1 CONJUGATED POLYMERS IN BULK HETEROJUNCTION PHOTOVOLTAIC DEVICES By NATHAN C. HESTON A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE O F DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2 2009 Nathan C. Heston
3 To my family and in loving memory of my father
4 ACKNOWLEDGMENTS I could not have completed this work without the support and hel p of my family, friends, and colleagues. Graduate school has been both fun and challenging Through out my time at the University of Florida many people have supported me in various ways. My family has always been vital in any success Ive had, and in th ese past five years their support continued to help me progress. Until my fathers death last year, he was a steadfast believer in me. I will always be grateful for the letters, conversations, support, and love that he constantly gave me. My mother, sister, brother -in -law, brother, and nephew have also been there to remind me the joys of life and family. For this I would like to offer them my sincerest thanks I would like to thank both Dr. Tanner and Dr. Reynolds who welcomed me into their research gr oups. They were supportive, helpful, and inspiring at all the right times. Their advice and guidance has been invaluable. I want them both to know that I have appreciated working for them. I am grateful for all of the work that they did to create the o pportunity for me to do mine. In the same regard, I am thankful to Dr Hebbard, Dr Rinzler, and Dr Hershfield for their willingness to serve as my committee. Each one of them contributed to my understanding of science through their teaching and advice F rom the beginning of my work at the University of Florida friends, classmates, lab mates, and collaborators have always been helpful. Beginning this list I would like to thank all members of the 2004 graduate physics class. My first year with them here was one of the most demanding scholastic times of my life and they made it fun. Of them, many became my close friends and for the past five years I owe many thanks for their help. Of my coworkers there are many that I owe thanks to. Jianguo Mei has provided me with excellent samples through his diligent and creative synthe tic efforts I would like to thank him for allowing me to collaborate with him and for his unparalled optimistic attitude. I also want to
5 thank Nicolas Berton for his friendship and c ollaboration. I owe him many thanks for his insights and the materials he has provided me. I would like to especially thank Ken Graham for his always helpful and pleasant attitude. He is an all around great guy who is hard not to like. I also appreciat e his recent assistance with AFM measurements. Less specifically, but with no less gratitude, I am thankful to all members of the Reynolds and Tanner research groups. I have learned many things from you. I cannot conclude this list without mentioning D r. Edward Rajaseelan. As an undergraduate student I was inspired by him. He was one of the greatest teachers I have ever had, and through his every action he showed how much he cared about his students Thank you Dr. R.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 8 LIST OF FIGURES .............................................................................................................................. 9 ABSTRACT ........................................................................................................................................ 14 CHAPTER 1 DISSERTATION SUMMARY .................................................................................................. 15 2 CONJUGATED POLYMERS: GENERAL PRINCIPALS AND APPLICATIONS ............ 16 2.1 Introduction ....................................................................................................................... 16 2.2 Fundamental Properties of Conjugated Polymers ........................................................... 17 2.3 Trans port and Conductivity .............................................................................................. 28 3 EXPERIMENTAL TECHNIQUES AND INSTRUMENTATION ........................................ 37 3.1 Introduction ....................................................................................................................... 37 3.2 Fundamentals of Photovoltaic Cells ................................................................................ 38 3.2.1 Doping 39 3.2.2 P N Junctions ........................................................................................................ 41 3.2.3 P N Junction Under External Bias ....................................................................... 45 3.2.3 Photovoltaic Cells ................................................................................................. 49 3.2.4 Detaile d I -V Behavior of a Photovoltaic Cell ..................................................... 52 3.2.5 Solar Cells ............................................................................................................. 55 3.3 Organic Solar Cells ........................................................................................................... 63 3.3.1 First Generation Organic Cells ............................................................................ 63 3.3.2 Bulk Heterojunction Organic Photovoltaic Cells ............................................... 69 3.3.3 O ptimizing Bulk Heterojunction Solar Cells ...................................................... 76 4 EXPERIMENTAL TECHNIQUES AND INSTRUMENTATION ........................................ 79 4.1 Introduction ....................................................................................................................... 79 4.2 Materials and Purification ................................................................................................. 79 4.3 Film Formation .................................................................................................................. 82 4.4 Characterization of Film Morphology and Thickness Using Atomic Force Microscopy ........................................................................................................................ 88 4.5 Optical Measurements on Films ....................................................................................... 91 4.6 Device Fabrication Techn iques and Construction .......................................................... 95 4.6.1 Glovebox ............................................................................................................... 95 4.6.2 Metal Vapor Deposition ....................................................................................... 97
7 4.6.3 Space Charge Limited Current Device Fabrication ........................................... 99 4.6.4 Photovoltaic Device Fabricat i on ........................................................................ 103 4.6.5 Details On Photovoltaic Device Patterns .......................................................... 106 4.7 Documentation of Polymer GPC and NMR Characterization ..................................... 107 5 CONDUCTIVITY AND MOBILITY ..................................................................................... 115 5.1 Material Properties and Carrier Mobility ...................................................................... 115 5.1.1 Intrinsic and Extrinsic Properties ....................................................................... 115 5.1.2 Challenges in Determining Charge Carrier Mobility in Conjugated Polymers .............................................................................................................. 118 5.2 Methods of Mobility Determination .............................................................................. 119 5.2.1 Time of Flight Measurements ............................................................................ 119 5.2.2 Modeling of FieldEffect Transistor Characteristics ........................................ 122 5.2.3 Field Dependent SpaceCharge -Limited Current Modeling ........................... 124 5.3 Mobility of Regioregular and Regiorandom Poly 3 hexyl Thiophene ....................... 127 5.3.1 Optical Properties of P3HT1, P3HT2, and P3HT3 ............................................. 129 5.3.2 SCLC Mobility Characterization of P3HT1, P3HT2, and P3HT3 .................... 131 5.3.3 SCLC Results ...................................................................................................... 133 5.4 Space Charge Limited Current Studies on Newly Developed Photovoltaic Polymers .......................................................................................................................... 141 5.4.1 PTVBT SCL C Diodes ........................................................................................ 142 5.4.2 3,6 PBTC SCLC Diodes .................................................................................... 1 45 5.4.3 P PtBTD Th SCLC Diodes ................................................................................ 148 5.5 Conclusions ..................................................................................................................... 149 6 BULK HETEROJUNCTION PHOTOVOLTAIC DEVICES ............................................... 151 6.1 Introduction ..................................................................................................................... 151 6.2 Performance Optimization of P3HT:PCBM BHJ Solar Cells ...................................... 151 6.2.1 Active Layer Film Composition ........................................................................ 152 6.2.2 Optimization of Active Layer Film Formation ................................................. 154 6.2.3 Optimization of Annealing Conditions ............................................................. 156 6.3 PTVBT:PCBM So lar Cells ............................................................................................ 162 6.4 PBTC:PCBM Solar Cells .............................................................................................. 168 7 CONCLUSIONS AND FUTURE WORK .............................................................................. 178 LIST OF REFERENCES ................................................................................................................. 180 BIOGRAPHICAL SKETCH ........................................................................................................... 189
8 LIST OF TABLES Table page 5 1 Mobility and field dependence extracted fr om SCLC diode characteristics. .................. 137
9 LIST OF FIGURES Figure page 2 1 Structure of Conju gated Polymers ....................................................................................... 18 2 2 Structural symmetry in a conjugated polymer backbone ................................................... 19 2 3 Repeat unit structur es for several different types of conjugated polymers ........................ 20 2 4 Electronic Energy Bands ....................................................................................................... 22 2 5 The two competing ground states (aromatic and quinoidal) of polypyrrole are not energetically equivalent. ........................................................................................................ 24 2 6 Different Linkages in poly(alkylthiophene where all configurations other than Head to Tail Head to Tail result in steric interactions between the alkyl groups ...................... 25 2 7 An illustration showing pi -stacking, the tendency of pi congjuated systems to form order struc -orbital overlap. ......................................................................... 26 2 8 An illustration relating electron transition energies to photon wavelengths for light propagating in free space ....................................................................................................... 27 2 9 Quasi -particle formation represented in segments of PPP and P A ..................................... 31 2 10 Metal insulator transition ...................................................................................................... 34 3 11 Doping of Silicon ................................................................................................................. 41 3 12 A p -n junction is formed when the doping is switched from p -type to ntype .................. 42 3 13 The density of holes in the p-doped region remain high until the depletion region where electron and hole combination leaves almost no free carriers ................................. 43 3 14 The variations in potential and energy in a p n junction. .................................................... 44 3 15 The effect of the bias voltage V on the electric potential of a p -n junction ....................... 46 3 16 Illustration showing the characteristic current -voltage behavior of a diode ...................... 49 3 17 I-V Characteristics of a photovoltaic cell ............................................................................. 51 3 18 Photovoltaic cell I -V behavior ............................................................................................... 54 3 19 AM0, AM1, and AM1.5 are differentiat ed by the amount of atmosphere that they pass through. ........................................................................................................................... 56 3 20 AM0 and AM1.5 are irradiance spectra ............................................................................... 57
10 3 21 Energy conversion from solar photons ............................................................................... 60 3 22 Normal ized AM1.5 photon flux spectrum ........................................................................... 61 3 23 Schemati c illustration of a two layer donor acceptor photovoltaic device similar to the one used by Tang. ............................................................................................................ 64 3 24 Energy level representation of a donor accep tor solar cell. ................................................ 65 3 24 Bulk heterojunction solar cell architecture ......................................................................... 70 3 25 Details of a bulk heterojunction organic solar cell ............................................................... 72 3 26 Folded cells can increase effective active area .................................................................... 78 4 1 ITO Patterns -designed and purchased from Kintec, Hong K ong ........................................ 81 4 2 Dust and film formation ......................................................................................................... 85 4 3 Spin Coater Chucks. .............................................................................................................. 86 4 4 Optical image displaying a cross -shaped scratch dual layer film (glass/PEDOT:PSS/ polymer) ................................................................................................................................. 89 4 5 Atomic force microscopy images used in calculati ng film thickness ................................. 90 4 6 Optical Characteri zation of Substrates used in this work .................................................. 94 4 7 Top schematic view of vacuum annealing chamber halves ............................................... 97 4 8 Evaporation mask and substrate holder designed for precise alignment of masks on substrates ................................................................................................................................ 98 4 9 SCLC Device Configuration ............................................................................................... 99 4 10 Schematic diagrams of aluminum evaporation holder used for creating substrates with regions of patterned contacts ...................................................................................... 101 4 11 SCLC shadow mask des ign ................................................................................................ 102 4 12 Photovoltaic device configuration ....................................................................................... 104 4 13 Photovoltaic shadow -mask design ..................................................................................... 105 4 14 Considerations in photovoltaic device design ................................................................... 106 4 15 Photovoltaic shadow -mask design ..................................................................................... 107
11 4 16 GPC analysis performed on the three different P3HT samples used in the mobility studies of Chapter 3 showing their measured molecular weights. .................................... 109 4 17 GPC analysis performed on PTVBT .................................................................................. 110 4 18 GPC analysis performed on PBTC .................................................................................... 110 4 19 H NMR spectrum of RR -P3HT1 ......................................................................................... 111 4 20 H NMR spectrum of RR -P3HT2 ......................................................................................... 112 4 21 H NMR spectrum of RRa -P3HT 3 ....................................................................................... 113 4 22 H NMR spectrum of PBTC ................................................................................................. 114 5 1 The schematic setup used for a TOF mobility measurement ........................................... 120 5 2 Simulated example of TOF data ......................................................................................... 121 5 3 Typical layout of a top contact transistor showing the dimensions of the channel following the typical convention ......................................................................................... 123 5 4 SCLC diode contacts ............................................................................................................ 125 5 5 Regioregularity in P3H T ...................................................................................................... 128 5 6 UV-Vis absorption spectra for thin P3HT films, where the increased regioregularity has an obvious impact on absorption specta. ...................................................................... 129 5 7 Band gap region of pure P3HT films on glass as well as optical band gap estimates of P3HT1, P3HT2, and P3HT3 ................................................................................................. 131 5 8 P3HT1 diode film absorpt ion spectra measured to verify AFM film thickness measurements ..................................................................................................................... 133 5 9 P3HT1 diode characteristics plotted in a form such that they should show linear behavior for regions in whi ch their current is space -charge-limited ................................ 134 5 10 P3HT2 diode characteristics plotted in the same manner, and on the same scale as the P3HT1 diodes in Figure 5.8 ................................................................................................. 135 5 11 P3HT3 diode characteristics plotted in the same manner, but on a different scale from those of P3HT1 and P3HT2 in Figures 5 9 and 5 10 .......................................................... 135 5 12 Extraction of the mobility value was made through applying a linear fit to the plot in the SCLC region, and using the relations in Eq. 5.7 to calculate the field dependence from the slope, and the zero-field mobility from the y intercept. ..................................... 136
12 5 13 Electric field dependence of the average mobilities before and after annealing The region from which the fit was made is indicated on the graph .......................................... 138 5 13 Atomic force microscopy height images of the polymer film surfaces of the diodes .... 141 5 14 Structures of polymers whose mobilities were investigated using SCLC ........................ 142 3 15 Absorption spectra of PTVBT diodes collected after device testing and used in conjunction with AFM measurements in deter mining film thicknesses. .......................... 143 5 16 PTVBT diode characteristics .............................................................................................. 144 5 17 Field dependence of the positive ca rriers in PTVBT as found through application of the SCLC model. .................................................................................................................. 145 5 18 Absorption properties of the 3,6 PBTC diodes ................................................................. 146 5 19 PBTC diode characteristics ................................................................................................. 147 3 20 The mobility of PBTC has strong field dependence, though the mobility values are relatively small. .................................................................................................................... 147 5 21 The field dependent SCLC mobility plots show slightly greater curvature than linear behavior .............................................................................................................................. 148 5 22 The field dependence of the mobility for P -PtBTD Th is positive, but the mobility remains low over a broad range of fields. ........................................................................... 149 6 1 The I -V characteristics for cells containing a varied amount of P3HT c ontent in the polymer:fullerene blend, under illumination with A.M. 1.5 simulated solar radiation. .. 153 6 2 Photovoltaic characteristics for cells containing a varied amount of P3HT con tent in the polymer:fullerene blend under simulated A.M. 1.5 solar radiation ............................ 154 6 3 Simulated A.M. 1.5 J -V behavior for a set of cells in which the blend film was composed of 55 % polymer, the active layers were spun at 700 RPM with 3 s ramps, and all devices were post -fabrication annealed at 150C for 30 min ................................. 155 6 4 The performance parameters for cells fabr icated under similar conditions, but containing variations in the spin rate and blend solution concentrations ......................... 156 6 5 Power conversion efficiencies showed improvements with anneali ng temperatures until peaking at ~168 C. ....................................................................................................... 157 6 6 Peformance characterics for cells annealed at various temperatures ................................ 158 6 7 Atomic Force Microscopy images showing the affect of annealing on P3Height image of cell before annealing ........................................................................................... 159
13 6 8 Optical response of pure films and blend films t o annealing. ........................................... 160 6 9 Current -voltage characteristics for cells illuminated under A.M. 1.5 conditions and post -fabrication annealed under different annealing condtions. ....................................... 161 6 10 External quantum efficiency of a P3HT:PCBM solar cell showing greater than 60 % incident photon conversion efficiency. ............................................................................... 162 6 11 Overlap of the absorption spectra of PTVBT and P3HT (arbitrary units) with the A.M. 1.5 solar photon flux spectrum. ................................................................................. 163 6 12 A.M. 1.5 illuminated J -V characteristics o f PTVBT:PCBM solar cells with varied amount of polymer content .................................................................................................. 164 6 11 Performance of PTVBT:PCBM solar cells ...................................................................... 165 6 11 A.M. 1.5 J -V characteristics of the best PTVBT whose active layer was composed of 10% polymer and 90% PCBM The power conversion efficiency was ~0.5 %. ................ 166 6 13 Tapping mode AFM images of PTVBT:PCBM blend film surfaces .............................. 168 6 14 Absorption profiles and structures of PCBM, 3,6 PBTC, and C60-PBTC. ...................... 169 6 15 Photovoltaic behavior of PBTC:PCBM solar cells ............................................................ 171 6 16 J -V characteristics of C60-PBTC:PCBM solar cells under A.M. 1.5 illumination. .......... 171 6 17 Tapping mode AFM morphology images of the surfaces of active layers in which PBTC makes up 22.5, 30, and 37.5 % of the blend .......................................................... 172 6 18 A.M. 1.5 illuminated J -V characteristics of C60PBTC:PBTC:PCBM solar cells in which the polymer fraction composes 30 wt. % of the active layer blend and PCBM composes 70 wt. %. .............................................................................................................. 173 6 19 The effect of differing weight percentages of C60PBTC in the active layers of C60PBTC:PBTC:PCBM solar cells ........................................................................................... 173 6 20 Tapping mode AFM scans of the sur face of C60-PBTC:pbtc:PCBM solar cells in which the polymers compose 30 wt. % of the blends and PCBM composes 70%. ........ 174 6 21 A.M. 1.5 illuminated J -V characteristics of (C60-PBTC):P3HT:PCBM solar cells ....... 175 6 22 A.M. 1.5 illuminated J -V characteristics of :PCBM solar cells showing the similar effect that adding PBTC has on the performance compared to the addit ion of C60PBTC. .................................................................................................................................... 175 6 20 Tapping mode AFM scans of the surface of P3HT:PCBM solar cells in which the additives have been incorporated into the blend ................................................................ 176
14 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CONJUGATED POLYMERS IN BULK HETEROJUNCTION PHOTOVOLTAIC DEVICES By Nathan C. Heston August 2009 Chair: David B. Tanner Major: Physics The main focus of this work has been the incorporation of conjugated polymers into photovoltaic devices. This application necessitates an understanding of the underlying optical an d electronic properties of conjugated polymers as well as morphological behavior in polymer fullerene blends. Optical properties of the polymers were studied through ultraviolet to near infrared (UV NIR) spectroscopy. Space charge -limited devices were f abricated and characterized to probe the transport properties of different pi -conjugated systems Bulk heterojunction photovoltaic cells were made utilizing new polymer s and fabrication techniques in attempts to optimize device performance and elucidate t he photovoltaic process in polymer based cells. Atomic force microscopy studies were made to examine the relationship between blend -film morphologies and photovoltaic performance. Results of these experiments are presented in this work along with suggest ions for future studies.
15 CHAPTER 1 DIS S ERTATION SUMMARY This work investigated the transport properties of neutral pi -conjugated polymers and the incorporation of these materials into bulk heterojunction photovoltaic devices. This c hapter outlines the l ayout of this dissertation. Chapter 2 will give an overview of s ome of the fundamental pr operties of conjugated polymers, especially those which are related to photovoltaic cells. Chapter 3 develop s the principals of photovoltaic cells, and gives a brief outline of the use of p olymers in the field. Chapter 4 describes the experimental methods and techniques employed in the laboratory, making specific references to materials, techniques, and instrumentation used while completing the work in the Materials Chemistry Characterization Laboratory (MCCL). Chapter 5 presents the results of experiments which sought to investigate the bulk electronic properties of conjugated polymers through examination of the current -voltage characteristics of polymer diodes. T o extract the mobility the I -V behavior was fitted to the field -dependent mobility model discussed in Section 2 .3. Further research was dedicated to optimization of photovoltaic cells through the examination of photovoltaic behavior with variations in fil m thickness, film roughness, donor acceptor blend ratios, and processing additives. It also explored new materials for solar cells, alternative methods for film deposition. The results of these findings are presented in Chapter 6 Chapter 7 discusses th e results of this work and suggestions for future experiments.
16 CHAPTER 2 CONJUGATED POLYMERS: GENERAL PRINCIPALS A ND APPLICATIONS 2.1 Introduction From its surprisingly recent start, around thirty years ago, the study of conjugated polymers has developed into a broad field. Arguably, it was the discovery that doping could increase the conductivity of polyacetylene by nearly 10 orders of magnitude that provided the initial impetus to the growth of the field. This finding was made in the late 1970s by Shi rakawa, MacDiarmid, and Heeger.13 Since then, progress in the field has been made at an ever quickening pace. The numb er of published scholarly articles on conjugated polymers has approximately doubled in each five year period since this initial work. Though much of the early research on conjugated polymers focused around the development of doped polymer systems for use as lightweight and flexible conductors, a large portion of current research seeks to utilize their semiconducting nature. Many properties of conjugated polymers make them well -suited for applications in display technologies4, field effect transistors5, capacitors6, light emitting devices7, electrochromic windows8, and photovoltaic cells9, among others. The abundant worldwide use of plastics makes many attractive features of polymers apparent, such as high strength to weight ratio, biocompatibility, and excellent low -temperature, low -cost processibility.10 Many of these attributes are shared by conjugated polymers. A key feature, important for electronic applications, is that the optical, electrical, and electrochemical properties of conjugated polymers can be tuned through modification of the polymer structure. This capability provides a unique pathway towards adapting conjugated polymers for specific electronic functions.
17 The large variety of ways that conjugated polymers can be constructed and modified has led to a vastly diverse class of polymer s Though no book or publication can completely describe this rapidly evolving field, there are many excellent sources of information available.11 2 .2 F undamental Properties of Conjugated Polymers The defining feature of a con jugated polymer is the presence of alternating single and double bonds along the polymer backbone as is illustrated in 2 1 A Each carbon atom along the polymer chain has valence e lectrons in one 2s and three 2p (px, py, and pz) atomic orbitals When a c arbon atom is bonded to three other atoms the 2s atomic orbital combines with two of the 2p atomic o rbitals resulting in the formation of three equivalent sp2 hybridized orbitals. These three hybri d orbitals lie at approximately 120 angles in a plane (though the angles can be influenced by neighbori ng atoms) T he remaining 2p orbital, which is nominally taken as the pz orbital, is perpendicular to this plane.12 Overlap of the sp2 hybridized orbitals along the carbon chain Electrons filling the energetically favorable bonding orbitals are responsible for the strong sigma bonds between the atoms of the polymer backbone This is illustrated in F igu re 2 1 B. The remaining pz orbitals between consecutive atoms overlap to a lesser degree resulting in the formation of weaker bonding orbitals orbitals Electrons fill ing these bonding orbitals are responsible for the formation -bonds a long the chain interactions between chains. 2 1 C illustrates -bonding along a polymer chain. Examination of the valence electron configuration of carbon atoms along the backbone of a conjugated polymer chain shows, that aft er the formation of the three sigma bonds, each carbon -electron. Based on a symmetric model of the polymer chain, as illustrated in Figure 2 2 A, these electrons would be in a quasi -one dimensional periodic
18 A B C Figure 2 1. Structure of Conjugated Polymers. A) Illustration of bond alternation in polyacetylene. B) Overlapping sp2 hybridized orbitals where the red regions indicate regions of sp2 orbital overlap and the formation of in -plane sigma bonds. Note that the pz orbitals are not shown here. C) Weak overlap of pz orbitals forming pi bonds. lattice with lattice constant a Such a crystal, with the carbon atoms occupying the positions in the Bravais lattice would exhibit a periodic potent ial and have Bloch solutions w ith energies given by (k) = 2k2/2m .13 If each carbon atom of the lattice contributed one electron the band would be half filled with a Fermi surface at The electron density of states would be continuous, thus allowing for electron delocalization along the polymer chain.14 However, it turns out that such a one dimensional lattice is unstable and subject to a structural distortion, known as a Peierls Transition, after physicist Rudolf Peierls who disco vered it in the 1930s when writing a section on one dimensional models.15 Instead of maintaining equal spacing along the polymer backbone, the carbon atoms rearrange slightly by
19 forming pairs such that the spacing between them alternates as illustrated in Figure 2 .2 B. This restructuring of the pol ymer chain i s known as dimerization. A B Figure 2 2 Structural symmetry in a conjugated polymer backbone. A) I llustration of the structurally symmetric conjugation of the unstable metallic state in which atoms are equally spaced. This would lead to a continuous density of states B) The configuration of the symmetric state is subject to a Peierls distortion where the atoms of the backbone dimerize. This renders the material a semiconductor. Because of the symmetry of the sigma bonds there is a n elastic energy increase associated with this departure from equally spaced carbon atoms Therefore, in order for dimerization to occur -electrons. The result of dimerization on the ene electrons is a reduction in occupied -energy states near the Fermi surface and an increase in the unoccupied -energy states above the Fermi surface.15 The competing driving forces for the distortion reach an equilibrium state whose result is bondlength modulation along the polymer backbone. In summary the overall energy of the polymer configuration is lowered by the Peierls distortion. This distortion results in formation of an --orbitals. -bonding orbitals in pairs between alternating pairs of carbon atoms producing a lternating single and double bonds (bond c onjugation) along the polymer chain as is illustrated in Figure 2 2 B.
20 There are many ways in which the alternating single and double bond structure of a conjugated polymer backbone can be attained. Thus, there are a huge variety of conjugated polymers, some of which are illustrated in Figur e 2-3. The polymer chains are constructed from a buildup of repeat base units or monomers. There is no precise guideline for determining the number of monomer units needed to make a polymer. Obviously, th e chains are not infinite, and in some cases less then 10 monomer units result in the saturation of specific properties such as absorption spectra.16 However, this number varies from one material to another, and properties Figure 2-3. Repeat unit structur es for several different types of conjugated polymers. PA = polyacetylene, PT = polythiophene, PPP = poly(p-phenylene), PPy = polypyrrole, PheDOT = polyphenylenedioxyt hiophene, PEDOT = poly(3,4ethylenedioxythiophene), PPV = pol y(p-phenylenevinylene), PBTD = polybenzothiadiazol, PTQ = polythiadiazolquinoxaline. are often dependent upon the conditions in which they are measured. It can also be the case that different physical properties will saturate at different conjuga tion lengths. Thus, the dividing line between an oligomer and a polymer is by no means sharp.
