1 NANOSTRUCTURED THIN FILMS FOR ORGANIC PHOTOVOLTAIC CELLS AND ORGANIC LIGHT -EMITTING DIODES By YING ZHENG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2 2009 Ying Zheng
3 To my family and friends
4 ACKNOWLEDGMENTS I would like to express my sincere gratitude to Prof. Jiangeng Xue, for his guidance and support during my four years gra duate study at University of Florida His substantial k nowledge in the field and high standard s on academic research has sharpened my m ind and helped me a lot to accomplish my goals I would also like to thank my supervisory committee, Prof. Paul Holl o way, Prof. Franky So, Prof. Susan Sinnott and Prof. John Reynolds, for their advice and support on my research I owe many people for thei r active collaboration on accomplishing this thesis. Among them I would like to thank Sharon Pregler Jason Myers and Prof. Sinnott for conducting the M olecular D ynamic s simulation work Sang Hyun Eom for extensive discussion on phosphorescent OLEDs, and L ei Qian for synthesis of various ZnO material s. I am grateful to the fellow graduate students and post -docs that I work with during my graduate study. I would like to thank William Hammond Yixing Yang, Weiran Cao, Edward Wrzesniewski, Renjia Zhou, Zhifeng Li, Jiaomin Ouyang, Tengkuan Tseng, and Debasis Bera for the joyful discussions with them about the research projects and beyond. I also need to thank Peng Xu, Lu Cai, Wei Qiu, Haixuan Xu Song Xue, Liwen Jin and Yuping Li for sharing the joy and support outside the lab. I also need to acknowledge the financial support from US Department of Energy and National Science Foundation. Finally, this thesis would not have been complete without the support and understanding from my parents. I show my deepest gr atitude and appreciation to them for their love, trust and encouragement throughout my college study.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 8 LIST OF FIGURES .............................................................................................................................. 9 ABSTRACT ........................................................................................................................................ 13 CHAPTER 1 INTRODUCTION TO ORGANIC SEMICONDUCTORS ..................................................... 15 Introduction ................................................................................................................................. 15 Electronic Structures of Organic Semiconductors .................................................................... 16 Electronic State of Isolated Molecule ................................................................................. 16 Energy Level and Charge Carriers Transport .................................................................... 19 Excitons ................................................................................................................................ 22 Energy Transfer ................................................................................................................... 23 Growth and Structure of Organic Thin Film ............................................................................. 24 Vacuum Thermal Evaporation and Spin-Coating .............................................................. 24 Amorphous and Crystalline Thin Films ............................................................................. 26 Heterojunctions .................................................................................................................... 28 Organic Photovoltaic Cells ......................................................................................................... 29 Progre ss of Organic Photovoltaic Cells .............................................................................. 29 Principles of Organic Photovoltaic Cells ........................................................................... 30 Photocurrent Generation ..................................................................................................... 32 Organic Light Emitting Diodes .................................................................................................. 35 Progress of Organic Light -Emitting Diodes ...................................................................... 35 Principles of Organic Light Emitting Diodes .................................................................... 36 Phosphorescent Organic Light -Emitting Diodes ............................................................... 37 Overview of This Thesis ............................................................................................................. 39 2 MEASUREMENT OF ORGANIC PHOTOVOLTAIC CELLS AND ORGANIC LIGHT EMITTING DIODE S .................................................................................................... 41 Introduction ................................................................................................................................. 41 Measurement of Organic Photovoltaic Cells ............................................................................. 41 Power Conversion Efficiency Measurement with Spectrum Mismatch Factor ............... 42 External Quantum Efficiency Measurement ...................................................................... 45 Measurement of Organic Light Emitting Diodes ...................................................................... 48 Measurement Setup for Organic Light Emitting Diodes .................................................. 48 Conclusions ................................................................................................................................. 51 3 PHASE SEPARATION IN MOLECULAR DONOR -ACCEPTOR MIXTURE .................. 52
6 Introduction ................................................................................................................................. 52 Thermodynamics and Kinetics of Phase Separation ................................................................. 55 Behavior of Regular Solution ............................................................................................. 55 Spinodal Decomposition and Nucleation ........................................................................... 57 Structural and Morphological Control of Pentacene:C60 Mixtures .......................................... 59 Effect of Mixing Ratio on Structure and Morphology ...................................................... 59 Effect of Deposition Rate on Structure and Morphology ................................................. 63 Molecular Dynamics Simulation of Pentacene:C60 Mixtures .................................................. 68 Method of Molecular Dynamics Simulation ...................................................................... 68 Structural Evolution in Pentacene:C60 Mixtures ................................................................ 70 Effect of Mixing Ratio and Deposition Rate ..................................................................... 74 Photovoltaic Performance of Pentacene:C60 Mixtures ............................................................. 76 Discussion .................................................................................................................................... 79 Conclusions ................................................................................................................................. 83 4 ORGANIC PHOTOVOLTAIC CELLS WITH ALIGNED CRYSTALLINE MOLECULAR NANORODS .................................................................................................... 84 Introduction ................................................................................................................................. 84 Oblique Angle Deposition .......................................................................................................... 87 Structur e and Morphology Study of CuPc Nanorods ............................................................... 90 Structure and Morphology Evolution of Nanorods ........................................................... 91 Effect of Incident Angle ...................................................................................................... 93 Effect o f Substrate Rotation ................................................................................................ 95 Effect of Surface Property and Deposition Rate ................................................................ 97 Structure of CuPc Nanorods ............................................................................................. 100 Orga nic Photovoltaic Cells Based on CuPc Nanorods and PCBM ........................................ 101 Infiltration of PCBM into Nanorod Arrays ...................................................................... 101 Optimization of Cupc Nanorods/PCBM Cell .................................................................. 103 Planar Plus Nanorods Cupc/PCBM Cell .......................................................................... 104 Discussion .................................................................................................................................. 107 Conclusions ............................................................................................................................... 109 5 EFFICIENT DEEP -BLUE PHOSPHORESCENT ORGANIC LIGHT -EMITTING DIODES ..................................................................................................................................... 110 Introduction ............................................................................................................................... 110 Charge Carrier and Exciton Blocking in Phosphorescent OLED .......................................... 112 Effect of Electron and Exction Blocking Layer ............................................................... 112 Evidence for Effective Electron and Exciton Blocking .................................................. 118 Effect of Hole and Exciton Blocking Layer ..................................................................... 119 Reducing Driving Voltage of Phosphorescent OLED ............................................................ 121 Efficient Electron and Hole Injection Layer .................................................................... 122 Device Using p -i -n Structure ............................................................................................ 126 Discussion .................................................................................................................................. 130 Conclusions ............................................................................................................................... 132
7 6 LOW TURN ON VOLTAGE OF MEH -PPV POLYMER LIGHT -EMITTING DIODES ..................................................................................................................................... 134 Introduction ............................................................................................................................... 134 Low Turn-on Voltage with ZnO Nanoparticles as Electron Injection Layer ........................ 136 Morphology and Structure of ZnO Nanoparticles ........................................................... 136 LiF versus ZnO Nanoparticle as Electron Injection Layer ............................................. 138 Application of ZnO Nanoparti c les on Other Polymer Light Emitting Diodes .............. 142 Mechanism Study for Low Turn -on Voltage Phenomenon.................................................... 143 Auger Assisted Injection Model ....................................................................................... 144 Comparison of Different ZnO Materials .......................................................................... 146 Temperature Dependence of Turn on Voltage ................................................................ 149 Electron -only Device ......................................................................................................... 152 Discussion .................................................................................................................................. 153 Conclusions ............................................................................................................................... 155 7 CONCLUSIONS AND FUTURE WORK .............................................................................. 157 Conclusions ............................................................................................................................... 157 Future Work ............................................................................................................................... 158 Control Growth of Organic Nanostructure ...................................................................... 159 Organic Light Emitting Diodes ........................................................................................ 160 Af terword ................................................................................................................................... 161 LIST OF REFERENCES ................................................................................................................. 162 BIOGRAPHICAL SKETCH ........................................................................................................... 172
8 LIST OF TABLES Table page 1 1 Molecular structure of the family of polyacene molecules. ................................................ 19 1 2 Crystallographic data on anthracene, teteracene and pentacene crystals. ........................... 27 3 1 Calculated and exper imental determined cohesive energy for C60 and pentacene. ........... 69 3 2 Comparison of PV performance of pentacene:C60 BHJ devices ........................................ 78 4 1 Comparison of PV performance of CuPc nanorod/PCBM devices ................................. 107 5 1 Comparison of performance of three FIr6 PHOLEDs. ...................................................... 130
9 LIST OF FIGURES Figure page 1 1 A flexible small molecule OLED display and plastic solar cell ........................................ 16 1 2 Molecular structure of some organic semiconductors. ........................................................ 17 1 3 -conjugated ethene molecule. ........................................................... 18 1 4 Energy level diagram of an isolated molecule, a molecular crystal a nd an amorphous solid. ........................................................................................................................................ 21 1 5 Schematic of three different types of exciton in molecular solid. ....................................... 23 1 6 Schematic of vacuum thermal evaporation (VTE) and spin -coating. ................................. 25 1 7 Schematic structure of unit cell of an anthracene crystal. ................................................... 26 1 8 Energy conversion efficiency progress of state -of -the art research on diff e rent photovoltaic devices. .............................................................................................................. 29 1 9 Typical structure of organic photovoltaic cells and its equivalent circuit. ......................... 31 1 10 Four steps in the photocurrent generation process in OPV cells ........................................ 33 1 11 Schematic of bulk heterojunction with percolated donor accepter phase. ......................... 34 1 12 Principle of operation of an organic light -emitting diode (OLED). .................................... 36 1 13 Jablonski diagram for the fluorescence and phosphorescence. ........................................... 38 2 1 Diagram of power conversion efficiency (PCE) measurement for OPV cells. .................. 42 2 2 Calculation of spectrum mismatch factor with specific solar spectrum and reference cell. .......................................................................................................................................... 45 2 3 Diagram of external quantum efficiency (EQE) measurement for OPV cells. .................. 46 2 4 External quantum eff iciency as a function of the wavelength for ITO/CuPc (20 nm)/C60 (40 nm)/BCP (8 nm)/Al OPV cell ......................................................................... 47 2 5 Luminance measurement of OLEDs. .................................................................................... 49 3 1 Different morphological structure in molecular donor accetpor mixtures and the charge transport within. ......................................................................................................... 53 3 2 The variation, with composition, of the molar Gibbs free energy of the formati on of a binary regular solution. .......................................................................................................... 56
10 3 3 Phase separation through spinodal decomposition. ............................................................. 58 3 4 Phase separation through nucleation. .................................................................................... 59 3 5 X ray diffraction patterns for 30 nm thick pentacene and pentacene:C60 mixed films. .... 60 3 6 AFM images of 3 nm thick pentacene and pentacene: C60 films on Si (100) substrates ... 62 3 7 AFM and SEM images of 50 nm thick pentacene:C60 films deposited at 0.6 /s on Si substrate ................................................................................................................................. 64 3 8 AFM and SEM images of 50 nm thick pentacene:C60 films deposited at 6 /s on Si substrate ................................................................................................................................. 66 3 9 X ray diffraction patterns for 50 nm pentacene:C60 (1:1) mixed films fabricated under 0.6 /s and 6 /s. ........................................................................................................ 67 3 10 Molecular model of petancene:C60 (2.3:1) (by weight) before and after 100 ps equilibration. ........................................................................................................................... 71 3 11 Pentacene stacks in the equilibrated pentacene:C60 (2.3:1) structure ................................ 72 3 12 C60 aggregation in pentacene:C60 (2.3:1) before and after 100 ps equilibaration. ............. 73 3 13 Number of two-molecule pentacene stacks versus the maximum allowed angle offset ....................................................................................................................................... 75 3 14 J V characteristics for devices with pentacene:C60 =1:1, 1:2 and 1:5.5 in dark and under illumination. ................................................................................................................. 77 3 15 J V characteristics for pentacene:C60 = 1:5.5 devices with the mixed layer deposited at ~ 0.6 /s and ~ 6 /s under illumination. ....................................................... 79 4 1 Schematic of ideal interdigitated bulk heterojunction. ........................................................ 85 4 2 Schematic of oblique angle deposition. ................................................................................ 87 4 3 Different nanorod structure a chieved under OAD process. ................................................ 90 4 4 SEM images of approximately 150 nm thick CuPc nanorod film at different stages of growth. .................................................................................................................................... 92 4 5 SE M and AFM images of 100 nm thick CuPc nanorod film fabricated with different incident angle. ......................................................................................................................... 94 4 6 Topographic and cross -sectional SEM images of CuPc nanorods grown on a stationary or rotational substrate. .......................................................................................... 96
11 4 7 SEM images of thick and thin CuPc nanorod films gr own on Si and SiO2 substrate under identical OAD conditions. ........................................................................................... 98 4 8 SEM images of CuPc nanorod film grown under identical OAD conditions but different deposition rate. ........................................................................................................ 99 4 9 XRD pattern of a CuPc nanoro d film with 300 nm long nanorods and a 100 nm thick flat CuPc film. ...................................................................................................................... 100 4 10 Cross -sectional SEM images of CuPc nanorod/PCBM composited films. ...................... 101 4 11 Absorption spectra of CuPc nanorod film before and after infiltration of PCBM. .......... 102 4 12 J V characteristics of three CuPc/PCBM photovoltaic cells under 1 sun AM 1.5 illumination. .......................................................................................................................... 103 4 13 J V characteristics of three planar plus nanorod CuPc/PCBM photovoltaic cells under 1 sun AM 1.5 illumination. ....................................................................................... 105 4 14 Photovoltaic performance versus incident irradiance ( Pin) of PlanarNR, NR R, NR -S and Planar device. ................................................................................................................ 106 5 1 Schematic energy level diagram of the deep -blue phosphorescent organic light emitting diodes. .................................................................................................................... 113 5 2 EL spectra of two PHOLEDs with the structures of ITO/NPD or TAPC (40 nm) /mCP (15 nm)/UGH2:10% FIr6 (25 nm)/BCP (40 nm)/LiF (1 nm) /Al .......................... 114 5 3 Performance of PHOLEDs with either NPD or TAPC as hole transport layer (HTL). ... 115 5 4 Luminance power and external quantum efficiency versus current density ( J ) for TAPC -based devices with a varying mCP layer thickness. ............................................... 117 5 5 EL spectra of TAPC -based and NPD -based PHOLEDs without mCP layer. .................. 118 5 6 Performance of PHOLEDs with ITO/TAPC (40 nm)/mCP (10 nm)/UGH2:10 wt % FIr6 (20 nm)/HBL (5nm)/BCP (40 nm)/LiF (1 nm) /Al. ................................................... 120 5 7 Conductivity study of F4 TCNQ doped MeO TPD films ................................................ 123 5 8 J V characteristics of ITO/TAPC or MeO TPD: 3 mol% F4 TCNQ (20nm)/TAPC (20 nm)/UGH2: 10 wt % FIr6 (20 nm)/Au (50 nm) hole -only device. ............................ 124 5 9 J V characteristics of ITO/UGH2: 10 wt % FIr6 (20 nm)/BCP (40 nm)/LiF/Al and ITO/UGH2: 10 wt % FIr6 (20 nm)/Bphen (20 nm)/Bphen:Li =1:1 (20 nm)/LiF/Al electron -only device. ............................................................................................................ 125
12 5 10 Device structure of the conventional NPD -based deep -blue PHOLEDs and the two p i -n structure PHOLEDs under study as well as the energy level diagram of p -i -n PHOLEDs. ............................................................................................................................ 127 5 11 Performance of PHOLEDs of conventional NPD based device and two p -i -n devices .. 128 6 1 Morphology of ZnO nanoparticle s. ..................................................................................... 137 6 2 X ray diffraction (XRD) pattern of ZnO NPs al ong with that of bulk ZnO .................... 138 6 3 Performance of devices using ZnO NPs/Al versus LiF/Al as cathode ............................ 140 6 4 Electroluminescence (EL) spectra of ZnO NPs device under different forward bi as ..... 141 6 5 Performance of other PLEDs with ZnO NPs as electron injection layer ........................ 142 6 6 Interfacial Auger recombination in ZnO NPs device. ....................................................... 144 6 7 Morpholo gies of different ZnO materials .......................................................................... 147 6 8 Performance of device with different ZnO electron injection layer. ................................ 148 6 9 Temperature dependence of J V characteristics of ZnO NPs and LiF device. .............. 150 6 10 Temperature dependence of L V characteristics of ZnO NPs and LiF device. ............. 151 6 1 1 J V characteristics of ZnO NPs and LiF electron -only device. ...................................... 153 7 1 Potential templates for controlled growth of organic nanorods. ....................................... 160
13 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 NANOSTRUCTURED THIN FILMS FOR ORGANIC PHOTOVOL TAIC CELLS AND ORGANIC LIGHT -EMITTING DIODES By Ying Zheng August 2009 Chair: Jiangeng Xue Major: Materials Science and Engineering Achieving e fficient organic optoelectronic devices, such as organic photovo ltaic (OPV) cells and organic light -emitting diodes (OLEDs), relies on the understanding of the formation of various organic nanostructure s as well as the fundamental of physical process es in device operation. The research presented in this thesis systemat ically investigate s the controlled growth of organic nanostructure through different approaches and their relationship to OPV cell performance. M oreover, new materials and device structure are explored to achieve efficient OLEDs which also provide further insight of the physical processes governing the performance of these devices We first investigated the phase separation process in a molecular mixed donor acceptor (D A) bulk heterojunction (BHJ) composed of pentacene and C60 suing a combination of experimental and computational approaches Both experiment characterization and the MD simulation reveals that strong aggregation of pentacene exists in the pentacene:C60 mixtures due to the among pentacene molecules By controlling the process ing conditions to suppress the pentacene aggregation to nanoscale leads to higher device efficiency as the more photogenerated ex citons are able to reach the D A interface and contribute to the photocurrent To circumvent the limits on phase separated D A mixed heterojunction an
14 interdigitated D A BHJ is synthesized through the oblique angle deposition (OAD) of copper phthalocyanine (CuPc). The morphology of CuPc nanorod arrays grown under the OAD process can be c ontrolled by careful selection of the processing conditions and we have achieved a high density, vertically aligned, polycrystalline CuPc nanorod array with nanorod size as small as 20 30 nm Successful infiltration of [6,6] phenyl -C61-butyric acid methyl ester (PCBM) into the optimized CuPc nanorod arrays has resulted in doubling of the power conversion efficiency of the OPV cell over planar heterojunction device based on the same D A materials We also show that the efficiency of a deep -blue phosphore scent OLED (PHOLED) can be significantly enhanced by improving the exciton and charge confinement in the multilayer organic stack A peak external quantum efficiency of (20 1) % is achieved, which approaches the theoretical maximum of PHOLED without specific out -coupling mechanism s W e further demonstrate PHOLEDs with enhanced power efficiency by using the p -i -n device structures to reduce driving voltage and achieved a maximum of (14 1) lm/W and (12 1) lm/W at a luminance of 100 cd/ m2. Moreover an ultra low turn-on voltage of ~ 1.3 V is observed in an orange -emitting polymer light -emitting diode (PLED) using ZnO nanoparticles as the electron injection layer. An Auger assisted electron injection mechanism is proposed to explain the low turn on vol tage. The novel ZnO nanoparicles electron injection layer opens a new way to reduce driving voltage in PLED.
15 CHAPTER 1 INTRODUCTION TO ORGA NIC SEMICONDUCTORS Introduction Research on organic semiconductors has made significant breakthrough s in the past two decades due to increasing knowledge of material synthesis device physics and nanostructure d thin film s F ruitful advances have been made especially on organic photovoltaic cells and organic light -emitting diodes.1 10 The future decrease in supply of fossil fuels coupled with the rapid growth of global demand on energy will boost the commercialization of energy related organic semiconductor products. The purpose of this chapter is to provide a brief overview of the knowledge related to science and technology on organic semiconductors The contents do not intend to cover every aspect of organic semiconductors but what is needed t o help interpret the works achiev ed in this thesis This review will be broken down several sections. The electronic structure of isolated organic molecule and their solid -state a ggregates will be presented in the first s ection, where intramol e cular and intermolecular interaction s and their effect on the energy transfer, exciton and charge carrier transport are briefly reviewed The following s ection will talk about the growth of organic thin film s and different type s of structure that can be achieved ins ide the organic thin film. The next two sections will discuss two of the most active ly studied organic optoelectronic devices: organic photovoltaic (OPV) cells and organic light -emitting diodes (OLED s ). Here, the operation principles along with typical opt ical and electrical characteristics of these two types of device are introduced. Moreover, the existing challenges in achieving efficient OPV cells and OLEDs will also be summarized. The outline of each chapter is presented in last section which serves as a reading guide for this thesis
16 Electronic Structure s of Organic Semiconductors Organic semiconductors are a class of semiconductor materials that exhibit different properties from the conventional inorganic semiconductors, such as Si and Ge.11 13 T he unique properties of organic materials have enabled fascinating application s that are v ery difficult to achieve by conventional inorganic semiconductor s such as large area, flexible lighting and displays, low -cost plastic solar cells and ink jet print ing of integrated circuit s (see Figure 1 1) All t hese revolutionary products could not hav e been realized without understanding t he peculiar electronic propertie s of organic semiconductors. Figure 1 1. A flexible small molecule OLED display made by Universal Display Corporation (left), and plastic solar cell s developed by Konarka (right). Image s courtesy of Universal Display Corporation and Konarka Technologies, Inc. Electronic State of Isolated Molecule Most organic semiconductors can be divided into two major categories: small molecular weight materials and polymers. Figure 1 2 lists a nu mber of typical organic molecules that are used in organic semiconductor devices. Note that all of these organic compound s are compose d of conjugated electron system s which corresponds to by the alternative appearance of C C and C = C in their molecul ar structure. I ncreased conjugated length within a molecular structure
17 Figure 1 2. Molecular structure of some organic semiconductors: MEH PPV: poly[2 -methoxy5 (2 -ethylhexyloxy) 1,4 -phenylenevinylene], fullerene C60, PVK: poly(9 vinylcarbazole) Alq3: tri(8 hydroxyquinoline)aluminum, CuPc: copper phthalocyanine, and pentacene generally lead s to stronger delocalization of electrons within the molecule and produces high charge mobility The simplest and most representative conjugated molecule is ethe ne, which is composed of two carbon atoms and four hydrogen atoms The sp2 hybridization of each carbon atom gives rise of three identical sp2 orbitals, leaving one pz orbital intact. The six sp2 orbitals of two carbon atoms will allow the formation of fiv bonds, four C H bonds and one C C bond The two remaining pz -bond between t wo carbon atoms in addition to the C C bond. Fig ure 1 3 shows the relative energy level of atomic orbitals of sp2 hybridized carbon atom and the mol conjugated molec ule of ethene -bonding is much weaker due to less overlap of the two parallel pz atomic orbitals compared with two sp2 orbitals forming C C O O n N n NN N N N N N N Cu N O N O N O A l MEH PPV PVK C60Alq3 CuPc Pentacene
18 Figure 1 conjugated ethene molecule with sp2 hybridized carbon and bonding. Therefore, T he energy for corresponds to the lowest energy needed to generate electronic excitation in conjugated molecu le s In most case s molecular orbitals (HOMO) and lowest unocc upied molecular orbitals (LUMO), respectively. The ene rgy difference between H OMO and LUMO strongly depends on the level of conjugati on in a molecule. Table 1 1 summarizes the wavelength of absorption and emission maximum for the family of polyacenes.14 It can be seen that as the degree of conjugation increase s (more benzene rings) the absorption and emission maxima red -shift accordingly, indicating a smaller energy gap between HOMO and LUMO. Thus, the optoelectronic properties of molecules can be tuned by adding different functional group to modify the conjugation level, reflecting the versatility of organic materials. ** Energysp2sp2p p antibonding orbitals bonding orbitals
19 Table 1 1. Molecular structure of the family of polyacene molecules, together with the wavelength of the main absorption and emission peak. Molecule Structure Absorption maximum Emission Maximum Benzene 205 nm 278 nm Naphthalene 286 nm 321 nm Anthracene 365 nm 400 nm Tetracene 390 nm 480 nm Pentacene 580 nm 640 nm Energy Level a nd Charge Carriers Transport The macroscopic optical and electrical properties of organic semiconductors are not only determined by the electronic structure of isolated molecules but also by t heir solid -state aggregates, whereas inorganic semiconductors are entirely formed by co va lent bonding. T here are two types of bonding present in molecular solid s The strong covalent -type interaction with typical interaction energy of 2 4 eV is present in the intramolecular interaction, while much weaker Van der Waals type of interaction with interaction energy of 103 102 eV dominates the in termolecular interaction.11 Generally, the total Van der Waals interaction energy UVdW between molecules can be expressed as: = + +
20 where Udi s is energy from dispersion interaction, Uind from inductive interaction and Udd from dipole -dipole interaction. In the case of non -polar molecules, to which most of the organic semiconductor materials belong, the inductive and dipole dipole interaction i s negligible. Therefore, the total intermolecular interaction energy mainly depends on the attractive dispersion interaction Udis and repulsive interaction Urep. The Lennard -Jones equation is an empirical formula that describe s the potential energy of intermolecular interaction :15 ( ) = ( ) + ( ) = 6 + 12 where r is the distance between the molecules and A and B are empirically derived constant s Potential energy of the system is minimize d at certain distance of r0, at which point attractive and repulsive interaction s equals. The weak intermolecular interaction produce s only slight changes in the electronic structure of molecules on fo rmation of the solid phase, and as a result, molecules retain their identity. Due to the molecular nature of the solid the band structure and charge carrier transp ort in organic semiconductors behave different ly from inorganic semiconductor s Figure 1 4 indicates the difference between LUMO and HOMO level s of isolated molecule and the valence and conduction band in molecular crystals and amorphous solids. It can be seen that the energy level of conduction and valence band in molecular crystals are slightly shifted compared with the LUMO and HOMO level of the i solated molecule, ind icating lower hole and electron ( charged molecules ) energy in the solid This is because the charged molecules are stabilized by the surrounding polarized molecules in the solid. When moving from crystals to amorphous molecular solids, the highly localized molecular environments lead to a Gaussian distribution of density of state for the electron and hole transport level s
21 Figure 1 4. Energy level diagram of an isolated molecule (left), a molecular crystal (middle) and an amorphous solid (right). The width of the Gaussian density of states in an amorphous solid depends on the level of disorder inside the solid. Based on the above two different types of energy level present in organic solid, the charge carrier transport characteristic s can be divided into two categories: band or hopping transport. The band transport is only observed in highly purified molecular crystal and the charge carrier mobility can usually reach as high as 1 to 10 cm2/Vs.16 18 However, mos t of organic semiconductors exhibit hopping transport ch aracteristics due to the difficulty to obtain molecular crystals. In the hopping transport, the charge carrier has to hop from one molecule to the next. Because of the weak coupling of adjacent molecules, the hopping barrier is sometimes so high that it in duce s very low mobility ranging from 106 to 103 cm2/Vs.19,20 Because of the activated nature of the hopping transport, the mobility is temperature and electric field dependent :20 ( ) exp exp with T as the temperature, F as the electric field, k E as the activation energy for intermolecular hopping and as constant. Energy LUMO HOMO CB VB Isolated molecule Molecular crystal Amorphous solid
22 Excitons The exciton, a bound electron -hole pair, can be created in organic semiconductors through optical excitation or electrical injection of hole s and electron s In the optical excitation, a photon is absorbed by the molecule, followed by excitation of one electron from HOMO level on to LUMO level, leaving one hole in the HOMO level. In electrical injection hole s and electron s are injected into the HOMO and LUMO level from anode and cathode and move towa rds opposite electrode s under an applied electric field. As they meet each other a portion of electrons and holes bound to each other and form excitons due to Coulombic interaction. Based on strength and distance between the bounded electron and hole, excitons in organic semiconductors can fall into two extreme categories: Frenkel exciton and Wannier -Mott exciton. As shown in Figure 1 5 a Frenkel exciton i s strongly bounded electron and hole pair being confined within one molecul e with exciton binding energy of 0.5 2 eV.12,21,22 The Frenkel exciton can diffuse around in the organic semiconductors through hopping between neighbor molecules. However, the mobility of exciton could be very low due to weak intermolecular interaction. In Wannier -Mott exciton, however, electron and hole are weakly bounded (a binding energy of a few meV) and reside on two separated molecules.13 The distance between the electron and hole in Wannier exciton s could be ten times of the intermolecular distance, making them unstable and easily dissoci ated. The third type of exciton, charge -transfter (CT) exciton, has a binding energy between Frenkel and Wannier exciton. In a CT exciton, the electron and hole also reside on separated molecules, but the separation distance is only a few times of the inte rmolecular distance. Thus, the CT exciton is stable and can diffuse around in very similar way to Frenkel exciton.
