Bulk-heterojunction Hybrid Solar Cells Based on Colloidal CdSe Quantum Dots and Conjugated Polymers

Transcription

1 Bulk-heterojunction Hybrid Solar Cells Based on Colloidal CdSe Quantum Dots and Conjugated Polymers DISSERTATION ZUR ERLANGUNG DES AKADEMISCHEN GRADES EINES DOKTOR-INGENIEUR DER TECHNISCHEN FAKULTÄT DER ALBERT-LUDWIGS-UNIVERSTÄT FREIBURG IM BREISGAU YUNFEI ZHOU FREIBURG IM BREISGAU, 2011

2 DEKAN: REFERENT: KOREFERENT: Prof. Dr. Hans Zappe Prof. Dr. Gerald A. Urban (University of Freiburg) Prof. Dr. Klaus Meerholz (Unviersity of Cologne) DATUM DER DISPUTATION:

3 Abstract Emerging alternative photovoltaic technologies such as dye sensitized solar cells (DSSCs) and organic solar cells (OSCs) have recently gained much attention and are on the step of being commercialized. Bulk-heterojunction hybrid solar cells containing inorganic nanoparticles and semiconducting polymers are still lagging behind the DSSCs and fullerene derivative-based OSCs in respect of device performance. Nevertheless, hybrid solar cells have the potential to exceed better performance while still retaining the benefits such as low-cost, thin and flexible, and easy to produce, because NCs have the features of tunable bandgap, high absorption coefficient, and high intrinsic charge carrier mobility. In addition, it is possible to synthesize stable elongated or even branched- nanostructures on the length scale of nm with desirable exciton dissociation and charge transport properties. In this dissertation, the results of a research aiming at the development of bulk-heterojunction hybrid solar cells based on colloidal CdSe quantum dot (QDs) and conjugated polymers are presented. Both the materials and device structures are investigated and optimized systematically in respect of QD synthesis and post-synthetic modification, hybrid nanocomposites formation, and device fabrication, leading to an improvement of hybrid solar cells power conversion efficiency (PCE). This dissertation begins with a general introduction of solar cells and organic/hybrid solar cells. The state-of-the-art development of bulk heterojunction hybrid solar cells is reviewed. Critical factors limiting the solar cell device performance are highlighted and strategies for further device improvement are demonstrated by giving recent examples from literature. Highly reproducible synthesis methods for CdSe QDs are applied, leading to a narrow size distribution and excellent photophysical properties. Pre-heating of the hexadecylamine (HDA) ligand and aging of the Se-TOP precursor are proven as two critical parameters for synthesizing high quality QDs. The influence of the QD characteristics such as diameter, photoluminescence (PL) peak wavelength, and PL intensity on the performance of hybrid solar cells is studied, revealing that the synthesis conditions have a crucial impact on the QD surface quality, which can be partially detected by the PL intensity. As a result, high quality QDs are desirable for achieving efficient photovoltaic devices. I

4 An effective post-synthetic hexanoic acid treatment on HDA-capped CdSe QDs before their integration into photovoltaic devices is demonstrated. Solar cells with optimized ratios of QDs to poly(3-hexylthiophene) (P3HT) exhibit PCEs of about 2.0%. A simple ligand sphere model is derived from PL quenching, TEM and dynamic light scattering results. The results indicate that an effective reduction of the immobilized ligand sphere is a crucial factor to enhance the device performance. Furthermore, extended investigations on applying the hexanoic acid treatment to different ligand-capped (i.e. mixture of trioctylphosphine (TOP) and oleic acid (OA)) CdSe QDs are presented. The comparable performance of devices based on P3HT and different ligand capped QDs indicates that the acid treatment is generally applicable to QDs with TOP/OA ligands for improving device performance. In addition, lower bandgap polymer Poly[2,6-(4,4-bis-(2-ethylhexyl)-4H-cyclopenta[2,1-b;3,4-b']-dithiophene)-alt-4,7-(2, 1,3-benzothiadiazole)] (PCPDTBT) has been used instead of P3HT as polymer part for the formation of the photoactive hybrid film. Here, optimized devices exhibit PCEs of 2.7% after spectral mismatch correction. This value is the highest reported one for spherical QD based hybrid solar cells. Comparison studies of P3HT and PCPDTBT based devices revealed that the improved PCEs in PCPDTBT:CdSe device can be mainly attributed to the increased short-circuit current density (J sc ) as a result of the improved match of the blend absorption with the solar emission spectrum, as supported by UV-Vis absorption and external quantum efficiency (EQE) measurements. In addition, it is demonstrated that low bandgap polymers which can harvest photons at longer wavelength region and have adequate energy levels are promising to be incorporated into hybrid solar cells. Finally, a summary of the results is presented and an outlook for further investigations is given. In the additional appendix part, the pre-evaluation setup and procedure for solar cells fabrication and measurement in our lab are demonstrated. II

5 Deutsche Zusammenfassung Aufstrebende alternative Photovoltaik Technologien wie die Farbstoffsolarzellen und die organischen Solarzellen haben in letzter Zeit viel Aufmerksamkeit auf sich gezogen und sind auf dem Wege kommerzialisiert zu werden. Bulk-Heterojunction Hybridsolarzellen enthalten anorganische Nanopartikel und halbleitende Polymere und hinken in der Entwicklung noch hinter den Farbstoffsolarzellen und den Fulleren basierten organischen Solarzellen in puncto Solarzelleneffizienz hinterher. Dennoch habenhybridsolarzellen das Potential ihre Effizienz zu steigern unter Beibehaltung ihrer Vorteile wie Kostengünstigkeit, Flexibilität, geringe Dicke und leichte Herstellung. Desweiteren haben Nanokristalle den Vorteil, dass sich ihre Bandlücke einstellen lässt, sie einen großen Absorptionskoeffizienten besitzen und eine hohe intrinsische Mobilität der Ladungsträger vorweisen. Zusätzlich ist es möglich, elongierte oder sogar verzweigte Nanostrukturen auf einer Längenskala zwischen nm herzustellen, die eine verbesserte Trennung der Ladungsträger sowie einen besseren Ladungstransport ermöglichen. In dieser Dissertation werden Resultate über die Entwicklung von Bulk-Heterojunction Hybridsolarzellen basierend auf kolloidalen CdSe Quantum Dots (QDs) und konjugierten Polymeren präsentiert. Beides, die Materialien und der Solarzellenaufbau sind untersucht und systematisch optimiert worden. Es werden die QD Synthese, die post-synthetische Modifikation der QDs, die Herstellung der hybriden Nanokomposite und die Herstellung der Solarzellen beschrieben, die zu einer Verbesserung der Effizienz der Hybridsolarzellen führt. Die Dissertation beginnt mit einer generellen Einführung über Solarzellen und organischen- wie auch Hybridsolarzellen. Der Stand der Technik in der Entwicklung von Hybridsolarzellen wird zusammengefasst. Entscheidende Faktoren, die die Solarzelleneffizienz limitieren werden aufgezeigt und Strategien zu ihrer Verbesserung werden anhand von neueren Beispielen aus der Literatur gegeben. Eine hoch reproduzierbare Synthesemethode für CdSe QDs wurde angewandt, die zu QDs mit einer geringen Grössenverteilung und exzellenten photophysikalischen Eigenschaften führte. Das Vorheizen des Hexadecylamin (HDA) Liganden sowie das Altern der Se-TOP Ausgangsverbindung haben sich dabei als notwendige Faktoren III

6 zur Synthese hochqualitativer QDs gezeigt. Der Einfluss der QD Größe und des resultierenden Emissionssignals einschließlich der Signalintensität auf die Effizienz entsprechender Hybridsolarzellen wurde untersucht. Es hat sich herausgestellt, dass die Synthesebedingungen einen großen Einfluss auf die Qualität der QD-Oberfläche hat, welche sich zum Teil in der Intensität der Photolumineszenz widerspiegelt, mit dem Resultat, dass hochqualitative QDs notwendig sind, um effiziente photovoltaische Zellen herzustellen. Eine effektive postsynthetische Behandlung der HDA bedeckten CdSe QDs mit Hexansäure und die Integration der behandelten QDs in Solarzellen wird aufgezeigt. Solarzellen mit optimierten Mischungsverhältnissen aus QDs und Poly-3-hexylthiophen (P3HT) führen zu Zellen mit Effizienzen von 2,0%. Ein einfaches Liganden-Sphären Modell wurde zur Erklärung herangezogen, basierend auf Photolumineszenz Auslöschungsexperimenten, Ergebnisen der dynamischen Lichtstreuung und elektronenmikroskopischen Untersuchungen. Diese Experimente zeigten, dass die effektive Reduzierung der immobilisierten Ligandensphäre ein entscheidender Faktor für die Effizienzsteigerung der Solarzellen ist. Desweiteren wurden intensive Untersuchungen des Hexansäure Behandlungsschrittes auch an CdSe QDs die die Oberflächenligandenmischung Ölsäure/Trioctylphosphin enthalten durchgeführt. Es wurden vergleichbare Solarzelleneffizienzen bei Verwendung dieser QDs mit P3HT gefunden, was die generelle Anwendbarkeit des Säurebehandlungsschrittes zeigt. Zusätzlich wurde das P3HT durch das Polymer Poly[2,6-(4,4-bis-(2-ethylhexyl)-4H-cyclopenta[2,1-b;3,4-b']-dithiophene)-alt-4,7-(2, 1,3-benzothiadiazol)] (PCPDTBT) zur Bildung des photoaktiven Filmes ersetzt, welches eine kleinere Bandlücke besitzt. Hier zeigten optimierte Testsolarzellen Effizienzen von 2.7% nach Berücksichtigung des spektralen Korrekturfaktors ( spectral mismatch correction ). Dieser Wert ist der höchste bisher erzielte für Hybridsolarzellen basierend auf sphärischen QDs. Vergleichende Studien zwischen P3HT und PCPDTBT basierten Hybridsolarzellen zeigen, dass die Effizienzsteigerungen von PCPDTBT:CdSe Solarzellen auf eine höhere Kurzschlussstromdichte zurückzuführen ist, die aufgrund der verbesserten Absorption der Energie des solaren Spektrums resultiert, was durch UV-Vis Absorptionsspektroskopie sowie durch Bestimmung der externen Quanteneffizienz bestätigt wird. Somit ergibt sich, dass Polymere mit einer kleinen Bandlücke, die IV

7 zusätzlich Photonen mit größeren Wellenlängen absorbieren können und geeignete HOMO-LUMO Energienieveaus besitzen, vielversprechend für die Integration in Hybridsolarzellen sind. Am Ende werden die Ergebnisse zusammengefasst und ein Ausblick für weitere Untersuchungen wird gegeben. Im Anhang werden der Messaufbau, der für die Vorevaluationen der Hybridsolarzellen verwendet wurde, beschrieben, ebenso wie die Fabrikation der Solarzellen und die Details zu den Solarzellenmessungen. V

8 Table of Contents Abstract Deutsche Zusammenfassung I III 1. Introduction Introduction to solar cells Organic/hybrid solar cells Motivation and context Outline Bulk-heterojunction hybrid solar cells Colloidal semiconductor nanocrystals (NCs) Devices based on CdSe NCs Strategies for efficiency improvement Hybrid solar cells based on other semiconductor NCs Synthesis of CdSe quantum dots (QDs) CdSe QD synthesis Characterization of CdSe QDs Critical parameters for high quality CdSe QDs synthesis Surface modification of CdSe QD Hexanoic acid treatment P3HT:CdSe composites Solar cell performances Ligand sphere model Influence of the QD synthesis on the device performance Hybrid solar cells based on acid treated CdSe QDs and low bandgap polymer PCPDTBT 74 VI

9 5.1. Hexanoic acid treatment on TOP/OA-capped CdSe QDs Solar cells based on low bandgap polymer PCPDTBT Summay and outlook Summary Outlook 91 Appendix 98 A1. Hybrid solar cells fabrication 98 A2. Solar cell performance test program based on LabVIEW 102 Abbreviation 110 References 112 Curriculum vitae 122 List of publications 123 Acknowledgements 126 Erklärung 129 VII

10

11 1.1 Introduction to solar cells Chapter 1 Introduction 1.1 Introduction to solar cells Renewable energy resources from natural sources, like sunlight, wind, rain, tides, and geothermal heat, will neither run out nor have any significant harmful effect on our environment. Photovoltaic (PV) energy conversion is attracting more and more attention as the need for renewable energy sources replacing current environmental critical technologies becomes more urgent. PV is an attractive way of producing electrical energy directly from sunlight, without producing noise, toxic substances and greenhouse gas emission, while requiring very little maintenance. PV technologies have found markets in various fields, ranging from consumer electronics and small scale distributed power systems to megawatt scale power plants. The discovery of the PV effect can be traced back to 1839 when a French physicist A. E. Becquerel s performed pioneering studies in liquid electrolytes 1. Until 1883, Charles Fritts built the first solar cell by coating the semiconductor selenium with an extremely thin layer of gold to form the junctions. Albert Einstein explained the photoelectric effect in 1905, and he received the Nobel Prize in Physics in 1921 for it. In the modern era, Chapin et al. 2 reported on a silicon based single p-n junction device with a solar power conversion efficiency (PCE, defined as the percentage of maximum output of electrical power to the incident light power) of 6%, which is believed to be the tipping point that transformed photovoltaic into practical technology to convert solar energy into electricity. 1

12 Chapter 1. Introduction Development of solar cells 1) First generation solar cells Currently, the PV market is still dominated by crystalline silicon wafer based solar cells, so-called first generation solar cells. The best single crystalline silicon solar cells exhibited PCEs of about 25% 3, which is approaching the theoretical Shockley-Queisser limit efficiency of 31.0% for single junction solar cells without light concentrating 4. Commercial products typically achieve module efficiencies of about 15-18% 5. However, silicon wafers are fragile, making the manufacturing processes complex limiting potential applications. In addition, high purity crystalline silicon wafers are very expensive, resulting in inherently high-cost and it may take several years to gain the payback for their purchasing and installation costs. 2) Second generation solar cells In order to simplify manufacturing and reduce costs, second generation solar cells, which are mostly associated with thin film solar cells, were developed. Second generation solar cells are much cheaper to produce than first generation cells since they are fabricated by depositing thin films of photoactive materials on substrates, using less amount of materials and cheaper manufacturing processes. Copper Indium Gallium Selenide (CIGS), CdTe, amorphous silicon (a-si), and nanocrystalline silicon (nc-si) are the most commonly used materials 6. As direct band gap semiconductors, the thin film semiconductor materials have much higher absorption coefficients than silicon, therefore much thinner semiconductor layers (< 1 um), which is times less than for Si, are required. The highest confirmed efficiencies by certified test centers for CIGS cell, CdTe cell, a-si cell, and nc-si are 20.1% a, 16.7±0.5% 7, 9.5±0.3% 8, and 10.1±0.2% 9, respectively. However, the PCEs remain lower than for the first generation solar cells, and the promise of low cost power has not been realized yet. a Press release 05/2010, Zentrum für Sonnenenergie- und Wasserstoff-Forschung Baden-Württemberg (ZSW), Germany. 2

13 1.1 Introduction to solar cells 3) Third generation solar cells Third generation solar cells are the cutting edge of solar technology and still in the research phase. They are expected to achieve reasonable efficiencies at lower costs than first and second generation technologies 4 and contain a wide range of promising technologies including multijunction tandem cells 10, dye-sensitized solar cells (DSSCs) 11 12, 13, organic solar cells based on small-molecule and polymer 14-22, and organic/inorganic hybrid solar cells 21, conjugated For example, multijunction inorganic solar cells are currently the most efficient solar cells. It consists of multiple photoactive thin films to absorb nearly the entire solar spectrum, thus generating electric power from as much of the solar energy as possible. In single band gap solar cells, efficiency is limited due to the inability to convert a broad range of photons of the solar emission into electricity. Photons with lower energy than the band gap of photoactive materials are lost, since they are not able to excite electron over the band gap. This part of photons are either not absorbed or converted into heat within the materials. Energy in the photons above the band gap energy is also lost, since only the energy necessary to generate hole-electron pair is used, and the remaining energy is converted into heat. In multijunction solar cells, the band gap of each layer can be tuned to absorb a specific range of the solar spectrum by using different alloys of III-V semiconductors. Recently, a research team from Boeing Spectrolab claims commercial availability of cells at 40.7±2.4% efficiency in GaInP/GaAs/Ge (2-terminal) device structure under an intensity of concentrated sun illumination (240 suns) 10. Fraunhofer Institute for Solar Energy Systems (FhG-ISE) in Germany and National Renewable Energy Laboratory (NREL) in United States also reported 40.8±2.4% 27 and 41.1%±2.5% b efficiencies of multijunction solar cells, respectively. With such high efficiencies, multijunction solar cells are well-suited for space application and have a good development potential. However, the cost is too high to allow a large scale use of it due to the complex device structure and the high price of materials. Figure 1.1 summarises the progress of the best research-cell PCEs based on various materials and technologies updated till June, b ached-for-multi-junction-solar-cells-at-fraunhofer-ise 3

14 Chapter 1. Introduction Figure 1.1 Overview of the best research-cell efficiencies of various PV technologies (This graph was created and prepared by the National Renewable Energy Laboratory (NREL) for the U.S. Department of Energy) Photovoltaic market Nowadays, solar cells have found markets in various application fields ranging from consumer electronics and small scale distributed power systems to megawatt scale power plants. According to a Solarbuzz report c, global PV market installations reached a record high of 6.43 gigawatt in 2009 (Fig. 1.2), representing a growth of 6% over the previous year. The top three European countries are Germany, Italy, and Czech Republic. The PV industry generated US$ 38 billion in global in 2009, while successfully raising over US$13.5 billion in equity and debt up 8% on In 2010 and also over the next 5 years, the PV industry is expected to return to high growth, and the global market will be 2.5 times its current size by 2014, and the annual industry revenues will approach US$ 100 billion then. c 4

15 1.1 Introduction to solar cells Figure 1.2 Photovoltaic market in 2009 (Data taken from Solarbuzz, Renewable energy As a renewable energy technology, PV technology has to be compared with other technologies based on certain indicators, such as energy payback time, CO 2 emissions, and the recycling management at their end of life. Energy payback time is defined as the recovery time required for generating the energy input during the whole life cycle, which includes the energy requirement for manufacturing, installation, energy use during operation, and energy needed for decommissioning. Depending on the solar cells module and installation location, the typical energy payback times are ranging between 1 to 4 years 28. With a typical lifetime of 20 to 30 years, modern solar cells generate significantly more energy over their lifetime then energy spent for their production, thus they are net energy producers. Opposite to fossil energy sources, the operation of solar cell systems are CO 2 -free, and the greenhouse gas emission occurs almost entirely during their manufacturing. In this term, the CO 2 emissions (g/kwh) of the present grid connected roof-top systems have been estimated to be significantly lower than those of fossil fuel power plants. Another important issue is the recycling of solar cells after its lifetime. Depending on the solar cells technology, they might contain small amounts of hazardous and toxic materials, such as Cd, Pb, and Se. However, there are already existing technologies 5

16 Chapter 1. Introduction for recycling solar cells and it can be therefore considered economically feasible Solar cells characterization 1) Power Conversion Efficiency Figure 1.3 Current density-voltage (J-V) characteristic of a typical solar cell in the dark (dashed line) and under illumination (solid line). Typical solar cell parameters such as short-circuit current density J sc, open-circuit voltage V oc, and the maximum power point P m are illustrated on the graph. (Image taken from Ref. 29 ) Power conversion efficiency (PCE) is one of the most import parameter to characterize solar cell performances. It is defined as the percentage of maximum output of electrical power to the incident light power. Fig.1.3 shows the current density-voltage (J-V) characteristic for a typical hybrid solar cell in the dark and under illumination. The PCE can be described as PCE = P m P in = J sc V oc FF P in (1.1) where P m is maximum power point, P in is the incident light intensity, J sc is the short-circuit current density, and V oc is the open-circuit voltage, and FF is the fill factor, which is defined as the ratio of P m to the product of J sc and V oc FF = P m J sc V oc (1.2) 6

17 1.1 Introduction to solar cells 2) Quantum efficiency External quantum efficiency (EQE) is another important parameter for solar cell characterization. It is calculated by the number of electrons extracted in an external circuit divided by the number of incident photons at a certain wavelength under short-circuit condition EQE(λ) = number of electrons number of photons = J sc(λ)/e P in (λ)/( hc λ ) = Jsc(λ)hc P in (λ)eλ (1.3) where λ is the wavelength, e is the elementary charge, h is the Planck constant, and c is the speed of light in vacuum. Note that the EQE represents the external quantum efficiency, meaning that the losses due to reflection at the surface, and/or the transmission through the device are also included in EQE. Considering the fraction of the actually absorbed photons by the photoactive layer, EQE can be converted into the internal quantum efficiency (IQE) IQE(λ) = EQE(λ) 1 Ref(λ) Tran(λ) (1.4) where Ref(λ) is the fraction of reflected light and Tran(λ) is the fraction of the transmitted light. The IQE is also very helpful in organic solar cells to investigate the physical processes occurring in the organic semiconductor materials. 3) Spectral mismatch correction A sun simulator which can simulate natural sunlight is usually used as a light source for repeatable and accurate indoor testing solar cells. In order to compare results from various devices, solar cells are characterized based on international accepted standard reporting conditions (SRC), which are referred to a cell temperature of 25 C under air mass 1.5 global (AM1.5 G) illumination spectrum at an intensity of 1000 W/m 2. This AM1.5 G condition corresponds to the spectrum and irradiance of sunlight incident 7