21 -orbitals in monomer units can be described in terms of the highest occupied molecular orbital (often referred to as the HOMO level) and the lowest unoccupied molecular orbital (referred to as the LUMO level). As monomer units are linked together into t he formation of oligomers -orbitals results in a splitting of the energy levels. As more and more monomer units are joined together the allowed energy levels become clo ser and closer Eventually as a polymer is formed there is a near continuum of filled energy levels and another of unfilled energy levels. This process is illustrated in a simplified form in Figure 2 4 A Of course the density of states in a real pol ymer is more complex, but the end result is the same. A more detailed and accurate account of the buildup of energy levels from monomer units into the polymers polyacetylene, polyborole, polycyclopentadiene, polypyrrole, polyfuran, polysilole, polyphosphole, polythiophene, polyselenophene and polytellurophene can be found in reference 17.) Adopting the common semi conductor language of physics the set of bands formed from the filled HOMO levels are known as the valence band a nd those that are formed from the unfilled LUMO levels are known as the cond uction band. Figure 2 4 B illustrates the basic differences between metals, semiconductors, and insulators. In the case of metals, there is no difference in energy between the top of the filled valence band and the empty levels of the conduction band. For this reason, electrons in a metal are likely to found the conduction band. At T = 0 some materials have a band gap between the valence and conduction bands. This defines the material classes of semiconductors and insulators. The difference between th ese materials is the size of the band gap. The border between semiconducting and insulating materials is somewhat arbitrarily defined by the magnitude of the band gap and the value typically taken to be around 3 eV. As the size of the
22 band gap is increas ed, thermal population of the conduction band decreases and eventually becomes negligible. Some semiconductors have small enough band gaps that they allow for a small number of electrons to be excited at room temperature. These materials are referred to as intrinsic semiconductors. A B Figure 2 4 Electronic Energy Bands. A) Simplified diagram illustratin g the buildup of energy bands in a conjugated polymer. B) Band gaps of metals insulators and conductors. This energy gap defines a notable difference between metals and semicon ductors. In metals, due to decreased scattering interactions between the lattice and conduction electrons, conductivity typically increases as the temperature is decreased As the temperature is decreased in semicondu ctors thermal excitation of electron s into the conduction band decrease s which resul ts in decreased conductivity with temperature. Also to be noted, is that both electrons and holes can act as charge carriers in a semi conductors. Electrons travel throu gh the conduction band, whereas, holes travel through the valence band. This results in different transport dynamics. Hall measurements in most metals show that the dominant carriers are electrons. In t he case of semiconductors, hole transport can signi ficantly contribute to the overall conductivity of the material Additionally, Hall measurements on intrinsic semiconductors become more complicated by the presence of both
23 holes and electrons that can build up in different concentrations and m ove with different mobilities. The population of thermal carriers in the conduction band is dependent upon the Fermi Dirac distribution function: T k E fB g FD) ( exp 1 1 (2 1 ) At room temperature, the value of kbT is ~0.025 eV which is much smaller than the b and gap of most semiconductors and the population of carriers that are thermally excited from the valence to the conduction band is exceedingly low. Semiconductors can also be doped by introducing impurities that provide states inside the band gap. Add ing impurities in the band gap by doping with materials that introduce filled states is referred to as n -doping, and doing so with materials that introduce empty states is referred to as p -doping.18 Another method by which charge carriers can be introduced in semiconducting ma terials is by absorption of electromagnetic ra diation. Absorption of photons with energies greater than or equal to the semiconductor band gap excite s electrons into the conduction band. Sometimes this is called optical -doping or photo -doping The effec t, known as photoconductivity, is the result of increased conductivity of a material when exposed to radiation. Typically this radiation corresponds to ultraviolet, visible, or near infrared light. The classical example is the case of polyvinylcarbazole which is used in the photocopying process of xerography machines As has already been mentioned, the e ffect of the structural distortion caused by the Peierls transition is the opening of a band gap in conjugated polymers. In their neutral state most co njugated polymers fall into the category of semiconductors. Th r ough the use of dopants conjugated polymers can reach high levels of conductivity approaching the conductivity of
24 metals. Because of this, conjugated polymers ar e often referred to as conducting polymers, though this is a bit of a misnomer. There are many factors in addition to the Peierl s distortion that affect the magnitude of the band gap in conjugated polymers. One such factor of importance is that almost all conjugated polymers have nondegenerate ground states. Th is means that if th e double and single bonds along the conjugated backbone were to be ex changed, the resulting ground state energy configuration would not be equal to the initial ground state. This is illustrated in Figure 2-5 for the case of polypyrrole, where th e quinoidal form has the higher ground-state energy. The aromatic ground state is more energetically favorable, hence, the double bonds of the polymer are stabilized in the aromatic form. This decreases the amount of el ectron delocalization, and increases the polymer band gap. There are ma ny different ways to minimize the energy difference between the two ground states. The addition of electron-w ithdrawing and electrondonating groups can be added to th e structures in ways to stabili ze one ground state form relative to another. N H PPyn N H PPyn Aromatic Quinoidal Figure 2-5. The two competing ground states (a romatic and quinoidal) of polypyrrole are not energetically equivalent. The degree of -orbital overlap, and hence, electr on delocalization is dependent on backbone planarity. Steric inte ractions between different segm ents of a polymer chain can reduce planarity and result in an increased band gap. This is evid ent in the case of polythiophene where the regioregularity can be directly linked to the polymer ba nd gap. The addition of alkyl
25 chains onto polythiophene is what leads to soluble forms of the polymer in organic solvents. These alkyl chains can be added in different pos itions on the repeat unit. A poly(alkylthiophene) with a high degree of regioregularity has the alky l chains symmetrically oriented in a head-totail, head-to-tail fashion, as is shown in Figure 2-6 A. A regi orandom version will have these linkages oriented in a random fashi on containing different combinati ons of the linkages. This is shown in Figure2-6 B,C, and D. Due to interactions between the alkyl gr oups in the cases of TTHT, HT-HH, and TT-HH linkages, there is a loss of planarity and a reduction in -orbital overlap. As an example of the effect on the polymer band gap, regioregular poly(3butylthiophene) has a band gap of 1.7 eV while re giorandom poly(3-butylthiophene) has a of 2.1 eV.19 A B C S S S (HT-HH)R R R D Figure 2-6. Different Linkages in poly(alkylthiophene where all c onfigurations other than Head to Tail Head to Tail result in steric inte ractions between the al kyl groups. A) Head to Tail Head to Tail configur ation. B) Tail to Tail Head to Tail configuration. A) Head to Tail Head to Head configuration. B) Tail to Ta il Head to Head configuration. Interchain interactions can al so affect the polymer band gap. Along a planar conjugated polymer backbone, the -orbitals are oriented in an ort hogonal position relative to the chain. Therefore, -orbitals of different chains can overlap in a stacking fashion. The overlap produces
26 an interchain-attraction (pi-bonding) that tends to cause diffe rent chains to st ack on top of each other in a regular fashion known as pi-stacking. Pi-stacking is illustrated in Figure 2-7. In addition to increasing structural order, pi-stacking signi ficantly decreases the separation distance between different pi-conjugated systems, resulting in increased electron delocalization, decreased band gaps, and increase d interchain electron transfer.20 Figure 2-7. An illustration showing pi-stacking, the tendency of pi-congjuated systems to form order structures based on -orbital overlap. The band gap size plays an important role in optoelectronic applications because electron excitations from -orbitals to *-orbitals (often referred to simply as transitions) often correspond to photon energies in the visible portion of the elec tromagnetic spectrum. The energy of a photon propagating in free space is given by E = h where h is Planks constant and is the photons frequency. An electron abso rbing a photon will make an energetic transition equal to the photons energy. Figur e 2-8 shows a plot of electr on transition energies (in eV) versus the corresponding photon wavelengths (in nanometers). Visible light corresponds to el ectron transitions from about approximately 1.75 to 3.1 eV. The band gap of a conjugated polymer determines the minimum photon energy that is able to excite a pi-electron. Photons with energy greate r than or near to that of the band gap often correspond to transitions. For this reason, conjugated polymers are often inte nsely colored. Though the absorption spectrum of a polymer will de pend on its density of electronic states, as
27 well as other factors, most conjugated polymers tend to be transparent to photons of energy less than the band gap (longer wavelengths). Figure 2-8. An illustration relati ng electron transition energies to photon wavelengths for light propagating in free space. If a semiconducto rs band gap is greater than the energy of a photon of given wavelength of light, then the material will not absorb the light. Conjugated polymers can be doped through the charge-balancing cations or anions. They differ from other semiconductors in that this do ping process can be done in a reversible fashion through electrochemical doping. The process of doping a conjugated polymer introduces electronic states inside the band gap of the polymer. The energy st ates are the result of relatively strong electron-phonon coupling in which the lattice di storts in the presence of a charge. This stabilizes and lowers the energy state. These di stortions are discussed in more detail in the following section. The intermediate states (polarons and bipol arons) change the absorption spectrum of the original neutral polymer and allo w for absorption at longer wavele ngths. This is the basis for electrochromic applications of conjugated pol ymers in which the polymers spectrum can be modulated by applying of an external voltage.
28 2 .3 Transport and Conductivity In cla ssical semiconductors the lattice is composed of tightly bound atoms in a crystal. The electrons and holes in these systems move through the conduction and valence bands without having a large impact on the local electronic and geometric states. This is evident from the typical exciton binding energies (~0. 0 1 e.V.) in traditional semiconductors.21 In pi conjugated systems the story is much different and typical exciton binding energies are much higher (~0.5eV) .22 The presence of an electron, hole, or excitation can have a significant impact on the local electronic and geometric structure of the underlying lattice, i.e. the polymer .23 Conjugated polymer systems are further complicated by the fact that bulk samp les are composed of multiple chains that may be in semi -ordered or highly disordered arrangements. Interchain interactions and morphology can therefore play significant role s in the charge transport properties of a sample. A detailed understanding of the underlying processes is needed in order to understand the optoelectronic processes such as exciton formation and dissociation as well as charge carrier mobilities. There have been significant advances in the understanding of transport over the past thi rty years of research, yet each polymer sample is somewhat different In addition, the extraction of general parameters such as m obility i s not always a straight forward or easy task. There are a number of central issues that should be addressed in the di scussion of charge carriers in conjugated polymers. Neutral conjugated polymers have no free carriers. Charges must be added by optical excitation, electron transfer, or through chemical doping. One of the key points to be made is that an excitation or the addition of a charged species to a conjugated polymer will result in a disturbance of the local electrical and geometric structure T here will be a mutual influence of the structure on the charged species and the charged species on the structure. Th is coupling of the structure to the charge carriers is often referred to as electron -
29 lattice coupling All dynamic process es in conjugated polymers will therefore involve the motion of not only the charge carriers, but also the atoms of the lattice.24 To retain the simple conceptual picture of free carriers in valence and conduction bands in conjugated po lymer syste ms, one must suppose the bands to bend near the presence of charge d or excited species. In this sense, conjugated polymer lattices tend to be electrically soft. Thus, charge carriers in conjugated polymers should not simply be regarded as electrons or holes, but as a coupled local disturbance and a charge or excitation Originally it was thought that the addition or removal of a n electron from a conjugated polymer chain would simply create a charged species in the bottom of the conduction band or the to p of the valence band. The contrary discovery, that it is energetically favorable to localize the charge on the chain and have the charge surrounded by a structural relaxation of the lattice resulted in theoretical attempts to account for the electron l attice coupling through the development of the concept of quasi -particle carriers, representing the lowest energy eigenstates of the coupled-electron (hole) lattice system.2527 The term polaron is used to describe the quasi particle state of a charge -lattice coupling. Further introduction of charged species on the polymer chai n can lead to the formation of bound polaron pairs (bipolarons) if the gained energy from the relaxation of the lattice is greater than the Coulomb repulsion between the two confined charges. Formation of bipolarons is supported by electron spin r esonance (ESR) measurements .25 Since a polaron contains an unpaired charge it will have an associated spin. As doping levels are increased an increase is seen in spin resonance signals until with higher levels of doping this sign al decreases as polarons form spinless bipolarons. Figure 2 9 illustrates the polarons and bipolarons in segments of polyparaphenylene.
30 Degenerate ground state polymers, such as trans -polyacetylene, form a special case because of the equivalence of their ground state energies. A result from the degeneracy is that would -be polaron pairs on the polymer chain separate because there is no increase in distortional energy so that bipolarons do no t form. As shown in Figure 2 9 E -F the presence of an extra char ge or an unpaired electron will represent the region between two different phases of bonding. The region represents a domain wall and the associated quasi particle is called a soliton. Because the energetic configurations of these phases are equivalent there is very little binding energy for the soliton and it is delocalized along the chain. The term soliton is used because the particle has properties of a solitary wave that can propagate without deformation or dissipation.25 Prop agation of a soliton along the polymer cha in results in the movement of the domain wall, that is, the separation between the two phases. Therefore, the movement of a soliton corresponds to conversion of the chain from one phase to another. The existence of polarons, bipolarons, and solitons become s particularly relevant in the description of optical transitions. Associated with each of these species are localized allowed energy levels within the band gap.23 The presence of a soliton creates a mid -gap state that can be occupied by zero, one or two electrons. The presence of a polaron results in two states symmetrically placed within the band gap region. In the formation of a bipolaron the states are shifted deeper toward the center of the band gap Each of these states can also be em pty or occupied by one or two electrons. The symmetrically allowed optical transitions to and from these states are indicated in red in Figure 2 9 These transitions become evident in absorption spectra by the emergence of absorption peaks occurring with in the band gap region with increased amounts of doping.
31 Figure 2 9 Quasi particle formation represented in segments of PPP and PA The domain walls are drawn unrealistically sharp for illustration purposes. Red arrows indicate symmetry permitted tra nsitions. A) Positive polaron B)Negative polaron C) Positive Bipolaron D) Negative bipolaron E) Neutral soliton F) Positive soliton G) Negative soliton
32 The term self localization is often used to describe the quasi -particles because the presence of the charge produces a local geometric relaxation, which must be overcome in order for the species to move. It should also be noted that though the transitions from one bonding phase to the next are illustrated as sharp differences in Figure 2 9, that cal culations show that they are often spread out over many repeat units. Despite the associated lattice defects, polarons and bipolarons are relatively mobile along polymer chains, and with sufficient pi -overlap between chains, the charge carriers can propag ate with fairly high mobilities. In the case of chemical doping, however, the dopant counter ions are much less mobile. Charged species in the polymer tend to remain near these counter ions, and, for this reason, high levels of doping are needed to creat e high conductivities. As applications began to emphasize the use of neutral (undoped) materials for thin film transistor and photovoltaic applications, focus shifted from characterization of conductivity to understanding charge -carrier mobility in polymer films. Though conductivity is a bulk property of a sample, charge mobility is an intrinsic property defin ed by the relationship between drift velocity of the charge carrier and the electric field through = v/E where E represents the electric field and is the drift velocity For many electronic applications of conjugated polymers knowledge of the carrier mobility is key to impr ovement of device performance. In light emitting diodes a balanced injection of charges is necessary for efficient devices so that each injected electron can combine with a hole for radiative emission. If mobilities are not balanced the recombination can occur at the electrodes instead of in the emissive layer. In the case of photovoltaic applications high carrier mobility is necessary to enable efficient charge collection and eliminate electron -hole recombination. High carrier mobility parallel to the substrate is key to increasing drive current as applications demand smaller channel lengths .28
33 In thin film applications etc., the neutral semiconducting polymer lay er is sandwiched between metal electrodes This is illustrated in Figure 2 10. Before electrical contact with the metals the Fermi level of the neutral semi conductor should be at mid -gap.29 The Fermi level of the metal is very near the metal work function. Figure 2 -10 B illustrates the formation of two potential barriers after the materials are brought i nto contact and the Fermi levels align. These barriers arise as the charges near the interface rearrange and result in band bending in the semiconductor. The potential barrier seen by electrons moving from the metal into the conduction band of the semic onductor is known as a Schottky barrier and is given by, ) ( m bn (2 2 ) where n represents the work function of the metal and represents the electron affinity of the semiconductor. A second barrier will be seen by electrons moving from the semiconductor to the metal which is given by, ) ( e E E VF c bn bi (2 .3 ) where Ec represents the conduction band energy of the semiconductor and EF is the Fermi level. However, if the work function of the semiconductor is greater than the work function of the metal, then an ohmic contact can be formed where the charges see essentially no barriers at the interface. (A more thorough treatment of the Schottky barriers can be found in reference 30.) In a m etal -semiconductor -metal configuration, the dominant carriers are injected from the contacts, because thermal charge carriers are generally negligible. If the injection barriers are suitably small, voltage applied across the film will result in current li mited by injected space charge (space charge is charge that is thermo ionically injected from the metal contacts in the
34 A B Figure 2 10. Metal -insulator transition. A) Metal -insulator transition before electrical contact is made. B) Metal insul ator transition after electrical contact is made where an electron injection barrier has formed. presence of a strong electric field, as in the case of vacuum diodes) which leads to a quadratic dependence upon the voltage 3 2 089 L V Jr (2 .4) w here J is the current density, represents the carrier mobility, represents the relative dielectric, V the voltage across the layer, and L the layer thickness. 30 This nonlinear curre nt behavior known as Motts Law, can be c omplicated by the presence of traps in the semiconducting material and by the existence of field dependence in the mobility. Analytical treatments are therefore not always possible. Only in cases where there is a simple dependence of the mobility on the electric field (such as in the Pool -Frenkel model where there is an exponential behavior of the mobility on the square root of the electric field) do approximate analytical solutions exist that allow for extraction of the mobility from I -V behavior.31
35 The theory of space -charge limited curr ent was extended by Murgatoyd to account for a set of shallow traps .32 In disordered materials with the mobility has b een found to have an exponential dependence on the electric field of the form, ) exp(0E (2 .5 ) where 0 represents the zero -field mobility, and represents the field dependence parameter .32 The current density can be expressed in terms of the carrier density p(x) the mobility (E) and electric field E (x), as J = p(x)e (E(x) ). The Poiss on equation is ) ( ) (x p dx x dE e (2 .6 ) Combing these equations yields an expression for the current density: EdE E ) exp( Jdx0 (2 .7 ) Since the current can be assumed to be independent of x (for area >> thickness) this yields, LE OEdE E J ) exp( L0 (2 .8 ) T he voltage across the film is given by, L OEdx V (2 .9 ) S ubstituting dx from above gives LE OdE E E J2 0) exp( V (2 .10) Both 2 .7 and 2 .9 have the form of an incomplete gamma function that can be expanded. Keeping only the first terms returns quadratic dependence of Eq. 2 .3. The expansion has the approximate solution
36 3 2 0) 89 exp(. 8 9 L V E Jr (2 11) For a material containing a series of shallow traps at a depth of energy A below the conduction band the net effect will be to reduce the amount of free charge carriers. The proportion of total charges that are free will then be ) exp(0T k A N NB t c t f f (2 .1 2 ) where f represents the density of free charge carriers, t represents the density of trapped carriers, Nc denostes the density of states in the conduction band and Nt is the density of traps. Frenkel pointed out in 1954 that the effective trap depth can be dec reased in the presence of a strong field.33 T his is now know n as the Frenkel effect. If t he Frenkel effect is used in conjuction with Eq. 2 .12 a numerical solution can be obtained with the form, 2 / 1 0 3 3 2 0 0891 exp() exp( 8 9 L V e T k T kA N N L V JB B t c r (2 .13) This is more commonly written as 3 2 0 0 0) 891 exp(. 8 9 L V E Jr (2 .14 ) It ha s been experimentally found that transport through a metal -conjugated polymer -metal diode exhibits this form for sufficiently low injection barriers.3437 It has also been pointed out by several other publications that this expression neglects the density depend ence of charge 38,39 Nevertheless, it is the generally accepted modeling method for polymer diodes.
37 CHAPTER 3 EXPERIMENTAL TECHNIQ UES AND INSTRUMENTATION 3.1 Introduction A very notable application of semiconductors is their use as the photoactive material in solar cells. The photovoltaic process involves the conversion of electromagnetic radiation into an electric current and corre sponding voltage ; in other words, it is the conversion of electromagnetic energy into electrical energy. It was first observed in 1839 by the French physicist Alexandre Edmond Becquerel, when at the age of 19, he was conducting experiments with silver c hloride -coated platinum electrodes and observed the generation of current upon exposure to sunlight.40,41 (In stricter terms, Becquerel observed a photo-electrochemical effect, but he is generally ascribed with the discovery of photovoltaics.) It was not until 1954, that the first practical photov oltaic devices were made out of silicon at Bell Telephone Laboratories, achieving efficiencies of around 6% and leading to commercial use of solar panels.42 During the following decades t echnological improvements led to increased performance efficiencies, but progress in commercial solar cells has lagged behind the rapid growth of the transistor, the true offspring of the semiconductor revolution. This in part must be considered due to the low -cost of fossil fuels. As evidence of global climate change continues to mount it has become more an d more apparent that non -carb on-emitting sources of energy are needed. In addition to increased interest in environmentally friendly energy sources, recent trends in energy cost s suggest that crystalline silicon -based photovoltaic technologies will be competitive with traditional gri d power prices within the next decade, and that new thin-film solar cell technologies (with lower efficiency and larger area needs) are currently able to compete within this price range.43
38 One of the obstacles to overcome in reducing the price of photovoltaic energy is the cost of silicon which accounts for about 40% of the final manufacturing cost of crystalline silicon cells.43 Further efforts to reduce the price of solar energy have resulted in exploration of lower cost materials and device fabrication possibilities As the price of fossil fuels continue to climb it seems inevitable that solar energy generation will no longer be limited by cost. Comparing renewable energy resources, the potential for solar energy far outweighs other carbon -free sources. According to estimates of global energy use (averaging about 13 TW), more solar energy strikes the earth in one hour than is currently consumed in one year by the entire planet.44 The difficulty, of course, is in finding ways to harness this energy in a cost effec tive way. Conjugated polymers are excel lent candidates for low cost photovoltaic cells. In contrast to the high temperature processing required for silicon photovoltaic cells, polymer -based cells can be processed at low temperatures through solutionbased methods. The development of polymer -b ased cells also opens up the possibility of lightweight flexible solar cell arrays which could be easily transported. Current power conversion ef ficiencies for the best polymer -based cells are still low (~ 6.5%)45 compared to typical commercial crystalline silicon cells (12 18%), but there are many facets of research being explored for conjugated polymer -based cells that wil l likely close this gap. Though it is difficult to estimate the cost of up-scaling polymer -based cells, the widespread use of plastics in the modern world is indicative of the low ma nufacturing costs of polymers. 3.2 Fundamentals of Photovoltaic Cells Pho tovoltaic cells utilize the fundamental electronic properties of semiconductors. The operation of traditional silicon photovoltaic cells is based upon the physics of the p-n junction,
39 which is also known a p -n heterojunction. Formation of p-n junctions results when the doping profile of a semiconductor is v aried from p -doped to ndoped. It is instructive to begin by considering doping i n silicon, the most widely used and the prototypical semiconductor. Silicon has four valence electrons. In its pure cr ystalline state each of these four electrons participates in four covalent bonds to their nearest neighbors in the crystal lattice. This is illustrated schematically in Figure 3 11 A. These electrons fill the valence band of the crystal which is separa ted from the conduction band by a band gap of about 1.1eV. 3.2.1 Doping The crystal lattice of silicon can be n -doped with phosphorous (or a similar group V element) which has five valence electrons. This process is illustrated schematically in Figure 3 11 B. The first four valence electrons participate in the bonding structure of the crystal. The energy level of the extra valence electron from the dopant species occurs very near in energy to the empty conduction band. At room temperature the energy difference is typically on the order of kBT and some of the extra electrons can easily be converted into free carriers by excitation into the conduction band.13 Because the gap between the impurity band and the conduction band is much less than the band gap, the density of electrons from the impurities is many orders of magnitude greater than the density of electron-hole pairs produced from thermal excitation of electrons from the valence band. For this reason, the carriers don ated from the impurities are known as majority carriers and those excited from the valence band are known as minority carriers. The net effect of adding the n dopant impurities is the addition of negative carriers into the conduction ban d of the crystal. It is important to keep in mind that though each dopant atom adds an extra valence electron to the crystal, the dopant species is adding the same number of protons as electrons, and so,
40 regardless of the doping level, the crystal remains electrically neut ral. When the extra valence electrons leave the dopant species, they are leaving behind an ionized atom. This atom remains bonded in the crystal as a fixed positive charge. An n -doped crysta l can be visualized as a sea of electrons moving freely in a la ttice filled with fixed positive ions whose net charges cancel each other. The crystal of silicon can also be p doped, which is done in a similar fashion to n-doping. In p -doping, which is shown in Figure 3 11 C, some of the lattice sites of the crystal a re filled by adding gallium (or a similar group III element). Group III elements have 3 valence electrons which will contribute to the crystal bonds and lie in the valence band. The presence of the dopant species also contributes an unfilled energy level wh ich is just above that of the valence band. The small energy difference between the valence band and the dopant band means that at a finite temperature, some electrons will fill the empty dopant bands leaving electron holes behind in the valence band. The holes can then become delocalized from the dopant species and propagate freely through the crystal. In p -doped silicon, the holes resulting from the dopant species are the majority carriers and their concentration is many orders of magnitude greater than the minority carriers resulting from thermal excitation of electrons from the valence to the conduction bands. P -doped crystals are also electrically neutral. The delocalized holes leave behind a net negative charge on the dopant species, which remai ns bonded in the crystal as a fixed negative charge. A pdoped crystal can be visualized as a sea of free holes moving freely in a lattice filled with fixed negative ions. Once again the net charges cancel each other.
41 A B C Figure 3 11. Dop ing of Silicon A) Undoped silicon with empty band gap ~1.1 eV. B) n -doped silicon, where an extra electron appears in the lattice, producing a new filled energy band very near in energy to the conduction band. Electrons from this impurity are easily tra nsferred to the conduction band and delocalized throughout the crystal. C) p doped silicon, where a hole is present near the impurity creating an unfilled energy state near in energy to the conduction band. Electrons from neighboring atoms can fill this vacant state allowing the hole to become delocalized and move through the crystal. 3.2.2 P -N Junctions A p -n junction is formed where the doping is abruptly switched from p type to ntype (such that the transition width << depletion region width). Near to the transition (taken as x = 0 ) free electrons from the n type side will move to fill the holes from the p type side. Their combination results in a region that is deficient of majority charge carriers and i s known as a
42 depletion region (a lso, someti mes referred to as a space-charge region because of the extremely low concentration of free charge carriers and the high electric fields that can be present). As was just described in the previous paragraphs movement of the delocalized carriers leaves beh ind fixed negative ions in the lattice of the p -type region and fixed positive ions in the lattice of the n type region. Because the majority carriers combine in the depletion region, within this region they no longer balance the fixed charges in the crys tal. Thus, on the p -type side of the junction, fixed negative charge begin to accumulate and in the n -type region fixed positive charges build up. As more and more charges build up, an electric field grows, which points from the positive charges on the n -type side of junction towards the negative charges on the p type side. The growing electric field acts to oppose further charges from diffusing and eventually an equilibrium is reached. The charge distribution is illustrated in Figure 3 12, and the ca rrier concentration is shown in Figure 3 13. Figure 3 12. A p -n -junction is formed when the doping is switched from ptype to n-type. Within the depletion region ions fixed in the lattice are left behind by the movement of electrons and holes t o combine. The fixed ionic charges left in the crystal are no longer balanced by the presence of free charge carriers and a build up of fixed charges results. These charges create an electric field which points from the n type region to the p -type region and upon equilibrium acts to oppose further charges from diffusing.
43 At equilibrium the fixed charges produce a potential barrier across the depletion region whose height will be slightly less than the band gap. This can be quantified for an abrupt chang e in doping (taken at x = 0 ) by application of a semiclassical treatment. (This approach is well B Figure 3 13. The density of holes in the p-doped region remain high until the depletion region where electron and hole combination leaves almost no free carriers. Similarly electron density is high in the n-doped region, but virtually nonexistent in the depletion layer. described in reference 13.) Following the semiclassical approach, the density of carriers nc(electr ons) and pv (holes) are shifted by the energetic change due to the potential according to: ) ] ) ( [ exp( ) ( ) ( T k x e T N x nb c c c ) )] ( [ exp( ) ( ) ( T k x e T P x pb v v v (3 .12) where Nc ( Pv) is the density of negative (positive) carriers in the conduction (valence) band, is the chemical potential (Fermi level), is the electric potential, T is the temperature, and c (v) represents the conduction (valence) band energy. If the practical case is taken in which impurities far from the transition region are fully ionized the density of impurity donors ( Nd) should be equa l to the density of conduction electrons and similarly the density of impurity acceptors should be equal to the density of impurity acceptors ( Na):
44 ) ] ) ( [ exp( ) ( ) ( T k e T N n Nb c c cd ) )] ( [ exp( ) ( ) (T k e T P p Nb v v v a (3 .13) These equations can be easily combined by solving for the chemical potential. The size of the barrier can then be written as: ] [v c d a b gP N N N Ln T k E e (3 .14) where Eg represents the band gap. The resulting potential barrier is illustrated in Figure 3 .14 A. If the energy bands are plotted allowing for equalization of the Fermi energy levels, the bands will bend in the depletion region as is illustrated in Figure 3 .14 B. A B Figure 3 14. The variations in potential and energy in a p-n ju nction. (Note that because the sign of the electron is negative, the electronic energy level show n an opposite trend to the voltage trends, based on the relation ) A) Potential barrier as s een in an unbiased p -n junction at equilibrium B) The density holes in the p-doped region remains high until the depletion region where electron and hole combination leaves almost no free carriers. Similarly electro n density is high in the n -doped region, but virtually nonexistent in the depletion layer. Application of Poissons equation, which relates the potential to the underlying charge distribution, can be made to determine the width of the depletion region. A pplying the Poisson equation and noting that the potential shift is much greater than kBT results in the following nonlinear differential equation for (x):
45 0 ], 4 [ ], 4 [ 0 d deN eN for x d d x x d d xp p n n0 0 (3 .15) Where dn represents the extent of the depletion region into the n -type region and dp represents the extent into the p type region. This differential equation can easily be integrated. Requiring the first and second derivatives to be continuous results in two simple equations that can be combined (see reference 13) to determine the depletion widths: 2 / 1 0 1 ,2 ) ( )/ ( e N N N N dd a d a p n (3 .16) If typical values are i nserted the extent of the depletion layer is found to be around 101000 nm in width, depending of course, on the level of doping. 3.2.3 P -N Junction Under External Bias An external voltage can be applied to the p -n junction. By convention, positive vol tage is defined as a voltage which raises the potential of the p-side relative to the n -side. It has already been established that before the application of a bias voltage a depletion region exists in which the density of carriers is far less than their density elsewhere in the material. The depletion region will therefore be a region of much greater electrical resistance than the regions on eithe r side of the depletion layer. When an external bias voltage ( V ) is applied, the potential will therefore only vary appreciably across the depletion layer. It was found that at equilibrium wi th no bias voltage applied, the potential increased from the p -side of the junction to the n-side by an amount
46 (determined by Eq. 3 .14) that will be d esignated as 0 If the external voltage V is not zero, than the potential across the depletion region will be modified such that: eV P N N N T k E V e ev c d a b g ] ln[ ] ) [(0 (3 .17) The modification of the barrier height by the bias voltage V is shown in Figure 3 .15. The effect of the bias ( V ) is to raise the potential ( ) on the p -side relative to that on the n -side. The barrier height is what prevents the high concentrations of holes on the p -side from diffusing into the region of low concentration on the n-side an d vice versa for electrons. If the bias voltage is zero, then the resulting electric potential, is simply the unmodified form of the equilibrium potential. For negative values of the bias voltage ( V<0 ), the potential barrier will increase. When th e bias voltage is positive the barrier height will decrease. Once the bias voltage nears the equilibrium barrier height ( 0 ), the effective barrier height becomes negli gi ble. Large diffusive currents result, as will be discussed subsequently. For positive bias voltages larger than 0 the diffusion of charges is driven by an electric field ( )] ( [ x E ). Figure 3 15. The effect of the bias voltage V on the electric potential of a p -n junction. Positive bias voltages act to reduc e the barrier height limiting diffusive currents across the depletion region. (It is noted that the potential through the depletion region for the case of zero barrier height will not be as simple as illustrated in this figure )
47 Since the width of the depletion region, determined by Eq. 3 .16, is dependent upon the potential difference, and all other parameters are fixed there will also be a corresponding change in the width which can be written as: 2 / 1 0 ,) ( 1 ) 0 ( ) ( V d V dp n p n (3 .18) The depletion wi dth is also affected by temperature, since the equilibrium potential barrier, 0 is changed with increasing temperature. The current that flows under the influence of an external bias can also be determined. To deduce the behavior of the current, the transport of electrons and holes mus t be considered independently. The case of zero external bias is considered first. The net current across the p n junction is zero, but there are still charges flowing. Consider the thermal generation of a hole (a mino rity carrier) within the depletion region. The electric field will quickly sweep the hole across the depletion region, to the p -doped side. Initially the charge configuration was in equilibrium, but the addition of a minority carrier hole to the p -doped side must now be compensated by the return of a majority carrier hole from the p -side to the n -side of the junction to maintain the equilibrium. A hole traveling back from the p -side to the n-side of the junction is going against the electric field and mu st have sufficient energy to be able to make it across the potential barrier. This is a slow process and there is only a small flow of current, just enough to compensate for the minority carriers. The same thing can be said for minority and majority electrons traveling in the opposite direction. This balance between thermally generated minority carriers being swept across the junction and majority carriers returning against the electric field is what determines the equilibrium charge distribution.