23 Figure 1 5. Schematic of three different types of excit on in molecular solid: a Frenkel exciton with tightly bounded electron -hole pair re sides on one molecule (left); a Wannier Mott exciton with loosely bounded electron-hole pair, which is delocalized over a distance longer than lattice distance (middle); a charge-transfer exciton, which is in between the previous two types of exciton (right). Energy T ransfer Once excitons are created in the organic semiconductors, the excitons will diffuse around inside organic solid before their energy released in radiative (emit photon) or non-radiative (generate phonon) ways if they are not dissociated into free charges within their lifetime. However, the diffusion of an exci t on is accompanied by an energy tr a nsfer between molecules which is an essential process during the operation of OPV cells and phosphorescent OLEDs. T here are two types of energy transfer mechanisms in organic semiconductors: Fster and Dexter energy transfer.23 In Fster energy transfer, a photon is emitted through the exciton recombination on a donor molecule. The photon is then absorbed by an acceptor molecule and simultaneously creates an exciton. The efficiency of Fster energy is proportional to the overlap between the emission spectru m of the donor molecule and the absorption spectrum of the acceptor molecule. However, instead of two individual emission and absorption process observed conventional, the recombination and excitation during Fster energy transfer occur
24 simultaneously and no real photon is emitted. The dipole -dipole interaction and the distance between the donor and acceptor molecules play an important role in determining the energy transfer rate. Efficient Fster energy transfer can occur over a distance of tens of Angstro m s The Dexter energy transfer, on the other hand, is realized through direct charge exchange between neighboring molecules. An electron or hole on a donor can tr ansfer to an acceptor when there is strong overlap between the orbitals of the two molecules Therefore Dexter energy transfer can only occur in a range of a few Angstrom s Growth and Structure of Organic Thin Film Currently, the active layer of most organic semiconductor devices composed of thin film with thickness ranging from 10 to a few hundred nanometer Therefore, growth of high quality organic thin film with desired structure is the key to achieve certain electrical and optical properties in the devices. Vacuum Thermal Evaporat ion and Spin-C oating There are many ways of growing organic thin film s such as molecular bea m epitaxy organic vapor phase deposition etc.24 T wo of the most widely used methods to grow organic thin film are vacuum thermal deposition and spin -coating. Generally, vacuum thermal evaporation is used to grow small molecular weight based organic thin film s while spin -coating is used to deposit polymer based thin film s Fig ure 1 6 a shows the schematic of vacuum thermal evaporation ( VTE ). In VTE, the substrate and shadow mask are placed above the source boats at a distance of tens of centimeter s During the deposition the boats (usually made of Tungsten or Tantalum) are heated by applying s ufficient electrical current and the organic materials inside sublime under high vacuum. The evaporated organic molecules travel ballistically and finally deposit on the cold substrate above as well as on chamber wall s A quartz crystal monitor (QCM) and a shutter are combined to
25 accurately control the film thickness. During the whole deposition process, the chamber pressure is kept at 107106 Torr Polymer thin film s however, are mostly prepared through solution process es where spin coating prevails. Fig ure 1 6b show s the schematic of the spin -coating method. Here, the solution c ontaining pol ymer materials is dispensed on a spi nning substrate. The centrifugal force spread s the solution evenly over the surface. After the solvents are driven away by baki ng, polymer film s with various thicknesses can be achieved. S pin rate and solution viscosity are two important parameters that determine the film thickness. The VTE method has advantages over solution spin casing, such as high vacuum clean environment, ab ility to fabricate of multilayers as well as doped organic thin films. However, the low -cost spin -coating method is suitable for large area deposition and has much higher material use efficiency. Other methods of organic thin film deposition, such as ink-j et printing, organic vapor phase deposition (OVPD), laser induced thermal imaging (LITI), are also widely use to Figure 1 6. Schematic of vacuum thermal evaporation (VTE) and spincoating. Substrate Organic solution Source boats Shutter QCM Substrate Mask Organic vapor Vacuum chamber ~ 1 10-6Torrb) a)
26 deposit organic thin film with desired structure and morphology .2426 Therefore, careful selecti on and combination of different thin film deposition techniques should provide versatile ways to achieve organic thin films with various structures Amorphous and Crystalline Thin Film s In general, t he structure of organic thin film s can be either amorphous or crystalline, which show different electrical and optical properties. As a matter fact, early studies on organic semiconductors started with applying ve ry high voltage across anthr a cene crystals and light emitting behavior of the materials was observed.27,28 Due to the conjugation natu re for most of the organic semiconductor materials, polycrystalline organic thin film s is commonly observed. Figure 1 7 shows the unit cell of anthracene crystals. The anthracene crystals are monoclinic (a b c, = = 90, 90) with two molecules in each unit cell. Molecule I is located at the position (0,0,0) of the unit cell is transformed into molecule II at (1/2,1/2,0) by a glide operation. Figure 1 7. Schematic structure of unit cell of an anthracene crystal [Ref 29,30]
27 Table 1 2. Crystallographic data on anthracene, teteracene and pentacene crystals [Ref 11] Parameters Anthracene Teteracene Pentacene Crystal structure Monoclinic Triclinic Triclinic Space group P 2 1 /a P 1 P 1 a () 8.56 7.90 7.90 b () 6.04 6.03 6.04 c () 11.16 13.53 16.01 90 .0 100.3 101.9 124.7 113.2 112.6 90 86.3 85.8 z 2 2 2 Density (g/cm 3 ) 1.25 1.29 1.32 z: number of molecules in unit cell Table 1 2 lists the crystallographic data on crystals from polyacenes family. It can be seen that all polyacenes show very similar crystal structure and increased unit cell volume with larger molecular size. The similar crystal structure results in very similar electrical and optical properties of the polyacene molecules. Amorphous thin film s, on the other hand, also play an important role in organic semiconductor devices. Compared with polycrystalline thin film s the amorphous structure may lead to much lower carrier mobility. However, the high ly localized electronic states present in the amorphous film could be good for charge trapping and exciton confinements, which is beneficial for efficient light -emitting devices. For instance, in phosphorescent OLEDs, which adopts a host and dopant mixture as the emitting layer, the amorphous structure of the activ e layer can secure good dispersion of dopants in the host and reduce quenching from aggregates. Whether the organic molecule will form crystalline or amorphous state depends on the molecular structure. Planar molecule s -electron orbitals tend to stack on each
28 other in their aggregates, which forms polycrystal line thin film s .31 33 Polyacenes and p hthalocyanine molecules are two of the representative materials that have a str ong tendency to form polycrystalline thin film. Non -planar molecules with large steric hindrance, such as Alq3 and C60 for instance tend to form amorphous structure due to difficulty to c losely pack with each other. Heterojunctions Hetero jun ctions, which are the junctions between two different semiconductor materials, are of great importance for the optoelectronic properties of organic and inorganic semiconductor s C. W. Tang first report ed the usage of organic heterojunc tions on organic light emitting diodes (OLEDs) and organic photovoltaic (OPV) cells in the 1980s .1,2 Since then the research on understanding the properties of organic hete rojunctions has brought forward significant enhancement on device performance. In OLEDs, heterojunctions play an important role in confining or blocking charge carrier s due to the energy offset between the HOMO and LUMO level s of the two materials. The ch arge confinement effect due to energy offset at the heterojunction can enhance the quantum efficiency of the device significantly. Moreover, in phosphorescent OLEDs, the emitting layer can form a heterojun c tion with materials with high triplet exciton en ergy. This confines the exciton within the emitting layer, which increases quantum efficiency remarkably. H eterojunction s form ed between donor and acceptor materials, on the other hand, is the key component of the OPV cells due to its capability to dissoci ate the photo-generated excitons. In a donor acceptor heterojunction, if the energy offset of between LUMO or HOMO level of the two materials is higher than the exciton binding energy a charge transfer process will take place at the junctions, leaving sep arated holes on the donor and electrons on the acceptor.
29 Organic Photovoltaic Cells One of the most widely studied optoelectronic organic semiconductor devices is the organic photovoltaic cells due to potential to provide low -cost solar energy, which is an important technology to resolve the global warming issue and increasing energy demand. Progress of Organic Photovoltaic Cells The study on organic photovoltaic s has spurred significantly since C. W. Tang first introduced a bilayer heterojunction solar cell composed of copper phthalocyanine (CuPc) and 3,4,9,10perylenetetracarbo xylic bis -benzimidazole (PTCBI) in 1986.1 The i ntroduction of the bulk heterojunction and successful realization of the tandem structure in the last two decades have further boost ed the efficiency of OPV cells.34 36 Figure 1 8 c ompares OPV cells with other type s of solar cells; it can be seen that the current state -of -the art efficiency of organic solar cells is only around 5%,37 which is much lower than other inorganic and dye -s ensitized solar cells. Even though the efficiency of OPV cells is low, the fabrication cost of OPV cells could also be Figure 1 8. Energy conversion e fficiency progress of state of -the art research o n different photovoltaic device s. [Ref. 38]
30 very low. That makes OPV cell s a very competitive candidate for the third generation solar cells, whose average cost for generating 1 watt electrical power falls in the $0.2 ~ 0.5/watt category.39 However, there ar e still plenty of challenges to overcome in the OPV cell, which hinder its commercialization. One big challenge is efficient collection of exciton s generated in the active layer. The exciton diffusion length in organic materials is only tens of nanometer s preventing most of the excitons from reach ing the donor acceptor interface. To overcome this shortcoming, bulk heterojunctions composed of nanostructured donor acceptor phase is highly desired and a number of research groups have demonstrated how such st ructure can be generated For instance, people have shown that bulk heterojunctions composed of nanoscale percolated donor and acceptor phase can be achieved by controlling the phase separation in donor acceptor mixtures.4,6,40 47 This method applies t o both the small molecular weight based OPV cells composed of CuPc/PTCBI and polymer based OPV cells composed of P3HT/PCBM. O ther challenge s in OPV cells include low open circuit voltage, short lifetime, etc.4851 More fundamental understanding of the material properties and physical proces s of the OPV cells is important to further enhance the device performance P rinciple s of Organic Photovoltaic Cells The typical structure of OPV cells, as shown in Figure 1 9 a is compose d of an organic active layer sandwiched between a layer of transparent conductive oxide (TCO) as anode and a metal layer as cathode. The light enters the device from the glass /TCO side and gets absorbed by the organic layer. In addition to working as cathod e, the metal layer can reflect the transmitted light, increasing the optical path and absorption. The active layer of OPV cells usually uses organic donor acceptor heterojunctions to generate free holes and electrons after absorbing
31 Figure 1 9. (a) Typi cal structure of organic photovoltaic (OPV) cells and (b) equivalent circuit for an organic heterojucntion to describe the current -density voltage ( J V ) model of OPV cells. The series and parallel resistances are Rs and Rp, respectively, and Jdark and Jph the dark and photocurrent density, respectively. photons. The holes and electrons are collected by the anode and cathode respectively to produce photocurrent. The photovoltaic behavior of organic PV cells can be characterized in the same way as inorgani c PV by short -circuit current (JSC), open circuit voltage ( VOC), and fill factor (FF) T he equivalent circuit of OPV cells is shown in Figure 1 9b ,52 where the Jdark is the dark current density and Jph as photocurrent density. After taking the series ( Rs) and shunt resistances (Rp) into consideration, the current -density voltage ( J V ) characteristic of OPV cells can be expressed in the form of the Schockley equation :53,54 = + exp ( ) 1 + ( ) where n is the ideality factor of the diode, kB is Boltzmanns constant, T is the temperature, and JS is the reverse saturation current of the diode. The VOC is obtained when J = 0 Thus, = ln ( ) + 1 It can be seen that VOC depends on Jph(V) when the temperature is kept constant. Since Jph(V) is proportional to the incident power density P0, (i.e., Jph(V ) P0) the VO C shows dependence on P0 as VOC ln( P0). This rule a pplies in OPV cells until the product JSRp become s Glass ITO Organic D-A Metal V + J JdarkJphRpRs V
32 comparable with VOC due to Rp P0 1. The maximum VO C is achieved when the quasi -Fermi levels of donor and acceptor are pinned at high photocurrent density. Photocurrent Generation The process of photocurrent generation in OPV cells is quite different from conventional inorganic photovoltaic cells wher e a free electron and hole are generated in the depletion region after absorbing a photon.53 The photocurrent generation process in OPV cells can be genera lly divided into four consecutive steps (see Figure 1 10), which includes generation of an exciton after absorbing a photon (step 1), exciton diffusion (step 2), exciton dissociation at the D -A interface through charge transfer (step 3), and charge collect ion by electrodes (step 4).9,55 57 Therefore, the external quantum efficiency ( EQE) of the OPV cell is determined by the above four steps, viz.: = where A, ED, CT and CC are the efficiency of absorption, exciton diffusion, charge transfer and charge collection respectively. In most of the cases, CT is unity due to the extremely high charge transfer rate. Therefore, the EQE of OPV cells is mainly limited by A, ED, and CC. The absorption of a photon and generation of exciton is accompanied by the electronic transition of organic molecules from the ground state to the excited state. The presence of various vibronic levels in the molecular excited state leads to several narrow bands in the absorption spectrum of organic materials.12 This is different from inorganic materials, where the high density of states results in a continuous absorption spectrum with a clear band -edge absorption cut off. The 5 cm1. Since absorb light is generally on the order of 100 nm.
33 Figure 1 10. Four steps in the photocurrent generation process in OPV cells, namely 1) light absorption and exciton generation, 2) exciton diffusion 3) exciton dissociation and 4) charge collection. Th e efficiencies of above four steps are labeled as A, ED, CC and CT, respectively. The black circles represent electrons, while the open circle s are hol es. Excitons are indicated by a dash line between electrons and holes. The dip of the electron and ho le energy level at the vicinity of the exciton reflects the lower energy state of the exciton compared to an unbounded electron hole pair, which corresponds to a exciton binding energy of 0.5 1 eV. Even though A can be enhanced by simply increasing the film thickness, the dependence of exciton diffus ion and charge collection efficiency on the morphology and thickness has put another restriction on achieving high EQE. Because of the short exciton diffusion length (~ 10 nm) in organic materials,9,56 only excitons generated at the vicinity of the donor acceptor heterojunction can reach the interface and dissociate into free charge carriers Therefore, the exciton diffusion efficiency in bilayer donor acceptor heteroj unction can be very low. In order to circumvent this limitation a bulk heterojunction composed of donor acceptor mixture can be D A ALUMO HOMO ED CT CC 1) 2) 3) 4)
34 used. However, even if the exciton diffusion efficiency c an reach nearly 100% in this structure, the charge collection efficiency can be very low due to poor charge transport. Xue et. al. have studied the electron and hole mobility in CuPc and C60 mixtures, in which they found that high er charge carrier mobility is observed in neat film s while the presence of traps and defects in mixed films lead to lower mobility.58 In order to secure both efficient exciton diffusion and charge collection, a bulk heterojunction with percolated donor acceptor phase shown in Figure 1 11 is preferred. The width of the percolated donor acceptor network phases should be on the order of tens of nanometer s so that all the excitons, regardless of their generation location, can be collected. This struct ure also allows efficient charge transport due to the existence of pure donor and acceptor phase s However, creation of such an ideal nanostructured b ulk heterojunction is challenging. Chapter 3 and Chapter 4 of this thesis mainly focus on addressing this issue through contr olling the phase separation in donor acceptor mixed heterojunctions and controlled growth of one -dimensional organic nanorods using oblique angle deposition. Figure 1 11. Schematic of bulk heterojunction with percolated donor accepter phase. photonElectron transport Hole transportCathode Anode Donor Acceptor
35 Organic Light -Emitting Diodes Organic light -emitting diode s (OLED s ) are very promi sing for next generation lighting and display applications and have attracted a lot of research attention over the past decade. In this part we will mainly dis cuss th e knowledge regarding OLEDs Progress of Organic Light Emitting Diodes The first electroluminescence (EL) effect from organic materials dates back to 1963, when Pope et. al. m thick anthra cene single crystal .27 It was not until 1987 when C. W. Tang from Kodak report ed a n OLED with a heterojunction composed of Alq3 as the emitting layer and NPD as the hole transport layer that low driving vo ltage and high EL efficiency were achieved.2 This has led to the prosperous development of OLED s with a lot of new observation and technologies Among the most important findings was the work by Frrest et. al. in which OLED phosphorescence was first observed.5 Phosphorescence OLEDs are able to break the theoretical limit of 25 % on internal quantum efficiency of fluorescent OLED s Even though OLED s have significant advantages over conventional displays and lighting technologies, there are still some challenges that must be overcome before large scale commercialization of OLED s For instance, the current ef ficiency and power efficiency are still l ow compared with florescent lighting. In order to improve the current efficiency, better confinement of charge carrier s and excitons in the emitting layer using charge transport layer s with appropriate LUMO/HOMO lev el is suggested. High power efficiency, on the other hand, requires efficient electron and hole injection layer to reduce driving voltage Therefore, new materials and device architectures based on better understanding of molecular properties and device physics are essential for realizing high efficiency OLED s Furthermore, there exist still more challenges such as device lifetime, pack ag ing, etc, before large scale commercial development.
36 Principles of Organic Light -Emitting Diodes The photon energy emitt ed by organic materials strongly depends on the energy difference between the ground state and excited state of the molecule. Therefore, various color emission from organic materials can be tuned by changing the bandgap o f molecules. Thompson, et. al. show that the emission spectra of platinum -based organic compounds, which cover nearly the whole visible region, can be varied through replacing the functional groups.59 This indicates the versatility in using an organic chromophore as the emitter. The operation principle of OLED s is illustrated in Figure 1 12, where a simple three layer device composed of electron and hole transporting layers (ETL and HTL) and emitting layer (EML) is shown Under an applied bias holes and electrons are injected into the HTL and ETL from the anode and cathode, respectively. Holes will transport through the HOMO level of the HTL while electron s through the LUMO level of the ETL. If injection of charge carrier s is balanced most of the electrons and holes will form excitons inside the EML and emit light. In order to further improve OLED performance, additional layers such as charge injection layer s Figure 1 12. Principle of operation of an organic light -emitting diode (OLED ), where the injected charges are transport ed through the HTL and ETL. The holes and electrons form emissive single or triplet exciton s within EML
37 and blocking layer s could also be used. The external quantum efficiency (EQE) of an OLED which is define d as the ratio of photons emitted to charges flowing in the circuit can be expressed as: = w here created to charges injected, rst is the fraction of excitons which can radiatively decay, q is the efficiency of radiative decay of singlet excitons, and EXT is the optical outcoupling efficiency. F or fluorescent emission, rst is limited to 25%, while rst can potentially reach 100% for phosphorescent emission. Phosphorescent Organic Light -Emitting Diodes Light emission in OLEDs originates from radiative decay of excited molecules. S ince the two electrons in excited molecule occupy two different MOs i.e. LUMO and HOMO, respectively, the net spin value of the excited molecule can be either S = 0 or 1 depending on the spin angular momentum of the two electrons. An e xcited state with net spin value S = 0 is called a singlet excited state, while S = 1 corresponds to a triplet excited state. Any radiative decay process of a singlet excited state is called fluorescence, whereas radiative decay of a triplet state is calle d phosphorescence. Figure 1 13 shows the Jablonski diagram of the fluorescence and phosphorescence process In general, the rate of fluorescent decay is much faster than phosphorescent decay, in which the spin angular momentum needs to change in order to g enerate a singlet ground state. Typical lifetimes of the singlet excited state are in the range of 1 10 ns, whereas the lifetimes of triplet excited state s can be in the millisecond range for pure aromatic hydrocarbons. Therefore, m ost of the light -emitting process in organic molecule originates from fluore scence rather than phosphorescence. Since the number of singlet to triplet excited states
38 Figure 1 13. Jablonski diagram for the fluorescence and phosphorescence of organic molecules with S0 as the ground state and S1, T1 as the singlet and triplet excited state, respectively. The rate of fluorescence, phosphorescence and intersystem crossing is kF, kP and kISC, respectively. created during electrical injection is 1:3, any OLED whose emission completely relies on fluorescence can have a maximum of 25 % internal quantum efficiency. However, Forrest et. al. found that by introducing a heavy metal atom onto the organic molecule, th e rate of intersystem crossing between singlet and triplet states is significantly enhanced, while the lifetime of the triplet excited state is also shortened. All these lead to efficient phosphorescent OLEDs (PHOLEDs) with a maximum100 % internal quantum efficiency. PHOLED s are usually composed of a guest -host t ype emitting layer with phosphorescent molecules doped into a host matrix. The host matrix plays a couple important roles to help obtain efficient phosphorescent emission from dopants. During operation electrons and holes are injected into the host materi al on which the excitons are formed and transferred to phosphorescent dopants through Dexter or Foster energy transfer. In some cases, the excitons are directly form ed on the dopants through direct trapping. Due to a cascade energy transfer process in th e EML of PHOLED, the host molecules should have higher exciton energy than phosphorescent dopants in order to secure efficient energy transfer. However, excitons may recombine non-radiatively due to transfer ring to the S1S0 T1 Phosphorescence Fluorescence ISC kISCkPkF
39 adjacent HTL or ETL layer through For ster and Dexter process if the HTL or ETL mater ials have lower exciton energy. This will significantly redu ce quantum efficiency of the device. Therefore, HTL s and ETL s will high exciton energy are desired to prevent such exciton loss. Overview of This T hesis The works present ed in this thesis can be divided into two major parts with focus es on organic photovoltaic (OPV) cells and organic light -emitting diodes (OLEDs), respectively First accurate measurement of efficiencies of OPV cells and OLEDs are in troduced in Chapter 2. Here, we describe the method to measure the external quantum efficiency of the OPV cells and to use the spectrum mismatch factor to correct ly characterize an OPV cell efficiency. At the same time, the method of measuring OLED efficie ncy using a Si photodiode and a luminance meter is presented by assuming a Lambertian source. Next Chapter 3 and 4 focus on improving the power conversion efficiency of the OPV cells through creation of nanostructured bulk heterojunctions. Specifically Chapter 3 investigate s the phase separation process in donor acceptor molecular mixture s composed of pentacene and C60. Here, a molecular dynamic (MD) simulation is used to monitor the nanoscale structural evolution in the pentacene:C60 mixture computati onally, while the structure and morphology of the mixed films are studied using X ray diffraction (XRD), scanning electron microscopy (SEM) and atomic force microscopy (AFM). The results reveal that the performance of the OPV cells relates to the degree of phase separation inside the donor acceptor mixture. Chapter 4 demonstrates a facile way to fabricate onedimensional organic nanorods using oblique angle deposition. The morphology of the nanorods grown under different conditions is investigated By changing the substrate rotation mode, CuPc nanorods with different diameter and packing density can be obtained. The power conversion efficiency of a bulk hererojunction
40 utilized the organic nanorods show two times enhancement over the planar heterojunction device Chapter 5 and 6 of this thesis cover studies on organic light -emitting diodes. In Chapter 5, new ETL and HTL materials are applied to blue PHOLED s The charge carrier and ex citon confinement properties from different charge transport layers a re compared and their effect s on im proving the PHOLED efficiency are summarized. Moreover, p and n -type doping of the charge injection layer are investigated to further reduce the overall driving voltage of PHOLED s Chapter 6 discusses a novel electron in jection layer compris ing ZnO nanoparticles applied to polymer light emitting diode s (PLED s ). It is found that the turn -on voltage of the PLED is significantly reduced due to an electron injection process through interfacial Auger recombination. F ina lly, Ch apter 7 of the thesis consists of conclusions and future outlook. Conclusions are drawn b ased on the results from studies on OPV cells and OLEDs and the limitation s that remain for these two types of optoelectronic device are discussed. Future research wo rks that could further enhance the performance of OPV cells and OLEDs are also suggested.
41 CHAPTER 2 MEASUREMENT OF ORGAN IC PHOTOVOLTAIC CELL S AND ORGANIC LIGHT EMITTING DIODES Introduction Organic photovoltaic (OPV) cells and organic light -emitting diodes (OLEDs) have attracted substantial research attention in the past a few years due to their poten tial to meet future needs for clean and efficient energy devices. However, since various devi ce area s and pattern s are us ed, the measurement procedures for OPV cells and OLED are inconsistent between different research groups. Therefore, it is critical to accurately measure the efficiency of these two types of device s so that direct comparison s ca n be made betw een different groups .60,61 In this chapter, setups for efficiency measurement s of OPV cells and OLED s are presented. We also provide detail s of our method s to calibrate the measurement systems so that accurate measur ements can be achiev e d The first section of this chapter will focus on the measurement of OP V cells, where the spectrum mismatch factor is considered based on our measurement system. The measurement of OLED efficiency using a planar geometry by assuming a Lambertian emission pattern is given in the second section Finally, the conclusio ns are pro vided in the last section. Measurement of Organic Photovoltaic Cells In the past two decades, the power conversion efficiency (PCE) of OPV cells has gradually progressed from less than 1 % to nearly 6 % today. The highest PCE reported so far for the polymer bulk heterojunction (BHJ) OPV cells is around 6%.8 For small -molecular -base d OPV cells, PCE up to 5.7% has been reported for device based on tandem cells using copper phthalocyanine (CuPc).34 Since the PCE is now approachin g the threshold for commercial applications of OPV cells, it is of great importance to accurately determine the efficiency values so that fair comparison s of results from different research groups can be achieved.
42 Power Conversion Efficiency Measurement w ith Spectrum Mismatch Factor Generally, the efficiency measurement of OPV cells concerns two parts of information, which include s the power conversion efficiency (PCE) and the external quantum efficiency (EQE) as function of the wavelength. To obtain an a ccurate PCE value, IV characteristics of OPV cells under standard reporting conditions (SRC) suggested by the American Society for Testing and Materials (ASTM) are required And t he SRC recommended by ASTM for rating the performance of terrestrial PV cell s include: 1000 W/m2 irradiance, AM 1.5 (AM: air mass) global reference spectrum, and 25 C cell temperature.62,63 In reality, a solar simulator is widely used to conveniently generate a simulated AM 1.5G spectrum with various irradiance. Figure 2 1 shows the diagram of I V characteri stics measurement under simulated AM 1.5G illumination for OPV cells. Here, a n Oriel solar simulator equipped with a Xenon lamp with spectr al coverage from 250 nm to 1100 nm is used as the light source. In front of the lamp an AM 1.5G filter is used to adjust the raw spectrum into a simulated AM 1.5G solar spectrum. The beam then goes through a set of irises and neutral density (ND) filters to achieve a uniform light spot with desired irradiance At the end, the OPV cells or a silicon reference cell are mounted on a three dimensional manipulator to give a relatively fixed position. With certain irradiance, the I V Figure 2 1. Diagram of power conversion efficiency (PCE) measurement for OPV cells. Solar simulator Xenon lamp AM 1.5 filter Iris Neutral density filter Iris PV cells or Si reference cell Semiconductor analyzer
43 measurements of OPV cells are taken by a semiconductor analyzer. Therefore, the PCE of the OPV cell is given as:63 PCE = 100 = 100 here, Pmax is the measured maximum power output reading from the I V curve of test cell, A is the device active area, Etot is the incident irradiance ISC is the short -circuit current, VOC is the open -circuit voltage, and FF is the fill factor. In order to obtain the real PCE value of the solar cell, it is crucial to accurately secure the right Etot based on the reference AM 1.5G spectrum provided by ASTM standard G159.62 The Etot incident on the PV cell is typically measured with a calibrated Si reference cell, which is suggested by ASTM standard E948.63 However, f or I V measurements of OPV cells under the simulated AM 1.5G spectrum, a spectral error could exist in the measured I V which leads to inaccurate VOC, FF and especially, ISC. Such spectral e rror could originate from the spectrum difference between the simulated spectrum and the reference spectrum, and the spectral respons e difference between the Si reference cell and our OPV cell.64 To correct this spectral error on ISC, the spectrum mismatch factor ( M ) is introduced by ASTM standard E973, which is expressed as:64 = ( ) ( ) 21 ( ) ( ) 21 ( ) ( ) 21 ( ) ( ) 21 = , where ( ) is the reference spectral irradiance, ( ) is the simulated spectral irradiance, ( ) is the spectral responsivity of the reference cell, and ( ) is the spectral responsivity of the test cell. All of them are as a function of the w limits 1 and 2 should encompass the spectral responsivity limits of the OPV cells and Si
44 reference cell. It can be seen that the real ISC of test cell under the reference spectrum can be expressed as: = therefore, by adjusting the simulator spectral irradiance so that is equal to can be directly obtained by dividing the measured by the mismatch factor M In the real measurement setup, it is more con venient to adjust the simulator spectrum and responsivity of the reference cell so that M is close to unity. In this case, the measured closely equals Here, we showed the calculation value of the spectrum mismatch factor M for a bilayer OPV cell of ITO/CuPc (20 nm)/C60 (40 nm)/ bathocuporine (BCP ) (8 nm)/Al with a Si reference cell. Figure 2 2 a compares the standard AM 1.5 G reference spectrum and the simulated AM 1.5 G spectrum from Oriel solar simulator. And Figure 2 2 b shows the responsiv ity of the silicon reference cell with and without th e KG1 color filter It can be seen that M = 0.8 0 with the unfiltered Si reference cell, which originates from the large difference in the spectral responsivity between the OPV cells and the reference cel ls. The unfiltered Si reference cell has a spectral response over a range of 400 1100 nm, wher e as the CuPc/C60 cell only shows response from 300 to 800nm (see Figure 2 4) Therefore, in order to obtain M close to unity, it is important to choose a reference cell whose s pectral response closely matching that of the OPV cell. Therefore a KG1 color filter with a cut off at around 800 nm is used to correct the responsivity of the Si reference ce ll. It can be seen from Figure 2 2 b, the KG1 filtered Si referen ce cell show s much narrower spectral response of 400 800nm, which is very close to that of the CuPc/C60 cell. This leads to an almost unity mismatch factor of M = 0.99. Clearly, the KG1 -filtered Si reference cell is more suitable for use as a reference c ell in measuring the spectral irradiance from the solar simulator.
45 Figure 2 2. Calculation of spectrum mismatch factor with specific solar spectrum and reference cell (a) Comparison of reference AM 1.5G solar spectrum and simulated AM 1.5 spectrum. (b) Responsivity of Si reference cell with and without KG 1 filter. Inset: the transmission of KG 1 filter. External Quantum Efficiency Measurement By using a KG1 -filtered Si reference cell, a spectrum mismatch factor close to unity can be obtained. The PCE of the OPV cell under standard reference spectrum can be directly inferred from the M value for pa rticular simulated spectrum and an OPV cell/reference cell combination. It is also important to obtain the external quantum efficiency (EQE) of the OPV cell so that 1) the spectral responsivity of the OPV cell can be calculated from the EQE curve by ( ) = 300 400 500 600 700 800 900 1000 1100 0 1 2 3 4 5 Reference AM 1.5 solar spectrum Simulated AM 1.5 solar spectrumRelative spectral distribution (nm) 300 400 500 600 700 800 900 1000 1100 0.0 0.2 0.4 0.6 0.8 M = 0.99 with KG-1 without KG-1Responsivity (A/W) (nm) M = 0.80 500 1000 0 50 100 Transmission (%) (nm)
46 ( ) and 2) the short circuit current under reference spectral irradiance can be calculated to compare with to confirm the solar simulator system is correctly set up. Moreover, the EQE curve provides useful information about the physical process es inside the OPV cell, such as exciton diffusion, charge collection, etc. Figure 2 3 shows our setup for EQE mea surement based on the guidelines from ASTM E1021,65 where a monochromatic, chopped beam of light is incident at normal direction onto the OPV cell. At the same time, a continuous white bias light is used to illuminate the entire device at an irrad iance approximately identical to normal operating conditions intended for the OPV cell (1000 W/m2 irradiance in most case s ). T he spectral dependence of the AC (chopped) component of the short circuit current is picked up by the lock in amplifier as the wav elength of the monochromatic light varies over the response band of the cell. The power of the incident monochr omatic light is determined by Newport 818-UV Si detector with known spectral responsivity. In the current setup, two condensing len ses combined w ith slits on the monochromator are used to adjust the beam size to an appropriate lev el so that all Figure 2 3 Diagram of external quantum efficiency (EQE) measurement for OPV cells. MonochromatorChopper controller Lock -in amplifier Current amplifier Condensed Lens Condensed Lens Chopper PV device or Si detector Xenon lampsource input input output White light bias
47 monochromatic light is incident on the devices active area. With th is information, the ( ) of the OPV cell can be obtained through: ( ) = ( ) ( ) / ( ) here, ( ) and ( ) are the short circuit current s generated b y the monochromatic beam for on OPV cell or detector respectively ( ) is the spectral responsivity of the detector. With EQE information, the short circuit current under reference spectral irradiance can be calculated through the equation: = ( )21( ) Fi gure 2 4 shows the measure d EQE curve of a bilayer CuPc/C60 OPV cell along with the current density-voltage ( J V ) characteristics of the same device at 1000 W/cm2 simulated AM Figure 2 4. External quantum efficiency as a function of the wavelength for I TO/CuPc (20 nm)/C60 (40 nm)/BCP (8 nm)/Al OPV cell. Inset: current densityvoltage ( J V ) characteristics of same device under calibrated AM 1.5 G 1000 W/cm2 irradiance. 300 400 500 600 700 800 0 5 10 15 20 25 30 External Quantum Efficiency (%) (nm) -1.0 -0.5 0.0 0.5 1.0 -6 -4 -2 0 2 J ( mA/cm2)V (V) JSC (direct)= 3.70 0.09 mA /cm2 JSC (cal)= 3.46 0.03 mA /cm2
48 1.5G irradiance measured through the KG1 filter Si reference cell mentioned above. It can be seen that the direct measurement gives a short circuit current density of JSC (direct) = 3.70 0.09 mA/cm2 based on 0.04 cm2 device area, whereas the calculated JSC (cal) = 3.46 0.03 mA/cm2 based on EQE curve. The difference between the two values is around 5%, which suggests good consistency between two measurement systems. Measurement of Organic Light Emitting Diodes OLED technology has made significant advance s in th e last decade, and different commercial products have recently been available on the market. However, the efficiency measurement setup adopted by different research groups varies significantly, which makes valid comparison between different laboratories di fficult. Here, we present our simple yet accurate way to measure the efficiency of OLEDs. Measurement Setup f or Organic Light -Emitting Diodes For OLEDs targeted for display and lighting application s it is of key importance to accurately measure the lumi nance from the OLEDs so that accurate value s for the different types of efficacy can be realized. One important assumption we make in our OLED measuremnts is that the emission p attern of the devices resembles that from an ideal Lambertian emitter.66 Basically, a Lambertian emitter is isotropic, emitting with equal luminance into any solid angle within the forward viewing hemisphere. Nevertheless, such assumption becomes violated when microcavity out -coupling or other structure s are applied to the OLEDs, which can significantly change the emission pattern from Lambertian type .67 69 The luminance of OLED s with Lambertian emission pattern s can be easily measured using a commercial luminance meter. With the recorded luminance value L (cd/m2), the current efficiency of the device L (cd/A) can be simp ly calculated by dividing L by the drive cur rent density of the device JOLED. The power efficiency P and external quantum efficiency EQE can
49 also be calculated based on L with knowledge of the driving voltage, emission spectrum and photopic response of the human eye. In practical measurement s th e current -density -luminance voltage ( J L V ) characteristics of the OLEDs are desired, which is obtained by sweep ing the OLEDs across a certain voltage range and record ing JOLED and L at each corresponding voltage. Using a luminance meter to measure the luminance over the whole scan range can take substantial amount of time due to integration time needed for ea ch measurement. One accurate yet con venient way for measuring the l uminance is shown in Figure 2 5 a, which employs a photodetector to collect the emitting light. Here, only a fraction of light is being collected by Figure 2 5. Luminance measurement of OLEDs (a) Diagram of the luminance measurement setup for organic light -emitting diodes (OLEDs) with Si photodetector collected fraction of forward emitting light. (b) Linear dependence of the luminance (measured by luminance meter) and photocurrent (measured by Si photodetector) using the above measurement diagram. Semiconductor analyzer Si detector Glass OLED 60 70 80 90 100 110 120 130 140 150 25 30 35 40 45 50 55 L ( cd/m2)Idet (nA) a) b)
50 the photodetector and the corresponding luminance is related to the photocurrent through = The conversion factor is determined by as well as the spectral responsivity of the photodetector L and over a certain range of luminance. As shown in Figure 2 5 b, exhibit s linear dependence vs. luminance over the range of 25 55 cd/m2 for a deep blue phosphorescent OLED, with = 0.391 cd/m2A from curve fitting. Generally, is a parameter that only depends on the geometric arrangement of the photodector and the OLEDs by assuming the emission pattern does not change at different driving voltage nor with different OLED emission spectra Therefore, by fixing the relative position of the photodetector to the OLEDs, should be a constant. With knowledge of the emission spect rum of an OLED can be expressed as: = 0 ( ) ( ) ( ) ( ) where 0 is the peak value of the photopic response curve, i.e., 683 lm/W,70,71 ( ) is the normalized photopic response curve and A is the device area. Since is only determined by the geometry of the measurement setup, the L of any OLEDs with such geometry can be obtained by simply measuring the photocurrent and the corresponding emi ssion spectrum: = 0 ( ) ( ) ( ) ( ) while P and EQE can be related to L through the equations: = = ( ) ( ) ( )
5 1 One should be cautious since all of the above equations are spectrum sensitive. Therefore, if the emission spectrum changes with respect to the driving voltage it is required that the right spectrum be plug ged in to get the corresponding efficiency at certain driving condition s The efficiency measur ed using the current method only account s for the forward direction, so any photons emitted from the substrat e edge and backward direction are missed. The forward efficiency is widely used in display application s .60 The total external quantum efficiency which account s for the photons emitted from the device in all directions, can be measured by placing the OLEDs into an integrating sphere.60 Conclusions In this chapter, we provide d accurate ways to measure the efficiency of OPV cells and OLEDs. F or OPV cells, it is important to select a reference cell with similar spectral response compared to the OPV cells being test ed. By using a KG1 -filter ed Si reference cell, the spectr al mismatch factor reaches 0.99, which suggests the efficiency measured by such a solar simulator system accurately reflects the performance of the device under reference spectral irradiance. The short circuit current calculated by integration of the product of external quantum efficiency and reference spectral irradiance is also a reference for the measured short circuit current, and the difference should be le ss than 10 % to indicate good agreement between the two setup s. In the measurement of OLEDs efficiency, a simple method involving a calibrated photodetector is su ggested. By fixing the geometry-determined parameter and measuring the emission spectrum, the current, power and external quantum efficiencies in the forward direction can be readily obtained through existing equation s by assuming a Lambert ian emission pattern of. All the OPV and OLED efficiencies provided in this thesis are obtained through the methods outlin ed in this chapter, which should serve as a good reference base for understanding the physical process es in the devices.