18 Chapter 1. Introduction upon an inclined plane at 37 o tilt towards the equator with an elevation of o above the horizon d. A Si reference solar cell is mostly used as the reference cell for calibrating a sun simulator. Usually there is a spectral mismatch in the measured short-circuit current of the solar cell with respect to a AM1.5 G reference spectrum. The reasons for this spectral mismatch are: 1) the spectrum mismatch between the light source and the AM1.5G reference spectrum, and 2) the difference on spectral response between Si reference cell and testing cell. The mismatch value can be calculated by using the spectral responsively data for the testing cell and reference cell combinations. For most organic solar cell, the short-circuit current density shows a linear relationship with the incident light intensity 30. The open-circuit voltage and fill factor are much weaker dependent on the light intensity 31, 32. Therefore, once a mismatch factor is known, the J sc and the PCE of the testing cell can be calibrated. The spectral mismatch correction especially for organic solar cells are described in detail elsewhere 33. d ASTM Standard G173, Standard Tables for Reference Solar Spectral Irradiances: Direct Normal and Hemispherical on 37 Tilted Surface, American Society for Testing and Materials, West Conshocken, PA, USA. 8

19 1.2 Organic/Hybrid solar cells 1.2 Organic/Hybrid solar cells Generally, PV technologies are currently dominated by inorganic semiconductor based solar cells. However, the manufacturing processes of traditional inorganic solar cells often involve elevated temperature, high vacuum, and numerous lithographic steps, resulting in high production costs and energy consumption. Alternatively, solar cells based on organic materials such as small molecules and conjugated polymers offer a cost-effective way to convert solar energy into electricity. In contrast to their inorganic counterparts, organic solar cells (OSCs) have several advantages. They are able to be manufactured by low-temperature processing: either from small molecules from the vapor phase, or polymers from solution. The organic materials are relatively inexpensive, and only nm thick films are required due to the high optical absorption capabilities of organic semiconductors. Additionally, roll-to-roll processing like low cost printing techniques 34 can be used for manufacturing. In Fig. 1.4a a photograph is shown illustrating a roll-to-roll manufacturing machine for fabricating flexible polymer solar cells 35. Therefore, OSCs have the promising potential to be applied in consumer products with the features of thin, flexible, light weight and low cost. Additionally, the properties of small molecules or polymers can be tailored by modifying their chemical composition, resulting in greater customization than traditional inorganic solar cells. Since Tang s pioneering work on the bilayer heterojunction OSC reported in , intensive investigation have been performed and the PCEs of devices have been improved substantially. Currently, OSCs with small active area have reached a PCE level 8% (announcement by Konarka with a certified PCE of 8.3% e ), which are close to the requirement for entering the business market 37. Several companies such as Konarka, Heliatek, Solarmer, and Plextronics have already started or are preparing to make commercial available products to offer alternative inexpensive solar power modules. It has to be mentioned that the PCEs of solar power modules based on OSCs are currently in the range of 2-4%. For example, solar bags with a polymer based OSC panel on the surface are already commercially available on the market based on the technology developed by Konarka. They are able to provide electricity to charge mobile phones and other handheld devices from solar e accessed on December 3,

20 Chapter 1. Introduction power (shown in Fig. 1.4b). The light-weight, thin-film photovoltaic material is much more versatile for various applications than traditional solar panels. Figure 1.4 (a) Photograph of a roll-to-roll manufacturing facility for fabricating flexible polymer solar cells 35.(b) Solar bag with an integrated polymer based OSC panel on the bag flap ( Device structure OSCs are typically thin film devices consisting of photoactive layer(s) between two electrodes of different work functions (see Fig. 1.5). Figure 1.5 Schematic structure of a typical Organic solar cell. (Modified image according to ref. 29 ) High work function, conductive and transparent indium tin oxide (ITO) on a flexible plastic or glass substrate is often used as anode. The conducting polymer poly(3,4-alklenedioxythiophenes):poly(styrenesulfonate) (PEDOT:PSS) is usually 10

21 1.2 Organic/Hybrid solar cells used as anode buffer material for smoothing the ITO surface, enhancing the adhesion to the upper light absorbing layer, better energy level matching, and improving the device stability by hindering oxygen and indium diffusion through the anode The photoactive layer can be either deposited thermally (small molecules) in vacuum or spin-coated (polymer) onto the ITO substrate to form a film thickness of around nm. Finally, a top metal electrode (e.g. Al, LiF/Al, Ca/Al) is vacuum deposited onto the photoactive layer. The photoactive layers can be constructed in a variety of ways, leading to single layer, bilayer, or bulk-heterojunction devices. 1) Single layer device Single layer cells are the oldest and simplest OSCs. In 1959 Kallman and Pope discovered that anthracene can be used to make OSCs, exhibiting an efficiency of only % 41. In such single layer device, the internal electric field which arises from the work function difference between two electrodes is much too weak to overcome the exciton binding energy to dissociate the excitons into free electrons and holes. For instance, only 10% of the excitons dissociate into free carriers in a pure poly[2-methoxy-5-(2-ethylhexyloxy)-1,4-phenylenevinylene] (MEH-PPV) film, while the remaining excitons decay via radiative or nonradiative recombination 42. Thus the efficiencies of single-layer polymer solar cells exhibit rather low efficiency in the 43, 44 order of 0.1% 2) Bilayer device A major advancement in organic solar cells was realized in 1986 by Tang 36, who firstly introduced the bilayer heterojunction structure by stacking two materials with suitable energy levels offsets. Solar cells with a planar junction of CuPC as donor and a perylene tetracarboxylic derivative as acceptor reached an efficiency of about 1% under AM2 75 mw/cm 2 illumination. This dramatically improvement on efficiency is mainly due to the exciton dissociation at the heterojunction which is much more efficient than in the bulk organic layer or in the organic/metal interface of a single layer device. A schematic energy diagram of a device with a bilayer heterojunction structure consisting of a donor and an acceptor is shown in Fig. 1.6, where HOMO is the highest occupied molecular orbital, and LUMO is the lowest unoccupied molecular orbital. 11

22 Chapter 1. Introduction Figure 1.6 Schematic energy level diagram of a device with a bilayer heterojunction consisting of a donor and an acceptor material. The process of electron charge transfer from the donor to the acceptor is shown in this case. (Modified image according to ref. 29 ) In the heterojunction structure, an electron donor material (D) and electron acceptor material (A) are used. Here, the HOMO and the LUMO of the donor material is higher than that of the acceptor material. Photons are absorbed by the donor materials, thus the energy from incident photons excite electrons from HOMO to LUMO to generate excitons. When they diffuse close to the D/A interface, photo-generated excitons can be efficiently dissociated into free carriers by charge transfer. The charge transfer process, i.e. electron transfer from the LUMO of the donor to the LUMO of the acceptor is energetically favorable. The exciton dissociation by charge transfer in bilayer heterojunction device is intrinsically more efficient than that in the single bulk organic semiconductor. Furthermore, during the charge transport process after exciton dissociation, the possibility for recombination losses is significantly reduced since electrons or holes transport to their respective electrodes in pure n-type or p-type layers. However, the exciton dissociation efficiency is limited due to the limiting interfacial area in bilayer heterojunction device. Since the exciton diffusion lengths in conjugated polymers are typically around nm 45-47, the optimum distance of the exciton to the donor/acceptor (D/A) interface, where charge transfer takes place and excitons dissociate into free charge carriers, should be in the same length range. This 12

23 1.2 Organic/Hybrid solar cells requirement limits the part of the active layer which has contribution to photocurrent to a very thin region near the D/A interface. In other words, excitons generated in the remaining area of the device are lost. Therefore, for efficient exciton dissociation at the heterojunction, the donor and the acceptor materials have to be in a suitable distance. 3) Bulk-heterojunction device In order to overcome the problem that not all excitons are able to reach the D/A interface, a so-called bulk-heterojunction structure 48, 49 was introduced by Yu et al. 48 and Halls et al. 49 independently. The bulk-heterojunction composite is made by mixing both the electron donor and acceptor intimately together, thus the interfacial area is dramatically increased and the distance that excitons have to travel to reach the interface is reduced. After exciton dissociation into free charge carriers, holes and electrons are transported via polymer and NCs percolation pathways towards the respective electrodes. Fig. 1.7 shows the schematic illustration of device structure and energy levels of a bulk-heterojunction solar cell. Compared to the bilayer heterojunction structure where donor and acceptor phase contact the respective anode and cathode selectively, the bulk-heterojunction requires percolated pathways for the charge carrier transporting phases to the respective electrodes. Thus the donor and acceptor phases should form a bicontinuous and interpenetrating network for efficient charge transport after exciton dissociation. Therefore, nanoscale morphology control in the blend is very important for a bulk-heterojunction device. Figure 1.7 Schematic illustration of a) device structure, and b) energy level diagram of a bulk-heterojunction solar cell. The photoactive layer is made up of blend solution consisting out of donor and acceptor materials. (Modified image according to ref 29 ) The bulk-heterojunction device structure has been used extensively since its 13

24 Chapter 1. Introduction introduction and state-of-the-art polymer based solar cells are primarily using this bulk-heterojunction device structure Recent progress reported on bulk heterojunction solar cells with internal quantum efficiency approaching 100% 50 by using poly[n-9''-hepta-decanyl-2,7-carbazole-alt-5,5-(4',7'-di-2-thienyl-2',1',3'-benzothiadia zole) (PCDTBT) as donor the fullerene derivative [6,6]-phenyl C-70-butyric acid methyl ester (PC70BM) as acceptor. It implies that essentially every absorbed photon results in a separated pair of charge carriers and that all photo-generated carriers are traveling to the respective electrodes. This result also indicates that the exciton dissociation and charge carrier transport can be very efficient in such bulk-heterojunction structure after optimization of the materials and device manufacturing procedures Working principle As shown in Fig. 1.8 photocurrent generation is a multistep process in OSCs for a typical (a) bilayer heterojunction device and (b) bulk-heterojunction device, where photons are mainly absorbed in the donor material. The physics of organic solar cells is reviewed in detail in dedicated review articles 15, 25. In short, there are following four main steps which have to be considered: photon absorption, exciton diffusion, charge transfer, charge carrier transport and collection. Figure 1.8 Schematic diagram of the photocurrent generation mechanism in typical (a) bilayer heterojunction and (b) schematic of bulk-heterojunction organic solar cells, where photons are mainly absorbed in the donor material. Photogeneration process: exciton generation (1), exciton diffusion (2), charge transfer (3), charge carrier transport and collection (4). (Image taken from ref. 29 ) 14

25 1.2 Organic/Hybrid solar cells 1) Photon absorption First, incident photons with an energy hν are absorbed mainly by the donor material and excite the electrons from the HOMO to LUMO level, creating excitons with a certain binding energy (typically mev 43, 58 ). 2) Exciton diffusion In order to generate separated negative and positive charges, the excitons need to diffuse to the D/A interface. Since excitons are neutral species, their motion is not affected by any electric field and they diffuse via random hops driven by the concentration gradient. Note that the exciton diffusion lengths are typically around 10 nm for most conjugated polymers before recombination takes place. Excitons that do not reach the D/A interface are lost for the energy conversion and have no contribution to the photocurrent. 3) Charge transfer Excitons dissociate at the D/A interface by charge transfer which is energetically favored: the LUMO level of the NCs is lower than that of the polymer, and the HOMO level of the polymer is higher than that of the NCs. The offsets in both HOMO and LUMO levels must be larger than the exciton binding energy minus the columbic binding energy of the charge-separated state 59. 4) Charge carrier transport and collection In the final step, once charge transfer has occurred at the D/A interface, separated holes and electrons are distributed within the donor and acceptor phases, respectively. Holes and electrons are then transport towards their respective electrode driven by an internal electric field deriving from the Fermi level difference of the electrodes with efficiencies depending on their mobilities during the hopping process, and consequently being collected at the respective electrodes Organic solar cells vs. inorganic solar cells Compared to the classical crystalline inorganic semiconductors (e.g. silicon), organic semiconductors are different in some fundamental aspects. The differences between organic and inorganic solar cells have significantly consequences for understanding the fundamental mechanisms of photoconversion in the two different systems. 15

26 Chapter 1. Introduction First of all, the absorption of a photon from the incident sunlight in organic solar cells does not create free charge carriers, but strongly bound electron-hole pairs so- called excitons 60. The binding energy is typically in the range of mev 43, 58, which is much higher than k B T at room temperature (k B T (300K) =26 mev). Due to their low dielectric constants and weaker non-covalent electronic interactions between organic molecules, organic solar cells are also classified as excitonic solar cells Second, the charge carrier mobilities in organic semiconductors are much lower than those in crystalline inorganic semiconductor. For example, hole mobilities for conjugated polymers range from 10-1 to 10-7 cm 2 /(Vs) 64-67, while electron mobilities are typically even lower ( cm 2 /(Vs)) 65, 68. By contrast, the hole and electron mobilities in crystalline silicon are 475 and 1500 cm 2 /(Vs) 69. The disordered structure of organic semiconductors causes the transport of carriers through hopping mechanism rather than through band-like transport. Therefore, these low mobilities of organic semiconductors limit the feasible thicknesses of the active layer to only few hundred nanometers. Fortunately, organic semiconductors have relatively high optical absorption coefficients (usually>10 5 /cm) in the UV-Vis regime, thus only ca nm thick organic layers are needed for effective absorption. Figure 1.9 schematic diagrams of the differences of band gap structures and the distribution of photo-generated carriers between (a) a conventional solar cell and (b) an organic solar cell. In the case of an organic solar cell, the concentration gradients are the driving force for carriers transport. (Images drawn according to Ref. 60 ) 16

27 1.2 Organic/Hybrid solar cells Fig. 1.9 shows the schematic diagram of the differences on band gap structure and the distribution of photo-generated carriers between conventional solar cell and organic solar cell. In conventional inorganic solar cells, the charge carriers are separated from each other by the built-in electric field of the device and travel to their respective electrode. The photo-generation is distributed throughout the active layers and photo-generated carrier concentration gradients are negligible. Therefore, photocurrent is dominantly driven by drift of minority carriers in the built-in electric field (Fig. 1.9a). In organic solar cells, the charge carriers are tightly bound to each other in the form of excitons. The excitons diffuse to the donor/acceptor (D/A) interface where they dissociate into free carriers and create a majority of carriers, leading to large concentration gradients. These high interfacial concentration gradients promote the separation of carriers. The diffusion and drift driving forces act in the same direction to separate the charge carriers (Fig. 1.9b) Photoactive material Based on the organic semiconductor components used for photoactive layers, organic solar cells can be divided into three main types: small molecule solar cells, polymer solar cells, and organic-inorganic hybrid solar cells. 1) Small molecule solar cells In 1986, Tang s 36 firstly demonstrated the bilayer OSCs with a planar heterojunction between copper phthalocyanine (CuPc) as electron donor and perylene derivative as acceptor. Later on, in order to reduce the exciton lost due to quenching at the cathode metal contact, a wide bandgap electron blocking layer (EBL) for excitons was introduced between the acceptor materials and the cathode 70, 71. Excitons are blocked at the interface between the acceptor and the EBL because of the offset in energy levels. Besides, the EBL prevents damages of the active layer due to the cathode deposition. Bathocuproine (PCB) and 4,7-dipheny1-1,10-phenanthroline (BPhen) are two commonly used materials as EBLs. Fig shows the chemical structures of commonly used materials for small molecule solar cells. A major improvement on PCEs of small molecule solar cells was the introduction of a so-called pin structure, where p, i and n refer to the type of the three different layers as predominately p-type (donor), intrinsic absorber, and n-type (acceptor). Maennig et 17

28 Chapter 1. Introduction al. 72 have reported the first OSCs based on a pin structure where the Fermi levels in the transport layers are controlled by molecular doping. In such pin devices, the contact between the transport layer and the electrode is typically ohmic due to the doping of the transport layers 73. Additionally, the doped transport layer only permits either electron or hole transport, thus blocks the opposite charge carriers and excitons. Furthermore, the position of absorber can be optimized to the place where the absorption in the optical interference pattern forming due to the reflecting back contact, since the doped transport layers do not significantly absorb light from the sun spectrum above 400 nm 72. Recently, a record PCE was announced by Heliateck GmbH, Germany for small molecules solar cells based on a pin tandem structure with certified efficiency of 8.3% and an active area of 1.1 cm 2f. Figure 1.10 Chemical structures of commonly used materials for small molecule solar cells. Shown are copper phthalocyanine (CuPc), zinc phthalocyanine (ZnPC), fullerene (C 60 ), bathocuproine (BCP), and 4,7-dipheny1-1,10-phenanthroline (BPhen).(Images taken from 2) Polymer solar cells Polymer based solar cells have several attractive features. Because the active materials used for device fabrication are soluble in most of the common organic solvents, it has the potentials to be flexible and to be manufactured in a roll-to-roll processing like low cost printing techniques 34. Conjugated polymers generally have relatively high hole mobilities but low electron mobilities. This intrinsic imbalance in carrier mobility can be overcome by the incorporation of an n-type semiconductor f accessed on December 3,

29 1.2 Organic/Hybrid solar cells material as electron acceptor to provide a pathway for electron transport. Figure 1.11 Chemical structures of commonly used materials for polymer solar cells. Shown are Poly(3-hexylthiophene-2,5-diyl) (P3HT), [6,6]-Phenyl C61 butyric acid methyl ester (PCBM), [6,6]-Phenyl C71 butyric acid methyl ester (PC 71 BM), Poly[2-methoxy-5-(2-ethylhexyloxy)-1,4-phenylenevinylene](MEH-PPV), and Poly[2-methoxy-5-(3,7 -dimethyloctyloxy)-1,4-phenylenevinylene](mdmo-ppv). (Images taken from Since the first report of photoinduced charge transfer from a conjugated polymer to a buckminsterfullerene (C60) in 1992 by Sariciftci et al. 74, the field of polymer based solar cells has been through an explosive development. After the bucky ball C60 showed a strong tendency to crystallize in the polymer matrix, a new fullerene derivative [6,6]-phenyl C61 butyric acid methyl ester (PCBM) with increased solubility was developed as electron acceptor material 48. A successful method to dissociate excitons and generate free charge carriers in conjugated polymer based solar cells was reported by Yu et al. in MEH-PPV as the electron donor and PCBM as the electron acceptor were blended to form a bulk-heterojunction photoactive layer. The solar cells based on MEH-PPV:PCBM composite showed an efficiency of about 1%, which was a major step for polymer based solar cells. In 2001, the device with poly[2-methoxy-5-(3,7-dimethyloctyloxy)-1,4-phenylenevinylene] (MDMO-PPV):PCBM blend eventually reached a benchmark PCE of 2.5% 75. During the last 6 years, research efforts have focused on poly(alkyl-thiophenes), in particular on regioregular poly(3-hexylthiophene) (P3HT) as electron donor, because of its higher hole mobility and its light absorption ability at longer wavelength compared to PPV derivatives. In 2002, the first encouraging results for P3HT:PCBM solar cells were published 76. Due to the efforts of several groups worldwide 50, 53-55, 77-84, PCE 19

30 Chapter 1. Introduction records up to 4-5% was reported in , 54, 76 and to around 6% as reported recently in , 84. Fig shows the chemical structures of commonly used materials for polymer solar cells. The main development over the last years consisted in understanding and optimizing the processing of the active layer. It was found that the optimum P3HT:PCBM weight ratio is between 1:0.8 and 1:1, and the best solvents to dissolve the polymer and PCBM are chlorobenzene (CB) and ortho-dichlorobenzene (odcb). Moreover, the device annealing conditions appeared to be mandatory to achieve high efficiency 30, In order to ensure maximum exciton dissociation at the D/A interface, as well as an efficient charge carriers transport to respective electrodes, morphology control on the nano-scale of the bulk-heterojunction composite was found to be the key parameter to reach high performance devices. After device annealing, the open-circuit voltage (V oc ) usually slightly decreased, while both the short-circuit current density (J sc ) and the fill factor (FF) increased significantly. The organization of the P3HT:PCBM is modified upon annealing 54, 56, 92, and fibrillar-like P3HT crystals embedded in a matrix believed to comprise mostly PCBM and amorphous P3HT 56. Later on, it was noticed that the molecular weight (M w ) of the polymer part has some influence on the solar cell device performance. The ideal morphology was observed for P3HT with an average M w in the range of Too low M w P3HT has inferior mobility, most likely because of the main-chain defects 66, 94, 95. On the other hand, too high M w P3HT is less soluble in DCB 93 and produces highly entangled polymer networks, requiring higher temperatures and/or longer annealing times for crystallization 96. Figure 1.12 Chemical structures of polythiophenes with head-to-tail (left) and head-to-head (right) monomer configurations. (Images taken from In addition, the influence of the polymer s regioregularity (RR), which is defined as the percentage of monomers with a head-to-tail configuration rather than head-to-head (Fig. 1.12), is found to be critical as well. A specific threshold of about 95% RR 20

