ENGINEERED FILM GROWTH AND INTERFACES IN SMALL MOLECULE ORGANIC PHOTOVOLTAIC CELLS By JOHN P HILLIP MUDRICK 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 2014
Â© 2014 John P hillip Mudrick
To Mom and Dad
4 ACKNOWLEDGMENTS I am extremely grateful for having such a helpful group of colleagues to work with during my time in Gainesville. First, my doctoral adviser Dr. Jiangeng Xue has been a tremendous influence. This has been a perfect e nvironment for the beginning of a caree r in engineering and I owe it to being accepted into this research group. The freedom that I have had to conduct experimental studies balanced with the challenge of defending my results and plans has been in credi bly r ewarding. Thank you for this opportunity. I was initially trained and tutored by Dr. Jason Myers , and I sincerely enjoyed the challenge of learning the fundamentals organic photovoltaics and the minutiae of experimental detail . Seeing Dr. Ying Zheng, Dr. Sang Hyun Eom, Dr. Bill Hammond, Dr. Yixing Yang, and Dr. Renjia Zhou successfully earn their doctorate degrees when I was a young group member was a very beneficial experience. I acknowledge Dr. Wei Zhao for a very productive ye ar as a post doctoral member of our group; her contributions to the PbPc portion of the studies in Chapters 3 and 4 were extremely valuable. Dr. Ed Wrzesniewski, my brother in arms, was a great lab mate whose support was very valuable. Nate Shewmon and Dr. Weiran Cao have been great co workers ; I appreciate your ability and willingness to help keep the lab running for the last three years. Finally, Adharsh Rajogopal, Chechen Yang , and m y undergraduates Megan Hamilton and Dan Schulman really helped me see wh you for your dedicated work in the lab . This work would not have been possible without significant financial support. I gratefully acknowledge funding from the National Science Foundation CAREER and Major Research Instrumentation programs, Florida Energy Systems Consortium, the
5 Energy Solar Energy Technologies Program, and the University of Florida Office of Science. My time in Gainesville was also shaped by my friends outside of the lab, and I acknowledge their support, occasional commiseration, and ultimately help with morale. To my brother Mark, some of my favorite memories are our long weekend visits . It was great to have spent time living in the same region of the country and I hope we can do it again . I also acknowledge the continuing love and support of my fianc Ã©e Alexandria. Your example has imparted to be the b est. I love you and am ready for our life together, at last. Fin ally, I acknowledge my parents. Mom and Dad, you taught me right from wrong and instilled in me the value s of honesty, dedication , and perseverance . I w as equipped to complete my education on ly because of your lifelong guidance and unwavering support. This work is dedicated to you.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 10 LIST OF ABBREVIATIONS ................................ ................................ ........................... 15 ABSTRACT ................................ ................................ ................................ ................... 17 CHAPTER 1 INTRODUCTI ON TO ORGANIC PHOTOVOLTAICS ................................ ............. 19 1.1 Motivation ................................ ................................ ................................ ......... 19 1.2 Inorganic and Organic Semiconductors ................................ ............................ 22 1.3 Brief History of Organic Photovoltaics ................................ ............................... 25 1.4 Optical and Electrical Processes in Organic Semiconductors ........................... 26 1.4.1 Chemical and Electronic Structures ................................ ......................... 26 1.4.2 Optical Absorption ................................ ................................ ................... 31 1.4.3 Electrical Conduction ................................ ................................ ............... 35 1.4.4 Excitonic Characteristics ................................ ................................ ......... 37 1.5 Organic Photovoltaic Devices ................................ ................................ ........... 40 1.5.1 Donor/Acceptor Heterojunction ................................ ............................... 40 1.5.2 Photocurrent Generation Processes ................................ ........................ 41 1.5.3 OPV Device Architectures ................................ ................................ ....... 43 1.6 Dissertation Overview ................................ ................................ ....................... 44 2 FABRICATION AND CHARACTERIZATION OF ORGANIC THIN FILMS AND PHOTOVOLTAIC DEVICES ................................ ................................ ................... 46 2.1 Thin Film and Device Fabrication Methods ................................ ....................... 46 2.1.1 Vacuum Thermal Evaporation ................................ ................................ . 46 2.1.2 Thermal Gradient Sublimation ................................ ................................ . 49 2.2 OPV Device Characterization ................................ ................................ ........... 52 2.2.1 OPV Performance Metrics ................................ ................................ ....... 52 2.2.2 Measurement and Calibration Methods ................................ ................... 58 2.3 OPV Device Characteristics ................................ ................................ .............. 63 2.3.1 Quantifying Conventional versus Organic PV ................................ .......... 63 2.3.2 Fundamental Limitations of OPV Devices ................................ ............... 65 2.4 Bridging the Gap: Routes Toward Enhancement ................................ .............. 70 2.4.1 Current State of the Art ................................ ................................ ............ 72
7 3 EFFECTS OF ORGANIC ANODE INTERLAYERS ON DONOR FILM PROPERTIES ................................ ................................ ................................ ......... 73 3.1 Introduction ................................ ................................ ................................ ....... 73 3.2 Donor and Interlayer Materials Selection ................................ .......................... 73 3.2.1 Donor Materials ................................ ................................ ....................... 73 3.2.2 Interlayer Materials Selection Criteria ................................ ...................... 75 3.2.3 Pairing Donor and Interlayer Materials ................................ .................... 76 3.3 The Effects of a Polycrystalline Pentacene Interlayer on Lead Phthalocyanine Film Properties ................................ ................................ ........... 78 3.3.1 Optical Absorption ................................ ................................ ................... 78 3.3. 2 Surface Morphology ................................ ................................ ................ 81 3.4 The Effects of Tetracene Anode Interlayers on Zinc Phthalocyanine Film Properties ................................ ................................ ................................ ............ 83 3.4.1 Optical Absorption ................................ ................................ ................... 84 3.4.2 Bulk Structure: X ray Diffraction ................................ .............................. 85 3.4.3 Surface Topography: Atomic Force Microscopy ................................ ...... 91 3.5 The Effects of PEDOT:PSS/Tetracene and PEDOT:PSS/NPB Interla yer Structures on DBP Film Properties ................................ ................................ ...... 94 3.5.1 Optical Absorption ................................ ................................ ................... 95 3.5.2 Photoluminescence Quenching ................................ ............................... 95 3.6 Summary of AIL/Donor Multilayer Structures ................................ .................. 101 4 EFFECTS OF ORGANIC ANODE INTERLAYERS ON PHOTOVOLTAIC DEVICE PERFORMANCE ................................ ................................ .................... 103 4.1 Introduction ................................ ................................ ................................ ..... 103 4.2 Planar Heterojunction PbPc based OPV Devices with a Pentacene Template Layer ................................ ................................ ................................ .. 103 4.3 ZnPc based OPV Devices with a PEDOT:PSS/Tetracene Anode Interlayer Structure ................................ ................................ ................................ ............ 110 4.3.1 Planar Heterojunction Devices ................................ .............................. 110 4.3.2 Bulk and Planar Mixed Heteroju nction Devices ................................ ..... 113 4.4 DBP based OPV Devices with Exciton Dissociating and Exciton Blocking Anode Interlayers ................................ ................................ .............................. 120 4.4.1 Planar Heterojunction Device Performance ................................ ........... 121 4.4.2 Comparison of Planar and Bulk Heterojunction Device Characteristics 126 4.4.3 Simulation of Photocurrent Generation at Planar Heterojunction Interfaces ................................ ................................ ................................ .... 129 4.5 Summary ................................ ................................ ................................ ........ 135 5 COMPARISON OF ORGANIC PHOTOVOLTAIC DEVICES WITH EXCITON DISSOCIATING AND EXCITON BLOCKING INTERLAYER STRUCTURES ...... 138 5.1 Introduction ................................ ................................ ................................ ..... 138 5.2 Review of Cascade based OPV Devices ................................ ........................ 139 5.3 Illumination Intensity Dependent Behavior ................................ ...................... 140
8 5.4 Spectrally Resolved, Voltage Dependent Photocurrent Behavior ................... 144 5.5 DBP Thickness Dependence ................................ ................................ .......... 149 5.6 Excitonic Photocurrent Calculations of DBP based Devic es with Exciton Quenching, Blocking, and Dissociating AIL Structures ................................ ... 155 5.7 Comparison of AIL/DBP Structures with Tetracene/Zn Pc Devices ................. 158 5.8 Summary of Cascade and Blocking Structures ................................ ............... 165 6 TANDEM SMALL MOLECULE ORGANIC PHOTOVOLTAIC CELLS FOR BROAD SPECTRAL RESPONSE AND HIGH OPEN CIRCUIT VOLTAGE ......... 169 6.1 Introduction ................................ ................................ ................................ ..... 169 6.2 Tandem Device Fundamentals ................................ ................................ ....... 169 6.3 Donor Materials Selection and Preliminary Tandem Devices ......................... 173 6.3.1 Initial Tandem Device Structure ................................ ............................ 176 6.4 Individual Subcell Characterization ................................ ................................ . 177 6.5 Tandem Device Optimization ................................ ................................ .......... 183 6.6 Conclusions ................................ ................................ ................................ .... 188 7 CONCLUSION S AND OUTLOOK ................................ ................................ ........ 189 7.1 Conclusions ................................ ................................ ................................ .... 189 7.2 Future Work and Outlook ................................ ................................ ................ 192 LIST OF REFERENCES ................................ ................................ ............................. 195 BIOGRAPH ICAL SKETCH ................................ ................................ .......................... 209
9 LIST OF TABLES Table page 2 1 List of organic semiconductor materials used in this work and their purification methods. ................................ ................................ .......................... 52 2 2 Photovoltaic performance parameters of the confirmed highest performing single junction PV devices based on various active layer materials. .................. 64 2 3 Photovoltaic performance parameters of CuPc/C 60 OPV de vices with PHJ and BHJ active layer structures under 100 mW/cm 2 simulated AM 1.5G illumination. ................................ ................................ ................................ ......... 66 3 1 Calculated and observed diffraction peak positions for ZnPc films. .................... 88 4 1 Photovoltaic performance parameters of PbPc based OPV devices grown on bare ITO and on Pentacene under 100 mW/cm 2 simulated AM 1.5G illumination. ................................ ................................ ................................ ....... 105 4 2 Photovoltaic performance parameters of PHJ ZnPc devices with various AIL structures under 100 mW/cm 2 simulated AM 1.5G illumination. ....................... 111 4 3 Photovoltaic performance parameters of BHJ and P M HJ ZnPc devices under 100 mW/cm 2 simulated AMA 1.5G illumination. ................................ ..... 116 4 4 Photovoltaic performance parameters of DBP (17 nm)/C 60 (30 nm) devices with various AIL structures under 100 mW/cm 2 simulated AM 1.5G illumination. ................................ ................................ ................................ ....... 122 4 5 Photovoltaic performance parameters of various DBP based OPV devices under 100 mW/cm 2 simulated AM 1.5G illumination. ................................ ........ 127 6 1 Photovoltaic performance parameters of single D/A HJ devices with different donor species under 100 mW/cm 2 simulated AM 1.5G illumination. ................ 175 6 2 Photovoltaic performance parameters of tandem devices under 100 mW/cm 2 simulated AM 1.5G illumination. ................................ ................................ ....... 188
10 LIST OF FIGURES Figure page 1 1 Energy consumption in the United States throughout history, sorted by energy source. ................................ ................................ ................................ .... 20 1 2 ASTM solar spectral irradiance and total absorbed power as a function of semiconductor band gap energy. ................................ ................................ ....... 22 1 3 Highest achieved efficiencies for various PV technologies. Image courtesy of the National Renewable Energy Laboratory. ................................ ...................... 25 1 4 Chemical structure of various classes of organic materials. ............................... 27 1 5 Schematic of the bonding in ethylene. ................................ ................................ 29 1 6 Molecular st ructures and frontier energy levels of several organic semiconductors. ................................ ................................ ................................ . 30 1 7 Trends in the optical absorption and E HOMO levels in terminally dicyano substituted oligothiophenes with different conjugation lengths. .......................... 32 1 8 Absorption efficiency and ex tinction coefficient for DIP and ZnPc films, respectively, grown on different substrates. ................................ ....................... 33 1 9 Absorption coefficient spectra of six organic semiconductor thin films. .............. 34 1 10 Excitons in a solid. ................................ ................................ .............................. 38 1 11 Energy level diagram of a donor/acceptor heterojunction. ................................ .. 41 1 12 Multi step photocurrent generation mechanism in a donor/acceptor heterojunction. ................................ ................................ ................................ .... 42 1 13 P lanar and Bulk heterojunction OPV device architectures used in this work. ..... 44 2 1 Schematic of a vacuum thermal evaporation chamber. ................................ ...... 48 2 2 Summary of thermal gradient sublimation. ................................ ......................... 50 2 3 Experimental and fitted dark current for an archetypal OPV device. .................. 54 2 4 Photovoltaic behavior of a typical CuPc/C 60 OPV device. ................................ .. 55 2 5 Short circuit EQE spectrum of a CuPc/C 60 OPV device alongside the AM 1.5G solar spectrum. ................................ ................................ .......................... 57 2 6 Measurement set up for illuminated J V data collection. ................................ .... 58