52 CHAPTER 3 PHASE SEPARATION IN MOLECULAR DONOR -ACCEPTOR MIXTURE Introduction O rganic photovoltaic (OPV) cells commonly employ a donor acceptor (D -A) heterojunction structure to provide the driving force for the dis sociation of photogenerated excitons due to an approp riate energy level offset at the D -A interface. However, the excit on diffusion length for most organic materials is on the order of a few nanometers.9,72 74 Thus, only exciton s generated in the vicinity of the D -A interface can be collected, which limits the efficienc y of OPV cells that employ a bilayer heterojunction structure. Therefore, it is crucial to properly design the morphological structure of the active layer to maximize the amount of the photogenerated excitons reaching the D -A interface and contribute to th e photovoltaic process. It has been shown that a bulk heterojunction (BHJ) com posed of a mixture of donor and acceptor molecules with percolated D -A phase c an enhance the exciton diffusion efficiency and hence the OPV efficiency.3,4,6,40 42,45,7578 In an optimized BHJ structure, nanoscale threedimensional percolated networks of the donor and acceptor phase lead to both good exciton diffusion and charge collection. Various approaches have been reported to realize the BHJ structure. Two of the most common methods include deposition of a thin film by spin -casting a blended solution with a polymer donor and small molecular acceptors3,4,41,43 or by co -evaporation of two molecular species in vacuum.6,45 It has been found that the morphology of the D -A networks strongly depends on the donor to acceptor mixing ratio and various processing conditions during film deposition as well as the mo lecular properties.6,47,58,79 The film morphology may even be further optimized by employing thermal or solvent post annealing processes.80,81 One important parameter for determining the film morphology, the degree of phase separation in a D -A mixture strongly depends on both the intrinsic molecular properties
53 Figure 3 1. Different morphological structure in molecular donor accetpor mixtures and the charge transport within. (a) complete mixing of two molecular species with discontinued conducting path; (b) molecular level conducting paths are formed with weak phase separation in the mixture; (c) ideal bulk heterojunction with nanoscale phase separation in D -A mixtures. [Ref 45] and the thermodynamic and kinetic processes involved during the mixing of the two molecular species.6,58 Fig ure 3 1 illustrates three different morphological structure of the molecular D -A mixtures. It can be seen that a very weak phase separation results in a high degree of intermixing between molecular species (Figure 3 1 a). Although efficient exciton diffusion is obtaine d in this case, the discontinuous conducting path ways for electrons and holes result in poor charge collection. A medium level of phase separation could produce morphology like Figure 3 1b, where continuous conducting path based ways on molecular level interconnections are formed. Ideally, the degree of phase sep aration should be neither too weak nor too strong so that a pure donor and acceptor phase with structural dimension of tens of nanometers can be obtained as shown in Figure 3 1c. The nanoscale phase separation could secure both good exciton diffusion and c harge collection. Although there have been intensive research efforts to correlate the degree of phase separation with experimental observations, it is still challenging to understand the phase Hole Electron Hole Electron Continuing transport Terminated transport Continuing transport Terminated transport Donor Acceptor Donor Acceptor
54 separation process and how it determines the resulting morphol ogy within the mixture, such as the domain size and degree of percolation. In this chapter, the morphological behavior between donor and acceptor molecules at the atomic scale is explored using computational methods. Specifically, the results obtained from molecular dynamics simulations indicate how the molecules behave with a given concentration and how their initial neighbors affect the final outcome. At the same time, experimental studies of the structure and morphology of the donor acceptor mixed films are conducted. The combined computational and experimental results give a profile of the phase separation behavior between pentacene and C60. It is therefore the purpose of this work to combine experimental and computational methods to understand the phase separation process in organic D A mixtures. The pentacene:C60 molecular system was chosen for this study due to the simplicity in dealing with small hydrocarbon molecules computationally. Efficient OPV devices based on bilayer pentacene:C60 heterojunct ions have also been demonstrated previously.82,83 Here, we use classical molecular dynamics (MD) simulations84,85 to examine morphology evolution in various pentacene:C60 mixtures. A quantitative way to describe the phase separation between pentacene and C60 in equilibrated mixtures w as suggested. Experimentally, X ray diffraction (XRD), scanning electron microscopy (SEM), and atomic force microscopy (AFM) were employed to characterize the crystal structure and surface morphology of mixed films prepared with various processing parameters. Photovolta ic (PV) devices using pentacene:C60 BHJs were fabricated based on the information obtained from simulation and morphology characterization. We found that devices constructed with D -A mixed films in which nanoscale phase separation was successfully controll ed exhibit the highest efficiency amongst all BHJ devices considered in the work.
55 This chapter is organized as follows. A general introduction of the thermodynamics and kinetics involved in phase separation process is given first, followed by the molecular dynamic simulation and discusses the computational results of structural evolution of pentacene:C60 mixtures. With the simulation results the experimental study of phase separation degree in pentace ne:C60 mixed films are conducted followed by the PV per formance comparison of devices based different pentacene:C60 mixed film. The last two sections summarize the co ntents in this chapter and draw a conclusion based on the results. Thermodynamics and Kinetics o f Phase Separation Behavior of Regular Soluti on In general, whether phase separation can happen in a mixture is determined by the energy level of each thermodynamic state adopted by the mixture. The lowest thermodynamic energy state corresponds to the most stable structure, which the mixture will fin ally adopt if there is no kinetic limitation during the structure evolution. One of the most widely used parameters to describe the energy change during the mixing and phase separation of various components is the Gibbs free energy. The Gibbs free energy c hange ( ) due to the mixing of different components can be expressed as:86 = where and are the enthalpy and entropy change due to mixing, respectively. And T is the temperature. In a regular solution model consisting of only two components, the enthalpy and entropy change of the mixing can be obtained statis tically by considering the interaction energy between some species as well as between different species. For instance, in a binary system composed of donor and acceptor molecule with concentration of XD amd XA, the and can be expressed as:
56 = 0 1 2 ( + ) = 0 ( ln + ln ) where 0 is Avogadros constant, k is the Boltzmanns constant, z is coordinates, and EDA, EAA and EDD are the energies of donor acceptor interaction s acceptor acceptor interaction and donor donor int eraction s respectively. Figure 3 2 shows the plots of as a function of mixing ratio, where three different scenarios can be recognized in term s of phase separation level. As indicated by the blue curve, is negative with onl y one global minimum when 2 > + suggesting complete intermixing of the mixture and no phase separation will be observed. On the other hand, if 2 < + and the temperature is not very high, two Figure 3 2. The variation, with composition, of the molar Gibbs free energy of the formation of a binary regular solution. The three scenarios suggest the thermodynamic stable states of the solution with (green curve) and without (blue and red curve) phase separation in the solution. 0.0 0.5 1.0 0 EDA < EAA+EDD small T EDA < EAA+EDD large T GM XA EDA > EAA+EDD
57 saddle points correspond ing to two different stable phases with different mixing ratios can be observed along the plot (see green curve). In this case, if a mixtu re with a 1:1 mixing ratio of donor and acceptor is created, the system is unstable due to a relatively high Gibbs free energy. Therefore, the mixture tends to have phase separ ation so that two phases, one donor -rich and the other acceptor rich, will form in the mixture. Although phase separation could happen under this condition, increasing the temperature will cause the two species to mix with each other again, which results in no phase separation (see red curve). Therefore, a complete and stable mixing o f donor and acceptor molecules happens when the donor acceptor interaction is stronger than the donor donor and acceptor acceptor interaction s ( 2 > + ) To realize phase separation in a D -A mixture, the donor -donor and acceptor acceptor inte raction s should be stronger than the donor acceptor interaction (2 < + ) and at the same time the system temperature should not be very high. In the current pentacene:C60 mixture, the pentacene pentacene interaction is expected to be strong er than the pentacene:C60 Therefore, phase separation is expected to be seen in the pentacene:C60 mixtures. F urther computational study to calculate the interaction energy between the molecular species involved will provide a more profound understanding of the phase separation in the pentacene:C60 mixtures. Spinodal Decomposition and Nucleation The Gibbs free energy curve of the mixture provides an insight of whether a phase separated structure is thermodynamic favorable state of the mixture. Nevertheless, how the molecules evolve in the mixture during the phase separation process is of great importance in determining the microstructure in the final phase separated mixture. Basically, t here are two
58 Figure 3 3. Phase separation through spinodal decomposition (a) The spinodal decomposition (b) spatial composition variation during the spinodal decomposition with system time t3 > t2 > t1. distinct mechanisms for phase separation; namely, spinodal decomposition and nucleation plus growth. Although the final thermodyanic state could be the same, different routes took by the mixture during the phase separation could result in different structure and morphology. The spinodal decompos ition happens at a part of the Gibbs free energy curve where the curvature is negative, which means 2 2< 0 Figure 3 3 a shows a part of the Gibbs free energy curve with negative curvature. It can be seen that a small deviation away from the init ial concentration point can lower the free energy of the system i.e. (unmixing) < 0 Therefore, one part of the system gets more concentrated of one species at the expense of another. The spatial composition change of the system with respect to the time is illustrated in Fig ure 3 3 b. It can be noticed that one part of the system become more concentrated with one species, while another part gets less concentrated as time proceed s Finally, two different phases with concentrations corresponding to the saddle points on the free energy curve are reached. The nucleation and growth mechanism is quite different from spinodal decomposition. As shown in Figure 3 4 a, the nucleation and growth takes place when a part of the free energy curve has positive curvature, i.e., 2 2> 0 As a result, a small fluctuation of the concentration will increase the system energy instead of lower the system energy as spinodal
59 Figure 3 4. Phase separation through nucleation (a) The nucleation process, (b) spatial compos ition variation during the nucleation process. decomposition did, i.e. (unmixing) > 0 Therefore, the phase separation in this case requires large composition fluctuations to decrease energy, which means the new phase must start with a composition far from the parent phase. This suggests a nucleation process of a new phase within an existing phase, followed by cryst al growth of the new phase. Figure 3 4 b illustrates the spatial composition change during the nucleation process, where a tiny part of the s ystem become highly concentrated with concentration of other part nearly unchanged. Generally, nucleation is a phase transition that is large in degree in terms of composition change but small in extent, whereas spinodal decomposition is small in degree but large in extent. Therefore, in order to achieve morphology in the D -A mixture that is favorable for OPV application, phase separation through nucleation and growth is highly desired due to the capability to generate smaller domains in the mixture. Howev er, whether phase separation will take the route of spinodal decomposition or nucleation and growth depends on the Gibbs free energy curve of the mixture, which is again determined by the temperature and interaction energy between the molecular species. Structural a nd Morphological Control of Pentacene:C60 Mixture s Effect o f Mixing Ratio on Structure and Morphology To understand the phase separation in the pentacene:C60 mixtures, t he structure and morphology of different pentacene:C60 mixed films are firs t investigated using X ray diffraction
60 and atomic force microscope. Pentacene:C60 films with various mixing ration were deposited via vacuum thermal evaporation in a custom made vacuum chamber with a background pressure of ~2 107 T orr. For structural an d morphological studies, all films were deposited on Si (100) substrate s, which is obtained by immersion in diluted hydrofluoric acid (HF:H2O = 1:50) to etch away the native oxide layer. To investigate the structure of pentacene:C60 mixtures X ray diffraction (XRD) is first used to characterize films with d if ferent mixing ratios. Figure 3 5 shows diffraction patters of 30 nm thick films of pentacene, C60 and pentacene:C60 deposited on an Si substrate. The XRD patters were obtained using a Phi lips Xpert MRD diffracto meter in the geometry with a 45 film shows diffraction peaks from the thin film (00 n ) and bulk (00n ) (n = 1,2,3) phases of pentacene. And the peaks for the Figure 3 5 X -ray diffraction patterns for 30 nm thick pentacene and pentacene:C60 mixed films with various mixing ratios, taken in the geometry. 4 6 8 10 12 14 16 18 20 (003) ( 003')( 002')(002) (001)1:2 1:1 2:1 Pentacene:C60 = 3:1 Intensity (a.u.)2 (degree) Pentacene ( 001' )
61 thin film phase are more prominent due to the small thickness of the film.16,87 The presence of higher order diffraction peaks indicate the good crystalline order in the neat pentacene film. The pentacene (001 ) and (001) peaks, located at 2 = 5.7 and 6.1, respectively, correspond to the interplanar s pacing ( d ) of 15.5 and 14.5 which agrees well with the values reported in the literature.87,88 When pentacene and C60 are mixed with a weight ratio of 3:1, the less prominent pentacene (001) peak disappears, and the intensity for the (001 ) peak reduces significa ntly. Additionally, all higher -order diffraction peaks, except the (002 ) peak, cannot be discerned anymore, an indication of the increased disorder among pentacene stacks. As the concentration of C60 continues increasing in the mixture, the (001 ) peak in tensity further decreases and it eventually disappears in the XRD pattern for the 1:2 mixed film. These data suggest that the pentacene aggregat es in the form of crystalline domains exist in pentacene rich films, although the disorder increases with the C60 concentration. It is noted that the general trend observed in the XRD patterns here is similar to that for CuPc:C60 mixtures.58 However, the CuPc agg regation can only be found when the CuPc concentration is high enough, i.e., a CuPc to C60 mixing ratio of 6:1 or higher. All these suggest that pentacene molecules have a stronger tendency to aggregate than CuPc. To gain an insight of the nanostructure i nside the phase separated pentacene:C60 mixtures AFM is used to investigate 3 nm thick neat or mixed films grown on Si. The AFM images are obtained using a Veeco Dimension 3100 AFM working in tapping mode with Si probe tips. As shown in Figure 3 6 the la yer plus -island growth mode can be observed in neat pentacene films, which is consistent with previous reports.8992 The first monolayer covers most of the area with m -sized domains, on top of which islands of second and subsequent monolayers are grown. The neat C60 film, on the other hand, shows a smooth and featureless surface under AFM. The
62 pentacene:C60 mixe d films (as shown in Figure 3 6 ), however, show remarkably different morphologies from those of the two neat films. As the mixing ratio chang es from pentacene :C60 (4:1) to pentacene:C60 (1:2), the morphologies change accordingly. The root mean square Figure 3 6 Tapping mode atomic force microscope (AFM) images of 3 nm thick films on Si (100) substrates: neat pentacene (a), neat C60 (e), pentacene:C60 = 4: 1 (b), 1:1 (c) and 1:2 (d) mixed films. The scanning areas is 55 2, and the height scale is 25 nm for all the images 25 nm 25 nm 25 nma) c) d) e) 25 nmb) 25 nm
63 roughness ( Rrms)increases from 2.0 nm for pentacene:C60 (4:1) to the maximum value of 2.9 nm for pentacene:C60 (1:2). The Rrms value of all mixed films is higher than that of either neat film (1.4 nm for pentacene and 1.5 nm for C60). Furthermore, two different types of features can be identified in the AFM image for pentacene:C60 (4:1): islands with a typical size of 200 nm an d an average height of a few nanometers as well as relatively flat domains in between islands. As the C60 concentration increased, the island features become more prominent with increased height The flat domains, on the other hand, gradually diminish and become almost nonexistent in the mixed films with more than 50 % C60. Based on the XRD patters from the 30nm films shown above, we believe that the flat domains correspond to the crystalline pentacene aggregates which contribute to the characteristic penta cene diffraction peaks, whereas the islands could be an amorphous pentacene -rich phase. Effect of Deposition Rate on Structure a nd Morphology The AFM study of ultrathin films (3 nm thick) indicates the existence of two different features in the pentacene:C60 mixtures, which leads to a rough surface morphology. We have found that these two typical features can evolve into significantly different structure s in much thicker films. Moreover, the phase separation can be further affected by the deposition rate, w hich provides a route to control the degree of phase separation. Figure 3 7 shows the SEM and AFM images of three 50 nm thick mixed films with mixing ratio of 1:1 (a, b), 1:2 (c, d), and 1:4 (e, f), deposited at a total rate of 0.6 /s. Here, the thickness of 50 nm is chosen as it is very close to the typical thickness used in PV devices. Numerous ridgelike structures on the surface are found on these films. The height of the ridges reaches nearly 200 nm in pentacene:C60 (1:1), substantially higher than the nominal thickness of 50 nm. The density and height of the ridges decrease as the pentacene concentration is decreased from 50 % to 20 %, and the ridge height
64 Figure 3 7 [(a), (c), and (e)] Tapping mode atomic force microscope (AFM) and [(b), (d), and (f)] scanning electron microscope (SEM) images of 50 nm thick pentacene:C60 = 1:1 [(a) and (b)], 1:2 [(c) and (d)] and 1:4 [(e) and (f)] deposited at 0.6 /s (total rate) on e scanning area in the AFM images is 55 2 while the height scale is 300 nm. b ) a ) d ) f) c ) e )
65 reaches around 50 nm in pentacene:C60 (1:4). This leads to a reduction of Rrms from 50 nm in pentacene:C60 (1:1) to 20 nm in pentacene:C60 (1:4). Besides the ridgelike structure s, many bumps with size of about 200 nm between the ridges can be identified in the SEM image for the 1:1 mixed film (Figure 3 7 a) although similar features become indiscernible in the films with higher C60 concentration. By comparing the morphologies o bserved in the ultrathin film and considering the dependencies of these features on the mixing ratio, we believe that the ridgelike structures found in the thicker film are evolved from the coalescing of the pentacene rich islands observed in the ultrathin films, whereas the bumps between ridges correspond to the flat crystalline domains from the ultrathin film. Similar ridgelike structures also appear in neat pentacene films grown under certain extreme condition where the growth of crystalline pentacene domains is disturbed.93 The presence of C60 molecules in the mixtures certainly represents one of the scenarios that the crystalline grow th of pentacene is disturbed. The very rough surfaces of pentacene:C60 mix ed films observed in Figure 3 -7 could be practically undesired for OPV device as they could result in poor metal/organic contact and low device yield. One way to obtain smoother m orphology is to utilize the non-equilibrium nature of the vacuum deposition processes and push the growth of thin film into the kinetic regime. Here, we increase the deposition rate so that the molecules arrived at the substrate surface have less time to d iffuse on the surface before being buried the later arriving molecules. Figure 3 8 shows the SEM and AFM images of three mixed films deposited at a much higher rate of 6 /s, with a mixing ratio of 1:1 (a, b), 1:4 (c, d), and 1:5.5 (e, f). It can be seen t hat those long ridges observed in films deposi ted at low rate (see Figure 3 7 ) break down into much shorter and smaller ones and no bumps can be identified in areas between the ridges. For the 1:1 mixed
66 Figure 3 8 [(a), (c), and (e)] Tapping mode atomic force microscope (AFM) and [(b), (d), and (f)] Scanning electron microscope (SEM) images of 50 nm thick pentacene:C60 = 1:1 [(a) and (b)], 1:4 [(c) and (d)], and 1:5.5 [(e) and (f)] deposited at 6 /s (total rate) on a Si substrate. The scale bare and the scanning area in the AFM images is 55 2 while the height scale is 300 nm. b ) d ) a ) c ) e ) f)
67 film, the surface roughness is significantly reduced from Rrms = 50 nm for the film deposited at 0.6 /s to 26 nm for the film with a 6 /s deposition rate. To further confirm that the high deposition rate can suppress the stacking of pentacene molecules in the mixed film, XRD measurement of 1:1 mixed films fabricated at 0.6 /s and 6 /s is c arried out. As sho wn in Figure 3 9 a small diffraction peak corresponds to the crystalline phase pentacene can be identified in the diffraction pattern of pentacene:C60 (1:1) deposited at 0.6 /s, whereas no peak is observed in 1:1 mixed film grow n with 6 /s. All these su ggest that the mixed films are more uniform and that the formation of both crystalline pentacene domains and pentacenerich structures is significantly suppressed. In the 1:5.5 mixed film ( Figure 3 8 e and f) almost no ridgelike features can be recognized under SEM or AFM, resulting in a low Rrms = 6 nm. Figure 3 9 X -ray diffraction patterns for 50 nm pentacene:C60 (1:1) mixed films fabricated under 0.6 /s (red curve) and 6 /s (blue curve). Inset: the corresponding SEM image of the 4 5 6 7 81000 High deposition rate Low deposition rate Intensity (a.u.)2 (degree)
68 Molecular Dynamics Simulation of Pentacene:C60 Mixtures In this section Molecular Dynamics (MD) simulation is used to investigate the structural evolution within the pentacene:C60 mixtures. The MD simulation works presented here is conducted by Sharon Pregler Jason Myers, and Dr. Susan Sinnott Method of Molecular Dynamics Simulation Molec ular dynamics (MD) simulation is a technique that widely used to compute the equilibrium and transport properties of classical many -body systems.84,85 In MD simulations, a model system composed of N particles is first built and initialized. The MD process then numerically inte grates Newtons equations of motion to predict particle responses to applied forces until their properties no longer change with time, which is also known as equilibrating the system. After equilibration, atom displacements, bond lengths, crystal structure or other macroscopic properties of interest can be calculated. In the current MD simulations of structural evolution in pentacene:C60 mixtures empirical many -body potentials are employed because it can allow the MD simulations to model the relaxation of thousands of atoms in a relatively short time. Here, the adaptive intermolecular reactive empirical bond order (AIREBO) potential is used.84,94 Compared with the secondgeneration reactive empirical bond order (REBO) potential, which was designed to model short range interactions of carbon and silicon including hybridization and bond breaking and reformation, AIREBO can account for the nonbonding interactions via an adaptive method, hence overcoming the limitations that REBO has on longranged interactions. A 6 12 Lennard Jones (LJ) potential ( ELJ) and torsion potential ( Etors) are used to described the nonbonded interactions through the equation: = + + where
69 = ( > ) The functions ( ) and ( ) describe pair additive interactions for core core repulsive and attractive interactions, respectively, accounts for the bondi ng between atoms i and j while is the distance between these atoms. A screening function that smoothly transitions between the REBO and LJ potentials is used in AIREBO. The molecular forces depend on the local chemical environment of the atoms, making the AIREBO potential well suited for modeling liquid hydrocarbons, thin films, and hydrocarbon molecules and mixtures. In order to check the validity of the computational method we first used AIREBO to calculate the cohesive energies for bulk C60 and bulk pentacene. As summarized in Table 3 1, the experimental cohesive energy determined from heats of sublimation of bulk C60 and pentacene is 1.74 eV and 1.30 eV, respectively.95,96 These two experiment values match very well with the cohesive energy obtained through AIREBO, which is 1.70 eV for C60 and 1.38 eV for pentacene. This suggests that AIRE BO is an adequate tool for studying the short range and long range evolution of the molecular systems investigated here. To simulate the structural evolution of pentacene:C60 mixtures, relaxation of the molecules were carried out in a 5 5 7 nm supercell w ith periodic boundary conditions due to the complexity in handling a larger system computationally. To accelerate the molecular relaxation, Table 3 1. Calculated and experimental determined cohesive energy (in eV/molecule) for C60 and pentacene molecules. Molecule Calculated Experimental C60 1.70 1.74 a Pentacene 1.38 1.30 b a Ref. 96 b Ref. 95.