31 1.2 Organic/Hybrid solar cells seems to be necessary to give best performance because of the better transport 97, 98 properties of highly RR P3HT However, the further improvement space on P3HT:PCBM devices seems to be limited, because the internal IQE for the >5% efficient devices is already approaching 100% 99, 100. Two reasons are considered as the limiting factors of P3HT:PCBM system. First, the open circuit voltage reaches only 0.7 V, which is quite small compared to the bandgap of P3HT (1.9 ev). A large amount of energy is lost when the photoexcited electron transfer from the LUMO of P3HT (-3 ev) to the LUMO of PCBM (-3.8 ev). This mismatch of LUMO offset can be overcame either by increasing the PCBM LUMO level to about 3.3 ev or by lowering both LUMO and HOMO level of the polymer to be better aligned the relative energy levels of the PCBM 86, 91, 101. The second limiting factor is the narrow absorption range of P3HT. It can absorb light within the visible spectrum up to about 650 nm, which means that most of the low energy photons of the sun emission cannot be harvested. Efforts have been taken to increase the absorption range by synthesizing novel low-bandgap polymers. For instance, poly[2,6-(4,4-bis-(2-ethylhexyl)-4h-cyclopenta[2,1-b;3,4-b]-dithiophene)-alt-4,7-(2,1, 3-benzothiadiazole)] (PCPDTBT) is one promising low-bandgap polymer (see Fig. 1.11). Muehlbacher et al. reported on devices based on this polymer reaching an efficiency of 3.2% 57. Recently, devices based on PCPDTBT:PC 70 BM system achieved PCEs up to 6.1% 102. In addition, solar cells based on thieno[3,4-b]thiophene and benzodithiophene polymer (PTBs) family exhibited record efficiency of 7.4% 103, showing the bright potential of bulk-heterojunction polymer solar cells reaching the threshold for commercialization. 3) Hybrid solar cells Replacing the fullerenes as organic nanoparticles for polymer solar cells by inorganic nanocrystals (NCs) such as colloidal inorganic semiconductor NCs as electron acceptors is an alternative approach leading to so-called hybrid solar cells 24-26, 104, 105. NCs based on CdS, CdSe, CdTe, ZnO, SnO2, TiO2, Si, PbS, and PbSe have been used so far as electron acceptors. Colloidal NCs synthesized in organic media are usually soluble in common organic solvents thus they are able to be incorporated into conjugated polymers which are soluble in the same solvents. By tuning the diameter 21

32 Chapter 1. Introduction of the NCs, their band gap as well as their energy level can be varied based on the quantum size effect 106. Furthermore, quantum confinement leads to an enhancement of the absorption coefficient compared to that of the bulk materials 107. As a result, in the NCs/polymer system, both components have the ability to absorb incident light, unlike the typical polymer/fullerene system where the fullerene contributes very little to the photocurrent generation 108, 109. In addition, NCs can provide stable elongated structures on the length scale of nm with desirable exciton dissociation and charge transport properties 110. The overview of the development of bulk-heterojunction hybrid solar cells and strategies for device performance improvement will be discussed in detail in Chapter Motivation and context This PhD work was performed within the PhD program of the German Research Foundation (DFG) graduate school GRK 1322 Micro Energy Harvesting. This graduate school is supported by the Department of Microsystems Engineering (IMTEK) and the Freiburg Materials Research Center (FMF), the Albert-Ludwigs-University Freiburg, and the Fraunhofer Institute for Solar Energy Systems (FhG-ISE) which is an associate partner of the graduate school. Three research areas are included in this graduate school: A. Conversion Mechanism, B. Materials and Storage Mechanism, and C. Energy and System Management. In the frame of the research area B: materials and storage mechanism, this PhD work is aiming to develop bulk-heterojunction nanocomposites for photovoltaics. These nanocomposites consist out of conjugated polymers as electron donors and inorganic NCs (e.g. CdSe) as electron acceptors. Both materials and device structure are investigated and optimized. The intention is to enhance the carrier mobility and balance the electron and hole transport within the system. The increase of the solar cell efficiency and the investigation of fundamental principles are targeted. 1.4 Outline Chapter 1 presents a general introduction of solar cells and organic/hybrid solar cells, their history and development. 22

33 1.4 Outline Chapter 2 reviews the state-of-the-art development of bulk heterojunction hybrid solar cells based on colloidal nanocrystals and conjugated polymers. Critical factors limiting the solar cell device performance are highlighted and strategies for further device improvement are demonstrated by giving recent examples from literature. Chapter 3 presents highly reproducible synthesis methods for CdSe QDs, leading to a narrow size distribution and excellent photophysical properties. Pre-heating of the hexadecylamine (HDA) ligand and aging of the Se-TOP precursor are proven as two critical parameters for synthesizing high quality QDs. The influence of the QD characteristics such as diameter, photoluminescence (PL) peak wavelength, and PL intensity on the performance of hybrid solar cells is studied, revealing that the synthesis condition has a crucial impact on the QD surface quality, which can be partially detected by the PL intensity. Chapter 4 presents an effective post-synthetic hexanoic acid treatment on HDA-capped CdSe QDs before their integration into photovoltaic devices is demonstrated. Solar cells with optimized ratios of QDs to P3HT exhibit PCEs of about 2.1%. A simple ligand sphere model is derived from PL quenching, TEM and dynamic light scattering (DLS) results to explain the improved PCEs. The results indicate that an effective reduction of the immobilized ligand sphere is a crucial factor to enhance the solar cell performance. Chapter 5 describes extended investigations on applying the hexanoic acid treatment to trioctylphosphine (TOP)/oleic acid (OA capped CdSe QDs. The comparable performance of devices based on P3HT and different ligand capped QDs indicates that the acid treatment is generally applicable to QDs with TOP/OA ligands for improving device performance. In addition, by using the low bandgap polymer Poly[2,6-(4,4-bis-(2-ethylhexy)-4H-cyclopenta[2,1-b;3,4-b']-dithiophene)-alt-4,7-(2,1,3-benzothiadiazole)] (PCPDTBT) as electron donor, optimized devices exhibited PCEs of 2.7% after spectral mismatch correction. This value is the highest reported one for spherical CdSe QDs based hybrid solar cells. Comparison studies on devices using P3HT and PCPDTBT as donor material reveal that the polymer has an essential impact on the absorption properties of the blend film as well as the device performance consequently. Finally, chapter 6 summaries the results presented in this dissertation, and describes 23

34 Chapter 1. Introduction an outlook for further investigation and potential device improvement. In an additional appendix, the pre-evaluation procedure for solar cells fabrication (A1) and measurement setup (A2) in our lab are presented. 24

35 2.1 Colloidal semiconductor nanocrystals Chapter 2 Bulk-heterojunction hybrid solar cells (The main content of this chapter was published in Energy & Environmental Science 3, 1815 (2010) entitled Bulk-heterojunction hybrid solar cells based on colloidal nanocrystals and conjugated polymers as a review article.) 2.1 Colloidal semiconductor nanocrystals Due to the decreased size of semiconductor nanocrystals (NCs) down to the nanometer scale, quantum effects become dominant thus a number of physical (e.g. mechanical, electrical, optical, etc.) properties change when compared to those of bulk materials. For example, the quantum confinement effect 106 can be observed once the diameter of the material is in the same magnitude as the wavelength of the electron wave function. Along with the decreasing size of NCs, the energy levels of NCs turn from continuous states to discrete ones, resulting in a widening of the band gap apparent as a blue shift in optical properties. Figure 2.1 Schematic illustration of the energy level distribution of a bulk semiconductor and a semiconductor NC. (Image taken from ref. 111 ) 25

36 Chapter 2. Bulk-heterojunction hybrid solar cells As shown in Fig. 2.1, a bulk semiconductor has continuous conduction and valence energy bands separated by a fixed energy gap, whereas a semiconductor NC is characterized by discrete atomic-like states and a size dependent energy gap. In general, there are two distinct routes to produce NCs: by physical approach where they can be grown by lithographic methods, ion implantation, and molecular beam deposition; or by chemical approach where they are synthesized by the method of colloidal chemistry in a solvent medium. Due to their scalability and the relative simplicity of the process involved, colloidal synthetic methods are widely used and are promising for large batch production and commercial applications. Due to their unique optical and electrical properties, colloidal semiconductor NCs have attracted numerous interests and have been explored in various applications like light-emitting diodes (LEDs) 112, 113, fluorescent biological labeling 114, lasers 115, and solar cells 110. Colloidal semiconductor NCs with suitable bandgap and energy levels can be incorporated into conjugated polymers to form so-called bulk-heterojunction hybrid solar cells. Fig. 2.2 shows the energy levels (in ev) of commonly used conjugated polymers as donors and NCs as acceptors for bulk-heterojunction hybrid solar cells. The Fermi levels of the electrodes and the energy levels of PCBM are shown as well. The variation of the values for the energy levels derived from Ref. 57, is due to different applied measurement methods for extracting the respective values of the HOMO-LUMO levels such as cyclic voltammetry (CV), X-ray photoelectron spectroscopy (XPS), ultra-violet photoelectron spectroscopy (UPS), and due to different experimental boundary conditions. In this chapter, recent state-of-the-art results will be summarized and concepts for improving the device performance are highlighted. 26

37 2.1 Colloidal semiconductor nanocrystals Figure 2.2 Fig. 6 Energy levels (in ev) of commonly used conjugated polymers and semiconductor NCs, as well as the electrode Fermi levels (in ev). Values for CdSe are for 4.8 nm QD and 4.6 nm QD, respectively. ZnO values correspond to NRs with diameters of 55 nm and 100 nm, respectively. In the literature the published values for the energy levels of the donor and acceptor materials differ slightly depend on e.g. different applied experimental methods to extract the HOMO-LUMO levels such as cyclic voltammetry (CV), X-ray photoelectron spectroscopy (XPS), ultra-violet photoelectron spectroscopy (UPS) and due to different experimental boundary conditions. (Image taken from Ref. 29 ) 27

38 Chapter 2. Bulk-heterojunction hybrid solar cells 2.2 Devices based on CdSe NCs CdSe NCs were the first NCs being incorporated into solar cells which still exhibit the highest PCEs compared to devices with NCs from other materials, and are under extensive studies for utilization in hybrid solar cells. CdSe NCs have some advantages: they absorb at a useful spectral range for harvesting solar emission from 300 nm to 650 nm (shown in Fig. 2.3a), they are good electron acceptors in combination with conjugated polymers, and the synthetic methods for their synthesis are well-established. In 1996, Greenham et al. 127 firstly reported on the incorporation of CdSe spherical quantum dots into MEH-PPV. At a high concentration of NCs of around 90% by weight (wt%), external quantum efficiencies (EQE) up to 10% were achieved, indicating an efficient exciton dissociation at the polymer/ncs interface. Although the phase separation, between the polymer and the NCs was observed to be in the range of nm, the PCEs of devices were very low of about 0.1%. This was attributed to an inefficient electron transport via hopping from NCs to NCs. After the breakthrough synthetic work of Peng et al. 128, shape control of CdSe NCs was introduced and different elongated CdSe structures were obtainable. Fig. 2.3b illustrates different shapes of NCs used in hybrid solar cells as electron acceptor materials such as spherical quantum dots (QDs), nanorods (NRs) and tetrapods (TPs). Figure 2.3 (a) Absorption spectrum of CdSe QDs with different sizes. Inset: Photoluminescence (PL) of differently sized QDs (3 nm - 6 nm) under UV irradiation. (b) Schematic illustration of different shapes of NCs, from left to right: quantum dot (QD), nanorod (NR), and tetrapod (TP). (Images taken from Ref. 29 ) Numerous approaches were published regarding synthesizing various morphologies and structures of CdSe NCs and their application in hybrid solar cells. A significant 28

39 2.2 Devices based on CdSe NCs advance was reported in 2002 by Huynh et al. 110, who demonstrated efficient hybrid solar cells based on elongated CdSe nanorods and P3HT. Elongated nanorods were used which naturally provide directed pathways for effective electron transport. Additionally, P3HT was used as donor material since it has a comparatively high hole mobility and absorbs at a longer wavelength range compared to PPV derivatives 76. By increasing the nanorod length, improved electron transport properties were demonstrated resulting in an improvement of the EQE (Fig. 2.4a). The optimized devices consisting out of 90wt% pyridine treated nanorods (7 nm in diameter and 60 nm in length) and P3HT exhibited an EQE over 54% and a PCE of 1.7%. Fig. 2.4b shows a J-V characteristic of an optimized device under AM1.5 G illumination. This work highlights the importance of shape control of materials on the nanometer scale, which opens new opportunities for the development of future generation solar cells. Later on, numerous literatures have been published for hybrid solar cells based on conjugated polymers and QDs 52, NRs 110, , TPs 51, 130, and hyperbranched 139 CdSe NCs, exhibiting the highest PCEs of 2.7% 140, 2.6% 132, 3.2% 51, and 2.2% 139, respectively. The selected performance parameters of hybrid solar cells based on CdSe NCs and conjugated polymers are summarized in Table 2.1. Figure 2.4 (a) Dependence of EQE on the NR lengths. (b) Current density (J)-voltage (V) characteristic of a CdSe NR (7 nm in diameter, 60 nm in length) containing device under AM1.5 G illumination. (Images taken from Ref. 110 ) 29

40 Chapter 2. Bulk-heterojunction hybrid solar cells Table 1. Selected performance parameters of hybrid solar cells based on CdSe NCs and conjugated polymer. All measurements were performed under AM1.5 G at intensities of mw/cm 2 illumination. CdSe Aspect treatment NC Mismatch Polymer d(nm) Solvent Cathode NCs ratio on NC wt% Correction Voc(V) Jsc(mA/cm 2 ) FF PCE(%) Ref TP PCPDTBT pyr CHCl 3 /Pyr/TCB 90 LiF/Al yes TP OC 1 C 10 -PPV 5 10 pyr CHCl 3 /Pyr/TCB 86 LiF/Al yes QD PCPDTBT hex acid CB 87.5 Al yes NR P3HT pyr/dithiol CHCl 3 /TCB Al no NR P3HT 5 13 pyr CHCl 3 /Pyr/TCB 90 Al yes TP APFO pyr CHCl 3 /Pyr/p-xylene 86 Al yes Hybrch P3HT pyr CHCl 3 /Pyr/TCB 86 Al no QD P3HT hex acid DCB 87 Al yes TP OC 1 C 10 -PPV 5 10 pyr CHCl 3 /Pyr/DCB 86 Al no QD P3HT 1 pyr/btylmn CHCl 3 / btylmn 92 Al no NR P3HT pyr CHCl 3 /Pyr 90 Al no NR P3HT a pyr CHCl 3 /Pyr 40 b Al no TP MEH-PPV pyr CHCl 3 /Pyr/CB 90 Al no TP P3HT pyr CHCl 3 /Pyr/TCB 90 Al yes QD MEH-PPV pyr CB/Pyr 89 Ba/Al no QD P3HT pyr CB/Pyr 89 Al no NR P3HT tboc 90 Al no Abbreviations: TP: Tetropods, NR: Nanorods, Hybrch: Hyperbranched, QD: Quantum Dots, MEH-PPV: poly(2-methoxy-5-(2`-ethyl)-hexyloxy-p-phenylene vinylene), OC 1 C 10 -PPV: poly(2-methoxy-5-(3,7 -dimethyloctyloxy)-p-phenylenevinylene), APFO-3: poly(2,7-(9,9-dioctylfluorene)-alt-5,5- (4,7 -di-2-thienyl-2,1,3,-benzothiadiazole)), pyr: pyridine, dithiol: benzene1,3-dithiol, hex acid: hexanoic acid, btylmn: butylamine, tboc: tert-buthoxycarbonyl. a functionalized P3HT, b volume percent 30

41 2.3 Strategies for efficiency improvement 2.3 Strategies for efficiency improvement NC surface modification In principle, polymer/nc hybrid solar cells should perform better compared to polymer/fullerene systems due to the additional higher absorption coefficient of inorganic semiconductor NCs and potential higher intrinsic electron mobility compared to that of PCBM (10-3 cm 2 /V -1 s ). Nevertheless, there has been no higher PCEs reported in hybrid solar cells compared to fullerene based OSCs so far. One important reason is that despite the relatively high intrinsic conductivity within the individual NCs, the electron mobility transport through the NC network in hybrid solar cells is quite low, which could be mainly attributed to the electrical insulating organic ligands on the NC surface 52. In most cases, the ligands used for preventing aggregation during the growth of the NCs contain long alkyl chains, such as oleic acid (OA) or trioctylphosphine oxide (TOPO), which form electrically insulating layers thus impedes an efficient charge transfer between NCs and polymer, as well as electron transport between NCs 127, 144. Ginger et al. 145 have investigated charge injection and charge transfer in thin films of spherical CdSe NCs covered with TOPO ligand sandwiched between two metal electrodes. Very low electron mobilities in the order of 10-5 cm 2 V -1 s -1 were measured, whereas the electron mobility of bulk CdSe is in the order of 10 2 cm 2 V -1 s In order to overcome this problem, extensive investigations on the surface modification of NCs have been reported based on ligand exchange approaches using various shorter capping ligands. The interparticle distance is expected to be reduced, thus facilitating the electron transport through the NC domain phases. For example, post-synthetic pyridine treatment of the NCs is the most commonly used and effective procedure, leading to the state-of-the-art efficiencies for hybrid solar cells 51, 110, 129, 130, 132, 133, , 142. Generally, as-synthesized NCs are washed by methanol several times and consequently refluxed in pure pyridine at the boiling point of pyridine around 115 o C under inert atmosphere overnight. Afterwards, the NCs are precipitated with hexanes, recovered by centrifugation, and then dispersed into a mixture of chloroform/pyridine (90:10, vol/vol). This pyridine treatment is believed to replace the synthetic insulating ligand with shorter and more conductive pyridine molecules. Treatments with other materials such as chloride 147, amine 131, and thiols 148, 149 were also investigated. Aldakov et al. 149 systematically investigated CdSe NCs modified by various small ligand molecules with nuclear 31

42 Chapter 2. Bulk-heterojunction hybrid solar cells magnetic resonance (NMR), optical spectroscopy and electrochemistry, although their hybrid devices exhibited low efficiencies. Olson et al. 131 reported on CdSe/P3HT blended devices exhibiting PCEs up to 1.77% when butylamine was used as a shorter capping ligand for the NCs. In an alternative approach, shortening of the insulating ligands by thermal decomposition was demonstrated and led to a relative improvement of the PCEs of the CdSe/P3HT-based solar cells 134. However, NCs after ligand exchange with small molecules tend to aggregate and precipitate out from the organic solvent once the long alky chain ligands are replaced 110, 150, resulting in difficulties to obtain stable mixtures of NCs and polymer. Recently, we have demonstrated a novel post-synthetic treatment on spherical CdSe QDs using a non-ligand-exchange approach 52, where the NCs were treated by a simple and fast hexanoic acid-assisted washing procedure. Figure 2.5 (a) J-V characteristic of a device containing 87 wt% CdSe QDs and P3HT as photoactive layer under AM1.5G illumination, exhibiting a PCE of 2.1% after spectral mismatch correction (Inset: Photograph of the hybrid solar cell device structure). (b) Schematic illustration of the proposed QD sphere model: an outer insulating HDA ligand sphere is supposed to be responsible for the insulating organic layer in untreated QDs directly taken out of the synthesis matrix and is effectively reduced in size by methanol washing and additional acid treatment (Images taken from Ref. 52 ). 32