11 2 7 Equipment components and experimental arrangement for spectral measurements. ................................ ................................ ................................ ... 59 2 8 Reference AM 1.5G and Xe solar simulator spectra. ................................ .......... 60 2 9 Short circuit EQE spectrum and J V characteristics of a DBP/C 60 based OPV device. ................................ ................................ ................................ ................ 63 2 10 OPV device performance of CuPc/C 60 devices with PHJ and BHJ active layer structures. ................................ ................................ ................................ ........... 66 2 11 Schematic of a D/A HJ energy level struct ure and corresponding OPV device characteristics. ................................ ................................ ................................ .... 69 3 1 Molecular structures of donor materials used in this study. ................................ 74 3 2 HOMO and LUMO energy levels of materials investigated in this study. ............ 77 3 3 Thin film absorption spectra of PbPc films. ................................ ......................... 80 3 4 AFM height images of 10 Ã… thick PbPc films grown on bare SiO 2 . ..................... 81 3 5 Evolution of surface morphology with PbPc thickness on Pentacene. . ............... 83 3 6 Optical absorption spectra of ZnPc films.. ................................ .......................... 84 3 7 Schematics of ZnPc molecular orientations and XRD spectra.. .......................... 86 3 8 X ray diffraction spectra of 80 and 40 nm thick ZnPc films on ITO substrate. .... 90 3 9 AFM height images of 25 nm thick ZnPc films. ................................ ................... 92 3 10 AFM height images of 10 nm thick Tetracene films. . ................................ .......... 93 3 11 Optical absorption of 15 nm thick DBP films grown on different underlying layer structures. ................................ ................................ ................................ .. 95 3 12 Schematic representations of energy level diagrams of three combinations of quenching and blocking DBP interfaces. ................................ ............................ 97 3 13 PL spectra of DBP films surrounded by different interlayer species. .................. 98 3 14 Calculated steady state exciton concentrations within a DBP film under = 530 nm illumination. ................................ ................................ .......................... 100 3 15 Total exciton concentrati on as a function of L D normalized to the case of two quenching interfaces. ................................ ................................ ....................... 101 4 1 Photovoltaic performance characteristics of PbPc based OPV devices. .......... 104
12 4 2 Short circuit EQE spectra of PbPc based devices with varying PbPc layer thickness. ................................ ................................ ................................ ......... 107 4 3 Comparison of the PbPc layer thickness dependencies of PbPc/C 60 solar cell performance characteristics for devices wi th and without a Pe ntacene template layer. ................................ ................................ ................................ .. 109 4 4 Photovoltaic performance characteristics of ZnPc based OPV devices. .......... 111 4 5 Two components of EQE spectra as a function of AIL structure.. ..................... 113 4 6 J V characteristics of ZnPc based OPV devices with a mixed ZnPc:C 60 layer. 114 4 7 IQE spectra of ZnPc based devices with different active layer structures. ....... 117 4 8 Relative IQE enhancement in the ZnPc absorption region for the three ZnPc:C 60 active layer architectures. ................................ ................................ . 118 4 9 Photovoltaic performance characteristics of optimized P M HJ OPV device. ... 119 4 10 J V characteristics of DBP based PHJ devices with different AIL structures under 100 mW/cm 2 simulated AM 1.5G illumination. ................................ ........ 122 4 11 Short circuit EQE and absorption coefficient spectra of DBP based PHJ devices and organ ic semiconductors present in said devices. ......................... 123 4 12 Spectral distrib utions of absorption and internal quantum efficiencies for DB P based OPV devices with different AIL structures. ................................ .... 125 4 13 J V characteristics of DBP based P M HJ and PHJ devices under 100 mW/cm 2 simulated AM 1.5G illumination. ................................ ......................... 127 4 14 Quantum efficiency spectra of DBP:C 60 P M HJ devices with three different DBP:C 60 blend ratios and a PHJ device with PEDOT:PSS/NPB AIL structure. 12 9 4 15 Equilibrium exciton concentration profiles within the DBP layer of a PHJ OPV device excited by = 615 nm illumination. ................................ ....................... 132 4 16 Calculated EQE ED as a function of L D in a DBP based PHJ device with various AIL structures under = 615 nm incident light. ................................ .... 134 5 1 Incident power dependence of photocurrent generation for DBP based devices with different AIL structures. ................................ ................................ 142 5 2 Voltage dependent EQE corresponding to light absorption in the acceptor and donor layers. ................................ ................................ .............................. 146 5 3 Voltage dependent EQE normalized to short circuit values. ............................ 148
13 5 4 Photovoltaic performance summary of DBP based devices with different anode/AIL structures. ................................ ................................ ....................... 151 5 5 Voltage dependent EQE corresponding to C 60 absorption, normalized to the short circuit value. ................................ ................................ ............................ 152 5 6 Voltage dependent EQE corresponding to DBP absorption, normalized to the short circuit value.. ................................ ................................ ........................... 154 5 7 Calculated maximum EQE corresponding to = 615 nm illumination as a function of exciton diffusion length L D for quenching, dissociating, and blocking anode/DBP interfaces across a range of DBP layer thicknesses. ...... 156 5 8 Illumination intensity dependent photovoltaic performance parameters of ZnPc based PHJ devices grown on bare ITO and ITO/PEDOT:PSS/Tetracene AIL structure. ................................ ....................... 159 5 9 Voltage dependent EQE corresponding to donor absorpt ion for DBP and ZnPc based PHJ devices under = 615 and 630 nm illumination, respectively. ................................ ................................ ................................ ...... 160 5 10 EQE corresponding to = 63 0 nm illumination for PHJ ZnPc based devices with different ZnPc layer thicknesses and AIL structures. ................................ 163 5 11 EQE(V) normalized to short circuit values for PHJ devices. ............................. 165 5 12 EQE corresponding to donor absorption for devices with blocking and cascade AIL structures. ................................ ................................ .................... 167 6 1 Trade offs in OPV devices. ................................ ................................ ............... 170 6 2 J V characteristics of a single device and a tandem device with CuPc /C 60 active layer structures. ................................ ................................ ...................... 171 6 3 PbPc/C 60 device performance with and without a 50 nm thick transparent spacer layer. ................................ ................................ ................................ ..... 173 6 4 PV performance of single donor devices. ................................ ......................... 175 6 5 Photovoltaic response and layer structures of OPV devices. ........................... 177 6 6 Spectra of device EQE and LED bias spectra for individual subcell characteri zation. ................................ ................................ ............................... 178 6 7 J V characteristics of OPV devices under various illumination sources. ........... 179 6 8 Illustration of the need for voltage bias in tandem subcell EQE measurements. ................................ ................................ ................................ . 180
14 6 9 Short circuit EQE spectra o f single devices and tande m device subcells ......... 181 6 10 Position dependence of the optical electric field intensity calculated by the transfer matrix method within a single D/A HJ SubPc device an d a tandem dev ice. ................................ ................................ ................................ .............. 183 6 11 Exciton generation rate within the mixed SubPc:C 60 layer of a SubPc device and tandem devices with varying front ce ll C 60 layer thic kness ....................... 184 6 12 Short circuit EQE spectra of PbPc and SubPc sing le D/A HJ devices and tandem su bcells with varying front cell C 60 layer thickness. .............................. 186 6 13 J V characteristics of tandem cells with varying front cell C 60 layer thickness. . 187
15 LIST OF ABBREVIATIONS AFM Atomic force microscopy AIL Anode inter layer AM 1.5G Air mass 1.5, global axis tilt BHJ Bulk heterojunction CIGS Copper indium gallium diselenide CuPc Copper Phthalocyanine DBP Tetraphenyl dibenzoperiflanthene EQE External quantum efficiency FF Fill factor GaAs Gallium arsenide HOMO Highest occupied molecular orbital InPc Indium Phthalocyanine Chloride IQE Internal quantum efficiency IR Infrared ITO Indium tin oxide J SC Short circuit current density L D Exciton diffusion length LED Light emitting diode LUMO Lowest unoccupied molecular orbital P Power conversion efficiency NIR Near infrared NPB Bis(naphthalene 1 yl) bis(phenyl) benzidine OPV Organic photovoltaic OSC Organic semiconductor
16 PbPc Lead Phthalocyanine PEDOT:PSS Poly(3,4 ethylenedioxythiophene):poly(styrenesulfonate) PHJ Planar heterojunction PL Photoluminescence P M HJ Planar mixed heterojunction PV Photovoltaic Si Silicon SubPc Boron Subphthalocyanine Chloride VB Valence band V OC Open circuit voltage VTE Vacuum thermal evaporation XRD X ray diffraction ZnPc Zinc Phthalocyanine
17 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 ENGINEERED FILM GROWTH AND INTERFACES IN SMALL MOLECULE ORGANIC PHOTOVOLTAIC CELLS By John P hillip Mudrick August 2014 Chair: Jiangeng Xue Major: Materials Science and Engineering Organic photovoltaic (OPV) devices have gained significant academic interest over the past three decades because of their potential as a low cost energy source with uniquely tuneable properties. While recent reports of devices with greater than 10 percent power conversion efficiency are very encouraging to this young technology, OPV devices are still limited by intrinsic properties of organic semiconducting materials. In this work, two strategies for c ircumnavigating the inherent tradeoffs of these materials are explored in detail. First, the insertion of an anode interlayer (AIL) structure between th e substrate and active layers wa s investigated as a means to alter the optoelectronic properti es of sub sequently deposited films and ultimately to improve OPV device performance. Lead Phthalocyanine (PbPc) thin films grown on a Pentacene AIL exhibit ed increased near infrared (NIR) absorption resulting from an increased fraction of the triclinic crystal phas e. Compared with thin films and devices grown on oxide surfaces, spectral response wa s extended well into the NIR range, increasing photocurrent generation by more than 30 % . Next, the crystalline regions of Zinc Ph thalocyanine (ZnPc) thin films we re shown to shift from exclusively edge on to a combination of edge on and face on
18 orientations with the introduction of a PEDOTPSS/Tetracene AIL structure. By varying the active layer structures of ZnPc based OPV devices, significant enhancemen ts in photocurrent generation wer e demonstrated and a ZnPc record 5.8 perce nt power conversion efficiency wa s achieved. Finally, the amorphous d onor Tetraphenyldibenzoperiflanthene (DBP) wa s used in conjunction with exciton dissociating and exciton blocking AIL structures in order to directly compare their effects on OPV device performance. While both AIL structures improve d photocurrent generation, we show ed that the exciton blocking AIL structure is best suited for maximizing power conversion efficiency. Finally, tandem de vices we re constructed with two complimentary donor materials to achieve a device with high open circuit voltage and spectral response past one micron. By measuring the photocurrent behavior of the individual subcells within the tandem stack, we we re able to validate optical electric field calculations and improve power conversion efficiency to a material record 4.5 percent.
19 CHAPTER 1 INTRODUCTION TO ORGANIC PHOTOVOLTAICS 1.1 Motivation Dating back to the discovery of fire and tools many thousands of years ago, humans have distinguished ourselves as a species by our ability to use natural resources to improve our quality of life. As time has gone on, our understanding of physical processes has vastly increased both the amount of natural resources that we are able to harness and the methods by which we harness them. This ability to learn and adapt has opened many possibilities to mankind, none of which would be possible without vast amounts of available energy and the abi lity to put that energy to use. Nevertheless, energy consumption changed relatively little from the discovery and control of fire until the discovery of the steam engine in the 18 th century. Once the potential of naturally occurring fossil fuels was unlocked, human civilizations have been able to achieve unprecedented levels of widespread prosperity and quality of life. In the United States, the diver sity of available energy to shift between different sources as they come available. A historical timeline of United States energy consumption by source is shown in Figure 1 1. Dominated by wood burning prior to the discovery of coal some 150 years ago , the utilization of the higher energy density fossil fuels coal, oil, and natural gas came to compose the majority of energy production in the United States throughout the 20 th century. These tried and true methods continue to provide for the vast majorit y of present energy consumption in many developed countries. Whether or not it is in the best interests of regional, national, and global societies to rely on these traditional energy sources well into the future is a very fascinating
20 debate. On the one h and, developed countries have successfully built robust infrastructures for capturing and deploying energy from fossil fuels. As a result, the financial costs of energy derived from these sources are relatively low. Further advantages in technology have se rved to further improve our ability to extract fossil fuels from the earth and spread power efficiently. On the other hand, the infrastructures that keep costs down in developed countries are not in place in underdeveloped or developing regions of the worl d. Furthermore, the finite amount of traditional energy resources makes relying on them a dubious proposition on any length scale over 100 years. Finally, the scale and reversibility of the environmental impact resulting from traditional energy sources cas ts a shadow over their widespread use in the future. Figure 1 1. Energy consumption in the United States throughout history, sorted by energy source. Several renewable, clean energy sources have emerged as viable large scale alternatives to traditional sources in the last few decades. Together, the renewable sources hydropower, wind, biomass and waste, solar photovoltaic, geothermal, and
21 concentrated solar combined to generate approximately 4.5 terrawatt hours (TWh) in 2011  , representing a 55% increase in global renewable power generation since 2000. While this is a staggering number, it still represents only a small portion of total energy supply: a case study conducted by the United States Energy Information Administration projects that 12% of worldwide energy consumption in 2014 will come from renewable sources  . Clearly, if renewable sources are going to be a factor in the future they must be available on a very large scale. Among the renewable power options, solar is unique because of the vast amount of energy available. Disregarding the angular and spatial distributions of optical power and the intermittent nature of solar availability optical power intensity is 1 W/m 2 . Accounting for the entire surface area of the planet, this corresponds to roughly 50,000 TW. If we consider the 29% of surface that is not covered by water and assume a feasible 10% power conversion efficiency , we arrive at 1450 TW of usable solar power. That same E nergy Information Administration case study projects a global energy demand of 30 TW i n the year 2050. So, covering roughly demand when the sun is shining. T hese back of the envelope calculations have omitted important details such as distribution efficien cy, temporal mismatch between supply and consumption, and stability. But the fact that the amount of solar energy incident on the earth represents a substantial fraction of energy consumption on a global scale merits attention in itself. The work presented in this dissertation is motivated by this vast potential of solar energy as a solution to local, regional, domestic, and global power requirements.
22 1.2 Inorganic and Organic Semiconductor s In the most basic sense, a photovoltaic (PV) device directly conve rts solar energy to electrical energy by converting photons to electrons. This process can be further broken into two steps: light absorption and charge carrier conduction. I n the first step, i ncident photons with sufficient energy can promote electrons in a solid from the ground state to a quantum mechanically allowed excited state. Semiconductors are ideal candidates for maximizing the efficien cy of this process because the energy gaps between their ground and excited states, or band gaps , are well aligne d with the energetic distribution of solar energy. A reference solar irradiance spectrum generated by the American Society for Testing and Materials  is shown in Figure 1 2. Figure 1 2. ASTM solar spectral irradiance and total absorbed power as a function of semiconductor band gap energy.