70 the system was place d at an elevated Langevin thermostat temperature of 600 K, which allows the molecules to shift out of their initial positions faster. After that, all the atoms in the system were set to 300 K. When the desired temperature was reached within the system, the thermostat was lifted, setting all the atoms to active. The molecules were then allowed to evolve in time according to Newtons law of motion. The mixtures were set to relax for 100 ps until the energy of the system reached a constant level. We believe th at such a simulation method is appropriate to predict the final thermodynamic stable structure of the selected pentacene:C60 mixtures. Structural Evolution in Pentacene:C60 Mixtures The computational studies begin with a virtual mixed film of pentacene and C60 built from an ordered structure approach. A low density structure was built with a starting unit cell and a given mixing ratio, and was then duplicated in three dimensions to create a supercell. The pentacene:C60 mixture with a weight mixing ratio of 2.3:1, or 6:1 by molar ratio, was first chosen to simulate the morphology evolution. For consistency, all mixing ratios referred to in this chapter will be by weight unless otherwise noted. In the pentacene:C60 (2.3:1) structure, the unit cells were arranged in a simple cubic (SC) structure with a density of 1.0 g/cm3. Figure 3 10a shows a snapshot of the SC structure pentacene:C60 (2.3:1) mixture before relaxation, where each fullerene molecule is surrounded by six pentacene molecules. It can been seen from Figure 3 10b that after 100 ps of equilibration the structure of the mixture changed significantly, where molecules of the same type clump together but are not completely phase separated. Extending equilibrat ion time up to 300 ps does not lead to complete phase separation. The potential energy curve as a function of molecular position is considered as a flat potential energy surface. Nevertheless, varying the initial structures of the molecules could lead to n umerous local minima. We found that in different initial structures, face -centered cubic (FCC) or body -centered cubic (BCC) for example, the pentacene and C60 equilibrate in the same
71 manner as mentioned above. The planar pentacene molecules have a tendency to stack on each 32 Due to the presence of much more pentacene than C60 molecules (6 to 1 in numbers) in the mixt ure, it is clear from Figure 3 10b that some of the pentacene molecules form stacks in addition to simple agglomerating. Figure 3 10. Molecular model of petancene:C60 (2.3:1) (by weight) (a) before and (b) after 100 ps equilibration. Periodic boundaries are applied in three dimensions. The unit cell dimension is 557 nm with a density of 1.0 g/cm3. (Courtesy of S Pregler) a) b)
72 In order to quantitatively describe the stacking nature of the pentacene molecules from the simulation, we use a combination of pair -wise distance calculation and molecular vector agreement. In this method, each pentacene molecule is assigned a longitudinal and transverse vector, along the benzene rings and across them, respectively, and the corresponding vectors for all molecules in a stack should be the same. Here, pentacene molecules are considered as forming stacks if their corresponding vectors are aligned w ithin an angle of 10. Additionally, intermolecular center to -center distances are also checked to make sure that molecules with aligned vectors are indeed stacked closely, not just coincidentally aligned. We choose a cutoff interplanar distance of 4.1 which is 0.5 greater than the closest interplanar stacking distance between neighboring molecules in a pentacene crystal.97 T he histogram shown in Figure 3 11a indicates that 22% of the pentacene molecules in the equilibra ted (2.3:1) mixture form pairs, and approximately another 8% of the pentacene molecules form either three or four molecule stacks. The inset of Figure 3 -11b shows all the Figure 3 11. Pentacene stacks in the equilibrated pentacene:C60 (2.3:1) structure (left) Percentage of pentacene molecules forming stacks versus number of molecules in each stack, (right ) isolated pentacene stacks in the film. (Calculation done by J Myers) 2 3 4 0 10 20 30 40 Percentage of pentacene / %Number of Molecules/Stack
73 pentacene molecules that form stacks inside the mixture. Thi s suggests that the pentacene molecules tend to form aggregates in the mixture after equilibration. The aggregation nature of C60 molecules, however, can be quantified using the pair distribution functions (PDFs).98 Figure 3 12a compares the PDFs of the C60 molecules in the penacene:C60 (2.3:1) mixture before and after relaxation as well as that of a pristine C60 crystal, which adopts a FCC lattice structu re.99,100 It is found that the center to -center nearest neighbor distance of C60 molecules is reduced from approximately 1.5 nm to 1 nm after system is equilibrated. This nearest neighbor distance after relaxation is very close to that in an FCC lattice of C60 crystal, which indicates the presence of short range order in the relaxed mixed film. The oscillation in the PDF of the equilibrated structure becomes relatively small at distance > 1.5 nm, suggesting the lack of long-range ordering of C60 molecule in the equilibrated mixed film. Figure 3 12. C60 aggregation in pentacene:C60 (2.3:1) before and after 100 ps equilibaration, (left) Pair distribution functions (PDFs), g(r), of the C60 molelcues as a function of the pair distance, r. The PDF of the face -centered -cubic C60 crystal is aslo shwon as comparsion. (right) isolated C60 m oleculs in the equilibrated film. (Courtesy of S Pregler) 0 1 2 3 4 5 6 7 0 0.5 1 1.5 2 2.5 g ( r ) r ( nm) After Relaxation Before Relaxation FCC Crystal
74 Effect of Mixing Ratio and Deposition Rate In the above structural evolution simulation of the penacene:C60 (2.3:1) mixture, we found that both pentacene and C60 molecules have certain degree of tendency to agglomerate with themselves. Here, the simulation is conducted by first building the bulk mixture, which is then allowed to relax. Such simulation resembles the annealing process in the real case and helps to understand the thermodynamic stable s tructure of the mixture. However, consideration of kinetic process involved during the mixture formation, deposition process for instance, is missing in this simulation. Moreover, the simulated structure is a pentacene rich mixture, which is biased to clai m general aggregation nature of pentacene or C60 molecules under other mixing ratio. Therefore, we investigated the effect of mixing ratio and deposition rate on the morphology of pentacene:C60 mixed films computationally. A Monte Carlo101 random film builder (RFB) program is employed to simulate the experimental film deposition process. In the RFB program, pe ntacene and C60 molecules, at a predetermined ratio but with a random order, appear on random locations on a virtual flat surface with in a supercell area. The program also randomizes the translation and rotation of the molecules. Once a layer of molecules is created by the RFB, the films were then relaxed suing MD simulations. Here, films with two different mixing ratios of pentacene:C60 = 1:2.6 and 1:5.5 deposited under low and high deposition rate were simulated. The condition of low deposition rates is mimicked by first building one layer using RFB, followed by equilibrating the layer at 300 K. This process is repeated for two more layers, i.e. another RFB layer is deposited on top of the previous equilibrated one and is equilibrated together with the underlying layer(s) before next layer is deposited. The high deposition rate condition in experiment, however, is approximated by generating three different layers with the RFB consecutively and equilibrating the entire structure altogether. Depending on how the randomization of molecules affects the interface of
75 the layer underneath, the distance between the centers of each layer range from 11 to 13 Similar to simulation in low density structures, the whole system was set to thermostat at 300 K, wh ich was then lifted after system reached the desired temperature. Figure 3 13 shows the simulation results of the petancene stacks formed in the pentacene:C60 (1:2.6) and (1:5.5) mixed films deposited at the low rate (LR) condition. Because of the higher C60 concentration in these films, only two -molecule statcks, i.e. pairs, are found after 100 ps relaxation. It can be seen that with 10 as the upper limit for the angle offset of the molecular vector only one stack is observed for the (1:2.6, LR) film. However, relaxing the stack detection criterion by increasing the angle offset results in more pentacene stacks being detected, and six stacks are observed when the angle offset is set to 30. An even higher C60 concentration in the (1:5.5, LR) mixed film le ads to substantially less stacks, with only one stack at 30 Figure 3 13. Number of two -molecule pentacene stacks versus the maximum allowed angle offset to qualify as stacks for the pentacene:C60 (1:2.6) and (1:5.5) deposited at low and high deposition rates (LR and HR, respectively) from simulation. 10 15 20 25 30 0 1 2 3 4 5 6 Number of Pentacene StacksAllowed Angle Offset (Deg) LD 1:2.6 HD 1:2.6 LD 1:5.5 HD 1:5.5
76 allowed angle offset. Figure 3 13 also reflects the pentacene stacking condition in two mixed films deposited at high rate (HR). Similar to the situa tion in low deposition rate case, significantly fewer stacks are found in the (1:5.5 HR) film than in the (1:2.6 HR) film. However, no stacks is observed in either film if the angle offset is set to 15 or lower, while two stacks are found in the (1:2.6 LR ) film. For either (1:2.6) or (1:5.5) mixing ratio, the variation in the number of detected pentacene stacks in the LR and HR films is no more than one when the maximum allowed angle offset is 20 or higher. We believe this is due to the limitations of the computational system where only three layers of molecules are involved. If more molecular layers were considered in the simulation, it is expected to see a more remarkable difference between the two cases, i.e., the movement of underlying molecules would be suppressed by the overlayers more significantly under high deposition rate condition. Nevertheless, the computational results from different mixing ratio and deposition rate scenario indicate that both higher C60 concentration and higher deposition rate lead to a lower degree of pentacene aggregation. However, approximately five pair -stacks out of a total 51 pentacene molecules in one supercell still exist in the (1:2.6) mixed film, indicating the strong aggregation nature of the pentacene. Photovoltaic Performance of Pentacene:C60 Mixture s To correlate the photovoltaic performance of pentacene:C60 mixtures to the degree of phase separation in the mixture, OPV cells with a device structure of ITO/pentacene:C60 (50 nm)/BCP (8 nm)/Al were fabricated. All OP V devices were fabricated on glass substrates pre -coated with was tested in air without encapsulation. An Agilent 4155C semiconductor parameter analyzer was used t o measure the current density -voltage ( J V ) characteristics of devices in the dark and under simulated AM 1.5 solar illumination.
77 Figure 3 14. Current density versus voltage ( J V ) characteristics for devices with pentacene:C60 =1:1 (circles), 1:2 (t riangle) and 1:5.5 (sqares) in dark (open symbol and under 120 mW/cm2 of simulated AM 1.5G illumination (solid symbol). The device structure is ITO/pentacene:C60 (50 nm)/bathocuproine (BCP) (8 nm)/Al (100 nm), and the active layers were deposited at 6 /s. Here, the 50 nm pentacene:C60 active layers with three different mixing ratios (1:1, 1:2 and 1:5.5 by weight) deposited at ~ 6 /s were used to fabricate the OPV cells. The J V characteristics of these three devices in dark are shown in Figure 3 14. The device with a 1:5.5 mixing ratio shows a rectification ratio of more than 104 (at 1 V), while the 1:1 and 1:2 devices do not show any rectification. The J V characteristics of the three devices under illumination intensity of 120 mW/cm2 are shown in Figure 3 14 as well It can be seen that the short -circuit current (JSC) and fill factor (FF) for the 1:5.5 device are 1.48 mA/cm2 and 0.29, respectively. Both the 1:1 and 1:2 devic es show much lower JSC and FF, especially for JSC which is reduced by factor of approximately 150 compared with the 1:5.5 device. The open circuit voltage ( VOC), on -0.8 -0.4 0.0 0.4 0.8 10-610-510-410-310-210-1100101102 J ( mA/cm2)V (V) (1:1) (1:2) (1:5.5) Dark Under illumination
78 Table 3 2. Comparison of the short -circuit current density ( JSC), opencircuit voltage ( VOC), fill factor (FF), and power conversion efficiency (PCE) of two bulk heterojunction devices based on pentacene:C60 = 1:1, 1:2 and 1:5.5 (by weight) under 120 mW/cm2 simulated AM 1.5G solar illumination. J SC ( m A/cm 2 ) V OC (V) FF PCE (%) Pentacene:C 60 (1:1) Pentacene:C60 (1:2) 9.7 10 3 1.7 102 0.45 0.56 0.14 0.12 5.1 10 4 9.5 104 Pentacene:C 60 (1:5.5) 1.48 0. 60 0.29 0.2 1 the other hand, gradually decreases from 0.6 V in the 1:5.5 device to 0.56 V in the 1:2 device and 0.45 in the 1:1 device. The substantial change of VOC with respect to the mixing could possibly be due to the increase of photocurrent in devices with higher C60 concentration. This can be associated with our previous understanding that reduced domain size can be achi eved by increasing the C60 concentration in the mixed film, leading to improved exciton dissociation. Also, the VOC of current pentacene:C60 bulk heterojunction devices is higher than the typical bilayer heterojunction device ( VOC ~ 0.4 V),82,83 which can be attributed to the much lower dark current observed in the bulk devices than that in bilayer devices. The 1:5.5 device shows significant improvement of JSC, VOC and FF over 1:1 and 1:2 devices, which results in 0.21 % power conve rsion efficiency (PCE). Table 3 2 compares a few importan t PV parameters of the three devices under illumination. The PCE of the 1:5.5 device is approximately 400 times higher than the other the other two mixing ratio devices. Since th e 1:5.5 mixing ratio provides a working PV device we also compared the effect of deposition rate on the pentacene:C60 (1:5.5) devices. Figure 3 15 shows the J V characteristics of two pentacene:C60 (1:5.5) devices with the mixed layer fabricated at ~ 0.6 /s and ~ 6/s. It can be seen that JSC = 0.65 mA/cm2 in the ~0.6 /s device, which is less than half of the ~ 6 /s device. The VOC = 0.64 V, however, which
79 is higher in the ~ 0.6 /s device due to suppressed dark current. The resulting PCE is only 0.10 % in the ~ 0.6 /s device, which is approximately 50 % lower than the ~ 6 /s device. Figure 3 15. Current density versus voltage ( J V ) cha racteristics for pentacene:C60 = 1:5.5 devices with the mixed layer deposited at ~ 0.6 /s and ~ 6 /s under 120 mW/cm2 of simulated AM 1.5 illumination. Discussion In the current MD simulation, the 557 nm supercell is relatively too small to simulate the formation of ridgelike structures (which are a few hundred nanometer is size), but we still gain useful insights into the phase separation process between the pen t acene and C60 in the mixture. The XRD patterns have confirmed the existence of crystalline pentacene domains in mixtures with more than 50 % pentacene. This is somehow consistent with the MD simulation results, which show that nearly 30 % of the pentacene molecules form stacks in the (2.3:1) mixed film after relaxation. Moreover, both computational and experimental studies indicate that the portion of pentacene molecules forming stacks decreases as the pentacene concentration in the -1.0 -0.5 0.0 0.5 1.0 -4 -3 -2 -1 0 1 Pentacene:C60 (1:5.5) (~ 0.6 A/s) Pentacene:C60 (1:5.5) (~ 6 A/s) J ( mA/cm2)V (V)
80 mixed film decreases. Although the XRD pattern indicates crystalline pentacene domains do not exist in when the pentacene concentration is lower than 50 %, the stacked pentacene molecular pairs may still be present even in the 1:5.5 mixed films as from the simulation results. Here, simulation serves as a powerful complement to experimental study. Previous studies on CuPc:C60 mixtures also suggest the presence of CuPc aggregates in the mixture. However, the aggregation of CuPc can be strongly suppressed by introducing even a small amount of C60 (25 % by weight) into the mixture.58 Compared to CuPc, the pentacene molecules exhibit a stronger tendency to aggregat e which could be attribute d to the strong intermolecular interaction between pentacene molecules.102 In addition, the energy barrier for pentacene molecules to rearrange and form stacks in the mixtures is much lower than that of CuPc due their smaller size. The aggregation of C60 molecules is difficult to characterize experimentally due to their amorphous nature. However, the PDFs from MD simulation do show that the first nearest neighbor distance of C60 in the mixtures reduces after relaxation, approaching the value in a FCC C60 crysta l. Such short range order (but no long range order) of C60 molecules in the mixture could be viewed as indicative of aggregation to some extent. Even though a certain level of phase separation in a D A bulk heterojunction is considered to be beneficial to enhancing PV performance, too strong phase separation could also be problematic.79,103 Strong phase separation, for instance, may produce pure domains with size s significa ntly larger than the exciton diffusion length. This will lead to lower exciton diffusion efficiency with a large portion of excitons unable to reach the D A interface and dissociate into free charge carrier. In other cases, the phase separated large doma ins may contain a tiny amount of the other molecular species because of the spinodal decomposition process, forming donor or acceptor rich domains. In this scenario, the exciton diffusion efficiency could be still be high
81 initially due to the presence of minority molecules with the domains, but the dissociated charge carrier s will be trapped on these isolated molecules which prevents the further dissociation of excitons at the D A heterojunction. Too strong phase separation in the mixed film will also l ead to significantly increased surface roughness causing poor contact between the top electrode and organic layer as well as a greater chance of electrode shorting between the electrodes. In the current pentacene:C60 mixtures, the pentacene has such strong aggregation that crystalline pentacene domains with size s between 100 and 200 nm can be readily observed in the mixed films with pentacene concentrations higher than 50 %. Moreover, ridgelike long features can also be found in pentacene:C60 (1:2 ) and (1:4) films due to the strong aggregation of pentacene. XRD pattern without any characteristic diffraction peak suggests these ridgelike structures are amorphous. Since the prevalence of ridgelike structures decreases with pentacene concentration, it is believed that the ridges are composed of amorphous pentacene -rich clusters. Amorphous pentacene clusters were previously found in neat pentacene films as well, where the pentacene molecules are lying flat in the clusters instead standing on their edges as in the crystalline phases.93 Rrms of 50 nm pentacene:C60 (1:1) film reaches 56 nm with 300 nm peak to valley distance of the surface features. Depositing 100 nm thick Al on such rough surface could still produce incomplete coverage, causing poor contact and erratic J V curves of the device. The CuPc:C60 films with similar mixing ratio, however, show very smooth ( Rrms < 1 nm) and featureless surface under SEM or AFM is observed.58 This again suggests stronger tendency of aggregation of pentacene than CuPc. Because of the too strong phase separation in pentacene:C60 mixed, it is necessary to suppress the phase separation to nanoscale in the mixture to obtain better PV performance. It has been demonstrated in the above studies that higher C60 concentration could hinder the stacking of
82 p en t acene, hence suppressing the phase separation. With more than 50 % C60 in the mixture, the crystalline phase pentacene become almost nonexistent, while amorphous pentacene rich domains could still exist in the mixture. In order to fully suppress the pe ntacene aggregation, C60 concentration as high as 80 % is required. Another way to manipulate the degree of phase separation is through varying the deposition rate. Basically, the low deposition rate allows molecules to diffuse longer distance over the sur face and reconfigure themselves into a lower energy state, which promotes the growth of a structure closer to a thermodynamic ally stable state. A high deposition rate, on the other hand, will hinder the diffusion or rearrangement of molecules on the surfac e, leading to weaker phase separation. Therefore, strong suppression of phase separation in pentacene:C60 mixed films is achieved with a 1:5.5 mixing ratio and high deposition rate. With such mixed film, the bulk heterojunction OPV device shows a PCE of 0.21 %, which is a 400-fold increase of efficiency compared to a pentacene:C60 (1:1) device. Such significant improvement in PV performance is related to better morphology in the pentacene:C60 (1:5.5) film. The 1:1 mixed film show s a rough surface with large domain (100 to 200 nm). The 1:5.5 film, in contrast, is much smoother and show s very rare large domains. All this indicates the successful suppression of phase separation in the 1:5.5 mixed film, whose morphology is more favorable for effi cient excit on diffusion and charge collection. Nevertheless, the PCE of the pentacene:C60 (1:5.5) BHJ device is still low compared with a bilayer pent acene/C60 device, which is between 1.1 and 1.6 %.104,105 This is due to the fact that only a small amount of pe n tacene molecules are allowed in the mixture (only ~ 15 % by weight) to successfully suppress the formation of large pentacene aggregates. As a result, the absorption of incident photons in the region of 500 to 700 nm will become very weak. Moreover, the low
83 pentacene concentration will also lead to poor hole transport across the film, which is reflected by the low fill factor in the OPV device (FF = 0.29). Conclusions In this chapter, the phase separation process in pentacene:C60 molecular donor acceptor mixtures is investigated using both computational and experimental methods. A lso, results from simulation and different experimental characterization su ggest that there is pentacene aggregation in the mixtures. The strong aggregation of pentacene molecules in the mixture results in an extremely rough surface and poor photovoltaic performance. Such observation is quite different from previous studies on Cu Pc:PTCBI and CuPc:C60 mixtures, where the phase separation is very weak in the as -deposited films and the mixed films present smooth and featureless surfaces. Moreover, phase separation in CuPc:PTCBI mixture s is enhanced through thermal annealing. In the current pentacene:C60 mixture, however, the phase separation can be controlled by tuning the pentacene to C60 mixing ratio and the deposition rate. By adopting a 1:5.5 mixing ratio and high deposition rate, we can successful suppress the phase separation in the pentacene:C60 mixed film and obtain a smooth surface. The efficiency of the device with an optimized level of phase separation is 400 times higher than that of a device based on a 1:1 mixed ratio, in which the scale of phase separated domains reaches ~100 nm. Further investigation of the relation between phase separation degree and other process parameters, such as substrate temperature, molecular properties, etc., is of great importance in engineering the nanoscale morphology of the bulk heterojucntio n for more efficient energy conversion.
84 CHAPTER 4 ORGANIC PHOTOVOLTAIC CELLS WITH ALIGNED C RYSTALLINE MOLECULAR NANORODS Introduction The research on organic photovoltaic (OPV) cells has made significant progress in the past two decades since the first report of organic bilayer donor acceptor (D A) heterojunction photovoltaic (PV) cells. However, the power conversion efficiency of OPV cells is still much lower compared with their inorganic counterpart. One of the limiting factors in achieving hig h efficiency OPV cells is the short exciton diffusion length in most organic materials, typically a few nanometers. Currently, bulk D -A heterojunc t i on, which is achieved by well controlling phase separation in a D -A mixture, is widely used to circumvent th e exciton diffusion bottleneck. However, the random interpenetrating D -A network obtained from phase separation could lead to poor charge transport due to charge trapping at the bottlenecks or cul -de -sacs along the conducting path. This could lead to low c harge collection efficiency. Another drawback of relying on phase separated D -A bulk heterojunction (BHJ) for efficient exciton diffusion and charge collection is the lack of control over morphology inside BHJ. As what we learned in Chapter 3 the degree o f phase separation in D -A mixture strongly depends on the strength of the interaction between the molecules as well as the processing conditions during the formation of BHJ. For instance, the aggregation of pentacene molecules in pentacene:C60 BHJ is so st rong that large domains (a few hundred nanometers) a re formed, which results in low exciton diffusion efficiency Moreover, the increased surface roughness induced by the strong aggregation also leads to poor contact between the electrode and the organic l ayer. On the other hand, very weak phase separation is observed in CuPc:C60 BHJ, where highly intermixing between two species is observed.58 Although t he exciton can all be dissociated in this case, the charge collection efficiency can be very low due to discontinued conducting paths. Therefore, the degree of phase
85 Figure 4 1. Schematic of i deal interdigitated bulk heterojunction, where the lateral di mension of the donor and acceptor phase is close the exciton diffusion length. The presence of straight conducting path for both electrons and holes secure efficient charge collection. separation is mostly determined by the nature of the organic molecules, which put s a heavy burden on the synthesis of molecule s with desired properties. Even though phase separation can be controlled through varying the processing condition s the level of controllability is limited. One way to overcome the above limitations i n D -A mixed BHJ is to use a bulk heterojunction composed of nanoscale interdigitated D -A phase, which is indicated in Fig ure 4 1 Here, the lateral dimension of the interdigitated structure may be composed of pure phase donor or acceptor materials Such st ructure can secure both efficient excition diffusion and charge collection if the lateral dimensions are small enough Since the charge transport within a pure phas is improved compared with mixed BHJ, thicker film s can be used to increase absorption. Eve n though such interdigitated D -A BHJ is ideal fo r efficient OPV cells, fabrication of such nanostructure is challenging. The interdigitated BHJ is usually formed by infiltrating the second phase into the existing one -dimensional nanostructure template comp osed of another phase of material. Recently, extensive research has focused on growing one -dimensional organic or inorganic nanostructure, followed by infiltration of acceptor or donor phase into the Electron transport Hole transport Cathode Anode Photon Electron Hole Acceptor Donor
86 nanostructure to achieve OPV device.77,106 109 For instance, Yang et. al. use organic vapor phase deposition (OVPD) to grow copper phthalocyanine (CuPc) nanorods, which then infiltrate with PTCBI.77 The BHJ achieved using this method showed a nearly two-fold enhancem ent over the planar heterojunction device. Hsu, et. al. also showed growth of ZnO nanorod arrays through a hydrothermal approach, followed by loading P3HT into the ZnO nanorods to achieve a hybrid PV device.108 However, the efficiency of device s using interdigitated BHJ is still relatively low compared with the best mixed BHJ device s. One reason for this is the later al size of one -dimensional nanostructure s is still too big for efficient exciton diffusion. Another reason may be poor infiltration of the second phase into the nanostructure so that excitons are not dissociated in the unfilled region. In this chapter, we demonstrate that one dimensional nanorod arrays composed of CuPc can be easily fabricated on different substrates using oblique angle deposition. The process of CuPc nanorod growth is studied and the dependence of nanorod morphology under different deposit ion conditions is compared. By controlling the processing conditions, the [6,6] phenyl C61-butyric acid methyl ester (PCBM) is successfully infiltrated into the nanorod arrays to achieve an interdigitated BHJ OPV cell. The optimized nanorod CuPc/PCBM devic e achieves a maximum power conversion efficiency (PCE) of p = (1.8 1) %, which is approximate ly twice that of the bilayer CuPc/PCBM device. The contents in this c hapter are organized as follows: the first section introduces oblique angle deposition (OAD) ; the second section is t he structure and morphology study of the OAD grown CuPc nanorod arrays under different condition followed by CuPc nanorod/PCBM device study; t he third section discuses the advantages and existing pro blems in the current approach
87 and points out possible solution to fu rther enhance device efficiency; t he last section summarize the concl usions Oblique Angle Deposition Oblique angle deposition (OAD), which is also known as glancing angle deposition, has been widely used to grow inorganic one -dimensional nanostructures with potential applications in memory, photovoltaic, and sensor devices.110114 Figure 4 2 illustrates the schematic of the OAD process. Unlike conventional high vacuum deposition pr ocesses use the ballistic molecular/ atomic beam s that arrive at the substrate surface from near normal directions, an OAD process, the molecular/atomic beam s that arrive at the surface with a large angle from the substrate normal Due to the shadowing effect by the deposited molecules and the limited diffusion of absorbed molecule on the surface, nanorod arrays with various morphologies can be obtained by controlling th e incident angle of the molecular flux as well as the surface diffusivity of the molecules. The formation of nanorod structure under OAD can be divided into two major steps, which include (1) nucleation and initial growth and ( 2) nanorod growth and evolution under ballistic shadowing effect. Figure 4 2. Schematic of oblique angle deposition (a) Nuclei dis tributed across the surface lead to ballistic shadowing of the surrounding regions during oblique angle deposition. ion of one -dimensional nanostructure. The nanorods will grow oriented toward the source, Vapor flux Vapor flux a) b)
88 The nucleation and initial growth of thin film under OAD process is in fact similar to the deposition in normal direction, especially on planar and featureless surface s An absorbed molecule will diffuse around on the surface until it meets another molecule to form a nucleus, or attach to existing nuclei, or reevaporate from the surface. The nucleus become st able once it reaches the critical size ; only after this point does nanorod growth occur The criti cal size of the nucleus plays in important role i n determining the nanorod morphology under OAD process and it depends strongly on the intrinsic characteristi cs of the molecule as well as the surface properties. Depending on the surface and interfacial energy between the substrate surface and absorbed molecules, three different modes for nucleation layer growth can be observed.115 When the interaction between the surface and the absorbed molecules is stronge r than that between the molecules, two -dimensional growth of monolayer with only one molecule in height will occur. This is known as Frank -van de Merwe or layer growth. On the other hand, if the absorbed molecules have a stronger tendency to bind to each other than to the surface, separated three dimensional nuclei will form. This is known as Volmer Weber or island growth. The third type of growth mode falls in the regime between the previous two modes, where absorbed molecules initially cover the surface i n a monolayer and then form islands formation on top of the monolayer. This growth mode is called Stranski -Krastanov or layer plus -island growth. For organic molecules, layer plus -island growth is observed in those with planar structure, such as sexithioph ene, pentacene, etc.31,47,91 For practical OAD application, the Volmer Weber growth mode is preferred due to formation of microscopic topologies which lead to ballistic shadowing. Once a nucleation layer as shown in Figure 4 2a is formed, the nuclei distributed on the surface will lead to a ballistic shadowing effect of their peripheral area, promoting na norod growth on top of the nuclei. Since the molecules only deposit on top of the nuclei, they will
89 develop into nanorods which are tilted toward the direction of the incoming flux. The nanorod empirical equation proposed to is :110 tan = 2 tan Although the ballistic shadowing effect promotes the growth of nanorod structure there are other processes that may alter the morphology of the nanorod. One important consideration is the diffusion of the absorbed molecule on the surface of existing nanorod structure.116 117The surface diffusion of molecules strongly depends on the substrate tempera ture a nd deposition rate. L ow substrate temperature and fast deposition rate will reduce the surface diffusivity of an absorbed molecule; therefore these conditions will facilitate the nanorod growth along the elongated direction. In contrast, elevated sub strate temperature and slow deposition allow the molecule diffuse longer distance over the surface so that a structure resembled that deposited at normal incident angle is obtained. As shown in Fig ure 4 2b, nanorod growth under OAD process is a competitive growth process, meaning that shorter nanorods will be shadowed by longer nanorods and cease growth. S uch extinction is accompanied by an increase in the diameter of the surviving nanor o d s to maintain constant planar density. Therefore, thicker nanorod fil m s will necessarily have thicker nanorods In fact, there are different ways to control the nanorod morphology in OAD process. For instance, instead of using an intrinsic nucleation layer as seed for nanorod growth, a substrate with regular pre patterned f eature can be used to tailor the distribution of nanorod s on the surface. Moreover, by rotating the substrate in a controlled fashion, diversified nanorod morphology can be achieved, which is shown in Figure 4 3. It can be seen that OAD is a versatile and convenient way to fabricate one -dimensional nanorod, and
90 such nanorod structure has potential to generate efficient interdigitated BHJ in OPV cell application. Figure 4 3. Different nanorod structure achieved under OAD process. [Ref 110] Structure and Morphology Study of CuPc Nanorods From the previous Section, we know that in order to grow nanorod structure suitable for OPV cell application, a few criteria have to be met. First nucleation layer grown with Volmer Weber mode and nuclei with 10 to 20 nm size are preferred for forming se eds for nanorod growth. Second, the absorbed molecule should have limited surface diffusivity to promo te one dim ension growth. Third, the nanorod arrays should have optimal packing density so that both good infiltration of the second phase and maximized D -A interface area can be secured. Finally, crystalline nanorod is desired for high charge mobility, hence efficient charge collection. In this section, we investigate the OAD growth of organic nanorod arrays composed of CuPc, which is a donor material widely used in small -molecular -weight base d OPV devices. The morphologies of CuPc nanorod s grown under different OAD conditions are compared to provide an insight on how to control the morphology to suit OPV applications.
91 Structure and Morphology Evolution of Nanorod Films In order to study the growth of CuPc nanorods under OAD condition, we monitor ed the morphologi es of the nanorod film at different stages of the growth. In this study, the CuPc nanorod films were deposited on glass substrates pre coated with a layer of indium tin oxide ion ized water, acetone, and isopropanol consecutively for 45 minutes before loading into a custom -ma de high vacuum deposition chamber with OAD setup (base pressure ~ 1 107 Torr). The molecular g the nanorod growth. The morphology of the nanorod film was investigated using a JOEL 6335F field-emission gun scanning electron microscope (SEM) and a Veeco Dimension 3100 atomic force microscope (AFM). Figure 4 4 shows a series of SEM images indicating the surface morphologies for different stage s of CuPc nanorod growth on ITO. As can be seen in Figure 4 4 a, the bare ITO surface is almost featureless under SEM with vaguely seen large domains with size more than 100 nm. A scrutiny of the ITO surface usin g AFM reveals the existence of smaller domains on the larger domains (see inset of Figure 4 4a) The small domains have size ranging from a few nanometers to tens of nanometers, and the root mean square roughness ( Rrms) considering both type of features is only Rrms = 2.0 nm. Figure 44 b shows that after deposition of a thin layer of CuPc under OAD condition, CuPc domains with size s f rom 10 to 30 nm can be observed, which is similar to planar film deposited at normal incident. Such observation is consist ent with the layer plus island film growth mode observed in a variety of organic films deposited by vacuum thermal evaporation (VTE).25,46 The lattice mismatch between the polycrystalline ITO and the
92 Figure 4 4. Scanning electron microscope (SEM) images of approximately 150 nm thick CuPc nanorod film at d ifferent stages of growth. (a) Bare indium tin oxide (ITO) surface. Inset: atomic force microscope (AFM) image of same substrate. (b) (e) N anorod films at different stage of the growth. The scale bare s are 100 nm in all images. CuPc crystals puts strains in the continuous organic layer, resulting in island formation. With continued oblique angle deposition upon such structure, the CuPc domains develop into pro trusions with maximum diameter of ~ 50 nm (see Figure 44c), indicating the onset of shadowing effect from CuPc domains The increased diameter during the transition from Figure 4 4b to Figure 4 4c is due to the diffusion of molecule s deposited on top of t he domains, promoting three -dimen sional growth rather than two dimensional M oreover, such increased CuPc domain size suggests the surface diffusivity of molecules on the existing CuPc surface is higher than that on the ITO surface. As shown in Figure 4 4d and e, further deposition leads to even rougher surface and short tilted nanorods 50 nm in length can be observed, indicating an a) b) c) d) e) f)
93 enhanced ballistic shadowing effect One -dimensional growth under OAD finally results in the morphologies shown in Figure 4 4f, where elongated nanorods with more than 100 nm in length and 20 to 50 nm in diameter can be observed. Note that existence of smaller and shorter nanorod s versus larger and longer nanorods confirms the competitive growth characteristic of the OAD process, where the survived nanorods become larger and longer to compensate the extinction of smaller nanorod s falling in their shadows. The growth of CuPc nanorod arrays on ITO substrate s was accompanied by the significantly increase of surface roughness, which increased from a few n ano m eters in bare ITO to tens of n anom eters in the final nanorod film Effect of Incident Angle Based on the previous explanation of morphology evolution during the growth of CuPc nanorod s we conducted another study regarding the morphology dependence on the incident angle 65 and 80. In order to mak e meaningful comparison between different nanorod films, the nominal thickne ss of the film was fixed to be 1 00 nm for all four samples. Figure 4 5 shows the SEM images of 100 nm CuPc nanorod films fabricated with four different incident angles It can be observed in Figure 4 5 a that t he bumps rather than nanorods are mps ranges from 20 to 50 nm, and no observable gap existed in between the bumps due to the weak shadowing effect at this incident angle. The film shows a roughness of Rrms = 9.5 nm obtained from AFM, which is not much greater than reported roughness of pla nar CuPc film s (Rrms = 4.8 nm).58 As the incident angle seen on the su rface (see Figure 4 5 b). Also, visible gaps can be observed across the film, indicating the presence of more prominent shadowing effect. Such a corrugated surface leads to a significant
94 Figure 4 5. Scanning electron microscope (SEM) and atomic force microscope (AFM) images of 100 nm thick CuPc nanorod fil m fabricated with different incident angle of = 45 [(a) and (b)], 55 [(c) and (d)], 65 [(e) and (f)] and 80 [(g ) and (h)]. The scale bar is 100 nm for all SEM images. The scanning area of the AF M images is 22 2, and the height is 300 nm. a) b) c) d) e) f) g) h)
95 increase of the roughness to Rrms = 14 nm, which is twice than that of the 45 film. Further and a larger prominence of nanorods as shown in Figure 4 5c. Nanorods with less than 100 nm in length and 20 to 50 nm in diameter can be seen. Larger gaps are observed in 65 film compared with 55 one, with maximum gap size of ~ 60 nm. All these contribute to a continued increase of roug hness to Rrms = 17 nm. Finally, CuPc nanorod arrays with nearly 200 nm long and 20 to 50 nm wide nanorods are ob 5 d). It can be seen that al l nanorods are tilted toward the direction of incoming flux, and the maximum separa tion distance between the nanor od is as high as 100 nm. These are evidence of a strong shadowing effect as goes to 80. The 80 film shows a roughness of Rrms = 73 nm, which is the highest among the four angles. Note that such high Rrms value s obtained through AFM may underestimate the actual Rrms due to detection limit of AFM techniqu e for such fine nanorod arrays; nevertheless, it yields a semiquantitative comparison between different nanorod films. It is also worth noticing that since the fo ur nanorod films have the same nominal thickness, increased porosity fr om 45 to 80 films suggest lower optical density in the high angle sample. Effect of Substrate Rotation In this study, we studied the effect of substrate rotation on the morphology of CuPc nanorod arrays. Compared with OAD with stationary substrate, the rotational substrate allow s the top of the nanorod to receive the oblique angle molecular flux from all directions. Therefore, instead of creating a shadowing effect along one direction in stationary substrate mode, the shadowing effect can be created in all orientations Here, CuPc nanorods are grown on top of wth modes.