43 2.3 Strategies for efficiency improvement Devices with optimized ratios of QDs to P3HT exhibited reproducible PCEs up to 2% after spectral mismatch correction which could meanwhile be exceeded (Fig. 2.5a). This is the highest reported value for a CdSe QDs based hybrid solar cell so far. It is notable that the FF is relatively high up to 0.54, implying a good charge carrier transport capability in the devices. A simple reduced ligand sphere model was proposed to explain the possible reason for improved photovoltaic device efficiencies after acid treatment as shown in Fig. 2.5b. By the assistance of hexanoic acid this immobilized insulating spheres formed by HDA ligands are effectively reduced in size due to the salt formation of HDA. This organic salt is also much more easily dissolved in the supernatant solution than unprotonated HDA and can be separated easily from the QDs by subsequent centrifugation. Another advantage of avoiding the exchange of the synthesis capping ligands is that the QDs retain a good solubility after acid treatment, resulting in reproducible performance as well as allowing a high loading of the CdSe QDs in the blend, which is preferable for an efficient percolation network formation during the annealing of the photoactive composite film Polymer functionalization From the polymer side, modifications such as end-group functionalization have been demonstrated as a route for improving the dispersion in solvents and electronic interaction between polymer and NCs 151. Liu et al. 133 have shown that pyridine-treated CdSe NRs can form well dispersed composite films with amine-terminated regioregular P3HT, exhibiting a maximum PCE of 1.4%. The end-group amine-functionalized P3HT is expected to provide intimate contact between NCs and polymer through covalent interactions, and thereby enhancing the NRs miscibility with P3HT, resulting in a favorable morphology improving the charge transfer between them. Other approaches to end-group functionalization of P3HT such as using H/(-SH), H/Br, and Br/allyl as terminating chemical groups have been reported 152. Hybrid solar cells with CdSe QDs and Br/allyl-terminated and H/(-SH)-terminated P3HT led to efficiencies of 0.9% and 0.6%, respectively. Zhang et al. 153 have demonstrated a route for directly attaching P3HT on CdSe NR surfaces, by coupling vinyl-terminated P3HT to CdSe NRs with arylbromide-functionalized phosphine oxides and thiols. Additionally, the direct synthesis of uncapped NCs inside the polymer phase is expected as an ideal situation, since the effects of the capping 33

44 Chapter 2. Bulk-heterojunction hybrid solar cells ligand on charge transfer and charge transport are eliminated and the step for transferring NCs to the polymer solution can be bypassed. Dayal et al. 154 have reported on the direct synthesis of homogeneously dispersed CdSe QDs in a P3HT solution. Photoinduced charge separation was observed, indicating that this composite could be a promising material for hybrid solar cells Photon absorption In order to harvest the maximum possible amount of the solar energy, absorption of a large fraction of the incident photons is required. Generally incident photons are mainly absorbed by the donor polymer materials and partially also from the inorganic NCs. For example in blends containing 90 wt% CdSe nanoparticles in P3HT, about 60% of the total absorbed light energy can be attributed to P3HT due to the strong absorption coefficient 141. Previous work mostly focused on systems based on CdSe NCs and P3HT as a high hole mobility polymer. For examples, by using P3HT as donor material, hybrid solar cells with spherical QDs, NRs, and hyperbranched CdSe NCs exhibited the best efficiencies of 2.0% 52, 2.6% 132, 135, and 2.2% 139, respectively. However, due to the poor overlap between the P3HT absorption spectrum and the solar emission spectrum 124, further improvement on PCEs seems to be difficult to obtain with this polymer system. Figure 2.6 AM1.5 G photon flux (black) and the integrated values (red) of photons as well as the maximum theoretical achievable short circuit current densities as a function of the maximal absorbed wavelengths (which results from the band-gap of the respective material). Crystalline silicon, P3HT, and PCPDTBT are shown as examples for comparison. (Irradiance data is taken from NREL, Image taken from Ref 29 ). 34

45 2.3 Strategies for efficiency improvement Fig. 2.6 shows the photons flux and its integral under AM1.5 G conditions (the AM1.5 G solar irradiance data is taken from NREL g ). Two different scales are shown on the right side: one axis shows the integral from nm of the number of photons which can be absorbed by having a certain band gap. The second axis shows the maximum theoretical J sc calculated by assuming that all photons are absorbed up to the band gap and converted into electrons without any losses (e.g. EQE is constant 1). Crystalline silicon has a band gap of 1.1 ev and can absorb up to 64% of the photons under AM1.5 G illumination, with a theoretical achievable current density J sc of about 45 ma/cm 2, while in the case of P3HT having a band gap of 1.85 ev, only 27% photons can be absorbed, resulting in a maximal J sc of 19 ma/cm 2. By using the low band gap polymer PCPDTBT with a band gap of about 1.4 ev, more photons can be absorbed theoretically, leading to a maximum J sc up to 32 ma/cm 2. Therefore, polymers with a smaller band gap absorbing at longer wavelengths are promising donor materials to increase the PCE of devices. Dennler et al. 12 demonstrated that for a minimum energy offset of 0.3 ev between the donor and acceptor LUMO level, PCEs of >10% are practical available for a donor polymer with an ideal optical band gap of ~1.4 ev. Most low band gap polymers are from the material classes of thiophene, fluorene, carbazole, and cylopentadithiophene based polymers, which are reviewed in detail in review articles 12, 23, 124. Among those low band gap polymers, poly[2,6-(4,4-bis-(2-ethylhexyl)-4h-cyclopenta[2,1-b;3,4-b]-dithiophene)-alt-4,7-(2,1, 3-benzothiadiazole)] (PCPDTBT, chemical structure shown in Fig.3) with a band gap of ~1.4 ev and a relatively high hole mobility up to cm 2 V -1 s appears to be an excellent candidate as a photon-absorbing and electron donating material 156. OSCs based on PCPDTBT:PC 70 BM system achieved already efficiencies up to 5.5% 157 and 6.1% 102. Recently, a bulk-heterojunction hybrid solar cell based on CdSe tetrapods and PCPDTBT was reported by Dayal et al. 51 with a certified PCE of 3.13% (measured by NREL). This is up to date the highest efficiency for colloidal NCs based bulk-heterojunction hybrid solar cells. As shown in Fig. 2.7, devices out of PCPDTBT and CdSe tetrapods, exhibited an EQE of >30% in a broad range from 350 nm to 800 g 35

46 Chapter 2. Bulk-heterojunction hybrid solar cells nm, which is the absorption band of the polymer. It is notable that the devices reached very high J sc values above 10 ma/cm 2, indicating that the broad absorption ability of the photoactive hybrid film consequently contributes to the photocurrent. Figure 2.7 EQE and absorption spectra of a device containing CdSe tetrapods and PCPDTBT exceeding a PCE of 3 % (Image taken from Ref. 51 ). It should be noted that lowering the band gap of photo-absorbing materials below a certain limit will lead to a decrease in efficiency, because the maximum V oc is limited by the band gap. In addition, the energy of absorbed photons with a larger energy than the band gap will be wasted as the electrons and holes relax to the band edges. This limitation might be overcome by multiple exciton generation in semiconductor NCs, where a single high-energy absorbed photon can produce multiple electron-hole pairs. This phenomenon has been studied using PbSe, CdSe, and PbS NCs Very high efficiencies exceeding 42% for solar cells with NCs having a band gap of 0.45 ev are theoretically possible, assuming that additional carriers from multiple exciton generation can be extracted efficiently 161. Nevertheless the contribution of multiple exciton generation in inorganic NCs to the solar cell efficiency is quite controversial in literature up to date 162 since initial results could not be verified and it needs further more convincing experimental verification. So far there is no practical evidence that the additional carriers can be extracted quickly enough to compete with the rapid exciton recombination processes within the NCs Control of the nanomorphology Generally, by controlling the morphology of the active layer, the performance of 36

47 2.3 Strategies for efficiency improvement bulk-heterojunction solar cells can be increased, because the efficient charge transfer, transport and collection strongly depend on the nanoscale morphology of the composite film 163, 164. For example, the crystallization of P3HT, induced during a thermal annealing step during device preparation, improves the light absorption property and hole mobility of the polymer 53, 54, 56. There are many factors which can have an influence on the formation of phase separated morphologies in the polymer/ncs blends, including the different solubilities of polymers and NCs in solvents, differential evaporation in solvent mixtures, NC aggregation in concentrated solutions, entropic effects, spinodal decomposition, convective instabilities during spin coating, and surface energies at the substrate and air interfaces 105. Huynh et al. 144 have investigated intensively the morphology control in P3HT/CdSe NC hybrid solar cells. Solvent mixtures of chloroform and pyridine were demonstrated as an effective way to control the morphology of polymer/ncs films. The nanoscale phase separation between NCs and polymer could be adjusted by varying the concentration of the solvent mixture. An optimum concentration for pyridine in chloroform of about 8% by volume was observed. Thermal annealing has been found to be an effective method to control the nanoscale morphology of photoactive films thus enhancing the performance of hybrid solar cells 131, 144. Besides the effect of polymer crystallization, the strong enhancement of the photocurrent after annealing is explained by removal of both interfacial and excess ligands, which act as a barrier for charge transfer from polymer to NCs and NCs to NCs. Furthermore, ligands can act as non-radiative recombination sites for excitons in the polymer 144. In addition, the use of suitable solvents for processing NC/polymer blends has been demonstrated as another crucial approach for nanomorphology control. Sun et al. 132 reported on an improved PCEs of P3HT/CdSe nanorod devices by using the high boiling point solvent 1,2,4-trichlorobenzene (TCB). Fig. 2.8 shows the comparison of EQE (a) and J-V characteristics (b) for devices fabricated by using TCB, thiophene, and chloroform as solvents for the materials during the spin coating of the respective photoactive layers. By using TCB as solvent, the EQE and the J sc of the device have significantly improved. It is proposed that the slow evaporation rate during the spin-coating process results in a favorable ordering of P3HT, which improves the hole mobility. By using branched CdSe NCs, e.g. tetrapods and poly(2-methoxy-5-(3,7 -dimethyloctyloxy)-p-phenylenevinylene) (OC 1 C 10 -PPV) as 37

48 Chapter 2. Bulk-heterojunction hybrid solar cells the donor polymer processed with high boiling point solvent TCB, further improvement on the PCE up to 2.8% has been achieved 136. It is explained that the tetrapods seem to preferentially segregate towards the vertical direction to the electrodes, which is beneficial for efficient electron transport through the NC phases. Figure 2.8 (a) EQEs of hybrid solar cells with 90wt% CdSe NRs and P3HT fabricated by using chloroform (solid line), thiophene (dashed line) and TCB (dotted line) as solvents for the materials during the spin coating process. (b) J-V characteristics for the same devices under AM1.5 G illumination (Images taken from Ref. 132 ). Similar to thermal annealing, solvent annealing is another effective way to improve the device performance, while the film is still in the liquid phase after spin-coating stored in a confined volume (such as a glass Petri dish) allowing the solvent to evaporate very slowly 53, 165. Recently, Wu et al. 135 have demonstrated an approach to enhance the charge separation and charge transport by using chemical vapor annealing treatment. The film of pyridine treated CdSe NRs mixed with P3HT was put in a closed glass vessel while the film was still wet. Benzene-1,3-dithiol was then added at the bottom edge of the vessel and the whole vessel was heated under 120 o C for 20 min. Compared to the non-treated devices, the PCEs have been significantly improved from 1.56% to 2.65%. The effective ligand exchange reaction and related phase separation during the annealing process are suggested to be the responsible reason for the performance enhancement. Ideally, an interdigital donor acceptor configuration would be a perfect structure for efficient exciton dissociation and charge transport (Fig. 2.9a). In such an ideal structure, the distance from exciton generation sites, either in the donor or the acceptor phase, to the D/A interface would be in the range of the exciton diffusion length. After exciton dissociation, both holes and electrons would be transported within their pre-structured donor or acceptor phases and find their direct percolation pathway to the respective electrodes easily. 38

49 2.3 Strategies for efficiency improvement Fig. 2.9 (a) Schematic illustration of an ideal interdigital structure of the light absorbing layer in a hybrid solar cell. (Image taken from Ref. 29 ) (b) SEM image of nanostructured P3HT film formed by nano-imprinting with a silicon mold (Image taken from Ref. 166 ). One way to approach this nanostructure is to use a polymer brush as the electron donor in the device, where polymer chains are grown directly from a substrate 167, 168. Snaith et al. 167 have demonstrated polyacrylate brushes with triphenylamine side groups grown onto ITO substrates, showing improved hole transport properties compared to spin-coated films of the same polymer. Afterwards the brushes were soaked in the CdSe NCs solution, resulting in a nanostructured photoactive layer. Although the PCEs of devices are relatively low due to the insufficient light absorption from a large band gap polymer, internal quantum efficiencies as high as 50% were measured, indicating good exciton dissociation and charge collection properties within the nanostructured film. More practically, He et al. 166 have demonstrated a nano-imprinting process to form a nanostructured polymer blend film. The spin-coated film of the electron donor P3HT was firstly patterned by a structured Si mold, as shown in Fig. 2.9b. Then the patterned P3HT film was used as a mold in the second imprinting step for the acceptor polymer poly((9,9-dioctylfluorene)-2,7-diyl-alt-[4,7-bis(3-hexylthien-5-yl)-2,1,3-benzothiadiaz ole]-2,2 -diyl) (F8TBT), which was spin-coated on a thermally evaporated Al cathode. This so-called double imprinting process can produce an interdigitated polymer bilayer film with a precisely defined structure. The smallest feature sizes of nanostructure are as small as 25 nm on a 50 nm pitch, leading to the distance between neighboring heterojunctions at or below the exciton diffusion length. OSCs with optimized interpenetrating nanostructures exhibited PCEs of 1.9%, which is among the best for polymer-polymer blend devices. Although only few systematic approaches have been reported, such kind of device structuring could be also applied to hybrid solar cells. 39

50 Chapter 2. Bulk-heterojunction hybrid solar cells Figure 2.10 (a) TEM picture of a cross-section of a P3HT/ZnO NC hybrid solar cell. (b) Restructured three-dimensional volumes of P3HT/ZnO layer with a thickness of 100 nm based on TEM tomography. ZnO NCs appear yellow, P3HT appears transparent (Images taken from Ref. 169 ). In addition, three dimensional TEM tomography has been demonstrated helpful to get feedback about the nanoscale morphology towards understanding the structure function relations 170. For bulk heterojunction solar cells this method delivers the possibility of an accurate three dimensional analysis of the nanophase distribution of the donor-acceptor blend. Therefore, charge extraction pathways as well as the volume fractions of donor and acceptor phase can be completely visualized. Oosterhout et al. 169 demonstrated the visualization of P3HT/ZnO NC bulk heterojunction nanoscale morphologies in three dimensions (Fig. 2.10). By statistically analyzing the three-dimensional morphology, information about the spherical contact distance and percolation pathways can be extracted, obtaining a consistent and quantitative correlation between device performance and nanoscale morphology. 2.4 Hybrid solar cells based on other semiconductor NCs ZnO NCs have also attracted a lot of attention for being used in hybrid solar cells because they are less toxic than typical II-VI semiconductors and are relatively easy to synthesize in large quantities. Devices based on blends of MDMO-PPV and ZnO NCs at an optimized NC content (67wt%) presented the best PCE of 1.4% 171. By using P3HT as donor polymer which has a higher hole mobility, the efficiency was optimized up to 2% with a composite film containing 50 wt% ZnO NCs 169. However, because of the relatively large band gap, the contribution to the absorption of light from ZnO NCs is very low. Another disadvantage is the low solubility of ZnO NCs in 40

51 2.4 Hybrid solar cells based on other semiconductor NCs solvents which are commonly used for dissolving conjugated polymers 172. This problem of processing ZnO NCs together with polymers to obtain well-defined morphologies hindered up to now further improvement of the solar cell performance of ZnO based hybrid solar cells. Low band gap NCs such as CdTe, PbS, and PbSe NCs are promising acceptor materials due to their ability of absorbing light at longer wavelengths which may allow an additional fraction of the incident solar spectrum to be absorbed. For instance, CdTe NCs have a smaller band gap compared to CdSe NCs, while their synthesis routes including shape control are similar to CdSe NCs 173. However, suitable CdTe/Polymer systems have not yet been found, and reported PCEs based on CdTe/MEH-PPV are quite below 0.1% 174. A systematic investigation on hybrid solar cells based on MEH-PPV blended with CdSe x Te 1-x tetropods demonstrated a steady PCE decrease from 1.1% starting from CdSe to 0.003% with CdTe 130. The reason of the dramatical decrease in efficiency could be attributed to the possibility that energy transfer rather than charge transfer could occur from the polymer to CdTe NCs in CdTe/Polymer blends, resulting in an insufficient generation of free charge carriers 130, 175. Further lowering of the NC band gap could be achieved by using semiconductors such as PbS or PbSe. Zhang et al. have reported hybrid solar cells based on blends of MEH-PPV and PbS NCs, but unfortunately the PCEs were very low (0.001%) 176. Watt et al. have developed a novel surfactant-free synthetic route where PbS NCs were synthesized in situ within a MEH-PPV film 177, 178. In general, using low band gap NCs as electron acceptor in polymer/ncs system has been not successful yet, because energy transfer from polymer to low band gap NCs is the most likely outcome, resulting in inefficient exciton dissociation. Recently it has been demonstrated that Si NCs are promising acceptor material for hybrid solar cells due to the abundance of Si compounds, non-toxicity, and strong UV absorption. Hybrid solar cells based on blends of Si NCs and P3HT with a PCE above 1% have been reported 179. Si NCs were synthesized by radio frequency plasma via dissociation of silane, and the size can be tuned between 2 and 20 nm by changing chamber pressure, precursor flow rate, and radio frequency power. Devices made out of 35 wt% Si NCs, 3-5 nm in size, exhibited a PCE of 1.15% under AM1.5 G illumination which is a promising first result. The J-V characteristic is demonstrated in Fig

52 Chapter 2. Bulk-heterojunction hybrid solar cells Figure 2.11 J-V characteristics of hybrid solar cells based on 35 wt% Si NCs/P3HT (main window) and P3HT-only (inset) (Image taken from Ref. 179 ). The selected performance parameters of hybrid solar cells based on NCs (except CdSe) and conjugated polymers are summarized in Table

53 Table 2.2. Selected performance parameters of hybrid solar cells based on NCs (except CdSe) and conjugated polymer. NCs Polymer d(nm) Aspect ratio NC wt% Cathode P(mW/cm2) Voc(V) Jsc(mA/cm 2 ) FF PCE(%) Ref CdS MEH-PPV Al CdTe MEH-PPV LiF/Al PbS MEH-PPV Al PbSe P3HT Al Si P3HT Al Si P3HT Al SnO2 MDMO-PPV Au TiO2 P3HT Al TiO2 P3HT Al ZnO P3HT 50 Al ZnO MDMO-PPV Al ZnO P3HT 15 a LiF/Sm/Al ZnO MDMO-PPV 15 a Al ZnO P3HT 1 65 Al ZnO APFO-3 67 Al a volume percent 43

54 Chapter 3. Synthesis of CdSe QDs Chapter 3 Synthesis of CdSe QDs (The main content of this chapter will be summarized in a manuscript which is under preparation) CdSe NCs are one of the most widely investigated II-VI semiconductor NC materials. Due to their good electron acceptor properties in combination with conjugated polymer such as P3HT and their broad absorption at useful range for solar energy conversion, CdSe NCs were the first to be used with conjugated polymers in solar cells application 127. Usually colloidal semiconductor NCs are composed out of an inorganic core with a diameter usually ranging between 2 to 10 nm, and an organic outer layer of surfactant molecules (i.e. ligands) which are important for the colloidal stabilization and for maintaining their optical properties 190, 191. They are synthesized from precursor compounds which are dissolved in solutions. The synthesis of colloidal NCs is based on three components: precursors, surfactants (i.e. ligands), and solvents. Fig. 3.1b shows a typical hot-injection synthesis approach for colloidal NCs introduced by La Mer and Dinegar 192. When the precursor materials are injected quickly into a hot solvent, the precursors decompose to form monomers, which are the basic growth materials for the NCs. Once the monomers reach a high enough super saturation level, the growth of NC starts with a nucleation process. The temperature during the synthesis process is very critical. It must be high enough for rearrangement and annealing of atoms while being low enough to promote controlled crystal growth. In addition, the monomer concentration also plays an important role. Generally, there exist two different regimes during the synthesis process: the focusing and the defocusing regime. When the monomer concentration is high enough, NCs growth is preceded by the addition of monomer from solution to the NCs nuclei. Meanwhile the concentration of monomers is below the critical 44

55 3.1 CdSe QD synthesis concentration for nucleation, thus these species only add to the growth of existing NCs rather than forming new nuclei 146. In the focusing regime, the radius of the smaller particles increase faster than large ones because of the larger volume/surface ratio, resulting in focusing of the size distribution and forming monodisperse NCs. When the monomer concentration is depleted, defocusing of the size distribution takes place as a result of Ostwald ripening 13, 193, where the sacrificial dissolution of smaller NCs results in growth of fewer larger NCs 158. Figure 3.1 (a) Schematic illustration of the stages of nucleation and growth during the formation of monodisperse colloidal particles. (b) Representation of the synthesis apparatus using for the NC preparation. (Image taken from ref. 35 ) The initial synthesis milestone for highly crystalline CdSe NCs with narrow size distribution has been reported by Murray et al Trioctylphosphine (TOP) and TOPO were used as coordinating solvent. Dimethylcadmium ([Me] 2 Cd) was used as Cd precursor. Bis(trimethylsilyl)-selenium ([TMS] 2 Se) was used as Se precursor. In 2000, Peng et al. 128 demonstrated that the shape of the NCs can be controlled by carefully choosing the reaction conditions, reactant concentrations, and the ligands system. For instance, the growth of nanorods with one dimensional elongated structure relies on the anisotropic crystal structure of the wurtzite form of CdSe, which has a unique c axis. The growth rate can be controlled by adding alkylphosphonic acids which bind more strongly to cadmium than TOPO does. Later on, Peng et al. 173 developed a synthesis method which has been widely adopted as it 45