23 The majority of solar irradiance is distributed between photon energies of 1 and 3 eV, in good agreem ent with the allowed interband transition energies of the inorganic semiconductors silicon (Si, 1.1 eV), gallium arsenide (GaAs, 1.4 eV) and copper indium gallium diselenide (CIGS, 1.0 1.7 eV). Put another way, if we assume continuous light absorption of photon energies greater than the band gap, the band gap energy of Si allows for 800 W/m 2 , or 80%, of the incident optical power to be absorbed . In addition to advantageous optical properties, semiconductors also exhibit excellent electrical properties for efficient charge carrier transport . Inorganic semiconductors are c haracterized as covalent solids where the valence electrons of neighboring atoms are shared within the crystal. The se three dimensional crystalline networks result in well defined energ y bands which are filled with electrons from the principle . The highest lying fully occupied band is termed the valance band, and the lowest lying empty (at T = 0 K) band i s termed the conduction band. By introducing dopants, atoms with more or fewer valence electrons than the host material, into substitutional lattice sites the occupancies of the valence and conduction bands can be engineered to achieve the desired electric al properties. Owing to their deep knowledge base dating back to the 1950s and excellent electrical properties which arise from extremely high levels of purity, power conversion efficiencies ( P ) over 20% have been achieved in cells based on these materia ls  . While these efficiencies are high enough to satisfy significant energy needs, the extent o f f uture enhancement is relatively limited as these efficiencies are very close to the theoretical maximum , first calculated by Shockley and Queisser  . A fter more than a
24 half century spent collectiv ely researching and developing photovoltaics, the cost of these modules is still high enough to limit their market share to well less than 1% of global and do mestic power generation [1,2] . A large part of these costs stem s from the complexity and high thermal budgets of processing steps that are required to achieve semiconductor films with the optoelectronic properties requir ed for high conversion efficiency . The relatively slow pace of inorganic PV development and market penetration merits the consideration and study of other classes of photoactive materials. Polymer and molecular solids with semiconducting properties present an alternative to inorganic semiconductors. Organic semiconducting (OSC) materials can be prepared via synthetic chemistry from inexpensive source materials. They can also have a broad range of optoelectronic properties which can be tuned by altering the chemical structure of the constituent materials and p rocessing methods of films and devices. Furthermore, semiconducting polymers and oligomers are compatible with high throughput processing methods and non rigid geometries [6,7] . Altogether these advantageous properties of OSCs can reduce processing and module costs, ultimately improving such parameters as the energy payback time  , required energy input  and final cost of electricity  . While the materials and processing methods re present an advantage for organic photovoltaics, the ultimate performance of their incorporated devices lags behind that of inorganic photovoltaic devices. While P values have surpassed 20 % in devices based on inorganic semiconductors, champion organic ph otovoltaic devices have only recently surpassed the 10 % mark. A timeline of maximum P values achieved for various PV
25 technologies is shown in Figure 1 3  . The reasons for relatively low conversion efficiencies in organi c photovoltaics will be discussed in Chapter 2. Figure 1 3. Highest achieved efficiencies for various PV technologies. Image courtesy of the National Renewable Energy Laboratory. 1.3 Brief History of Organic Photovoltaics The field of organic photovoltaics (OPV) largely traces itself back to the development of the so published by C. W. Tang in 1986  . I f we take a more simplistic view, carbon based organisms have captured e nergy from the sun for billions of years . Recalling this fact alone gives the scientist reason to believe that photovoltaics, specifically OPVs, can play a major role in energy production for human consumption. More practically, photoconductivity in an org anic material was first observed in A nthracene in the early 20 th century  , but the first functioning PV cells based on organic active materials were not discovered until the 1950s [14 16] . While
26 interesting as a proof of concept, these device were not efficient ( P << 1 %) enough for practical consideration. Efficiencies remained well below 1 % until 1984 and 1986 when Harima et al.  and Tang  , respectively, sandwiched two organic layers with different ionization potentials between metal electrodes, forming what would later become known as the donor/acceptor (D/A) heterojunction (HJ). The development of new organic materials with desirable photovoltaic properties quickly pushed record efficiencies past 1 % in the early 1990s. In the two decades since, further imp rovements in physical, optical, and electronic properties, as well as new fabrication and processing concepts, have pushed the record OPV efficiencies past the 10 % mark. 1.4 Optical and Electrical Processes in Organic Semiconductors The structural, electr ical, and optical properties of OPVs arise from the underlying characteristics of their constituent materials. In this section, the physical properties of OSCs will be introduced in order to build a framework for discussing the performance characteristics of OPV devices in Chapter 2 . From these fundamental properties, the challenges associated with ach ieving high performance OPVs will become apparent. 1.4.1 Chemical and Electronic Structures By a strict definition, organic materials are those which are comp osed entirely of carbon and hydrogen. Many of the materials used in t his field of study would not fit that definition, so in the context of this work organic will refer to any carbon containing material. Carbonaceous solids come in many forms of varying complexity , as illustrated in Figure 1 4. Here we break organic materials into three groups: small molecules, polymers, and biological molecules. Small molecules and polymers have similar
27 optoelectronic properties in the solid state but feature different processing techniques for film formation. S mall molecules and polymers with semiconducting characteristics feature atoms. Biological molecules can be envisioned as an extended netw ork of oligomeric or polymeric units, as exemplified by molecules such as DNA. Figure 1 4. Chemical structure of various classes of organic materials. A) Small molecular weight molecule zinc phthalocyanine, B) polymers poly (ethylene dioxythiophene) and poly (styrene sulfonate), and C) deoxyribonucleic acid (DNA).
28 The double bonded hydrocarbon molecule ethylene, C 2 H 4 , is a useful model for describing organic solids because it has a relatively simple chemical structure and embod ies the electronic structur e of larger molecules with extended electron systems , including OSC material s . A lone c arbon atom has the electronic configuration 1s 2 2s 2 2p xyz 2 . When two CH 2 moieties are brought together, the four valence electrons within the n = 2 shell hybridize in order to reduce the energy of the two body system. The four valence electrons of each carbon atom hybridize into three sp 2 electrons and one p z electron: the sp 2 electrons form coplanar atoms, one with the opposite carbon atom) while the p z electrons can form a stabilizing with one another. The three in the same plan e at approximately 120Â° angles with one another , while t he two p z electrons are then located above and below the pl ane of the molecu le and can carbon atoms. This bonding scheme is illustrated schematically in Figure 1 5. ons within the bond are * orbital, which is the next highest lying energetic configuration allowed in the system. Together, these two configurations form the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO). These energy levels * transitions dominate the optoelectronic properties of OSC materials and devices. When considering assemblies of conjugated oligomers and polymers in the solid state , the frontier molecular orbital energies will be offset from those of the isolated
29 molecules by di fferent amounts depending on material dependent intermolecular interaction energies. However the basic transport energy levels sh own in Figure 1 5 will schematically look the same. The molecular structures and frontier energy levels, E HOMO and E LUMO , of several OSC materials determined by ultraviolet photoelectron spectroscopy and optical absorption are shown in Figure 1 6. These values have been the topic of extensive study within the OSC community and measured values differ slightly between laboratories. As a result, the values collected from the literature [1 8 31] and presented here are approximate and are assigned error tolerances on the order of 0.1 0.2 eV. Figure 1 5. Schematic of the bonding in ethylene. A) Atomic configuration of ethylene. B) Hybridized electrons in carbon: the electron distributi ons of hydrogen s electrons are shown in gray, hybridized carbon sp 2 electrons are blue, and carbon p electrons are purple. C) Molecular orbital energies of and * configurations of ethylene.
30 The frontier energy levels of organic solids exhibit a strong dependence on small changes in chemical composition. The family of oligoacenes demonstrate s this complexity . When a fused phenyl group is added to Tetracene to form Pentacene, the band gap is reduced by ~ 0.3 eV and the E HOMO level is shifted closer to vac uum. When the fused Tetracene core is maintained and four phenyl groups are added to form Rubrene, the band gap is instead increased and the resulting E HOMO level is deeper . The family of metal phthalocyanines also embod ies a complex relationship between m olecular chemistry and physical properties, as demonstrated by the shifting energy levels and band gaps for Zinc Phthalocyanine (ZnPc) and Lead Phthalocyanine (PbPc) . Figure 1 6. Molecular structures and frontier energy levels of several organic semiconductors.
31 1.4.2 Optical Absorption The close relationship between molecular structure and physical properties such as energy levels and optical band gap presents a tremendous opportunity to optimize these structure property relationships for incorpor ation in optoelectronic devices. In the case of photovoltaics, one of the most important characteristics of semiconductor materials is optical absorption: the more photons a material can absorb , the more electricity it can produce . For achieving the best p erformance in OPV devices, an understanding of the physical relationships and familiarity with the properties of available materials are equally important. The structure property relationships of the oligoacene family have been studied for several decades and are well known. As a general rule of thumb, as the molecular conjugation length increases (i.e. moving from the two membered Naphthalene to the five membered Pentacene) the E HOMO level shifts closer to the vacuum level and the optical absorption onset is red shifted to lower photon energies  . This trend also holds in the more structurally complex dicyano v inyl substitu ted oligothiophene family. An excellent summary of the s y nthesis and characterization of these materials is given by Fitzner et al.  . These results are summarized in Figure 1 7. As we will see in Chapter 2, the careful selection of materials with different optoelectronic properties is key to maxi mizing OPV performance.
32 Figure 1 7. Trends in the optical absorption and E HOMO levels in terminally dicyano substituted oligothiophenes with different conjugation lengths. Data and figures from ref.  . In addition to the individual molecular structures, the absorption efficiency is also st rongly dependent on molecular arrangement in thin films. For planar OSC molecules such as ZnPc and D iindenoperylene (DIP), the orientation of the molecules within a film tical properties. As will be shown in Section 1.5, the configuration of OPV devices is such that incident light propagates in the direction perpendicular to the substra te surface. O ptical transitions are strongest when the optical electric field (perpendic ular to the , i.e. E is maximized. I t is therefore advantageous to have absorbing films with molecules arranged such that their transition dipoles are oriented in t he plane of the growth surface. For planar molecules, the transition dipole moment is most often along the conjugated molecular plane, so the case where planar molecules lie down flat on the substrate is ideal for efficient light absorption. SchÃ¼nemann et al.  have demonstrated this phenomenon nicely in a study of DIP and ZnPc films grown on oxide , gold, and organic surfaces. The peak extinction
33 coefficient of thin DI P films increased three fold when the molecules were oriented in a more flat lying configuratio n, while that of ZnPc increased by a factor of two . Rand et al.  reported similar findings for ZnPc films deposited on a CuI template layer. In both cases, the molecular orientations were verified with spectroscopic ellipsometry and/or x ray diffraction experiments. These characteristics are shown in Figure 1 8 . Figure 1 8 . Absorption efficiency and extinction coefficient (k) for DIP and ZnPc films , respectively, grown on different substrates. A) DIP grown on a PTCDA molecular template layer absorbs light more efficiently, from ref.  . B) ZnPc grown on a CuI template has increased optical constants, from ref.  . Given the sensitivity of optical properties to film structure exhibited in these well co ntrolled case studies, one can imagine that a truly vast spectrum of optical properties are exhibited by different OSC materials . The breadth of these characteristics are demonstrated by the optical absorption coefficient, which is related to the imaginary component of the complex refractive index k through the relation k T he
34 measured absorption coefficient spectra of six OSC materials are shown in Figure 1 9. OSC materials have very strong optical transitions resulting in peak excess of 10 5 cm 1 , which are more than an order of magnitude greater than most conventional semiconductor materials. These high characteristic absorption lengths less than 100 nm, thereby making OPV and ultrathin film technology . Figure 1 9. Absorption coefficient spectra of six organic semiconductor thin films. Similar to the frontier energy levels and optical band gaps depicted in Figure 1 6, the features of OSC absorption spectra are heavily dependent on the chemical stru ctures and film properties of the constituent molecules. For instance, the phthalocyanines ZnPc, PbPc, Boron Subphthalocyanine chloride (SubPc), and Indium Phthalocyanine chloride (InPc) have very similar molecular structures with the exception of the cent ral metal atom, yet their absorption spectra feature different peak absorption
35 energies and spectral shapes. While the magnitudes of the strength of OSC materials and OPV devices, the absorption bandwidths are relatively narrow. In particular, the band widths of the prima ry absorption transitions of DBP and SubPc films are less than 200 nm. An OPV device with either of these active layer materials would be transparent to a substantial portion of the solar spectrum. However, combining the absorption spectra of multiple OSC materials in a single device can circumvent this limitation. This topic will be addressed in Chapter 2. 1.4.3 Electrical Conduction The one constant between all of the molecular structures shown in Figure 1 6 is unhybridized 2p z electrons in these bonds are shared amongst neighboring atoms and their respective wavefunctions are delocalized throughout the molecule. These strong overlaps result in covalent C H and C=C bond energies on the order of 100 200 kcal/m ol  , or 5 10 eV/atom  . On the other hand, molecule molecule electroni c interactions are relatively weak due to their non covalent nature . For uncharged molecules, stabilizing intermolecular interactions are limited to van der Waals forces where small local charge redistributions accommodate molecular packing. The strength o f the se interactions drops with intermolecular separation distance r as 1/ r 6 . The equilibrium intermolecular distance can be empirically quantified by the Lennard Jones potential, which balances the stabilizing van der Waals force with electrostatic repuls ion, (1 1)
36 where A and B are material specific constants and n intermolecular repulsion due to the Pauli exclusion rule. In uncharged, weakly polar systems such as OSC films the magnitude of van der Waals stabilization forces is < 0.1 eV/atom  . The relative extents of electron orbital overlap resulting in drastically different intramolecular and intermolecular bonding energies suggest that charge conduction in organic films is limited by intermolecular c harge transfer. Charge carrier mobility in OSC films and devices is cha racterized by an intermolecular hopping mechanism where a significant potential barrier must be surmounted in order for charge carrier s to flow between adjacent molecules. As a result, the mobility in many OSC films exhibits a temperature dependence approximated by  (1 2) In contrast with conventional semiconductors, the mobility of carriers in OSC materials is enhanced with increasing temperature because of increased phonon activity. These phonon vibrations assist charge transfer processes between molecules rather than scatter deloca lized charge carriers as in conventional semicond uctors, whose mobility decreases with temperature. Despite being limited by intermolecular hopping processes, OSC materials can achieve charge carrier mobilities above 0.1 cm 2 V 1 sec 1 [39,40] , although values in the 10 7 to 10 3 cm 2 V 1 sec 1 are more common [41 43] . Carrier mobility can in some cases be enhanced by altering molecular orientations within a film  or by the incorporation of molecular dopants [26,45,46] .
37 1.4.4 Excitonic Characteristics After an energetic photon promotes an electron across the optical band gap, the differences between conventional and organic OSC materials become even more apparent. The attractive force between an excited electron and its parent lattice or molecular site can be approximated by the Bohr radius and associated Coulombic attractive energy. The Bohr radius r B of an excited electron in a material is related to the ground state hydrogen radius, r o , according to (1 3) where r is the relative dielectric constant, m e is the mass of an electron, m e * is the electron effective mass in the material , and r o is given by (1 4) where o is vacuum permittivity, e is the elementary charge. In Si, r = 11 and m e * = m e /3 , giving r B (Si) = 17 Ã…. This is large relative to the Si lattice const ant, signifying that the excited electron is weakly bound to its starting location. For OSC materials, the relatively weak intermo lecular interaction energies lead to low bulk dielectric constants in the 2 5 range [25,47,48] and electron effective masses > m e [32,49] . As a result, r o values are on the order of one nanometer or less and excitons are localized to a single molecule. These excitonic phenomena can also be quantified in terms of the Coulomb attractive energy between an electron and hole, given by (1 5 )
38 For conventional semiconductors, E B is comparable with or less than k B T near room temperature . These weakly bound excitons are called Wannier Mott excitons. The correspond ing exciton binding energy in OSC materials is a few hundred meVs, an order of magnitude greater than the thermal ener gy in the system at room temperature. T hese t ightly bound, localized excitons found in OSC materials are classified as Frenkel excitons. A charge transfer exciton is an intermediate case where the exciton is localized to neighboring molecules. These are sc hematically illustrated in Figure 1 10. Figure 1 10. Excitons in a solid. In order of increasing binding energy: A) Wannier Mott, B) Charge transfer, and C) Frenkel excitons. Excitons are chargeless particles which can diffuse throughout a solid and ar e unaffected by electric fields less below the MV/cm range  . Similar to charge carrier conduction, the diffusion of excitons is characterized by a n in termolecular hopping process which can be described by two mechanisms: FÃ¶rster transfer and Dexter
39 transfer . In FÃ¶rster transfer, dipole coupling between adjacent molecules facilitates non radiative resonant energy transfer between them. Energy or exciton transfer occu rs when the fluorescence spectrum of the excited molecule over laps with the absorption spectrum of the ground state molecule, and the intermolecular transfer rate between the pair, k F , is expressed as  (1 6) where is the mean exciton lifetime, R is the intermolecular separation distance , and R 0 is given by  (1 7) where PL = k R /( k R + k NR ) is the photoluminescence efficiency, is the dipole orientation factor, n is the refractive index, is the wavelength, F D is the normalized fluorescence spectrum, and A is the absorption cross section. The R 6 dependence of k F highlights the importanc e of the electronic coupling between the two bodies. This transfer mechanism can be efficie nt at length scales up to 10 nm  . Exciton migration in a solid can also occur by direct exchange of electrons between molecules, as de scribed by Dexter  . The rate of this process is determined by the electron wavefunction overlap between neighboring species . Since this mechanism requires the physical overlap of electron clouds, trans fer is only effective at separation distances up to a couple of nanometers. The rate is exponentially dependent on intermolecular separation R and takes the form, (1 8)
40 The efficiency of these transfer processes is highly dependent on the physical properties of OSC materials. These can be practically and quantitatively described by the exciton diffusivity D = L D 2 / , where L D is the characteristic exciton diffusion length and is the exciton lifetime after which the exciton decays to the ground state, typically on the order of nanoseconds. Values of L D are highly material dependent and most OSC materials have diffusion lengths < 20 nm [55 57] . This parameter is also highly sensitive to molecular orientation and film crystallinity, which can be alt ered by various processing methods. For instance, Verreet et al.  engineered Rubrene film growth to increase the size of crystalline domains, resulting in L D values in excess of 200 nm. 1.5 Organic Photovoltaic Devices The optical and electronic processes d escribed in Section 1.4 show that OSC materials can effectively perform the two major processes required for efficient photovoltaic power conversion: light absorption and charge carrier conduction. However, the route between photons and electrons/holes is blocked by the excitonic nature of the excited state. In order to deliver current to an external load, excitons must be separated into electron and holes. The barrier to this challenge was first overcome by Harima in 1984  and Tang in 1986  with the development of the donor/acceptor heterojunction. 1.5.1 Donor/Acceptor Heterojunction As discussed in Section 1.4, OSC materials are ex citonic in nature in that the excited state is a tightly bound electron hole pair, or exciton. Unlike conventional semiconductors and solar cell s, photocurrent generation in OPV devices requires an exciton dissociating interface where excitons can separate into charge carriers. A type II heterojunction between two materials with E HOMO and E LUMO offsets fulfills this
41 requirement, as illustrated in Figure 1 11. When an exciton reaches the interface, the energetic alignment of the frontier energy levels facilitates a charge transfer process between the two materials at the molecular level. Figure 1 11. Energy level d iagram of a donor/acceptor heterojunction. During this charge transfer, or exciton dissociation, process , the electron is transferred to the material with a deeper E LUMO or greater electron affinity, called an electron acceptor. The material with a lower electron affinity is therefore referred to as an electron donor, or donor. Exciton dissociation can occur from either side of the heterojunction, so absorption by either material can result in charge carrier generation. For efficient exciton dissociation, the LUMO and HOMO offsets should b e greater than the exciton binding energies of the donor and acceptor materials , respectively. 1.5.2 P hotocurrent Generation Processes Now, we can envision photovoltaic conversion of photons to electrons as a multi st ep process in OPV devices. Light absorption results in the creation of excitons, which must arrive at a donor/acceptor interface in order to dissociate into charge carriers.