48 Now i f the case in which an applied voltage is considered, it is quickly noted that the generation of minority carriers does not change. They are dependent on the availability of energy to excite charges from the valence to the conduction band. Therefore the current due to minority carriers (often called the generation current) is independent of applied voltage. However, the flow of majority carriers (often called the recombination current) is significantly affected by the applied voltage, which changes the barrier height, consequently the applied bias voltage disrupts the equilibrium. In order for the majority carriers to cross the depletion region they must have enough energy to overcome the potential barrier. Considering first the holes: those with su fficient energy are determined according to the proportion: 2 / 1 0 Re] ) ( [ exp T k V e JB c h (3 .19) where Jh Rec represents the number current density of holes. This can be compared to the generated holes by noting that at V=0 the two must balance so that Jh Re c(0)= Jh Gen. Using this in Eq. 3 .19 gives : ] exp[ Jh ReT k eV JB Gen ch (3 .20) Noting that the net current from the p -side to the n -side is Jh= Jh Rec Jh Ge n, and that the exact same argument can be applied to electrons results in an expression for the tot al current density ( j ) from the p -side of the junction to the n-side under the influence of an electric potential: ) 1 )( ( ) (/ T k eV Gen e Gen hBe J J e V j (3 .21) This current trend is graphed in Figure 3 .16. For reverse bias conditions the value approaches the limit e( Jh Gen Je Gen)
49 Figure 3 16. Illustration showing the characteristic current voltage behavior of a diode. In reverse bias the barrier height grows with larger negative bias voltages so very little current flows. In the forward bias direction, the barrier height decreases with increasing voltage. When the voltage becomes close to that band gap of the material a very rapid increase of current with voltage is observed. This is ty pically referred to as the turn-on voltage. 3.2.3 Photovoltaic Cells A photovoltaic cell is simply a p -n junction formed close to the surface of a semiconductor. Typically this is done by having a relatively thick n -doped layer and adding a very thin layer of p-doped material on top. If the top layer is sufficiently thin, light can penetrate into the p -n junction and be absorbed. If the energy of the incident light is greater than the band gap energy of the polymer, the photons will be absorbed and will create electronhole pairs (excitons) within and near to the depletion region. The strong electric field of the depletion region will split the pairs and push the holes to the p -side of the device and electrons to the nside of the device. This nonequilibrium process results in the generation of a negative (moving from n -side to p -side) photocurrent that is not compensated for by thermal equilibration. The photocurrent is limited by the absorption of the cell, as well as by radiative recombination of electrons and holes. In general, the greater the amount of light that i s absorbed, the greater the number of carriers generated in the depletion region and the higher the resulting photocurrent current. If a bias voltage is applied and swept from negative to positive, while the
50 cell is illuminated with radiation of sufficie nt energy, the resulting current will be the combination of the photocurrent and the equilibrium current (which in solar cell terminology is most often called the dark current). The effect of the photocurrent on the current -voltage characteristics of the device is to shift the curve downward. The amount by which the curve is shifted depends on how much photocurrent the cell generates. Figure 3 17 A illustrates the dark and illuminated beha vior of a photovoltaic device. There are a number of important fea tures that are used to characterize solar cells that can be seen in this figure. The value of the current when the bias voltage is zero is called the short circuit current. It is the current that you would see if the device w ere under illumination and t he contact to the p -side and n -side were connected with a wire (short -circuited). The open circuit voltage is the value of the voltage where the curve crosses the bias voltage axis. This is the value that the voltage would be if the device were illuminated but not connected. The point of maximum power in Figure 3 .17 C is indicated where the product of I and V on the curve is maximum. The fill factor is defined as the maximum power divided by the product of the open circuit voltage and the short circuit current. The efficiency of the device is determined by the ratio of the maximum output power of the device to the input power received from the radiation source. These relations are summarized in Figure 3 .17 B. It is easy to see that the shape of the cur ve in the fourth quadrant of the illuminated I -V traces indicates how the device would behave if used to generate power. Instead of applying a bias voltage to the photovoltaic cell, consider that it is connected to a load resist o r that could extract energ y from the circuit as in Figure 3 17 C. With the device illuminated the current can be measured as the resistance load is adjusted. Power will be extracted from the photovoltaic cell according to P = I2R If the current is monitored as the resistance is continuously increased
51 from zero to infinity, the power will be known, and there will be a point in which the power is a maximum. This resistance determines the best op erating condition for the cell. A B C D Figure 3 17. I -V Characteri stics of a photovoltaic cell. (Note that the current is plotted here instead of the current density so that the power is simply I V instead of J A V ) A) The characteristics of a cell with no illumination, illumination with Intensity1, and illumination with Intensity2>Intensity1. B) Simple factors used to characterize the performance of photovoltaic cells. C) The short circuit current is the point where the curve crosses the current axis. The open circuit voltage is the point where the curve crosses the Vol tage axis. The fill factor can be visualized as the ratio of the areas of the two boxes. D) Illustration of circuit that could be used to extract electrical power from a solar cell. For a fixed intensity of incident radiation, the output power will depe nd on the resistance value of the resistor. Since the voltage drop across the resistor is related to the current by V = IR R can be eliminated from the equation determining the output power, leaving a function of only V and I. This means that the maximum output power of the cell is also contained in the I -V information
52 from the fourth quadrant of I -V curve under bias. The power is simply, P = IV The open-circuit voltage and short circuit current can be related to the variable resitor circuit. When t he resistor is at zero, the current value will be the short circuit current. When the resistor is infinite, the cell will be at its open circuit voltage. 3.2.4 Detailed I V Behavior of a Photovoltaic Cel l The response of an illuminated photovoltaic cell t o an external bias is very similar to the response of a p-n junction with the exception of the photocurre nt. The convention followed is that the bias is taken to be positive when it increases the potential of p-side relative to the n -side. Figure 3 18 il lustrates the measurement and contains data taken from an actual cell fabricated and characterized in this research. A detailed explanation of this typical I -V behavior will be made and the following paragraphs refer to the regions marked in Figure 3 18. If there is no illumination and the bias voltage (V) is swept, the device exhibits a dark current characteristic of diodes (blue curve). If the device is illuminated and the I -V characteristics are measured again, the red curve will be obtained which conta ins two current contributions, a diffusive current (or dark current) and the photocurrent. If it assumed that the two currents do not significantly affect each other, then the dark current can be subtracted from the illuminated current (subtraction of the blue curve from the red curve) and the result should be the photocurrent (black curve). To explain the behavior of these curves, the graph in Figure 3 18 has been broken into four regions. In Region 1, the bias voltage is negative. For a negative bias t he barrier height across the depletion region is large, corresponding to a strong electric field. The large barrier prevents diffusive currents, but it increases the strength of the electric field of the depletion (since the change in potential is increas ed). This large electric field, pointing from the n doped region to the p doped region, quickly splits photon-generated pairs in or near to the depletion region and sweeps holes to the p-side and electrons to the n-side. For this reason there is a large
53 photocurrent for negative bias. As the bias is brought towards zero, the photocurrent only decreases slightly since there is still a significant field remaining from the charge imbalance of the p n junction. When the bias voltage is zero, the device is in short -circuit condition and the barrier height is equal to the equilibrium barrier height, (0 This is labeled as Point A on the graph. In Region 2 the bias voltage is positive. On the left side of this region, the photocurrent is still appreciable since there is still a significant electric field across the p -n junction that can split the electron -hole pairs and sweep them across the depletion region. Moving towards higher positive bias values in this region leads to decreased barrier heights and correspondingly smaller electric fields across the depletion region. This can be seen as cau sing two effects. The first is that with decreasing barrier heights the diffusive current is increasing. The diffusive current is coming from the high concentration of holes in the p-doped region diffusing towards the n-doped region and vice versa for el ectrons. Therefore, this current is going in the opposite direction from the photocurrent and it is positive. The second effect is that the electric field becomes smaller and less effective at splitting the electron -hole pairs. The magnitude of the neg ative photocurrent begins to decrease in size. The combined result is that the total current begins to decrease more rapidly than in Region 1. As the bias becomes more and more positive, eventually there becomes a value at with the positive diffusive curr ent is equal in magnitude to the negative photocurrent and they cancel leaving the total current equal to zero. This point is labeled as Point B in the figure and it is known as open circuit voltage. Region 3 lies at bias potentials above the open circuit voltage. In this region the electric field still points from the n -side to the p -side, but the field strength is small and so is the potential
54 A B C Figure 3 18. Photovoltaic cell I -V behavior. A) The I -V trace of a photovoltaic cell can be sp lit into different regions leading to a better understanding of the underlying process. B) Configuration of a device under illumination and bias C) Summary of I -V behavior barrier. Photogenerated charge carriers are still split by the electric field, con sequently, there is still a negative photocurrent present. However, the weaker field strength is not as effective at splitting the electron -hole pairs, and larger amounts of recombination result in a much smaller
55 photocurrent than at negative bias values. In Region 3, the diffusive current is larger than the photocurrent, so the total current will be positive. Moving towards higher bias values further decreases the barrier height and the strength of the electric field. Eventually, the electric field bec omes negligible, and the photocurrent goes to zero. This is labeled as Point C in the graph. At this point the bias voltage is equal in magnitude to the equilibrium barrier height, 0. At bias values above this point, the electric field is forced to switch directions and point from the p type region to the n-type region, and the photocurrent becomes positive. This region is labeled as Region 4 in the graph. Figure 3 18 C summa rizes the results in table form. 3.2.5 Solar Cell s A solar cell is a photovoltaic cell which is designed to absorb sunlight. Sunlight is essentially the radiation spectrum of a 5800K blackbody, with differences due to spectral lines and absorption. The i rradiance that is received by the earth is fairly constant, though there is a small ~2% variation associated with the Earths elliptical orbit. The intensity of sunlight at the surface of the Earth is different from the intensity outside the atmosphere du e to both scattering and absorption. The intensity spectrum also varies with location on the earths surfa ce and time of day, year, etc. To provide a standard set of conditions under which to measure solar cells, researchers have adopted two common irradi ance spectra. These spectra are reflective of the typical conditions under which a solar cell might be used. The first of these two spect ra is referred to as AM0 and it represents the radiation seen outside the Earths atmosphere. Solar cells that are developed for space applications use this reference. The letters AM stand for air mass and indicate the amount of typical atmosphere through which the radiation passes (AM0 passes through zero atmospheres). The second and most common reference spectr um is AM1.5. It represents light that travels through approximately one and a half Earth atmospheres. Sometimes
56 a third reference spectrum (AM 1 ) is used which represents the solar radiation that would be received if sunlight w ere directly overhead, but thi s reference is becoming less common. Figure 3 .19 illustrates the difference between the three Figure 3 19. AM0, AM1, and AM1.5 are differentiated by the amount of atmosphere that they pass through. The AM0 and AM1.5 reference irradiance data are p lotted in Figure 3 .20. If the total amount of energy is summed over the frequency range the intensity of light outside the earths atmosphere is about 1350 W/m2. A similar sum for AM1.5 radiation results in about 890 W/m2. This is the typical amount of radiation received on a clear sunny day in the United States and it includes contributions from both direct sunlight and scattered light. About 40% of the AM0 intensity is lost from scattering and absorption by the atmosphere. At low wavelengths ozone i n the outer atmosphere absorbs most of the ultraviolet light. The greater difference seen between the two spectra at lower wavelengths as opposed to longer wavelengths is due to the fact that smaller wavelengths are more susceptible to scattering. (This is also why the sky is blue, as blue light scatters more than other visible light.) The other major differences are d ue to absorption bands of atmospheric gases, mostly water. The dip in the AM1.5 spectrum at ~750nm is due to oxygen and ozone, and those at ~825 nm, ~940 nm, and ~1140 nm are due to water vapor.46 Most solar cells are develop ed targeting uses on the Earths surface and characterizing these sources under AM1.5 radiation gives a good indication of how well they will perform. In
57 Figure 3 20. AM0 and AM1.5 are irradiance spectra. AM0 is the sprectrum of solar light that w ould be seen just outside the Earths atmosphere. AM1.5 is the spectrum seen after sunlight passes at a slant through 1.5 atmospheres. the A.M. 1.5 spectrum 99% of the light reaching Earth is at wavelengths less than 2500 nm, and 88% is below 1350 nm. T o obtain the most efficient solar cell, it is desirable to absorb the largest number of photons possible otherwise their energy will be lost. Because the absorption spectrum of a semiconductor is limited by its band gap, this suggests materials with sma ller band gaps are more ideal. Only photons with energies greater tha n the band gap will be absorbed, but there is a downside to having a small band gap. The maximum operating voltage of the cell is limited by the size of the band gap; the smaller the ban d gap, the smaller the operating voltage. The energy that the cell converts from absorbing a photon and generating an electron, will be equal to the operating voltage of the cell times the charge of the electron ( Eout = eVoperating). This means that al l of the energy delivered by a photon, in excess of the band gap energy, will be wasted as heat. In summary, a larger band gap can lead to a higher operating voltage, but it limits the number of photons that the cell can absorb.
58 This consideration impose s an upper limit on the efficiency of a s ingle cell which depends on the band gap of the semiconductor The limit can be obtained mathematically by following the two assumptions that only photons of energy greater than the band gap can be absorbed, and th at excess energy of photons is lost as heat. The efficiency of the device is given by ff V Joc sc (3 .23) The short circuit current can be calculated from integration of the photon flux up the band gap energy, GE E ph scdE EQE E N J0) ( (3 .24) Where Nph represents the number density of photons at a given energy, and the EQE is the efficiency at which incident photons are turned into electrons. Ideally, the EQE is one. Following this simple method for the case of silicon with a band gap of 1.1eV gives an the upper limit of about around 42%. Further thermodynamic limitations such as radiative and nonradiative recombination further reduce this level down to 30.1%.47 For a silicon cell, the band gap is about 1130 nm, which corresponds to an energy of 1.1 eV, thus, the ideal and maximum operating voltage would be 1.1 eV. If this ideal cell absorbs a 500 nm photon which has 2.5 eV of energy, it will generate the same amount of energy as the absorption of a 1000 nm photon whose corresponding energy is 1.24 eV. As noted above t his is because the absorption of a photon will only excite one electron across the band gap (while the extra energy of the photon will be wasted as heat.) Th e actual operating vo ltage of a typical solar cell is less than the band gap at around 0.5 eV for silicon H igher -end single junction commercial cells are around 20% efficient. To improve upon this efficiency, multi junction cells are made to capture a larger fraction of th e energy from lower wavelength photons.
59 Figure 3 .21 A shows the AM1.5 solar spectrum graphed from 300 to 1400 nm with the photon energy, Ephoton overlaid. The extra energy of the shorter wavelength photons can be captured using a series of solar cells s tacked on top of each other Such cells are referred to as multi -layer or tandem cells. This is done by using semiconductors with successively decreasing band gaps for lower layers. Each layer will have its own operating voltage which will be limited b y the layers semiconductor band gap. If the largest band gap is used for the top layer, then this layer will have the greatest operating voltage, but it will be transparent to longer wavelength (lower energy) photons. These photons can be absorbed by lower layers that have smaller band gaps and lower operating voltages. In a multi layer cell the energy obtained from the highest energy photons will be limited by the band gap of the semiconductor of the top layer, whereas the longest wavelength absorption will be limited by the smaller band gap of the bottom layer. This means that electrons generated in the higher layers. These cells offer improved power conversion efficiencie s over single layer cells because the band gap of a single -layer cell limits both the operating voltage and the longest wavelength absorption. A multi layer solar cell is illustrated in Figure 3 .21 B. The top layers will absorb the higher energy photons, but they are transparent to photons of lower energy and these longer wavelength photons are transmitted through to the bottom layers. If the layers are connected in series, the resulting voltage of the cell will be the sum of the layer voltages By opt imizing this type of cell, through careful match of current densities, energy conversion from the solar spectrum can be more closely matched while still maintain ing higher operating voltages.
60 A B Figure 3 21. Energy conversion from solar photons. A) AM1.5 solar spectrum with the energy of each photon being displayed. B) Structure of a multi -layer solar cell. Multilayer cells are more expensive to make but represent the highest achievable efficiencies. The theoretical limit on the efficiency by stac king an infinite number of cells has been calculated at 68% unconcentrated solar radiation and 86% for concentrated sunlight.48 The highest recorded efficiencies are based on concentrated sunlight. Multijunction cells have been made with solar power conversion efficiencies above 40%,4952 and three different sources claim to have achieved the current World Record at slightly above 40%.49,51,52 Most photovoltaic cells are single layer cells wher e the excess photon energy is lost. Each electron can only absorb one photon, and because of the excess energy loss, all absorbed photons produce the same result consequently, the important information is not the energy distribution of the solar spectrum but the number distribution of photons as a function of wavelength. So, a more elucidating way to represent the AM1.5 spectrum is to plot the photon flux. The photon flux can be determined by dividing the energy flux (irradiance) by the energy per phot on ) 1 )( / ) ( ( ) ) ( (2 2nms m hc E E Irradiance s nm m photonsphoton 3 .25
61 Since longer wavelength photons have less energy than those of short wavelengths, a plot of the photon flux is more heavily weighted at the red end of the spectrum relative to the Energy flux. The AM1.5 normalized photo n flux density is graphed in Figure 3 22 along with the integrated approximately 75% of all solar photons occur at energies less than 1350 nm. Figure 3 22. Normalized AM1.5 photon flux spectrum illustrating the variation in solar photon flux with wavelength, where the black trace indicates the photon flux and the red trace represents the totals percentage of solar photons below the given wavelength as obtained by integrating the photon flux. The photon flux is valuable for evaluating the potential photovoltaic application of a material based on its absorption spectrum. The standard method of characterizing a solar cell is to record I -V characteristics of the cell when exposed to AM1.5 radiation at normal incidence. Though the summation of the AM1.5 irradiance spectrum over the entire distribution results in a total intensity of about 890W/m2, it is more co mmon to characterize cells under AM1.5 light at 1000 W/m2 intensity. This intensity is sometimes referred to as 1 Sun. Because typical laboratory cells are often on
62 scales ~1cm, it is also common to express the intensity in mW/c m2, where 1 S un is 100mW /c m2. The efficiency of the cell can then be obtained from the I -V curve with the relations given in Figure 3 17. The AM1.5 measurements do not give any information about the wavelength dependence of the cells power -conversion efficiency. This information can be very useful to know, and it can be examined by illuminating the cell with monochromatic light of known intensity and monitoring the output current of the device under short circuit conditions. This type of measurement is known as an external quant um efficiency (EQE) measurement or equivalently as an incident photon conversion efficiency (IPCE) measurement because it measures the efficiency at which external photons are converted to electrons. It is expressed in terms of a percentage and can be cal culated by converting incident power into number of photons and outgoing current into number of electrons according to the following equation: ) )( ( ) ) /( / ( ) ) /( / ( 100 # # (%) P I e hc hc P e I hf P e I photons electrons IPCEsc sc sc 3 .26 Sometimes the internal quantum efficiency is also reported. The internal quantum efficiency (IQE) tells the number of absorbed photons that are converted to electrons and it is related to the EQE by the absorption profile of the cell. The fraction of absorbed photons can be obtained from subtracting the cells transmission (T) and re flection (R) from 1. Since most cells have a highly reflective bottom electrode, the internal and external quantum efficiencies are related by the equation IQE = EQE/(1 -R) If it is assumed that cells conversion efficiency for photons into electrons ( EQ ) is independent of the presence of photons of other wavelengths, then in principal, the AM1.5 power conversion efficiency of the cell can be calculated from a weighted sum of the EQE that takes into account the loss of excess energy from each photon whose wavelength is above the band gap. Because electron hole
63 recombination can be expected to increase under radiation of full spectral width, this limit sets an upper bound on the efficiency of the device based on the EQE. It can be calculat ed by taki ng a weighted average (over all photons of the AM1.5 spectrum) of the EQE and assuming photon energy in excess of the band gap is lost. This can be expressed with the following equation: d E V EQE d PhotonFlux PhotonFlux PCEphoton op) ) ( ( ) ( ) ( (%) 3 .27 where Vop is the expected operating voltage of the cell which will be equal to the band gap of an ideal cell. Integration of the first two products inside the integral simply gives the total photon conversion efficiency. Multiplication by the third product accounts for the fractional perce ntage of each photons energy that will be acquired. This can be simplified to: d EQE Flux PhotonFlux V PCETotal op) ( ) ( 1240 (%)) ( 3 .28 This method is useful for using the EQE to estimate the PCE efficiency of cells numerically, where the integral is replace by a sum. 3.3 Organic Solar Cells 3.3.1 First Generation Organic Cells The first reported observation of photoconductive behavior in organic materials dates back to at least 1906 when photoconductivity was discovered in anthracine.53 But it was not until 1979, well after the first commercial uses of photovoltaics began, that the first organic breakthrough came when Tang54 discovered that by using two organic layers he could achieve efficiencies of one percent. This significant improvement over other attempts at the time (whic h only incorporated single organic layers between electrodes) was not published until 1986.55 The principal discovery of Tangs work was that the interface between two organic materia ls of
64 dissimilar work functions could be used to separate singlet excitons. Figure 3 23 illustrates a device s structure simila r to the one employed by Tang. Tangs findings were a huge step forward for organic photovoltaics they got to the root difference between organic photovoltaic materials and inorganic materials. The absorption of a photon in inorganic semiconductors results in the formation of an exciton with very weak electron -hole binding energy. This binding energy can easily be overcome and the essential product is a free electron and a free hole. As discussed in section 2 .3 the coupling between charge carriers and the lattice of organic materials, as well as a lower degree of screening, results in a significantly higher binding energy that is not easily overcome thermally The essential product of photon absorption is a mobile exciton that is not easily separated. The results of Tangs work showed that the interface between two materials was an ideal site for exciton dissociation. Figure 3 23. Schematic illustration of a two layer donor acceptor photovoltaic device si milar to the one used by Tang. There is another significant difference between traditional semiconductors and organic conductors. Inorganic semiconductors are crystalline and have high mobility rates. Organic materials usually show a very high degree of disorder and transport takes p lace through successive hoppingresulting in lower rates of charge transport. This is well -phrased in a quote by Carston Deibel ,56 comparing the two Crystals are like the autobahn for charges, whereas
65 disordered matter are country roads at best. For this re ason mobility of the exciton plays an important role if the exciton is to get to an interface for dissociation and because excitons are neutral their movement is limited to diffusion. Another key factor is the lifetime of the singlet exciton state, whic h is typically on the order of nanoseconds. In contrast, charge transfer between a polymer and fullerene can be as fast as 100 fs.57 A limiting factor in organic materials is therefore the diffusion length of the exciton, which is typically on the order of 10 nanometers in conjugated polym ers.58 A redeeming factor for organic materials is that their absorption coefficients tend to be high compared to inorganics and a very significant fraction of solar photons can be absorbed by very thin layers (~100 nm). A B Figure 3 24. Energy level representation of a donor acceptor solar cell. (Note the difference in the direction of the electric potential and the energy axes due to the negative charge of an electron.) A) Representation of the energy levels prior to electrical contact. B) Simplified illustration showing band bending due to Fermi level alignment after electrical contact. Illumination will result in electron gen eration in the donor and a subsequent photocurrent. The photovoltaic devices used in organic solar cells are often known as donor acceptor devices because of the underlying charge transfer process. The energetic configuration of a simple donor acceptor de vice is illustrated in Figure 3 24. An important feature in such a device is the offset between the relative energy bands of the donor and the acceptor. In order that exciton s be dissociated at the interface between the materials, the valence band offset must be
66 greater than the binding energy of the exciton. Typical binding energies are on the order 0.3 0.5 eV and unless the band offsets are in excess of this amount efficient charge transfer will not result. This suggests having a high valence band for the donor material, but there is a drawback to this. A larger valence band will increase the band gap of the material and limit the absorption to shorter wavelengths. Of course this could be compensated for by developing a material with a high HOMO le vel. However, the higher the HOMO level of the polymer the more sensitive the material is to oxidation. In order that the polymer be air stable the HOMO level of the polymer needs to be below 5.0 eV. A good material will then have a LUMO level high eno ugh above the LUMO level of the acceptor to be able to overcome the binding energy, but low enough that the EG is not large and that absorption is not hindered. The ideal LUMO level will therefore be dependent on the LUMO level of the acceptor. Absorption of a photon by the donor results in the formation of an exciton in the donor. The neutral exciton migrates to an interface between the donor and acceptor via diffusion. At the donor acceptor interface, strong differences in electron affinity result in b and bending and the formation of a junction. If the energy difference is greater than the exciton binding energy, it can result in the dissociation of the exciton, where the electron is transferred to the acceptor and the hole remains in the donor. (Alte rnatively this process could be reversed in which case the acceptor material could absorb light, forming excitons in the acceptor, which could be dissociated at the interface by transfer of the hole from the acceptor HOMO to the donor HOMO. However, highl y absorbing acceptor s are not yet common though they would be beneficial T his type of transfer, known as back transfer only accounts for a small fraction of the photocurrent, usually at short wavelengths.) The remaining hole in the donor forms a posi tively charged polaron and the electron forms a negatively charged polaron in the acceptor material.
67 These oppositely charged polarons are still attracted to each other, but the Coulombic attraction between them is significantly reduced and the pol aron pair can easily be split. The depletion of excitons by dissociation at the donor acceptor interface results in an uneven distribution of excitons throughout the donor material and further diffusion results in a net flow of excitons toward the donor and ac ceptor interface. Positive polarons build up in the donor and negative polarons build up in the acceptor. Because the attraction between the polaron pairs is significantly reduced the positive polarons can diffuse through the donor material and the negat ive polarons can diffuse through the acceptor material. However, the pairs can also recombine in geminate recombination. The dissociation of the polaron pairs is significantly increased by the presence of an electric field which can be created by appropr iate choice of electrodes with dissimilar work functions. Another important consideration is the choice of the el ectrodes, at least one of which needs to be transparent. Typical organic cells use a transparent high work function conducting electrode such as indium tin oxide, as the anode and a low work function metal such as aluminum, for the cathode. It is important to have the work function of the anode near or slightly above the conduction band of the donor to facilitate charge transfer to the electro de. Similarly the work function of the anode is chosen to be near to or slightly under the valence band of the acceptor material. Figure 3 24 A illustrates a good match as can be seen by the energy levels before electrical contact of the layer s is made. Once the layers of materials are in electrical contact as in Figure 3 24 B, the Fermi levels of the materials must be in alignment. Due to the difference in the work functions of the electrodes, this results in band bending of the energy levels of the or ganic valence and conduction bands. An internal electric field will then be present inside the device which is
68 known as the built in field. Unlike the localized field of present at the donor acceptor interface, this built in field can extend deep into the device from the edges both contacts, though it is undoubtedly not constant as the linear depiction of the band bending illustrated in Figure 3 24 B would suggest. The built in field is important in aiding in the dissociation of the polaron pairs which bu ild up at from exciton separation across the donor acceptor interface. The presence of the built in field is responsible for causing electron migration from interface to the electrodes. The application of the dual layered organic device introduced by Tang resulted in considerable improvement in organic based photovoltaics, particularly in the case of small molecules where efficiencies of small molecule based devices efficiencies reached greater than 3.5%.59 The dual layer design however has the drawback o f requiring high large exciton diffusion distances. In order to absorb enough sunlight to produce reasonable efficie ncies the donor layer must be of thicknesses which typically exceed the diffusion length of the excitons. This can result in recombinative loss and ultimately limit the efficiency of the device. The dual layer device architecture is particularly poorly suited for polymers whose morphologies are much less crystalline than small molecules. The true advantage of using organic materials for sol ar cells lies in their ability to be solution processed. Typically, absorption is very strong in organic materials and this means that extremely thin layers (~100 nm, much thinner than inorganics where thicknesses are 1 micron, or greater) can be used which require very little material.57 The strong absorption is often accompanied by thinner absorption widths, but the tunabiltiy of polymers and the prospect of multilayered devices with organic materials is still a viable pa thway to low -cost solar cells.
