimized and tested in the roll-to-roll process.

Only one sample per additive was investigated with AFM, and with few measurements. This means that there is large uncertainty in the results. More samples need to be investigated, and each sample must be measured in several places in order to exclude local eﬀects and be able to draw general conclusions.

(38)

6 Conclusion

The results from these experiments show lots of variation. Some additives work well with diﬀerent types of donor-acceptor blends, and small variations in the amount of additives can have a big impact on the performance. Therefore, it can be diﬃcult to predict the results when working with organic photovoltaics and extensive testing is necessary.

The results from the second experiment show that both the type and the amount of additive affects the initial performance as well the stability. For the D:A3 blend, additive2 with amounts of 5 % and 10 % showed the highest inital P_max as well as overtime. The AFM measurement indicated that the device with highest P_max had the roughest surface topography. This is somewhat unexpected, but the investigated samples had not been laminated. During the lamination process the device is exposed to heat and pressure, which may affect the roughness. Therefore, it is difficult to relate the electrical performance of the devices to the surface topography. Further measurements are required before general conclusions can be made.

In these experiments, the Reference devices with Epishines standard active layer showed the highest P_max. Of the new active layers, the donor D with the acceptors A3, A6 and A7 showed the highest electrical performance, especially together with additive2 and additive4. They showed potential to reach the same electrical performance as the Reference devices, and could probably be further optimized by testing other amounts of additives, donor-acceptor concentrations, spin-coating speeds and annealing conditions.

P_max of the device that was stored in ambient conditions for 10 days decreased with about 50 % compared to inert conditions. The shape of the J -V curve shows that the major loss comes from a decrease in JSC and FF , and a clear eﬀect of increased series resistance Rs. Proper encapsulation is therefore required for commercial products.

7 Future work

It would be interesting to investigate various amounts of additive2 and additive4 for the donor D with the acceptors A3, A6 and A7. This is similar to the second experiment but also including additive4 and the acceptors A6 and A7.

The active layers can further be optimized with other donor-acceptor ratios and concentrations in the solvent, spin-coating speeds and annealing conditions. The most promising active layers can further be investigated and optimized by slot-die coating, to see the potential for a large scale manufacturing process.

(39)

8 Appendix

8.1 Active layers in each experiment Second experiment

Table A1: List of the diﬀerent active layers investigated in the second experiment.

(40)

Table A2: List of the diﬀerent active layers investigated in the third experiment.

8.2 Additional results

Here additional results from the second and third experiment are presented. The ﬁgures illustrate the stability of the maximum generated power of the devices. Each line is the average of the three devices in the respective active layer (except in Figure A3, where all devices are presented).

(41)

Figure A1: Stability of D:A3 with additive1, where the lines are the average of the three devices in the respective active layer.

Figure A2: Stability of D:A3 with additive2, where the lines are the average of the three devices in the respective active layer.

(42)

Figure A3: Stability measurement for D:A3 with additive3 over 15 days, with three devices for each active layer (no average).

Figure A4: Stability of D:A3, where the lines are the average of the three devices in the respective active layer.

(43)

(44)

(45)

References

[1] L. Schmidt-Mende and J. Weickert, Organic and hybrid solar cells: an introduction. Walter de Gruyter GmbH & Co KG, 2016.

[2] A. Smets, Solar energy : the physics and engineering of photovoltaic conversion, technologies and systems. Cambridge, England: UIT Cambridge Ltd, 2016.

[3] J. Zhao, Y. Li, G. Yang, K. Jiang, H. Lin, H. Ade, W. Ma, and H. Yan, “Eﬃcient organic solar cells processed from hydrocarbon solvents,” Nature Energy, vol. 1, no. 2, pp. 1–7, 2016.

[4] M. C. Scharber and N. S. Sariciftci, “Eﬃciency of bulk-heterojunction organic solar cells,”

Progress in polymer science, vol. 38, no. 12, pp. 1929–1940, 2013.

[5] R. Søndergaard, M. Hösel, D. Angmo, T. T. Larsen-Olsen, and F. C. Krebs, “Roll-to-roll fabrication of polymer solar cells,” Materials today, vol. 15, no. 1-2, pp. 36–49, 2012.

[6] K. Orgil, “Comparison of organic and inorganic solar photovoltaic systems,” 2018.

[7] Y. Cui, Y. Wang, J. Bergqvist, H. Yao, Y. Xu, B. Gao, C. Yang, S. Zhang, O. Inganäs, F. Gao, et al., “Wide-gap non-fullerene acceptor enabling high-performance organic photovoltaic cells for indoor applications,” Nature Energy, vol. 4, no. 9, pp. 768–775, 2019.

[8] Y. Cui, H. Yao, T. Zhang, L. Hong, B. Gao, K. Xian, J. Qin, and J. Hou, “1 cm2 organic photovoltaic cells for indoor application with over 20% eﬃciency,” Advanced Materials, vol. 31, no. 42, p. 1904512, 2019.

[9] R. Chang, Chemistry. Boston: McGraw-Hill, 2010.

[10] R. Hansson, Morphology and material stability in polymer solar cells. PhD thesis, Karlstads universitet, 2015.

[11] T. Oku, Solar Cells and Energy Materials. Walter de Gruyter GmbH & Co KG, 2016.

[12] S. Li, Z. Zhang, M. Shi, C.-Z. Li, and H. Chen, “Molecular electron acceptors for eﬃcient fullerene-free organic solar cells,” Physical Chemistry Chemical Physics, vol. 19, no. 5, pp. 3440–

3458, 2017.

[13] O. Ostroverkhova, Handbook of organic materials for optical and (opto) electronic devices:

properties and applications. Elsevier, 2013.

[14] F. Zhao, C. Wang, and X. Zhan, “Morphology control in organic solar cells,” Advanced Energy Materials, vol. 8, no. 28, p. 1703147, 2018.

[15] C. Sprau, F. Buss, M. Wagner, D. Landerer, M. Koppitz, A. Schulz, D. Bahro, W. Schabel, P. Scharfer, and A. Colsmann, “Highly eﬃcient polymer solar cells cast from non-halogenated xylene/anisaldehyde solution,” Energy & Environmental Science, vol. 8, no. 9, pp. 2744–2752, 2015.

[16] A. Armin, A. Yazmaciyan, M. Hambsch, J. Li, P. L. Burn, and P. Meredith, “Electro-optics of conventional and inverted thick junction organic solar cells,” ACS Photonics, vol. 2, no. 12, pp. 1745–1754, 2015.

Electrical performance study of organic photovoltaics for indoor applications