42 Charge carriers must then be conducted outside of the active layers to do work on an external system. We can thus define the external quantum efficiency (EQE) of this process, which is the ratio of photogenerated electrons to incident photons, as t he product of the efficiencies o f these four individual steps : light absorption ( A ), exci ton diffusion to an interface ( ED ), exciton dissociation or charge transfer ( CT ), and charge carrier collection ( CC ). These processes are illustrated schematically for donor absorption in Figure 1 12. Figure 1 12. Multi step photocurrent generation mechanism in a donor/acceptor heterojunction.
43 1.5. 3 OPV Device Architectures One limiting factor in this multi step process is the difference in the characteristic length scales of light absorption and exciton diffusion in OSC materials. Peak absorption co efficients on the order of 10 5 cm 1 (Figure 1 9) correspond to characteristic absorption lengths > 50 nm. On the other hand, most OSC materials have exciton diffusion lengths < 10 nm. In order to achieve efficient photocurrent generation, both A and ED should be maximized. For a multilayer structure with neat donor and acceptor layers, high efficiencies for both of these processes are mutually exclusive. This phenomenon is known as the exciton diffusion bottleneck. The bottleneck can be overcome, howeve r, by forming an intimately mixed donor: acceptor blend, or bulk heterojunction (BHJ) [46,59,60] . In the BHJ structure, the donor and acceptor materials each form ~nanometer size d domains which allow for efficient exciton diffusion to a D/A interface. Furthermore, the relative amounts of donor and acceptor in the blend layer can be engineered to alter the optoelectronic properties of BHJ films  and OPV devices  . Optimized BHJ devices can achieve both high A and ED and have become the structure of choice in most high efficiency OPV devices. An OPV device is formed by sandwiching the donor/acceptor heterojunction structure between electrodes with different work functions. One of the two electrodes must be transparent so that visible light can enter the device and be absorbed by the o rganic layers. In the standard device configuration used in this work, a glass substrate pre coated with a transparent, high work function indium doped tin oxide (ITO) film is used as the anode. Organic photoactive layers are grown on this substrate by met hods discussed Chapter 2, and an opaque, low work function Aluminum cathode is deposited
44 to complete the device. Buffer layers can also be inserted between the electrodes and photoactive layers in order to improve charge carrier extraction, energy level al ignment, and performance stability. Schematics of planar heterojunction (PHJ) and BHJ device stacks in this standard configuration are shown in Figure 1 13. Figure 1 13. A) Planar and B) Bulk heterojunction OPV device architectures used in this work. 1. 6 Dissertation Overview Fabrication and characterization methods used to form OSC films and OPV devices will be de scribed in Chapter 2, along with a discussion of two key inherent limitations to OPV device performance. Chapter 3 will investigate methods of modifying donor film and interlayer properties by inserting multi functional anode interlayer s (AIL) between the ITO anode and subsequently deposited donor layers. The extension to OPV device performance will be the topic of Chapter 4. Chapter 5 will extend on this
45 analysis by focusing on a single donor material and isolating individual AIL/donor phenomena and their effects on OPV device performance. Experimental work on series connected tandem devices will be presented i n Chapter 6 . Here, two individual D/A HJ devices are combined in a single tandem cell in order to harness the optoeletronic prope rties of multiple donor materials within the same device. Finally, Chapter 7 summarizes the work in the context of the current status of OPV and suggests further work in this field.
46 CHAPTER 2 FABRICATION AND CHARACTERIZATION OF ORGANIC THIN FILMS AND PHOT OVOLTAIC DEVICES This chapter will explore the characteristics and performance of OPV devices from a more practical angle. First, the major fabrication methods used in this thesis will be described. Next, film and device characteri zation methods will be su mmarized with an emphasis on OPV device characterization and figures of merit. After developing this quantitative framework, the historical progress of OPV performance and continuing challenges to be addressed in this thesis will be discussed in more detai l. 2.1 Thin Film and Device Fabrica tion Methods Two major classes of OSC materials can be identified based on their different physical structures and associated processing methods: polymers and small molecular weight oligomers, or small molecules. Both cla sses of OSC materials exhibit the optoelectronic properties and characteristics introduced in Chapter 1, but the way in which they can be processed to fabricate optoelectronic devices are different. Polymers are soluble in many organic solvents and can be d eposited into thin films from solution. Compatibility with solution processing requires that both donor and acceptor species are soluble in a common solvent. Solutions of donor and acceptor, as well as different buffer layers, can be deposited by spin coa ting followed by evaporation of solvent material to leave a film of the desired species. While spin coating and spray deposition are attractive methods compatible with high volume manufacturing methods, the studies described in this thesis primarily rely o n vacuum processing of small molecules. 2.1. 1 Vacuum Thermal Evaporation Small molecules have historically been insoluble in solution and are instead deposited onto substrates by thermal evaporation in the gas phase. The recent
47 development of several solut ion processable small molecules has bridged this gap, but none of these new hybrid materials are studied in this work. A widespread method for depositing films of metals and small molecu les is vacuum thermal evaporation (VTE). Film growth proceeds in a hig h vacuum chamber (base pressure ~ 10 7 Torr) by resistive heating of source material containers, or boats. Once the material specific sublimation temperature is reached, a molecular (or metallic) vapor is formed and films are grown on downward facing subst rates positioned above the source material . The molecular mean free path ( MFP ) in such a system can be expressed as (2 1) Here, k B T is the gas temperature, P is the system pressure, and 2 is the collision area between two molecules  . At a pressure of 10 6 Torr, nanometer sized molecules which sublime at 150 500Â°C have mean free paths in excess of 1 m. The distance between the source materials and substrate in a typical VTE syst em is on the order of a few tens of centimeters, so transport can be considered ballistic in nature. A schematic of a VTE deposition system is shown in Figure 2 1. Quartz crystal microbalances (QCM) are used to monitor growth rates and film thicknesses, an d the power supplied to individual electrodes can be adjusted to control the rate of deposition. Furthermore, multiple materials can be deposited simultaneously to form doped or mixed layers for a bulk heterojunction structure. In this case, QCMs are caref ully positioned to only observe one material at a time so that two or more such QCMs can monitor separate materials individually. Once the desired deposition rate(s) is (are) achieved, a shutter below the s ample is opened to begin film growth on the substr ate .
48 Using th is strategy, multiple layers can be deposited sequentially to form the desired film or device stack. Figure 2 1. Schematic of a vacuum thermal evaporation chamber. As suggested by Figure 2 1, VTE is not very economical in terms of material s usage. In the configuration shown, o nly a small fraction of the sublimed cone is collected on the substrate for analysis. Linear sources can be substituted for point sources to increase this yield. Nonetheless, VTE is an excellent method for fabricating precise multilayer films and device structure s. Several organic, metal, and oxide materials can be deposited in the system in a single batch without breaking vacuum, allowing complex structures which cannot be achieved by solution processing alone . VTE als o offers a greater degree of control over doped and mixed layer structures as well, which can be crucial in certain device structures.
49 2.1. 2 Thermal Gradient Sublimation As is the case for conventional semiconductors, OSC film and device properties are str ongly dependent on the composition of the semiconducting active layers. While defects in conventional semiconductors are readily ionized and affect the bulk conductivity, the weak intermolecular forces in OSC materials make electrical defect incorporation and controlled doping an ongoing challenge. Instead, impurities usually occupy molecular interstitial sites and disrupt molecular packing. This can have drastic effects on charge carrier mobility, as evidenced by studies of impuri ty incorporation in PTCDA films  . One preven ta tive step which can reduce impurity incorporation is purification of organic powders by thermal gradient sublimation. In this process, source materials purchased from external suppliers (typical starting purities range from 85% to 99.9%) or synthesized in house are loaded into a quartz tube . This tube is one of three, the other two being used as collection tubes for purified m aterial, which are placed within a larger diameter tube located in a three zone furnace. Quartz wool is placed at the end of the large tube between the purification zones and the pumps to keep organic material from getting into the pumps, and a metal scree n is placed downstream of the quartz wool to keep it from doing the same. Purifications can be run at high vacuum with a turbomolecular pump or in the presence of a controlled inert gas flow. The experimental set up and progression is illustrated in Figure 2 2, along with source and purified C 60 .
50 Figure 2 2. Summary of thermal gradient sublimation. Progression of purification from the beginning A) to the final purified product B). C) Source C 60 material (MER Corp., 99.9%, left) and high purity C 60 after one purification run with N 2 flow (right). Image courtesy of author. The quartz tubes are etched with diluted hydrofluoric acid and cleaned with organic solvents between runs before being baked out at T > 600Â°C to remove any remaining organic residues. Or ganic source material is then loaded and heated to its
51 sublimation temperature while the temperatures of the other two sections of the furnace are controlled to create a decreasing temperature gradient. Sublimation temperatures are material specific and range from 150 to 450 Â°C. The temperature gradient is optimized experimentally so that more volatile impurities are deposited farther down the temperature gradient outside of the collection zones while he avy impurity molecules remain in the source tube. The number of required purification steps varies from one to four depending on the purity of the starting material and the difficulty of separating pure material from impurities; subsequent steps are carrie d out at s uccessively elevated temperatures to further remove light impurities. Y ields typically range from 50 90 % by weigh t for a single purification run. Information regarding the source of organic materials and their purification is summarized in Table 2 1. Purified organic powders are loaded into evaporation source boats illustrated in Figure 2 1. These powders are deposited sequentially to form a multilayer structure illustrated schematically in Figure 1 13. In order to form a photovoltaic device, the se organic layers must be sandwiched between electrode layers. Once the desired organic layers have been deposited onto the substrate pre patterned with a transparent electrode, an opaque metal cathode layer is deposited through a finger like mask to form a crossbar geometry where the transparent anode, organic layers, and cathode from a 4 mm 2 overlapping active layer.
52 Table 2 1. List of organic semiconductor materials used in this work and their purification methods. Material Supplier Source purity (w t. %) No. purification runs Vacuum or Nitrogen Bathocuproine (BCP) Sigma Aldrich 96% 0 N/A Bathophenanthroline (BPhen) Sigma Aldrich 97% 1 Vacuum Boron subphthalocyanine chloride (SubPc) Sigma Aldrich 85% 2 Vacuum C 60 MER Corp. 99.9% 1 Nitrogen Copper Phthalocyanine (CuPc) Sigma Aldrich 95% 2 Vacuum Tetraphenyldibenzoperiflanthene (DBP) Luminescence Technology Corp. >99% 1 Vacuum Indium Phthalocyanine chloride (InPc) Sigma Aldrich 90% 2 Vacuum Lead Phthalocyanine (PbPc) Sigma Aldrich 80% 2 Vacuum Bis(naphthalene 1 yl) bis(phenyl) benzidine (NPB) Luminescence Technology Corp. >99.5% 0 N/A Pentacene Sigma Aldrich 99% 1 Vacuum Tetracene Sigma Aldrich 99.99% 0 N/A Zinc Phthalocyanine (ZnPc) Synthesized in house >99% 1 Vacuum 2.2 OPV Device Characterization This section wi ll be dedicated to defining metrics for evaluating photovoltaic device performance and describing how key figures of merit are determined in the laboratory. Once these have been defined, a more quantitative comparison between conventional and organic PV technologies will follow. This will lead to a discussion of key limitations of OPV materials and devices, which motivate the scope of this work. 2.2.1 OPV Performance Metrics In the dark, OPV devices based on a donor/ac ceptor (D/A) heterojunction (HJ) function as p n junction diodes with rectification ratios often > 10 5 at V = Â± 1 V at room
53 temperature [50,65,66] . A popular equivalent circuit model, the modified Shockley diode equation (2 2 ), approximates an OPV device as a diode characterized by saturation current density J S and ideality factor n , in parallel with a photocurrent generation source. Large shunt and relatively small series resistors with resistances R SH and R S approximate leakage pathways and bulk resistivity in the device , respectively . (2 2 ) More detailed analysis of OPV dark current behavior has led to the development of a more rigo rous treatment of diode characteristics [66,67] ; however this analysis will be omitted herein. A semi log plot of dark current behavior (i.e. J PH = 0) typical of OPV devices is shown in Fig ure 2 3. In addition to the dark current, O PV devices exhibit strong photocurrent generation under illumination (with intensity P O ) as discussed in Section 1.5 . The total current density flowing through a PV device is therefore given by the sum of the dark current and photocurrent ( J PH ): ( 2 3 ) The most popular metric for evaluating solar cells is the power conversion efficiency, P , defined as the maximum power generated by the cell to the power incident on the device, ( 2 4 ) where P MAX is the maximum power output and P 0 is the incident optical power. This quantity is determined by measuring the current voltage characteristics of the device under simulated solar illumination. Two key characteristics are the s hort circuit current
54 density ( J SC ), the current density flowing through the external circuit under no applied voltage bias, and the open circuit voltage ( V OC ) , the potential at which no current flows through the cell. Sample J V characteristics of a donor/ acceptor Copper Phthalocyanine (CuPc)/fullerene C 60 OPV device are shown in Figure 2 4. Figure 2 3. Experimental and fitted dark current using ( 2 2 ) for an archetypal OPV device. After determining the current density and voltage corresponding to maximum power generation, the fill factor FF is defined as (2 5 ) Now, (2 4 ) can be expressed as (2 6 ) This relation shows that P is equally dependent on th ree key performance parameters: J SC , V OC , and FF .