56 Chapter 3. Synthesis of CdSe QDs removes the need of using highly toxic and pyrophoric dimethyl cadmium. In this method, CdO is heated together with TOPO and a phosphoric acid ligand, e.g. tetradecylphosphoric acid (TDPA), to form a Cd-TDPA complex. Then the Cd-TDPA precursor is heated to around 300 o C followed by an injection of Se-TOP (trioctylphosphine) precursor, resulting in NC nucleation. In the same year Talapin et al. reported on an enormous PL enhancement of CdSe QDs by introducing HDA as coordinating agent 195. In this study, high quality spherical CdSe QDs were synthesized by using Cd(CH 3 (CH 2 ) 16 COO) 2 (Cd-stearat) as the Cd- precursor, Se-TOP as the Se- precursor, and HDA/TOPO acting as ligands as well as solvent matrix. Details about the optimization of this synthesis method are described in a recent publication CdSe QD synthesis The setup for the CdSe QD synthesis includes a 25 ml three-neck flask, a glass stopper of slot shape, a small spin-bar, a nitrogen floating system, a temperature sensor, a hotplate, a metal bath, and a magnetic stirring apparatus. Fig. 3.2 shows a picture of the synthesis setup. Figure 3.2 A picture of the synthesis setup 46

57 3.1 CdSe QD synthesis Cd-stearat was prepared from 32.1 mg CdO (0.25 mmol) and mg stearic acid (0.875 mmol) with a catalytic amount of succinic acid at 200 o C under nitrogen atmosphere until a clear colorless solution appeared. Elemental Se dissolved in TOP (97% from ABCR, Germany) at a concentration of 1M was used as Se-TOP precursor. Briefly, 222 mg (0.2 mmol) Cd-stearat, g (12 mmol) HDA (98% from Molekula) and g (8 mmol) TOPO (99% from Aldrich) were loaded into the reaction flask and heated up to 300 C, then 0.2 ml of a 1 M solution of Se in TOP was rapidly injected. The solution was then maintained at this temperature to allow the growth of QDs. 3.2 Characterization of CdSe QDs The absorption and emission spectra of CdSe QDs were recorded by a J&M TIDAS diode-array spectrometer (Spectralytics, Aalen, Germany). A halogen/deuterium combined light source was used for the absorption spectroscopy, and a J&M FL3095 monochromator was used for the emission spectroscopy, where the excitation wavelength range is between 318 nm and 595 nm. The samples taken from reaction solution were dissolved in chloroform and the absorbance of first excitonic peak was adjusted to 0.5 for equal comparison under the same condition. Transmission electron microscopy (TEM) measurements in the transmission mode were performed using a Zeiss (LEO) 912 Omega transmission electron microscope with an acceleration voltage of 120 kv and zero loss filtering. For TEM measurement, the samples were purified in order to remove the excess ligands which disturb the measurement. A methanol washing procedure was applied by dissolving 1 ml of the synthesis product in 3 ml chloroform, and then 6 ml methanol was added afterwards. The excess of ligands was removed after centrifugation at 14.5 krpm. The purified samples were prepared on carbon coated copper grids before investigation. 47

58 Chapter 3. Synthesis of CdSe QDs Figure 3.3 (a) TEM image (average diameter of about 5.5nm) and (b) absorption and PL spectrum of CdSe QDs after 120 min reaction time, which are typical for the NC material used for hybrid solar cells later on. Fig. 3.3a shows a TEM image and Fig. 3.3b an absorption spectrum as well as a photoluminescence (PL) spectrum of CdSe QDs after 120 min reaction time under optimized synthesis conditions. It represents the typical NC material for the fabrication of hybrid solar cells which will be described in chapter 4. TEM investigation reveals a uniform size distribution of round shaped NCs with an average diameter of around 5.5 nm. The first absorption peak occurs at 607 nm with a corresponding PL emission at 622 nm. The spectrum peak shift between absorption and PL is known as Stokes shift, and the value of 15 nm shift is consistent with the data from the literature 197. In addition, the existence of an absorption fine structure and the narrow full width at half maximum (FWHM) of 29 nm revealed that the QDs had a very narrow size distribution and a uniform size, which was also observed by TEM measurement. Fig. 3.4 shows the normalized absorption and PL spectra of CdSe QDs after different 48

59 3.2 Characterization of CdSe QD reaction times. As mentioned above, due to the quantum confinement effect, the band gap of NCs decreases along with the growth of diameter. Thus the red shift of absorption and PL along the synthesis reaction time can be observed. Figure 3.4 Normalized (a) absorption and (b) PL spectra of CdSe QDs after different reaction times. Both the absorption and PL spectra are red shifted indication the size growth of QDs. Fig. 3.5 shows the time dependence evolution of the PL peak and intensity of CdSe QDs. The size development of QDs shows an exponential like behavior. The initial growth rate in the first 2 min is very high, and it reaches a threshold after 5 min. Afterwards no significant increase in size was observed, which could be attributed to the depletion of precursors. The PL intensity reached the maximum value within the first one minute, and then rapidly dropped to 75% of its maximum value at two min. Along with the reaction time, PL intensity recovered steadily and is almost stable after 60 min. 49

60 Chapter 3. Synthesis of CdSe QDs Figure 3.5 Time dependence of the PL emission wavelength (a) and PL intensity (b) of CdSe QDs. Inset: picture of time aliquot QD samples taken at different reaction times illuminated under 312 nm UV irradiation. 3.3 Critical parameters for high quality CdSe QDs synthesis In Ref. 196, some critical parameters for CdSe QDs synthesis have been extensively investigated by Yuan et al., including the reaction temperature, Se-TOP injection speed, stirring velocity, and the reagent grade of the chemicals. Based on the established synthesis approach, the procedures for CdSe QDs synthesis were further optimized especially in the aspect of PL intensity by investigating two parameters: HDA pre-heating and Se-TOP aging HDA pre-heating In order to remove the oxygen and water residuals in the relatively low grade (98%) HDA before mixing with Cd-stearat and TOPO, HDA was pre-heated in the flask separately under 300 o C in a protective nitrogen atmosphere. Fig. 3.6a shows the PL peak wavelength evolution as a function of time of reactions with different HDA 50

61 3.3 Critical parameters for high quality CdSe QD synthesis pre-heating times. CdSe QDs synthesized from non-pre-heated HDA reach a PL peak plateau at a wavelength about 660 nm showing a large red shift compared to QDs synthesized after 30 min pre-heating of HDA (630 nm), implying that the diameter of QDs from pre-heated HDA is smaller than that of the non-pre-heated one. The reduced diameter of QDs could be explained by a larger number of nuclei formed in the initial nucleation period with the consequence that a lower amount of monomers are available during the growth stage. Figure 3.6 Time dependence of (a) PL emission wavelength and (b) PL intensity of CdSe QDs synthesized by using pre-heated HDA at 300 o C for different pre-heating times. Moreover, as shown in Fig. 3.6b, a remarkable enhancement in PL intensity over the whole reaction period was achieved after pre-heating HDA for 30 min (red) and 60 min (blue) in comparison with non-pre-heated one (black). For example, for the samples after 60 min reaction time, the PL intensities increase 2.6 times for 30 min pre-heated HDA and 2.1 times for 60 min pre-heated HDA in comparison with non-pre-heated samples, respectively. The enhanced PL intensity indicates a higher crystalline quality and less surface defect sites of the synthesized QDs. 51

62 Chapter 3. Synthesis of CdSe QDs Se-TOP Aging The Se-TOP precursor aging conditions have been investigated by allowing fresh Se-TOP solution to be exposed to air for different times at temperature and room humidity (experiments were performed in May 2010 in Freiburg, relative humidity at 60-80%). Figure 3.7 Time dependence of (a) PL emission wavelength and (b) PL intensity of CdSe QDs synthesized by using different time-aged Se-TOP as precursor. 30 min pre-heated HDA was used for the following synthesis experiments. Fig. 3.7 shows the time dependence of the PL peak wavelength and intensity of CdSe QDs synthesized by using different time aged Se-TOP precursor. Similar to the effect of HDA pre-heating, QDs synthesized from fresh Se-TOP exhibit lower PL emission peak wavelengths and smaller sizes. To get the highest PL intensities the optimized aging time for Se-TOP was found between 49 and 78 hours. For example, QDs synthesized from 49 hour aged Se-TOP show a more than three times higher PL intensity over the ones prepared with freshly made Se-TOP. In addition, the PL intensities are almost the same for QD aliquot samples taken between 60 min and 120 min, while the fresh Se-TOP samples show a rapid decrease of the PL intensity after 60 min of the reaction. 52

63 3.3 Critical parameters for high quality CdSe QD synthesis Figure 3.8 (a) PL peak emission wavelength and (b) PL intensity of CdSe QDs samples taken after 120 min of the reaction by using different time aged Se-TOP as precursor. The comparison of samples taken after 120 min of the synthesis with different aging times of the Se-TOP precursor is shown in Fig It is observed, that the tendency to get to higher PL intensities is corresponding to lower PL emission wavelengths. For example, optimized QDs synthesized from 78h aged Se-TOP show a large blue shift of the PL emission peak from 655 nm to 626 nm. In addition, the PL intensity is significantly increased by using aged Se-TOP, which is 10 times higher than that of the fresh Se-TOP. The optimized synthesis approach is summarized as following: 1) g (12 mmol) HDA (98% from Molekula, Germany) was loaded into the reaction flask and heated up by using a metal bath (Wood s metal from Fisher scientific, UK, containing Bi 50%, Pb 26.7%, Sn 13.3%, Cd 10% by weight, melting point: 70 o C) on a hotplate. In order to remove water and oxygen, the flask was evacuated three times when the temperature was around 100 C. 2) HDA was pre-heated at 300 C for 30 min in the flask under nitrogen 53

64 Chapter 3. Synthesis of CdSe QDs atmosphere. Once the HDA solution was cooled down to 120 C, 222 mg (0.2 mmol) Cd-stearat, and g (8 mmol) TOPO were melted by a heating gun and injected into the flask to be mixed with HDA. The flask was evacuated at again 100 o C and afterwards heated up to 300 C. 3) The flask temperature was hold at 300 C for 5 min. Then 0.2 ml of 78h air-aged 1M Se-TOP was rapidly injected. 4) The solution was maintained at this temperature and aliquots were taken at different time for absorption and emission spectroscopy measurements. If not mentioned, QDs with 2 hours reaction time are used for solar cells application. In summary, highly reproducible synthesis methods for CdSe QDs have been developed, leading to a narrow size distribution and excellent photophysical properties. Pre-heating of HDA ligands and aging of the Se-TOP precursor have been proven as two critical parameters for synthesizing high quality QDs with high PL quantum yields and less surface defect sites. De-adsorption of CO 2 from HDA during the pre-heating process 198, and the reaction between Se-TOP and water/oxygen in air during aging process are proposed to explain the improved quality of synthesized QDs. The detail mechanism of these two critical parameters is still under investigation. The synthesis of reproducible and high quality QDs is the basis of fabricating efficient hybrid solar cells. The influence of the QD characteristics such as diameter, PL peak wavelength, and PL intensity on the performance of hybrid solar cells is discussed in detail in chapter 4. 54

65 4.1 Hexanoic acid treatment Chapter 4 Surface modification of CdSe QDs (The main content of this chapter was published in Applied Physics Letters 96, (2010) entitled Improved efficiency of hybrid solar cells based on non-ligand-exchanged CdSe quantum dots and poly(3-hexylthiopehen).) CdSe QDs directly after synthesis are not suitable to be incorporated into conjugated polymer for fabricating hybrid solar cells. Because the ligands used for preventing aggregation during the growth of QDs usually hinder the electron transfer between QDs and polymers and the electron transport between neighboured QDs. Therefore, post-synthetic surface modifications on the QDs are desirable. For example, pyridine treatment of the NCs is the most commonly used and effective procedure for improving device performance. However, NCs after ligand exchange with small molecules tend to aggregate and precipitate out from the organic solvent once the long alky chain ligands are replaced 110, 150, resulting in difficulties to obtain stable mixtures of NCs and polymer. In this study, we demonstrate a novel post-synthetic treatment on spherical CdSe QDs with non-ligand-exchange approach, where the NCs were treated by a simple and fast hexanoic acid-assisted washing procedure. Devices with optimized ratios of QDs to P3HT exhibited reproducible PCEs up to 2.1% after spectral mismatch correction which could meanwhile be exceeded. 4.1 Hexanoic acid treatment Before being incorporated into polymer to form hybrid blends, CdSe QDs synthesized in our lab with a diameter of around 5.5 nm and HDA capping ligand (the synthesis of QDs is described in detail in Chapter 3) were treated by a simple and fast hexanoic 55

66 Chapter 4. Surface modification of CdSe QDs acid-assisted washing procedure. Typically, 0.6 ml as-synthesized CdSe QDs were added to 6 ml hexanoic acid (>99.5% from Aldrich) at 105 C and stirred for 10 min. 12 ml anhydrous methanol (99.8% from Aldrich) was added afterwards and stirred for 8 min to precipitate the QDs, which were recovered by centrifugation. Then 1.5 ml chloroform (>99% from Aldrich) and 4.5 ml methanol were added at 90 C to precipitate the QDs again. After being recovered by centrifugation, QDs were finally dispersed into anhydrous DCB (99% from Aldrich) at a concentration of about 15 mg/ml, showing a very good solubility without adding any additive. Photovoltaic devices were fabricated according the structure shown in Fig.4.1. Figure 4.1 Schematic structure of a typical fabricated solar cell, chemical structure of P3HT, TEM image of HDA-capped CdSe QDs, and photograph of a typical device fabricated in our lab. PEDOT:PSS dispersions (Baytron AI4083, H.C. Starck) were spin-coated onto the UV-ozone treated indium tin oxide (ITO) glass substrates to form a film thickness of about 70 nm. Typically, the solution of QDs (15 mg/ml) and 15 mg/ml P3HT (regioregularity > 98.5%, ~ 50,000 MW from Aldrich) in anhydrous DCB were mixed at various ratios, and spin-coated at 500 rpm for 2 min onto the surface of the PEDOT:PSS layer. The typical P3HT:CdSe film thickness was nm measured by a profilometer Dektak Aluminum cathodes with thicknesses of 100 nm were deposited by thermal evaporation (Edwards Auto 306 coating system) with an active area of 0.08 cm 2. The device fabrication is described in detail in Appendix A1. Subsequently devices were annealed at 145 C for 30 min in a nitrogen floated 56

67 4.1 Hexanoic acid treatment glove-box and cooled down to room temperature before measurement. Current density-voltage (J-V) measurements were performed in the glove-box under AM1.5G illumination using a solar simulator (LOT-Oriel sun simulator system) at an intensity of 100 mw/cm 2, which had been calibrated by a certified photodiode (Thorlab S121B). 4.2 P3HT:CdSe composites As shown in Fig. 4.2, the UV-vis absorption spectra of CdSe QDs in chloroform solution before and after applying the acid washing procedure were similar, and were not affected by the washing procedure, exhibiting the first absorption peak at 606 nm. In contrast, strong PL quenching was observed in the case of the hexanoic acid washed QDs which could be attributed to the removal of excess ligands accumulated around the QDs, resulting in the increase of non-radiative decay processes through surface defects and/or energy traps formed by the surrounding solvent 199. NMR studies of H 1, C 13 and P 31 isotopes revealed that the synthesis ligand HDA still remained on the surface, and no ligand exchange was identified. Figure 4.2 Absorption and photoluminescence (PL) spectra of CdSe QDs before (dark, solid line) and after (red, dashed line) hexanoic acid treatment. Fig. 4.3 shows the energy level diagram of the P3HT:CdSe composite system. In such system where the bandgap of polymer P3HT is similar to CdSe QDs, the excitons are mainly formed in the polymer donor material, when they reach the D/A interface, instead of direct electron transfer from donor to acceptor (Fig. 4.3a), energy transfer, 57

68 Chapter 4. Surface modification of CdSe QDs i.e. Förster resonance energy transfer (FRET) could also occur (Fig.4.3b) 200. This energy transfer process leads to the formation of excitons in the acceptor QDs, followed by hole transfer from QDs to polymer. The final state is the same as for the direct electron transfer process, however, the rates involved in the energy transfer as well as hole transfer processes could be significantly different 105. In addition, some part of lower energy excitons are formed in the NCs as shown in Fig.4.3c, then free charge carriers are generated by hole transfer from the HOMO of NCs to the HOMO of the polymer. Figure 4.3. Schematic routes for exciton dissociation and charge transfer in a P3HT:CdSe composite system. (a) Exciton formation in P3HT, followed by electron transfer to CdSe. (b) Exciton formation in P3HT, followed by energy transfer to CdSe, then by hole transfer to P3HT. (c) Exciton formation in CdSe, followed by hole transfer to P3HT. Fig. 4.4 shows the absorption and PL spectrum of P3HT:CdSe composites with various CdSe QDs loadings in chloroform solution. Spectra deriving from pure polymer are also shown for comparison. Along with the increase CdSe loading in the composites, the feature of QDs exciton absorption peak at 609 nm is more and more pronouced (Fig. 4.4a). Meanwhile, the intensity of the PL emission from P3HT at 577 nm is decreasing. For example, the intensity of PL from P3HT is only 6.6% of its initial intensity when 87.5 wt% CdSe QDs was blended in the composite. The PL 58

69 4.2 P3HT:CdSe composites quenching phenomenon indicates that more excited electrons decayed via nonradiactive transitions, which could be attributed to charge transfer and/or energy transfer between P3HT and CdSe QDs. Figure 4.4. (a) Absorption and (b) PL spectra of P3HT:CdSe composites with various CdSe loadings. (c) PL and absorption intensity dependence as a function of CdSe loading in the composites. Fig. 4.5 shows the TEM image and AFM (non contact mode, surface morphology) image of a P3HT:CdSe composite film with 87 wt% CdSe loading. The TEM sample was prepared by immersing a device in water in order to dissolve the PEDOT:PSS layer under the P3HT:CdSe film. Afterwards the P3HT:CdSe film was collected by a TEM grid from the water surface. For the TEM image (Fig. 4.5a), the black areas represent the QD domain, and the white areas represent the P3HT domain. The P3HT and CdSe domains are relatively well-distributed, showing a phase separation n the scale of nm. The AFM measurement reveals that although the surface of the composite film is quite rough in the range of 50 nm, the distribution of peaks and valleys is still homogenous throughout the whole scanning area. The rougher but still homogeneous surface could be also beneficial for the electrical contact to the Al cathode due to the increased interfacial contact area. 59

70 Chapter 4. Surface modification of CdSe QDs Figure 4.5. (a) TEM image and (b) AFM surface morphology image of a P3HT:CdSe composite film with 87wt% CdSe loading. 4.3 Solar cells performance Fig. 4.6 shows the variations of values for V oc, J sc, FF, and PCE under illumination as a function of increasing CdSe QD wt% loading. The average values (symbols) and standard deviations (error bars) are derived from three individual devices located on the same substrate (Fig. 4.1). One can clearly observe that the PCEs of devices are very sensitive to the QD loading concentration in the blend. An optimum QD loading was found in the range of 84-87wt% (Fig. 4.4d), which is in agreement with results reported in literature 110, 127, leading to reproducible devices with average PCEs exceeding 2.0%. Lower QD loading below this optimum range leads to a significant decrease in PCE, because the amount of the QDs seems not to be sufficient to form efficient percolation pathways minimizing hopping distances between the QDs for extracting electrons out of the device. While in the case of higher QD loading, the polymer part is too less, leading to inefficient photon absorption and hole transport, which result in inferior device performance. 60

71 Current density (ma/cm 2 ) 4.3 Solar cells performance Figure 4.6 Variation of values for (a) V oc, (b) J sc, (c) FF, and (d) PCE based on different devices fabricated in our lab with various CdSe QD wt% loading concentrations under AM1.5G 100 mw/cm2 illumination. The average values (symbols) as well as standard deviations (error bars) are shown. Devices were fabricated and measured in our lab without spectral mismatch correction V oc =0.58 V J sc =7.34 ma/cm 2 FF=0.54 =2.28% Voltage (V) Figure 4.7 J-V characteristic of the best device fabricated in our lab from a blend containing 87 wt% QDs, measured in the dark (dashed line) and under AM1.5G 100 mw/cm 2 illumination (solid line). Devices were fabricated and measured in our lab without spectral mismatch correction. 61