96 Figure 4 6. Topographic [(a) and (b)] and cross -sectional [(c) and (d)] scanning electron microscope (SEM) images of CuPc nanorods grown on a stationary [(a) and (c)] or rotational [(b) and (d)] substrate with speed of ~ 5 rpm. The scale bar is 100 nm for all four images. Figure 4 6a and b show the topographic SEM images of CuPc nanorod arrays grown with a stationary or rotational substrate, respectively, whereas Figure 4 6c and d are the corresponding cross -sectional images. As we have learn ed in the previous two parts, the stationary substrate mode generates slanted nanorods with 20 50 nm in diameter, tilting towards the direction of the incoming molecular flux (see Figure 4 6a and c). On the other hand, the CuPc nanorods grown with rotati onal substrates have larger diameters, typically 40 to 70 nm but in some cases up to 100 nm (see Figure 46b and d). Moreover, the nanorods are mostly in the up right orientation instead of tilting in the rotational substrate mode, which is due to the cons tantly changing flux a) b) c) d)
97 direction. Closer observation of the upright nanorod reveals that the larger nanorods are in fact composed of several smaller nanorods with diameter around 20 nm clumping together. This is consistent with the proposed morphology evolution model of OAD growth on rotational substrate. Another significant difference between the two growth methods is the packing density of nanorods. It can be seen from the cross -sectional SEM images that the nanorods are close packing with gap size of arou nd 30 nm with stationary substrate, whereas the nanorods are further separated and gaps as large as 100 nm can be observed using rotational substrate mode. Effect of Surface Property and Deposition Rate Two steps that significantly influence the nanorod mo rphology include the formation of nucleation layer and nanorod formation on the existing nucleation layer, both of which strongly depend on the surface diffusivity of absorbed molecule s It is known that the surface diffusivity are mainly affected be the surface properties, substrate temperature and deposition rate. Therefore, the effects of different surface properties and deposition rate s on the morphology of CuPc nanorod are investigated in this part. Figure 4 7 a and b show the SEM images of CuPc nanor od arrays grown on top of stationary Si and SiO2 Surprisingly, both nanorod films display very similar morphology with 20 50 nm in diameter and around 100 nm in length, seemingly not affected by different substrate properties To better understand why such a phenomenon is present, the morphologies of thin CuPc films deposited on Si and SiO2 substrates under the same OAD condition are compared. As shown in Figure 4 7c and d, the SEM images indicate that the CuPc film grown on SiO2 s how larger average grain size than that grown on Si. The grain size ranging from 40 to 100 nm can be observed in film deposited on SiO2 substrate, whereas the film on Si substrate shows grain size of 10 50 nm. This suggests the re are different wetting co ndition s for CuPc on hydrophobic (Si) and hydrophilic (SiO2) substrate s which has been
98 Figure 4 7. SEM images of t hick [(a) and (b)] and thin [(c) and (d)] CuPc nanorod films grown on Si and SiO2 substrate under identical OAD conditions. The scale bare is 100nm. observed in many vacuum deposited organic films .118,119 Howev er, such difference becomes hard to distinguish after nanorod arrays are fully developed. This can be explained by different surface diffusivity of CuPc on Si and SiO2, which leads to different nucleation morphology. However, as the nanorod start forming, molecules will only deposit on the top of CuPc nanorod, where the surface diffusivity are identical, ignoring the different substrate properties. Therefore, the size of nanorod becomes solely dependent on the surface diffusion distance of CuPc on their ow n domains. Since the temperature and deposition rate are the same for OAD on Si and SiO2 substrate, the identical CuPc surface diffusion distance leads to similar nanorod size. Based on the above study, it is more practical to control the CuPc nanorod morp hology through manipulating the surface diffusivity of the molecule. It is known that fast deposition rate will result in short surface diffusion length because the absorbed molecule has less time to a) b) c) d)
99 Figure 4 8. Scanning electron microscope ( SEM ) images of CuPc nanorod film grown under identical OAD conditions but different deposition rate of (a) ~ 10 /s and (b) ~ 0.5 /s. The scale bar is 100 nm. rearrange their position before being buried by the incoming molecules. Here, we comp are the effect of deposition rate on the nanorod morphology, where nanorod films deposited at ~ 0.5 /s and at ~ 10 /s are fabricated on ITO 8 a and b show the SEM images of CuPc nanorod film s deposited at two different rates but with the same t arget thickness (determined by quartz crystal monitor). It can be seen that the shorter nanorod with lower packing density is observed in film s deposited at ~ 0.5 /s, whereas dense nanorod arrays with longer nanorod is obtained using deposition rate of ~ 10 /s. Nevertheless, the diameters of nanorods obtained under two conditions are very similar, ranging from 20 to 50 nm. Such observation is consistent with the proposed morphology evolution under OAD condition, where a longer surface diffusion distance p romotes the growth of morphology close to t hat deposited at normal incidence In the current study, this is indicated by the shorter and sparse CuPc nanorod using a low deposition rate In some extreme cases, the surface diffusivity of the molecule is so s trong that no nanorod structure is formed under OAD process. a) b)
100 Structure of CuPc N anorods So far, only the morphologies of CuPc nanorods grown under different OAD conditions have been investigated. However, the application of such nanorod fabrication on efficient interdigitated BHJ OPV cell s also requires good charge transport across these nanorods. Here, xray diffraction (XRD) patterns of the CuPc nanorod and planar films were investigated using a Philips Xpert MRD diffractometer in the geometry with a Cu K radiation source. The XRD pattern of a CuPc n anorod film grown with a rotational substrate is shown in F igure 4 9 along with that for a CuPc flat f ilm deposited at normal incidence The XRD patterns are very similar, showing diffract ion peaks at 2 CuPc phase in the nanorods as well as in the flat film.120 The polycrystalline structure of CuPc nanorods is important t o ensure high mobility for hole transport through the donor material, leading to high collection efficiency for photogenerated holes in the PV devices. Figure 4 9. X -ray diffraction (XRD) pattern of a CuPc nanorod film with 300 nm long nanorods (blue) and a 100 nm thick flat CuPc film (red). 4 6 8 10 12 14 Nanorod film Diffraction Intensity (a.u.)2 (deg) Planar film -CuPc -CuPc (100)
101 Organic Photovoltaic Cells Based on CuP c Nanorods and PCBM With the obtained CuPc nanorod arrays from OAD process, an interdigitated bulk heterojunction can be achieved by infiltrating the acceptor materials into the nanorod arrays. CuPc and C60 have been demonstrated to be one of th e most efficient small -molecule -based D A heterojunction for OPV appl ication, and a maximum PCE of 4.2 % have been achieved by Xue et. al.121 Nevertheless, infiltrating of C60 into the current CuPc nanorod arrays through vacuum thermal evaporation method is problematic due to the ballistic transport of molecule s in high vacuum.107 Therefore, solution processing with PCBM, which is soluble fullerene derivative was selected to achieve better infiltration in to the nanorod film. Here, performance of OPV cells based on the resulting CuPc/PCBM interdigitated BHJ is investigated. Infiltration of PCBM i nto Nanorod Arrays Before preparing OPV device s based on CuPc nanorod/PCBM BHJ, it is first important to ensure good infiltration of PCBM into the nanorod arrays. Here, the interdigitated BH J was achieved by spin-coating a chlorobenzene solution of PCBM on to the nanorod arrays. Figure 4 10a and b show two cross -sectional SEM images of CuPc nanorods/PCBM composite films Figure 4 10. Cross -sectional SEM images of CuPc nanorod/PCBM composited films with PCBM concentration of (a) 15 mg/mL and (b) 30 mg/mL in the chlorobenzene solution. The scale bar is 100 nm. a) b) Nanorods /PCBM ITO ITO Nanorods /PCBM
102 with different PCBM loading. The length of the CuPc nanorods grown with a stationary substrate both cases. No obvious voids or pin -holes are observed in either film, indicating good infiltration of PCBM into the spacing between CuPc nanorods. With a high PCBM loading (30 mg/mL in chlorobenzene), the gap between CuPc nanorods is completely filled, re sulting in a relatively smooth top surface ( see Figure 4 10b) However, when a much lower PCBM loading (15 mg/mL in solution) was used, the amount of PCBM molecules deposited was not sufficient to completely fill the spacing between the CuPc nanorods, lead ing to a corrugated and rough surface as shown in Figure 410 a. This also suggests that the CuPc nanorods still stand on the substrate and the spin-coating process does not damage the contact of the nanorods with the underlying ITO electrode, which is impo rtant for hole co llection in PV devices. Figure 4 11 also shows the absorption spectra of a CuPc nanorod film before and after spin = 500 nm in CuPc nanorod/PCBM composite film compared with neat CuPc nanorod film indicates reasonable amount of PCBM is loaded into the spacing between the nanorods. Figure 4 11. Absorption spectra of CuPc nanorod film before (red dash line ) and after (blue solid line ) infiltration of PCBM. 400 600 800 0.0 0.4 0.8 CuPc nanorods Absorbance (nm) CuPc nanorods/PCBM
103 Optimization of Cupc Nanorods/ PCBM Cell To fabricate CuPc nanorods/PCBM PV device s CuPc nanorod films with 40 60 nm long concentration was spin-coated onto the nanorod films at 1000 rpm. The CuPc/PCBM composite films were then annealed at 90 C for 15 min to remove solvent residual before the deposition of an 8 nm thick bathocuproine (BCP) exciton -blocking layer122 and a 100 nm thick aluminum cathode in high vacuum. Figure 4 12 shows the current density-voltage ( J V ) ch aracteristics under simulated 1 sun AM 1.5 illumination for CuPc nanorods/PCBM devices based on 60 nm long nanorods grown with a stationary or rotational substrate (labeled as NR -S and NR R, respectively). The structure of these nanorod-based devices i s shown in the inset of Figure 412 S hown for Figure 4 12. Current density voltage ( J V ) characteristics of three CuPc/PCBM photovoltaic cells under 1 sun AM 1.5 illumination: a bilayer cell (labeled as Planar) with a 30 nm thick flat CuPc film, two devices with CuPc nanorods grown on ITO with a stationary (NR -S) or rotational (NR R) substrate. The inset sche matically illustrates NR -R device structure -1.0 -0.5 0.0 0.5 1.0 -8 -6 -4 -2 0 2 J ( mA/cm2) 30 nm planar CuPc ~ 60 nm nanorod CuPc (NR-S) ~ 60 nm nanorod CuPc (NR-R)V (V) Al BCP PCBM CuPc ITO
104 comparison are the J V characteristics of a bilayer, planar heterojunction CuPc/PCBM devi ce (labeled as Planar) with a 30 nm thick flat CuPc film and a PCBM layer deposited under the same condition as the nanorod -based devices. The short -circuit current density reaches JSC = (4.4 0.2) mA/cm2 for both NR -S and NR R devices, which is higher than that of the planar heterojunction device, JSC = (3.4 0.2) mA/cm2. On the other hand, all three devices show the same open circuit voltage of VOC = 0.57 V, which is close to the previously reported val ue based on the same D A materials system.123,124 The fill factor (FF) of the NR S device is, howe ver, much lower than that of the NR R device, 0.40 vs. 0.55. This could be attributed to the differences in the morphology of nanorod arrays. The tilted nanorods obtained using stationary substrate mode may present more difficulties for the complete infilt ration with PCBM molecules than the mostly up right nanorods grown with a rotational substrate as well as a less direct path for charge transport to respective electrodes. Nevertheless, the power conversion efficiency reaches PCE = (0.95 0.05) % in the N R S device and (1.4 0.1) % in the NR -R device, both of which are higher than PCE = (0.85 0.05) % in the bilayer CuPc/PCBM device. Planar Plus Nanorods Cupc/ PCBM Cell To further improve the efficiency of the nanorod based devices, a planar CuPc layer ca n be inserted between the nanorods and the ITO to further enhance the hole transport across the film and also provide additional absorption of the incident photons. Similar structure composed of planar -mixed heterojunction was also applied in the CuPc/C60 PV cell by Xue et. al ., and enhanced power conversion efficiency is observed.45 Figure 4 13a shows the J V characteristics under simulated 1 sun AM 1.5 illumination for three NR -R devices with 8 nm planar CuPc inserted between nanorod and ITO (labeled as Planar NR). It can be seen that by varying the nanorod height from 40 to 80 nm JSC
105 Figure 4 13. Curren t density voltage ( J V ) characteristics of three planar plus nanorod CuPc/PCBM photovoltaic cells (label as Planar NR) under 1 sun AM 1.5 illumination: (a) 8 nm thick planar layer plus 40, 60 and 80 nm height nanorod, (b) 5, 10 ,20 nm thick planar pl us 40 nm height nanorod. Inset: Schematic of Planar NR device structure. decreases from (5.0 0.2) mA/cm2 for the 40 nm long planar NR device to (4.3 0.2) mA/cm2 for the 80 nm long planar NR device. With similar VOC for the three device s maximum p = ( 1.7 0.1) % is achieved in the 40 nm long planar NR device. This observation suggests that the -1.0 -0.5 0.0 0.5 1.0 -8 -6 -4 -2 0 2 ~ 40 nm ~ 60 nm ~ 80 nm Nanorod Height: J ( mA/cm2)V (V) -1.0 -0.5 0.0 0.5 1.0 -8 -6 -4 -2 0 2 5 nm 10 nm 20 nm Planar CuPc thickness:J ( mA/cm2)V (V) a) b) Al BCP PCBM CuPc ITO
106 additional planar layer can help improve the device efficiency and PCBM infiltration is better with 40 nm long nanorods. Further optimization of the plana r NR d evice involved varying the planar CuPc layer thickness with fixed 40 nm long nanorods on top. The J V characteristics of planar NR devices with 5, 10 and 20 nm planar CuPc under simulated 1 sun AM 1.5 il lumination are shown in Figure 4 13b The three devices exhibit similar VO C = 0.6 V, whereas the JSC increases with increased planar layer thickness, exhibiting JSC = (4.7 0.2) mA/cm2 with 5 nm thick planar layer and JSC = (5.6 0.2) mA/cm2 with a 20 nm thick layer With a high FF of 0.53, power conversion efficiency of PCE = (1.8 0.1) % is achieved in the planar NR device using a combination of 20 nm thick planar layer and 40 nm long CuPc nanorods as donor layer. Figure 4 14 shows the PV parameters o f planar NR device as a function of i ncident irradiance ( Pin) along with those of NR S, NR -R and planar heterojunction devices. It is found that all four devices have almost constant responsivity ( JSC/Pin) over the entire illumination range though the value drop slightly higher intensities region (> 40 mW/cm2), indicating linear Figure 4 14. Photovoltaic performance versus incident irradiance ( Pin) of Planar NR, NR R, NR S and Planar device. PCE : p ower conversion efficiency FF: fill factor Jsc/Pin: responsivity and Voc : open circuit voltage. 0.5 1.0 1.5 Pin ( mW/cm2 ) PCE (%)0.4 0.5 0.6 FF 1 10 100 0.3 0.4 0.5 0.6 Voc (V)Pin ( mW/cm2 )0.04 0.06 Jsc/Pin ( A/W ) Planar-NR NR-R NR-S Planar
107 dependence of JSC on Pin. The two nanorod devices show the same level of responsivity, which is higher than that of the planar heterojunction device but lower than that of the planar NR device. The VOC increases logarithmically with Pin for all devices, which follows the trend of conventional p-n junction solar cell s All four devices show lower FF at higher intensity (> 20 mW/cm2) and the low FF for NR S device is due to poor hole transport. The maximum PCE = (1.80 0.1) % is achieved at 1 sun intensity of the planar NR device, which is approximately a two -fold enhancement compared with the planar heterojunction device, PCE = (0.85 0.05) %. Table 4 1 compare s some photovoltaic parameters of the Planar, NR S, NR -R and Planar -NR device under 1 sun illumination. Table 4 1. Comparison of the open-circuit voltage ( VOC), short -circuit current density ( JSC), fill factor (FF), and power conversion efficiency (PCE) of Planar, NR -S, NR R and Planar NR devices under 1 sun illumination. Device VO C (V) JSC (mA/cm2) FF PCE (%) Planar 0.57 3.4 0.2 0.47 0.85 0.05 NR S 0.57 4.4 0.2 0.40 0.95 0.05 N R R 0.57 4.4 0.2 0.55 1.4 0.1 Planar NR 0.60 5.6 0.3 0.53 1.8 0.1 Discussion In this work, CuPc nanorods with 20 50 nm in diameter are successfully grown on top of different substrate s using the oblique angle deposition method. In order to promote the growth of one -dimensional nanorod s under OAD condition, it is important to maintain a large incident angle. It can be seen from Figure 4 5 that as the incident angle becomes smaller the trend of one dimensional growth becomes less prominent, leading to shorter nanorods and reduced surface
108 roughness. However, a large incident angle will increase the spacing between the nanorods due to a strong er shadowing effect. For application of CuPc nanorod arrays in OPV device s it is desired that not only the diameter of the nanorod is comparable to the exciton diffusion length of CuPc but also that the spacing between the nanorods is close to the exciton diffusion length of PCBM. Therefore, nanorod arrays with appropriate spacing can be achieved by tuning the incident angle of molecular flux In the current study, the diameter of the CuPc nanorods achieved using stationary substrate mode is 20 50 nm, while it reaches 40 70 nm using rotational substrate mode. Therefore the nanorod diamete r obtained here is still too large to generate efficient exciton diffusion. To further reduce the nanorod diameter, it is crucial to limit the surface diffusivity of the absorbed molecules. As indicated in the nanorod morphology evolution study, the onset of twodimensional growth is accompanied by initial three -dimensional growth of the existing nuclei due to the diffusion of absorbed molecule s from the top of the n u clei to the bottom (see Figure 4 4 b and c). Since the diameter of the nanorod grown afterward is determined by the size of the pr e -formed nuclei, it is essential to reduce or prevent the initial three -dimensional growth of existing nuclei by limiting the s urface diffusivity of molecules. Here, deposi tion rate as high as ~ 10 /s was used to suppress the molecule surface diffusivity. Although denser nanorod arrays with increased nanorod length was obtained using h igh deposition rate (see Figure 4 8 ), there w a s no significant change in the nanorod diameter. This suggests increasing deposition rate alone may not be sufficiently suppress the molecular surface diffusivity, and lower substrate temperature may be desired as an alternative. Comparison between the pe rformance of NR S and NR -R OPV devices suggest that infiltration of PCBM is vital to achieve high efficiency in interdigitated BHJ. The up -right orientation of CuPc nanorods in NR -R device results in better
109 PCBM infiltration, therefore, higher fill factor and power conversion efficiency can be obtained. Even though larger nanorod (40 70 nm) is pres ent in the NR -R device, the better PCBM infiltration lead s to better performance than the NR -S device, where smaller but tilted nanorod s are used. Based on this investigation, it is believed that further enhancement in the CuPc nanorod/ PCBM OPV cell efficiency requires smaller up -right oriented CuPc nanorod s This can be achieved by using OAD with rotational substrate mode, while at the same time keeping the substrate at low temperature to limit the surface diffusivity of molecule. Additionally, planar CuPc and PCBM layer s can be used to sandwich the interdigitated BHJ layer to further improve device efficiency. Conclusions In conclusion, we have demonstra ted that aligned polycrystalline CuPc nanorods with diameters as small as 20 nm can be grown using the oblique angle deposition method. While the growth on a stationary substrate lead s to slanted nanorods, grow ing on a rotational substrate yields mostly u p -right, though somewhat larger nanorods An interdigitated bulk heterojunction structure was realized by infiltrating the CuPc nanorod array with solution processed PCBM molecules. A maximum power conversion efficiency of p = (1.8 0.1) % at 1 sun AM1.5 illumination was achieved in such nanorod based device, approximately twice of that of a bilayer CuPc/PCBM device.
110 CHAPTER 5 EFFICIENT DEEP -BLUE PHOSPHORESCENT ORGANIC LIGHT -EMITTING DIODES Introduction Research on organic light -emitting diodes (OLEDs) has gained significant advancement over the last two decades since the first heterojunction OLED report by C. W. Tang.2 Today, OLEDs have found their way in to the flat -panel display (FPD) and lig hting application s However, one of the remaining challenges for OLEDs in widely commercial application is the efficiency and stability of blue -emitting OLEDs.125 128 Phosphorescent OLEDs, which utilize triplet exciton s to emit light, have been widely studied to enhance the OLED efficiency.5 And PHOLEDs with nearly 100% internal quantum efficiency have been d emonstrated in green emitting device.129,130 For blue PHOLEDs, metal organic complexes of iridium (III) bis[(4,6 difluorophenyl) -pyridinoto N,C2]picolinate (FIrpic) and iridium (III) bis(4,6difluorophenylpyridinato)tetrakis(1 -pyrazolyl)borate (FIr6) are two blue phosphoresc ent emitters that widely employed.131 134 The FIrpic -based devi ce exhibits an emission spectrum with two vibronic peaks at 475 and 500 nm and Commission Interntionale de LEclairage (CIE) coordinates of approximately (0.17, 0.34).134,135 Due to the greenish -blue color, FIrpic -based device is not an ideal blue source for display and lighting application. FIr6 -based device, however, exhibits deeper blue color emission than FIrpic -based device with two major vibronic peaks at 458 and 489 nm and CIE coordinates of (0.16, 0.26).132,136 FIr6 has a higher triplet energy ( T1) of 2.72 eV, than that of FIrpic ( T1 = 2.65 eV) which in turn produce deeper blue color .131,136 Therefore, a host material with an even higher tr iplet energy is needed in order to ef fectively confine the exciton to the phosphorescent emitter. In a previous work, Holmes et al. showed that direct charge trapping by the FIr6 could be an effective way to improve the device efficiency, which avoids the energy loss during host -guest exothermic
111 energy transfer commonly found in FIrpic -based device.132 In their work, a wide bandgap host layer of p bis(triphenylsilyly)benzene (UGH2)136,137 doped with FIr6 was employed as the emissive layer (EML). In addition a 15 nm thick layer of N,N -dicarbazolyl 3 ,5 -benzene (mCP) is inserted between the hole transport layer (HTL) of bis[N (1 naphthyl) -N -phenyl amino]biphenyl (NPD) and EML as the exciton blocking layer (XBL) to confine the exciton as well as to facilitate the hole injection into the EML. On the cath ode side, a layer of bathocuporine (BCP) is used as electron transport layer (ETL) and hole blocking layer (HBL). The device achieved a maximum external quantum efficiency (EQE) of EQE = 12 % and power efficiency of P = 14 lm/W. To further enhance the efficiency of FIr6 -based PHOLEDs, there are still a number of issues to be addressed. In the current FIr6-based device, a 25 nm thick EML is needed to achieve effici ent charge trapping for exciton generation, whereas ano ther 15 nm XB L is required to effective ly blocking the exciton. It is known that the UGH2 and mCP have relativel y low charge carrier mobility a nd such thick layers will inevitably increase the overall driving voltage and reduce the power efficiency of the devic e, which is an important factor in determining the practical application of OLEDs. In this chapter, we focus on improving the performance of the FIr6 PHOLED based on the existing structure, where new materials and device architecture s are explored. In the first se ction the effect of several new electron and hole blocking materials on improving the device efficiency are compared. A direct proof of efficient electron blocking is shown by comparing the emission spectra of two controlled devices. The second section in troduces two major ways to lower the overall driving voltage of the blue PHOLED, which includes optimization of the active layer thickness and using p / n -doped charge injection layer to achieve p -i -n structure. The third section
112 is a discussion of the results in this chapter and how to further improve the blue OLED performance. Finally, the conclusi ons will be given in the last section Charge Carrier and Exciton Blocking in Phosphorescent OLED As shown in the energy level diagram of the FIr6 device (Fig ure 5 1 ), the electrons are injected from the LiF/Al anode into the lowest unoccupied molecular orbital (LUMO) of BCP, followed by injection into the FIr6 LUMO. Similarly, the holes are injected into highest occupied molecular orbital (HOMO) of FIr6 through N PD and mCP layer. The electrons and holes, which are trapped by the FIr6 phosphorescent molecules, will recombine to emit photons. Nevertheless, it is found that part of the electrons injected into the EML can overcome the energy barrier at mCP/EML interfa ce and are injected into the NPD layer when mCP layer is thinner than 15 nm. This suggests that the 0.4 eV barrier at the mCP/EML interface cannot efficiently confine the electron with the EML. On the other hand, the triplet exciton formed in the EML will diffuse to the BCP layer due to slightly lower triplet energy of BCP ( T1 = 2.5 eV)138 than FIr6. Such an exciton loss mechanism can also happen between FIr6 and NPD ( T1 = 2.29 eV)139 when the mCP layer thickness is lower than the exciton diffusion length. Effect of Electron and Exction Blocking Layer Here, we investigate the charge carrier and exciton blocking effect of a new hole transport material 1,1 -bis[(di 4 tolyamino)phenyl]cyclohexane (TAPC). It is found that a device using TAPC as the ETL shows significantly enhanced performance over the NPD -based device, which could be due to higher triplet en ergy ( T1 = 2.87 eV) yet comparable electron mobility of TAPC compared with NPD.140142 Fig ure 5 1 show s the molecular structure of active ma terials involved in the current FIr6 based device as well as their energy level diagram and triplet energy. The blue PHOLEDs were fabricated on glass substrates commercially precoated with a layer a layer of indium tin oxide
113 Figure 5 1. Schematic energy level diagram of the deep -blue phosphorescent organic light emitting diodes (PHOLEDs) and molecular structure of the materials involved in the devices. acetone, and isopropanol consecutively for 45 min, and then treated in ultraviolet ozone for 15min immediately before loading into a high vacuum deposition chamber (background pressure ~ 3 107 Torr). Deposition of all materials, including the Al cathode, was co nducted in succession w ithout breaking the vacuum. F irst, devic es using NPD and TAPC as HTL were compared. The 40 nm thick HTL consists of TAPC or NPD, followed by 15 nm mCP. The 25 nm EML consisted of FIr6 doped into UGH 2 with 10 wt % concentration. 40 nm thick BCP was deposited as the electron transport and hole blocking layer. TAPC, mCP, UGH2 and FIr6 were purchased from Luminescence Technology Corp., BCP from TCI America, and NPD from e Ray. All organic materials were used as obtained without further pu rification. The cathode consisting of 0.5 nm thick layer of LiF followed by a 50 nm thick Al was deposited through an 2.0 2.4 5.5 ITO TAPC NPD 2.4 5.9 mCP (x nm) 2.8 3.1 6.1 7.2 UGH2 FIr6 (25 nm) 3.0 6.5 BCP (40 nm) 4.1 LiF /Al 4.7 (40 nm)FIr6 BCP UGH2 mCP TAPC NPD
114 Figure 5 2. Electroluminescence (EL) spectra of two PHOLEDs with the structures of ITO/NPD or TAPC (40 nm)/mCP (15 nm)/UGH2:10 wt % FIr6 (25 nm)/BCP (40 nm)/LiF (1 nm) /Al in situ shadow mask, forming a 4 mm2 active device area. Current -densityluminance voltage ( J L V ) characteristics were measured using an Agilent 4155C semiconductor parameter analyzer with a calibrated Newport silicon diode. The electroluminescence (EL) spectrum and CIE coordinates of the device were collected through a high sensitivity Ocean Optics spectrometer. The luminance was calibrated using a Minolta luminance meter (LS 100). The external quantum efficien cy ( EQE) was determined by assuming Lambertian emission and following the reported method.60 The EL spectra of the NPD and TAPC -based devices at J = 3 mA/cm2 are shown in Figure 5 2. The two devices show very close CIE coordinates, (0.16, 0.28) for th e NPD device and (0.16, 0.27) for the TAPC device. However, there is a smaller shoulder peak appeared in EL 300 400 500 600 700 800 0.0 0.2 0.4 0.6 0.8 1.0 EL Intensity (a.u.) (nm) NPD TAPC NPD
115 Figure 5 3. Performance of PHOLEDs with either NPD or TA PC as hole transport layer (HTL). (a) Current density voltage ( J V closed symbols) and luminance-voltage ( L V open symbols) characteristics and (b) Luminance power ( P, closed symbols) and external quantum ( EQE, open symbols) efficiencies as a function of current density layer becomes thinner than 15 nm. This has been attributed to the emission from NPD when a portion of electrons overcome the barrier at the mCP/EML interaface and enter into NPD, where they recombine with holes injected from th e ITO and form exciton s However, no TAPC driving voltage or with a much thinner 10-310-210-11001010 5 10 15 EQE (%)J (mA/cm2)P (lm/W) TAPC NPD0 5 10 15 20 25 0 2 4 6 8 10 10-610-510-410-310-210-1100101 J (mA/cm2) TAPC NPD V (V)10-210-1100101102103104 L ( cd/m2) a) b)
116 mCP layer. Figure 5 3 a compares the J L V characteristics of the NPD and T AP C -based devices. It can be seen that the TAP C -based device show s a slightly lower current density at V > 5 V, but the difference in the L -V characteristics is small The current -density dependencies of the external quantum and power efficiencies of these t wo devices are shown in Figure 5 3 b. A peak external quantum efficiency of EQE = (12 1) % and a peak power efficiency of P = (14 1) lm/W are obtained in the NPD device, which are very similar to the previous report values. For the TAPC device, on the other hand, EQE reaches a maximum of (18 1) %. The peak power efficiency of P = (16 1) lm/W, which is nearly 15 % higher than the maximum of the NPD based device, is achieved at a luminance of L = 4 cd/m2. Moreover, the TAPC device shows a P = (13 1) lm/W at L =100 cd/m2 and P = (8.9 0.6) lm/W at L =800 cd/m2, approximately 60% higher than the NPD device at these two luminances (7.9 and 5.6 lm/W, respectively). We believe the significant improvement in the TAPC device is due to better electron and exciton blocking provided by the TAPC. The electron affinity of TAPC is 0.4 eV smaller than that of mCP (2.0 eV vs 2.4 eV), i.e., the LUMO level of TAPC is 0.4 eV higher than that of mCP.132,140 Hence, a TAPC layer should provide an additional barrier height for preventing electrons injected from cathode, which is not available in NPD. Moreover, the higher triplet energy of TAPC should also enable us to reduce the mCP layer thickness without compromising t he exciton confinement in the EML. Figure 5 4 shows the performance of TAPC -based devices with the mCP layer thickness varied from 0 to 15 nm. Here, the EML layer thickness is reduced to 20 nm in an effort to reduce the overall driving voltage. It can be s een that with same 15 nm thick mCP layer, reducing the EML thickness from 25 nm to 20 nm results in a decrease of maximum EQE from (18 1) % to (15 1) %. This is due to incomplete exciton formation in the
117 Figure 5 4. Luminance power ( P, closed symbols) and external quantum ( EQE, open symbols) efficiency versus current density ( J ) for TAPC -based devices with a varying mCP layer thickness. thinner EML. However, the peak P is slightly increased from (16 1) to (18 1) lm/W because of the reduct ion in device driving voltage With the EML thickness fixed at 20 nm, as the mCP layer thickness is reduced from 15 to 10 nm, EQE is almost unaffected, and the maximum P remains at (18 1) lm/W. However, P shows slight improvement at high current densi ties due to the lower driving voltage with reduced mCP layer thickness. Further reducing the mCP layer thickness to 5 nm or lower will lead to a significant reduction in EQE. The mCP plays an important role in facilitating the injection of hole s into the EML, as its highest occupied molecular orbital (HOMO) level sits between those of TAPC and FIr6.136,143 Without the mCP layer, electrons and holes would accumulate at the TAPC/EML interface due to the presence of larg e energy offset, resulting in a very thin recombination zone close to this interface. Therefore, polaron quenching of exciton and exciplex formation at the interface could significantly lower the quantum efficiency.144147 The device with 5 nm thick mCP exhibits a much faster efficiency 10-310-210-11001010 5 10 15 20 25 EQE ( % ) P ( lm/W ) 0 nm mCP 5 nm 10 nm 15 nmJ ( mA/cm2) 0 5 10 15
118 roll -off at high current densities compared to other devices with a thicker mCP layer although they show almost identical efficiency at low current densities ( J ~ 103 mA/cm2). Since the surface roughness of ITO substrate is typically a few nanometers,148 the 5 nm thick mCP may not form complete coverage over the whole device area, leading to efficiency reduction in spots where the mCP layer thickness ap proaches zero. Evidence for Effective Electron and Exciton Blocking To confirm the improvement on device efficiency is due to better electron and exciton confinement properties of TAPC over NPD, we compared NPD and TAPC -based devices without the mCP l aye r. Inset of Fig ure 5 5 shows the plot of EQE as a function of the current density for the two devices. It can be seen that the EQE of the TAPC is approximately 5 times higher than that of the NPD device. In addition, as shown in Figure 5 5 strong NPD fl uorescent emission ( = 435 nm) can be observed in the EL spectrum NPD device, while no TAPC Figure 5 5. Electroluminescence (EL) spectra of TAPC based and NPD -based PHOLEDs without mCP layer. Inset: external quantum efficiency vs current density of the two devices. 300 400 500 600 700 800 0.0 0.2 0.4 0.6 0.8 1.0 NPD Fluo. EL intensity (a.u.) (nm) TAPC NPD 10-310-210-11001010.1 1 10 TAPC NPDEQE (%)J (mA/cm2)
119 fluorescence ( = 360 nm) is found.149 This indicates that electron leakage into the HTL is rather significant in the NPD device, but not present in the TAPC device. In the current investigation, the re is no discernable triplet emission from NPD ( = 541 nm) or TAPC ( = 430 nm).139 Therefore, it is difficult to directly compare the exciton b locking properties of NPD and TAPC. However, it has been demonstrated previously that the higher triplet energy of TAPC than NPD can enhance the efficiency of green PHOLEDs because of better exciton confinement.139 Effect of Hole and Exciton Blocking Layer So far, th e optimized TAPC device composed of 10 nm mCP and 20 nm EML, which has EQE of (15 1) % and P of (18 1) lm/W. Since TAPC alread y provide s good electron and exciton confinement, further enhancement of the device efficiency may be achieved by improved hole and exciton blocking at the EML/ETL int erface. Currently, the BCP has T1 = 2.5 eV, which is slightly lower than that of FIr6. Moreover, the energy barrier of hole s is only 0.4 eV at the EML/BCP interface. Both factors may result in inefficient hole and exciton blocking at the EML/BCP interface. Therefore, a thin layer of UGH2 and 1, 3 -bis[(4 tert -butylphenyl) 1,3,4 oxadiazolyl] phenylene (OXD 7)150 is inserted between EML and BCP layer in order to further improve the hole and exciton blocking within the EML. Figure 5 6 a shows the L J V characteristics of devices with 5 nm UGH2 or OXD 7 inserted between EML and BCP layer. It can be seen that the UGH2 device has better current injection an d luminance than the OXD 7 device, reaching J = ~ 5 mA/cm2, L = ~ 1000 cd/m2 at 10 V for UGH2 device but only J = ~ 3 mA/cm2 and L = ~ 700 cd/m2 for OXD 7 device. It is kno wn that OXD 7 has better electron transport than UGH2.151 Therefore, the OXD 7 device should show lower driving voltage than the UGH2 device. The presence of high driving voltage in the current OXD 7 device could possibly due to the poor electron injection at the additional EML/OXD 7 and OXD 7/BCP interfaces, which are not present in the UGH2 device. The
120 dependencies of external quantum and power efficiency on current densi ty for OXD 7 and UGH2 devices are shown in Figure 5 6b. It can be seen that both devices show higher maximum EQE and P compared with the optimized TAPC device without any hole blocking layer. The Figure 5 6. Performance of PHOLEDs with ITO/TAPC (40 nm) /mCP (10 nm)/UGH2:10 wt % FIr6 (20 nm)/HBL (5nm)/BCP (40 nm)/LiF (1 nm) /Al (a) Current density -voltage ( J V closed symbols) and luminance -voltage ( L V open symbols) characteristics and (b) Luminance power ( P, closed symbols) and external quantum ( EQE, open symbols) efficiencies as a function of current density. 2 4 6 8 10 0 2 4 6 8 V (V)J (mA/cm2) no HBL UGH2 HBL OXD-7 HBL10-1100101102103 L ( cd/m2) 10-310-210-11001015 10 15 20 25 30 35 J (mA/cm2)P (lm/W)0 5 10 15 20 no HBL UGH2 HBL OXD-7 HBL EQE (%) a) b)
121 OXD 7 device has peak EQE of (15 1) % and P of (21 1) lm/W, and the UGH2 device show peak EQE and P of (20 1) % and (25 1) lm/W, respectively. This suggests the inserted OXD 7 and UGH2 layer, with their higher triplet energy, do help improve the device efficiency by inducing better exciton blocking. Moreover, due to the higher HOMO level of UGH2 with respect to the OXD 7 and BCP (both have the same HOMO level), the addition hole blocking effect could be expected from a neat UGH2 layer, which contributes the highest peak efficiency achieved. However, it is worth noticing that both OXD 7 and UGH2 devices have very fast efficiency roll -off at high current densities, which is so significant that the efficiency become s lower than the optimized TAPC device. This suggests that the charge injection become more unbalanced in the high voltage regime.152 Considering an additional OXD 7 or UGH2 layer will reduce the electron transport, the EML could become hole dominant at high voltage. Since charge balance also plays an important role in determining the device efficiency, the high triplet energy of the ETL should be the only key factor affecting the device efficiency. Reducing Driving V oltage of Phosphorescent OLED In the last section we showed that charge carrier and exciton confinement is of great importance to obtain blue PHOLEDs. And significant improvement of external quantum efficie ncy has been achieved using TAPC as HTL and UGH2 as exciton blocking layer. Nevertheless, one of the key parameter s to determining the practical application of OLED s is luminance power efficiency, which is the light output of OLED with unit energy input, o r say, lumens per watt (lm/W). Therefore, the work in this section focuses on reducing the driving voltage of the blue PHOLEDs by optimizing the active layer thickness and using efficient charge injection layer.