55 Figure 2 4. Photovoltaic behavior of a typical CuPc/C 60 OPV device. A) J V characteristic and B) Power generation quadrant. A closer look at the J V characteristics in Figure 2 4 shows that the voltage dependence of current density is greatest under forward voltage biases approaching the V OC . This curvature in the J V characteristic has the strongest influence on the V OC and FF parameters. Comparing the currents in F igures 2 3 and 2 4, we notice that the dark current and photocurrent reach similar magnitudes at forward bias voltages approaching the V OC . In this regime, the two currents compete with one another, and understanding the interplay between the opposing driv ing forces is critical to OPV device performance. Solving ( 2 2) and (2 3) for the V OC (i.e. J = 0, V = V OC ) under idealized conditions (low R S , high R SH ) gives (2 7 ) where q is the electron charge and J PH ( V OC ) is the photocurrent under open circuit conditions. Note that the V OC corresponds to the voltage bias where no total current flows through the device, i.e. photocurrent and dark current sum to zero. While rarely considered, photocurrent behavior under for ward bias also plays a role in determining A B
56 the V OC  , as demonstrated in (2 7 ). The V OC is also dependent on the dar k saturation current density J S . This parameter has been shown to be heavily dependent on the energy level offsets and intermolecular D/A interactions at the D/A HJ interface [69 71] . The nontrivial absorption spectra of OPV active layer materials (Figure 1 9) result in a strong spectral depend ence of photocurrent generation. D ifferent photoactive materials within the same OPV device will often have distinct absorption spectra. Therefore, spectral photocurrent measurements are an im portant tool for evaluating the behav ior of not only the OPV device but also individual layers within the multilayer structure . The most ubiquitous metric describing PV spectral response is the external quantum efficiency (EQE), defined as the fraction of incident photons that are converted t o electrons and collected at the electrodes. This is calculated as ( 2 8) where I PH is the measured photocurrent, P 0 is the incident optical power, h constant, and is the photon frequency. The internal quantum efficiency (IQE) can be calculated by measuring the absorption efficiency ( A ) of the device, allowing separation of the optical and electronic behaviors of the device. The EQE spec trum of a CuPc/C 60 OPV device is shown in Figure 2 5.
57 Figure 2 5. Short circuit EQE spectrum of a CuPc/C 60 OPV device alongside the AM 1.5G solar spectrum. Furthermore, the EQE and irradiance spectr a are related to the J SC according to the relatio n (2 9 ) Clearly, increasing EQE is essential to maximizing P by way of maximizing J SC . Equation (2 9 ) also illustrates the importance of a broad spectral response: the breadth of the product has a strong bearing on J SC . Focusing on the EQE and AM 1.5G spectra in Figure 2 5, for a given maximum EQE the J SC of this device could be increased by a broadening of the spectral response in order to harness low energy near infrared ( NIR ) photo ns. Since the flux of photons below the band gap of these materials is still considered in the P calculation, they are in a sense wasted and leave room for improvement. Furthermore, the agreement between the J SC obtained from (2 9 ) and that obtained from illuminated J V characteristics of the same device is a good measure of
58 the calibration of the light source used for illuminated J V measurements. This procedure will be addressed in the next section . 2.2.2 Measurement and Calibration Methods Current densi ty voltage ( J V ) measurements conducted in this work use an Agilent 4155C semiconductor parameter analyzer for electrical measurements and a 150 W Newport Xenon arc lamp as illumination source. Proper alignment of the reflecting mirror within the lamp hous ing and focusing optics result in a circular beam much larger than the 4 mm 2 devices used in this work. A small iris between the source lamp and device under test removes stray photons from the edges of the beam. Neutral density fi lters are also placed in the beam path to change the illumination intensity, which can also be altered by fine adjustments of the input power to the lamp. A schematic of this set up is shown in Figure 2 6. Figure 2 6. Measurement set up for illuminated J V data collection. Spectral measurements of both OSC films and OPV devices require monochromatic illumination and are performed with the lock in technique. White light generated by a quartz tungsten halogen lamp (Newport 70613NS) is monochromated
59 by dif fraction gratings in a monochromator (Newport 74100) before being chopped at 400 Hz to create an alternating signal . The resulting beam is collimated and focused onto a test device , film under investigation, or calibrated Si photodiode with known responsivity R measured with a current amplifier (Keithley 428) and a lock in amplifier (Stanford Research Systems SR 830 DSP). This measurement set up can be used to measure EQE from (2 8), as well as transmission and reflection through and by films. Furthermore the current amplifier allows for voltage biasing as well as current measurement, thereby allowing illumination wavelength and voltage dependent device measurements. The experimental set up for these spectral measurements is depicted in Figure 2 7. Figure 2 7. Equipment components and experimental arrangement for spectral measurements.
60 One issue complicating PV device characterization is the irradiance spectrum of the illumination source used for measurements. The true solar spectral irradiance depends strongly on cloud cover, atmospheric conditions, time of day, latitudinal location, etc. For this reason the photovoltaic community has an agr eed upon standard irradiance spectrum so that laboratories can follow similar characterization methods and generate reproducible results. For terrestrial applications like those studied in this work, the AM 1.5G spectrum is used according to the American S ociety for Testing and Materials (ASTM) G173 03 standard reference tables for hemispherical axis tilt  . Integrating the entire spectrum gives the total incident optical power used for P calculations: 100 mW/cm 2 . The AM 1.5G spectrum is plotted alongside the output spectrum of the Xenon lamp used in this work in Figure 2 8. Figure 2 8. Reference AM 1.5G and Xe solar simulator spectra.
61 While the Xe arc lamp does a good job of reproducing the reference AM 1.5G 800 nm, the spectra differ significantly in the NIR region. This can be mitigated by placing a color filter in front the device under test, such as the KG1 filter shown (broken blue line) . In order to report a meaningful P value, the differences between reference and source illumination spectra must be addressed. For a perfectly calibrated illumination source, the J SC values obtained from J V mea surements and short circuit EQE measurements by way of (2 9) would agree. When working with materials with different characteristic EQE spectra, this can be achieved by calculating the spectral mismatch factor M , given by [72,73] (2 10) where E and S refer to spectral ir radiance and responsivity, respectively, where I PH 0 . The subscripts ref , R , S , and T refer to the reference AM 1.5G spectrum, a reference detector (here, we use a calibrated Si PV cell), the lamp source used in the laboratory, and the cell under test, respectively. Since the pro duct of irradiance and respon siv ity spectra is a current I , M can also be expressed as (2 11) This equation can be broken into two halves: on the left hand side, the I ref,R /I S,R ratio describes the ratio of the photocurrents generated by the reference cell under the reference AM 1.5G and laboratory illumination source spectra, respectively. The right hand side is analogous to a test device. A standard method for calibrating a solar simulator is to experimentally set the left hand side of (2 1 1 ) , I ref,R /I S,R , equal to unity . This is achieved by measuring the EQE
62 J SC value from (2 9). Next, the reference cell is placed in front of the solar simulator lamp an d its position , the power supplied to the lamp, and/or neutral density filters are adjusted until the I SC = J SC A = I ref,R agrees with the calculated value. The location and calibrated I SC,R value are recorded and used until another calibration is performe d . In order to achieve AM 1.5G, 100 mW/cm 2 equivalent illumination intensity for a given OPV device with a different EQE spectrum, M is calculated from (2 10) and the right hand side of (2 1 1 ) can be forced to equal unity . Then, the reference cell photocurrent corresponding to 100 mW/cm 2 incident power under reference AM 1.5G conditions will be given by (2 11) The importance of this calibration procedure is illustrated in Figure 2 9. The device under test had M = 1.10 due it its relatively strong blue green response, coinciding with the spectral region where our simulator lamp generates more photons than the AM 1.5G standard. From (2 11), failing to reduce t he solar simulator intensity would cause an over excitation of the test device and artificially high reported J SC and P values would result . Indeed, the J SC calculated from (2 9) is roughly 10% less than that calculated from a J V measurement using an unco rrected light source. When the optical power, measured by the reference cell, is reduced according to (2 11) by either reducing the power supplied to the lamp or with neutral density filters between the lamp and test location , the difference between EQE and J V measurements disappears . The proper calibration of the solar simulator light source is key to obtaining correct and reliable
63 values in OPV device measurements. Even when a precise illumination intensity or reference cell photocurrent cannot be obtained, knowledge of the mismatch factor can explain the difference between J SC values calculated from EQE spectra and J V measurements. Furthermore, the output spectra of solar simulat ors is known to change with service time  , so regularly checking the lamp output spectrum and re calibrating is good practice. Figure 2 9. Short circuit EQE spectrum and J V characteristics of a DBP/C 60 based OPV device. In B) , the solar simulator lamp intensity was reduced according to (2 11) in the text, resulting in close agreement between J SC values obtained from the EQE and J V measurements. 2.3 OPV Device Characteristics 2.3.1 Quantifying Conventional versus Organic PV W hile significant progress has been made in the past three decades, OPV devices are still relatively inefficient compared with other PV technologies (Figure 1 3) . In the nearly 30 years since the introduction of the Tang cell  , champion OPV devices have only recently surpassed the P = 10 % mark, and the number of published reports of such devices can be counted on one hand [7 5,76] . By contrast, the first reported Si based PV cell had an P of 6 %  and cells with P > 10 % were fairly A B
64 common within the scientific community less than a decade layer  . In other words, it has taken more th an two decades for OPV technology to achieve th at which was achieved by Si based PV cells in less than ten years. For a better understanding of why OPV performance has improved relatively slowly, we can consider the performance characteristics of the best performing cells from the most recent compilation by Green et al.  . The confirmed J SC , V OC , fill factor, and P of cells composed of different active layer materials are listed in Table 2 2 . Clearly, the large st different between the best organic cell and the more efficient inorganics is the J SC , which is more than doubled in the Si based device. All three listed inorganic devices have a greater fill factor, but this difference accounts for at most a 25 % differ ence. Besides, OPV devices with fill factors > 0.7 are being reported more frequently [21,79 82] and will also be demonstrated in the results of this work. While the V OC of the best OPV device is actually greater than those of Si and CIGS based devices, the J SC in OPV cells is clearly the main difference between organic and conventional photovoltaics: the two to three fold difference in P is also characteristic of the difference in J SC values for these various devices. Table 2 2 . P hotovoltaic p erformance parame ters of the confirmed highest performing si ngle junction PV devices based on various active layer materials. Material Crystalline Si  GaAs  CIGS [8 5] Organic  J SC (mA/cm 2 ) 42.7 29.7 34.9 17.8 FF 0.83 0.87 0.79 0. 69 V OC (V) 0.71 1.12 0.72 0.87 P (%) 25.0 Â± 0.5 28.8 Â± 0.9 19.9 Â± 0.8 10.7 Â± 0.3
65 This difference in J SC can be directly traced to the fundamental differences between the physical properties of inorganic and organic semic onductors described in Chapter 1 . T he three dimensional crystalline covalent bonding network formed in inorganic semiconductors results in continuous bands of allowed energy states through which charge carriers can travel. This leads to high carrier mobilities and, more crucially in the context of p hotocurrent generation, large enough dielectric constants so that electrons excited across the band gap are not bound to their parent hole as is the case in OPV materials and devices. The requirement of the D/A HJ for efficient charge carrier generation li mits the efficiency of photocurrent generation in OPV compare d with conventional PV devices, thereby limiting J SC . 2.3.2 Fundamental Limitations of OPV Devices The D/A HJ requirement for efficient exciton dissociation in OPV devices leads to two key trade offs which must be mitigated in order to achieve devices with high quantum and power conversion efficiency devices. The first of these, the exciton diffusion bottleneck, was touched on briefly in Section 1.5. Since the length scales of light absorption and exciton diffusion differ by an order of magnitude in OSC materials, EQE is limited by the trade off between the efficiencies of light absorption and exciton diffusion, A and ED , in PHJ devices . Alternatively, donor and acceptor species can be deposited simultaneousl y from the vapor phase or in solution to form a bulk heterojunction (BHJ) where ED approaches 100% [86 88] . Figure 2 10 shows the A) J V characteristics and B) EQE spectra of PHJ and BHJ devices based on the CuPc/C 60 D/A pair.