69 Small molecule organic solar cells typically employ evaporative techniques which is a major drawback. The role of organic materials in commercial solar cells seems to be greatly dependent on solution -based techniques. The disorder of conjugated polymers renders them poor candidates for bilayer devices. Fortunately, a second major breakthrough was discovered which truly opened the door for conjugated polymer photovoltaics. 3.3.2 Bulk Heterojunction Organic Photovoltaic Cells In 1995 a publication in Science from the group of Allan Heeger60 reported that a blend of MEH -PPV with C60 had a photoconductivity that was increased by an orde r of magnitude. Through the development of more soluble derivatives of fullerene they were able to use a polymer fullerene blend as the active layer between two contacts and achieve an external quantum efficiency of 29% and a monochromatic power conversi on efficiency of 2.9%. Heeger coined this new architecture for polymer photovoltaics, a bulk heterojunction (BHJ) In the same year the group of Fred Wudl developed a highly soluble derivative of fullerene known as PCBM ([6,6]phenyl C61 butyric acid methyl ester) which could be used to deposit polymer -fullerene blends from solution to form active photovoltaic layers.61 As the name suggests, a bulk heterojunction is a mixture of donor and acceptor materials in which the interface between the materials is spread throughout the bulk of the blend. This means that instead of having a low interfacial surface area to volume ratio, as in the case of the planar separation between two films, that phase separation throughout the blend results in a greatly enhanced interfacial surface to volume ratio. A si mplified schematic representation of a bulk heterojunctionerojunction blend appears in Figure 3 24. This type of active layer is very advantageous for polymers because even with thick polymer layers, excitons which are generated throughout the blend are not far from an interface at which they will be dissociated. Phase separation is not easy to control and continuous pathways to the contacts may not always
70 be present, nevertheless the advantage of the ability to dissociate low mobility excitons throughou t the film le d to very significant improvements in conjugated polymer -based photovoltaic performance. Figure 3 24. Bulk heterojunction solar cell architecture. By blending a conjugated polymer with PCBM phase separation can result throughout the bulk of the film -providing improved donor acceptor interfacial surface to volume ratio. This greatly enhances the ability of the blend to dissociate excitons before the y recombine radiatively. Operation of a bulk heterojunction cell is very similar to operation of a bilayer organic solar cell. The obvious difference is that both the donor and the acceptor are in contact with the cathode as well as the anode. This merits the obvious need for two different work function electrodes. Typically a transparent high work function electrode is used as the anode and a reflective low work function metal is used as the cathode. Figure 3 25 A illustrates the energy levels of a bulk heterojunction solar cell before contact where both the donor and acceptor conduction b ands and valence bands are shown. After the device is assembled there will be a built in electric field due to Fermi level alignment which causes band bending across the blend film. In Figure 3 .25 B, this is represented by a constant downward slope of the energy bands, which is an oversimplification. The degree to which the field extends into the blend from each electrode will be different in the donor and in the acceptor and will surely not be linear. The common method of illustrating the energy leve ls of both the donor and the acceptor as done in
71 Figure 3 .25 A can be confusing as it seems to represent a homogenous blend with two valence bands and two conduction bands, which is not the case. Figure 3 25 C illustra tes showing the inhomogeneity of the b lend with regions of phase separation. If a constant path through the film is followed as is indicated by the arrow, there will be a corresponding variation of voltage especially in crossing the donor acceptor interfaces. The valence and conduction bands along this line would show a similar trend to the voltage behavior as is indicated in Figure 3 .25 D. The marks on the xaxis correspond to the transitions from one material domain to the next in the arrow. The overall average downward trend of the ener gy bands represents the built in voltage of the device. The regions of steep slope at the donor acceptor interfaces are regions of strong electric field due to the different electron affinities and ionizati on energies of the materials. In analogy to inor ganic solar cells t hese are the corresponding pn junctions of BHJ solar cells. The photovoltaic process in a bulk heterojunction cell is illustrated in Figure 325 E. When the cell is illuminated, photons are absorbed by the donor material (and by the ac ceptor material to a smaller extent). This results in the formation of excitons inside the donor and acceptor domains. Migration of the excitons to a junction (if the excitons do not radiatively recombine first) will lead to separation into a positive polaron in the donor material (polymer) and an negative polaron in the acceptor material (PCBM). The binding energy of the pair is greatly reduced, and the built in electric field due to the different work functions of the electrodes pushes the carriers tow ard opposite electrodes. If a continuous pathway exits, positive polarons will move towards the higher work function anode where they produce a build up of holes. Migration of negative polarons towards the lower work function cathode results in a build u p of electrons on the cathode. The build up of charges on the cathode and anode act to oppose
72 Figure 3 25. Details of a bulk heterojunction organic solar cell. A) Energy levels before electronic contact is made. B) Energy levels after contact is made where the difference in electrode work functions has caused the conduction and valence bands of the donor and acceptor to bend. C) An arrow representing a direct pathway through a bulk heterojunction film with markers indicating the change from one material to another. D) Simplified representation of the conduction and valence band energies as a function of the distance along the arrow of C. For simplicity band bending has been drawn extending throughout the entire blend film, though it is more likely that the bending occurs mostly at the electrode interfaces. E) A bulk heterojunction solar cell under illumination where the photovoltaic process is illustrated for both forward and back transfer.
73 A B C D E
74 the built in electr ic field, and increased positive charge buildup in the donor, and negative charge buildup in the acceptor act to oppose the effectiveness of the exciton dissociation. Additionally, the increased presence of charges can lead to increases in polaron recombi nation. So, eventually the photovoltaic process saturates and the net current flow onto the electrodes becomes zero. At this point the voltage difference between the cathode and anode is the open circuit voltage. If the electrodes are connected to an e xternal load, the charge buildup of the cell c an be used as a power supply. There is an obvious dependence of the open circuit -voltage on illumination intensity due to a greater density of photons creating a greater density on the electrodes. This can be realized from the fact that with zero light intensity the open circuit voltage is zero. As the intensity is increased the voltage will build up. At some point the process will saturate. The relative dependence of the o pen circuit voltage of the cell on t he electrode work function difference and on the donor acceptor electron affinities is not easily separated often resulting in confusion in the field on work function origins .62,63 Open circuit voltage has been found to hav e a strong correlation to the difference between the donor valence band and the acceptor conduction band.64 The strong degree of separation at this interface results in positive charge buildup in the polymer and negative charge buildup in the acceptor. There will be a corresponding voltage difference between the two materials. The charge transfer from the donor to the anode is more effective than from the donor to the cathode because of more closely matched energy levels and vice versa for the acceptor. This will r esult in an open circuit voltage d ependence on the HOMO -LUMO gap. However, open circuit voltage has also has also been found to have a strong dependence on difference in work electrode work functions .62,63,65 The built in electric field due to the Fermi
75 level differences facilitates transport of charge from the donor to the anode and acceptor to the cathode resulting in an open circuit voltage depen den ce on the work function difference. The explanation presented here explains the mutually observed depend ence on both the HOMO LUMO gap and the electrode work function difference. This would suggest a connection between the donor -HOMO acceptor -LUMO gap and the open circuit voltage that would be heavily dependent on incident light intensity. However, open cir cuit voltage saturates at high intensities indicating that both the gap and the electrode work function difference play a role in open circuit voltage. The prototypical polymer fullerene bulk heterojunction solar cell is built starting with an indium -tin oxide (ITO) coated glass substrate to act as the transparent anode, followed by deposition of a layer of poly(3,4 ethylenedioxythiophene): poly(styrenesulfonate) known more commonly as PEDOT:PSS. (This layer is a highly doped and therefore conducting c onjugated polymer layer which is predominately transparent to solar photons. ) It is deposited from an aqueous solution, and is insoluble in organic solvents, thus allowing the deposition of additional layers from organic solvents. The PEDOT:PSS layer has a low lying work function which lower s the work function of the cathode as well as increasing the wettability of the substrate surface for active layer deposition. This reduces shunts, or small shorts, caused by incomplete active layer coverage. The thi ckness of this layer can also be adjusted and used as an optical spacer, to slightly modify the optical properties of the cell. The polymer fullerene blend film, often called the active layer, is deposited on top of the anode. Typically the active layer i s deposited through spin coating, but other deposition methods have also been explored with success. An aluminum cathode layer is then added on top of the active layer. In many cases it has been found to be beneficial to add a thin lithium flouride layer
76 (LiF) or calcium layer prior to deposition of the cathode which can enhance efficiencies by around 20% for some cells.64,66 The effect of lit h ium fluoride is not entirely understood but it is thought to reduce reactions that break down the organic layers near the contacts67, though other explanations have been proposed.68,69 The layered configuration is often represented as glass/ ITO/ PEDOT:PSS/ Polymer: PCBM / (LiF)Aluminum. 3.3.3 Optimizing Bulk Heterojunction Solar Cells There are many factors that affect the performance of bulk heterojunction cells consequently routes towards optimization are not simple. There are key material properties suc h as conjugation lengths, planarity, solubility, polydispersity, material purity, and HOMO -LUMO offsets. Charge carrier mobility is important in films to effectively transport photo-generated charges. Absorption intensity and band gap size play roles of obvious importance. There are many obvious parameters that can be varied when optimizing cell performance, such as blend film donor to acceptor ratios, film thickness, deposition methods and drying rates. It is important to realize though, that many var iations that lead to maximum performance relative to one parameter dont correspond to the maxima of other parameters. For instance, if the film thickness is adjusted to obtain maximum performance for a certain weight ratio of donor matarial to acceptor m aterial that film thickness will not be optimal for a different ratio. The process of optimizing bulk heterojunction cells involves tradeoffs, nevertheless, considerable improvements have been made. Morphology plays an important role in device performanc e and there are many different processing parameters that can affect the morphology. A number of researc h efforts have found that by simply varying the type of solvent used to dissolve and deposit the blend film materials they could achieve considerable improvements in performance Most of these performance enhancements could be traced to morphology differences.7072 In the case of poly(3 -hexyl -
77 thiophene) cells large improvements can be realized through annealing methods which increase crystallinity, red -shift the absorption spectrum, and improve donor acceptor domain separation, leading to doubled efficiencies in some cases. Methods of attaining these r esult s have included solvent -vapor annealing,73, thermal annealing74,75, and even microwave annealing.76 Another aspect of the bulk heterojunction blend that has considerable impact on the device efficiency is the interface between the donor and acceptor domains. Through the use of additives in the polymer fullerene blend s, research has found that these interfaces can be modified resulting in higher charge separation efficiencies.77,78 Further progress on photovoltaic cells includes methods of optimizing the thin -film interference properties. Interference is de pendent upon layer thicknesses by using thin -film optical spacers. Interference of light in thin films varies with wavelength. For the wavelengths of peak exciton generation, interference can be influenced through the use of optical spacers to achieve i nterference maxima within the active layer.40,79,80 The less broad, but intense absorption of polymer thin films mean s that it is likely that the best cells will be made with multi layer architectures and there have been a number of studies exploring tandem cells.80,81 The most efficient cells to date have been made by layering bulk heterojunction blends with a thin separating layer of titanium oxide.45 Recently an excellent review of organic tandem solar cells was published which; it can be found in reference 82. Further developments on the materials synthesis side, where there is considerable focus on developing materials extending the absorption spectrum towards longer wavelengths, have l ed to low band-gap solar cells.83,84 An important design exists that has been show n to increase performance by geometrically reducing the total cell area, but increasing the effective active area
78 throu gh the use of folding. As seen in Figure 3 2 6 this shape can increase path length and decrease reflective losses.85 Figure 3 26. Folded cells can increase effect ive active area. This figure illustrates a folded tandem cell in which the layer s absorb in different regions of the spectrum. Layers can be connected in series or parallel. The above mentioned avenues toward s enhancing performance as well as others tha t have not been discussed, have led to a broad array of published research associated with many advancements in the field of conjugated polymer solar cells. This work explores some of these techniques and seeks to develop others.
79 CHAPTER 4 EXPERIMENTAL T ECHNIQUES AND INSTRUMENTATION 4 .1 Introduction Chapter 1 provided background information on general aspects of conjugated polymers, charge transport, and photovoltaics. This chapter will provide detail and specifics about the materials, instruments and m ethods used for this work. While sometimes there were slight variations in the fabrication and characterization processes the general method s outlined here applied to nearly all of this work 4 .2 Materials and Purification Purity of materials is crucial to the production of high quality organic electronic devices. Because of the sometimes detrimental effects from the presence of dissolved water and oxygen, it was necessary to remove these gases from the solvents. Solvents were purchased from Sigma Aldric h unless otherwise noted. Anhydrous c hlorobenzene (99.8%) was t ransferred into a Schlenk flask and freeze -pump thawed ( frozen with liquid nitrogen and put under vacuum to remove oxygen and then thawed) for three cycles without exposing to air. The Schlenk flask was then backfilled with argon and taken into an argon filled glovebox. Anhydrous 1 2 dichlorobenzene was bought from Sigma Aldrich and freeze -pump thawed(3 4 cycles) before bringing the solvent into the glovebox. O lder work that utilized chloroben zene was done with a different supply that was not anhydrous. To remove water as well as oxygen this solvent was distilled over phosphorous pentoxide (P2O5) under vacuum then transferred into a Schlenk flask and freeze -pump -thawed (3 4 cycles) without ex posing to air. The Schlenk flask was then backfilled with argon and taken into the glovebox. Chloroform was distilled(at atmospheric pressure under argon). The
80 chloroform was then freezepump thawed(3 4 cycles) before bringing the solvent into the glove box. Indium tin oxide (ITO) was used in all devices as the transparent electrode. All other electrodes used in this work were deposited using metal -vapor -deposition. It is well known that the work functions of metals are dependent upon their purity Whe re possible, the highest purity materials were obtained The metals and electrodes used in this work were obtained from variety of suppliers. U npatterned ITO on polished float glass (8 x 25 x 1.1mm was purchased from Delta Technologies (Still water, MN http://www.delta -technologies.com product #X140) Patterned ITO o x 25 x 1.1 mm was purchased from Kintec (Hong Kong, China www.kintech.hk) Figure 4 1 illustrates the patterns where the black region repre sents the ITO and the white region has been etched Lithium Flouride (LiF 97% purity ) was purchased from Fisher Scientific (item #AC218271000 ) and used as recieved. This material is extremely hygroscopic and was stored inside the glovebox. Aluminum slu gs (99.99% purit y) were purchased from Alfa Aesar( Item #40417) and used as recieved Gold was purchased in the form of high purity coins from a local supplier (National Coins, 2007 NW 43rd Street, Gainesville, FL 32605, 3523783983). The coins purchased w ere Canadian Maple Leaf (1982 -present mints) coins at 99.99% purity. These are the highest purity co ins in mint. In 2007 a special edition of this coin was minted at 99.999% purity It is recommended that the 2007 mint be purchased in the future as bot h are sold at the same high purity rate. After purchase t he coin s were cleaned by sonication in a solution (~30mg/mL) of sodium dodecyl sulfate
81 (1 2 minutes each), sonication in Acetone, and sonication in distilled isopropanol. Prior to cutting the coin the sheer was thoroughly cleaned with hexanes, acetone, and isopropanol to remove all traces of cutting oil. The coin s were then cut into approximately 1g pieces using a m etal sheer in the machine shop. Alumina coated tungsten evaporation boats were purchased f rom R.D. Mathis Company (PART #S35B -AO -Mo) Alumina coated boats are mo re inert and do not wet by most metals making them more stable to multiple evaporations. These boats were used for gold, aluminum, and lithium flouride evaporations and are suitable for sources with melting points below 1200C. Figure 4 1. ITO P atterns d esigned and purchased from Kintec, Hong Kong The white regions of the substrate have been etched leaving the ITO pattern represented by the black regions in the figures. All new polymers utilized in this work were syn thesized by Prof. J.R. Reynolds gr oup. The last section of this chapter contains gel permeation chromotography (GPC) analysis and nuclear magnetic resonance ( NMR ) r esults on most of the materials used Those that are not documented here are documented in the published literature indicate d in the relevant chapters. The syntheses of these materials are also documented in the literature which can be found in the
82 references. Other materials not synthesized in the group, were purchased from a variety of suppliers. Several grades of [6,6] phenyl C61 butyric acid methyl ester ( PCBM ) were purchased. Both 99% (product # 91015001G) and 99.5% (product # 91015051G) PCBM was purchased from SES Research and used as received. Material was stored in desiccant dry box. Regioregular (~94%) P3HT was purchased from Rieke Metals. The P3HT was purified by Soxhlet extraction with methanol, acetone, hexanes and chloroform. The chloroform fraction was condensed under reduced pressure, filter and the polymer was obtained after vacuum drying Regioregular (>98.5%) P3HT BASF Sepiolid P200 was p urchased from Rieke Metals and used as received. RegioRandom (~60% head to tail) P3HT was purchased from Rieke Metals and used as received. This p olymer w as also stored under argon atmosphere inside the glovebox. The instability of P3HT in air is widely noted and all three samples were stored under argon inside the glovebox. GPC results for each of these three polymers appear in Figure 4 16, and NM R resul ts appear in Figures 4 18, 4 19, and 4 20, at the end of this chapter. The poly(3,4 -ethylenedioxythiophene): poly(styrenesulfonate) (PEDOT:PSS) used for this work was (Bayer Baytron P VP Al 4083). It was kept refrigerated when not in use. Prior to use a liquots of PEDOT:PSS were gradually warmed to room temperature while stirr ing 4 3 Film Formation The low -cost importance of solution processibility was mentioned in chapter 1. There are a variety of methods that can be used to create the thin conjugated polymer films necessary for electronic applications. These include dip -coating, blading, screen printing, spin coating, and spray casting. Though eventually it is expected that various screen printing methods will be useful in the commercialization of roll to roll processed devices, by far the most widely used
83 laboratory method is spin coating. This is because of the easily replicated conditions and of the high degree of uniformity of the resulting films. Spray casting is also a useful technique for pro ducing thin -films, but the films produced from spray casting offer lower repeatability and have a greatly increased surface roughness (which for some applications may be beneficial.) Regardless of the method used to create films, there are many important factors that come into play. Clean substrates are essential in the formation of good polymer films. In this work all substrates were thoroughly cleaned following a previously established explained here After labeling the backside of substrates using a d iamond tipped scribe, each was thoroughly rinsed in deioniz ed water to remove any glass chips that might scratch the surface in later cleaning steps. (It should be noted that use of powder -free gloves is essential at all times during cleaning and that som e latex gloves, though powder -free, contain a layer of aloe that can be dissolved by the cleaning solutions. This aloe layer can simply be washed off the gloves with soap and water prior to starting the cleaning process. ) The cleaning process began by cr eating a surfactant solution made by dissolving approximately 30 mg of sodium dodecyl sulfate in ~150 mL of 18 Substrates were then dipped in this solution and gently rubber with a cotton tip swab. All substrates were th en sonicated in the surfactant solution for 15 Sonication was then repeated for 15 minutes each in: pure 18 distilled Isop ropanol. U pon removal from the isopropanol (using clean tweezers) each substrate was rapidly dried by blowing with filtered nitrogen and immediately transferred into a plasma cleaner (Harrick PDC 32G) The plasma chamber was then evacuated using a pumping station w ith a
84 dry -ice/isopr op anol cold trap and back -filled with pure oxygen. The cycle was repeated once to ensure a pure oxygen plasma and the substrates were then cleaned by turning the plasma cleaner (18 Watts applied to RF coil) to high for 15 minutes. Film s were cast immediately after removal from the plasma cleaner (It was noticed that use just 1 hour later resulted in greatly reduced surface wettability and poorer quality films .) Microscopic particulates (i.e. dust) from the air can create large visible defects (referred to as comets, see Figure 4 2 A ) in polymer films by altering local surface tensions and affecting solvent flow patterns during spin coating. Due to early difficulties in obtaining comet -f ree films, a positive pressure cleanbox was designed and fabricated to allow for the deposition of these films in a particle -free environment. This was made by creating a clear, acrylic enclosure into which a high volume in line fan pushed air through a Hepa filter (0.3 micro n filter). The design incl uded a variable current allowing for flow adjustment. When not in use the filtered airflow was left on at a reduced setting to prolong filter life. Placing both the plasma cleaner and the spincoater inside this remarkably s imple design eliminated virtual ly all comets in the resulting films. The process of spin coating is affected by many factors including viscosity, surface tension, temperature, solvent volatility, substrate wettability, aggregation, temperature variations, ambient conditions and edge eff ects, among others. Yet, if done carefully it produces highly consistent and repeatable results. In general the spin coating process can be broken down into four general stages. The first is the application of the solution onto the substrate. The second stage involves accelerating the substrate and solution from rest to the desired number of revolutions per minute (RPMs). The time that is taken for this acceleration process to complete is referred to as the ramp time.
85 A B C Figure 4 2 Dust and film formation. A) Optical image taken of a comet in a P3HT: PCBM film caused by a ~100 micron particle. B) Schematic diagram of cleanbox featuring 350 CFM enclosure fan and 12 x 12 x 3 Hepa filter (McMaster Carr part # 2153k31) According to manufacturer filter eliminates 99.98% of particles down to 0.3 microns in size. C) Photo Cleanbox with heat lamp spin coater and plasma cleaner inside. During this phase there is a rapid discharge of fluid from the substrates surface. This phase generally takes from 2 5 seconds depending upon the instrument and the final spin rate. The third stage is when the substrate is spinning at a constant angular rate at the desired RPM and the effects of fluid flow and viscosity dominate the loss of solvent .86 During this stage the layer of solution on top of the substrate should be fairly uniform in thickness. This layer gradually thins as solution flows to the edges builds up, and is flung off. Edge effects are almost always visible in spin coated films due to this buildup. The fou rth stage begins as fluid flow diminishes and solvent evaporation governs the drying of the film. It has already been noted that clean surface conditions are extremely important. If the entire surface is not wetted incomplete films can result. Similar to the manner in which comets can result from dust, aggregates in the solution can lead to comet formation. It is essential that polymer solutions be completely dissolved before application to the substrate surface. Even though larger particles can be filtered out, aggregates small enough to fit through filter pores can still cause relatively large film def ects. It is also important to properly center the slide on the spin coater. An off -center slide can result in an unbalanced film. Another very important
86 consideration when applying the solution to the substrate is to ensure there are no bubbles or foam on the film before it is spun. Bubbles greatly alter local surface tensions and inevitably lead to large film variations. Surface roughness is known to be highly dependent upon spin rate and the volatility of the solvent used, where more volatile solvent s and lower spin rates produce rougher films.87 W hen forming films for this work, freshly cleaned slides were centered on the spin coater chuck and then a vacuum was applied to secure t he substrate to the chuck when spinning. Films were spun using a Chemat model KW 4A spincoater. F ully dissolved solutions were loaded into a syringe leaving a n air -gap between the solution and the plunger of the syringe to prevent any contamination. The volume of solution used varied between 50 400 L per substrate depending on surface wettabilit y and substrate size (substrate sizes ranged from 1.0 x 1.0 cm through 2.5 x 2.5 cm). All solutions were filtered as they were applied to the substrates. N ylon filters (0.2 0.45 m) were used for aqueous solutions and polytetrafluoroethylene (PTFE) filters (0.2 0.45 m) were used when filtering organic solutions. A B Figure 4 3 Spin Coater Chucks. A) Old chuck design which pulled the substrate onto the sur face by sucking air through the channels and into the vacuum port in the center B) New chuck design greatly reduced air flow and surface defects by incorporating O rings. Besides the previously mentioned comets in early work, some films also produced a def ect in the form of a cross pattern in the center of the slide, mimicking a pattern beneath the slides on the spin coater chuck as shown in Figure 4 3 A. This pattern was determined to be resulting
87 from the flow of air into the inner vacuum hole of the chu ck. The air in these regions was expanding because of the vacuum and thus resulting in local cooling on the above substrates. This problem was remedied by redesigning the chuck with vacuum groo ves for O rings. Two different sized O rings (as labelled in Figure 4 3 B) were incorporated so that both large and small substrates could be used. Films produced using the newer chuck design that were fully constant across the substrate surface. Some materials were not soluble at room temperature and for this rea son it was necessary to heat the solutions before spin coating. Because these solutions quickly aggregated when deposited on to a room temperature substrate a heat lamp was used to achieve similar substrate temperatures. A thermocouple (Omega model HH12) was used to determine the proper distance away from the bulb to achieve the desired temperature. The substrates were then left under the lamp for several minutes before spinning to allow the temperature to equilibrate. A second method utilized for creati ng films was spray -casting. Due to the hazardous nature of many organic solvents used, a ventilated spray -casting both was designed and built and a respirator mask was worn. Spray casting films using an airbrush offers a large degree or variability throu gh solvent choice and combinations air pressure levels, and fluid flow adjustments, as well as external conditions such as temperature. The films spray -cast for this work did not explore many of these variations because the films were being made in compar ison to films spin coated under specific conditions. These sp ray -cas t films were cast from the same solvents as were used to create the corresponding spin coated films This was done to preserve comparability so that any solvent related effects to morphol ogy were the same in both types of devices. All spray casting was done using a suction fed Iwata airbrush. PEDOT:PSS films were sprayed from undiluted stock PEDOT:PSS (Baytron
88 P). Active layer films were sprayed from solutions of polymer in chlorobenzene at varied concentrations (5 17mg/mL). In all instances nitrogen was used as the compressed gas source With airbrushes, higher air pressure settings result in smaller size droplets of sprayed solutions. For this reason, the nitrogen tank regulator was set at its ma ximum value of 60 PSI. 4 .4 Characterization of Film Morphology and Thickness Using Atomic Force Microscopy The importance of film morphologies was addressed in the preceding chapter. Atomic force micr oscopy (AFM) was used to investigate the morphology of the films in this work. Because of the relatively soft nature of polymer surfaces, nearly all images were taken using tapping mode. This method of microscopy involves driving the tip at oscillations near the tips resonanc e frequency. As th e sample is scanned beneath the tip, interactions cause deviations in the tip s oscillations. The microscope uses a piezoelectric actuator to maintain the amplitude of these oscillations. In this way the scanning process measures the interaction forces be tween the tip and the sample. Images used for this research were taken using a Veeco Innova (Model 840-012711) atomic force microscope. Surface features differ over a very large scale with the roughness from spray cast films beco ming evident on scales be tween 2 0 100 microns, whereas polymer -fullerene phase separation regions may not be apparent until scales as small as 1 micron. Hence to investigate morphology differences often times multiple images were taken from large (20 100 micron) to small scales ( 0.5 2 microns). Film thickness pla ys an important role in device performance. A ccurate characterization of film thickness is essential. Thickness measurements were made using the Veeco diInnova a tomic force microscop e in tapping mode. The se measurements were made by using a sharp razor blade to cleanly scratch the polymer film down to the substrate. Multiple AFM scans
89 found that this scratching (even with pressures greater than needed to remove the polymer layer) produced no detectible effects on either the glass or ITO su bstrate s. Producing good scratches was important in accurately determining the film thickness es The scratch es needed to satisfy two main criteria. They need ed to be clean in the sense that the scratched area did not have any remainin g polymer. The scratch also needed to be free of polymer buildup at the edges. Good scratches could be detected with the optical microscope of the AFM Figure 4 4 shows an optical image of a cross -shaped scratch on a dual layer film (glass/ PEDOT:PSS/ polymer) where both polymer layer s are visible Scans of 5 15 micron s in width were run back and forth from the scratch which exposed the substrate surface up onto the surface of the polymer film. The difference in height of these two regions could then be used to determine the film thickness. Figure 4 4. Optical image displaying a cross -shaped scratch dual layer film(glass/PEDOT:PSS/polymer). This image shows the exposed glass. The AFM tip visible in the image, has an approximate width 35 microns and provides a sense of the size scale. An example of a n AFM step height measurement appears in Figure 4 5 A The difference in the height of the two surfaces is apparent. Figure 4 .5 B illustrates the same scan in two
90 dimensions. A histogram of the heights of the polymer film (red box on left) and the substrate (red box on right) can be used to calculate the height difference, which is the thickness of the film. The results of the distribution analysis are shown in Figure 4 .5 C. For this particular case th e film thickness was ~127 nanometers. A B C Figure 4 5 Atomic force microscopy images used in calculating film thickness. A) Three dimension view showing the drop from polymer film (yellow) to the substrate (black). Film thickness step height measureme nt B) Two dimensional view from the same scan showing the regions used to generate a histogram of the height distribution. C) Histogram results showing the average heights of the two surfaces and their difference as used in the determination of film thickness.
91 Some materials produce scratches better than others and so technique varies from sample to sample. If the scratch contains buildup at the edges the tip jumps and a comparison of forward and return scans will show slightly different values. In t hese cases it was necessary to scan very slowly (0.05 0.2 Hz) and to use the average value from the forward and return traces. One final note is that in very large scans (> 20 microns) the AFM images will begin to show a natural bowing resulting from th e bending of the microscopes peizoelectric scanner. The sample is mounted on top of the scanner and it is moved back and forth underneath the tip. The scanner can be visualized as a peizoelectric cylinder with contacts to bend it from side to side (as we ll as elongate and compress it). The movement of the scanner forms an arc which is not noticeable in small scans. Normally this arc is removed by fitting, but in the case of step height measurements these fits should not be used because they distort the image. 4 .5 Optical Measurements on Films Electromagnetic characterization is of obvious importance opto -electronic applications. Any study of the photovoltaic process in polymer solar cells would be incomplete without thorough characterization of the spec troscopic properties of the polymers. Optical measurements were performed using a dual beam Cary 500 UV NIR spectrophotometer. A Zeiss MPM 800 microscope photometer was also used for UV NIR reflectance and transmittance characterization. All measurements were taken at normal incidence in air. The Cary 500 spectrophotometer has a measurement range from 200 to 3300 nm Transmission and absorption measurements were taken by first using empty sample and reference compartments for both 100% and 0% (parasitic) correction files. After measuring for the correction files samples were placed in the front of the measurement beam and reference beam was left unobstructed. Reference measurements (blanks) were taken before and after
92 sample sets to ensure there was no instrument drift. In almost all cases measurements were repeatable with less than 1% deviation. The transmittance of the polymer film could then be calculated using: % 100 ( substrate total filmT T T (4 .1) where Tfilm is the transmittance of the film, Ttotal i s the transmittance of the film as measured on the substrate and Tsubstrate is the transmittance of the substrate. The measurement range of the Zeiss microscope is from 240 to 2200 nm. This instrument contains a precise motorized translation stage allowin g for marking and returning to the same sample position. The microscope was used to characterize the reflectance from photovoltaic cells. In this setup the light generated from the sources (tungsten or xenon lamps) passes through a beamsplitter and objec tive lens before being reflected off the sample and returning through the objective to be directed to the detector. The reflectance data is corrected us ing ( ( ( ( R P S P O Q (4 2 ) w here Q (Quotient) is the reflectance spectrum after spectral cor rection, O (object is the single beam spectra of the sample, S (Standard) is the single beam spectra of the source lamp, R (Reference) is the reflectance of the reflectance standard (Al mirror), and P is the parasitic measurement (0% reference reflectance) Absorption coefficients were calculated using the relationship where A is the absorption, x is the film thickness and is the absorption coefficient. When doing these measurements, absorption spectra were taken after background spectra wer e run with the substrates. It is noted that this method of calculating the absorption coefficient does not account
93 for the changes that arise from the modification of second substrate /air interface by the presence of the polymer, and the addition of a sec ond interface, polymer/air. When the reflectance and thin film interference are non -negligible, a more accurate method of calculating the absorption coefficient was utilized In this instance both the transmission and reflection are taken of the uncoated substrate as well as several films of different thicknesses. These can then be treated using a Drude Lorentz model for thin film analysis. This method has been documented in detail in previous Tanner lab dissertations (see, for instance, Hwang) When t he reflectance is small (as is normally the case with conducting polymers) a nother useful technique for estimating the absorption coefficient is to use the thinnest sample as a reference and thickness difference between this and the target sample. In this instance the reflectance from both samples should be the same since they both have similar interfaces and the difference in the spectra should be due to absorption of the film. This method is an improvement over simply utilizing glass as the reflectance r eference, but in regions where glass become absorbing (outside the region 360 2700nm) the absorption of the glass must be subtracted using d x dglass d uncorrecte corrected ( ( ( (4 .2) W her e (corrected) is the correct absorption coefficient of the film, (uncorrected) is the absorption coefficient of the film calculated without correcting for glass, d is the thickness of the film, (glass) is the absorption coefficient of glass and, x is the thickness of the glass substrate. Typically for photovoltaic application we are most interested in the regions from 360 1000 where glass has a nearly zero absorption coefficient. Figure 4 6 displays the optical properties of the substrates used in this work. Kintec ITO substrates were used in the fabrication of all devices. The optical properties of the ITO/glass substrates are indicated by the red curves. The optical properties of the device films were made
94 A B Figure 4 6 Optical Characterizatio n of Substrates used in this work. A) Graph contains spectra of the transmittance of glass substrates from Kintec, transmittance of ITO coated glass substrates from Kintec, and transmittance spectra of Corning microscope slide glass substrates (0215 Corni ng Soda Lime Glass). B) Corresponding absoption spectra fro each of the same substrates.