122 Efficient Electron and Hole Injection Layer One way to effectively reduce driving voltage of the OLEDs is using p and n doped charge injection/transport layer s It is believed that the doped organic layer will have higher intrinsic carrier density than the undoped layer. On one hand, such increa sed carrier density can improve the conductivity of the film. On the other hand, the charge injection barrier between the electrode and the organic layer can be reduced due to band bending at the metal -organic interface, therefore, forming ohmic contact. I n the current study, N,N -diphenyl N,N bis(3 methylphenyl) -[1,1 -biphenyl] 4,4 -diamine (MeO TPD) doped with tetrafluorotetracyanoquinodimethane (F4TCNQ) is used as a p doped hole injection layer, whereas Li doped 4,7 diphenyl 1,10 phenanthroline (BPhen) layer is used as n -doped electron injection layer. The improved conductivity of the F4TCNQ doped MeO TPD originates from the efficient electron transfer from MeO TPD to F4TCNQ molecules due to the strong electron affinity of F4TCNQ. As a result, additional holes are created in the MeO TPD layer, increasing the conductivity of the film. Figure 57a shows the J V characteristics of 250 nm F4 TCNQ doped MeO TPD films sa ndwiched between gold electrodes, which generate hole only devices. It can be seen that the current density increased as the F4TCNQ dopant concentration increases from 0 mol % to 9.0 mol %. The conductivity of the MeO TPD layer is obtained by fitting the ohmic region, i.e., of the J V curve. As shown in Figure 57b, the conductivity is improved from ~ 1010 Scm1 in the undoped MeO TPD film to more than 105 Scm1 in the 9.0 mol % doped sample, which is five order of magnitude higher. Therefore, t he voltage drop across the doped MeO TPD will be significantly lower. Even though a high conductivity p -doped layer is successfully obtained through doping MeO TPD, it is still unclear how it works on the current blue PHOLEDs. Therefore, two hole
123 Figur e 5 7. Conductivity study of F4 TCNQ doped MeO TPD films (a) Current density -voltage (J V ) characteristics of doped and undoped MeO TPD films. The dots are measured data point, and the lines are the linear fit of the ohmic regime. (b) Conductivity of the doped MeO TPD films as a function of the F4 TCNQ doping concentration. (Courtesy of T Tseng and J Xue) only devices are fabricated to investigate the hole injection and transport in the FIr6 device. The two hole only devices involve structure s of ITO/TAPC(20 nm) or MeO TPD:F4 TCNQ (3 mol %, 20 nm)/TAPC (20 nm)/UGH2:FIr6 (10 wt %, 20 nm)/Au (50 nm). Here, BCP is removed to make sure holes can be fully collected at the cathode and Au is used as cathode to reduce electron inject ion under forward bias. Figure 5 8 compares the J V characteristics of the two hole -only devices under forward bias. It can be seen that the device using 20 nm doped MeO TPD as the hole injection layer shows significantly higher current density than the TAPC only device. T he current density reaches J = ~ 50 mA/cm2 at 10 V for the doped MeO TPD device, whereas J = ~ 0.2 mA/cm2 for the TAPC at same voltage. This suggests that hole injection and transport in the doped MeO TPD is more efficient than in the TAPC only device. Her e, only 3 mol % F4 TCNQ doping concentrati on is used in the p -doped layer because further increasing the dopant concentration will not significantly enhance the hole current further. Moreover, high dopant concentration will induce strong exciton quenching when the dopants manage to diffuse into the EML with thinner TAPC and mCP layer. 10-210-110010-810-610-410-2100 9.0 mol % 4.9 mol % 2.4 mol %1.3 mol % 0.6 mol % J (A/cm2)V (V) Undoped 1 10 10-1110-910-710-510-3 Conductivity (S/cm)F4-TCNQ Doping Concentration (mol %) Undoped MeO-TPD a) b)
124 Figure 5 8. Current density-voltage ( J V ) characteristics of ITO/TAPC (red) or MeO TPD: 3 mol % F4 TCNQ (blue) (20nm)/TAPC (20 nm)/UGH2: 10 wt% FIr6 (20 nm)/Au (50 nm) ho le -only device. Inset: same plot in linear scale. The n -doped electron injection and transport layer composed of Li doped BPhen layer, has been successfully used on different OLEDs to reduce driving voltage of the device. It has been shown that the extremely active Li atoms doped into the BPhen will donate electrons to BPhen molecules. As a result, the intrinsic carrier density in BPhen is increased significantly, and high conductivity could be obtai ned. It has been shown that the conductivity of intrinsic BPhen is approximately 1010 Scm1, which is increased by 4 order s of magnitude to ~ 106 Scm1 in the 1:1(by molar ratio) Li doped BPhen layer.153 Therefore, similar to F4TCNQ doped MeO TPD, improvement on electron injection and transport is expected in the device using Li doped BPhen layer. To confirm our hypothesis, two electron-only devices wi thout electron blocking mCP and nm)/Bphen (20 nm)/BPhen:Li (1:1 by molar, 20 nm)/LiF/Al and ITO/ UGH2:FIr6 (10 wt %, 20 nm)/BCP (40 nm)/LiF/Al. Here, the ITO substrate is directly used without UV -ozone treatment, 0.1 1 10 10-510-410-310-210-1100101102103 V (V)J (mA/cm2) F4-TCNQ:MeO-TPD HIL/HTL TAPC HTL 0 5 10 0 50 V (V)J (mA/cm2)
125 Figure 5 9. Current density-voltage ( J V ) characteristics of ITO/UGH2: 10 wt % FIr6 (20 nm )/BCP (40 nm)/LiF/Al (red) and ITO/UGH2: 10 wt% FIr6 (20 nm)/Bphen (20 nm)/Bphen:Li =1:1 (20 nm)/LiF/Al (blue) electron-only device. Inset: same plot in linear scale. which result s in a lower work function of ITO to reduce hole injection from ITO under forward bias. Moreover, a 20 nm thick neat BPhen layer is inserted between the UGH2 and n-doped Bphen layer to prevent diffusion of Li cations into the EML. It is believed that the Li cation can be the quenching center for the triplet exciton, which reduces the luminance efficiency.154,155 Figure 5 9 shows the J V characteristics of the two electron -only devices under forward bias. The device with Li doped BPhen layer show higher current density than the BCP device at the same voltage. For instance, J = ~ 15 mA/cm2 at 10 V for the BCP device, whereas J increases to more than 100 mA/cm2 for the n -doped device at the same voltage. This suggests more efficient ele ctron injection and transport is achieved in the device using Li doped BPhen layer as n -doped layer. Another important observation by comparing the two sets of single carrier devices is that 0.1 1 10 10-510-410-310-210-1100101102103 V (V)J (mA/cm2) Bphen:Li EIL/ETL BCP ETL 0 5 10 0 50 100 150 V (V)J (mA/cm2)
126 the electron and hole current is more balanced under high driving voltage in the devices with doping layer s It can be seen from Figure 5 8 and Figure 5 9 that J = ~ 50 mA/cm2 and ~120 mA/cm2 at 10 V for the p-doped hole -only and n-doped electron-only devices, respectively, whereas J = ~ 0.2 mA/cm2 and = ~ 15 mA/cm2 for the undoped hole -only and electron -onl y devices. The current density difference in terms of percentage is much less after using doped charge injection layer s This suggests more balanced charge injection into the EML is achieved; hence improved device eff iciency is anticipated. Device Using p -i -n Structure Based on the previous understanding on the performance of pand n -doped charge injection/transport layers, it is advisable to incorporate the two efficient dop ing layers into the blue PHOLED structure to reduce driving voltage and improve luminance power efficiency. The so -called p -i -n structure typically is composed of an intrinsic EML sandwiched between a pdoped hole injection/transport layer and an n -doped electron compliment In this Section, the performance of p -i -n blue PHOLEDs using F4 TCNQ doped MeO TPD as a p -doped layer and Li doped BPhen as an n -doped layer is investigated. Here, a p -i -n structure based on the conventional NPD device studied previously is fabricated. F igure 5 10 shows the energy level diagr am of the p -i n devices as well as the device structure under investigation. The conventional NPD device composed of 40 nm NPD, 15 nm mCP, 25 nm 10 wt % FIr6 doped UGH2 as EML and 40 nm BCP, which is reported as the o ptimized conventional NPD device. In the NPD p -i -n device, the 40 nm NPD is replaced with 25 nm thick 2 mol% F4TCNQ doped MeO TPD and 15 nm pure NPD, whereas the 40 nm BCP is replaced with 20 nm pure BPhen and 20 nm thick Li doped BPhen (1:1 molar mixing ratio).
127 Figure 5 10. Device structure of the conventional NPD -based deep -blue PHOLEDs and the two p -i -n structure PHOLEDs under study as well as the energy level diagram of p -i -n PHOLEDs. Figure 5 11a shows the J L V characteristics of conventional NP D device and the p -i -n NPD device. It can be seen that, with same mCP and EML layer, the p -i -n device has higher current density and luminance than the conventional device. At 10 V the current density and luminance reach J = ~ 50 mA/cm2 and L = ~ 4300 cd/m2 for the p -i -n device, whereas J = 15 mA/cm2, L = ~ 2000 cd/m2 for the conventional device. Moreover, the p -i -n structure has significantly reduced the driving voltage of the d evice. The driving voltage at L = 1000 cd/m2 is reduced from V = 10 V in the co nventional device to V = 7.5 V in the p -i -n device. This has led to enhanced luminance power efficiency in the p -i -n device. Figure 5 11b shows the plot of external quantum and luminance power efficiency as a function of luminance. It should be noted ITO substrate NPD (40 nm) UGH2:10 wt% FIr6 (25 nm) mCP (15 nm) BCP (40 nm) LiF (1 nm) Al ITO substrate NPD (15 nm) UGH2:10 wt% FIr6 (25 nm) mCP (15 nm) Bphen (20 nm) LiF (1 nm) Al Bphen:Li (1:1 molar) (20 nm) MeO TPD:F4 TCNQ (2 mol%) (25 nm) ITO substrate NPD (15 nm) UGH2:15 wt% FIr6 (30 nm) mCP (15 nm) Bphen (20 nm) LiF (1 nm) Al Bphen:Li (1:1 molar) (20 nm) MeO TPD:F4 TCNQ (2 mol%) (25 nm) MeO -TPD F4-TCNQ NPD mCP UGH2 FIr6 Bphen Bphen Li 4.1 LiF/Al ITO 4.7 5.1 1.9 2.4 5.5 2.4 5.9 2.8 3.1 6.1 7.2 3.0 3.0 6.5 6.5
128 Fi gure 5 11. Performance of PHOLEDs of conventional NPD -based device and two p -i -n devices (a) Current density -voltage ( J V closed symbols) and luminance -voltage ( L V open symbols) characteristics and (b) Luminance power ( P, closed symbols) and external quantum ( EQE, open symbols) efficiencies as a function of luminance that the external quantum efficiency of the p -i -n device is slightly lower than the conventional device at certain voltage s with peak EQE = (10 1) % for the p -i -n device and EQE = (12 1) % for the conventional device. Although the peak P of the p -i -n device is still lower than the 0 2 4 6 8 10 10-610-510-410-310-210-1100101102 V (V)J (mA/cm2) Conventional NPD Normal p-i-n Modified p-i-n 10-1100101102103104105106 L ( cd/m2) a ) 10-1100101102103104-5 0 5 10 15 0 5 10 15 20 25 L ( cd/m2) EQE (%)P (lm/W) Conventional NPD Normal p-i-n Modified p-i-n b )
129 conventional device, (12 1) lm/W versus (14 1) lm/W, the p -i -n device shows higher P than the conventional one as L > 4 cd/m2, which is m ore relevant to practical display or lighting applications. In the current p -i -n and conventional devices, a common 25 nm 10 wt % FIr6 doped UGH2 layer is used as the EML. Since significantly more charge carriers are injected into the EML in the p -i -n device, such low FIr6 doping concent ration and thin EML layer may not be sufficient to e ffectively trap the injected carriers and generate triplet exciton s By considering this, we increased the FIr6 doping concentration to 15 wt % as well as the EML laye r thickness to 30 nm. As shown in Figure 511a, with increased FIr6 concentration and EML layer thickness, the modified p -i -n device shows slightly lower current density yet higher luminance compared with the normal p -i -n device, reaching J = 40 mA/cm2 and L = 4700 cd/m2 at 10 V. Moreover, due to the presence of more trapping and emitting centers in the EML, the modified p -i -n device achieves higher external quantum efficiency than the conventional and normal p -i -n device s with peak EQE = (13.0 1.0) %. T he enhancement in power efficiency of modified p -i -n device s over conventional one s is more significant, especially a t high luminances (see Figure 5 11b). Noticeably, both p -i -n devices show very fast roll -off of the efficiency curves at L > 1000 cd/m2. Th is could due to both unbalanced charge injection under high voltage and triplet -triplet quenching at high luminances when the triplet exciton density is too high within the EML. The comparison of the efficiencies at different luminance and CIE coordinates of the three devices is listed in Table 5 1. The p -i -n structure is very effective in improving the power efficiency at the high luminance, which is much more valuable for applications in displays ( L ~ 100 cd/m2) and lighting ( L ~ 1000 cd/m2). And all these are achieved without a significant change of the EL spectra and CIE coordinates of the devices.
130 Table. 5 1. Comparison of the turnon voltage ( VT), external quantum efficiency ( EQE), power efficiency ( P), and CIE coordinates of three devices Device VT (V) EQE (%) (at peak) P (lm/W) (at peak, 100 cd/ m2, 1000 cd/ m 2 ) CIE (x, y) (at 1 mA/c m2) Conventional (NPD) 4.7 12 1 14, 8, 5 1 (0.16, 0.28) Normal p i n 4.6 11 1 12, 9 6 1 (0.17, 0.29) Modified p i n 4.8 1 3 1 14, 12 8 1 (0.16, 0.29 ) Discussion In this chapter, several methods have been succes sfully applied to the existing deep blue PHOLEDs to improve the device efficiency. Due to the nature of the blue emitter, achieving efficie nt blue OLEDs is challenging. For phosphorescent blue OLEDs in particular the high triplet energy phosphorescent emitter requires even higher triplet energy for the host materials as well as the peripheral charge transport materials. Also, the wider bandgap of the blue emitter usually possess lower HOMO and higher LUMO level s which makes the charge injection into the corresponding energy level s more difficult. Here, using TAPC as the HTL instead of NPD can increase the efficiency of the FIr6 PHOLED significantly. The improvement originat es from better exciton and electron blocking provided by TAPC, who has a T1 = 2.87 eV and a LUMO level of 2.0 eV. With TAPC, the thickness of mCP and EML can be reduced without largely sacrificing the external quantum efficiency, hence higher luminance power efficiency is achieved. Similarly, high triplet energy ETL of OXD 7 and UGH2 is used to improve the hole and exciton blocking on the cathode side.152,156 Even though improv ed efficiency is observed at low current density, the efficiency ac tually becomes lower than the device without OXD 7 or UGH2. This is attributed to the poor electron transport after inserting any of these two layers between EML and BCP. To overcome this drawback, a single ETL with high triplet energy and
131 high electron mobility is desired. In the work after, we used electron transport material tris[3 (3 pyridyl)mesityl] borane (3TPYMB) with electron mobility of e = ~ 105 cm2/Vs, triplet energy of T1 = 2.95 eV and HOMO level of 6.77 eV to replace BCP (e = ~ 106 cm2/Vs, T1 = 2.5 eV, HOMO of 6.7 eV) ,157 which led to a peak EQE = (20.0 1.0) %.153 Ther efore, by incorporating high triplet energy materials as the peripheral charge transport layers, further enhancement of the efficiency of current blue PHOLEDs should be achievable. Another route demonstrated to be effective in improving device efficiency involves usage of n and p -doped charge injection layer s The doped charge injection/transport layers are able to achieve lower injection barrier and higher conductivity, which result in reduced driving voltage and higher luminance power efficiency. In the current p -i -n blue PHOLEDs, the F4 TCNQ doped MeO TPD as p -doped layer and Li doped BPhen as n-doped layer not only reduced the driving voltage but also significantly increased the amount of charge injected into the EML. Since FIr6 is the phosphorescent e mitter as well as the charge transport/trapping center, higher FIr6 doping concentration is desired to maximize the exciton generation in the p -i -n device. That is why higher efficiency in the p -i -n device is obtained by increasing the FIr6 concentration i n the EML. The current modified p -i -n device possesses a 30 nm thick 15 wt% FIr6 doped UGH2 as EML, whose thickness can be further reduced to lower the driving voltage In order to reduce the EML layer thickness without compromising the quantum efficiency of the device, it is possible to further increase the FIr6 concentration to 20 wt % or even greater. In the current blue PHOLEDs, electrons can still be injected in to the LUMO level of the mCP even though TAPC is used to improve electron confinement. In the next step of research work conducted majorly by Sang -Hyun Eom, we demonstrated that a TAPC device with dual emissive layers consisting of 4 wt % FIr6 doped mCP and 25 wt % doped UGH2 can further
132 enhance the PHOLED efficiency by utilizing the electron i njected into mCP layer. Moreover, application of p i -n structure on such dual emissive layer device resulted in further improvement of the power efficiency to (25 2) lm/W at 100 cd/m2 and (20 2) lm/W at 1000 cd/m2.158 I n the current p -i -n devices, a layer 20 nm thick of pure BPhen is inserted between the n -doped BPhen layer and EML to prevent the diffusion of Li ions into the EML, which can cause significant luminance quenching. To reduce the pure BPhen layer or even completely remove it requires way to hinder the diffusion of ion dopant into EML. By doping the BPhen with Cs instead of Li the diffusion of ion dopants into the EML should be reduced due to the large size of the Cs ion. In another work we published, we achieved higher efficiency in the p -i -n device using Cs as dopants than that with Li dopants.153 With efficient deep -blue PHOLED, a white PHOLED can be easily achieved by doping certain amount of the green and red phosphorescent emitters into the EML.159 With current efficiency blue PHOLEDs, we achieved a white OLED by introducing fac -tris(phenylpyridine) iridium [Ir(ppy)3] as green emitter and iridium(III) bis (2 -phenylquinoly -N, C2) acetylaceto -nate (PQIr) as red emitter. The while PHOLED achieved a peak power efficiency of (40 2) lm/W, a color rendering index of 79 and CIE of (0.37, 0.40).160 Conclusions We have shown that the efficiencies of the FIr6 -based deep -blue PHOLEDs significantly depend on the proper ties of the peripheral charge injection/transport materials. A maximum EQE = (18 1) % was achieved using TAPC as HTL, which is approximately 50% higher than the device using NPD as HTL. This improvement is attributed to the better electron and exciton c onfinement provided by the TAPC. The maximum EQE is further e nhanced to (20 1) % by introducing a thin neat UGH2 as hole and exciton blocking layer. We also have shown that incorporation of F4TCNQ doped MeO TPD hole injection and Li doped BPhen electron
133 injection layer can substantially reduce device driving voltage The modified p -i -n device achieved a power efficiency of P = (11 1) lm/W at 100 cd/m2, which is nearly 45% higher than the conventional device at the same luminance.
134 CHAPTER 6 LOW TURN ON VOLTAGE OF MEH -PPV POLYMER LIGHT -EMITTING DIODES Introduction One important category of organic light -emitting diodes (OLEDs) is the polymer light emitting diodes (PLEDs), which differ from small molecule OLEDs by the use of conjugated polymers as emissive layer. The research on PLEDs started in 1990 when the first electroluminescence from a polymer was observed in poly( p -phenylene vinylene) (PPV),161 and significant advances have been achieved in PLEDs in the two decades since. Compared with small molecule OLEDs, PLEDs have an advantage of being able to be fabricated through solution processes where the active layers are deposited through spin-casting or ink jet printing from a polymer solution. Therefore, the PLEDs are more suitable for large area low -cost fabrication by eliminating the need for expensive material -costly vacuum deposition. Nevertheless, the current PLEDs suffer a few drawbac ks compared with small molecule OLEDs and one of the i ssues remaining is the relatively high driving voltage As we learned in Chapter 5, a multi layer p -i -n structure was achieved through vacuum thermal evaporation to successfully reduce the driving voltage With s uch a multi -layer structure the variou s fun ctional materials separate the role s of charge injection, transport and light emission, which shows the ve rsatility of the small molecule OLEDs. Achieving such a multi layer structure, however, is rare in PLEDs. This is mainly due to the difficulty of achieving solvent orthogonality, which requires the solvent during the subsequent layer deposition does not dissolve the previously deposited layer. Because of this, most of the PLEDs are composed of poly(ethylenedi oxythiophene):polystryrene sulphonate (PEDOT :PSS) as hole injection layer, a polymer as the emitting layer (EML) and a vacuum deposited small molecule electron injection/transport layer.162164
135 Since most of the light -emitting conjugated polymers are p type semiconductor s balanced injection and transport in the hole -dominant PLEDs requires sufficient electron injection into the EML during device operation. Even though low -work -function cathodes of Ca, Ba and LiF/Al are widely used to reduce the electron injection barrier,165168 their air and water sensitivity significantly reduce device stabili ty and lifetime. Over the past few years, great efforts have been made by different research group s to provide an efficient electron injection cathode with ambie nt stability and ease of fabrication. Wu et. al. reported a bilayer cathode consisting of aluminum and alcohol /water -soluble conjugated polymer, which shows comparable performance to the Ba/Al cathode.168 Cao et. al. showed that addition of an organic surfactant between the polymer EML and Al can improved PLED performance.169 Another way to improve th e electron injection includes inserting a layer of polystyrene sodium sulfonate between the polymer EML and Al.170 In this chapter, we present a novel and efficient electron injection cathode composed of ZnO nanoparticles (NPs) and Al. The poly[2 methoxy5 (2 -ethylhexyloxy) 1, phenylene vinylene] (MEH -PPV) PLED with ZnO N Ps/Al cathode show s remarkably improved electron injection over the conventional LiF/Al cathode. Electroluminescence from MEH -PPV is observed at a driving voltage as low as (1.3 0.1) V. It is believed that the ultra low turn-on voltage achieved by the Zn O NPs/Al cathode could be attribute d to an efficient Auger process that took place at the MEH -PPV/ZnO NPs heterojunction. An electron can gain additional energy during the Auger process and can overcome the injection barrier at the MEH PPV/ZnO NPs interfac e Therefore, electrons can be injected into the MEH -PPV at a much lower driving voltage Such a method to improve electron injection is significantly different from the conventional fashion of seeking low -work -function cathode to reduce injection barrier. The content in this chapter is arranged in the following order. Structure and morphology
136 characterization of the ZnO NP layer as well as device performance of MEH -PPV PLED with ZnO NPs/Al as cathode are presented in the first part. The second part will focus on the mechanism study of the low turn -on voltage phenomenon, which is conducted based on the hypothesis of Auger process at the MEH -PPV/ZnO NPs interface. The discussion and partial conclusion will be given in the last part. Low Turn -on Voltage with ZnO Nanoparticles as Electron Injection Layer To study the mechanism of efficient electron injection, hence the ultra low turnon voltage, it is of great importance to first gain information of the particular structure and morphology of Zn O NPs as well as the device performance of the MEH -PPV PLED with ZnO NPs/Al cathode. Morphology and Structure of ZnO Nanoparticle s The ZnO nanoparticles were synthesized through the sol -gel method. In a typical experience, ZnO NPs were prepared by dropwise addition of a stoi chiometric amount of tetramethy l ammonium hydroxide dissolved in ethanol (concentration 0.55 M) to 30 mL of 0.1 M Zinc acetate (dehydrate) dissolved in dimethyl sulfoxide (DMSO) followed by stirring for an hour. The ZnO nanoparti cles were then precipitated and washed thoroughly using heptanes/ethanol twice to remove the residual chemicals. ZnO NPs were then dispersed in ethanol with a concentration of ~ 30 mg/mL. Here, transmission electron microscopy (TEM) and X ray diffraction (XRD) are employed t o characterize the morphology and structure of ZnO NPs. The TEM measurement is conducted using JOEL TEM 200CX, whereas the XRD pattern is recorded using Philips Xpert MRD diffractometer with Cu mode. Figure 6 1a shows the transmission electron microscopy (TEM) images of ZnO NPs synthesized through the above method. The nearly spherical shape ZnO NPs have diameter ranging from 2.5 nm to 3.5 nm. Figure 6 1b
137 Figure 6 1. Morphology of ZnO nanoparticle s. (a) Transmission electron microscope (TEM) image of ZnO NPs. Inset: high resolution TEM image of same ZnO NPs, where ZnO NPs form peanut like pair -wise particle. (b) Histogram for size distribution of the ca. 50 ZnO NPs. The scale bar is 20 nm. (Courtesy of Lei Qian) shows a histogram of ZnO NP sizes for a population of ~ 50 isolated nanocrystals, which suggests a fairly narrow distribution of the nanoparticle size with an average diameter of 3.2 0.3 nm. The XRD patterns of the ZnO NPs along with that of bu lk ZnO single crystal are shown in Figure 6 2. Bulk ZnO is known to adopt a wurtzite structure with space group of P63mc and lattice constant of a = 3.250 and c = 5.207 .171 Comparison of the XRD patterns of the ZnO NPs with tha t of bulk ZnO suggests that the ZnO NPs are crystalline with similar wurtzite structure and lattice constant to bulk ZnO. However, the diffraction peaks are significantly broadened in the XRD pattern of ZnO patterns, which could due to the reduced size of these nanoparticles. By measuring the full width at half maximum (FWHM) through curve fitting, the size of the ZnO nanoparicle can be obtained through the Scherrer formula:172 = cos 2 3 4 5 0 10 20 30 Percentage (%)Diameter (nm) 3.2 0.3 nma) b)
138 Figure 6 2. X -ray diffraction (XRD) pattern of ZnO NPs (red) along with that of bulk ZnO (blue) Here, d is the diameter of the nanoparticle, is the x -ray wavelength, which is 1.54 for Cusource, B K is a shape factor, which equals 1.2 assuming a spherical shape.171 The calculation using the above formula suggests an average particle size of 3.1 nm, which matches very well with the value obtained through TEM study. LiF versus ZnO Nanopar ticle as Electron Injection Layer The performance of two typical MEH -PPV PLEDs using LiF/Al and ZnO NPs/Al as cathodes is studied in this part. The devices are fabricated on glass substrate s pre -coated with a layer of indium -tin -oxide (ITO) with as sheet r cleaned by de -ionized water, acetone and isopropanol consecutively followed by ultraviolet ozone treatment for 15 minutes. A 40 nm thick PEDOT:PSS (Baytron AI 4083) is then spincoated on to the ITO followed by baking at 150 C in air for 15 minutes to remove residual water. MEH PPV with a concentration of 5 mg/mL in chloroform is spin-coated at 4000 rpm on 30 40 50 60 70 ZnO NPs Bulk ZnO(201) (200) (002) (112) (103) (110) (102) (101)Intensity (a.u.)2 (degree)(100)
139 the PEDOT:PSS to provide a ~ 80 nm thick film. For the device using LiF/Al as the cathode, the MEH -PPV layer is annealed at 150 C in nitrogen for 30 minutes before loading into a custom made vacuum chamber (background pressure: ~ 1 106 Torr) for LiF and Al deposition. For the device using ZnO N Ps/Al as cathode, the ZnO NPs are spin -coated at 4000 rpm on top of the fresh MEH -PPV layer from an ethanol solution to provide a 40 nm thick film. The MEH PPV plus ZnO NPs layer are then annealed at 150 C for 30 minutes in nitrogen to remove residual solvent followed by Al deposition. The device area is 4 mm2. Current -densityluminance voltage (J L V ) characteristics were measured using an Agilent 4155C semiconductor parameter analyzer with a calibrated Newport 818 -UV silicon diode. The luminance was calibrated using a Minolta luminance meter (LS 100). The external quantum efficiency ( EQE) was determined by assuming Lambertian emission. The electroluminescence (EL) spectrum is obtained through a highly sensitive JASCO FP 6500 fluorescence spectrometer. All measurements are conducted in air without encapsulation of the device. Figure 6 3a compares the J L V characteristics of MEH -PPV PLEDs with ZnO NPs/Al and LiF/Al as cathodes. The ZnO -based device always displays higher current injection than the LiF -based device at the same voltages and the difference is more s ignificant at the low voltage region. At the driving voltage of 2 V and 9 V current densities J = ~ 6 mA/cm2 and ~ 2.47 103 mA/cm2 were obtained in the ZnO -based device, whereas J = ~ 0.03 mA/cm2 and ~1.29 103 mA/cm2 in LiF -based device. On the other hand, the ZnO -based device shows higher luminance than the LiF -based device when V < 5.5 V but lower luminance when V > 5.5 V. It is noted that the ZnO -based device has remarkably lower turn-on voltage VT (defined as the v oltage at L = 0.1 cd/m2) than the LiF -based device, with VT = 1.36 V in the ZnO -based device and VT = 2.08 V in the LiF -based device. Figure 6 3b shows the EQE and power efficiency P of the two devices.