72 Chapter 4. Surface modification of CdSe QDs Fig. 4.7 shows the J-V characteristic of the best device fabricated from a blend of 87 wt% CdSe QDs and P3HT. The measurements were performed in the dark (dashed line) and under illumination (solid line). An open-circuit voltage (V oc ), of 0.58 V, a short-circuit current density (J sc ) of 7.3 ma/cm 2, a fill factor (FF) of 0.54, and a PCE of 2.28% was measured and calculated. Figure 4.8 Variation of values for (a) V oc, (b) J sc, (c) FF, and (d) PCE based on different devices with various CdSe QD wt% loading concentrations under AM1.5G 100 mw/cm2 illumination. The average values (symbols) as well as standard deviations (error bars) are shown. Devices were fabricated and measured at ISE after spectral mismatch correction. Furthermore, we performed a completely new measurement of freshly prepared photovoltaic devices together with the Fraunhofer Institute of Solar Energy Systems (ISE) under more accurate characterization conditions including spectral mismatch corrections. In this set of experiment, patterned ITO glass substrates after UV-ozone treatment were supplied by the ISE. The PEDOT:PSS and P3HT:CdSe blend were spin-coated onto the substrates consequently. Afterwards, they were transfer from the glove-box in our lab to the ISE lab. Aluminum cathodes with thicknesses of nm were then deposited by thermal evaporation with an active area of 0.08 cm 2. An image of a typical photovoltaic device is shown in Fig. 4.5 (inset). Subsequently devices 62

73 4.3 Solar cells performance were annealed at 145 C for 10 min followed by 160 C for 10 min in a glove-box and cooled down before measurement. J-V measurements were performed in a glove-box. All the PCE data measured at ISE are corrected for spectral mismatch. Similar to Fig. 4.6, variations of values for Voc, Jsc, FF, and PCE based on different devices fabricated at ISE as a function of CdSe QD wt% loading concentration is shown in Fig The average values (squares), standard deviations (error bars), maximum values (triangles) and minimum values (dots) are derived from six individual devices located on the same substrate (Fig. 4.8 inset). An optimum QD loading was found in the range of 85-89wt% (Fig. 4.9d), leading to devices with average PCEs approaching 2.0%. Table 4.1 summarizes the performance of devices with different CdSe loading under illumination. Table 4.1 Summarized performance parameters of devices with different CdSe QDs loadings. CdSe Loading V oc (mv) J sc (ma/cm 2 ) FF (%) PCE (%) 85 wt% 624.0± ± ± ± wt% 625.2± ± ± ± wt% 598.6± ± ± ±0.04 Fig. 4.9 shows the J-V characteristic of the best device fabricated from a blend of 87 wt% CdSe QDs and P3HT obtained under AM1.5G 100 mw/cm 2 illumination. An open-circuit voltage (Voc) of 623 mv, a short-circuit current density (Jsc) of 5.8 ma/cm 2, a fill factor (FF) of 0.56 and a PCE of 2.0% was measured and calculated. This was the highest reported value for a CdSe QDs based hybrid solar cell after spectral mismatch correction at that time. 63

74 Current density (ma/cm 2 ) Chapter 4. Surface modification of CdSe QDs V oc = 623 mv J sc = 5.8 ma/cm 2 FF = 0.56 Eff = 2.0% Voltage (mv) Figure 4.9 J-V characteristic of the best cell fabricated from a blend containing 87 wt% QDs under AM1.5G 100 mw/cm 2 illumination. Inset image: photograph of a typical device fabricated at Fraunhofer ISE. Devices were fabricated and measured at ISE after spectral mismatch correction. Additionally, the influence of different cathode materials on the device performance has been investigated. Although Ca/Al 201 cathodes are widely used for polymer/pcbm photovoltaic devices exhibiting better performances than Al cathode devices, our results showed that devices with an Al cathode exhibited higher PCEs in such polymer/cdse system as shown in Fig Figure 4.10 J-V characteristics of devices containing 87wt% CdSe QDs with different cathode materials. Devices were fabricated and measured at ISE after spectral mismatch correction. Typical devices containing 87wt% CdSe QDs with Al cathode exhibited PCE of 2.0% with V oc = 625 mv, J sc = 5.7 ma/cm 2, and FF=0.55. However, devices with Ca/Ag cathode exhibited PCE of 0.42% with V oc = 339 mv, J sc = 2.8 ma/cm 2, and FF=

75 External Quantum Efficiency (%) 4.3 Solar cells performance The reason for the inferior device performance with Ca/Ag cathode is still not clear, it might be explained by the contact incompatibility between the CdSe phase and the Ca cathode. Table 4.2 summarizes the performance of devices with different cathode materials. Table 4.2 Summarized performance parameters of devices containing 87wt% CdSe QDs with different cathode materials. Cathode V oc (mv) J sc (ma/cm 2 ) FF (%) PCE (%) Al 625.2± ± ± ±0.06 Ca/Al 339.0± ± ± ±0.06 Fig shows the wavelength dependent short-circuit External Quantum Efficiency (EQE) measurement for an optimized device containing 87wt% CdSe QDs, exhibiting a maximum EQE of 45% under 0.72 mw/cm 2 at 450 nm, which is in agreement with results reported in literature 110. Due to the overlapping absorption of QDs (absorption edge at 650 nm) and P3HT (ca nm), the additional photocurrent contribution from QDs at wavelengths >660 nm, where P3HT is transparent to the incident radiation 132, 136, 139, 144, cannot be clearly distinguished in our study Wavelength (nm) Figure 4.11 Wavelength dependent short-circuit External Quantum Efficiency (EQE) spectrum for a typical device with 87 wt% CdSe QDs in a P3HT/QDs blend. 65

76 Chapter 4. Surface modification of CdSe QDs 4.4 Ligand sphere model TEM measurements in the transmission mode were performed using a Zeiss (LEO) 912 Omega transmission electron microscope with an acceleration voltage of 120 kv and zero loss filtering 4.12a for unpurified as-synthesized product. Samples were prepared on carbon coated copper grids and measured without any further treatment. For the electron loss imaging of the unpurified synthesis product an acceleration voltage of 52 ev was used (Fig. 4.12b). In this mode white areas show the particles and the organic compounds, while black areas represent the background. Fig. 4.12c is an overlaid image of Fig. 4.12a and 4.12b for the direct visualization of the organic compounds (HDA ligand) around the QDs. Figure 4.12 TEM images of the QDs directly taken from the synthesis. (a) Transmission mode. (b) Electron Loss Imaging mode. (c) Overlaid image of image (a) and image (b), where the organic compounds around the QDs are visible. In Fig TEM images of regular purified QDs and acid treated QDs are shown. For a regular purification of QDs, a methanol washing procedure was performed once by dissolving 1 ml of the synthesis product in 3 ml chloroform, then 6 ml methanol was added afterwards. The excess of ligands was removed after centrifugation. TEM investigations confirmed that the average particle to particle distance is reduced significantly after acid treatment of the QDs compared to untreated QDs. Without acid treatment, the particles are usually well separated as shown in Fig. 4.13a. While after acid treatment (Fig. 4.13b), the QD aggregation tendency is clearly visible, indicating the potential for the formation of efficient percolation pathways in the photoactive film which results in the improvement of the PCEs. 66

77 4.5 Influence of the QDs synthesis on the device performance Figure 4.13 (a) TEM image of CdSe QDs after applying a methanol washing procedure for one time without applying any acid treatment, and (b) with applying the acid washing procedure. Additionally, QD size distribution was investigated by dynamic light scattering (DLS) measurements (Zetasizer Nano Series ZS, Malvern) before and after acid treatment (Fig. 4.14). The effective particle size was reduced significantly by more than one order of magnitude after acid treatment, along with the improvement in the PCE from 0.01% to about 2%. The hydrodynamic diameter of around 200 nm for the untreated dots is indeed very large and the occurrence of agglomerates cannot be excluded. Hydrophobic interaction caused by additionally adsorbed ligand molecules, leading to aggregation in solution, is very likely. Nevertheless DLS shows the right tendency of the overall diminishing particle size and supports the proposed sphere model. The hydrodynamic diameter of 10 nm for the acid treated CdSe QDs is a more realistic value and in accordance with results from TEM and UV-vis investigations. Figure 4.14 QD size distributions before and after acid treatment measured by dynamic light scattering (DLS), correlated together with the PCEs of the devices based on the respective QDs. Based on the PL quenching, the TEM investigation and the DLS results a simple reduced ligand sphere model is proposed to explain the possible reason for improved 67

78 Chapter 4. Surface modification of CdSe QDs photovoltaic device efficiencies after acid treatment (Fig. 4.15). Immobilized HDA ligand spheres around the particles might be responsible for a poor device performance of non acid treated samples. By the assistance of hexanoic acid this insulating ligand sphere is effectively removed due to the salt formation of HDA. This organic salt is also much more easily dissolved in the supernatant solution than unprotonated HDA and can be separated easily from the QDs by subsequent centrifugation. The reduced insulating sphere barriers would allow for both, an improved charge transfer between P3HT and QDs, as well as an improved electron transport between QDs. Another advantage by avoiding the exchange of the synthesis capping ligands is that the QDs retain a good solubility in organic solvent (i.e. DCB) after acid treatment, allowing a high loading of the CdSe QDs in the blend resulting in an efficient percolation network formation during the annealing of the photoactive composite film. Due to the fact that the PCEs are sensitive to the ratio of QDs to P3HT, a good solubility of QDs in the polymer solution is required for the formation of high quality photoactive thin-films. Figure 4.15 Schematic illustration of the proposed QD sphere model: an outer insulating HDA ligand sphere is supposed to be responsible for the insulating organic layer and is effectively reduced by acid treatment. In addition, different acid chain lengths such as butanoic acid (referred as C4) and octanoic acid (referred as C8) have been applied for post-synthetic treatment in comparison with hexanoic acid (referred as C6). Fig shows the performance of devices based on different acid treated QDs. With C4 and C6 treated QDs, devices show a similar J-V characteristics with PCEs around 2.2%. However, the PCE of a device with C8 treated QDs is very low. This could be explained by using the ligand sphere model that acids with a chain length below a certain value (e.g. 6 carbon atoms 68

79 4.5 Influence of the QDs synthesis on the device performance in this case) could penetrate effectively into the HDA ligand shell and reduce it. Therefore the longer acid chain lengths show a poorer effect on the ligand sphere reduction than the shorter ones. Table 4.3 summarizes the performances of device with different acid treated QDs. Figure 4.16 J-V characteristics of device with different acid treated QDs. Devices were fabricated and measured in our lab without spectral mismatch correction. Table 4.3 Summarized performances of devices based on different acid treated QDs. Treatment V oc (V) J sc (ma/cm 2 ) FF PCE (%) C4 0.61± ± ± ±0.04 C6 0.58± ± ± ±0.05 C8 0.47± ± ± ± Influence of the QDs synthesis on the device performance As mentioned in chapter 3, the quality of CdSe QDs has crucial impact on hybrid solar cells performance. PL intensity is one of the important factors representing the quality of QDs, since a higher PL intensity indicates less surface defect sites of the synthesized QDs. Fig. 4.17a shows the absorption and PL spectra of QDs synthesized from freshly prepared Se-TOP and non-pre-heated HDA (black, referred as bad CdSe ), and from 2 days air-aged Se-TOP with pre-heated HDA (red, referred as good CdSe ). 69

80 Chapter 4. Surface modification of CdSe QDs Figure 4.17 (a) Absorption and PL spectra of QDs synthesized from freshly prepared Se-TOP and non-pre-heated HDA (black, referred as bad CdSe ), and from 2 days air-aged Se-TOP with pre-heated HDA (red, referred as good CdSe ). (b) J-V characteristics of devices fabricated with acid-treated bad CdSe (black) and good CdSe (red). Devices were fabricated and measured in our lab without spectral mismatch correction. The bad CdSe QDs exhibit an absorption peak at 640 nm and PL intensity of about 7000 counts. While the good CdSe QDs show an absorption peak at relatively shorter wavelength of 614 nm, indicating that the diameter of the good QDs is smaller than that of the bad one. Nevertheless, the PL intensity of the good QDs is about counts, which is almost double than that of the bad CdSe QDs (relative comparison, both PL measurements have been performed at similar time). Fig. 4.17b shows the typical J-V characteristics of devices fabricated with 87wt% acid-treated bad and good CdSe QDs with P3HT as polymer. It is worth to mention that although the PCEs shown in this section were not corrected for spectral mismatch, conclusions from these results are still valid for relative comparison. Devices based on good CdSe QDs show a much better PCE of up to 2.1% with V oc = 0.61 V, J sc = 7.23 ma/cm 2, and FF=0.48 than devices based on bad CdSe QDs with PCE= 0.9%, V oc = 0.58 V, J sc = 3.92 ma/cm 2, and FF=0.40. Table 4.4 summarizes the performance of devices with different synthesized CdSe QDs. Both devices show similar V oc, indicating that the energy levels of both kinds of QDs are comparable. The improvement in PCE is mainly attributed to the significantly increased J sc and FF. This result implies that high quality of CdSe QDs with less surface defect sites are required for hybrid solar cells, since surface defects might act as traps when electrons are transported between neighbored QDs, resulting in lower electron mobility in a photovoltaic device. In addition, surface defect on QDs can facilitate the nonradiative recombination processes of excitons during charge transfer from polymer to QDs, 70

81 4.5 Influence of the QDs synthesis on the device performance leading to inferior charge transfer efficiency and finally results in lower device efficiency. Table 4.4 Summarized performance parameters of devices with acid-treated bad CdSe and good CdSe QDs. CdSe synthesis PL peak wavelength (nm) PL intensity (a.u.) V oc (V) J sc (ma/cm 2 ) FF (%) PCE (%) Bad ± ± ± ±0.03 Good ± ± ± ±0.03 The influence of CdSe QDs with different synthesis reaction time (i.e. size) on the device performance has been investigated. The CdSe QDs absorption spectrum and the PL peak wavelength and intensity dependence on different synthesis reaction time are shown in Fig. 4.18a and 4.18b. Due to the quantum confinement effect, the bandgap of QDs decreases along with the growth of QDs diameter (synthesis reaction time), resulting in a red shift of the absorption and photoluminescence. According to ref. 202, the calculated diameter of QDs varies from 3.44 nm for the 0.5 min sample to 5.27 nm for the 120 min sample as summarized in table 4.5, along with optical bandgaps reduction from 2.12 ev to 1.96 ev. Additionally, the energy level of different sized QDs could also change slightly 120. It is notable that under good synthesis conditions as shown in Fig. 4.18b, the PL intensity increases dramatically within the first 1 min, followed by a considerable drop down to a minimum point at around 30 min, and recovers afterwards to the maximum point at 120 min. Fig. 4.18c shows the J-V characteristics of devices based on acid-treated CdSe QDs with different synthesis reaction times. For comparison, all the other parameters of device fabrication retained the same except using different sized QDs. The tendency to increase the device PCE along with the QDs reaction time is clearly observed. Table 4.5 summarizes the performance of devices based on QDs with different synthesis reaction times. 71

82 Chapter 4. Surface modification of CdSe QDs Figure 4.18 (a) Absorption spectra, (b) PL peak wavelength and intensity dependence, and (c) J-V characteristics of devices based on acid-treated CdSe QDs with different synthesis reaction times. Devices were fabricated and measured in our lab without spectral mismatch correction. Table 4.5 Summarized performance parameters of devices based on acid-treated CdSe QDs with different synthesis reaction times. synthesis time (min) Calculated size (nm) PL peak wavelength (nm) PL intensity (a.u.) V oc (V) J sc (ma/cm 2 ) FF (%) PCE (%) ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

83 4.5 Influence of the QDs synthesis on the device performance Devices with 0.5 min and 2 min synthesized QDs exhibit similar poor PCEs of about 0.09%, although their sizes are quite different varying from 3.4 nm to 4.5 nm. One possible reason could be that once the size of QDs is smaller than a certain threshold value, the percolation pathways for the electron transport are inefficient since electrons have to hop much more times from the exciton dissociation site until they reach the electrode in comparison with the big sized QDs. In addition, the device PCE improves significantly from 0.42% for the 30 min synthesized QDs based device to 1.7% for the 120 min synthesized QDs based device. Although the absorption spectrums are almost the same for these two types of QDs, the PL intensity of the 120 min synthesized QDs are much higher than that of the 30 min synthesized one. Therefore, devices based on QDs with higher PL intensity exhibit better performances. This result is in consistence with the results previously shown in Fig. 4.16, confirming that the quality of the QDs surface is a crucial factor for the hybrid solar cell performance. In summary, an effective post-synthetic hexanoic acid treatment of non-ligand-exchanged CdSe QDs before their integration into photovoltaic devices was demonstrated, leading to improved device performance and reproducibility. Bulk-heterojunction hybrid solar cells based on blends out of acid treated CdSe QDs and conjugated polymer P3HT were investigated and a simple ligand sphere model was derived from PL quenching, TEM and DLS results to explain the improved PCEs. Devices with optimized ratios of QDs to P3HT exhibit PCEs of about 2.0%. This indicates that an effective reduction of the immobilized ligand sphere without introducing surface defects on the QDs is a crucial factor to enhance the device performance. Moreover, comparison studies revealed that the CdSe QDs synthesis condition has a crucial impact on the QD surface quality, which can be partially detected by the PL intensity of QDs. High quality QDs with less surface defects are desirable for achieving high efficient hybrid solar cells. 73

84 Chapter 5. Hybrid solar cells based on acid treated CdSe QDs and low bandgap polymer Chapter 5 Hybrid solar cells based on acid treated CdSe QDs and low bandgap polymer PCPDTBT (The main content of this chapter was published in Solar Energy Materials and Solar Cells 95, 1232 (2011) entitled Efficiency enhancement for bulk-heterojunction hybrid solar cells based on acid treated CdSe quantum dots and low bandgap polymer PCPDTBT.) As discussed in previous chapters, surface modification on NCs has been considered as a crucial factor to improve charge transfer from polymer to NCs as well as electron 132, 203 transport between NCs. Devices based on traditional pyridine ligand exchanged and hexanoic acid treated 52 NCs exhibited the state-of-the-art PCEs due to the effective replacement or reduction of the insulating ligand shell. In addition, low band gap polymers which can absorb at longer wavelength regions are promising for harvesting more photons from the solar emission. Poly[2,6-(4,4-bis-(2-ethylhexyl)-4H-cyclopenta[2,1-b;3,4-b']-dithiophene)-alt-4,7-(2, 1,3-benzothiadiazole)] (PCPDTBT, which chemical structure is shown in Fig.1.10) with a band gap of ~1.4 ev and a relatively high hole mobility up to cm 2 V -1 s appears to be an excellent candidate as a photon-absorbing and electron donating material 204. Devices based on PCPDTBT:PC 70 BM system achieved PCEs up to 6.1% 102. Nevertheless, NCs/polymer hybrid solar cells using low bandgap polymers are seldom explored so far. Recently Dayal et al. reported the certified highest PCE of 3.13% for polymer:ncs hybrid solar cell based on PCPDTBT 51 and pyridine-treated CdSe tetrapods. 5.1 Hexanoic acid treatment of TOP/OA-capped CdSe QDs We extended our investigations by applying the hexanoic acid-assisted washing 74

85 5.1 Hexanoic acid treatment of TOP/OA-capped CdSe QDs procedure to different ligand-capped CdSe QDs. Here, highly crystalline CdSe QDs with a mean diameter of about 4.7 nm were synthesized by using a flow-through microreactor system and a mixture of trioctylphosphine (TOP) and oleic acid (OA) ligands (delivered from Bayer Technology Services (BTS)). Before being incorporated into polymer, as-prepared QDs were treated by hexanoic acid. Briefly, 16 µl CdSe QDs (134 mg/ml) were added to 6 ml hexanoic acid at 105 C and stirred for 10 min. 12 ml methanol was added afterwards to precipitate the QDs, which were recovered by centrifugation. Then 1 ml chloroform and 3 ml methanol were added at 90 C to precipitate the QDs again. After being recovered by centrifugation, QDs were finally dispersed into organic solvents (e.g. CB, DCB, and TCB) at a concentration of about 23 mg/ml. Afterwards, 15 mg/ml P3HT (regioregularity >98.5%, ~ 50,000 MW from Aldrich, molecular weight ~50,000) in DCB was mixed with QDs solution at certain ratios. The blend solution was spin-coated onto the surface of PEDOT:PSS covered ITO glass substrate to form photoactive films. An Al cathode was thermally deposited onto the photoactive film. Afterwards, devices were annealed at 145 o C for 10 min followed by 160 o C for 10 min. Device performances were characterized under AM1.5G 100 mw/cm 2 illumination in a nitrogen glove-box (most of the measurements were performed at ISE). Similar to the acid treated HDA-capped CdSe QDs, TEM investigations indicated that the average particle to particle distance is reduced significantly after treatment in comparison with the untreated QDs. As shown in Fig. 5.1a, without acid treatment particles are usually well separated. After acid treatment the aggregation tendency is clearly visible as shown in Fig. 5.1b, indicating the potential for the formation of efficient percolation pathways in the photoactive film. Figure 5.1 TEM pictures of CdSe QDs (a) without and (b) with hexanoic acid treatment. 75