95 in regions where there was no ITO present. The transmittance and absorption of these regions are indicated by the black traces in the Figure 4 6 A -B. For purely opt ical studies where a conducting substrate was not needed, Corning microscope slides, composed of 0215 soda lime glass, were used as the substrates. The optical properties of these substrates appear as the green curves in Figure 4 6 A B. 4 .6 Device Fabrica tion Techniques and Construction 4 .6.1 Glovebox Because of the sensitivity of many of the materials used in this work, it was necessary to carry out much of the fabrication and characterization under inert atmosphere. This was done by utilizing an argon f illed glovebox designed and built by Mbraun. During regular operation this box maintained an internal atmosphere of less than 0.1 parts per million (ppm) of oxygen and less than 0.1 ppm of water. The glovebox allowed the fabrication and testing of device s without exposing them to oxygen and water vapor. Fabrication of devices inside the glovebox was done utilizing two separated section. The left section of the box (designated the wet side) wa s used for solution processing, annealing, and spin coating. The right side of the box was used for deposition of metal contacts and device characterization. The high purity (99.999%) argon used by the glovebox wa s purchased from Air Gas South. This gas is used to both operate the valves and maintain the proper w orking pressure of the box. All objects and materials introduced to the box must first be placed into antechambers that are put through several evacuation cycles. The glovebox employs several methods of purification in order to maintain the extremely low levels of oxygen and water. The first stage in the purification process is proper circulation o f Argon throughout the box. This is done by a blower which pushes clean gas into the boxes through hepafilters. Gas from the box exits and passes through se veral types of purification
96 The first stage is hepafiltration to remove dust. The gas then enters a solvent trap which removes organic solvents from the box through the use of activated carbon that is changed every 3 4 months. The gas then passes both oxygen and water sensors. (The water sensor utilizes two separated wire coil winds t hat are coated in dried acid. Water molecules passing over the these coils changes the cond uctivity of the acid. It is necessary to recoat the winds every three months t o ensure proper operation.) Oxygen and water is then removed by molecular sieve and copper catalyst pellets. The cleaned argon is then reintroduced to the box. The purification column must be regenerated every 4 6 months as the oxygen and water lev els be gin to rise above 0.1 ppm This purification process has been the source of frustrations at times and here are a few words of advice regarding its maintenance. The valves on the regeneration valve block have a housing which is made from a grade of steel that is subject to corrosion Regeneration of the catalyst involves the expunging of water molecules which in the presence of organic solvents forms acidic conditions. Rusted valves on the valve block malfunction causing pollution of the box atmosphere. (Advice: test the valves for proper operation prior to starting the regeneration.) As is documented in Chapter 1 there is a strong connection between the electronic properties of conjugated polymer films and their morphologies. For this reason a custom a nnealing and drying oven was built which allowed vacuum drying, solvent -vapor annealing, and thermal annealing inside the glovebox. The chamber was constructed from brass to ensure even heating and good vacuum capabilities. A schematic view of this chamb er is pictured in Figure 4 7 Temperature of the chamber was controlled by through the use of a temperature probe connected to a digital 730 series Dataplate. The data plate provided precise control of the chamber temperature as well as allowing for use of temperature ramps and programmable cycles.
97 An external vacuum line was used to connect the chamber to the dry iced -trapped pumping station. Figure 4 7 Top schematic view of vacuum annealing chamber halves. System uses hightemperature(400C) silic one 046 O rings. Chamber is capable of annealing under tempera ture, vapor pressure, or vacuum 4 .6 .2 Metal Vapor Deposition The Mbraun glove box is equipped with a metal vapor deposition system which made it possible to deposit contacts without removing devices from the inert atmosphere of the glovebox. Metal vapor deposition (also know as physical vapor deposition) is a process in which metals are thermally evaporated from a source and condense onto the substrate surface forming a thin metallic film. Th is process is line of -sight in that the vaporized atoms move directly from the source to the substrate.88 It requires vacuum on the order of 106 mTorr in order to reduce gaseous contaminants in the resulting metal films. The Mbraun deposition system utilizes a turbo pump with dryscroll backing pump to achieve these pressures. Metal vapor is generated inside the deposition system by passing large currents through resistive tungsten evaporation boats containing the evaporation metal Evaporation rates are monitored using two independent quartz crystal monitors. A Sigma SQC 222 codeposition
98 controller uses a feedback loop to regulate the system power and maintain deposition rates. In the original configuration of the deposition system the tungsten boat s were secured to the top of stainless steel posts through the use small (4 mm) bolts. This system resulted in multiple difficulties including the melting and galling of the bolts as well as cracking of the tungsten boats due to torques from mounting bolt s. I t was redesigned replacing the thin steel posts with thick copper posts and a lever -camming boat mounting system. This design worked well, with the occasional (annual) need to clean deposited metal from the cams by sandblasting. This system was used to deposit gold, silver, lithium flouride, and Aluminum films onto devices. To ensure precise placements of contact evaporation patterns, a shadow mask and device holder was designed to accomodate 25 x 25 mm masks and devices as displayed in Figure 4 8 Following the examples set forth in the literature, evaporation of gold and silver was completed at 1 /s, lithium fluoride was deposited at 0.1 /s, and Aluminum was deposited at 2 /s. Typically 100 nm thick films were used for contacts. Figure 4 8 E vaporation mask and substrate holder designed for precise alignment of masks on substrates. Al ignment grooves were cut to 25.2 mm to accomodate 25.0 mm masks
99 and substrates. Circular notches cut into the outside edge allow for the removal devices and mas ks which reduces the likelihood of scratching. 4 .6 .3 S pace -Charge -Limited -Current Device Fabrication Carrying out the fabrication and characterization of devices inside the glovebox solved many reproducibility pro blems found in early work. In C hapter 1 th e theory and principles of space -chargelimited -current (SCLC) were developed. Hole -dominated devices were created by layering polymer films between two precisely patterned metal contacts in a sandwich style configuration as is illustrated in Figure 4 9 A Due to the many orders of magnitude difference between the conductivity of the semi -conducting polymer film and the top and bottom contacts conduction is limited to the regions of overlap of the contacts. The regions of overlap are referred to as activ e areas or p ixels These are illustrated in Figure 4 9 B. Typical polymer films used in the devices ranged in thickness from 100 to 300 nm The pixels width was either 3 or 5 millimeters, making the devices 6 orders of magnitude wider than they were t hick e ssentially result ing in two -dimensional verti cal transport through the film. A B Figure 4 9 SCLC Device Configuration. A) A side view of a pixel showing a thin polymer layer sandwiched between two metal contacts. B) Top view of a metal conta ct patterns used in the fabrication of SCLC devices. Devices for SCLC measurements were created using either a gold or ITO pattern to act as the bottom contact. Prior to purchasing patterned ITO substrates patterning was achieved through an etching process This process started with 25mm squares of ITO -coated glass
100 (Sigma) with nominal sheet resistances of 8 12 slides using 3M packaging tape. A positive etch pattern was produced by cutting the tape and pealing away areas from which ITO was to be removed. The masked slide was then acid -vapor etch ed by suspending it u pside -down above freshly prepared aqua regia solution (fuming mixture of 25% concentrated nitric acid and 75% concentrated hydrochloric acid) Approximate ly 10 minutes of exposure adequately removed unmasked ITO from the glass surface. The devices were t hen thoroughly rinsed with distilled water before removing the masks. Devices were then cleaned as described earlier in the chapter by using surfactant and cotton swabs, sonication in surfactant solutions, rinsing, and sonication in and isopropanol. Patterned ITO substrates were then blown dry with filt ered nitrogen gas and treated for 15 minutes in an o xygen plasma clean er Devices fabricated using substrates purchased with pre -patterned ITO were cleaned follow ing the same process. For devices in which gold was used as the bottom contact, 25mm squares of microscope glass (Corning) were cut and cleaned using the same process as used with the ITO substrates. The cleaned glass was then immediately transferred int o the glovebox evaporation chamber for metal vapor deposition of t he gold pattern. To create the pattern a n aluminum tray was designed with the p attern illustrated in Figure 4 .10 A machined into the aluminum. Figure 4 .10 B shows a three dimensional versi on of the tray. The black regions represent places where the metal was cut all the way through. The remaining regions acted as a mask to shadow the remaining regions of the 1 x 1 s ubstrates Utilizing this shadow mask allowed for very precise control of the device active area. Following the cleaning steps, a thin layer of PEDOT:PSS was deposited in most transport devices. This layer acted to reduce the work function of the bottom contact as well as smoothing
101 A B Figure 4 10. S chematic diagrams of aluminum evaporation holder used for creating substrates with regions of patterned contacts A) Dimensionalized schematic in which the black region designates the area of the substrate onto which gold will be deposited and the white regions represent masked regions B) 3 -dimensional Schematic diagram of the evaporation tray. Glass substrates were placed in the tray for deposition of patterned gold contacts.
102 the ITO surface. Aqueous PEDOT:PSS solution also wets ITO better than most organic solutions and it was found that devices formed containing PEDOT:PSS layers were much less likely to contain shorted pixels. The films were deposited at either 4000 or 5000 RPM for 60s resulting in approximately 35 and 40 nm thick layers. The devices were then drie d at 110 C under vacuum for 1hour before transferring to the glove box. Hole -transport layers were spin coated inside the glovebox at rates ranging from 500 to 2000 RPM for 30s, using 3s ramps(except where noted), followed by 30s at higher rates(to increase surface uniformity and aid in drying the films) from polymer solutions varying from 5 to 25 mg/mL in chloroform, chlorobenz en e, or dichlorobenzene. Figure 4 11. SCLC shadow mask design. Masks create 3 x 3 mm regions of overlap between the top and bottom contacts on SCLC devices. Devices were dried and/or pre annealed in the vacuum annealing chamber before transferring into evaporator for contact deposition. To create top contacts on the film surfaces which directly overlapped the bottom contacts as displayed in Figure 4 8 a shadow mask was designed and machined as shown in Figure 4 11. Devices were placed face down on the masks in the evaporation tray and positioned inside the evaporation chamber. After pumping down to
103 pressures on the order of 1 x 106 mTorr gold contacts were deposited at 1 /s. Devices were then removed from the evaporator for characterization. Current -Voltage characteristics of the devices were taken under dark conditions (to eliminate photo-generated carriers ) u sing a Keithly 2400 sourcemeter with identical leads. Curves were fit to field -dep endent SCLC model described in C hapter 1 using a least squares fitting routine as will be discussed in Chapter 3. 4 .6 .4 P hotovoltaic Device Fabricat i on Fabrication of photovoltaic device s was done in a similar manner to SCLC devices. Figure 4 1 2 A illustrates the layer ed configuration of the devices. (Note that a cross -bar design was not used for these devices because it has been shown to give inaccurate results due to current contribut ions from other pixels. More details regarding this pattern and the patterns used in this work are given in the following subsection. ) Pixels with active areas of 0.25 cm2 were formed from the overlap of the contacts as depicted in Figure 4 1 2 B. Typica lly devices were constructed starting with pre -patterned ITO purchased from Kintec. Following the same cleaning method as was used for SCLC devices, a 35 40 nm layer of PEDOT:PSS was deposited in the cleanbox at 4000 or 5000RPM The devices were dried at 110 C under vacuum for 1hour before transferring to the glove box. Polymers were blended with PCBM to form solutions in chloroform, chlorobenzene, or dichlorobenzene. Typically the concentrations of these solutions were around 15 mg/mL but they were va ried from 8 to 25mg/mL during device optimization experiments. Solutions were stirred overnight or longer to ensure they were fully dissolved. Spin coated active layers were deposited at spin rates from 500 to 2000 RPM for 60 seconds, using 3 second ramps. Spray casting was also used to form active layers and the devices were sprayed from
104 A B Figure 4 1 2 Photovoltaic device configuration. A. Illustration showing device layers: glass substrate, ITO, PEDOT:PSS layer, bulk heterojunction blend layer, and aluminum contacts. Squares are drawn to indicate pixels. B) Two dimensional view of the regions of contact overlap forming the active regions. See following sub section for more details on photovoltaic device patterns. solutions of the polymer in chlor obenzene at concentrations of 17 or 18 mg/mL. The active layers were then annealed or dried using the vacuum annealing chamber inside the box before transferring to the evaporator for deposition of the c ath ode Shadow masks that were machined to dimensio ns shown in Figure 4 1 3 were used for the cathode deposition. Devices were placed on top of the masks with the film side down. The deposition chamber was evacuated to pressures on the order of 1 x 106 mTorr. Some devices received a thin layer(0.5 nm ) o f lithium fluoride prior to deposition of 100 nm aluminum contacts. Devices were then removed from the evaporator for c haracterization and in some cases, post -fabrication annealing. Photovoltaic devices were characterized under A.M. 1.5 conditions from 15 0 Watt xenon powered solar simulator (Newport 66902 lamp housing and Newport 66907 power supply.) The radiation from the xenon lamp passes through a collimating lens before being filtered by consecutively stacked A.M. 0, A.M. 1D, and A.M. 1.5 filters (New port). Calibration of the simulator was done using a calibrated silicon photovoltaic cells and a calibrated thermopile
105 Figure 4 1 3 Photovoltaic s hadow -mask design. Masks create 0.5 x 0.5 cm pixels on photovoltaic devices where Aluminum is deposited in the black regions and the white regions are effectively masked by the metal. detector (Newport 70260). Current -Voltage characteristics of the devices were taken under both dark and illuminated conditions. Efficiency fill factors, and other characteristics were calculated according the methods described in C hapter 1. The incident photon conversion efficiencies were also characterized by using a Newport 70612 IPCE equipped with a Cornerstone 130 monochrometer. The IPCE was calibrated using a calibrat ed Newport Oriel thermopile detector. The detector was carefully positioned and equipped with a specially fabricated anodized aluminum mask identical in shape and size to 5 x 5 mm pixels of the devices. Photogenerated current was recorded as a function of monochromatic illumination and efficiencies were calculated based on the relation: (%) 100 (%) P e I c h IPCE (4 .3) Where e is the charge of an electron, h is Planks constant, c is the speed of light, I is the current, is the wavelength, and P is the power of the incident light.
106 4.6. 5 Details On Photovoltaic Device Patterns The continuous design of the aluminum used in fabricating the devices gave accurate characterization of individual pixels. The reason for use of the continuous aluminum contac ts present in the photovoltaic design is due to the historical use of the cross pat tern design shown in Figure 4 14 A. It was realized early in the work presented in this dissertation, that the crossbar pattern gives inaccurate results because of contribut ions from alternative current pathways (shown in Figure 4 13 B). These pathways were eliminated by changing the ITO pattern to the new one shown in Figure 4 1 2 To ensure the new design did not contain external current contributions the pixels were characterized both before and after separating the aluminum contacts by sc ratching as shown in A B C D Figure 4 14. Considerations in photovoltaic device design. A) Crossbar pattern (used historically). A) Alternative current pathways in terfering with individual pixel characterization. C) Method to test the design used in this work. A scratch was made across the aluminum contact to separate all pixels. As opposed to the old configuration in A the configuration in C showed no difference before and after scratching the aluminum. This confirmed that the results being measured were due to individual pixels. D) to create a cleaner design, future devices will be made with this pattern, for dimensions of the masks see F igure 4 14.
107 Figure 4 14 C. There was found to be no difference between devices with continuous aluminum compared to those with separated aluminum. Still, to eliminate all possibilities, and simply to present a cleaner looking design, new shadow mask designs are being fabric ated in which the aluminum is separated similarly to the ITO (Figure 4 12 D). The old and new mask dimensions are shown in Figure 4 15. A B Figure 4 15. Photovoltaic shadow -mask design. A) Mask used in this work. B) New mask designs that are underway Pixels are completely separated in the new design. The slight offset of the new masks is to ensure complete overlap of the aluminum and ITO contacts while allowing for the small amount of offset that can occur in the evaporation tray. 4 7 Documentatio n of Polymer GPC and NMR Characterization This section contains GPC and NMR data obtained for most of the materials used in this work. Those that are not documented in this chapter can be found in the relevant published literature that is referenced in th e relevant sections. Figure 4 16 shows GPC analysis of the three different P3HT polymers used in this work. This data shows the molecular weights to be ~29 kg/mol, ~30 kg/mol, and 76 kg/mol, respectively. Figure 4 17 contains the GPC analysis of PTVBT The molecular weight of the polymer was found to be ~130 kg/mol. The molecular
108 weight PBTC was found to be ~6 kg/mol, and the GPC data for this polymer is shown in Figure 4 18. Nuclear magnetic resonance (NMR) imaging data was also taken for each of thes e polymers. Figure 4 19, 4 20, and 4 21, show the hydrogenNMR analysis of P3HT1 P3HT2, and P3HT3, respecitively. These images were used to estimate the degree of regioregularity (which is dicussed in Chapter 3), as well as to confirm the polymer struct ures. The H NMR spectrum of PBTC, is shown in Figure 4 2 2
109 A B C Figure 4 16. GPC analysis performed on the three different P3HT samples used in the mobility studies of Chapter 3 showing their measured molecular weights.
110 Figure 4 17. GPC analysis performed on PTVBT The polymer was used in both mobility and photovoltaic studies. Figure 4 1 8 GPC analysis performed on PBTC. This polymer was also used in both mobility and photovoltaic studies.
111 Figure 4 1 9 H NMR spectrum of RR -P3HT1
112 Fig ure 4 20. H NMR spectrum of RR -P3HT2
113 Figure 4 21. H NMR spectrum of RRa P3HT3
114 Figure 4 22. H NMR spectrum of PBTC
115 CHAPTER 5 CONDUCTIVITY AND MOBILITY 5 .1 Material Properties and Carrier Mobility Charge carrier mobility can play an important role in the optimization of conjugated polymer -based electronic devices, such as organic light emitting diodes (OLEDs), thin film transistors (TFTs), and organic solar cells. Thus, characterization of charge carrier mobilities in new polymers is important in det ermining their potential use in applications. B efore delving into the discussion of experimental techniques, a sh ort but important section will be included on material properties and the nature of carrier mobility. 5.1.1 Intrinsic and Extrinsic Properties The current flow through an object is dependent on the potential difference across it. For many objects it is found that this relationship is linear over a broad range of potential differences. This statement is known as Ohms l aw an d the object is said to behave ohmically throughout the region for which the relation remains linear. In the o hmic region of the object the relation may be specified by the equation This statement defines a value, R, called the resistance which is constant in the ohmic region and is an extrinsic property of the object. Measurements of the resistance of objects made of a given material often lead to the conclusion that a s imple relationship exists between the geometry of the object and the measured resistance. The existence of this relationship allows the defining of an intrinsic property of the material known as the conductivity. This property is a measure of how easily charge can flow through the material and it is defined by the relation: ) / ( A R l in which l indicates the length of the object and A indicates the cross sectional area of the object. T he resistance is a property of the object, dependent upon the object s dimensions. In regions where it is well defined, the conductivity is an intrinsic property of the material; it is independent of dimensions.
116 The conductivity of a material only remains a useful quantity in cases where the simple relation between resistance and geometry holds. Using the intrinsic properties of a material, the relation between the flow of charge and the potential difference can be rewritten. If da represents an infinitesimal surface within a material the flow of current a cross this area may be represented by a current density, J The value of the current density is defined to be ) / ( da I d J where da is the surface area perpendicular to the current flow. The thre e di mensional version of Ohms l aw is )) ( ( r J (A simple integration where a d J I w ill show the equivalence to the V = IR form.) A mo re common way to express Ohms l aw is: E J in which the relationship be tween the flow of charges and the electric field driving their movement is apparent. In a segment of time dt the current density represents the amount of charge crossing a cross sectional area. It is due to the flow of free charge carriers in response to the presence of the electric field. Because holes propagate through the conduction band and electrons propagate through the valence band, their velocit y distributions will be different. If the average velocity of holes is vh and the average velocity of electrons charge carriers is ve then the total charge passing a cross section in time dt is, A dt v en A dt v en dqe e h h) ( ) ( (5 .1) where A is the area of the cross section, and e is the charge of an electron. Since J is the charge per unit area per unit time, ) (e e h hv en v en J (5 .2) where n represents the number of carriers per unit volume.
117 If it is assumed that the drift velocity of the carriers is proportional to the electric field, then the current density can also be written as E n E n Je e h h The quantity is known as the mobility and it specifies the relationship between the drift velocity of the carriers and the electric field, E vdrift / The mobility is also an intrinsic property of a mate rial and compariso n with Ohms l aw shows that it is related to the conductivity through e e h hen en (5 .3) In g eneral, most materials are not o hmic, and the intrinsic properties such as the conductivity and the mobility will depend on other factors such as temperature, carrier density, and the electric field. The values of mobility and conductivity can also strongly depend on direction. In these cases, t he constants representing the intrinsic properties must be replaced by functions whose values vary. The intrinsic properties of a material cannot b e directly measured. The value in defining intrinsic properties rests on the assumption that a simple relationship exists, which allows the intrinsic properties of a material to be determined from measurements of the extrinsic properties of a sample. The more thoroughly that an intrinsic property is specified (i.e. dependence on field, temp, etc) the more difficult it becomes to extract that property from measurable quantities. Another important point is that b ecause an intrinsic property is a property of the material independent of dimensions, the extraction of that property from a measurement assumes the homogeneity of the underlying material. This section has been included here because the mobility of a mater ial is an intrinsic property which must be inferred from extrinsic measurements. It is not easily done, particularly in the case of conjugated polymer systems. In addition, can be highly dependent on both
118 temperature and electric field, making the relation, E T E vdrift ) ( complicated, and diminishing the conceptual value in defining At best, mobility values for conjugated polymers must be viewed as approximatio ns, whose values should be considered along with the technique used to obtain them. 5.1.2 Challenges in Determining Charge Carrier Mobility in Conjugated Polymers There are several challenges inherent to all mobility characterization techniques. One of co nsiderable importance for conjugated polymers is the strong influence that film morphology has on transport properties, and the significant variations in film morphologies that can result with small changes in processing conditions. For this reason, methods of film formation should be an important consideration when comparing mobility values. Another factor making mobility determination challenging in conjugated polymers is the way in which interactions between the lattice and the charge carriers affect the carrier mobility. The underlying movement of charge in conjugated polymer systems was discussed in Chapter 1, where the chargelattice coupling effects were mentioned. This coupling effect significantly complicates the mobility properties of soft m aterials in comparison to those with crystalline lattices. In addition, the bonding between atoms within a polymer chain is very different from the way in which the atoms are connected to atoms outside of their chain. This difference inevitably leads to variations in intrachain and interchain transport properties. It is generally accepted that intrachain carrier transport happens on a much faster time scale then interchain transport.89 The transport process can be considered to be compos ed of rapid intrachain diffusion and slow interchain hopping steps. One last note of importance in the determination of polymer mobility values is that all samples are different. The polymer characteristics can vary from batch to batch. Chain lengths
119 in particular are an important property High molecular weight polymers will behave differently from low molecular weight polymers. The polydispersity index (a measure of the breadth of the molecular weight distribution) can also influence transport. These characteristics should be consider ed when comparing mobility values. 5 .2 Methods of Mobility Determination There are three principle techniques used to determine conjugated polymer mobilities: time of flight measurements (TOF), modeling of I -V characteris tics from TFTs, and modeling of the I-V characteristics from space-charge limited -current (SCLC) diodes. Each of these techniques can be useful in probing charge carrier dynamics in conjugated polymer systems. In addition, each presents its own set of ch allenges and drawbacks. 5.2.1 Time of Flight M easurements Possibly the most direct way of determining the mobility of charge carriers in a bulk polymer sample is through use of the time of flight method. F igure 5 1 illustrates the experimental setup. The TOF method uses a device structure similar to a solar cell in which there is a transparent conducting electrode on one side of the device.8991 A polymer film is sandwiched in between the conducting electrode and a thermally evaporated back electrode. A co ns tant potential is then applied across the device to create an electric field within the material. A short laser pulse (5 10ns) at a wavelength strongly absorbed by the polymer is directed through the transparent electrode. To ensure that the carriers all travel similar distances, the thickness ( L ) of the polymer film is typically mad e significantly larger than the polymers optical absorption depth (inverse of the absorption coefficient).92 If a n oscilloscope is used to monitor the pho tocurrent generated by the pulse the mobility can be extracted from the transit time using, ) /( /2 transit driftt V L E v where vdrift is the drift
120 Figure 5 1 The schematic setup used for a TOF mobility measurement. A short laser pulse is used to generat ed charge carriers. These carriers are separated, and transported to opposite electrodes by an applied electric field. The transit time is monitored through use of a high frequency oscilloscope. velocity, E is the applied field, V the applied voltage, L is the polymer later thickness, and ttransit is the measured transit time. By varying the sign of the bias voltage, the movement of either electr ons or holes can be selected. Varying t he magnitude of the bias voltage can be used to determine the field de pendence of the mobility. The transit time can be extracted from the observed photocurrent. In an amorphous material t he d isplacement of the average position of the charges can be represented by the Scher Montroll model,28 in which x (t) = t for ( 0 < < 1 ), whre x is the density t is time and is a constant. The current will be proportional to the derivative, I ~ t 1, for times less than the transit time, and I ~ t ( 1 ), for times greater than the transit time (since after t = ttransit charges will be reaching the electrode). If the curre nt is plotted versus time on a log -log scale there will be two linear regions of different slope from which the intersection will give the transit time. This is shown in Figure 5 2 where the simulated data takes the general form of TOF data. The TOF metho d for determining charge carrier mobility is probably the most valuable method because it offers the most direct path towards mobility extraction. It also is
121 Figure 5 2 Simulated example of TOF data. Photo current shows two distinct regions of slope from which the transit time for charge carriers may be extracted. Real data will be more noisy, but it will follow this same general form. The transit time will vary with the applied voltage allowing for the extraction of mobility at different electric f ields. advantageous in that it measures mobility perpendicular to the substrate surface. This is the direction of interest for solar cells since their layered configuration also results in transport perpendicular to the substrate surface. Another nice fe ature of this method is that both hole and electron transport can be analyzed in the same device simply by reversing the bias. There are also drawbacks to the TOF method. One important drawback is that the films needed for TOF are of much greater thickne ss (typically > 10 times thicker)28 than those used in actual solar cells. This means that the parameters used for the TOF film formation will be different from those used to make solar cell films. Since the film morphologies of most materials show a str ong dependence on film -formation parameters, there is no guarantee that the films will be equivalent to those used in solar cells. Another challenge to this approach is the equipment required to perform the measurements. Purchase of a laser and an oscill oscope capable of making the measur ements is relatively expensive.
122 5.2.2 Modeling of Field -Effect Transistor Characteristics Probably the most common method of characterizing polymer mobility values is through modeling the characteristics of TFTs.93 These measurements are of significant importance for improving the drive current in transistors. Polymer -based TFTs have the advantage over tradition al transistors in that they can be more easily used for large area electronics (LAEs). Combined with the low -cost of polymers, these advantages will inevitably mean that polymer transistors will continue to play an important role in display technology. The drive current and the operating frequency of transist ors are limited by the device geometry along with the charge carrier mobility.94 In inorganic transistors, improvements can be made through decreasing c hannel length. The precision of the printing process used in low -cost organic transistors limits the amount by which the cannel length can be decreased. This makes the carrier mobility a key factor in organic TFTs. Figure 5 3 shows the configuration of a typical top contact transistor. The voltage of the gate electrode modulates the resistance of the channel. Almost all polymer based transistors are p type charge accumulation devices, meaning that the transistor is turned on by increasing the concentrat ion of holes through application of a gate voltage. The concentration of holes is related to the gate voltage through, NCH = C0 |VG VT| where NCH represents the concentration of hole carriers, C0 the capacitance of the gate dielectric, VG the gate vol tage and VT is the threshold voltage. For VG greater than the threshold voltage the transistor is turned on. When the |VSD| < | VG VT | the transistor is in the linear regime and the drive current, ISD, is given by the gradual channel approximation, SD SD T GV V V V C L W 2 I0 SD (5.1)
123 Figure 5 3 Typical layout of a top contact transistor showing the dimensions of the channel following the typical convention. The voltages of the gate and source are measured relative to the drain. A dielectric mater ial separates the conductive gate from the semiconducting polymer layer. where W is the channel width, L is the channel length, and is the charge carrier mobility. If the source -drain voltage is increased to the point where |VSD| > | VG VT | the devic e enters the saturation regime and the current becomes independent of VSD. This behavior happens because the channel is pinched off near the drain contact.93 In this region the drive current simplif ies to: 2 0 SD2 IT GV V C L W (5 .2) If the square root of the drive current is plotted versus the gate voltage, the mobility can be extracted from the slope of the curve Analysis of the transistor characteristics can also be used to calculate the f ield effect mobility by using: 0 0 FE) ( DV G D D GV I V WC L V (5 .3) There is a large amount of research that has been published on TFTs and the method used for mobility extraction is common. However, mobility results from real TFTs are not always consistent The mobility values extracted from TFTs are well known to be several orders of magnitude larger than those determined from other methods.28 It is typical to explain this discrepancy by citing that TFTs measure in plane mobilities whereas other methods measure
124 mobilities perpendicular to the subs trate. But, mobilities measured in amorphous materials also show this discrepancy and morphology cannot account for this. A possible explanation is related to the effect of carrier concentration.38 The fact that TFTs measure mobility in -plane as opposed to perpendicular to the surface makes their value of reduced applicability to solar cells where the transpo rt is perpendicular. 5.2.3 Field -Dependent Space -Charge -Limited -Current Modeling The technique of modeling conjugated polymer diodes using the Murgatoyd field dependent space -charge-limited -current model has become an increasingly popular way of characteri zing charge carrier mobility. This technique is built on several assumptions that can affect the validity of the results. The voltage and electric field used in the Murgatoyd model of Eq. 1.14 are representative of the voltage across t he semiconductor ma terial. The model does not account for the possibility of v oltages drops at th e contacts. It assumes o hmic contacts, which may or may not be the case in real conjugated polymer devices. A second assumption of the model is that only one charge -carrying species is present. The presence of both free electrons and holes in the material would provide overestimated values for the extracted carrier mobility. These issues are addressed in the diode fabrication process by selecting electrodes with work functions as close as possible to either the conduction band or the valence band depending on whether the intent is to measure hole mobility or electron mobility. Figure 5 .4 A illustrates the ideal choice of electrode to create a hole dominated device by choos ing a metal with a work function at the valence band of the semiconductor. Figure 5 .4 B illustrate s the choice of low work functions contacts that would be used to create an electron-dominated device. It is assumed that population of the conduction band by electrons will be negligible if EG >> kBT (which at room temperature is ~0.025eV).