140 Figure 6 3. Performance of devices using ZnO NPs/Al versus LiF/Al as cathode. (a) Current density -voltage ( J V closed symbols) and luminance -voltage ( L V open symbols) characteristics. Inset: device structure of ZnO NPs device. (b) External quantum (EQE, close symbols) and luminance power ( P, open symbols) and efficiencies as a function of current density. Although higher current injection is achieved in the ZnO -based device, the ZnO -based device exhibit s lower EQE and P than the LiF based due to less improvement i n luminance. Maximum effic iencies of EQE = 0.20 % and P = 0.31 lm/W are obtained in the ZnO -based device, whereas of EQE = 0.85 % and P = 1.47 lm/W in the LiF -based device. 10010110210310-510-410-310-210-1 EQE (%) LiF/Al ZnO NPs/Al P (lm/W)J ( mA/cm2) 10-310-210-1100101102 0 2 4 6 8 10 10-610-410-2100102104 LiF/Al ZnO NPs/Al L ( cd/m2)V (V)J ( mA/cm2) 10-510-310-1101103105107 PEDOT:PSS Glass ITO MEH PPV ZnO NPs Aluminum a) b)
141 The surprisingly high current injection and low turn-on voltage in the ZnO -based device is intriguing. Ge nerally, the VT of light -emitting diode s should be close to the value of the emitted photon energy divided by electron charge (photon energy voltage). To ensure the light emission originated from excited state MEH PPV (emission peak at the wavelength of 580 nm) but not from any exciplex states between MEH PPV and ZnO N Ps, the EL spectrum at different driving voltage s of the ZnO based device is measured using a high sensitivity fluorescence spectrometer. Figure 6 4 shows the EL spectra from the ZnO -based device with driving voltage s from 1.3 V to 5 V. It can be seen that the ZnO -based device shows consistent EL spectra under all driving voltage Emission peaks located at the wavelength = 580 nm corresponds to a photon energy of ~ 2.14 eV, which is consistent with the reported EL spectrum from MEH -PPV PLED.173 Notably, weak light emission from MEH -PPV can be observed with a driving voltage as low as 1.3 V, which is approximately 0.8 V lower than the value of 2.14 V photon ene rgy voltage. Figure 6 4. Electroluminescence (EL) spectra of ZnO NPs device under different forward bias. Inset: normalized EL spectra. 300 400 500 600 700 800 0.0 0.5 1.0 1.3 V 1.5 V 1.8 V 3 V 5 VEL intensity (a.u.) (nm) 400 600 800 0 1 Normalized EL (nm)
142 Application of ZnO Nanoparti c les on Other Polymer Light -Emitting Diodes With the low turn-on voltage observed in ZnO -based MEH PPV PLED, the turn on voltages of two other PLEDs using ZnO NPs as electron injection layer are investigated. In the current study, two PLED s compris ing ITO/PEDOT:PSS/poly(9,9 -dioctylfluorene co benzothia diazole) (F8BT) or poly[2 -methoxy, 5 (3,7dimethyl octyloxy)] p -phenylene vinylene (MDMO PPV)/ZnO NPs/Al are fabricated. Figure 6 5a and b show the L V and J V characteristics of Figure 6 5 Performance of other PLEDs with ZnO NPs as electron injection layer (a) Luminance-voltage ( L V ) and (b) current density-voltage ( J V ) characteristics of ITO/PEDOT:PSS/MDMO PPV or F8BT/ZnO NPs/Al PLEDs. Inset: normalized EL spectra of the two devices. 0 2 4 6 8 10 10-510-310-1101103 MDMO-PPV F8BTL ( cd/m2)V (V) 10-110010110-410-2100102104 MDMO-PPV F8BTJ (mA/cm2)V (V) 400 600 0 1 MDMO-PPV F8BTNormalized EL (nm) a) b)
143 the two devices. Due to poor photoluminescence efficiency of MDMO -PPV compared with MEH -PPV, the MDMO -PPV device displays very low luminance with a peak value of L = ~ 30 cd/m2. Nevertheless, both devices show low turn-on voltage of VT = 2.8 V for the F8BT device and VT = 2.0 V, whi ch are close to the emitted photon energy voltage (2..33 eV for F8BT and 2.16 eV for MDMO -PPV based on wavelength of the emission peak). The inset of Figure 6 5a shows the normalized EL spectra of F8BT and MDMO PPV device at the driving voltage of 2.4 V an d 2.0 V, respectively, wh ich are consistent with the reported F8BT and MDMO -PPV spectra.166,174,175 On the other hand, the current density reaches J = ~ 1000 mA/cm2 at V = 5 V in the F8BT device. This value is much higher than the report value of F 8BT PLED using Ca as cathode, in which J = ~ 300 mA/cm2 at V = 5 V.166 Similarly, the current MDMO PPV PLED also achieves J = ~ 800 mA/cm2 at V = 5 V, which is twice higher than the report value of MDMO -PPV PLED with LiF/Al as cathode at same driving voltage.175 All these suggest that ZnO NPs is an efficient electron injection layer for a number of PLEDs. Improved charge injection and low turn-on voltage are realized in PLEDs using MEH PPV, F8BT or MDMO -PPV as emissive layer and ZnO NPs as electron injection layer. Mechanism Study for Low Turn-on Voltage P henomenon Usually, the VT of the organic or inorganic light -emitting device is close to the emitted photon energy.176,177 In the current ZnO -based MEH PPV PLED, however, 2.14 eV photon s originated from MEH PPV bandgap emission is observed with VT as low as 1.36 V, which is approximately 0.8 V lower than the photon energy voltage of 2.14 V. While such significant reduction of VT is not observed in the device with the LiF/Al cathod e, this could particularly relate to the ZnO NPs layer inserted between MEH -PPV and Al. In this section the mechanism for the ultra low turnon voltage in ZnO -based is investigated.
144 Auger Assisted Injection Model Ideally, the driving voltage necessary to o btain red, green, and blue emission in light emitting diodes is ~ 1.65, 1.95 and 2.28 V, respectively.178 Therefore, the thermodynamic limit for VT of the orange -emitting MEH PPV PLED sho uld fall in the region of 1.65 1.95 V. Observation of light emission at VT = 1.36 V in the ZnO based device definitely required an efficient energy up -conversion process to meet the thermodynamic requirement. Figure 6 6 a shows the energy level diagram of the ZnO -based device. It can be noted that the hole injection at the PEDOT:PSS/MEH PPV interface is very efficient due to the negligible hole injection barrier. Since MEH PPV has higher hole mobility than electron mobili ty, the injected holes can easily transport across the MEH PPV layer and be blocked at the MEH PPV/ZnO NPs interface due to the large energy offset between the MEH -PPV HOMO and ZnO valence band. On the other side, the electron s can be efficiently injected into ZnO NPs from the Al because of the matched ZnO conduction band level and Al work function. However, because Figure 6 6 Interfacial Auger recombination in ZnO NPs device. (a) Energy level diagram of ZnO NPs device. (b) Interfacial Auger recombinati on and electron injection process at the MEH -PPV/ZnO NPs interface, which involves 1) interfacial recombination, 2) energy transfer, 3) excitation of hot electron, 4) injection of hot electron across the MEH -PPV/ZnO NPs interface, and 5) radiative recombination in MEH PPV. Al MEH PPV ZnO NPs 3.0 5.1 4.2 7.6 PEDOT :PSS ITO 4.8 5.1 4.2 ZnO NPs MEH PPV hv = ~ 2.14 eV 4.2 3.0 5.1 7.6a) b)1 3 4 5 2
145 of the 1.2 eV barrier between the MEH -PPV LUMO and ZnO conduction, the electrons will also be blocked at the MEH -PPV/ZnO NPs interface. The blocked electrons and holes will create a strong electric field at the MEH -PPV/ZnO NPs interface under forward bias. Conceptually, ZnO -based devices should not display very efficient electron injection into the MEH -PPV layer due to the large electron injection barrier at the MEH PPV/ZnO NPs interface. However, experimental observation of MEH -PPV emiss ion at a driving voltage of 1.36 V suggests that electron s accumulated at the MEH PPV/ZnO NPs interface can gain additional energy to overcome the 1.2 eV barrier. One possible argument for electron injection at low driving voltage could be that the high energy electrons, which based on the Boltzmann distribution (assuming non -degenerate case), were swept into the MEH PPV by the strong electric field formed at the MEH PPV/ZnO NPs interface. Nevertheless, simple calculation suggests the chance of electrons of occupied energy states 1.2 eV above the bottom of ZnO conduction band is approximately 1020. By assuming effective density of state s in the conduction band at the level of ~1019 cm3, the amount of electron s injected through such a process should only support a maximum EQE of ~ 1020 %, which is significantly lower than the measured EQE = ~ 103 % at 1.6 V. To explain the origin of the additional energy for electron injection, we proposed an energy up -conversion process through interfacial Auger recombination at the MEH -PPV/ZnO NPs interface, which is depicted schematically in Figure 6 6 b. The electron injection and sequential emission process can be divided into the following steps. First, one electron from the ZnO conduction band and one hole from the MEH -PPV HOMO will recombine, and the energy released will transfer to another electron on the ZnO conduction band. With the additional energy the hot electron will occupy a higher energy state, from where the hot electron is quickly
146 swept into the MEH -PPV LUMO by the strong interfacial electric field. The injected electron will recombine with a hole accumulated on the MEH -PPV side and emit a photon. As matter of fact, such an energy up-conversion process through interfacial Auger recombination is not rare, and samples have been observed in a few inorganic heterojunctions .179 181 Seidel et. al. reported a n energy up -conversion phenomenon in a type -II inorganic heterojunction by optical ly pumping the heterojunction.180 In their case, the accumulated yet spatially separated pools of electrons and holes created through light excitation of the heterojunction is required for efficient Auger reco mbination to take place. This resembles the scenari o in our current MEH -PPV/ZnO NP heterojunction, where pools of electrons and holes are created at the heterojunction through electrical injection. Therefore, it is reasonable to believe such energy up conv ersion through the Auger process could be used to explain the ultra low turnon voltage in the ZnO -based device. Comparison of Different ZnO Materials In order to understand whether the Auger process is common to the MEH -PPV/ZnO heterojunction or only p articular to the current ZnO NPs, we compare effect of different ZnO electron injection layer s on the MEH -PPV PLED. Here, three additional types of ZnO are deposited on top of MEH PPV and their turn-on voltages are compared. The three additional types of Z nO layer involve: 1) ZnO NPs with average particle size of ~ 5 nm, 2) ZnO layer deposited through sol -gel precursor followed by annealing at 150 C for 30 minutes to transform the precursor into oxide, 3) electron beam deposited 15 nm thick continuous ZnO layer. Figure 6 7 shows the TEM image of 5 nm ZnO NPs along with their XRD pattern as well as the tapping mode atomic force microscope (AFM) images of sol gel ZnO layer. It can be seen that the 5 nm ZnO NPs display similar spherical shape to the 3 nm ZnO N Ps, and the XRD pattern indicates these larger particles adopt the same wurtzite structure as that of 3 nm ZnO NPs. The AFM
147 Figure 6 7 Morphologies of different ZnO materials. (a) Transmission electron microscope (TEM) image of ZnO NPs with 5 nm diamete r. Inset: the XRD pattern of same ZnO NPs. (b) Atomic force microscope (AFM) image of ZnO layer prep ared through sol gel precursor. The scale bar is 20 nm and the scanning area is 1.5 1.5 m2. (TEM image courtesy of Lei Qian) image, on the other hand, s uggests a fairly smooth surface of sol -gel deposited ZnO layer and grains with size as small as ~ 20 nm can be observed. With the same device structure but different ZnO electron injection layers, the four MEH PPV PLEDs display different J V and L V characteristics It can be seen from Figure 6 8 a that the devices using ZnO NPs and sol gel ZnO as electron injection layer s have a similar level of current injection at the same voltage, whereas the device with e -beam deposited ZnO layer show significantly lower current injection. The current density is J = ~ 9.0 102 mA/cm2 in the e beam deposited ZnO device, which is more than two orders of magnitude lower than the other three devices. The L V characteristics in Figure 6 8 b of the four de vices, however, indicate a gradual decrease of the luminance as the ZnO transformed from small nanoparticles to continuous film. The luminance at 9 V for 3 nm, 5 nm ZnO NPs, sol -gel ZnO and e beam ZnO devices is L = ~ 8700, ~100, ~60 and ~ 0.08 cd/m2, resp ectively. Such a decrease of luminance is 40 60 0 1 Intensity (a.u.)2 (Degree) a) b)
148 Figure 6 8 Performance of device with different ZnO electron injection layer. (a) Current density -voltage ( J V ) and (b) luminance -voltage ( L V ) characteristics. Inset: normalized EL spectra ZnO NPs, sol gel ZnO and e -beam ZnO device. achieved without significant change of EL spectra of the devices (see inset). Quite surprisingly, the ultra low VT phenomenon does not appear in the devices using larger ZnO NPs, sol gel and e beam deposited ZnO as electron injection layer. The VT actually increases from 1.36 V in 3 nm 10-110010110-610-410-2100102104 3 nm ZnO NPs 5 nm ZnO NPs Sol-gel ZnO E-beam ZnO J (mA/cm2)V (V) 1 10 10-310-210-1100101102103104 3 nm ZnO NPs 5 nm ZnO NPs Sol-gel ZnO E-beam ZnOL ( cd/m2)V (V) 400 600 800 0 1 ZnO NPs Sol-gel ZnO E-beam ZnONormalized EL (nm) a) b)
149 ZnO NPs device to 2.72 V, 5.04 V and 9.44 V in 5 nm ZnO NPs, sol gel ZnO and e beam ZnO device, respectively. Temperature Dependence of Turn-on Voltage In this part, the temperature dependence of the J V and L V characteristics of the device using 3 nm ZnO NPs as electron injection layer is investigated, while behavior of device us ing LiF/Al as cathode under the same condition s is also recorded for comparison. The temperature study is conducted in a low vacuum cryostat cooled by liquid nitrogen. The temperature of the device is monitored by a thermocouple and a temperature controlle r is used to accurately control the temperature to be 0.1 K within the setpoint. During the measurement, the temperature is gradually increased from 177 K to 357 K, and the J L V scan of the device is executed 2 minutes after temperature become s stabl e at the setpoint. Figure s 6 9 a and b is the comparison of the J V characteristics of the ZnO NPs based device and that of the LiF device. Both device s display ed improved current injection as the temperature increased, and such improvement becomes less s ignificant as the temperature goes beyond 297 K. In the LiF based device, the J V curves display two distinct regimes: (ohmic regime) at the low voltage and ( > 2 ) at high voltage. And the transition point between the two regimes decreases from 2.4 V at 179 K to 1.4 V at 357 K. On the other hand, the ZnO NP device always show s higher current injection than the LiF based device at the same temperature, especially at voltage of V < 2 V. Even though the ZnO NP based device displays a similar oh mic regime at low voltages, the range of the regime with ( > 2 ) characteristic is much smaller. The transition point between the two regimes is much lower than the LiF based device, which is pinned at 0.8 V when temperature T > 297 K. Moreover, f or the ZnO NP based
150 Figure 6 9 Temperature dependence of current density voltage ( J V ) characteristics of (a) ZnO N Ps and (b) LiF device. device, the J V curves show fast transition to space -charge limited (SCL) regime ( 2) at high voltages when temperature T > 297 K. The temperature dependent L V characteristics of the two devices are shown in Figure 6 10a and b. the ZnO NP based device show s improved luminance as the temperature increases from 177 K to 357 K, whereas th e luminance from the LiF based device actually decreases as the 10-110010110-510-310-1101103 177K 207K 237K 267K 297K 317K 337K 357K J ( mA/cm2)V (V) Increased temperature J ~ V2 10-110010110-510-310-1101103 177K 207K 237K 267K 297K 317K 337K 357K Increased temperature J ( mA/cm2)V (V) a) b)
151 Figure 6 10. Temperature dependence of luminance -voltage ( L V ) characteristics of (a) ZnO NPs and (b) LiF device. Inset: comparison of turn on voltage as function of temperature of the tw o devices. temperature increases when T > 297 K. The decrease of luminance at elevated temperature s in the LiF device may result from the fast degradation of the MEH -PPV/ LiF interface.167 Inset of figure 6 10 compares the VT of the two devices as function of the temperature. In the current case, the threshold for VT is set to be 1 cd/m2 due to the different luminance measurement geometry inside the cryostat. For LiF based device, the VT decreases from 4.2 V at 177 K to 2.2 0 2 4 6 8 10 10-2100102104 Increased temperature L ( cd/m2)V (V) 200 250 300 350 1 2 3 4 5 VT (V)T (K) ZnO NPs LiF a) b) 0 2 4 6 8 10 10-2100102104 177K 207K 237K 267K 297K 317K 337K 357K Increased temperatureL ( cd/m2)V (V)
152 V at 357 K, whereas the VT decreases from 2.9 V at 177 K to 1.25 V at 357 K for ZnO NP based device. Notably, VT become almost unchanged when T > 300K for both devices. Electron -only Device The current interfacial Auger recombination model requires the presence of a high density of electrons and holes at the MEH -PPV/ZnO NPs interface so that the three -particle process (two electrons and one hole) can effectively take place. Therefore, by significantly reducing t he density of one type of charges at the interface, the Auger recombination rate will become much lower, leading to reduced electron current coming from such a process. Base d on this consideration, electron only devices are fabricated to probe the validity of above assumption. The two electron -only devices comprise Al/MEH -PPV (80 nm)/ZnO NPs (40 nm) or LiF (1 nm)/Al deposited in sequence. Here, Al is used as anode to prevent hole injection into the MEH PPV layer. The i nset of F igure 6 11 shows the energy le vel diagram of the ZnO NPs based electron -only device. Under forward bias, electrons will accumulate at MEH -PPV/ZnO NPs interface. Since the hole density is much lower at the interface in this scenario, the interfacial Auger recombination rate is suppresse d. Figure 6 11 shows the J V characteristics under forward bias of the two electron -only devices along with that of the two normal bipolar MEH PPV PLEDs with ITO/PEDOT:PSS as anode. It is found that both electron -only devices display much lower current density than the bipolar ones d ue to the absence of hole current. It is noted that without the hole current, the ZnO NP based electron -only device actually has even lower electron current than the LiF based device. For instance, at V = 1.5 V the current density reduces from J = 0.85 mA/ cm2 in the normal PLEDs to J = 4.1 106 mA/cm2 in the electron -only device for ZnO NPs, whereas it reduces from J = 6.5 104 mA/cm2 to J = 8.9 105 mA/cm2 for LiF based devices. This
153 Figure 6 11. Current densityvoltage ( J V ) characteristics of ZnO NPs and LiF electron -only device. J V characteristics of bipolar devices are shown for comparison. Inset: energy level diagram of ZnO NPs electron -only device. suggests that without the hole current, electrons accumulated at th e MEH -PPV/ZnO NPs interface can not obtain additional energy from the Auger recombination to jump over the interfacial barrier, which lead s to such reduced current density. Plus, the Auger recombination itself in the bipolar device will also contribute to a certain portion of the forward current density. Moreover, the lower electron current device reflects that the electron injection barrier is actually higher in the ZnO NP based than that in the LiF based device, in which an interfacial dipole layer is forme d to reduce injection barrier.167 Discussion In the current study, an electron injection layer composed of ~ 3 nm ZnO nanoparticles is applied to a MEH -PPV PLED to successfully reduce the turn -on voltage of the device. It is 10-110010110-710-510-310-1101103 Bipolar device ZnO NPs device LiF deviceJ ( mA/cm2)V (V) Electron-only device Al MEHPPV ZnO NPs3.0 5.1 4.2 7.6 4.2 Al 4.2
154 proposed that an interfacial Auger recombination is responsible for the ultra low VT. As a matter of fact, the interfacial Auger recombination has been reported in a number of inorganic or even organic heterojuncti on, such as InP/AlInAs, rubrene/C60 heterojunction.180,182,183 N evertheless, observation of the Auger process at the organic/inorganic interface is rare. Owing to the negligible hole (electron) injection barrier at the PEDOT:PSS/MEH -PPV (ZnO NPs/Al) interface and large energ y barrier at the MEH -PPV/ZnO NP hererojunction, charges of opposite signs can accumulate at the interface under low forward bias ( V < 1.5 V). Electrons from the ZnO NPs and holes from MEH -PPV can form interfacial charge transfer (CT) state excitons, or say, exciplex.174,184 The energy released from the recombination of CT excitons is subsequently transferred to another electron on the ZnO NPs through a resona nt process.182,183 One important observation in the current MEH PPV/ZnO NPs heterojunction is that the appearance of interfacial Auger recombination strongly depends on the size of the ZnO NPs. The current minimum 3 nm ZnO NPs still have a size larger than the Bohr radius o f ZnO ( ~ 1.4 1.8 nm) and strong quantum confinement effect should not be expected.185,186 Nevertheless, the reduced particle size does provide a higher surface area -to -volume ratio and increased density of surface states. These surface states can serve as electron trapping centers Therefore, the incr eased density of trapped electrons could result in more efficient resonant energy transfer in the reduced size ZnO NPs, in which the coupling between the interfacial CT exciton and the electrons is stronger. As the particle size increases, the energy may r elease in the form of heat instead of transferred to the electron, which is less confined. This is indirectly proven in Figure 6 7a and b, where devices with different size of ZnO NPs display fairly similar J V characteristics but significantly different L V performance.
155 Temper ature study of the ZnO NP based device reveals that the current density increases as the temperature increases, a similar trend is also observed in LiF based device. This could be partially due to the common hole injection proce ss at the PEDOT:PSS/MEH -PPV interface through thermionic emission.187 Nevertheless, the transition point between the two regimes (ohmic regime and ( > 2 )) on J V curves of LiF based device is sensitive to temperature, and it shif ts to lower voltage s as the temperature increased. On the other hand, the transition point on J V curves of ZnO NP based device s is pinned at 0.8 when T > 297K, which support the fact that the interfacial Auger recombination is less sensitive to the temperature as long as sufficient charges are built up at the interface. Even though both ZnO NPs and LiF based devices display similar trend s on the VT versus temperature, the consistent ly lower VT in ZnO NPs under the same temperature may suggest the presence of the Auger process, which is insensitive to temper ature. Finally, investigation of electron -only devices further confirms the electron injection through the interfacial Aug er recombination. By removing the hole accumulation at the MEH -PPV/ZnO NPs interface, the Auger recombination rate is significantly reduced. And the current density show s approximately five orders of magnitude reduce in ZnO NPs device compared to only one order of magnitude change in the LiF device. Such abnormal yet significant reliance of electron injection on hole current may only be explained by the Auger process at the MEH PPV/ZnO NPs interface. A more direct way to probe the validity of the Auger proc ess may involve transient absorption measurement s of the interfacial CT excitons and the hot electron s on ZnO NPs, from which their corresponding lifetime can be extracted as well. Conclusions In this chapter, we demonstrated ultra low turn -o n voltage can be achieved in MEH -PPV PLED by using ZnO nanoparticles as electron injection layer. The ultra low turnon voltage can be explained by Auger recombination assisted electron injection at the MEH -PPV/ZnO NPs
156 interface. Such an electron injection process is peculiar to the current material system, which allows both electrons and holes to accumulate at the MEH PPV/ZnO NPs het erojunction. ZnO NP size dependence of the turnon voltage suggest strong coupling of electrons and CT excitons are required for efficient energy transfer during the Auger process. Temperature study and the electron -only device further support the hypothesis of interfacial Auger recombination for assisting electron injection. The current ZnO NPs can be applied to a variety of po lymer light emitting device s and pave a new way to achieve low turnon voltage through Auger assisted electron injection.
157 CHAPTER 7 CONCLUSIONS AND FUTU RE WORK Conclusions The rapid development of knowledge in the field of organic electronics has facilitated the transformation of lab -based research into commercial products. Nevertheless, there are still many issues need to be addressed before mass production of organic elect ronic devices. The work presented in this thesis has focused on investigation of novel material processing techniques and device architectures for efficient organic optoelectronic devices. Of particular interest, organic photovoltaic (OPV) cells and organi c light -emitting diodes (OLEDs), both of which are on the edge of commercialization, are extensively studied. Even though donor acceptor mixed bulk heterojunctions (BHJ) have been widely accepted for improving the efficiency of OPV cells, the thermodynics and kinetics process behind the formation of nanoscale phase separated mixed BHJ is still not clear. Molecular dynamic (MD) simulation combined with experimental investigation of pentacene:C60 mixtures reveals that the intermolecular interaction plays an major role on determining the degree of phase separation in the mixture (Chapter 3). And the strong interaction between pentacene molecules leads to large aggregates, which reduce exciton diffusion efficiency. Although processing conditions may still be va ried to manipulate the degree of phase separation, the controllability is quite limited. Based on the knowledge obtained from study of pentacene:C60 mixtures, we show in Chapter 4 that interdigitated BHJ composed of organic nanorod grown by oblique angle deposition (OAD) can overcome the disadvantages present in the molecular mixed BHJ and, at the same time, provide efficient exciton diffusion and charge collection. Copper phthalocyanine (CuPc) nanorod arrays with different morphology are obtained by cont rol the incident angle, substrate rotation and deposition rate during the OAD process. It is found that the optimized interdigitated BHJ
158 composed of vertical aligned CuPc nanorod and infiltrated PCBM show twice higher efficiency than the CuPc/PCBM planar h eterojunction device. Deep blue phosphorescent OLEDs (PHOLEDs) are studied in Chapter 5. Due to high triplet energy of the blue phosphorescent emitter, it is crucial to secure good exciton confinement by using high triplet energy charge transport layers. O n the other hand, improved charge confinement with EML can be obtained by increasing the energy offset between the LUMO or HOMO of EML and that of the adjacent layers. For practical application of OLEDs, high power efficiency of the device is more concerne d. Using pand n -doped organic layer can significantly reduce the driving voltage due to lower charge injection barrier at the metal -organic interface and increased conductivity of charge transport layer. Polymer light -emitting diodes (PLEDs), which is an other important category of OLEDs, have advantage of solution processability. In Chapter 6, an electron injection layer composed of ZnO nanoparticles is inserted between the MEH -PPV and the Al to significantly reduce the turn -on voltage. The observed ultra low turn-on voltage is attributed to the efficient electron injection through an Auger recombination at the MEH PPV/ZnO NPs interface. The work presented in this thesis is to understand the physical process behind the formation of different organic nanos tructure for efficiency OPV cells as well as operation principle of novel devic e architecture for efficiency OLEDs. Fruitful results are obtained from the current research work, which is helpful for future fabrication of efficient and low -cost OPV cells an d OLEDs. Future W ork Based on the knowledge obtained from the work present in this thesis, there is future research that can be conducted to expand the existing work so that large area, low -cost efficient OPV cells and OLEDs can realized.