66 Figure 2 10 . OPV device performance of CuPc/C 60 devices with PHJ and BHJ active layer structures. A) J V charact eristics and B) short circuit EQE spectra. Both devices shown in Figure 2 10 had the same total active layer thickness, with the PHJ consisting of 20 nm CuPc/40 nm C 60 and the BHJ consisting of a 60 nm thick , 1:1 by weight mixture of CuPc and C 60 . It is immediately apparent that the BHJ device generates significantly more photocurrent than the PHJ structure, wit h peak BHJ EQE J SC and ultimately P in the BHJ device. The photovoltaic performances of these dev ices are summarized in Table 2 3 . Table 2 3 . Photovoltaic performance parameters of CuPc/C 60 OPV devices with PHJ and BHJ active layer structures under 100 mW/cm 2 simulated AM 1.5G illumination. Active layer structure PHJ BHJ J SC (mA/cm 2 ) 4.4 Â± 0.1 7.8 Â± 0.1 FF 0.58 0.52 V OC (V) 0.47 0.50 P (%) 1.2 2.0 A B
67 However, there are also some advantages that the PHJ has over the BHJ device. The voltage dependence of photocurrent is lower for the PHJ device under reverse ad forward voltage bias . We can therefore conclude that a greater portion of the maximum generate d photocurrent is collected at short circuit in the PHJ device, whereas the BHJ device is characterized by stronger voltage dependent photocurrent loss mechanisms. This also manifests itself in the reduced FF exhibited by the BHJ device. Recalling (2 8 ) we can infer that since neither A nor ED are expected to show appreciable voltage dependencies in this voltage range, the CT CC product is diminished under forward bias in the BHJ device. Indeed, detailed analyses have shown that short circuit charge coll ection efficiency approaches 100% in PHJ devices and the minimal slope present in the J V characteristic can be attributed to photoconductivity in neat layers [66,89,9 0] . On the other hand, strong voltage dependent charge collection is common in BHJ devices [62,68,87,91] . This behavior can be approximated as (2 12 ) where d M is the BHJ mixed layer thickness and L C is the charge carrier collection length given by (2 13 ) where V bi is the built in potential and L C0 = L C ( V = 0 V). ( 2 12 ) predicts that CC will decrease with mixed layer thickness. Unlike a PHJ structure, in a mixed BHJ layer photogenerated electrons and holes occupy the same area and have a higher pr obability of recombining and diminishing photocurrent production . Under stronger
68 reverse voltage biases where carriers are extracted towards the electrodes, the probab ility of recombination decreases and CC increases as predicted by (2 12 ). While the original A versus ED bottleneck was overcome with the discovery and optimization of the BHJ structure, a different bottleneck is still present, this time featuring a trade off between A and CC . As discussed in Section 1.4, charge conduction b etween molecules is highly dependent on intermolecular orientation a nd interactions. When the ordering which is present in a neat film is disrupted by the addition of a second phase, interactions between like molecules are reduced and the charge carrier mo bility [41,61] as well as OPV device performance [62,87] suffer. A second trade off also arises from the nature of the D/A HJ struc ture. As discussed in Section 2.2, broad spectral overlap between OPV device EQE and AM 1.5G solar irradiance is necessary for efficient photocurrent generation. At present, a small minority of materials used in OPV device active layers have strong absorpt ion narrow band gap materials, another challenge which has slowed the realization of higher p in OPV devices is the mutual exclusivity of NIR absorption and high V OC . Several small molecule based OPV devices with NIR response have been reported, each with V OC < 0.6 V [92 94] , relatively low compared with more conventional devices . The requirement of a D/A HJ with sufficient offsets betw een frontier ener gy levels E HOMO and E LUMO places a constraint on the E LUMO of the donor material, E LUMO D . Specifically, it must be nearer to the vacuum level than E LUMO of the acceptor material, E LUMO A , by an amount greater than the donor exciton binding energy. The same is true for the
69 dissociation of excitons generated in the acceptor material, only in this case the E HOMO offset must be large enough for favorable dissociation. Furthermore, the V OC of OPV devices has been shown to be directly proportional to the D/A inte rface gap, DA = |E HOMO D | |E LUMO A | [69,70] . Therefore, a large | E HOMO D | is desirable in order to achieve a device with large V OC . For donor absorption, it follows that for a fixed minimum LUMO offset required for efficient exciton dissociation, donor materials can be chosen to have a shallow | E HOMO D | , narrow bandgap, and red shifted absorption or a deep | E HOMO D | , wide bandgap, an d high V OC . This trade off is schematically illustrated in F igure 2 11 A) . Figure 2 11. Schematic of a D/A HJ energy level structure and corresponding OPV device characteristics. A B C
70 Since the vast majority of OPV devices rely on strong donor absorption and use a fullerene as electron acceptor, this set of design rules severely limits the possible D/A combinations with both NIR absorption and high V OC , both of which are critical to max imizing P . Figure 2 12 B) shows the EQE spectra of four devices based on different donor materials, and their respective J V characteristics under simulated solar illumination are shown in Figure 2 12 C) . There is a clear correlation between the onset of ph otocurrent generation, which proceeds in terms of increasing wavelength from SubPc to DBP to ZnPc to PbPc, and V OC , which proceeds in the opposite order. As a result, all four of these devices achieved similar P values between 2.1 and 3.5 %. 2.4 Bridging the Gap: Routes Toward Enhancement Several approaches have been implemented to sidestep the inherent limitations to OPV device performance. Th e hybrid planar mixed (P M) HJ structure was first introduced by Xue et al. to directly circumvent the A CC chal lenge  . In the P M HJ structu re, a BHJ layer is sandwiched between neat donor and acceptor layers so that light absorption is economically distributed between neat and mixed layers in order to balance the critical components of EQE. Taking this concept a step further, graded heterojun ctions featuring a continually varying D:A ratio throughout the BHJ layer have also successfully been used in OPV devices [95,96] . Acceptor rich blends using the strongly absorbing fullerene C 70 have also resulted in highly efficient devices. Relying on acceptor absorption relaxes the constraint s of donor selection, and highly transparent donors with deep E HOMO energy levels have been used in devices with P > 5 % [97,98] . While this strategy has been effective thus far, questio ns regarding the scalability of C 70 production remain.
71 The development of new materials and processing methods will always be a crucial part of OPV development. The introduction of the squaraine family [99 ,100] has continued this trend, resulting in D/A HJ based devices with a unique combination of NIR absorption and V OC > 0.8 V. Solvent vapor annealing has also emerged as a facile method for controlling the properties of donor films in the bulk and at t he interfacial level [101 103] . The insertion of interl ayer s between the substrate and photoactive materials has also been demonstrated as an effective way to modify the optoelectronic properties of donor and acceptor layers. Organic exciton blocking [21,104] and exciton dissociating [105 107] interlayers directly alter excitonic behavior within the donor layer as a way to increase the fraction of excitons which dissociate into char ge carriers. Blended fullerene buffer layers [108,109] inserted between the acceptor and cathode have recently been implemented into OPV devices to reduce energetic barriers to charge collection. Analysis of various combinat ions of anode interlayer s and donor materials will be the focus of Chapters 3 5 in this work. The most successful development in circumnavigating the NIR/ V OC bottleneck has been the tandem device structure. By depositing two full devices sequentially with a series connection between them, the constituent materials can be chosen to broaden spectral response significantly. Xue et al.  pioneered this approach by altering the amounts of CuPc and C 60 in the two cells to tune the response of the indivi dual cells. A number of follow up studies have extended this approach to complimentary donor materials in the two subcells [111 113] . Tandem OPVs represent a tremendous opportunity to harness the improvements made in single junctions devices into an all -
72 encompassing device with greater ultimate efficiency. Tandem cells are investi gated in Chapter 6. 2 .4.1 Current State of the Art Through advances in device engineering, material synthesis, and novel processing methods, champion OPV device performance has come a long way since the first report of a P = 1 % cell. The current record is held by the German company Heliatek, who reported a certified P = 12.0 % multi junction device in 2013  . Among published reports, You et al. showed an NREL certified 10.6 % efficie nt tandem cell based on two polymers with absorption well into the NIR  , also in 2013. Among single cells, the highest published polymer based cell has P = 9.2 % with peak EQE of 0.80  . Xiao et al. recently published an evaporated small molecule based device with P M HJ structure having an impressive P = (8.1 Â± 0.4) %  , and Cnops et al. reported an 8.4 % efficient fullerene free cell in 2014  . S olution processed small molecule based devices have quickly reached similar levels with a maximum reported P = 8.9 %  .
73 CHAPTER 3 EFFECT S OF ORGANIC ANODE INTERLAYERS ON DONOR FILM PROPERTIES 3. 1 Introduction The optoelectronic properties of organic semiconducting materials are strongly dependent on film growth and processing conditions. There is a large amount of literature describing studies aimed at investigating the effects of different film and device fabrication methods on such properties as bulk film structure, surface morphology, optical properties, charge c arrier mobility, interface energetics, and more. In the context of organic solar cells, the conclusions of these studies are used in order to increase the efficiency of various processes necessary for power generation. The work described in the following two chapters will contribute to this knowledge base, both in terms of film properties and photovoltaic device performance. Specifically, the effects of inserting an organic interlayer between the ITO anode and the subsequently grown donor material are inve stigated in detail. The scope of these studies is limited to three vacuum deposited donor molecules, each of which is grown on multiple underlying interlayer structures. In Chapter 3 , the structural, optical, and electronic properties of neat donor films a re explored as a function of growth underlayers. In the Chapter 4 , the same underlayers are investigated in terms of their effect on photovoltaic device performance. 3. 2 Donor and Interlayer Materials Selection 3. 2.1 Donor M aterials The n u mber of donor ma terials used in OPV devices is vast and continually growing. When conducting a study such as this, several factors should be considered when choosing a set of donor species in order to ensure generalized conclusions.
74 These include optical absorption spectr a, energy level alignment, crystalline quality, and charge carrier mobility. After consideration of these factors, the donors Zinc Phthalocyanine (ZnPc), Lead Phthalocyanine (PbPc), and Tetraphenyldibenzoperiflanthene (DBP) were chosen for this study. The molecular structures of the se donor materials are shown in Figure 3 1. Figure 3 1. Molecular structures of donor materials used in this study. The phthalocyanine family of molecules has been studied extensively for optoelectronic device applications. In addition to being readily available, they exhibit strong optical absorption and sufficient conductivity for use in photovoltaic devices. The two features of metal phthalocyanines (MPcs) with the strongest effects on film growth are their strongly intera cting central metal atoms and relatively planar molecular structure. Both of these characteristics lead to strong dependencies of film characteristics on process conditions. Strong MPc molecule substrate interactions present in the early stages of film gro wth have significant consequences on film properties such as molecular orientation [34,64,117] , refractive index  , electrical conductivity  , and o ptical absorption  . Lead phthalocyanine, PbPc, has been shown to adopt various physical conformations in thin films grown under a range of processing conditions. This polymorphism has profound effects on the film
75 optoelectronic properties. Most importantly, the characteristic absorption spectra of PbPc films can be tailored by controlling the identity and amount of polymorphs within the PbPc film. This is critical to absorbing and converting near infrared solar radiation to electricity. PbPc film properties are investigated in Section 3. 3. Phthalocyanines with Cu and Zn central metal atoms, Cu Pc and ZnPc, respectively, have been studied for more than three decades and therefore have an especially deep knowledge base. ZnPc was chosen as a donor material in this study owing to the vast amount of scientific literature describing both its film prop erties and photovoltaic device characteristics. This collection of previous work serves as a platform for the results obtained in the studies described herein, both as a comparison for control studies and a benchmark for demonstrating the utility of the ne w approaches used in this study. These are described in 3. 4. Finally, DBP is an attractive material for OSC application because its energy level structure lends itself to OPV devices with high open circuit voltages. As such, it has been the subject of seve ral recent advances in OPV devices with higher power conversion efficiencies than those achieved by more commonly used small molecule donor materials. The effects of interlayer structures on DBP film properties are described in Section 3. 5 3. 2.2 Interlayer Materials Selection Criteria Before describing the effects of interlayer structures on donor film properties and OPV device performance, a list of criteria for the selection of anode interlayer (AIL) materials should be presented. As discussed in Section 1. 5 , the donor material is the primary absorber in the majority of OPV devices. As such, any additional species added to the control device structure should not reduce donor absorption by themselves being
76 strong absorbers. In the ideal case these materials should be transparent, wide bandgap semiconductors. Since these materials will be in place prior to donor layer deposition, the interlayer/donor heterojunction should not impede hole transport. Therefore the E HOMO levels of the interlayer and donor specie s should be well aligned. As a final criterion, hole transport should not be limited by the bulk conductivity of the interlayer film. In summary, interlayer structures should meet the following criteria : low absorption coefficient or be sufficiently thin t o minimize absorption form a hole conducting interface with the donor material hole mobility greater than or equal to the donor material 3. 2.3 Pairing Donor and Interlayer Materials Historically, the oligoacene family is one of the original classes of small molecule organic semiconductors to be studied for practical use. They are characterized by their highly ordered, polycrystalline film structures and exceptionally high charge carrier mobilities exceeding 1 cm 2 V 1 sec 1 . Their applicability to solar cells is however limited by characteristic low absorption coefficients, typically an order of magnitude less than those observed in more successful OPV materials. These properties make the polyacenes excellent candid ates for use as AIL species. Due to its extensive history in research, the optoelectronic properties of the donor ZnPc are well known. It is widely agreed that its E HOMO level is located (5.2 Â± 0.1) eV below vacuum [25 27] , while its field 3 cm 2 V 1 sec 1  . Both of these figures of merit are compatible with Tetracene, the four membered oligoacene , which has a comparable E HOMO [18,29] and greater hole mobility [39,40] than ZnPc. Similarly, the optoelectronic properties of the donor PbPc and the five ring oligoacene , Pentacene, are compatible with one another: the PbPc E HOMO of
77 (5.1 Â± 0.1) eV  is in good agreement with that of Pentacene [22,30,31] , while the high ly ordered film structure of Pentacene results in very high hole mobilities > 0.1 cm 2 V 1 sec 1 . The pr operties of AIL structures on PbPc and Zn Pc film properties are investigated in Sections 3. 3 and 3. 4 , respectively. Two characteristics differentiate DBP from the two MPc materials, thereby changing the AIL requirements. First, DBP has a significantly deeper E HOMO of 5.5 eV [20,21] , therefore requiring an AIL with a deeper E HOMO for proper energy le vel alignment . Second, the four out of plane phenyl g roups restrict structural ordering in DBP films. The consequences of these two key characteristics are investigated in Section 3. 5 by using Tetracene and Bis(naphthalene 1 yl) bis(phenyl) benzidine ( NPB ) as AIL species. The energy levels of AIL and donor species are shown in Figure 3 2. Figure 3 2. HOMO and LUMO energy levels of materials investigated in this study. Numbers are reported in eV relative to vacuum. HOMO values were measured by UPS with citations given in text, and LUMO values we re inferred from thin film absorption onset.
78 3. 3 The Effects of a Poly crystalline Pentacene Interl ayer on Lead Phthalocyanine Film Properties Lead Phthalocyanine (PbPc) is an attractive OPV donor material because it has an absorption range extending well i nto the near infrared (NIR), making it ideal for low energy photon absorption and photovoltaic response. PbPc films are known to crystallize in monoclinic and/or triclinic phases, which have different optoelectronic properties [117,118,120] . The structural properties of PbPc films exhibit strong dependence on substrate identity  , substrate temperature [117,120] , and film thickness [117,118] . These dependencies can be engineered to promote absorption well into the near infrared (NIR), which is desirable for low energy photon harvesting . With the goal of obtaining relatively thin (< 40 nm) PbPc films with strong red sh ifted absorption, we use a polycrystalline Pentacene template to guide PbPc growth in order to absorb NIR photons. Pentacene and PbPc films were grown at relatively low rates of 0.2 0.3 Ã…/sec. 3. 3.1 Optical Absorption The triclinic phase of the PbPc crys tal structure has 900 nm, and the amount of triclinic content of PbPc films has been enhanced by increasing the substrate temperature [118,120] , reducing the deposition rate [117,118,120] , increasing film thickness  , or any combination of these methods. To simplify the process of obtaining triclinic rich PbPc films, we keep the subst rate at room temperature and use a Pentacene interlayer between the ITO substrate and subsequently deposited PbPc films. The optical absorption spectra of PbPc films deposited on bare ITO and ITO/10 nm Pentacene are shown for a range of PbPc thicknesses in Figure 3 3 .
79 The absorption spectra of P b Pc films grown on ITO, shown in Figure 3 3 A) , all have a major peak at = 740 nm, corresponding to absorption of the monoclinic phase, and a shoulder extending past 900 nm, corresponding to the triclinic phase. The absorption spectrum of the 10 nm thick PbPc film indicates that the monoclinic crystal structure is dominant, with only a small contribution of the 890 nm centered triclinic phase absorption. This shoulder increases in magnitude relative to the monoclinic peak as PbPc thickness is increased from 10 to 40 nm, where the peak height of the triclinic phase absorption is still less than that of monoclinic phase absorption. These spectra indicate that the majority of the first 10 to 20 nm of the PbPc film grown on ITO adopts the monoclinic crystal structure. As more PbPc accumulates , a greater portion of the top section of the layer aggregates in the triclinic structure, in good agreement with previous absorption and x ray diffraction experiments [118,120] . When PbPc is deposited on 10 nm Pentacene covered ITO, the absorption spectra shown in Figure 3 3 B) indicate that both the monoclinic and triclinic phases are present in comparable amounts for all PbPc thicknesses investigated. For PbPc layers grown on Pentacene, the peak corresponding to the triclinic phase is very pronounced for just a 10 nm thick PbP c film: the triclinic phase contribution to the absorption spectrum is stronger relative to the monoclinic peak than it is for 40 nm thick PbPc grown on ITO. The relative amount of the triclinic phase again increases faster than the monoclinic phase with i n creasing PbPc thickness. In contrast with films grown on bare ITO , triclinic phase absorption accounts for a greater fraction of the total film absorption than the monoclinic phase for PbPc layers 20 nm or greater in thickness.