125 A B Figure 5 4 SCLC diode contacts. A) To create a diode to study the transport of holes a high work function electrode is used that will create an inject ion barrier for electrons. B) To create an electron dominated device a low work function metal is used that will create an injection barrier for holes. Mobility is extracted from the diodes by measuring the I -V characteristics of the d evice under dark conditions (to eliminate the possibility of photo induced charge carriers), and then fitting the resulting characteristics to the Murgatoyd SCLC model (described in the third section of Chapter 1) : 3 2 0 0 0) 891 exp(. 8 9 L V E Jr (5 .4) The model can be used to extract the mobility if the material and device parameters are known. An easy way to fit the data to this model can be realized by rewriting the field dependent SCLC equation as a linear expression of the variables J L and V Rearranging Eq 5 .4 gives, L V V JLr 891 exp 8 91 0 0 0 2 3 (5 .5) Taking the natural log of both sides and rearranging the terms results in, 8 9 ) 891 (.0 0 0 2 3 rLn L V V JL Ln (5 .6 )
126 If the diodes exhibits space charge limited current, then plotting quantity, L n (JL3/V2) versus (V/L)1/2 will res ult in a straight line in the SCLC region. This allows for easy extraction of the carrier mobility 0, and the field dependent prefactor, from the slope and y intercept using 891 0 slope and 0 0 09 8 r exp( y -intercept) (5 7 ) Typically, it is assumed that the trap depths are small and the factor 0 is dropped. A unique ad vantage of field -dependent SCLC modeling is that it allows the measurement of material properties in configurations vary similar to those used for solar cells. Another advantage is that it doesnt require sophisticated or expensive equipment to make the m easurements. In principal, any group characterizing bulk heterojunction cells will have access to the necessary equipment. One limitation of the method is that electron and hole mobilities cannot be measured in the same device. Another drawback is that t he carrier concentration may exert a significant influence on the mobility and separating the field dependence of the mobility from the carrier concentration dependence of the mobility cant be done by any straightforward method. Another factor affecting the accuracy of the results is that the mobility is dependent on the third power of the film thickness. Typically, films on the order of 100200 nm are used, and determining the film thickness can be a source of inaccuracy in the calculation of a polyme rs mobility. Considering the challenges to the SCLC mobility method, the accuracy in determining the mobility of a polymer may be limited. But the method is valuable, if in the very least, as a means of comparison between polymer samples We have employ ed it in this work to examine the mobility of a number of different polymers. The results of these studies are presented in the remainder of this chapter.
127 5 .3 Mobility of Regioregular and Regiorandom Poly -3 -hexyl -Thiophene In recent years, poly(3 -hexylthi ophene 2,5-diyl), know n commonly as P3HT, has been one of the most heav ily studied conjugated polymers. T his is partially to the high level of performance of P3HT in photovoltaics and field effect transistors. Regioregular forms of P3HT (RR P3HT) have be en found to achieve hole transport mobilities as high as 0.16 (cm2/Vs) in FETs.36,95,96 These relatively good mobilities are directly linked to the morphology of RR -P3HT films, where the p olymer displays a strong tendency to pack in highly-ordered forms. In the second section of Chapter 1 the regioregular ity of poly(alkylthiophenes) was discussed. Pure unalkylated polythiophene is not soluble, and so, the addition of alkyl chains is ne eded to achieve the highly desirable property of solution processibility. Figure 5 5 shows the structure of several chain segments of poly(3 alkylthiophene). In this figure the alkyl branch chains are designated by the symbol R. For the case of P3HT R represents a six -carbon long chain. The length of these chains means tha t there will be appreciable overlap between the alkyl groups on neighboring monomer units. The alkyl chains on the monomer units of the polymer segment in Figure 5 5 A are all orient ed in a symmetric fashion. This configuration is representative of the structure of RR P3HT The alkyl chains on the monomer units of the segment of Figure 5 5 B are all oriented randomly, which is representative of regiorandom P3HT (RRa -P3HT). Interac tions of the hexyl chains are especially prominent in RRa -P3HT, where they can lead to conformational twists in the polymer backbone. In turn, these conformation changes result in electronic effects through decreases in pi overlap, as well as morphologic al differences, due to decreased packing order. These effects are evident in both the charge carrier mobility and the optical properties of the materials.
128 A B Figure 5-5. Regioregularity in P3HT. A) RRP3HT has consecutive head-totail head-to-tail. B) RRaP3HT has a variety of linkages. The presen ce of head to head linkages results in steric interactions that can decrease eff ective conjugation length and affect packing. In this work, the field-dependent space-ch arge-limited-current model was employed to study the hole transport properties of three different P3HT polymers: a highly-regioregular (RR ~ 98.5 %) polymer, a regioregular polymer (RR ~ 94 %), and a regiorandom (RR ~ 55%) polymer. In this section the polymer sample with 98.5% regioregularity will be referred to as P3HT1, the sample with 94% regioregularity will be referred to as P3HT2, and the regiorandom sample will be referred to as P3HT3. The results of these findings along with relevant optical and morphological data are presented in this sec tion. Additionally, a discussion is included comparing the results with other work in the field and its relevance. The three versions of P3HT used in this study were all purchased from Reike Metals Inc. The polymer samples, P3HT1, P3HT2, and P3HT3 were specified to have molecular weights of ~40 (kg/mol), ~50 (kg/mol), and 70 (kg/mol), consecutively. The regioregularity of P3HT1, and P3HT2 were specified at > 98% and ~94%, consecutively. The regiorandom sample, P3HT3, did
129 not have a specified degree of regioregularity. In order to confirm these results GPC and NMR measurements were done on the polymers. Our NMR measurements found the regioregularity of P3HT1 P3HT2 and P3HT 3 to be ~98.5%, ~94%, and 55 % consecutively We found the molecular weights of P3HT1, P3HT2, and P3HT3 to be ~ 29 (kg/mol) 30 (kg/mol), and 76 (kg/mol) consecutively See Chapter 3 for NMR and GPC results. 5.3.1 Optical Properties of P3HT1 P3HT2 an d P3HT3 The optical properties of the polymers were examined through UV -Vis to NIR transmission measurements. Figure 5 6 displays the absorption spectra of films that were spin coated on glass substrates. The spectra have been corrected for the absorption of the glass substrate (see Chapter 2 for UV -Vis to NIR absorption spectra of the substrates used in this work.) The plots show a clear red -shift in absorption, correlating with increasing degree of regioregularity. The absorption maxima are indicated in the graph, where they can be see to shift from 444 nm in P3HT3 to 516 nm in P3HT2 to 552 nm in P3HT1 Figure 5 6 UV-Vis absorption spectra for thin P3HT films, where the increased regioregularity has an obvious impact on absorption specta.
130 In compa rison with the spectra of P3HT3 b oth spectra from polymers P3HT1 and P3HT2 show more structure associated with higher degrees of ordered packing. The spectrum from P3HT2 shows the emergence of shoulders around 490 nm 550 nm and 600 nm The spectrum o f P3HT1 display more finely resolved peaks The shoulde r around 490 is barely apparent, but t wo independent features are evident at 520 and 552 nm as well as another around 610 nm In comparison to the P3HT2 spectrum, i t seems that the higher degree of regioregular ity in P3HT1 has lead to greater increases in the intensity of the peak at 552 nm which has surpassed the peak at 520 nm and become the new maximum. The absorption features in the spectra of regioregular p3ht (although the degree of regioregul ar ity only reached ~96%) were analyzed in a study by Brown et al. in 2003.95 They give a convincing argument which concludes that the longest wavelength feature (visible in both the P3HT1 and the P3HT2 plots of Figure 5 6) was due to an interchain absorption state and that the featu re intensity increases with increased order. They conclude that the t transition is therefore represented by the slightly higher energy peaks occurring around 550 nm. The absorption spectra can be used to give an optical estimation of the band gap for each of the three polymers. Figure 5 7 show s the fits estimatin g the band gaps of t he polymers. The band gaps of P3HT 1 and P3HT2 were both found to occur at 644 nm. The band gap of the P3HT3 was found to reside at 557 nm, indicating the effect of the energetically unfavorable steric interactions. Figure 5 7 D shows the spectra of all three samples with near -zero absorption well into the IR region. The absence of polaron absorption peaks in this region is indicative of the neutral purity of the polymers. The presence of impurity states can radically change the cond uctivity of conjugated polymers. High purity is essential in preparing diodes for the investigation of hole transport properties.
131 A B C D Figure 5 7 Band ga p region of pure P3HT films on glass as well as o ptical band gap estimates of P3HT1 P3HT2 and P3HT3 A) For the highly RR P3HT1 sample the estimated band gap occures at 644 nm B) Similarly to P3HT1 the optical band gap of P3HT2 occurs at 644nm However, the optical density of P3HT2 is blue -shifted relative to highly regioregular P3HT1 C) The optical band gap of RRa P3HT3 is significantly blue shifted relative to the RR forms and occurs at 557 nm. D) Absorption spectra of the films extending into the NIR. Lack of absorption peaks in the band gap region is indicative of material purity 5.3.2 SCLC Mobility Characterization of P3HT 1, P3HT2 and P3HT3 The HOMO level of P3HT is accepted to be at about 5.0 eV and PEDOT:PSS, which has a work function around 5.1 eV can be used to create a close match of the energy levels for minimum barrier hole injection .96 Similarly, gold has a work function about 5.0 eV and will also create a close match with the HOMO level of P3HT.36 To examine the carrier mobilities of P3HT1 P3HT2 and P3HT3 diodes were fabricated in the layered structure, ITO/ PEDOT:PSS/ P3HT/Au. As a method of verifying the thickness
132 scaling of the mobility, devices were made with four various thicknesses for each polymer sample. In total, twelve diodes (each one containing four pixels enabling individual measurements), were fabricated and studied. The details of the fabrication process were described in Chapter 3 Due to the unstable nature of the polymer, all fabrication and measurement steps beginning with the deposition of P3HT were done inside the inert atmosphere of an argon-filled glovebox. The I -V characteristics of each device were measured under dark c onditions to eliminate the possibility of photoinduced charge carriers. F ollowing the first I -V characterization, all devices were annealed for 15 min at 150 C as a method to investigate the effect of thermal annealing on carrier mobility After post a nnealing measurements were complete, all devices were removed from the glovebox for further optical and morphological studies. Absorption measurements were made on each of the device s w ithin one hour of removal from the box The absorption was corrected f or the p resence of the PEDOT:PSS layers by subtracting the spectrum of a an independently prepared and measured PEDOT:PSS film. PEDOT:PSS has very limited absorption in the UV -Vis to NIR region and its effect on the absorption measurements is minimal. Th e films of the devices were then examined through tapping -mode atomic force microscopy measurements. Surface scans were made to investigate for the presence of nano-scale morphology features. Additionally AFM ste p height measurements were used to determ ine the thickness of the films. The relative absorption maxima of the spectra were used to confirm the accuracy of AFM film thickness measurements and were found to be within ~5% agreement. The absorption spectra appear in Figure 5 8.
133 A B C Figure 5 8 P3HT1 diode film absorption spectra measure d to verify AFM film thickness measurements. A) absorption spectra of P3HT1 films. B) Absorption spectra of P3HT2 films. C) Absorption spectra of P3HT3 films. 5.3.3 SCLC Results Prior to extracting the mo bility from the device characteristics, the built in voltage and the voltage loss due to resistive effects w ere determined This was done so that they could be subtracted from the bias voltage yielding values that would correctly represent the true volta ge across the semiconducting polymer layer For each device, t he value was extracted from the best -fit conditions in regions where the electric field exceeded 9.0 x 106 V/m. ( Plots of the data showed that at fields a bove th is value, the diode behavior wa s representative of space -charge limite d transport ) The extracted values for the built in voltage and the resistive voltage drop were found to be small, averaging 0.03 V for devices prior to annealing and 0.07 V for annealed devices. These values a re co nsistent with the fact that o hmic injection from PEDOT:PSS should provide a slightly positive built in voltage due to the slightly lower work function of gold. Because the extracted values for the voltage loss should be independent of layer thickness th e se average values were subtracted from all bias voltages used in each of the subsequent field dependent SCLC mobility modeling Figure 5 8 shows plot s of the characteristics from the diodes made with P3HT1 As indicated by t he field -dependent SCLC equatio n, written in the form of Eq. 5 .7 of this chapter,
134 regions where the current is space charge -limited should show linear behavior. Plotted in this manner, the diode characteristics displayed in the figure show a transition region between v alues of ~20003000 V1/2 m1/2 (or equivalently, electric fields between 4 to 9 MV/m) The linear behavior, apparent in each of the curves, indicates that above the transition region the diodes characteristics well -represent field dependent SCLC. A B Figure 5 9 P3HT1 d iode characteristics plotted in a form such that they should show linear behavior for regions in which their current is space -charge-limited. ( An example of the fits of this data used to extract the mobility appear s in Figure 5 12.) The characteristics o f the four diodes appearing in the traces are the average of 2 4 measurements made on independent pixels for each diode. A) Behavior prior to annealing for 15 min at 150 C. B) Post annealing behavior. The plots of the diode characteristics made with P3HT 2 are displayed in Figure 5 10, plotted on the s ame scale as those of Figure 5 9 Similarly to the case of P3HT1 diodes, t he transition region for P3HT2 diodes also appears between voltages ~20003000 V1/2m1/2. All P3HT2 diodes displayed SCLC behavior, with the exception of one (from which no consistent I V data could be rendered, perhaps do to shorts or other irregularities in the fabrication process ). Plots of the characteristics from P3HT3 diodes are shown in Figure 5 10 (Note the change in vertical s cale.) Conversely to the P3HT1 and P3HT2 diodes, which exhibited well -defined
135 A B Figure 5 10. P3HT2 diode characteristics plotted in the sam e manner, and on the same scale as the P3HT1 diodes in Figure 5 .8. The linear regions display SCLC behavior. A) Behavior befor e annealing at 150 C for 15 min B) Post annealing behavior. SCLC domains, the regio random P3HT3 polymer diodes, showed a much -less defined transition region, beyond which the I -V behavior was only proximally limited to SCLC behavior. The approximate SCLC domain can be seen to begin at around 3000 V1/2 m1/2 and it becomes more apparent at higher electric field s A B Figure 5 11. P3HT3 diode characteristics plotted in the same manner, but on a different scale from those of P3HT1 and P3HT 2 in Figure s 5 9 and 5 10. L inear regions in the plots are not well -defined, but the curves seem to approach SCLC behavior, especially for high voltages. A) Behavior before annealing at 150 C for 15 min. B) Post annealing behavior.
136 A comparison of the curves in the SCLC regions of Figures 5 9 5 10, and 5 1 1 obviates a strong difference in the linearity, slope, consistency, and relative vertical position of the traces from P3HT3 diodes as compar ed with those of P3HT1 and P3HT2 Additionally, a small di ffer ence is perceptible between the vertical positions of the annealed curves from P3HT1 diodes compared with those of P3HT2 These differences can be quantified by fitting the data. Figure 5 12 show an example of the l inear fits that were used to extract the mobility values. By applying the relations of Eq. 5 .7, t he y -intercept of the fitting line was used to calculate the zero -field mobility, 0, and the slope of the fitting line was used to calculate the field dependence. The fits were made from for v alues of VDiff / L > 3000 V1/2m1/2. Table 5 1 summarizes the results of the fits giving both the mobility values as well as the standard deviations Figure 5 1 2 Extraction of the mobility value was made through applying a linear fit to the plot in the SCLC region, and using the relations in Eq. 5 .7 to calculate the field dependence from the slope, and the zero -field mobility from the y -intercept.
137 Table 5 1. Mobility and field -dependence extracted from SCLC diode characteristics. Table includes val ues b oth before and after 15 minutes of thermal annealing at 150 C on a brass surface. Before Annealing After Annealing Material 0 (cm 2 / V s) (m 1/2 / V 1/2 ) 0 (cm 2 / V s) (m 1/2 / V 1/2 ) P 3 HT1 (171 nm) 4.3 x 1 0 4 2. 2 x 10 4 1.1 x 10 3 3.6 x 10 4 P 3 HT1 (142 nm) 1.1 x 10 4 3.1 x 10 4 4.4 x 10 4 3.1 x 10 4 P 3 HT1 (97 nm) 2.6 x 10 4 3.0 x 10 4 3.7 x 10 4 3.4 x 10 4 P 3 HT1 (62 nm) 1.2 x 10 4 3.1 x 10 4 2.2 x 10 4 4.0 x 10 4 P 3 HT2 (1 57 nm) 2.8 x 10 4 2.2 x 10 4 4.0 x 10 4 2 .4 x 10 4 P 3 HT2 ( 96 nm) 1.5 x 10 4 2.4 x 10 4 2.4 x 10 4 2.9 x 10 4 P 3 HT2 ( 56 nm) 2.8 x 10 5 3.2 x 10 4 1.1 x 10 4 3.5 x 10 4 P 3 HT3 ( 262 nm) 7.1 x 10 9 2.7 x 10 4 1.2 x 10 8 1.5 x 10 4 P 3 HT3 ( 178 nm) 5.0 x 10 9 2.8 x 10 4 3.8 x 10 9 5.1 x 10 4 P 3 HT3 ( 111 nm) 2.0 x 10 8 1.1 x 10 4 5.8 x 10 7 2.4 x 10 4 P 3 HT3 ( 68 nm) 5.0 x 10 7 1.7 x 10 4 5.6 x 10 5 4.7 x 10 4 The average mobility valu es prior to annealing were nearly equivalent for both regioregular samples. The zero field mobility for the regiorandom sample was approximately, three orders of magnitude lower. Annealing was found to improve the mobilities of all three polymers. Surpr isingly, the mobility of P3HT3 increased by two orders of magnitude, though the diodes of this material produced poor fits to the SCLC model and post annealed devices exhibit a broad standard deviation Figure 5 13 shows the electric field dependence of the mobility of P3HT1 and P3HT2 over the range from 0.5 MV/m to 10 MV/m. On this scale the transition range for the diodes occurred between 4 and 9 MV/m. Even though the extracted mobility is supposed to represent an intrinsic property of the polymers, reg ions close to zero electric field inevitably become more complicated and it is likely that in very low electric fields, the mobility will not hold the simple exponential dependence that is represented. However, for regions near and above the diode
138 tra nsition region the fields represented are similar to the one used for the mobility extraction from our diodes. T o help put this scale in perspective it is useful to not e that application of 0.55 volts across a 100 nm film, is equivalent to the region betw een 5 and 50 on this scale. These are typical fields used in polymer electronic devices and this region is of particular interest. Through this region the mobilities show an order of magnitude decrease Figure 5 13. Electric field dependence of the average mobilities before and after annealing. The region from which the fit was made is indicated on the graph. The plot shows a negative dependence of the mobility on electric field. Most polymers are found to exhibit increased mobilities with electric f ield. As is evident from the graph t he mobilities found here show negative field dependence. This negative dependence of P3HT mobility is not well established in the field. It is in contrast to the work
139 done by McGehee and Kline ,37 wh o reported positive field dependence (~2 x 104 m1/2V1/2) for low molecular weight P3HT samples and mobilities of 105 cm2/ Vs. They also noted a rise in mobility with polymer molecular weight For m olecular weights ~30 kg/mol (nearly equivalent to those used in this work) they reported increased zero -field mobilities with an average of 3.3 + 0.73 x 104 cm2/V s. However, they only examined these polymers with a field independent model, which is unfortunate because on close inspection, their high molecular weight data seems to exhibit negative field dependence. More recent TOF measurements have been published on regioregular P3HT mobilities by Ballantyne et. al.97 in which they made a thorough survey of the effect of molecular weight on mobility. Their findings were that beyond a molecular weight of ~18kDa hole mobility decreased with longer chain length. Through time of flight analysis, this work also found slight increases in mobility with electric field. They did, however, observe mobilities in P3HT : PCBM ble nd films with negative field dependence for low molecular weight P3HT samples. In 2004, Saraciftci and Mozer98 made a r eport of negative field dependence in P3HT using time of flight technique and attributed it to the significant spatial disorder in the material. Arguments have been made that the negative field dependence can be attributed to limitations of the TOF method. In 2005, Saraciftci and Mozer99 published a second repor t in which they found negative field dependence with both the TOF and the charge extraction by linearly increased voltage method (CELIV). This technique involves observing the time dependence of the current relative to rates of voltage increas e In 2008, a report100 was published finding ne gative mobility field -dependence in drop -cast samples of both regioregular P3HT and regioregular poly(3
140 octylthiophene) by analysis using CELIV H owever this same report found positive field dependence in spin -cast P3HT films. T he negative field depen dence factors reported by Sariciftci and Mozer cannot be compared to those found here because it appears there was a mistake in the units published where they report 1.4 3.6 x 104 ( cm V 2)1/2 which does not work dimensionally in their fitting equation Their reported zero -field mobilities are also an order of magnitude higher than those found in this work. It seems that the work presented here is the first instance of negative field dependence in P3HT as observed through use of the SCLC modeling te chnique. Figure 5 13 includes one micron morphology images from surfaces of the diode films. The ima ges of the surfaces from diodes containing regioregular P3HT1 and P3HT2 (Figure 5 13A,B) both showed distinct morphologies. As expected, the surface image of the P3HT3 film, plotted on the same scales shows very little feature. The existence of more ordered packing domains can be related to the higher carrier mobilities in the regioregular films. The films with the highest mobility corresponded to those w ith the most distinct domain features. These features are possibly evidence of a greater degree of crystallinity in the film s The origin of the negative field dependence is not well understood. It was mentioned that regioregular poly(3 -octylthiophene) s howed negative field dependence. A similar time of flight study101 found regioregular poly(3-butylthiophene). Th e conclusion of Mozer and Saraciftci was that it was due to large spatial disorder, but this explanation seems inadequate, as P3HT is known to be a highly ordered polymer. Most disordered polymers exhibit positive field dependence. It seems that it is so mething more intrinsic to P ATs but its origin is has not been well -explained.
141 A B C Figure 5 1 3 Atomic force microscopy height images of the polymer film surfaces of the diodes A) Film surface from a P3HT1 diode showing strong feature s which are possibly indicative of underlying crystallinity. B) Film surface from a P3HT2 diode showing similar features to those of P3HT1 C) Film surface from a P3HT3 diode which shows very little feature indicative of a less ordered film. 5 .4 Space -Charge -Limited -Current Studies on Newly -Developed Photovoltaic Polymers Field dependent SCLC measurements were also conducted on three other polymers which were developed for use in solar cells. The structures of these materials appear in Figure 5 14. Th e vinylene linked thiophene and benzothiadiazole polymer appearing in Figure 5 14 A is known as PTVBT (short for poly thiophene -benzothiadiazole). Figure 5 14 B contains a 3,6 -
142 linked thiophene -carbazole, which has been abbreviated as PBTC (short for poly bis -thiophene carbazole). The platinum -containing polymer of Figure 5 14 C is repre sented by the abbreviation P Pt BT D Th (short for poly -platinum benzothiadiazole thiophene). A B C Figure 5 14. Structures of polymers whose mobilities were investigated using SCLC. A) Vinylene -linked poly -benzothiadiazole thiophene. B) 3,6-poly bis thiophene carbazole. C) Poly -platinum -benzothiadiazole -thiophene. 5.4.1 PTVBT SCLC Diodes PTVBT is a donor acceptor polymer, meaning that it contains alternating electron donating units (thiophene) and electron accepting units (benzothiadiazole). The electron donating units act to increase electron density of the conjugated backbone, therefore establishing the HOMO level of the resulting polymer. The additional presence of electron accepting groups act to lower the LUMO level of the polymer, effectively decreasing the polymer band gap. Torsional interactions between neighboring monomer units can lead to decreased planarity and conjugation length. The vinylene l inkages have been incorporated into PTVBT to reduce these interactio ns through spatial separation.
143 Electrochemical experiments on PTVBT indicated that the HOMO level of PTVBT resides at approximately 5.2 to 5.4 eV relative to vacuum. The LUMO level was found to be at approximately 3.5 to 3.6 eV. The HOMO level should allow for positive charge carrier injection from PEDOT:PSS with limited injection barriers. Electron i njection into the LUMO level should be minimal because the large energy separating i t from the work function of PEDOT:PSS ( 5.1 eV). Diodes were fabricated with PTVBT using the typical hole -dominated transport configuration of ITO/PEDOT:PSS/ PTVBT /Au. The I -V characteristics of the diodes were measured under dark conditions inside an argon-filled glovebox. After removal from the box, absorption spectra were taken, followed by AFM step -height measurements. The corrected absorption spectra appear below in Figure 5 15. Thickness measurements on the diode layers using AFM proved difficult du e to surface roughness and step edge qualities. For this reason, the thickness of each polymer layer was calculated from the AFM measurements of the thickest layer and the relative height of the abso rption maxima using Beers law. Figure 3 15. Absorpti on spectra of PTVBT diodes collected after device testing and used in conjunction with AFM measurements in determining film thicknesses.
144 The I -V characteristics of the PTVBT diodes appear in Figure 5 16 along with an example of one of the fits. PTVBT diod es do not appear to reach field -dependent SCLC behavior until much higher fields than P3HT diodes. The characteristics were fit for electric fields beyond 16 MV/m. Positive field dependence was exhibited with an average field dependence factor of 6.6 + 2 .9 x 104 m1/2V1/2 The average zero field mobility was 1.6 + 2.0 x 107 cm2/Vs. A B Figure 5 16. PTVBT diode characteristics. A) Linear trends as predicted by the field -dependent SCLC model are only approximately followed by diodes. B) Corresponding fit of a 255 nm PTVBT thin -film diode. The average field -dependent mobility is displayed in Figure 5 16. The positive trend of mobility increasing with electric field strength is common behavior for conjugated polymers. The mobility values are on the low end of the spectrum, especially at low fields. The analysis of a similar vinylene linked phenylamine benzothiadiazole polymer102 found mobilities of 5 x 105 cm2/Vs, but their treatment did not use the field dependent model and so it is difficult to compare their v alues without knowing the field strength. It is possilbe that the low zero -field mobilities found to occur in PTVBT could be caused by charges being localized near acceptor units.
145 Figure 5 17. Field dependence of the positive carriers in PTVBT as fou nd through application of the SCLC model. 5.4.2 3,6 -PBTC SCLC Diodes The manner in which the thiophene and carbazole units are linked together in 3,6 PBTC is an important factor. In Figure 5 14 B, the different possible attachment locations on the carbazo le unit are numbered. Attaching through the 3 and 6 position (the case of 3,6 PBTC), means that the backbone of the polymer will no longer be fully conjugated. It can be seen by interchanging the single and double bonds along the polymer backbo ne, that i n the case of 3,6 linkage, there is a conjugation break that occurs in the carbazole unit. This broken conjugation will inevitably limit charge transport along the polymer backbone. The aromaticity of the carbazole units does provide a platform for pi -st acking and so it is possible that transport could still be strong through interchain transfer. The electronic levels of 3,6 PBTC were investigated through the use of cyclic voltametry. The HOMO level of the polymer was estimated to be at 5.0 eV relative to vacuum. The LUMO level is considerably higher at 2.4 eV, and thus diodes of the structure ITO/ PEDOT:PSS /
146 PTVBT / Au should be hole -dominated. This configuration was used in the fabrication of four 3,6 PBTC diodes for hole transport measurements. Af ter I -V characteristics of the diodes the film thicknesses and optical properties were measured. The thin -film absorption data appears in Figure 5 18. Figure 5 18. Absorption properties of the 3,6 -PBTC diodes. The low absorption maximum occurring arou nd 420 nm is indicative of the large band gap. The absorption tail extending through the visible was present in solid state measurements, but absent in solution. The 3,6 PBTC diodes exhibited good correspondence with the field-dependent SCLC model. The d iode behavior is shown in Figure 5 19, where a transition region to space charge limited current appears at around 3000 V1/2m1/2. As expected fits of the data yielded low mobilities with an average zero -field mobility of only 1.0 + 1.2 x 107 cm2/Vs. Th e field dependence averaged 6.8 + 2.4 x 104 m1/2V1/2. This is relatively strong field dependence. It is indicated in Figure 5 20.