159 Control Growth of Organic Nanostructure The organic photovoltaic (OPV) cells have been long viewed as a renewable energy to replace the current fossil fuels due to their potential to provide inexpensive and abundant electrical power from the sun, while generate zero emis sion of greenhouse gases. Even though the power conversion efficiency of OPV cells has been improved to more than 5% over the past two decades, it is still below the requirement for wide commercialization of OPV cells. In this thesis, we demonstrated that control growth of donor acceptor (D -A) nanostructure is crucial to achieve both efficient exciton diffusion and charge collection so that enhanced OPV cell efficiency can be obtained (Chapter 3 and 4). With the acquired knowledge, there is a number of othe r ways to better control the formation of desired organic nanostructure, where both efficient exciton diffusion and charge collection can be achieved. As we have learned in the study of pentacene:C60 mixtures and OAD grown CuPc nanorods, the surface and bu lk diffusivity of the molecules can strongly affect the morphology of the nanostructure. Since the surface diffusivity of the molecule can be affected by the deposition rate, the large portion of the work in this thesis has been focusing on the nanostructure morphology dependence on deposition rate. Nevertheless, another factor that deserves investigation is the substrate temperature, which strongly influences both surface and bulk diffusion of molecule. Low substrate temperature will suppress the diffusion of molecule and induce less segregated structure in the D -A mixtures or reduced nanorod diameter during OAD process. It is believed that different nanostructure can be grown by varying the substrate temperature and deposition rate over a large range. Ano ther way to control the formation of certain organic nanostructure involves utilizing pre existing template. Fi gure 6 1a shows the AFM image of 10 nm pentacene:tetracene mixed film and SEM images of gold nanoparticle deposited on top of the ITO substrate. It can be seen
160 Figure 7 1. Potential templates for controlled growth of organic nanorods (a) Atomic force microscope (AFM) images of 10 nm pentacene:tetracene = 1:1 film The scanning area is 55 2. Inset: same film with 11 2 area. (b) Si nanorod a rrays achieved by inductively coupled plasma ( ICP ) etching Si wafer using Au nuclei as masks (etching conducted by Weiran Cao). The scale bare is 100 nm. that the pentacene:tetracene mixed film display a surface morphology with uniformly distributed do mains with 50 70 nm. Figure 7 1b, on the other hand, shows the Si nanorod arrays achieved by inductively coupled plasma (ICP) ion etching the Si substrate using Au nucleation as masks. Both nanostructured thin films can be used as a template to replace the intrinsic nucleation layer during OAD process so that desired nanorod diameter and packing density can be obtained when appropriate incident angle and surface diffusivity are provided. Organic Light -Emitting Diodes The research on OLEDs has gained fru itful results over the past decade with commercial viable product appeared on the market already. Phosphorescent OLEDs (PHOLEDs) have been considered as the most promising technology due to their capability to achieve nearly 100% internal quantum efficienc y. In this thesis, we found that improved exciton and charge blocking is crucial to achieve high quantum efficiency, whereas doped charge injection layer can significantly reduce the driving voltage leading to higher power efficiency.
161 Nevertheless, there are a few challenges need to be addressed in the current deep -blue PHOLEDs. For the practical application of OLEDs, the power efficiency at hi gh luminance is more relevant ( 100 cd/m2 for display and 1000 cd/m2 for lighting). However, the fast roll -off of efficiency curve s observed in the current p -i -n device results in low power efficiency at high luminance. This fast roll -off is mainly attributed to the imbalanced injection of electrons and holes. One way t o address this issue is to control the doping con centration in the n and p doped injection layer so that more balance injection of opposite charges can be maintained at high luminance. This could partial ly address the issue of fast roll -off in the efficiency curve observed in p -i -n devices. Since 3 nm ZnO nanoparticles (NPs) can be used as efficient electron injection layer to significantly reduce the turn -on voltage of MEH -PPV PLED, it is worth applying such ZnO NPs on other types of PLEDs to see the reduction of turn -on voltage In the curr ent study, a few experiments have been conducted to demonstrate the presence of Auger recombination at the MEH -PPV/ZnO NPs interface. Nevertheless, a more direct proof of such process requires transient absorption measurement to confirm the existence of in terfacial charge transfer (CT) excitons or excited hot electrons generated by the Auger process and their lifetime as well. Afterword Instead of competing with the existing inorganic electronic device, the future of organic electronics lies in the fabrica tion of high-efficiency, low cost and large area devices with reasonable lifetime. The works presented in this thesis have paved the way to this goal so that organic photovoltaic cells and organic light -emitting diodes with improved efficiency are achievab le.
162 LIST OF REFERENCES 1 C. W. Tang, Appl. Phys. Lett. 48, 183 (1986). 2 C. W. Tang and S. A. Vanslyke, Appl. Phys. Lett. 51, 913 (1987). 3 J. J. M. Halls, C. A. Walsh, N. C. Greenham, E. A. Marseglia, R. H. Friend, S. C. Moratti, and A. B. Holmes, Nature 376, 498 (1995). 4 G. Yu, J. Gao, J. C. Hummelen, F. Wudl, and A. J. Heeger, Science 270, 1789 (1995). 5 M. A. Baldo, D. F. O'Brien, Y. You, A. Shoustikov, S. Sibley, M. E. Thompson, and S. R. Forrest, Nature 395, 151 (1998). 6 P. Peumans, S. Uchida, and S. R. Forrest, Nature 425, 158 (2003). 7 M. Pfeiffer, S. R. Forrest, K. Leo, and M. E. Thompson, Adv. Mater. 14, 1633 (2002). 8 S. H. Park, A Roy, S. Beaupre, S. Cho, N. Coates, J. S. Moon, D. Moses, M. Leclerc, K. Lee, and A. J. Heeger, Nat. Photonics 3, 297 (2009). 9 P. Peumans, A. Yakimov, and S. R. Forrest, J. Appl. Phys. 93, 3693 (2003). 10 R. H. Friend, R. W. Gymer, A. B. Holmes, J. H. B urroughes, R. N. Marks, C. Taliani, D. D. C. Bradley, D. A. Dos Santos, J. L. Bredas, M. Logdlund, and W. R. Salaneck, Nature 397, 121 (1999). 11 E. A. Silinsh and V. Capek, Organic molecular crystals: Interaction, localization, and transport phenomena AIP Press, New York, 1994. 12 M. Pope and C. E. Swenberg, Electronic processes in organic crystals and polymers 2nd Ed., Oxford University Press, 1999. 13 C. Kittel, Introduction to solid state physincs, Wiley, 7th Ed. New York, 1996. 14 W. Brutting, Physics of Organic Semiconductors Wiley -VCH, 2005. 15 J. E. Lennard Jones, Cambridge Philos. Soc. 1963. 16 C. D. Dimitrakopoulos and P. R. L. Malenfant, Adv. Mater. 14, 99 (2002). 17 S. R. Forrest, M. L. Kaplan, and P. H. Schmidt, J. Appl. Phys. 56, 543 (1984). 18 W. Q. Deng and W. A. Goddard, J. Phys. Chem. B 108, 8614 (2004). 19 A. J. Campbell, D. D. C. Bradley, and H. Antoniadis, Appl. Phys. Lett. 79, 2133 (2001). 20 P. W. M. Blom, M. J. M. deJong, and M. G. vanMunster, Phys. Rev. B 55, R656 (1997).
163 21 I. G. Hill, A. Kahn, Z. G. Soos, and R. A. Pascal, Chem. Phys. Lett. 327, 181 (2000). 22 M. Chandross, S. Mazumdar, S. Jeglinski, X. Wei, Z. V. Vardeny, E. W. Kwock, and T. M. Miller, Phys. Rev. B 50, 14702 (1994). 23 R. C. Powell and Z. G. Soos, J. Lumin. 11, 1 (1975) 24 S. R. Forrest, Nature 428, 911 (2004). 25 S. R. Forrest, Chem. Rev. 97, 1793 (1997). 26 G. B. Blanchet, C. R. Fincher, and I. Malajovich, J. Appl. Phys. 94, 6181 (2003). 27 M. Pope, P. Magnante, and H. P. Kallmann, J. Chem. Phys. 38, 2042 (1963). 28 W. Helfrich and Schneide.Wg, Phys. Rev. Lett. 14, 229 (1965). 29 V. C. Sinclair, J. M. Robertson, and A. M. Mathieson, Acta. Crystallogr. 3, 251 (1950). 30 A. M. Mathieson, J. M. Robertson, and V. C. Sinclair, Acta. Crystallogr. 3, 245 (1950). 31 M. A. Loi, E. Da Como, F. Dinelli, M. Murgia, R. Zamboni, F. Biscarini, and M. Muccini, Nat. Mater. 4, 81 (2005). 32 H. Sirringhaus, P. J. Brown, R. H. Friend, M. M. Nielsen, K. Bechgaard, B. M. W. Langeveld -Voss, A. J. H. Spiering, R. A. J. Janssen, E. W. Meijer, P. Herwig, and D. M. de Leeuw, Nature 401, 685 (1999). 33 R. A. Laudise, C. Kloc, P. G. Simpkins, and T. Siegrist, J Cryst Growth 187, 449 (1998). 34 J. Xue S. Uchida, B. P. Rand, and S. R. Forrest, Appl. Phys. Lett. 85, 5757 (2004). 35 J. Y Kim, K. Lee, N. E. Coates, D. Moses, T. Q. Nguyen, M. Dante, and A. J. Heeger, Science 317, 222 (2007). 36 A. Hadipour, B. de Boer, and P. W. M. Blom, Adv Funct Mater 18, 169181 (2008). 37 M. A. Green, K. Emery, Y. Hishikawa, and W. Warta, Prog Photovo ltaics 16, 435 (2008). 38 From National R enewable Energy Laboratory. 39 M. A. Green, Third generation photovoltaics: advanced solar energy conversion, Springer, 2003. 40 D. Chirvase, J. Parisi, J. C. Hummelen, and V. Dyakonov, Nanotechnol. 15, 1317 (2004).
164 41 G. Li, V. Shrotriya, J. S. Huang, Y. Yao, T. Moriarty, K. Emery, and Y. Yang, Nat. Mater. 4, 864 (2005). 42 W. L. Ma, A. Gopinathan, and A. J. Heeger, Adv. Mater. 19, 3656 (2007). 43 W. L. Ma, C. Y. Yang, X. Gong, K. Lee, and A. J. Heeger, Adv. Funct. Mater. 15, 1617 (2005). 44 J. Xue B. P. Rand, S. Uchida, and S. R. Forrest, J. Appl. Phys. 98, 124903 (2005). 45 J. Xue B. P. Rand, S. Uchida, and S. R. Forrest, Adv. Mater. 17, 66 (2005). 46 F. Yang, M. Shtein, and S. R. Forrest, J. Appl. Phys. 98, 014906 (2005). 47 Y. Zheng, S. K. Pregler, J. D. Myers, J. M. Ouyang, S. B. Sinnott, and J. Xue J. Vac. Sci. Technol. B 27, 169 (2009). 48 C. J. Brabec, A. Cravino, D. Meissner, N. S. Sariciftci, T. Fromherz, M. T. Rispens, L. Sanchez, and J. C. Hummelen Adv. Funct. Mater. 11, 374 (2001). 49 D. Veldman, S. C. J. Meskers, and R. A. J. Janssen Adv. Funct. Mater. 19, 1939 (2009). 50 N. Li, B. E. Lassiter, R. R. Lunt, G. Wei, and S. R. Forrest, Appl. Phys. Lett. 94, 023307 (2009). 51 M. Jorgensen, K. Norrman, and F. C. Krebs, Sol Energ Mat Sol C 92, 686 (2008). 52 B. P. Rand, D. P. Burk, and S. R. Forrest, Phys. Rev. B 75, 115327 (2007). 53 S. M. Sze, Physics of semiconductor devices, 2nd Ed. Wiley, New York, 1981 54 R. Bube and A. Fahrenbruch, Ad vanced in Electronics and Electron Physics, Vol. 56 Academic Press, New York, 1981. 55 F. Yang and S. R. Forrest, Acs Nano 2, 1022 (2008). 56 P. W. M. Blom, V. D. Mihailetchi, L. J. A. Koster, and D. E. Markov, Adv. Mater. 19, 1551 (2007). 57 S. E. Gledhill, B. Scott, and B. A. Gregg, J Mater Res 20, 3167 (2005). 58 B. P. Rand, J. Xue S. Uchida, and S. R. Forrest, J. Appl. Phys. 98, 124902 (2005). 59 J. Brooks, Y. Babayan, S. Lamansky, P. I. Djurovich, I. Tsyba, R. Bau, and M. E. Thompson, Inorg. Chem 41, 3055 (2002). 60 S. R. Forrest, D. D. C. Bradley, and M. E. Thompson, Adv. Mater. 15, 1043 (2003).
165 61 V. Shrotriya, G. Li, Y. Yao, T. Moriarty, K. Emery, and Y. Yang, Adv. Funct. Mater. 16, 2016 (2006). 62 Standard ASTM G159, Standard Tables for Reference solar Spectral Irradiance at Air Mass 1.5: Direct Normal and Hemispherical for a 37 degree Tilted Surface, American Society for Testing and Materials, West Conshocken, PA, USA. 63 Standard ASTM E948, Standard Test Method for Electrical Performanc e of Photovoltaic Cells Using Reference Cells Under Simulated Sunlight, American Society for Testing and Materials, West Conshocken, PA, USA. 64 Standard ASTM E973, Standard Test Method for Determination of the Spectral Mismatch Parameter Between a Photovo ltaic Device and a Photovoltaic Reference Cell, American Society for Testing and Materials, West Conshocken, PA, USA 65 Standard ASTM E1021, Standard Test Method for Measuring Spectral Response of Photovoltaic Cells, American Society for Testing and Mater ials, West Conshocken, PA, USA. 66 J. S. Kim, P. K. H. Ho, N. C. Greenham, and R. H. Friend, J. Appl. Phys. 88, 1073 (2000). 67 C. L. Mulder, K. Celebi, K. M. Milaninia, and M. A. Baldo, Appl. Phys. Lett. 90, 211109 (2007). 68 Y. Sun and S. R. Forrest, J. Appl. Phys. 100, 073106 (2006). 69 J. Lee, N. Chopra, and F. So, Appl. Phys. Lett. 92, 033303 (2008). 70 S. J. Williamson and H. Z. Cummi ns, Light and color in nature and art, Wiley, 1983. 71 A. Ryer, Light measuremen handbook, International Light, Newbury port, MA 1997. 72 V. Bulovic and S. R. Forrest, Chem. Phys. 210, 13 (1996). 73 J. J. M. Halls, K. Pichler, R. H. Friend, S. C. Moratti, and A. B. Holmes, Appl. Phys. Lett. 68, 3120 (1996). 74 A. L. Burin and M. A. Ratner, J. Phys. Chem. A 104, 4704 (2000). 75 W. Geens, T. Aernouts, J. Poortmans, and G. Hadziioannou, Thin Solid Films 403, 438 (2002). 76 Y. Kim, S. Cook, S. M. Tuladhar, S. A. Choulis, J. Nelson, J. R. Durrant, D. D. C. Bradley, M. Giles, I. Mcculloch, C. S. Ha, and M. Ree, Nat. Mater. 5, 197 (2006). 77 F. Yang, M. Shtein, and S. R. Forrest, Nat. Mater. 4, 37 (2005).
166 78 J. J. Dittmer, R. Lazzaroni, P. Leclere, P. Moretti, M. Granstrom, K. Petritsch, E. A. Marseglia, R. H. Friend, J. L. Bredas, H. Rost, and A. B. Holmes, Sol Energ Mat Sol C 61, 53 (2000). 79 S. C. Veenstra, G. G. Malliaras, H. J. Brouwer, F. J. Esselink, V. V. Krasnikov, P. F. vanHutten, J. Wildeman, H. T. Jonkman, G. A. Sawatzky, and G. Hadziioannou, Synthetic Met. 84, 971 (1997). 80 F. Padinger, R. S. Rittberger, and N. S. Sariciftci, Adv. Funct. Mater. 13, 85 (2003). 81 J. Peet, J. Y. Kim, N. E. Coates, W. L. Ma, D. Moses, A. J. Heeger, and G. C. Bazan, Nat. Mater. 6, 497 (2007). 82 A. C. Mayer, M. T. Lloyd, D. J. Herman, T. G. Kasen, and G. G. Malliaras, Appl. Phys. Lett. 85, 6272 (2004). 83 S. Yoo, B. Domercq, and B. Kippelen, Appl. Phys. Lett. 85, 5427 (2004). 84 S. J. Stuart, A. B. Tutein, and J. A. Harrison, J. Chem. Phys. 112, 6472 (2000). 85 W. F. Vangunsteren and H. J. C. Berendsen, Angew Chem Int Edit. 29, 992 (1990). 86 D. R. Gaskell, Introduction to the th ermodynamics of materials, 4th E d., Tayler & Francis, New York, 2003 87 M. Shtein, J. Mapel, J. B. Benziger, and S. R. Forrest, Appl. Phys. Lett. 81, 268 (2002). 88 T. Minakata, H. Imai, M. Ozaki, and K Saco, J. Appl. Phys. 72, 5220 (1992). 89 A. C. Mayer, R. Ruiz, R. L. Headrick, A. Kazimirov, and G. G. Malliaras, Org. Electron. 5, 257 (2004). 90 R. Ruiz, D. Choudhary, B. Nickel, T. Toccoli, K. C. Chang, A. C. Mayer, P. Clancy, J. M. Blakely, R. L. Hea drick, S. Iannotta, and G. G. Malliaras, Chem. Mater. 16, 4497 (2004). 91 R. Ruiz, A. C. Mayer, G. G. Malliaras, B. Nickel, G. Scoles, A. Kazimirov, H. Kim, R. L. Headrick, and Z. Islam, Appl. Phys. Lett. 85, 4926 (2004). 92 F. J. M. Z. Heringdorf, M. C. R euter, and R. M. Tromp, Nature 412, 517 (2001). 93 H. Yanagisawa, T. Tamaki, M. Nakamura, and K. Kudo, Thin Solid Films 464-65, 398 (2004). 94 A. Liu and S. J. Stuart, J Comput Chem 29, 601 (2008). 95 J. E. Northrup, M. L. Tiago, and S. G. Louie, Phys. Rev. B 66, 121404 (2002). 96 N. Troullier and J. L. Martins, Phys. Rev. B 46, 1754 (1992).
167 97 R. B. Campbell, J. Trotter, and J. M. Robertson, Acta. Crystallogr. 14, 705 (1961). 98 S. M. Allen and E. L. Thomas, The structure of materials Wiley, New York, 1999. 99 K. Tanigaki, S. Kuroshima, and T. W. Ebbesen, Thin Solid Films 257, 154 (1995). 100 W. I. F. David, R. M. Ibberson, T. J. S. Dennis, J. P. Hare, and K. Prassides, Europhys Lett 18, 735 (1992). 101 M. P. Allen and D. J. Tildesley, Computer simulation of liquids, Oxford University P ress, New York, 1987. 102 D. Nabok, P. Puschnig, and C. Ambrosch Draxl, Phys. Rev. B 77, 245316 (2008). 103 S. E. Shaheen, C. J. Brabec, N. S. Sariciftci, F. Padinger, T. Fromherz, and J. C. Hummelen, Appl. Phys. Lett. 78, 841 (2001). 104 A. K. Pandey and J. M. Nunzi, Appl. Phys. Lett. 89, 3 (2006). 105 J. Yang and T. Q. Nguyen, Org. Electron. 8, 566 (2007). 106 K. M. Coakley, Y. X. Liu, M. D. McGehee, K. L. Frindell, and G. D. Stucky, Adv. Funct. Mater. 13, 301 (2003). 107 Y. S. Hsiao, W. T. Whang, S. C. Suen, J. Y. Shiu, and C. P. Chen, Nanotechnol. 19, 415603 (2008). 108 D. C. Olson, Y. J. Lee, M. S. White, N. Kopidakis, S. E. Shaheen, D. S. Ginley, J. A. Voigt, and J. W. P. Hsu, J. Phys. Chem. C 111, 16640 (2007). 109 H. X. Xi, Z. M. Wei, Z. M. Duan, W. Xu, and D. B. Zhu, J. Phys. Chem. C 112, 19934 (2008). 110 M. M. Hawkeye and M. J. Brett, J. Vac. Sci. Technol. A 25, 1317 (2007). 111 K. Robbie, J. C. Sit, and M. J. Brett, J. Vac. Sci. Technol. B 16, 1115 (1998). 112 J. Q. Xi, M. F. Schubert, J. K. Kim, E. F. Schubert, M. F. Chen, S. Y. Lin, W. Liu, and J. A. Smart, Nat. Photonics 1, 176 (2007). 113 A. Kolmakov and M. Moskovits, Annu. Rev Mater Res 34, 151 (2004). 114 Y. P. Zhao, D. X. Ye, G. C. Wang, a nd T. M. Lu, Nano Lett. 2, 351 (2002). 115 J. A. Venables, Surf Sci 300, 798 (1994). 116 T. Karabacak, J. P. Singh, Y. P. Zhao, G. C. Wang, and T. M. Lu, Phys. Rev. B 68, 125408 (2003).
168 117 J. Zhang, I. Salzmann, S. Rogaschewski, J. P. Rabe, N. Koch, F. J. Zhang, and Z. Xu, Appl. Phys. Lett. 90, 193117 (2007). 118 D. Knipp, R. A. Street, A. Volkel, and J. Ho, J. Appl. Phys. 93, 347 (2003). 119 S. Pratontep, F. Nuesch, L. Zuppiroli, and M. Bri nkmann, Phys. Rev. B 72, 085211 (2005). 120 R. Prabakaran, R. Kesavamoorthy, G. L. N. Reddy, and F. P. Xavier, Phys Status Solidi B 229, 1175 (2002). 121 J. Xue S. Uchida, B. P. Rand, and S. R. Forrest, Appl. Phys. Lett. 84, 3013 (2004). 122 P. Peumans, V Bulovic, and S. R. Forrest, Appl. Phys. Lett. 76, 3855 (2000). 123 C. W. Chu, V. Shrotriya, G. Li, and Y. Yang, Appl. Phys. Lett. 88, 153504 (2006). 124 C. F. Zhang, S. W. Tong, C. Y. Jiang, E. T. Kang, D. S. H. Chan, and C. X. Zhu, Appl. Phys. Lett. 92, 083310 (2008). 125 S. J. Yeh, M. F. Wu, C. T. Chen, Y. H. Song, Y. Chi, M. H. Ho, S. F. Hsu, and C. H. Chen, Adv. Mater. 17, 285 (2005). 126 K. T. Wong, Y. Y. Chien, R. T. Chen, C. F. Wang, Y. T. Lin, H. H. Chiang, P. Y. Hsieh, C. C. Wu, C. H. Chou, Y. O. Su, G. H. Lee, and S. M. Peng, J Am Chem Soc 124, 11576 (2002). 127 D. Tanaka, Y. Agata, T. Takeda, S. Watanabe, and J. Kido, Jpn J Appl Phys 2 46, L117 (2007). 128 R. J. Holmes, S. R. Forrest, T. Sajoto, A. Tamayo, P. I. Djurovich, M. E. Thompson, J. Brooks, Y. J. Tung, B. W. D'Andrade, M. S. Weaver, R. C. Kwong, and J. J. Brown, Appl. Phys. Lett. 87, 243507 (2005). 129 C. Adachi, M. A. Baldo, M. E. Thompson, and S. R. Forrest, J. Appl. Phys. 90, 5048 (2001). 130 D. Tanaka, H. Sasabe, Y. J. Li, S. J. Su, T. Takeda, and J. Kido, Jpn J Appl Phys 2 46, L10 (2007). 131 R. J. Holmes, S. R. Forrest, Y. J. Tung, R. C. Kwong, J. J. Brown, S. Garon, and M. E. Thompson, App l. Phys. Lett. 82, 2422 (2003). 132 R. J. Holmes, B. W. D'Andrade, S. R. Forrest, X. Ren, J. Li, and M. E. Thompson, Appl. Phys. Lett. 83, 3818 (2003). 133 J. Li, P. I. Djurovich, B. D. Alleyne, M. Yousufuddin, N. N. Ho, J. C. Thomas, J. C. Peters, R. Bau, and M. E. Thompson, Inorg Chem 44, 1713 (2005).
169 134 S. Tokito, T. Iijima, Y. Suzuri, H. Kita, T. Tsuzuki, and F. Sato, Appl. Phys. Lett. 83, 569 (2003). 135 C. Adachi, R. C. Kwong, P. Djurovich, V. Adamovich, M. A. Baldo, M. E. Thompson, and S. R. Forrest, Appl. Phys. Lett. 79, 2082 (2001). 136 X. F. Ren, J. Li, R. J. Holmes, P. I. Djurovich, S. R. Forrest, and M. E. Thompson, Chem. Mater. 16, 4743 (2004). 137 H. Gilman and G. D. Lichtenwalter, J Am Chem Soc 80, 607 (1958). 138 M. A. Baldo and S. R. Forrest, Phys. Rev. B 62, 10958 (2000). 139 K. Goushi, R. Kwong, J. J. Brown, H. Sasabe, and C. Adachi, J. Appl. Phys. 95, 7798 (2004). 140 J. Kalinowski, M. Cocchi, D. Virgili, V. Tattori, and J. A. G. Williams, Adv. Mater. 19, 4000 (2007). 141 P. Strohriegl and J. V. Grazulevicius, Adv. Mater. 14, 1439 (2002). 142 S. Tokito, M. Suzuki, F. Sato, M. Kamachi, and K. Shirane, Org. Electron. 4, 1 05 (2003). 143 T. Matsushima, K. Goushi, and C. Adachi, Chem. Phys. Lett. 435, 327 (2007). 144 S. Reineke, K. Walzer, and K. Leo, Phys. Rev. B 75, 125328 (2007). 145 C. Gartner, C. Karnutsch, U. Lemmer, and C. Pflumm, J. Appl. Phys. 101, 023107 (2007). 146 J. D. McNeill, D. Y. Kim, Z. H. Yu, D. B. O'Connor, and P. F. Barbara, J. Phys. Chem. B 108, 11368 (2004). 147 K. Itano, H. Ogawa, and Y. Shirota, Appl. Phys. Lett. 72, 636 (1998). 148 J. Xue and S. R. Forrest, J. Appl. Phys. 95, 1869 (2004). 149 W. Stamp or, Chem. Phys. 256, 351 (2000). 150 T. Tsutsui, E. Aminaka, and H. Tokuhisa, Synthetic Met. 85, 1201 (1997). 151 X. H. Yang, F. Jaiser, S. Klinger, and D. Neher, Appl. Phys. Lett. 88, 021107 (2006). 152 F. So, B. Krummacher, M. K. Mathai, D. Poplavskyy, S. A. Choulis, and V. E. Choong, J. Appl. Phys. 102, 091101 (2007). 153 S. H. Eom, Y. Zheng, E. Wrzesniewski, J. Lee, N. Chopra, F. So, and J. Xue, Org. Electron. 10, 686 (2009).
170 154 K. Walzer, B. Maennig, M. Pfeiffer, and K. Leo, Chem. Rev. 107, 1233 (2007). 155 B. W. D'Andrade, S. R. Forrest, and A. B. Chwang, Appl. Phys. Lett. 83, 3858 (2003). 156 T. Sano, C. S. Tuan, R. E. Martin, B. S. Chuah, and A. B. Holmes, Synthetic Met. 121, 1701 (2001). 157 D. Ta naka, T. Takeda, T. Chiba, S. Watanabe, and J. Kido, Chem Lett. 36, 262 (2007). 158 S. H. Eom, Y. Zheng, N. Chopra, J. Lee, F. So, and J. Xue Appl. Phys. Lett. 93, 133309 (2008). 159 Y. R. Sun, N. C. Giebink, H. Kanno, B. W. Ma, M. E. Thompson, and S. R. Forrest, Nature 440, 908 (2006). 160 S. H. Eom, Y. Zheng, E. Wrzesniewski, J. Lee, N. Chopra, F. So, and J. Xue Appl. Phys. Lett. 94, 153303 (2009). 161 J. H. Burroughes, D. D. C. Bradley, A. R. Brown, R. N. Marks, K. Mackay, R. H. Friend, P. L. Burns, a nd A. B. Holmes, Nature 347, 539 (1990). 162 Y. Kawamura, S. Yanagida, and S. R. Forrest, J. Appl. Phys. 92, 87 (2002). 163 M. K. Mathai, V. E. Choong, S. A. Choulis, B. Krummacher, and F. So, Appl. Phys. Lett. 88, 243512 (2006). 164 B. Krummacher, M. K. Mathai, V. E. Choong, S. A. Choulis, F. So, and A. Winnacker, Org. Electron. 7, 313 (2006). 165 J. C. Scott, J. H. Kaufman, P. J. Brock, R. DiPietro, J. Salem, and J. A. Goitia, J. Appl. Phys. 79, 2745 (1996). 166 J. S. Kim, R. H. Friend, I. Grizzi, and J. H. Burroughes, Appl. Phys. Lett. 87, 023506 (2005). 167 X. H. Yang, Y. Q. Mo, W. Yang, G. Yu, and Y. Cao, Appl. Phys. Lett. 79, 563 (2001). 168 H. B. Wu, F. Huang, Y. Q. Mo, W. Yang, D. L. Wang, J. B. Peng, and Y. Cao, Adv. Mater. 16, 1826 (2004). 169 Y. Cao, G. Yu, and A. J. Heeger, Adv. Mater. 10, 917 (1998). 170 H. M. Lee, K. H. Choi, D. H. Hwang, L. M. Do, T. Zyung, J. W. Lee, and J. K. Park, Appl. Phys. Lett. 72, 2382 (1998). 171 R. Viswanatha, S. Sapra, B. Sa tpati, P. V. Satyam, B. N. Dev, and D. D. Sarma, J Mater Chem 14, 661 (2004). 172 A. L. Patterson, Phys. Rev. 56, 978 (1939).
171 173 D. Braun and A. J. Heeger, Appl. Phys. Lett. 58, 1982 (1991). 174 A. C. Morteani, A. S. Dhoot, J. S. Kim, C. Silva, N. C. G reenham, C. Murphy, E. Moons, S. Cina, J. H. Burroughes, and R. H. Friend, Adv. Mater. 15, 1708 (2003). 175 C. J. Brabec, C. Winder, N. S. Sariciftci, J. C. Hummelen, A. Dhanabalan, P. A. van Hal, and R. A. J. Janssen, Adv. Funct. Mater. 12, 709 (2002). 176 J. Shinar and R. Shinar, J Phys D Appl Phys 41, 133001 (2008). 177 G. H. Xie, Y. L. Meng, F. M. Wu, C. Tao, D. D. Zhang, M. J. Liu, Q. Xue, W. Chen, and Y. Zhao, Appl. Phys. Lett. 92, 093305 (2008). 178 R. Meerheim, K. Walzer, G. He, M. Pfeiffer, and K. Leo, Proc. SPIE 6192, 61920P (2006). 179 F. A. J. M. Driessen, Appl. Phys. Lett. 67, 2813 (1995). 180 W. Seidel, A. Titkov, J. P. Andre, P. Voisin, and M. Voos, Phys. Rev. Lett. 73, 2356 (1994). 181 A. Tit kov, W. Seidel, J. P. Andre, P. Voisin, and M. Voos, Solid State Electron 37, 1041 (1994). 182 A. K. Pandey and J. M. Nunzi, Adv. Mater. 19, 3613 (2007). 183 A. K. Pandey and J. M. Nunzi, Appl. Phys. Lett. 90, 263508 (2007). 184 A. N. Aleshin, I. P. Shche rbakov, E. L. Alexandrova, and E. A. Lebedev, Solid State Commun 146, 161 (2008). 185 C. Ton That, M. R. Phillips, and T. P. Nguyen, J. Lumin. 128, 2031 (2008). 186 Y. Danhara, T. Hirai, Y. Harada, and N. Ohno, physica status solidi (c) 3, 3565 (2006). 187 S. Barth, U. Wolf, H. Bassler, P. Muller, H. Riel, H. Vestweber, P. F. Seidler, and W. Riess, Phys. Rev. B 60, 8791 (1999).
172 BIOGRAPHICAL SKETCH Ying Zheng was born in Guangzhou, China in 1982. As a son of the Peoples Liberation Army (PLA) officer, he lived a happy childhood in the biggest military base in S outh China and finished his primary school in 1995. He graduated from Zhixin High School in 2001. With his outstanding academic performance, h e graduated summa cum laude from South China University of Technology in 2005 with a Bach elor of Engineering majored in polymer science and engineering. In fall 2005, he started his graduate study in Department of Materials Science and Engineering at University of Florida. With research topics on organic photovoltaic cells and organic light -emitting d iodes he is the author and co author of ten journal publications In 2009, he received his Doctor of Philosophy in materials science and e ngineering under the supervision of Prof. Jiangeng Xue.