80 Figure 3 3 . T hin film absorption spectra of PbPc fil ms. Films were deposited onto A) bare ITO and B) ITO/Pentacene (10 nm). Sketches for the relative amounts of monoclinic and triclinic phases are included below their respective absorption spectra. In contrast to the gradient of monoclinic:triclinic ratio with PbPc thickness in PbPc films grown on bare ITO, the relative amounts of the two phases in PbPc films grown on Pentacene are more homogeneous in the film thickness direction. Pentacene templated nm. This is promising for OPV devices because donor thicknesses on the order of the and carrier transport. These phenomena will be investigated in Chapter 4 . A B
81 3. 3.2 Surface Morphology The thin film absorption spectra shown in Figure 3 3 clearly show that the most important changes in PbPc properties occur during the first stages of film growth. Therefore we study the surface morphology of ultrathin PbPc films. We use Si (100) s ubstrates with native oxide SiO 2 because this substrate has significantly lower roughness than ITO, allowing more careful imaging and analysis of sub nm PbPc films. Figure 3 4 shows the AFM height images of 10 Ã… thick PbPc films on bare SiO 2 . At this thickness, the film shows incomplete coverage and an RMS roughness (r RMS ) of 2.8 Ã…. Interconnected PbPc grains with a typical size of 20 nm are evenly distributed on the surface, suggesting uniform nucleation of PbPc molecules throughout the sampled area. Figure 3 4. AFM height images of 10 Ã… thick PbPc films grown on bare SiO 2 . The surface morphologies of ultrathin PbPc films grown on Pentacene coated SiO 2 are shown in Figure 3 5 . The neat 5 nm thick Pentacene film, Figure 3 5 A) , forms well ordered, terraced molecular structures  with ledges < 2 nm in height, in agreement with the length of a single upright standing Pentacene molecule [121,122] .
82 When 2 Ã… PbPc is deposited on Pentacene, Figure 3 5 B) shows that the PbPc molecules aggregate and form clusters at the terrace ledges. The line scan across this film shows that while the terraced Pentacene structure is un changed with the addition of PbPc, 5 to 10 nm thick PbPc clusters form despite a nominal PbPc thickness of only 2 Ã…. These clusters average 50 nm in diameter, significantly larger than those formed on the bare SiO 2 substrate. As more PbPc molecules are dep osited on the surface in Figures 3 5 C) E) , more clusters form at Pentacene terrace ledges until the ledges are fully occupied. These clusters do not increase in size until the Pentacene ledges are completely filled, after which the PbPc clusters extend inw ard toward the centers of the larger Pentacene grains [ Figure 3 5 E) ] . Comparing the 10 Ã… thick PbPc films on bare SiO 2 ( Figure 3 4 ) and Pentacene coated SiO 2 [ Figure 3 5 D) ] , r RMS increases by an order of magnitude from 2.8 Ã… to 33 Ã…. This suggests that the crystalline Pentacene template layer imparts a higher degree of crystallinity in the subsequently deposited PbPc layer. Furthermore, the evolution of the Pentacene/PbPc surface with increasing PbPc thickness indicates weak interaction between Pentacene and PbPc molecules, allowing adsorbed PbPc molecules to diffuse across the flat terraces before encountering a high energy ledge. Once here, the PbPc molecules are immobilized and form triclinic rich aggregates, revealed by the absorption spectra described in Section 3. 3.1.
83 Figure 3 5 . Evolution of surface morphology with PbPc thickness on Pentacene. Films have A) 0 Ã…, B) 2 Ã… with line profile, C) 5 Ã…, D) 10 Ã…, and E) 20 Ã… thick P bPc layers. 3. 4 The Effects of Tetracene Anode Interlayer s on Zinc Phthalocyanine Film Properties In this section, the properties of ZnPc thin films grown on inorganic indium tin oxide (ITO) and silicon dioxide (SiO 2 ) substrates are compared with those of ZnPc films grown on Tetracene and poly(dioxythiophene):poly(styrene sulfonate) (PEDOT:PSS)/Tetracene anode interlayer s (AILs). PEDOT:PSS is a well known hole transport mate rial popularly used as an anode interlayer in polymer based OPV devices. Here it is used to modify the morphology of a subsequently deposited Tetracene film in order to create a dual AIL structure for subsequently deposited ZnPc layers. Tetracene and ZnPc film growth rates were in the range of 0.5 1 Ã…/sec.
84 3. 4.1 Optical Absorption ZnPc exhibits strong absorption across a broad spectral range of 600 nm < < 800 nm. The absorption spectrum of three identical 25 nm thick ZnPc films grown on bare ITO, ITO covered with 10 nm Tetracene, and ITO covered with 40 nm PEDOT:PSS + 10 nm Tetracene are shown in Figure 3 6 A) . The total integrated absorption is least for t he bare ITO substrate, and this figure increases by 6% for ITO/Tetracene and by 21% for ITO/PEDOT:PSS/Tetracene AIL structures. Stronger absorption in these films signifies an increased extinction coefficient, which is directly related to the extent of mol ecular packing and intermolecular interactions in absorbing organic films [34,35] . Figure 3 6 . Optical absorption spectra of ZnPc films. The data in A) was normalized to the low energy peak and re plotted in B) . The dominant features of the control (bare ITO) film are two peaks centered at = 640 and 710 nm corresponding to aggregated and dilute ZnPc species , respectively. Studies of ZnPc  and CuPc  based blends have demonstrated that monomolecular absorption (lower energy) increases with decreasing CuPc or ZnPc A B
85 content relative to transitions corresponding to higher order aggregates (higher energy) . This is more clearly observed in Figure 3 6 B) , where the absorption spectra have been normalized to the low energy peak. The aggregated state absorption peak of ZnPc grown on ITO is just 3% greater than that of the monomolecular peak. In ZnPc films grown on Tetracene or PEDOT:PSS/Tetracene AIL struct ures, aggregated state absorption reaches 22% and 23% greater, respectively, by this metric. In addition to the increased contribution of the intermolecular state to absorption, the spectra are also broadened relative to the bare ITO based film. ZnPc film s grown on Tetracene containing AIL structures show a wider shoulder on the high energy side of the absorption spectrum, signifying a greater extent of intermolecular interaction and aggregation. Furthermore, the separation between monomolecular and higher order aggregation peaks increases with the add ition of either AIL structure. Fitting these spectra with three peaks yields a high energy peak centered at = (637 Â± 2) nm for all three films. The film grown on ITO has a low energy peak center at = 710 n m compared with = 717 nm for ITO/Tetracene and = 716 nm for ITO/PEDOT:PSS/Tetracene. The increased peak separation is another indication of increased intermolecular interaction and ordering in the ZnPc films grown on AIL structures relative to the film grown on ITO. 3. 4.2 Bulk Structure: X ray Diffraction The bulk structure of planar phthalocyanine films have been studied extensively by x ray diffraction (XRD). Vacuum deposited films grown on non interacting substrates including ITO and SiO 2 under ambi ent conditions display prominent features in the 6 < < 9 degree range, corresponding to interplanar distances on the order of 10 15 Ã… [34,35,61,123,124] . These single peaks correspond to f
86 Pc stacking where the long molecular axis, on average, is aligned normal t o the substrate surface with molecular system overlap in the direction perpendicular to the substrate normal . Under certain conditions the Pc molecular orientation can be altered such that the system interactions dominate in the vertical direction, normal to the substrate plane. These motifs are sketched in Figures 3 7 A) and B) for edge on and face on, respectively. This face on stacking results in a significant reduction in intermolecular spacing in the direction normal to the substrate surface, or d spacing in terms of XRD. Therefore, the XRD signal for edge on molecular orientations is expected to have higher angle characteristic XRD responses [34,35] . Figure 3 7 . Schematics of ZnPc molecular orientations and XRD spectra. A) E dge on and B) face on ZnPc molecular orientation s relative to a flat substrate. C) X ray diffraction spectra of 150 nm thick ZnPc films grown on SiO 2 with an d without AIL structures. A B C
87 The XRD patterns of 150 nm thick ZnPc films grown on bare SiO 2 , SiO 2 /10 nm Tetracene, and SiO 2 /40 nm PEDOT:PSS/10 nm Tetracene are shown in Figure 3 7 C) . The control sample shows the predicted behavior for ZnPc films grown on non interacting substrates: a single peak centered at 2 = 6.9Â° corresponding to edge on growth and an interplanar molecular spacing of 12.7 Ã…. This peak has been assigned to first or der ( h 00) planes of edge on ZnPc film growth  . Focusing on the 2 < 10Â° peak, the feature evolves into a doublet of peaks at 6.9Â° and 7.3Â° when Tetracene and PEDOT:PSS/Tetracene AIL structures are introdu ced. The latter of these corresponds to (001) reflections of the highly ordered thin Tetracene layer  . For the ZnPc film grown on PEDOT:PSS/Tetracene, a weak feature is present at 2 = 13.6Â° suggesting second order (200) triclinic or (400) monoclinic reflections and a sign of increased crystallinity in this ZnPc film. The most clear effect of the addition of these AIL structures is the additional XRD peaks located at 25Â° < 2 < 30Â°. Peaks in this range have been attributed to third or higher order ( h 00) reflections, but the weak nature or complete absence of the second order peak in these film s suggests that a different growth mode is also present . To investigate, the lattice constants of triclinic and monoclinic ZnPc phases were collected from the literature  and inserted in to ( 3 1) ( 3 7) for the triclinic structure and ( 3 8) for the monoclinic structure. ( 3 1) ( 3 2) ( 3 3)
88 ( 3 4) ( 3 5) ( 3 6) ( 3 7) ( 3 8) Next, ( hkl ) indices observed previously in flat lying ZnPc films were inserted into the d the observed values. A comparison of these data is shown in Table 3 1. From these data, the bes t agreement is obtained with the (0 1 2) and ( 1 1 2) indices of the triclinic lattice. Table 3 1. Calculate d and observed diffraction peak positions for ZnPc films. Monoclinic Triclinic Observed ( hkl ) ( hkl ) 6.72 100 6.89 6.87, 6.93 200 13.81 7.02 7.03 400 27.83 26.7 0 1 2 26.79 27.8 1 1 2 27.69 These high angle diffraction peaks correspond to interplanar spacings of 3.2 to 3.3 Ã… , which is close to the interplanar distance in ZnPc crystals along the molecular
89 stacking direction. Given the geometry of 2 mode of these XRD measurements, this res ult suggests that the ZnPc films grown on Tetracene and PEDOT:PSS/Tetracene adopt both face on and edge on orientations, the latter of which have been attributed to improved charge carrier generation [104,125] and transport in organic films and devices [35,102] . However, a 150 nm thick ZnPc film is not relevant to OPV devices. In the context of this work, ZnPc thicknesses on the order of tens of nanometers are of much greater interest. This length scale is close to the characteristic exciton diffusion length of several organic semiconductors, inc luding ZnPc [35,55,56,126] . To address this conflict of int erests the ZnPc film thickness wa s reduced to 80 nm an d 40 nm while an ITO substrate wa s used in place of SiO 2 to better simulate ZnPc films used in OPV devices. In both cases, the change in ZnPc orientation is preserved. The XRD spectra of 80 nm and 40 nm ZnPc films on different underlayer s are shown in Figure 3 8 . In both 40 and 80 nm thick ZnPc films, the peak near 7 Â° corresponding to edge on ZnPc growth is maintained for films with and without AIL structures present. The films grown on PEDOT:PSS/Tetracene (blue dotted lines) show an incr ease in intensity, indicating a greater fraction of ordered upright standing domains. For the films grown on Tetracene (red dash dot lines), the high angle signal from flat lying ZnPc aggregates is no longer observed under these conditions. This could be due to the increased signal to noise ratio brought on by the rougher ITO surface, decreased signal from a thinner ZnPc film, or the absence of face on growth altogether. However, face on growth is still observed in 40 and 80 nm thick ZnPc films grown on PE DOT:PSS/Tetracene.
9 0 Figure 3 8 . X ray diffraction spectra of 80 and 40 nm thick ZnPc films on ITO substrate. Together, the opti c al absorption and XRD results give insight into the evolution of ZnPc growth modes with underlayer prope rties. The systems of ZnPc molecules, which extend above and below the flat plane of the molecule depicted in Figure 3 1 , do not interact strongly with passivated oxide surfaces such as SiO 2 and ITO. This results in the well known edge on growth mode wit h an absorption spectrum which has relatively balanced peak heights, depicted by black squares in Figure 3 6 . When grown on a substrate pre coated with an organic AIL structure, the substrate ZnPc adsorbate A B C D
91 interactions change: the arriving ZnPc molecules interact with Tetracene systems more strongly than they interact with oxide surfaces. This allows the ZnPc molecules to lie flat, as inferred from the optical absorption spectra and from the structural data from XRD experiments. 3. 4.3 Surface Topography: Atomic Force Microscopy In order to further study the effects of conditions on ZnPc growth, the surfaces of various film structures were characterized with atomic force microscopy (AFM). The tapping mode AFM micrographs of 25 nm thick ZnPc films grown on bare ITO, ITO/Tetracene, and ITO/PEDOT:PSS/Tetracene are shown in Figure 3 9 . For bare ITO, sized grains of ITO, forming a continuous film with a relativel y const ant height distribution [Fig. 3 9 A) ] an d root mean square roughness (r RMS ) of 2.4 nm. In contrast, ZnPc grown on ITO/Tetracene forms a discontinuous surface with islands of thicker ZnPc separated by lower lying ZnPc regions. These form a stepwise height distribution, as indicated by the relativ ely flat regions separated by 20 3 9 D) .The addition of PEDOT:PSS to the AIL structure drastically modifies the ZnPc surface: r RMS decreases from 13.6 nm to 4.3 nm, similar to that of ZnPc grown on bare ITO . The line profile shows that the peaks and valleys formed by ITO/Tetracene are replaced by a more gradually sloping terrain with the addition of PEDOT:PSS. Finer scans (not shown) reveal individual ZnPc grains (35 Â± 5) nm in size for all three growth cond itions.