147 A B Figure 5 19. PBTC diode characteristics. A) Beyond the transition region, the PBTC diodes well approximate SCLC behavior B) Field -dependent fit of 70nm thin-film PBTC diode. Figure 3 20. The mobility of PBTC has strong field-dependence, though the mobility values are relatively small. The class of 3,6 linked polycarbazoles has been relatively unexplored in neut ral semiconductor applications. The carbazole unit is a strong chromophore and 3,6 linked carbazoles have been investigated for electrochromic applications .103,104 The low mobilities observed here in the neutral form can be attributed to the broken conjugation of the polymer backbone. The 2,7 linked class of carbazoles have continued conjugation along the polymer
148 chain and hole mobilities as high as 0.1 cm2/Vs have been reported.105 The low mobilities of 3,6 -carbazoles will likely limit the applicati ons in semiconducting devices. 5.4.3 P -PtBTD -Th SCLC Diodes Interest in the possibility of triplet excited states contributing to photovoltaic behavior in conjugated polymers led to the incorporation of Platinum into a conjugated chain.106 The backbone of P PtBTD Th contains platinum within the conjugation. The presence of platinum may significantly affect the mobility of charge carriers along the chain by disrupting the effective conjugation length. Through cyclic voltametry the HOMO level of P PtBTD Th was found be -5.1 eV relative to vacuum, a good match for hole inje ction from PEDOT:PSS. Following the same general method, P -PtBTD Th diodes were made with a ITO/PEDOT:PSS/P PtBTD -Th/Au configuration. These diodes were tested inside the glovebox and film thicknesses were determined using AFM. The characteristics of th e diodes appear in Figure 5 21. It is interesting to note that even at very high fields it seem s that there is curvature displayed, suggesting the mobility is not entirely confined to SCLC. Figure 5 21. The field dependent SCLC mobility plots show sli ghtly greater curvature than linear behavior. A) Characteristics from all three devices. B) a linear fit in the high field region where the behaviors most closely approximates SCLC.
149 The field dependence of the mobility of P -PtBTD Th is small but positiv e as seen in Figure 5 22. The mobility remains low over a broad range of fields. The reason for low mobility is likely due to broken conjugation caused by the platinum. Additionally, the solubilizing butyl groups attached to the platinum are bulky and likely reduce both planarity and packing. Comparison of these mobility values with Metallated conjugated polymers are a relatively new class of materials and there are no reported mobility values on pure films. P PtBTD Th mobilities were studied in blend films with PCBM where the authors107 found the blends to achieve hole mobilities on the order of 105 cm2/Vs. However, this paper has been highly controversial and blend films typically showed improved hole transport mobilities. This is probably due to the polarizing nature of the polymer donor -PCBM acceptor interface. Figure 5 22. The field dependence of the mobility for P PtBTD Th is positive, but the mobility remains low over a broad range of fields. 5.5 Conclusions The field -dependent SCLC mode l can be implemented and used for the extraction of mobility values in conjugated polymer systems. There are challenges inherent to both the method and to defining the mobility of a conjugated polymer, due to the variations in polymer packing and to the s trong coupling between charge -car riers and the polymer lattice.
150 Mobility values in conjugated polymers are dependent on both the polymer and the underlying film morphology. The field dependence of mobilities in conjugated polymer films limits the determin ation of the mobility to the measurable range. Values of mobility should be accompanied by field strengths in which they were obtained.
151 CHAPTER 6 BULK HETEROJUNCTION PHOTOVOLTAIC DEVICES 6 .1 I ntroduction The end goal of almost all solar cell research is to find ways to improve power conversion efficiencies, air -stabilities, and fabrication pathways towards values that will make production commercially viable. In the case of polymer photovoltaics, low power conversion efficiencies are the biggest obstacle needed to be overcome Methods of optimization are numerous and the overwhelming majority of polymer solar cell research is focused on closing the efficiency gap between organic and inorganic solar cells. This chapter presents work that explored new mat erials for use in polymer -based bulk heterojunction solar cells, and employed a number of optimization techniques that focused on morphology and absorption related properties. 6 .2 Performance Optimization of P3HT: PCBM BHJ Solar Cells Regioregular P3HT has been one of the most successful donor -materials in polymer solar cells consequently, P3HT : PCBM BHJ solar cells have probably been the most thoroughly investigated of all polymer based photovoltaic devices Much of this success has been attributed to th e relatively highly ordered packing found in P3HT films, in comparison to other polymer systems. The variations between polymer samples have led to relatively broad guidelines for producing optimized P3HT cells. Additionally, it seems that different path ways (such as the cases of solvent, thermal, and microwave annealing mentioned in chapter 3) can l ead to the same end result This further complicates forming a precise set of conditions to be used in creating efficient P3HT cells. This work h as selected the use of a highlyregioregular (~98.5 %) P3HT polymer ( RR Rieke Sepiolid P200, which following the nomenclature of Chapter 3, will be
152 referred to as P3HT 1 ), with an end goal of finding a set of optimized conditions that would allow fabrication of rep roducible, efficient P3HT : PCBM solar cells. 6.2.1 Active Layer Film Composition There has been significant work examining the effect of the relative fractional content of polymer to PCBM in the blend films. Many groups have found small variations in the blend content to lead to noteworthy differences in performance.75 It has been suggested that balanced hole and electron transport properties, which occur in mixtures of around 4550% polymer should be used to select the proper ratio.72,108 However, other factors such as the strong blend composition-dependent domain separation need to be considered. The optimal ratio seems to vary slightly from one P3HT sample to another, with typical optimal performance being found for blends containing approximatel y 55 60 % P3HT and 45 40 % PCBM In this section, work is presente d in which the optimum proportional content of P3HT 1 is investigated It seems to generally be the case that optimal blend ratio is one of the important parameters that is least -dependent on other variables. This investigation in search of the optimal ra tio was done prior to other optimization studies on P3HT solar cells. Therefore, the fabrication techniques employed in this work were partly -based on the findings of other work in the literature on a different brand of P3HT .75 Following the cleaning of ITO -coated glass substrates (as described in Chapter 2), a thin layer(~30 nm) of PEDOT:PSS was deposited. All fabrication and measurement steps, following the deposition of P3HT, were done inside of an argon-filled glovebox. The active layers of devices in this set were spun at 700 RPM for 60s with 3s ramps. Post -fabrication ann ealing was done at 150 C for 30 minutes on a heated brass block. No lithium fluoride was used.
153 Several sets of devices were made in the optimization process. The first several sets suggested that the optimum amount of polymer content in the blend was n ear to 60 %. Figure 6 1 shows the A.M. 1.5 J -V characteristics of the final blend ratio optimization set containing seven different devices. There is a performance maximum at 55.5 % polymer content, though the open circuit voltage of this device is not a s high as others of the set. The maximum efficiency obtained in this set of devices was 2.7 %. Trends in the efficiency, short circuit current density, fill factor, and open circuit voltage are displayed in Figure 6 .2. Figure 6 1. The I -V characteris tics for cells containing a varied amount of P3HT content in the polymer:fullerene blend, under illumination with A.M. 1.5 simulated solar radiation. Efficiency and short circuit current both show trends from lower and higher polymer content leading to optimal values which seem to peak at ~ 55 58 % polymer. The fill factor also shows a maximum at 55 % polymer, but the trend is not consistent. Fill factor does show a very evident drop off for high polymer content. There was no general trend indicated by the variations in open circuit voltages. The results of this set indicate maximum will be achieved at polymer weight percentages of ~55% of the total blend.
154 A B Figure 6 2. Photovoltaic characteristics for cells containing a va ried amount of P3HT content in the polymer:fullerene blend under simulated A.M. 1.5 solar radiation. A) Current density and efficiency both show peaks near 55 % polymer. B) Fill factors showed marked drops for high polymer content, whereas open circuit voltages didnt seem to have a direct correlation to polymer content. 6.2.2 Optimization of Active Layer Film Formation In comparison to the blend composition of optimized P3HT:PCBM solar cells, there is a much greater variety of published values indicati ng optimized active layer spin -coating conditions. The variations reported reflect the fact that optimal thickness depends on both solution concentration and on the spin-coating rate. Typical optimized P3HT : PCBM cells have active layers of approximatel y 200 nm in thickness.74 This film thickness can be achieved by using low concentration solutions and low spin rates, or by us ing more highly concentrated solutions and higher spin rates. Common film formation techniques employ the use of solutions which have 1020 mg of donor acceptor mate rial (of which, approximately 5060 % is polymer and the other 50 40 % is PCBM ) per mg of solvent. The slower evaporation rates of high boiling point solvents have been found to produce the best results and chlorobenzene has become the solvent of choice .70,109 As with the case of blend film co mpositions, optimized conditions for active layer formation vary from one polymer sample to another. They depend on the polymer properties,
155 such as, regioregularity, molecular weight, and polydispersity.110 To find the optimal conditions for P3HT 1 several d ifferent concentration and spinrate combinations were investigated. Using the same device architecture as previous experiments and a blend film composition of 55% P3HT studies were carried out in which the concentration of the blend solution and the spin rate were varied. In the same manner as the earlier studies, the devices were annealed (post fabrication) at 150C for 30 minutes Figu re 6 3 shows the J -V traces of a set of cells in which all processing conditions were equivalent with the exception of the blend film concentration. The J V traces show p erformance peaks in both short circuit current and fill factor, and a slight peak of open circuit voltages for low concentration solutions. Figure 6 3. Simulated A.M. 1.5 J -V behavior for a set of cells in which the blend film was composed of 55 % pol ymer, the active layers were spun at 700 RPM with 3 s ramps, and all devices were post -fabrication annealed at 150C for 30 min. The curves indicate a maximum in performance for the device spun from a blend solution with a concentration of 17 mg/mL. The re sults of further studies are shown in Figure 6 4, in which many solar cells were made under the same conditions with the exceptions of varied concentrations ranging from 15 to 21 mg/mL and deposit ed at rates from 500 to 900 RPM The plot of power conver sion efficiencies versus concentration (Figure 6 6 A) shows an evident peak in performance at 17
156 mg/mL for active layers spun at both 500 and 700 RPM. The fill factor can be seen to show a similar trend to the efficiency, as indicated in Figure 6 6 B. Th e peak in short circuit current density is less clearly defined with maximum currents at 17 and 18 mg/mL being very similar. The open circuit voltage is not plotted, but remained fairly constant at about 0.6 V. A B C Figure 6 4 The performance param eters for cells fabricated under similar conditions, but containing variations in the spin rate and blend solution concentrations. A) A.M. 1.5 power conversion efficiencies. B) Fill Factors of the cells. C) Short circuit current variations. 6.2.3 Opt imization of Annealing Conditions There have been many studies on the effect of thermal annealing P3HT : PCBM solar cells.111 Other methods of annealing the blends in addition to thermal annealing have also been explored. Solvent vapor annealing has been employed by placing films in a solvent vapor saturated environment.112 The absorption of the solvent by the film likely increases the mob ility of the PCBM molecules in the polymer matrix allowing rearrangements in the underlying morphology towards more crystalline and lower energy configurations. A nnealing cells through exposure to microwaves has also produced improved efficiencies.113 It seems that each of the annealing methods present different pathways towards the same end result of modifying the blend packing arrangements.
157 To determine the optimum thermal annealing temperature for P3HT 1, multiple sets of devices were made following the previously found optimized blend composition of 55% P3HT and the optimized active layer deposition conditions of 17mg/mL solutions spun for 60s at 700RPM with 3 second ramps. Devices sets were pos t -fabrication annealed for 30 minutes on a heated brass block at temperatures ranging from 145 to 175 C. Figure 6 5 shows the power conversion efficiencies of two sets of these devices(red and black) where individual cells have been annealed at different t emperatures. A third set is included in the plot(green) in which the active layers were cast from 25mg/mL resulting in much thicker active layers. The power conversion efficiencies for the cells showed trends on increased improvement with annealing tempe rature until around ~168 C. Annealing at temperatures increasingly above this value led to a gradual decrease in cell performance, as well as an increased likelihood of cells with shorted pixels. It can also be seen that the thicker films showed a perfor mance maximum for an annealing temperature of around 170 C. The slightly higher value may be due to longer rearrangement times in thicker films. 140 150 160 170 180 0 1 2 3 4 5 A.M. 1.5 PCE (%)Annealing Temp (C) 17 mg/mL set Z 17 mg/mL set AA 25 mg/mL set X 2nd order Fit 3rd order Fit 2nd order Fit Figure 6 5 Power conversion efficiencies showed improvements with annealing tem peratures until peaking at ~168 C.
158 Figure 6 6 shows the trends in fill factor and the short circuit current. The fill factors (Figure 6 6 A) show very analogous behavior to the response of the efficiency, whereas the short circuit current s (Figure 6 6 B) show a drop in value for annealing temperatures above ~160C. The overall performance improvements with annealing temperatures beyond this point are, therefore, attributed to the increased fill factors which reached values in excess of 60 %. A 140 150 160 170 180 0 20 40 60 80 100 17 mg/mL set Z 17 mg/mL set AA 25 mg/mL set X 2nd order Fit 3rd order Fit 2nd order FitFill Factor (%)Annealing Temp (C) B 140 150 160 170 180 0 3 6 9 12 15 17 mg/mL set Z 17 mg/mL set AA 25 mg/mL set X 2nd order Fit 3rd order Fit 2nd order FitShort Circuit Current (mA/cm2)Annealing Temp (C) Figure 6 6 Peformance characterics for cells annealed at various temperatures. A) Fill factors show parallel improvements to the cell PCEs. B) Short circuit currents peak at ~160 C. The relationship betwee n annealing and morphology becomes apparent in the nanometer scale features of blend films. This can be seen by AFM surface topography scans. Figure 6 7 A and B show AFM height and phase images of a P3HT : PCBM blend film before thermal annealing. Figure 6 7 C and D show the height and phase images of the same film after thermal annealing. There is an obvious difference in the surface roughness that is induced by annealing. The formation of domains on the 3050 nm scale can also be seen especially in th e phase image Annealing also has a very significant impact on the optical properties of P3HT:PCBM blend films. The effect can be seen in the absorption plots of Figure 6 8. The absorption of both
159 A B C D Figure 6 7 Atomic Force Microscopy images showing the affect of annealing on P3Height image of cell before annealing. D) Phase image of cell after annealing. pure PCBM films and pure P3HT films are shown Figure 6 8 A before and after annealing at 150 C for 15 minutes. In both cases there are small optical density increases s een after annealing. Figure 6 8 B shows the absorption of a blend film (55% P3HT : 45%PCBM ) both before and after an nealing at 150 C for 15 minutes. The absorption of the blend film has a much larger optical density after annealing. Comparing the absorption changes in the blend film with those observed in the pure films makes it obvious that combination of the two mat erials greatly increases the annealing induced absorption changes. This means that there must be significant interactions of the two materials in the blend with increases with annealing.
160 A B Figure 6 8 Optical response of pure films and blend films to annealing. A) Pure films of PCBM and P3HT both show slight increases in absorption after annealing. B) The effect of annealing has a much greater impact on the absorption of a blend film. Further annealing studies were conducted to investigate the annea ling time on performance of cells with active layers made from 17 mg/mL solutions, spun at 700 RPM. Figure 6 10 shows the J -V characteristics for three cells (black, red, and green curves) which were annealed at 170 C for different lengths of time. The competing optimization between fill factor and short circuit current is noticeable in these films by comparing the 20 and 30 minute annealing times. The cell annealed for 20 minutes shows higher short circuit currents, but lower fill factors than the cell annealed for 30 minutes. Both cells have almost the same power conversion efficiency. The fourth curve (blue) appearing in Figure 6 9 corresponds to a cell that was slowly cooled during the annealing process. The method of slowly cooling the films during annealing was introduced with the idea that by beginning the annealing process at a high temperature the mobility would be introduced into the lattice and slow decreases in temperature would allow recrystallization of domains with higher degrees of order This done by beginning the annealing process at 170 C and decreasing the temperature to 150 C over the 30 minute annealing process. As can be seen from the blue curve, this annealing method produced the most efficient cell where both the fill factor an d the short circuit current are high. The external quantum efficiency
161 of this cell appears in Figure 6 10. The data shows that at wavelengths near the P3HT absorption maximum, over 60% of incident photons result in electrons at the contacts. The results of these studies suggest that the best performance of P3HT 1: PCBM solar cells result from active layers formed from ~17 mg/mL solutions, spun at 700 RPM with 3 s ramps, followed by post fabrication annealing beginning at 170 C and ramping down to 150 C over a 30 minute annealing time. The performance of these cells are slightly less that the highest reported values in the literature of nearly 5%. It is likely that future fine tuning of the optimization process will be able to slightly further increase the efficiencies obtained from P3HT 1: PCBM cells. It is also noted that the optimal annealing temperatures found for here for P3HT 1 are above those found elsewhere for other RR P3HT polymers. The higher annealing temperatures for P3HT 1 are attributed to its v ery high degree of regioregularity ~98.5 %, compared with other typical values of ~95 %. Figure 6 9 Current voltage characteristics for cells illuminated under A.M. 1.5 conditions and post -fabrication annealed under different annealing condtions.
162 Fi gure 6 10. External quantum efficiency of a P3HT:PCBM solar cell showing greater than 60 % incident photon conversion efficiency. 6 .3 PTVBT : PCBM Solar Cells Compared to P3HT l ow er band gap polymers are more desirable for use in solar cells because they have a better overlap of the solar photon flux spectrum. Figure 6 11 shows the absorption spectra of PTVBT (green) and P3HT (blue) overlaid onto the solar flux spectrum. The band gap of both of these polymers defines the longest wavelength for which the polymers will be able to absorb photons for conversion into electricity in solar cells. The band gap, consequently defines an ultimate limit on power conversion efficiency. In the case of P3HT the band gap is at a wavelength below which only ~22 % of s olar photons occur. The band gap of PTVBT corresponds to wavelength below which ~40 % of solar photons occur. This means that even at significantly reduced external quantum efficiencies PTVBT has a much greater optically defined potential for use in sola r cells. Of course, factors other than band gap are important in determining the actual efficiency of photovoltaic devices. This section provides the results of investigations exploring the use of PTVBT blended with PCBM in bulk heterojunction solar cells
163 Figure 6 11. Overlap of the absorption spectra of PTVBT and P3HT (arbitrary units) with the A.M. 1.5 solar photon flux spectrum. PTVBT was described briefly in Chapter 3. It is a donor acceptor polymer in which the electron rich thiophene units act a s donors and the electron deficient units act as acceptors. The mutual influence of the alternating units along the conjugated backbone act to reduce the polymer band gap. Electrochemical studies found the HOMO level of PTVBT to 5.2 to 5.4 eV relative to vacuum. This value is far enough below vacuum that PTVBT should be air -stable. The LUMO level of PTVBT was found to be at approximately 3.5 to 3.6 eV. This value is slightly above that of PCBM which is typically taken to be around 4.0 eV (though s ome estimates114 place it a s high as 3.7 eV, which would greatly reduce the potentia l for electron transfer to PCBM ). The offset of the LUMO levels between PTVBT and PCBM should result in electron transfer from the polymer to PCBM assuming that this offset is in excess of the exc iton binding energy in PTVBT PTVBT solar cells were fabricated using the common layered architecture of glass/ ITO/ PEDOT:PSS/ PTVBT : PCBM / Al. (It was noted that the addition of a thin LiF layer did not improve performance, hence, the majority of cells d id not employ this additional layer.)
164 Substrates were cleaned following the standard procedure described in Chapter 2. After the addition of the PEDOT:PSS layer (~30 nm) devices were dried and brought into the glovebox. The addition of the active layer and the remainder of the fabrication steps all took place under inert atmosphere. Stock solutions of PTVBT and PCBM were prepared in concentrations ranging from 15 20 mg/mL and were stirred at least 24 hours before combining in different ratios to form blend solutions for active layer spin coating. Multiple sets of devices were prepared to optimize the performance of the solar cells. Similarly to the methods used in optimization of P3HT solar cells, the optimization process began with the exploration of d evice performance based on the relative polymer to fullerene composition of the active layers. Figure 6 12 shows the A.M. 1.5 illuminated J -V characteristics of some of the initial cells which indicated optimal performance contained surprisingly low PTVBT percentages. The solar cells whose J -V behaviors are represented in the figure had polymer content of 55 % (blue) 35 % (red) and 20 % (black) The best performance was exhibited by the cell containing only 20% polymer, despite the fact that this cell had the most limited light absorption. Figure 6 12. A.M. 1.5 illuminated J -V characteristics of PTVBT : PCBM solar cells with varied amount of polymer content. The best performance is shown by the cell containing a surprisingly low percentage of polym er in the active layer blend ( 20 %).
165 Further optimization sets explored the range of polymer content in the active layer blends from 4 % to 50% polymer. Figure 6 .13 A shows a plot of the power conversion efficiencies of multiple cells from this range. A t only 4% polymer content the power conversion efficiencies exceed those of 35 % and 50% polymer. The results indicate that the optimal content of polymer in the blend lies in the range from about 10 % to 15% polymer. At such low polymer content, the amount of light that can harvested by the polymer is very low. This suggest that backtransfer, in which PCBM is absorbing photons and transferring holes to the polymer, might be the dominating photovoltaic mechanism for power conversion. Figure 6 13 B show s the external quantum efficiency of a cell containing 10% polymer and 90 % PCBM At low wavelengths (< 400 nm) the photocurrent response mimics the absorption spectrum of the PCBM (see Figure 6 8) indicating that back transfer is responsible for some of the power conversion. However, at wavelengths above 400 nm, the photocurrent spectrum closely parallels the absorption spectrum of PTVBT indicating the photons absorbed by PTVBT are responsible for the power conversion efficiencies. A B Figure 6 11. Performance of PTVBT : PCBM solar cells A) Power conversion efficiencies show optimal performance at low polymer contents percentages. B) External quantum efficiency measurements resemble both the PCBM absorption spectrum and the absorption spectrum on PT VBT which indicates photon absorption by both materials is contributing to the output power of the solar cell.
166 Adjusting the concentrations of the blend solutions and the spin rates led to only small additional improvements in efficiency. The open-circuit voltages of the cells mostly fell in the 5060% range which is typical for low band gap polymers.115 The fill factors also reached acceptable values, which fell in the 4 0 50% range. However, the short circuit currents were low, and the limiting factor in performance. Figure 6 12 shows the characteristics of the best performing cell, which had an efficiency ~0.5 %, short circuit current density ~1.5 mA/cm2. The reason for the low currents can be attributed to incomplete light absorption by the active layers which only contain a small percentage of polymer. The thicknesses of the active layers of the best performing cells were found to be ~100nm. The low fractional c on tent of polymer in the blends resulted in most of active layers having transmitting nearly 80% at the polymers absorption peak. However, the decrease in efficiency with increasing polymer content indicates strong, non-optical limiting factors which need a different explanation. Figure 6 11. A.M. 1.5 J -V characteristics of the best PTVBT whose active layer was composed of 10% polymer and 90% PCBM The power conversion efficiency was ~0.5 %. The morphologies of the blend films were examined with using t apping mode AFM. Figure 6 12 contains height images for blend films of vari ed polymer content ranging from 4%
167 to 50% PTVBT The images show a strong correlation between the content of polymer in the blend films and the surface morphologie s. For blends c ontaining just 4% polymer the surface of the films are very flat. As the polymer percentage is increased to 8% the films begin to show the formation or shallow round pits. At 12% polymer content the pits deepen with typical depths ~25 30 nm. For percent ages increased to 16 and 20% polymer the pits become shallower, but begin to broaden into the formation of valleys. For blends composed of 25% polymer the valleys have coalesced into the formation channels and there is strong and distinctive domain separa tion visible. At higher concentrations the domain separation remains evident with the size of the domains decreasing. The AFM analysis, which shows very strong phase separation beginning around 16 20% polymer, seems to coincide with the observed decreases in efficiency which occur at nearly the same polymer percentages. This domain separation possibly represents the lack of miscibility between PTVBT and PCBM It is a possibility that the polymer is collecting on the surface resulting in bi -layer devices as opposed to preferable bulk heterojunction erojunctions. However, it is also noted that the drop-off in power conversion efficiency is much less abrupt than the domain separations that are observed in the surface morphologies. The overall power conversi on efficiencies of PTVBT are low. This is likely due to a combination of low mobility and very poor miscibility of the PCBM and the polymer at higher percentages of polymer content. The fact that PTVBT:PCBM cells containing only 10% polymer were able to achieve external quantum efficiencies of ~10 % at the absorption maximum of the polymer in films as of only 100 nm in thickness indicates that charge transfer is occuring between the polymer and PCBM. Furthermore, at such low polymer percentages the films are only absorbing a small fraction of the incident radiation (~ 20%). This suggests that
168 internal quantum efficiencies may be as high as 50 % which is a very encouraging possibility. If the miscibility issues can be overcome, this polymer still represe nts the possibility for good power conversion efficiencies. Possible methods through which to address the phase separation include active layer formation with processing additives or the use of a different acceptor, such as PC70BM. Following these directions further work is likely to result in increased efficiencies. A B C D E F G H Figure 6 13. Tapping mode AFM images of PTVBT:PCBM blend film surfaces. A) 4 % PTVBT: 96 % PCBM. B) 8 % PTVBT: 92 % PCBM. C) 12 % PTVBT: 88 % PCBM. D) 16 % PTVBT: 84 % PCBM. E) 20 % PTVBT: 80 % PCBM. C) 25 % PTVBT: 75 % PCBM. A) 35 % PTVBT: 65 % PCBM. B) 50 % PTVBT: 50 % PCBM. 6 .4 PBTC:PCBM Solar Cells The domain boundaries between the donor and acceptor materials in bulk heterojunction cells p lay a crucial role in the photovoltaic process. These interfaces, occurring throughout the bulk of the blend, are responsible for exciton dissociation and preventing charge recombination. It is thought that in P3HT : PCBM blends annealing induces modification of the donor acceptor interfaces, resulting in enhanced charge transfer.116,117 The spectra appearing in Figure 6 8 are
169 supportive of this idea, where prior to annealing the blend film spectrum can be closely represented by the sum of the pure P3HT and PCB M spectra. After annealing however, the spectral weighting changes suggesting interaction between the materials of the the blend. When processing from solution, control over the donor acceptor domain formation and properties is limited. Recently it ha s been discovered that adding thiols to the blend solutions can effectively modify the interfacial properties of the blends, resulting in large increases in photovoltaic performance.77,78,118 An alternative approach to modify the do nor acceptor interface is to directly attach PCBM pendant groups onto the polymer chain. In this work this approach has been explored through the attachme nt of fullerene units onto the 3,6 linked PBTC. Figure 6 14 illustrates the stucture s of PCBM 3,6 PBTC and the new fullerene PBTC derivative (which will be referred to as C60-PBTC) along with their corresponding absorption spectra. Figure 6 1 4 Absorption profiles and s tructure s of PCBM 3,6 -PBTC and C60-PBTC The large band gap of 3,6 PBTC means t hat the absorption spectrum of the polymer is not well suited for absorption of the solar photons. However, polycarbazoles have been shown to be
170 very photoactive polymers, especially in the case of smaller band gap 2,7-linked derivatives.119,120 Currently, the highest efficiency, single layer, polymer solar cell published utilizes a small band gap carbazole derivative.121 The 3,6-class of poly-carbazoles has been largely unexplored for solar cell applications. The properties of 3,6 PBTC were described briefly in Chapter 3. The HOMO level of the polymer occurs ne ar 5.0 eV and the LUMO level is at 2.4 eV which is significantly higher than that of PCBM but should still lead to effective electron transfer. Solar cells were built with both 3,6 PBTC: PCBM blends, and C60-PBTC : PCBM blends. The potential use of C60-PB TC as an interfacial -modifier was explored through fabrication of solar cells with active layers composed of C60PBTC :PBTC: PCBM and C60PBTC: P3HT : PCBM blends. All cells utilized the layered architecture of glass/ ITO/ PEDOT:PSS/ PTVBT:PCBM / LiF/ Al. Approximately 40 solar cells were prepared with 3,6 PBTC: PCBM blends. Figure 6 15 A shows typical J -V characteristic s of the cells The power conversion efficiency was found to be optimal for blends composed of 30 % PBTC and 70 % PCBM by weight External quantum efficiencies reached above 40% in the best cells, however, poor overlap with the solar spectrum led to relatively low power conversion efficiencies ~0.25 0.5 %. This is evident in Figure 6 15 B, which shows the absorption spectrum of the polymer, the external quantum efficiency of a PBTC: PCBM solar cel l, and the solar flux spectrum. A set of six solar cells was made using C60-PBTC blended with PCBM. As can be seen from the J -V characteristics of several of these cells in Figure 6 16, the performance was worse than those made with PBTC, but this was expected because of the blue shifted spectrum of C60PBTC.
171 A B Figure 6 15. Photovoltaic behavior of PBTC: PCBM solar cells. A) A.M. 1.5 J -V characteristics of typical solar cells with active layers composed of 50, 33, and 20 % PBTC B) Absorption of PBTC, External quantum efficiency of a pbtc: PCBM solar cell, and solar flux spectrum. Figure 6 16. J -V characteristics of C60-PBTC:PCBM solar cells under A.M. 1.5 illumination. The surface morphologi es of the PBTC: PCBM cells were studied through tapping mode AFM Figure 6 17 shows AFM images of The blend films showed very little feature, and RMS roughness were small ~0.5 nm. No strong correlation between surface morphology and performance was found The sharp peaks occurring in the 30 % sample were less defined in other films, but they did represent the device showing the best performance.