86 Chapter 5. Hybrid solar cells based on acid treated CdSe QDs and low bandgap polymer Fig. 5.2 shows the V oc, J sc, FF, and PCE values as a function of increasing CdSe QD wt% loading under illumination. Similar to the devices based on HDA-capped CdSe QDs as depicted in chapter 4.2, device PCEs depend on the QD loading concentration in the blend as well. An optimum QD loading was found in the similar range of 87wt% (Fig. 5.2d), leading to the best PCEs approaching 2%. Table 5.1 summarizes the performance of devices with different CdSe loading under illumination. Figure 5.2 Variation of values for V oc, J sc, FF, and PCE based on different devices fabricated in our lab with various CdSe QD wt% loading concentrations under AM1.5G 100 mw/cm2 illumination. The average values (symbols) as well as standard deviations (error bars) are shown. Devices were fabricated and measured at ISE after spectral mismatch correction. Table 5.1 Summary of the performance parameters of devices with different CdSe QD loadings in P3HT:CdSe blends. CdSe Loading V oc (mv) J sc (ma/cm 2 ) FF (%) PCE (%) 87.5 wt% 649.8± ± ± ± wt% 661.2± ± ± ± wt% 658.2± ± ± ± wt% 648.3± ± ± ±

87 5.1 Hexanoic acid treatment of TOP/OA-capped CdSe QDs The J-V characteristic of the best device fabricated from a blend of 87.5 wt% TOP/OA-capped CdSe QDs and P3HT is shown in Fig. 5.3, the best device based on 87 wt% acid treated HDA-capped CdSe QDs is also shown for comparison 29. Although these two kinds of QDs have different sizes (5.5 nm for HDA-capped one and 4.7 nm for TOP/OA capped one) which could result in different energy levels of QDs as well, after acid treatment both devices exhibit PCEs of 2.1%. These performance parameters from both P3HT:CdSe devices are comparable, indicating that the acid washing procedure is generally applicable to CdSe QDs with TOP/OA ligands for improving the photovoltaic device performance. Table 5.2 summarizes the performance of devices based on two different ligands-capped CdSe QDs under illumination. Figure 5.3 J-V characteristics of the best devices fabricated from a blend of 87.5 wt% TOP/OA-capped CdSe QDs and P3HT. The best device based on 87 wt% HDA-capped CdSe QDs is also shown for comparison. Devices were fabricated and measured at ISE after spectral mismatch correction. Table 5.2 Summary of the performance of devices based on two different ligands-capped CdSe QDs under illumination. CdSe ligand V oc (mv) J sc (ma/cm 2 ) FF (%) PCE (%) HDA TOP/OA

88 Chapter 5. Hybrid solar cells based on acid treated CdSe QDs and low bandgap polymer In addition the influence of different solvents for the CdSe QDs on the device performance has been investigated. Fig. 5.4 shows the J-V characteristics of devices using different solvents (CB, DCB, and TCB) for the QD part and DCB for the polymer part (4.5:1, vol/vol) under illumination. The J sc, V oc, FF, and PCE values of these devices are shown in table 5.3. Although an improved PCE for P3HT:CdSe nanorod devices have been reported by using the high boiling point solvent TCB 132 due to the slow evaporation rate during the spin-coating process resulting in a favorable ordering of P3HT and improving the hole mobility, here device fabricated using CB as a solvent for the QDs exhibited the best PCEs. The reason could be attributed to the lower solubility of the QDs in TCB or DCB in comparison to that in CB. High solubility of the QDs is desirable for reproducible performance as well as allowing a high loading of the QDs in the blend, which is preferable for an efficient percolation network formation during the annealing of the photoactive film. Figure 5.4 J-V characteristics of devices fabricated from a blend of P3HT in DCB and 87.5 wt% TOP/OA-capped CdSe QDs in different solvents. Devices were fabricated and measured in our lab without spectral mismatch correction. Table 5.3 Summary of the performances of devices based on CdSe QDs in different solvents. Solvent for CdSe V oc (V) J sc (ma/cm 2 ) FF (%) PCE (%) CB 0.68± ± ±1 1.64±0.08 DCB 0.71± ± ±0 1.57±0.03 TCB 0.65± ± ±3 0.47±

89 5.1 Hexanoic acid treatment of TOP/OA-capped CdSe QDs Fig. 5.5 shows the device performance dependence on the solvent and shelf lifetime of QD inks after acid treatment. Devices fabricated immediately after acid treatment of the QDs in CB exhibited a higher PCE of 1.9% in comparison with using DCB as a solvent leading to a PCE of 1.6%. However, after acid treatment the QDs in solution are not stable. The PCE of a device fabricated from a two hours aged QDs ink in CB decreased from 1.9% to 1.6%. For two hours aged QD ink in DCB, the PCE decreased from 1.6% to 1.2%. A possible reason could be that the acid treatment reduces the TOP/OA ligand shell which is used for preventing aggregation, resulting in a reduction of the colloidal stability. Therefore, the QDs are less soluble in organic solvents and tend to precipitate more easily, resulting in large scale phase separation when they are mixed with the polymer. Inappropriate phase separation in the blend film leads to more dead-ends for the charge carrier transport both for holes in the polymer phase and electrons in the CdSe phase, thus being responsible for the decrease of PCEs of devices. Table 5.4 summarizes the performance of devices fabricated from different time aged CdSe QDs in different solvents. These results suggest that in order to get a better device performance, photoactive film formation should be done as soon as the QD ink is ready after acid treatment. Figure 5.5 J-V characteristics of devices fabricated from CdSe QDs immediately after acid treatment (solid line) and after two hours (dashed line) in CB (black) or DCB (red). Devices were fabricated and measured in our lab without spectral mismatch correction. 79

90 Chapter 5. Hybrid solar cells based on acid treated CdSe QDs and low bandgap polymer Table 5.4 Summary of the performances of devices based on different time aged CdSe QDs in different solvents. Solvent/shelf time V oc (V) J sc (ma/cm 2 ) FF (%) PCE (%) CB 0h 0.68± ± ±1 1.82±0.08 CB 2h 0.70± ± ±1 1.49±0.07 DCB 0h 0.71± ± ±0 1.46±0.09 DCB 2h 0.71± ± ±1 1.14± Solar cells based on the low bandgap polymer PCPDTBT The acid treated TOP/OA-capped CdSe QDs were blended together with the low bandgap polymer PCPDTBT to form photoactive layers. Photovoltaic devices have been optimized with respect to the polymer:cdse ratio and photoactive film thickness, leading to PCEs up to 2.7% after spectral mismatch correction for spherical CdSe QDs based hybrid solar cells. To date this efficiency is the highest reported value for CdSe QDs based photovoltaic devices. In addition, a comparison of devices using P3HT and PCPDTBT as donor material is also discussed, revealing that the polymer has an essential impact on the absorption properties of the blend film as well as device performance consequently. Fig. 5.6a shows the schematic energy level diagram for P3HT, PCPDTBT, and CdSe QDs, as well as the TEM image of the QDs and the chemical structure of the polymers. Due to the energy level offset between polymers and QDs, in both cases QDs act as electron acceptors and polymers as hole acceptors. Fig. 5.6b shows the UV-Vis absorption spectrum of P3HT, PCPDTBT, and TOP/OA-capped CdSe QDs after acid. P3HT films exhibit an absorption band at around 500 nm, while PCPDTBT films show two distinct absorption bands at 410 and 720 nm, respectively. CdSe QDs films show an excitonic absorption peak at 600 nm, and exhibits strong absorbance in the shorter wavelength region between 300 and 500 nm, complementing the inferior absorption from the polymers in this region. Fig. 5.6c shows the absorption spectrum of composite films with 87.5 wt% QDs loading in P3HT and PCPDTBT, respectively, showing a superposition of respective absorption spectra of pure polymer and QDs. 80

91 5.2 Solar cells based on low bandgap polymer PCPDTBT Figure 5.6 Fig.1. (a) Schematic energy level diagram of P3HT, PCPDTBT, and CdSe QDs, as well as the TEM image of the QDs and the chemical structure of the polymers. The bandgap energy values are taken from Ref 118, 116 and 126, respectively. (b) Absorption spectra of P3HT (dashed-dotted, black), PCPDTBT (dashed, red), and TOP/OA-capped QDs after acid treatment (solid, green) in thin films. (c) Absorption spectra of the composite films with 87.5 wt% QD loading in P3HT (dashed-dotted, black) and PCPDTBT (dashed, red), respectively. The current density-voltage (J-V) characteristics of the best cells fabricated from blends of P3HT:CdSe and PCPDTBT:CdSe under illumination are shown in Fig. 5.7a. The P3HT:CdSe cell exhibited a PCE of 2.1% with an V oc =628 mv, a short-circuit J sc =6.0 ma/cm 2, and a FF = Moreover, by using PCPDTBT, a PCE of 2.7% was achieved for the best cell with V oc = 591 mv, J sc =8.3 ma/cm 2, and FF=0.56. Fig.5.7b shows the EQE spectrum of photovoltaic devices based on different polymers. In comparison with the P3HT:CdSe device, the PCPDTBT:CdSe device showed a broader EQE spectrum from 300 to 850 nm, in consistence with the absorption spectrum of the composite film as depicted in Fig.1c. It is notable that a considerable photocurrent contribution from the QDs was observed in the PCPDTBT:CdSe device. In comparison with the absorption spectrum of the PCPDTBT film where the band around 700 nm dominates, the EQE spectrum showed an improved photocurrent generation in the region where the QD absorption is strong, leading to a maximum EQE of 55% at 400 nm. This implies that both components of the PCPDTBT:CdSe system contribute to the absorption of incident photons under photocurrent generation, in contrast to the typical polymer/fullerene system where the photocurrent contribution from the fullerene component is very weak

92 Chapter 5. Hybrid solar cells based on acid treated CdSe QDs and low bandgap polymer Figure 5.7 (a) J-V characteristics of the best solar cell fabricated from a blend of P3HT:CdSe with a PCE of 2.1% (dashed, black) and PCPDTBT:CdSe with a PCE of 2.7% (solid, red) under illumination. The inset shows the schematic structure of a photovoltaic device. (b) EQE spectrum of the P3HT:CdSe device (dashed, black) and the PCPDTBT:CdSe device (solid, red). Devices were fabricated and measured in ISE after spectral mismatch correction. The influences of solvents for the polymer, the polymer:cdse ratio, photoactive film thickness, thermal annealing treatment and the use of different cathode materials on the performance of the PCPDTBT:CdSe devices have been investigated systematically. Fig. 5.8 shows the performances of devices fabricated from solvent mixtures of DCB or TCB for the polymer and CB for the QDs (6:1, vol/vol). Both devices performed the same V oc, while a TCB based device exhibited higher J sc, resulting in a better PCE in comparison to a DCB based device. This result is in consistence with results reported in literature that TCB is usually used as solvent for PCPDTBT 51. Table 5.6 summarizes the device performance parameters. 82

93 5.2 Solar cells based on low bandgap polymer PCPDTBT Figure 5.8 J-V characteristics of devices fabricated from solvent mixture of DCB (black) and TCB (red) for the polymer and CB for QDs (6:1 vol/vol). Devices were fabricated and measured at ISE after spectral mismatch correction. Table 5.6 Summarized performance parameters of devices fabricated from solvent mixture of DCB or TCB for the polymer and CB for QDs (6:1 vol/vol). Solvent for polymer V oc (V) J sc (ma/cm 2 ) FF (%) PCE (%) DCB 582.2± ± ± ±0.16 TCB 592.0± ± ± ±0.14 Fig. 5.9 shows the variations of values for V oc, J sc, FF, and PCE as a function of CdSe QD wt% loading (a)-(d) and photoactive film thickness (e)-(h). The average values and error bars are derived from six individual cells on the same substrate. The device performance is sensitive to the QD loading and an optimum value was found around 87.5 wt%, which is similar to that of our P3HT:CdSe devices 52 and literature reports 51. Lower QD loading results in decreased PCEs due to the inefficient percolation pathways for electrons transport. While higher QD loading leads to insufficient photon absorption from the polymer part and lower hole mobility. The optimized composite film thickness was found to be around 100 nm, leading to a PCE of 2.66 ± 0.05% with V oc = 588 ± 2 mv, J sc = 8.16 ± 0.12 ma/cm 2, and FF = 0.55 ± In thicker films, the photocurrent reaches a plateau and no J sc increase was further observed (Fig.3f), while the FF drops simultaneously as a result of increased bulk resistance, leading to inferior PCEs. Additional photon absorption seems to be counterbalanced by diminishing charge transport properties. On the other hand, 83

94 Chapter 5. Hybrid solar cells based on acid treated CdSe QDs and low bandgap polymer thinner films result in insufficient photon absorption thus leading to a remarkable decrease in J sc as well as of the PCEs. Table 5.7 and 5.8 summarize the PCPDTBT:CdSe ratio and photoactive layer thickness dependence on the device performance, respectively. Figure 5.9 Variation of values for (a) V oc, (b) J sc, (c) FF, and (d) PCE of devices with various CdSe QD wt% loadings in the blend. Variation of values for (e) V oc, (f) J sc, (g) FF, and (h) PCE of devices with a fixed QD loading of 87.5 wt% in blends and various blend film thickness. Devices were fabricated and measured at ISE after spectral mismatch correction. Table 5.7 Summarized performance parameters of devices with different CdSe QD loadings in PCPDTBT:CdSe blends. CdSe Loading V oc (mv) J sc (ma/cm 2 ) FF (%) PCE (%) 85.7 wt% 601.8± ± ± ± wt% 610.0± ± ± ± wt% 607.7± ± ± ±

95 5.2 Solar cells based on low bandgap polymer PCPDTBT Table 5.8 Summarized performance parameters of devices with different PCPDTBT:CdSe photoactive layer thicknesses. Thickness (nm) V oc (mv) J sc (ma/cm 2 ) FF (%) PCE (%) 75.0± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±0.06 The effects of different cathode materials on the device performance have been investigated. Although LiF/Al 205 and Ca/Ag 201 cathodes are widely used for polymer/pcbm photovoltaic devices exhibiting better performances than Al only devices, our results showed that devices with an Al cathode exhibited higher PCEs in such polymer:cdse systems as shown in Fig A device with an Al cathode exhibited a PCE of 2.7% with V oc = 591 mv, J sc = 8.3 ma/cm 2, and FF=0.56 after the second post-annealing step (first step performed at 145 o C for 10 min followed by a second step at 160 o C for 5 min, Fig. 5.10, black curve). However, devices with a LiF/Al cathode exhibited a significant lower PCE of 0.9% after pre-annealing (single step at 145 o C for 10 min, Fig. 5.10, red curve) with V oc = 483 mv, J sc = 3.7 ma/cm 2, and FF=0.52, and a device with a Ca/Ag cathode exhibited a PCE of 0.8% after pre-annealing (single step at 145 o C for 10 min, Fig.5.10, green curve) with V oc = 469 mv, J sc = 3.2 ma/cm 2, and FF=0.51. Table 5.9 summarizes the dependence of the device performance from the cathode materials. 85

96 Chapter 5. Hybrid solar cells based on acid treated CdSe QDs and low bandgap polymer Figure 5.10 J-V characteristics of devices fabricated with different cathode materials. Devices were fabricated and measured at ISE after spectral mismatch correction. Table 5.9 Summarized performance parameters of devices with different cathode materials. Cathode V oc (mv) J sc (ma/cm 2 ) FF (%) PCE (%) Al 588.0± ± ± ±0.05 LiF/Al 481.6± ± ± ±0.04 Ca/Ag 464.4± ± ± ±0.04 Additionally, the effects of different thermal annealing treatments on the device performance have been investigated systematically. As shown in Fig.5.11, devices without annealing (No.1) and with pre-annealing before Al deposition (No. 2, 145 o C for 10 min) showed similar very poor PCEs around 0.5%-0.6%. With higher temperature pre-annealing at o C, devices (No. 3-6) show better PCEs with values around %. The optimum condition for device pre-annealing is at 210 o C for 10 min leading to a PCE of 1.7%. By applying post-annealing after Al deposition, device performances have been dramatically improved. After first post-annealing at 145 o C for 10 min (No. 7n), a device exhibited a PCE of 2.08±0.07% with V oc = 598.2±1.6 mv, J sc = 6.58±0.17 ma/cm 2, and FF=0.532± An additional post-annealing at 160 o C for 5 min (No. 8) further increases the device PCE to 2.43±0.05% as a result of J sc enhancement from 6.58±0.17 ma/cm 2 to 7.77±0.11 ma/cm 2. Table 5.10 summarizes the influence of the thermal annealing treatment on device performance. 86

97 5.2 Solar cells based on low bandgap polymer PCPDTBT Figure 5.11 Variation of values for V oc, J sc, FF, and PCE based on devices fabricated with different thermal annealing treatments. Devices were fabricated and measured at ISE after spectral mismatch correction. Table 5.10 Summarized performance parameters of devices with different annealing treatments. Device No. Annealing treatment Temp, Time ( o C, min) V oc (mv) J sc (ma/cm 2 ) FF (%) PCE (%) 1 No ± ± ± ± Pre-annl 145, ± ± ± ±0.0 3 Pre-annl 210, ± ± ± ±0.0 4 Pre-annl 220, ± ± ± ± Pre-annl 220, ± ± ± ± Pre-ann 220, ± ± ± ±0.0 7 Post-annl 145, ± ± ± ±0.07 8* Post-annl 160, ± ± ± ±0.05 * Additional annealing step of device no. 7 87

98 Chapter 5. Hybrid solar cells based on acid treated CdSe QDs and low bandgap polymer In summary, bulk-heterojunction hybrid solar cells based on TOP/OA-capped CdSe QDs after hexanoic acid treatment and the formation of photoactive hybrid films with the low bandgap polymer PCPDTBT were fabricated. Devices with an optimized PCPDTBT:CdSe ratio, photoactive film thickness, thermal annealing treatment, and cathode material exhibited a best PCE of 2.7%. Comparison studies of P3HT and PCPDTBT based devices revealed that the improved PCEs in PCPDTBT:CdSe device can be mainly attributed to the increased J sc as a result of the improved match of the blend absorption with the solar emission spectrum, as supported by UV-Vis absorption and EQE measurements. The results suggest that the hexanoic acid treatment is generally applicable to various ligand-capped CdSe NCs for improving photovoltaic device performances. In addition, low bandgap polymers which can harvest photons at longer wavelength region are promising to be incorporated into hybrid solar cells. 88

99 6.1 Summary Chapter 6 Summary and outlook 6.1 Summary There has been an impressive progress achieved for organic solar cells during the last couple of years regarding their device performance mainly referring to their PCE values. Hybrid solar cells based on semiconducting polymers and inorganic nanoparticles are still lagging behind in PCE values by roughly a factor of 2 and more compared to the best known organic solar cell systems based on polymer:pcbm composites. Besides size and morphology inhomogenities, the surface constitution of semiconductor NCs is usually hard to control which leads to fluctuations in material quality, solubility and device performance. The ligand surface constitution automatically affects the solubility and ink stability of NC solutions. This dissertation presents the results of a research aiming at the development of bulk-heterojunction hybrid solar cells based on colloidal CdSe QDs and conjugated polymers. Both the materials and device structures are investigated and optimized systematically in respect of QD synthesis and post-synthetic modification, hybrid nanocomposites formation, and device fabrication, leading to an improvement of hybrid solar cells PCEs. The state-of-the-art development of bulk heterojunction hybrid solar cells has been reviewed. Critical factors limiting the solar cell device performance were highlighted and strategies for further device improvement have been demonstrated by giving recent examples from literature. Highly reproducible synthesis methods for CdSe QDs are applied, leading to a narrow size distribution and excellent photophysical properties. Pre-heating of the HDA ligand and aging of the Se-TOP precursor have been proven as two critical parameters for synthesizing high quality QDs. The influence of the QD characteristics such as 89

100 Chapter 6. Summary and outlook diameter, PL peak wavelength, and PL intensity on the performance of hybrid solar cells was studied, revealing that the synthesis conditions have a crucial impact on the QD surface quality, which can be partially detected by the PL intensity. As a result, high quality QDs are desirable for achieving efficient photovoltaic devices. An effective post-synthetic hexanoic acid treatment on HDA-capped CdSe QDs before their integration into photovoltaic devices was demonstrated. Solar cells with optimized ratios of QDs to P3HT exhibit PCEs of about 2.1%. A simple ligand sphere model was derived from PL quenching, TEM and dynamic light scattering results. The results indicate that an effective reduction of the immobilized ligand sphere is a crucial factor to enhance the device performance. Furthermore, extended investigations on applying the hexanoic acid treatment to different ligand-capped (i.e. mixture of TOP and OA) CdSe QDs were presented. The comparable performance of devices based on P3HT and different ligand capped QDs indicates that the acid treatment is generally applicable to QDs with TOP/OA ligands for improving device performance. In addition, lower bandgap polymer PCPDTBT has been used instead of P3HT as polymer part for the formation of the photoactive hybrid film. Here, optimized devices exhibit PCEs of 2.7% after spectral mismatch correction. This value is the highest reported one for spherical CdSe QD based hybrid solar cells. Comparison studies of P3HT and PCPDTBT based devices revealed that the improved PCEs in PCPDTBT:CdSe device can be mainly attributed to the increased J sc as a result of the improved match of the blend absorption with the solar emission spectrum, as supported by UV-Vis absorption and EQE measurements. In addition, low bandgap polymers which can harvest photons at longer wavelength region and have adequate energy levels are promising to be incorporated into hybrid solar cells. Fig. 6.1 summarizes the development of the photovoltaic device PCEs during my PhD period from July 2007 to the February The PCEs improved dramatically from much less than 0.01% to over 3.0% due to the progress on the composite materials as well as device optimization. 90