172 A B C Figure 6 17. Tapping mode AFM morphology images of the surfaces of active layers in which PBTC makes up 22.5, 30, and 37.5 % of the blend. A. 22.5% PBTC: 78.5% PCBM blend film. B. 30% PBTC: 70% PCBM blend film. C. 37.5% PBTC: 62.5% PCBM blend film. Given the very similar structures of PBTC and C60PBTC, it was though t that the addition of C60PBTC i n small amounts would lead to interfacial modification between the donor acceptor blends, possibly leading to a peak performance at an optimal additive fraction. To test this possibility C60-PBTC :pbtc: PCBM cells were made, in which the relative content C60-PBTC to pbtc was varied, in which the total composition of polymer was 30 % polymer and the content of PCBM was 70 % PCBM The A.M. 1.5 PCE of the cells were recorded and surface morphology scans were performed to investigate for a relationship between performance and morphology. Figure 6 18 shows J -V characteristics of several representative solar cell s in which C60PBTC has been used as an additive. All of the cells with this composition showed the same general J -V trend with low fill factors (~30 %) Figure 6 19 A shows a plot of the PCE efficiencies fitted with a 3rd order power series. The fit weakly-indicates perfomance maximum for active layers in which ~6 % c60pbtc has been added. The were also differences noted in surface morphologies. The t rend in RMS roughness gathered from 1 micron AFM tapping mode scans is plotted in Figure 6 19 B. The corresponding AFM images appear in Figure 6 20. The maximum in surface roughness occurs when 8 % c60pbtc has been added, which suggests that
173 there is a correlation between the surface roughness and the weakly-defined peak in power conversion efficiency. Figure 6 18. A.M. 1.5 illuminated J -V characteristics of C60-PBTC: PBTC: PCBM solar cells in which the polymer fraction composes 30 wt. % of the active l ayer blend and PCBM composes 70 wt. %. A B Figure 6 19. The effect of differing weight percentages of C60-PBTC in the active layers of C60PBTC:PBTC: PCBM solar cells. A) Trend in power conversion efficiency with content of C60-PBTC The low overall perfor mance of PBTC makes the analysis of the effect of C60-PBTC on the system difficult to determine. In order to examine C60PBTC more explicitly as an interfacial modifying agent, it was added to P3HT:PCBM active layers. Solutions were made containing 0 10 % C60-PBTC by adding the polymer to a solutions whose original composition was 60% P3HT and 40% PCBM.
174 A B C D E F G H Figure 6 20. Tapping mode AFM scans of the surface of C60PBTC :pbtc: PCBM solar cells in which the polymers compose 30 wt. % of the b lends and PCBM composes 70%. The relative percentages of C60-PBTC:PBTC: PCBM are indicated consecutively for each scan. A) 0% : 30%. B) 2 % : 28 %. C) 4 % : 26%. D) 6 % : 24%. E) 8 % : 22%. F) 10% : 2 0%. G) 15% : 150%. H) 2 0% : 1 0%. The effect of the addition o f C60PBTC to the system was investigated by testing the A.M. 1.5 power conversion efficiencies before and after post -fabrication thermal annealing. The morphologies of the cells were then examined by tapping mode AFM. The addition of C60PBTC was found to have a detrimental effect on the PCE of the P3HT : PCBM solar cells. The A.M. 1.5 J -V characteristics of the cells are shown in Figure 6 21 A, prior to annealing. The
175 performance of the cells shows a steady decrease with increasing amounts of C60-PBTC a dditive. The characteristics of the cells after annealing 30 minutes at 150 C, are show in Figure 6 21 B. Annealing has resulted in slight improvements to the fill factors, but for cells containing C60PBTC there is also a strong drop off in current. A B Figure 6 2 1. A.M. 1.5 illuminated J -V characteristics of (C60PBTC ):P3HT : PCBM solar cells. A) Performance prior to annealing. B) Performance after annealing. To ensure that the effect was due to C60PBTC a control sample was made by adding 3 % PBTC (with no c60 pendant groups) to a 60% P3HT and 40% active layer blend. The control sample exhibited a very similar behavior to the behavior of the P3HT cell that had 3 % C60-PBTC added to it. This is illustrated in Figure 6 22. Figure 6 22. A.M. 1.5 illuminated J -V characteristics of : PCBM solar cells showing the similar effect that adding PBTC has on the performance compared to the addition of C60PBTC
176 A B C D E F G H Figure 6 20. Tapping mode AFM scans of the surface of P3HT : PCBM solar cells i n which the additives have been incorporated into the blend. A) 0 % additives. B) 0.5 % C60PBTC C) 1 % C60PBTC D) 2 % C60-PBTC E) 3 % C60PBTC. F) 5 % C60PBTC G) 10 % C60-PBTC H) 3% PBTC. The morphological changes of the solar cells are illu strated in Figure 6 23. The cell containing 0% additives can be seen to have the nanometer -scale, phase -separated domains characteristic of P3HT:PCBM cells. The addition of C60PBTC additive in increasing amounts corresponds to decreasing levels of small -scale surface roughness. In the case of the P3HT:PCBM solar cell with 3% PBTC additives the morphology is seen to take on very different surface features, but the phase separated domains have disappeared.
177 It is concluded from this work that both PBTC and C60-PBTC have a negative effect on the performance of P3HT:PCBM solar cells. This effect is likely related to the impact that the additives have on the P3HT:PCBM domain boundaries.
178 CHAPTER 7 CONCLUSIONS AND FUTU RE WORK The research fields that have deve loped around conjugated polymers represent many exciting new frontiers. Polymers offer a broad range of attractive properties that have gained the interest of many companies and fueled the rapid progress of conjugated polymer research. Applications of conjugated polymers in low -cost electronics have become a reality, and the commercial development of polymer -based solar cells is on the brink of occurring. This work has investigated the hole transport properties of several different species of conjugated p olymers. The hole -mobility of P3HT was found to be dependent upon regioregularity. Regioregular samples were found to exhibit a negative field -dependence. PTVBT, and PBTC, and p PtBTD Th, were all found to have relatively low hole -mobilities with positi ve field dependence. Optimization of P3HT : PCBM bulk heterojunction solar cells led to efficiencies of ~4 %, which is near to those of the best published results in the field (~4 5 %). Solar cells made using PTVBT reached power conversion efficiencies of 0.5 % and exhibited broad photovoltaic behavior extending to w avelengths greater than 800 nm. Further fine -tuning of the fabrication parameters of P3HT : PCBM solar cells should result in slight improvements in the power conversion efficiencies. The solar c ells made using PTVBT exhibited low power conversion efficiencies which have been attributed to phase poor miscibility of PCBM and PTVBT. Future work will address reducing the unfavorable, large -scale phase sep aration seen in the blend films. Exploration of PBTC as a donor for bulk heterojunction solar cells found that external quantum efficiencies of greater than 40 % could be achieved, but considerable mismatch with the solar spectrum makes attaining high efficiencies unlikely. The use of C60-PBTC as an interfacial modifying agent produced small effects in PBTC: PCBM cells and caused detrimental effects to
179 the performance of P3HT: PCBM solar cells. The decreases in performance are likely attributable to modification of the P3HT: PCBM domain boundaries
180 L IST OF REFERENCES 1. Chiang, C. K. et al. Electrical -conductivity in doped polyacetylene. Physical Review Letters 39, 10981101 (1977). 2. Shirakawa, H., Louis, E. J., Macdiarmid, A. G., Chiang, C. K. & Heeger, A. J. Synthesis of electri cally conducting organic polymers halogen derivatives of polyacetylene, (Ch)X. Journal of the Chemical Society -Chemical Communications 578580 (1977). 3. Chiang, C. K. et al. Synthesis of highly conducting films of derivatives of polyacetylene, (Ch)X. J ournal of the American Chemical Society 100, 10131015 (1978). 4. Barford, W. Electronic and optical properties of conjugated polymers (Clarendon Press, Oxford, 2005). 5. Bao, Z. & Locklin, J. J. Organic field -effect transistors (CRC Press, Boca Raton, 200 7). 6. Irvin, J. A., Irvin, D. J. & Stenger -Smith, J. D. in Handbook of conducting polymers (eds. Skotheim, T. A. & Reynolds, J. R.) Chapter 9 (CRC Press, Boca Raton, 2007). 7. Salaneck, W. R., Stafstrom, S. & Bredas, J. L. Conjugated polymer surfaces and interfaces : electronic and chemical structure of interfaces for polymer light emitting devices (Cambridge University Press, Cambridge ; New York, 1996). 8. Dyer, A. L. in Handbook of conducting polymers (eds. Skotheim, T. A. & Reynolds, J. R.) Chapter 2 0 (CRC Press, Boca Raton, 2007). 9. Nalwa, H. S. Handbook of organic electronics and photonics (American Scientific Pub, Stevenson Ranch, Calif., 2008). 10. Mark, J. E. Physical properties of polymers handbook (Springer, New York, 2006). 11. Skotheim, T. A & Reynolds, J. R. Handbook of conducting polymers (CRC Press, Boca Raton, 2007). 12. Brown, W. H. Introduction to organic chemistry (Saunders College Pub., Fort Worth, Tex., 2000). 13. Ashcroft, N. W. & Mermin, N. D. Solid state physics (Holt Rinehart an d Winston, New York, 1976). 14. Roth, S. & Carroll, D. L. One -dimensional metals : conjugated polymers, organic crystals, carbon nanotubes (Wiley -VCH, Weinheim, 2004). 15. Peierls, R. E. More surprises in theoretical physics (Princeton University Press, Pr inceton, N.J., 1991). 16. Hsu, J. H. et al. Correlation between optical properties and chain length in a quasi -one dimensional electronic polymer. Journal of Physical Chemistry B 106, 8582 8586 (2002).
181 17. Salzner, U., Lagowski, J. B., Pickup, P. G. & Poir ier, R. A. Comparison of geometries and electronic structures of polyacetylene, polyborole, polycyclopentadiene, polypyrrole, polyfuran, polysilole, polyphosphole, polythiophene, polyselenophene and polytellurophene. Synthetic Metals 96, 177189 (1998). 18. Kittel, C. Introduction to solid state physics (Wiley, Hoboken, NJ, 2005). 19. Reddinger, J. L. & Reynolds, J. R. Molecular engineering of pi -conjugated polymers. Adv. Polym. Sci. 10, 93 116 (1999). 20. Layachi, M., Savaria, Y. & Rochefort, A. The effect of pi -coupling on the electronic properties of 1,4-dithiol benzene stacking. International Conference on MEMS, NANO and Smart Systems .(Banff, Canada.2004). 21. Moses, D., Schmechel, R. & Heeger, A. J. Effects of structural disorder and temperature on the distribution of exciton binding energy in poly(phenylene vinylene) films. Synthetic Metals 139, 807810 (2003). 22. Van der Horst, J. W. et al. Excitons in conjugated polymers from first principles. Computer Physics Communications 147, 331334 (2002). 23. Salaneck, W. R., Friend, R. H. & Bredas, J. L. Electronic structure of conjugated polymers: consequences of electronlattice coupling. Physics Reports 319, 231251 (1999). 24. Stafstrom, S. in Handbook of conducting polymers (eds. Skotheim, T. A. & Reynolds, J. R.) Chapter 2 (CRC Press, Boca Raton, 2007). 25. Bredas, J. L. & Street, G. B. Polarons, bipolarons, and solitons in conducting polymers. Accounts of Chemical Research 18 309315 (1985). 26. Heeger, A. J., Kivelson, S., Schrieffer, J. R. & Su, W. P. Solitons in conducting polymers. Reviews of Modern Physics 60, 781850 (1988). 27. Su, W. P., Schrieffer, J. R. & Heeger, A. J. Solitons in polyacetylene. Physical Review Letters 42, 16981701 (1979). 28. Kline, R. J. & McGehee, M. D. Morphology and charg e transport in conjugated polymers. Polymer Reviews 46, 27 45 (2006). 29. Sapoval, B. & Hermann, C. Physics of semiconductors (Springer -Verlag, New York, 1995). 30. Mott, N. F. & Gurney, R. W. Electronic processes in ionic crystals (Clarendon Press, Oxford, 1948). 31. Natali, D. & Sampietro, M. Field dependent mobility from space -charge-limited current voltage curves. Journal of Applied Physics 92, 53105318 (2002).
182 32. Murgatroyd, P. N. Theory of space -charge -limited current enhanced by Frenkel effect. Journal of Physics D -Applied Physics 3 151156 (1970). 33. Frenkel, J. On pre breakdown phenomena in insulators and electronic semi conductors. Physical Review 54, 647648 (1938). 34. Bisquert, J., Montero, J. M., Bolink, H. J., Barea, E. M. & Garcia Belm onte, G. Thickness scaling of space -charge-limited currents in organic layers with field or density -dependent mobility. Physica Status Solidi a-Applications and Materials Science 203, 37623767 (2006). 35. Bozano, L., Carter, S. A., Scott, J. C., Malliara s, G. G. & Brock, P. J. Temperature and field -dependent electron and hole mobilities in polymer light -emitting diodes. Applied Physics Letters 74, 11321134 (1999). 36. Scott, J. C. et al. Charge transport processes in organic light -emitting devices. Synt hetic Metals 111, 289293 (2000). 37. Goh, C., Kline, R. J., McGehee, M. D., Kadnikova, E. N. & Frechet, J. M. J. Molecular weight dependent mobilities in regioregular poly(3-hexyl -thiophene) diodes. Applied Physics Letters 86, 1 3 (2005). 38. Tanase, C., Meijer, E. J., Blom, P. W. M. & de Leeuw, D. M. Unification of the hole transport in polymeric field effect transistors and light -emitting diodes. Physical Review Letters 91, 1 4 (2003). 39. Tanase, C., Blom, P. W. M., de Leeuw, D. M. & Meijer, E. J. Charg e carrier density dependence of the hole mobility in poly(p -phenylene vinylene). Physica Status Solidi aApplied Research 201 12361245 (2004). 40. Becquerel, A. E. Memoire sur les effets electiques produits sous l'influence des rayons solaires. Comptes R endus 9 561567 (1839). 41. Williams, R. Becquerel photovoltaic effect in binary compounds. Journal of Chemical Physics 32, 15051514 (1960). 42. Chapin, D. M., Fuller, C. S. & Pearson, G. L. A new silicon p -n junction photocell for converting solar radia tion into electrical power. Journal of Applied Physics 25, 676677 (1954). 43. Pinto, M. R. Has the sun finally risen on photovoltaics? Symposium on VLSI Technology .2 5 (Honolulu, Hawaii.June 17 19 2008). 44. Schiermeier, Q., Tollefson, J., Scully, T., Wit ze, A. & Morton, O. Energy alternatives: electricity without carbon. Nature 454, 816823 (2008). 45. Kim, J. Y. et al. Efficient tandem polymer solar cells fabricated by all -solution processing. Science 317, 222225 (2007).
183 46. Richard, A., Panofsky, H. A., Cahir, J. J. & Albert, R. A. The atmosphere (Charels E. Merrill, Columbus, 1978). 47. Shockley, W. & Queisser, H. J. Detailed balance limit of efficiency of p n junction solar cells. Journal of Applied Physics 32, 510519 (1961). 48. De Vos, A. Detailed balance limit of the efficiency of tandem solar cells. J. Phys. D: Appl. Phys. 13, 839846 (1980). 49. Schneider, K. World record: 41.1% efficiency reached for multi junction solar cells at Fraunhofer ISE 2009. http://www.ise.fraunhofer.de/press and -media/pdfs -zu presseinfos -englisch/2009/press release -wor ld record 41.1 -efficiency reached -for -multi junction -solar -cells at -fraunhofer -ise -pdf -file (Last Accessed: July, 2009) 50. King, R. R. et al. 40% efficient metamorphic GaInP/GaInAs/Ge multijunction solar cells. Applied Physics Letters 90 (2007). 51. NRE L Public Relations Office. NREL solar cell sets world efficiency record at 40.8 percent 2009. http://www.nrel.gov/news/press/2008/625.html (Last Accessed: July, 2009) 52. University of Delawar e Office of Public Relations. UD-led team sets solar cell record, joins DuPont on $100 million project 2007. http://www.udel.edu/PR/UDaily/2008/jul/solar072307.html (Last Accessed: July, 2009) 53. Pochettino, A. Sul comportamento foto -elettrico dell' antracene. Acad. Lincei. Rendic. 15, 35563 (1906). 54. Tang, C. W. Multilayer organic photovoltaic elements. United States Patent 4,164,431 (1979). 55. Tang, C. W. 2 layer organic photo voltaic cell. Applied Physics Letters 48, 183 185 (1986). 56. Deibel, C. How do organic solar cells function? part zero 2008. http://blog.d isorderedmatter.eu/2008/02/22/how do-organic -solar -cells -function -part zero/#more 11. (Last Accessed: July, 2009) 57. Deibel, C. Picture story how do organic solar cells function? 2008. http://blog.disorderedmatter.eu/2008/06/05/picture -story -how -do -organic -solar -cells function/ (Last Accessed: July, 2009) 58. Sariciftci, N. S., Smilowitz, L., Heeger, A. J. & Wudl, F. Photoinduced electron -tran sfer from a conducting polymer to buckminsterfullerene. Science 258, 14741476 (1992). 59. Peumans, P. & Forrest, S. R. Very -high -efficiency double -heterostructure copper phthalocyanine/C 60 photovoltaic cells. Applied Physics Letters 79, 126128 (2001).
184 6 0. Yu, G., Gao, J., Hummelen, J. C., Wudl, F. & Heeger, A. J. Polymer photovoltaic cells enhanced efficiencies via a network of internal donor acceptor heterojunctions. Science 270, 17891791 (1995). 61. Hummelen, J. C. et al. Preparation and characterization of fulleroid and methanofullerene derivatives. Journal of Organic Chemistry 60, 532538 (1995). 62. Yamanari, T., Taima, T., Sakai, J. & Saito, K. Origin of the open -circuit voltage of organic thin -film solar cells based on conjugated polymers. Solar Energy Materials and Solar Cells 93, 759761 (2009). 63. Uhrich, C. et al. Origin of open circuit voltage in planar and bulk heterojunction organic thin -film photovoltaics depending on doped transport layers. Journal of Applied Physics 104 (2008). 64. Bra bec, C. J., Shaheen, S. E., Winder, C., Sariciftci, N. S. & Denk, P. Effect of LiF/metal electrodes on the performance of plastic solar cells. Applied Physics Letters 80, 12881290 (2002). 65. Mihailetchi, V. D., Blom, P. W. M., Hummelen, J. C. & Rispens, M. T. Cathode dependence of the open circuit voltage of polymer : fullerene bulk heterojunction solar cells. Journal of Applied Physics 94, 68496854 (2003). 66. Jabbour, G. E. et al. Highly efficient and bright organic electroluminescent devices with an a luminum cathode. Applied Physics Letters 71, 17621764 (1997). 67. Turak, A., Huang, C. J., Grozea, D. & Lu, Z. H. Oxidation of LiF -coated metal surfaces. Journal of the Electrochemical Society 154, H691 H697 (2007). 68. Brown, T. M. et al. The influence o f LiF thickness on the built in potential of blue polymer light emitting diodes with LiF/Al cathodes. Synthetic Metals 124, 15 17 (2001). 69. Jonsson, S. K. M., Carlegrim, E., Zhang, F., Salaneck, W. R. & Fahlman, M. Photoelectron spectroscopy of the conta ct between the cathode and the active layers in plastic solar cells: The role of LiF. Japanese Journal of Applied Physics Part 1-Regular Papers Short Notes & Review Papers 44, 36953701 (2005). 70. Janssen, G. et al. Optimization of morphology of P3HT/PCBM films for organic solar cells: effects of thermal treatments and spin coating solvents. European Physical Journal Applied Physics 37, 287290 (2007). 71. Shaheen, S. E. et al. 2.5% efficient organic plastic solar cells. Applied Physics Letters 78, 841843 (2001). 72. Li, G. et al. High -efficiency solution processable polymer photovoltaic cells by self organization of polymer blends. Nature Materials 4 864868 (2005).
185 73. Campoy Quiles, M. et al. Morphology evolution via self -organization and lateral and v ertical diffusion in polymer: fullerene solar cell blends. Nature Materials 7 158164 (2008). 74. Li, G., Shrotriya, V., Yao, Y. & Yang, Y. Investigation of annealing effects and film thickness dependence of polymer solar cells based on poly(3 -hexylthioph ene). Journal of Applied Physics 98 (2005). 75. Ma, W. L., Yang, C. Y., Gong, X., Lee, K. & Heeger, A. J. Thermally stable, efficient polymer solar cells with nanoscale control of the interpenetrating network morphology. Advanced Functional Materials 15, 1 6171622 (2005). 76. Yoshikawa, O., Sonobe, T., Sagawa, T. & Yoshikawa, S. Single mode microwave irradiation to improve the efficiency of polymer solar cell based on poly(3 hexylthiophene) and fullerene derivative. Applied Physics Letters 94, 1 3 (2009). 7 7. Peet, J. et al. Efficiency enhancement in low -bandgap polymer solar cells by processing with alkane dithiols. Nature Materials 6 497500 (2007). 78. Lee, J. K. et al. Processing additives for improved efficiency from bulk heterojunction solar cells. Jo urnal of the American Chemical Society 130, 36193623 (2008). 79. Moule, A. J. & Meerholz, K. Minimizing optical losses in bulk heterojunction polymer solar cells. Applied Physics BLasers and Optics 86, 721727 (2007). 80. Hadipour, A., de Boer, B. & Blom P. W. M. Solution -processed organic tandem solar cells with embedded optical spacers. Journal of Applied Physics 102, 1 6 (2007). 81. Gilot, J., Wienk, M. M. & Janssen, R. A. J. Double and triple junction polymer solar cells processed from solution. Appl ied Physics Letters 90 (2007). 82. Ameri, T., Dennler, G., Lungenschmied, C. & Brabec, C. J. Organic tandem solar cells: A review. Energy & Environmental Science 2 347 363 (2009). 83. Zhang, F. L. et al. High photovoltage achieved in low band gap polymer solar cells by adjusting energy levels of a polymer with the LUMOs of fullerene derivatives. Journal of Materials Chemistry 18, 54685474 (2008). 84. Wienk, M. M., Struijk, M. P. & Janssen, R. A. J. Low band gap polymer bulk heterojunction solar cells. Che mical Physics Letters 422, 488491 (2006). 85. Tvingstedt, K., Andersson, V., Zhang, F. & Inganas, O. Folded reflective tandem polymer solar cell doubles efficiency. Applied Physics Letters 91 (2007). 86. Chemat Technology Inc. Spin coating. 2002. http://www.enlabproducts.com/spincoating. (Last Accessed: Jan 12, 2005)
186 87. Luo, S. C., Craciun, V. & Douglas, E. P. Instabilities during the formation of electroactive polymer thin films. Langmuir 21, 28812886 (2005). 88. Mattox, D. M. Handbook of physical vapor deposition (PVD) processing : film formation, adhesion, surface preparation and contamination control (Noyes Publications, Westwood, N.J., 1998). 89. Stoneham, A. M. et al. Understanding electron f low in conducting polymer films: injection, mobility, recombination and mesostructure. Journal of Physics -Condensed Matter 14, 98779898 (2002). 90. Choulis, S. A. et al. Investigation of transport properties in polymer/fullerene blends using time -of -fligh t photocurrent measurements. Applied Physics Letters 83, 38123814 (2003). 91. Bassler, H. Injection, transport and recombination of charge carriers in organic light emitting diodes. Polymers for Advanced Technologies 9 402418 (1998). 92. Scher, H. & Mon troll, E. W. Anomalous transit -time dispersion in amorphous solids. Physical Review B 12, 24552477 (1975). 93. Chabinyc, M. L. Characterization of semiconducting polymers for thin film transistors. Journal of Vacuum Science & Technology B 26, 445457 (2008). 94. Chabinyc, M. L. & Salleo, A. Materials requirements and fabrication of active matrix arrays of organic thin -film transistors for displays. Chemistry of Materials 16 45094521 (2004). 95. Brown, P. J. et al. Effect of interchain interactions on the absorption and emission of poly(3 -hexylthiophene). Physical Review B 67 (2003). 96. Brown, T. M. et al. Built -in field electroabsorption spectroscopy of polymer light emitting diodes incorporating a doped poly(3,4 ethylene dioxythiophene) hole injection l ayer. Applied Physics Letters 75, 16791681 (1999). 97. Ballantyne, A. M. et al. The effect of poly(3 hexylthiophene) molecular weight on charge transport and the performance of polymer : fullerene solar cells. Advanced Functional Materials 18, 23732380 ( 2008). 98. Mozer, A. J. & Sariciftci, N. S. Negative electric field dependence of charge carrier drift mobility in conjugated, semiconducting polymers. Chemical Physics Letters 389, 438442 (2004). 99. Mozer, A. J. et al. Charge carrier mobility in regiore gular poly(3 hexylthiophene) probed by transient conductivity techniques: A comparative study. Physical Review B 71 (2005).
187 100. Kazukauskas, V., Pranaitis, M., Cyras, V., Sicot, L. & Kajzar, F. Negative mobility dependence on electric field in poly(3 alky lthiophenes) evidenced by the charge extraction by linearly increasing voltage method. Thin Solid Films 516, 89888992 (2008). 101. Pandey, S. S., Takashima, W., Endo, T., Rikukawa, M. & Kaneto, K. Poly(3 butylthiophene): conjugated polymer with record high TOF mobility. Synthetic Metals 121, 15611562 (2001). 102. Huo, L. J., He, C., Han, M. F., Zhou, E. J. & Li, Y. F. Alternating copolymers of electron -rich arylamine and electron -deficient 2,1,3 benzothiadiazole: synthesis, characterization and photovolta ic properties. Journal of Polymer Science Part aPolymer Chemistry 45, 38613871 (2007). 103. Gaupp, C. L. & Reynolds, J. R. Multichromic copolymers based on 3,6 -bis(2 (3,4 ethylenedioxythiophene)) -N alkylcarbazole derivatives. Macromolecules 36 63056315 (2003). 104. Witker, D. & Reynolds, J. R. Soluble variable color carbazole containing electrochromic polymers. Macromolecules 38, 76367644 (2005). 105. Burgi, L. et al. High -mobility ambipolar near infrared light emitting polymer field effect transistors. Advanced Materials 20, 2217+ (2008). 106. Mei, J. et al. Low -band -gap platinum acetylide polymers as active materials for organic solar cells. ACS Applied Materials & Interfaces 1 150161 (2009). 107. Wong, W. Y. et al. Metallated conjugated polymers a s a new avenue towards high efficiency polymer solar cells. Nature Materials 6 521527 (2007). 108. Jin, S. H. et al. Optimization of process parameters for high -efficiency polymer photovoltaic devices based on P3HT : PCBM system. Solar Energy Materials and Solar Cells 91, 11871193 (2007). 109. Chang, J. F. et al. Enhanced mobility of poly(3 hexylthiophene) transistors by spincoating from high -boiling -point solvents. Chemistry of Materials 16, 47724776 (2004). 110. Woo, C. H., Thompson, B. C., Kim, B. J ., Toney, M. F. & Frechet, M. J. The Influence of Poly(3 -hexylthiophene) Regioreigularity on Fullerene Composite Solar Cell Performance. Journal of the American Chemical Society 130, 1632416329 (2008). 111. Jo, J., Kim, S. S., Na, S. I., Yu, B. K. & Kim, D. Y. Time dependent morphology evolution by annealing processes on polymer:fullerene blend solar cells. Advanced Functional Materials 19, 866874 (2009). 112. Miller, S. et al. Investigation of nanoscale morphological changes in organic photovoltaics duri ng solvent vapor annealing. Journal of Materials Chemistry 18, 306312 (2008).
188 113. Ko, C. J., Lin, Y. K. & Chen, F. C. Microwave annealing of polymer photovoltaic devices. Advanced Materials 19, 3520+ (2007). 114. Kim, J. Y. et al. New architecture for h igh-efficiency polymer photovoltaic cells using solution -based titanium oxide as an optical spacer. Advanced Materials 18, 572576 (2006). 115. Spanggaard, H. & Krebs, F. C. A brief history of the development of organic and polymeric photovoltaics. Solar E nergy Materials and Solar Cells 83, 125146 (2004). 116. Chen, L. M., Hong, Z. R., Li, G. & Yang, Y. Recent progress in polymer solar cells: manipulation of polymer: fullerene morphology and the formation of efficient inverted polymer solar cells. Advanced Materials 21, 14341449 (2009). 117. Kim, K., Liu, J., Namboothiry, M. A. G. & Carroll, D. L. Roles of donor and acceptor nanodomains in 6% efficient thermally annealed polymer photovoltaics. Applied Physics Letters 90 (2007). 118. Peet, J., Senatore, M. L., Heeger, A. J. & Bazan, G. C. The role of processing in the fabrication and optimization of plastic solar cells. Advanced Materials 21, 15211527 (2009). 119. Blouin, N. et al. Toward a rational design of poly(2,7 -carbazole) derivatives for solar cells. Journal of the American Chemical Society 130, 732 742 (2008). 120. Zou, Y. P. et al. A high -mobility low -bandgap poly(2,7 -carbazole) derivative for photovoltaic applications. Macromolecules 42, 28912894 (2009). 121. Park, S. H. et al. Bulk heterojunction solar cells with internal quantum efficiency approaching 100%. Nature Photonics 3 297301 (2009).
189 BIOGRAPHICAL SKETCH Nathan C. Heston was born in Coudersport, Pennsylvania in 1978. He grew up with an older sister and a younger brother on a farm near Galeton, Pennsylvania. He spent his entire youth there, attending Galeton Area High School, before leaving for college at the age of 18. Nathan attended Millersville University where he studied biology, chemistry, and physics from 1997 until graduating w ith B achelor of Science in physics in 2002. He completed a year of graduate study in physics at the Un iversity of Virginia, before beginning his doctoral degree at the University of Florida in 2004. At Florida, h e joined the research group of Dr. David Tanner of the Department of Physics and worked in close collaboration with the group of Dr. John Reynolds of the Department of Chemistry on research in the fields of conjugated polymer charge transport and organic solar cells.