101 6.1 Summary Figure 6.1 Development of device PCEs during the PhD period. 6.2 Outlook In order to form optimized photoactive hybrid films a systematic evaluation of the device performance as a function of various film processing parameters such as concentrations and donor/accepter material ratios within the composite, active layer film thickness, film annealing temperature, etc. have to be performed. Deeper understandings of the structure-function relationship within bulk-heterojunction composite layers are still lacking. Additionally, one critical aspect is the control over the nanomorphology of the donor and acceptor phases in the photoactive bulk-heterojunction films. Improved surface treatments of the NCs and surface active additives might help to promote phase segregation and charge dissociation between the donor and acceptor interface and increase the charge carrier mobilities within the donor and acceptor phases leading to an increase of the photocurrent. From the polymer side, the development and synthesis of novel low band gap semiconducting polymers will allow a more efficient energy harvesting from the solar radiation. Additional band gap engineering of polymers will help to promote the usability of nanocrystals systems based on e.g. PbS, PbSe and Si NCs leading to an improved utilization of the solar energy as well. 91

102 Chapter 6. Summary and outlook Fig. 6.2 summarizes the strategies for improving hybrid solar cell performance. Some aspects and concepts have been explored preliminarily in this dissertation. First results indicate that further improvement on device performance is possible. Figure 6.2 Strategies for further improving the performance of hybrid solar cells CdSe QD:Carbon nanotube composites Due to the high percolation threshold, high QD volume fractions in the hybrid composites are needed to overcome the limitation of inefficient electron hopping between QDs, leading to a small volume fraction of polymer lowing the hole mobility. Aiming to improve the carrier mobility, carbon nanotubes (CNTs) have recently been incorporated into such kind of devices due to their remarkable electronic, mechanical, and chemical properties 206, 207. Such QD-CNT composites are well suited for solar cells due to the tunable bandgap of QDs and the quasi-one-dimensional electron transport properties of CNTs. We investigated a non-covalent functionalization approach of multiwalled carbon nanotubes (MWNTs) by polymer wrapping with PAH (Poly(allylamine hydrochloride)) following a direct attachment of CdSe QDs. PL quenching in such composites has been studied, leading to first results indicating the promising potential for the efficiency improvement of solar cells by using such composites. MWNTs (produced by catalytic chemical vapor deposition, purity >99%) were 92

103 6.2 Outlook obtained from Bayer Material Science, Germany. Poly(allylamine hydrochloride) (PAH) was purchased from Sigma-Aldrich and used as received. The MWNTs were functionalized by polymer wrapping with PAH. Typically, MWNTs were dispersed in a 0.5 wt% PAH NaCl solution and sonicated for 5 h. The modified MWNTs were transferred from aqueous to organic solvent (e.g. chloroform) by repeated centrifugation and redispersed in ethanol. Consequently, an appropriate amount of CdSe QDs were added to polymer-wrapped MWNTs suspension for the reaction of non-covalent attachment. The mixture was treated in an ultrasonic bath for 15 min. As a result, the composites of MWNTs with a homogenous CdSe QDs decoration could be obtained due to the van der Waals interactions and mechanical anchoring. Direct evidence for the successful conjugation of CdSe QDs to the surface of MWNTs was given by TEM as shown in Fig.6.3. The random and homogeneous decoration of MWNTs with QDs is clearly visible. Figure 6.3 TEM image of CdSe QD-MWNT composites The normalized PL intensities of QD-MWNT and P3HT-MWNT composites as a function of the MWNT concentration are shown in Fig.6.4a. The observed PL quenching in QD-MWNT composites is much more pronounced than in P3HT-MWNT composites. It could be explained that in a solar cell device built up by a P3HT-QD-MWNT composite, P3HT is the hole carrier and the QD-MWNT-composite is the electron carrier, as illustrated in Fig.6.4b. A strong PL quenching effect is observed for the QD-MWNT composites. The excited electrons will be transferred to the MWNTs while the excited holes will be transferred to the polymer. This result suggests that with the incorporation of MWNTs at an appropriate 93

104 Chapter 6. Summary and outlook concentration, the optimized QD volume fraction in the device could be reduced and higher hole and electron mobilities can be expected, since MWNTs with a high conductivity and a much lower volume threshold will facilitate the electrons transport. Figure 6.4 (a) Normalized PL intensities of QD-MWNT and P3HT-MWNT composites as a function of the MWNT loading. (b) Schematic illustration of charge transfer and transport in P3HT-QD-MWNT composites. Further investigations on such composites are still in process. Improvements are expected by increasing the coating ratio of QDs on CNTs, investigating the influences of different CNTs (semiconducting, metallic, single walled or multi walled tubes) with different energy levels, as well as by studying the performance of solar cells devices using different QD-CNT composites Mixtures of QD and NR CdSe NCs with elongated structure such as nanorod (NRs) are preferable for hybrid solar cells since they naturally provide a direct pathway for electron transport. However, NRs tend be horizontal alignment 110 instead of vertically aligned. In such case, elongated structures are not superior to spherical QDs for extracting free carriers since electron needs to hop in vertical direction from the exciton dissociate site to the cathode. Therefore, mixtures of QDs and NRs could offer an alternative solution for improving electron transport within the bulk-heterojunction structure. 94

105 6.2 Outlook Figure 6.5 TEM image of CdSe QDs (left) and NRs (right). (Material and photo provide by Bayer Technology Services). The TEM image of CdSe QDs with a mean diameter of 4.7 nm and NRs with a diameter of 7 nm and a respect ratio of 4 to 6 are shown in Fig The NRs were synthesized in a mixture of TOP, TOPO, and tetradecylphosphonic acid (TDPA) as ligands. First results indicate that the hexanoic acid treatment protocols for QDs are not directly applicable to the NRs due to a different ligand system. Device fabricated from acid treated NRs and PCPDTBT exhibited relatively lower PCEs about 1%, while devices based on untreated NRs showed even much lower PCEs < 0.1%. Optimization of the acid treatment on NRs in respect of repetition and treatment time is under investigation. Fig. 6.6 shows the variation of values for V oc, J sc, FF, and PCE for different devices fabricated with various CdSe QD to NR weight ratio as acceptor (90 wt%) and PCPDTBT as donor. The photovoltaic devices based on a mixture of QD:NR show promising and better PCEs than QD only or NR only based devices in direct comparison, although the acid treatment for NRs is far from optimum. Table 6.1 summarizes the device performance. Further improvement could be expected by optimizing the acid treatment on NRs, and photoactive layer thickness. 95

106 Chapter 6. Summary and outlook Figure 6.6 Variation of values for V oc, J sc, FF, and PCE based on different devices fabricated with various CdSe QD to NR weight ratio. Table 6.1 Summarized performance of device with QD only and mixture of QD:NR (4:1 by weight) as acceptor and PCPDTBT as donor. QD:NR (by wt) V oc (mv) J sc (ma/cm 2 ) FF PCE (%) 100: : : : Stability of hybrid solar cells The shelf stability of a P3HT:CdSe device stored in nitrogen glove-box has been investigated preliminarily in comparison with that of a P3HT:PCBM device. Table 6.2 shows the device V oc, J sc, FF, and PCE variation on the shelf time. We observed the same tendency of device performance variation for both devices, e.g. V oc increasing, J sc, FF, and PCE decreasing. The relative degradation of P3HT:CdSe device (15% reduction) is comparable to that of the P3HT:PCBM device (20% reduction) after 14 days, although the initial PCE of P3HT:CdSe (1.63%) is lower the P3HT:PCBM 96

107 6.2 Outlook device (2.86%). This result might indicate that the stability of bulk-heterojunction solar cells with inorganic semiconductor nanocrystal as acceptor is similar to the one based on fullerene derivatives under inert conditions. Further investigation on the device stability under different test conditions is still ongoing. Table 6.2 Summarized performances of fresh and 14 days aged devices. Device V oc (mv) J sc (ma/cm 2 ) FF (%) PCE (%) P3HT:PCBM fresh P3HT:CdSe fresh P3HT:PCBM 14 days (+0.1%) 7.46 (-4.6%) 48.6 (-16.9%) 2.30 (-19.6%) P3HT:CdSe 14days (+0.1%) 5.13 (-4.1%) 41.0 (-10.7%) 1.38 (-15.3%) In summary, potential PCE values of 10% for OSCs as stated by Coakely and McGehee 22 and Scarber et al. 124 seems to be possible also for optimized hybrid solar cells which would allow to cross the threshold for commercialization, competing with already existing technologies such as DSSCs and thin film technologies. Additional factors have to be considered such as material toxicity, material costs and recovery, device long-term stabilities, achievable module efficiencies and energy payback time to directly compare different competing technologies. Combining all these efforts requires a highly interdisciplinary research collaboration of chemists, physicists and engineers. Nevertheless, with increased funding and research efforts in academia and in companies the probability for a sufficient device improvement for commercialization of hybrid solar cells and OSCs in general is realistic within the next couple of years. 97

108 Appendix Appendix A1. Solar Cells Fabrication A pre-evaluation setup for solar cells fabrication and measurement has been designed and established in our lab during the PhD period, allowing us to investigate various device parameters independently and systematically before more accurate measurements in cooperation with the Fraunhofer ISE were performed. Fig. A1.1 shows the main device fabrication procedures, including ITO substrate patterning and cleaning, spin-coating of the PEDOT:PSS anode buffer layer and polymer/cdse photoactive layer, deposition of the Al cathode, and finally measuring the device performance. Figure A1.1 Solar cells fabrication procedures. 98

109 A1. Solar cells fabrication A1.1 ITO substrate patterning and cleaning ITO glass substrates (CEC050S) with a diameter of 20 mm x 20 mm and a thickness of 1.1 mm were purchased from Praezisions Glas & Optik GmbH, Iserlohn, Germany. A 40 nm thick ITO film was coated on selected high quality float glass substrate, giving a good transmission property of >80% in the range between 450 nm and 1100 nm, and a typical sheet resistance 50 Ω/. As shown in Fig. A1.2, transparent pieces of tapes were used to cover the area where ITO was needed to remain. The substrate was then etched by hydrochloric acid (37%) solution for 10 min until the exposed ITO was totally removed. Afterwards, the substrate was rinsed by deionized (DI) water and the tapes were removed manually. Figure A1.2 ITO electrode structure (white) on the glass substrate (blue). After mechanical pre-cleaning by acetone stained fiber-free tissue, the substrates were ultrasonically cleaned in acetone, alcohol, and de-ionized water for 10 minutes sequentially in order to remove any fingerprints, body oils or residual organic contaminants from the transparent tape used in the patterning step. Finally they were dried by high purity nitrogen gas and put into an oven to dry at 150 o C for 30 min. The dried substrates were treated by oxygen plasma with an oxygen flux of 15 standard cubic centimeters per minute (sccm) for 2minutes in order to increase the surface work function of ITO as well as achieve hydrophilic surface for the following PEDOT:PSS spin-coating process. A1.2 Spin-coating PEDOT:PSS and photoactive layers Aqueous PEDOT:PSS dispersions (Baytron P4083, from H. C. Starck), which has 1.5% solid content and a 1:6 ratio of PEDOT:PSS, were spin-coated in a nitrogen glove box onto the pre-cleaned ITO glass substrates at a rotational speed of 3000 rpm for 70 seconds (WS-400A-6NPP/Lite, Laurell), then baked at 150 C on a hot plate for 30 99

110 Appendix minutes, forming thin films with a thickness of about 70 nm (measured by a Dektak 150 profilometer). After the formation of the PEDOT:PSS layers, blends of polymer and CdSe QDs solution were spin-coated at various speeds for about 2 min to form the photoactive layer. The thickness dependence of the film as a function of the spin-coating speed for blends of PCPDTBT and CdSe QDs (87.5 wt%) in a mixture of TCB and CB (6:1, vol/vol) with a final ink concentration around 20 mg/ml is shown in Fig. A1.3 as an example. The film thickness decreases rapidly between spin-coating speeds of 300 to 700 rpm, while it reaches a plateau value around 130 nm at higher speeds up to 1500 rpm. Further decreasing in film thickness can be achieved by using a higher acceleration rate at the beginning of the spin-coating process and/or using lower ink concentrations. Figure A1.3 Film thickness dependence as a function of spin-coating speed for blends of PCPDTBT and CdSe QDs (87.5 wt%) with a final ink concentration around 20 mg/ml. A1.3 Al cathode deposition The Al cathode was deposited in high vacuum with a base pressure of around mbar with a typical deposition rate of 1-2 nm/s. Fig. A1.4 shows the structure of the evaporation mask used to define the cathode structure and the schematic illustration of devices after Al deposition. The design of the mask allows up to 9 substrates to be loaded in one batch. 100

111 A1. Solar cells fabrication Figure A1.4 left: Al evaporation mask. right: Schematic illustration of the device structure after Al deposition. Fig. A1.5 shows a photo of a fabricated device and the corresponding schematic structures. Three individual cells are located on one substrate with a common cathode. The ITO anode is also covered by Al for getting a better electrical contact. The overlap between ITO and Al defines the active area, which is 0.08 cm 2 for each cell. Figure A1.5 Photo and schematic structure of a fabricated device. A1.4 Device measurement A holder providing electrical connections and light illumination was used for device measurement, as shown in Fig. A1.6. Light was induced from a solar simulator (LOT-Oriel, with an AM1.5G filter) by a liquid light guide into the nitrogen glove box where devices are measured. The detailed device measurement system is described in chapter A2. 101

112 Appendix Figure A1.6 left: schematics of a contacted device, right: photograph of a contacted device in a dedicated teflon-based sample holder. A2. Solar Cells Performance Test System A2. Solar cell performance test program based on LabVIEW In order to simplify the device test procedure as well as to improve measurement accuracy, a solar cell test system has been designed and established, which can achieve fully automatic measurement and data acquisition. The system includes a solar simulator (LOT-Oriel) as light source, a liquid light guide (LOT-Oriel), a device test holder, a Keithley source measure unit 2602A for I-V measurement, a GPIB-USB interface card, a computer, and a LabVIEW based program (Fig. A2.1). The system contains two main parts: 1) hardware for the I-V measurement and data transport, and 2) software for the graphical user interface, parameters setup, data calculation, display and storage. Figure A2.1 Diagram showing the individual items of the solar cell performance test system. 102

113 A2. Solar cells performance test program based on LabVIEW Hardware components: Hardware is an important part of the system. The selection of hardware should be considered comprehensively according to the need of measurement accuracy, speed, stability and system cost. Here a LOT-Oriel solar simulator with a Xe arc lamp and an AM1.5G filter is used as light source providing a standard test condition with AM1.5G spectrum under intensity of 100 mw/cm 2. The light intensity was calibrated by a certified photodiode (Thorlab S121B). An iris diaphragm is used for adjusting light intensity in a wide range. A liquid light guide ( nm transparent with a core diameter of 8 mm) is used for coupling light from solar simulator into the N 2 glove box where the devices have been tested in a dry and oxygen free environment (oxygen content < 10 ppm and water content < 5 ppm). A Keithley source measure unit 2602A is used for measuring I-V characteristics of the device. By using a GPIB-USB interface card, the source unit can be software controlled by a computer program to output voltage or current and measure corresponding currents or voltages simultaneously, combining voltage source, current source, voltage meter, and current meter all in one instrument. It has a very high-precision measurement capability: the maximum voltage resolution up to 1 µv as a current source, and the maximum current resolution up to 1 pa as a voltage source, which fully meet the needs of device testing. Software component: National Instrument (NI) LabVIEW 2009 is used as the development platform for the program design. It is a powerful and flexible graphical programming language which can integrate and communicate with various instruments via GIPB, VXI, RS-232, RS-485 etc. protocol interface. The solar cells test program consists out of six modules, namely user interface, parameter setting, data acquisition, data display, data storage, and error detection. The program has following features: 1) Wizard-style graphical control user interface. The measurement and sweep parameters can be input or modified in respective pages by clicking on the button Back and Next (Fig. A2.2a, b). The measurement will be started by clicking the button Run (Fig. A2.2c). 2) Real-time data acquisition, display, and record. Each measurement point will be 103

114 Appendix displayed in the I-V graph immediately, and meanwhile the data is saved in a previous specified file. The measurement can be stopped anytime by clicking the button stop. Figure A2.2 Program user interface: (a) welcome page, (b) measurement parameter setting page, (c) sweep setting page. 3) The measured data is saved as ASCII file including voltage, current, calculated current density and output power density. Device open-circuit voltage, short-circuit current density, fill factor and power conversion efficiency will be extracted and calculated consequently after a full sweep measurement and saved in the end of the data file, which can be imported directly to a dedicated data-processing software (e.g. Origin, MS Excel) for further analyzing. 4) The software can be compiled into an executable program (.exe) so that it can run independently from LabVIEW in various computer systems. 5) As shown in Fig. A2.3, the program is started by detecting if the connection 104

115 A2. Solar cells performance test program based on LabVIEW between a computer and 2602A is correct. If so the initialization will start by re-setting the measurement parameters back to their initial status and removing data in the buffer. Consequently, new measurement parameters, including measurement speed, filter type, number of values to be averaged, sweep start value, stop value, step value, delay time, device area, and irradiation light intensity will be sent to the Keithley 2602A sourcemeter. Afterwards, a measurement can be started by clicking the button Run. Each measurement point needs the trigger signal from the program and measured data will be displayed in the result window and saved in a file. Figure A2.3 Measurement process flow chart. A typical LabVIEW program consists of three parts: a front panel, a block diagram, and a connector panel. The front panel serves as a programmatic interface. Controls and indicators on the front panel allow a user to input data into or extract data from a running instrument. They are displayed on the front panel in the form of various icons such as knobs, switches, buttons, charts, graphs, and tec., which make it easy to understand. A block diagram can be considered as a traditional program s source code. Together with a front panel full functionality of a program can be achieved. A block diagram consists out of a port, node frame, and connection. The port is used for transferring date with the controls and indicators on the front panel. The node is used for functions 105

116 Appendix call. Frame is used for structured process control commands. Connection represents and defines the direction of the data flow within the block diagram. Connector is the interface where a sub-program can be called. Similar as function parameters, connector represents the input/output of the node data. The port of the connector should be assigned corresponding to its control and indicator on the front panel. Due to its hierarchical structure, users can create sub-programs as build blocks so that complex functionality can be easily realized. Since LabVIEW is a graphical programming language, the program code is similar to the block diagram flow chart connection lines. Some representative key codes are shown as following. 1) Connection detector As shown in Fig. A2.4, this diagram will be executed when the program starts. GPIB0::26::INSTR is the GPIB address of 2602A. ke2602reset block is the code for resetting all status to initial state. Connection status can be determined by detecting the return code from this block. If the returned Boolean value is true, it means that the connection between computer and 2602A is failed, and an error message will be displayed. Runtime counter is a variable value for recording the number of running times. Figure A2.4 Connection detector diagram. 2) User interface Tab control records the number of the currently displayed window. Run, Back, Next, and Run again are Boolean variable values with either true or false. For 106

117 A2. Solar cells performance test program based on LabVIEW example, if run button is pressed, its value is set to be true and result window will be displayed consequently. Figure A2.5 User interface diagram. 3) Measurement parameters setting The measurement parameters are sent to 2602A by the code shown in Fig. A2.6 by inputting desired values from the front panel such as measurement speed, filter type, number of averaged value. Figure A2.6 Measurement parameters setting diagram. 4) Data acquisition, display, and saving In the diagram shown in Fig. A2.7, measured voltage and current values are stored in a temporary buffer Read buffer in ASCII format and separated by comma. Then they are converted into numeral values and the current density and power density are 107

118 Appendix calculated. Meanwhile, all the data are displayed in the computer monitor stored in specified file. Figure A2.7 Data acquisition, display, and saving diagram. 5) Calculation Once the voltage, current density and power density are known, open-circuit voltage, short-circuit current density, fill factor, and power conversion efficiency can be calculated by the diagram shown in Fig. A2.8 based on the formula 1.1 described in chapter 1. As a result, these parameters will be displayed and stored and the device performance parameters are directly shown after each measurement. 108

119 A2. Solar cells performance test program based on LabVIEW Figure A2.8 Calculation diagram. 109