92 Figure 3 9 . AFM height images of 25 nm thick ZnPc films. Films were deposited onto A) bare ITO, B) Tetracene, and C) PEDOT:PSS/Tetracene. D) Line scans across the surfaces shown. It is clear from the AFM scans in Figures 3 9 B) and C) that the structural properties of the underlying Tetracene layer play a crucial role in determining ZnPc growth modes and surface morphology. The AFM scans of Tetracene films with 10 nm nominal thickness deposited on bare ITO and ITO/PEDOT:PSS are shown in Figure 3 A B C D
93 10 . Here, Tetracene forms a very rough, coarse grained structure on bare ITO, in agreement with previous studies of Tetracene films with similar thickness [39,40] . The film has a continuous height distribution with peaks reaching 50 nm above the measurement floor and r RMS = 13.0 nm. Figure 3 10 . AFM height images of 10 nm thick Tetracene films. Films were grown on A) bare ITO and B) ITO/PEDOT:PSS, with line scans in C) . Tetracene grown on PEDOT:PSS coated ITO contrasts sharply with that grown on bare ITO: r RMS decreases to 3.3 nm and the surface area is reduced by a factor of 3.7. Surface coverage improves dramatical ly and Tetracene domains are more continuous, suggesting a decrease in PEDOT:PSS Tetracene interaction energy relative to bare ITO Tetracene. These two modes can be characterized as three A B C
94 dimensional growth and two dimensional growth as the ITO/PEDOT:PSS/T etracene film is relatively uniform in the plane of the image, whereas the ITO/Tetracene film displays significant features in all three directions. In conclusion, the bulk and surface properties of the ZnPc donor and Tetracene AIL layers are each dependen t on the layer immediately adjacent. On a weakly interacting ITO substrate, Tetracene molecules prefer to self aggregate and form a rough film with peaks and valleys up to 60 nm in height. thiophene and styrene constituents of the PEDOT:PSS surface interact more strongly with Tetracene molecules, encouraging more uniform surface coverage and a four fold reduction in r RMS . These differences are then passed on to the nominally 25 nm thick ZnPc layer grown on top: ZnPc domains are when grown on ITO/Tetracene, while the surface of ITO/PEDOT:PSS/Tetracene/ZnPc is continuous and has roughness similar to that of ZnPc grown on bare ITO. 3. 5 The Effects of PEDOT:PSS/Tetracene and PEDOT:PSS/ NPB Inter layer Structures on DBP Film Properties As mentioned in Section 3. 2.1, the semiconductor Tetraphenyldibenzoperiflanthene (DBP) is a relatively new donor material with a deeper E HOMO than ZnPc and PbPc, leading to some of the highest reported power conversi on efficiencies in vacuum deposited planar heterojunction  and mixed heterojunction  small molecule OPV devices. DBP molecules adopt a face on conformation when deposited on oxide surfaces  , resulting in a very high absorpti on coefficient and efficient charge carrier transport. These characteristics lead to highly efficient OPV devices des pite a relatively large optical band nm. In these studies, all films were grown at rates between 0.5 and 1 Ã…/sec .
95 3. 5.1 Optical Absorption Unlike the planar phthalocyanines described in Section 3. 3 and 3. 4 , diffraction peaks are not observed for DBP films [21,80,128] and very small amounts of DBP inclu sion in DBP:fullerene blends have been shown to disrupt characteristic fullerene crystallinity  . Furthermore, the absorption spect ra shown in Figure 3 1 1 show that DBP absorption is unaffected by the presence of organic AIL species between DBP and bare ITO. The complex absorption spectra characteristic of DBP films is unchange d when film growth proceeds on top of ITO , Tetracene, or N PB . Therefore this study will not focus on modifying the molecular orientation or intermolecular interactions between DBP molecules because the amorphous nature of vacuum deposited DBP films suggests that the out of plane phenyl rings disrupt intermolecula r ordering. Figure 3 1 1 . Optical absorption of 15 nm thick DBP films grown on different underlying layer structures. 3. 5.2 Photoluminescence Quenching In addition to modifying the growth dynamics and bulk properties of layers as described in Sections 3 . 3 and 3. 4 , organic AILs have also been used to modify the anode/donor interface in OPV devices in order to provide a second exciton dissociating
96 heterojunction [80,105 107] or block excitons from quenching at the anode/donor interface [21,104,130] . In both cases the AIL structures are chosen such that hole transport from the donor to the anode is not impeded . In the former case, the AIL E HOMO lies closer to vacuum than that of the donor in order to efficiently dissociate excitons and generate additional photocurrent at the AIL/donor interface. In the c ase of an exciton blocking layer, the AIL E HOMO is chosen to lie even with or slightly deeper than the donor E HOMO so that excitons diffuse to and from the AIL/donor interface without quenching or dissociating there. In this study, quenching and blocking i nterfaces are investigated in terms of photoluminescence (PL) behavior of the donor material DBP. For a given illumination source and sample orientation, the PL signal is directly proportional to the steady state exciton concentration within the emitting f ilm. Therefore, by comparing the PL spectra of DBP films adjacent to different AIL structures it is possible to discern the exciton quenching, exciton dissociating, or exciton blocking behavior of said structures. First, we consider the hole transport mate rial NPB as an AIL species because its E HOMO of (5. 5 Â± 0.1) eV [22,23] is compatible with that of DBP (5.5 eV) [20,21] . For comparison, the perfectly quenching acceptor C 60  and perfectly blocking wide bandgap electron transport material Bathocuproine (BCP) [50,131 ] are also inserted in multiple combinations to compare NPB with known quantities. Schematics of the three quenching/blocking combinations are shown in Figure 3 1 2 . In terms of PL signal, we expect the DBP film surrounded by two quenching interfaces to exhibit the lowest PL, followed by one quenc hing interface, and two blocking interfaces would have the greatest PL.
97 Figure 3 1 2 . Schematic representation s of energy level diagrams of three combinations of quenching and blocking DBP interfaces. The PL spectra of 20 nm shown in Figure 3 1 3 . In order to maintain high transparency and minimize interference effects, an interlayer thickness of 5 nm as used for BCP and C 60 , while 10 nm NPB was used. As expected, the PL signal is least for two quenching interfaces and increases when the exciton blocking BCP replaces the exciton quenching C 60 (red circles and black squares , respectively). Interestingly, the PL spectrum of DBP with NPB and C 60 layers adjacent almost perfectly overlays that of ITO/DBP/BCP. This confirms that NPB not only blocks excitons generated within the DBP film but also, because ITO and C 60 behave identically as exciton sinks, that NPB blocks excitons with the same efficienc y as BCP. Integrating the spectra for ITO/DBP/BCP and NPB /DBP/C 60 structures yields PL ratios of 2.10:1 and 2.15:1, respectively, relative to the case of two quenching A B C
98 interfaces. Finally, the DBP film surrounded by two blocking layers exhibits a five fold enhancement in total PL signal compared with that of ITO/DBP/C 60 , with an integrated PL ratio of 5.17:1. Figure 3 1 3 . PL spectra of DBP films surrounded by different interlayer species. This data can be combined with quantitative optical and steady state diffusion calculations to verify the exciton diffusion length of the DBP f ilm. As described i n Section 1.4 , exciton s are chargeless particles which can diffuse through according to dynamics can be descr ibed by the steady state d iffusion equation in one dimension, ( 3 9) Here, x is the position within the film, e (x) and G(x) are the exciton concentration and exciton generation rate, respectively, as functions of depth within the film, L D is the exciton diffusion length, and is the exciton lifetime. This equation can be solved by
99 imposing various boundary conditions approximating exciton quenching and blocking behavi or. Taking x = 0 as the anode or AIL/donor interface and x = L as the end of the donor layer farthest from the anode, a quenching interface corresponds to e (x = 0 or x = L) = 0 whereas a blocking interface corresponds to (x = 0 or x = L) = 0. The g transfer matrix calculations [50,132] , leaving the exciton diffusivity D = L D 2 / as a fitting parameter. By integrating the calculated steady state exciton distributions across the DBP thickness and normalizing to the quenching/quenching case, can be eliminated, leaving L D as the lone fit parameter. The results of this treatment are plotted in Figure 3 1 4 . For a 20 nm think DBP film with an L D of 10 nm, for example, the steady state exciton concentration, whose distance profile is shown in Figure 3 1 4 A) , falls to zero at the film edges when both interfaces at x = 0 and x = L (here, 20 nm) quench excitons. Th e case of a blocking interface at x = 0 is also shown in Figure 3 14 A) , with the concentration normalized to the maximum of the ITO/DBP/C 60 case (two quenching interfaces). For the same L D , there is a greater total number of excitons within the DBP film at steady state because there is no quenching at one end of the DBP layer. Finally, the NPB /DBP/BCP structure (both blocking interfaces) shows the greatest total exciton concentration for the given L D , in agreement with PL experimental results.
100 F igure 3 1 4 . Calculated steady state exciton concentrations within a DBP film under = 530 nm illumination. A) Normalized concentration distribution for L D = 10 nm. B) Total exciton concentration for different L D values, normalized to the case of two quenching interfaces and L D = 2 nm. To extract an estimated L D value, the steady state exciton equation can be solved for a range of reasonable L D values and normalized to a constant concentration. In Figure 3 1 4 B) , the steady state exciton distribution is determined for 2 nm L D 15 nm and integrated over the entire DBP thickness (here, 20 nm) before being normalized to a common value: the exciton population corresponding to two quenching interfaces and L D = 2 nm. As L D increases in the cas e of quenching only (black squares ) and bl ocking plus quenching (red circles ), the total exciton count decreases with increasing diffusivity. This is expected as a longer L D corresponds to a higher likelihood of an exciton migrati ng to a quenching interface and not being present in the film bulk . For simpler comparison with experimental PL data, these values are normalized to the quenching only values in Figure 3 1 5 , where the PL ratios are shown as a function of L D . A B
101 Figure 3 1 5 . Total exciton concentration as a function of L D normalized to the case of two quenching interfaces. Here, the 2.10:1 and 2.15:1 (blocking + quenching):(quenching only) ratios obtained with PL experiments correspond to L D = 8 nm, while the 5.17:1 (block ing only):(quenching only) ratio corresponds to L D = 10 nm. Therefore we conclude the DBP film has an exciton diffusion length L D = ( 9 Â± 1) nm, in excellent agreement with values obtained from PL measurements  and simulated OPV device behavior [21,127] . 3.6 Summary of AIL/Donor Multilayer Structures In summary, we have shown that replacing ITO with an organic inter layer dramatically alters the bulk and interfacial properties of subs equently grown donor films. AFM images of PbPc show ed that molecules aggregate at ledges formed by large Pentacene grains. Furthermore, these first stages of film growth are crucial to the optical absorption spectra of PbPc films, which feature strong IR t ransitions at relatively thin films thicknesses in the range of 10 20 nm . ZnPc was shown to form a stronger aggregated, face on film when deposited on top of Tetracene. By adding a planarizing
102 PEDOT:PSS layer, Tetracene exhibited a more two dimensional gro wth mode, which then imparted a flatter surface on the subsequently grown ZnPc film. Finally, by depositing a transparent, wide band gap material between PEDOT:PSS and DBP, the intensity of characteristic DBP emission was enhanced in accordance with calcul ated exciton diffusivities and diffusion lengths.
103 CHAPTER 4 EFFECTS OF ORGANIC ANODE INTERLAYER S ON PHOTOVOLTAIC DEVICE PERFORMANCE 4. 1 Introduction After developing an understanding of the effects of different combinations of anode interlayer (AIL) and donor materials on donor film properties in Chapter 3 , this chapter will investigate how those properties affect photovoltaic device performance. As discussed in Chapter 1, the power conversion efficiency ( P ) in OPV devices is limited by a numb er of inher ent physical properties of the constituent OSC materials. Two of the major loss pathway s are due to characteristic narrow optical bandgaps and low dielectric constants of the major absorbing species. These properties lead to inefficient spectral coverage and charge carrier generation, respectively, in OPV devices. In this chapter, we show that AIL layers can help mitigate these loss mechanisms by boosting low energy photon absorption and increasing carrier generation in OPV devices based on diffe rent AIL/donor combinations. Section 4. 2 describes the effect of inserting a Pentacene template layer into a PbPc based OPV device. In Section 4. 3 , we use a series of device architectures to investigate the enhancement mechanisms in ZnPc based devices with a PEDO T:PSS/Tetracene AIL structure. DBP based devices with PEDOT:PSS/Tetracene and PEDOT:PSS/ NPB AIL structures are the focus of Section 4. 4 . 4. 2 Planar Heterojunction PbPc based OPV Devices with a Pentacene Template Layer To investigate the effect of a Pentacene interlayer on PbPc device performance, planar heterojunction OPV devices with the structure ITO/PbPc (20 nm)/C 60 (40 nm)/BCP (8 nm)/Al (100 nm) were fabricated with and without a 10 nm thick Pentacene
104 layer between ITO and PbPc layers . In this s tructure, BCP functions to block exciton generated in C 60 from nonradiatively quenching at the C 60 interface and to protect C 60 from damage and subsequent interdiffusion of Al atoms [50,131] . Pentacene and PbPc layers were deposited at 0.2 0.3 Ã…/sec and C 60 , BCP, and Al were deposited at 1 2 Ã…/sec. The current density voltage ( J V ) characteristics of these devices are shown in Figure 4 1. Table 4 1 summarizes the photovoltaic performance parameters (short circuit current density J SC , open circuit voltage V OC , fill factor FF , and power conversion efficiency P ) of these devices. With the same V OC and a slightly decreased FF , the power conversion efficiency increases from P = (1.6 Â± 0.1) % to (2.2 Â± 0.1) % with the use of the P entac ene layer , equivalent to a net increase of 38 %. The main driver of this improvement is the increase in J SC , which is 48% greater in the Pentacene templated device. Figure 4 1. Photovoltaic performance characteristics of Pb P c based OPV devices. A) J V characteristics and B) short circuit EQE spectra of PbPc devices grown on bare ITO and on Pentacene covered ITO. Since the majority of the enhancement in P is due to an increase in J SC , we shift focus to the short circuit EQE spectra shown in Figure 4 1 B) . We see that the A B
105 P entacene templated device has a higher EQE than the non templated device in the entire PbPc absorption region from = 550 nm to 1000 nm. Furthermore, this increased response is more significant in the longer wavelength portion corresponding to the triclinic PbPc phase absorption. On the other hand, there is virtually no difference in the C 60 absorption region ( = 400 nm to 550 nm). Note that the addition of the P entacene layer does not increase the series re sistance of the device. This is attributed to the high hole mobility in P entacene and the close alignment of the highest occupied molecular orbital (HOMO) energies of P entacene [22,30,31] and PbPc  . Table 4 1. Photovoltaic performance parameters of PbPc based OPV devic es grown on bare ITO and on Pentacene under 100 mW/cm 2 simulated AM 1.5G illumination. AIL J SC (mA/cm 2 ) V OC (V) FF P (%) J V EQE J V Corrected None 6.0 Â± 0.3 5.9 Â± 0.3 0.50 0.53 1.6 Â± 0.1 1.6 Â± 0.1 Pentacene 9.6 Â± 0.5 8.8 Â± 0.4 0.50 0.51 2.5 Â± 0.1 2.2 Â± 0.1 Note that we have adjusted the solar simulator intensity to account f or the spectral mismatch factor  for the device without any template. By integrating the EQ E spectrum of that device with th e standard AM1.5G solar spectrum  , we obtain a calculated J SC in excellent agreement with the measured J SC f or that device (see Table 4 1 ). However, due to the variation in the EQE spectra among different devices , the spectral mismatch factor varies slightly from 0.95 for the device on bare ITO to 1.05 for the device with Pentacene interlayer . As the solar simul ator intensity was set to be 1 sun for the non templated device in our experiments, this intensity is equivalent to approximately 1.1 suns for the device with the P entacene template. This agrees with the 10% higher calculated J SC than the measured value fo r th e Pentacene templated
106 device . In the subsequent discussions, we use the calculated J SC along with the measured V OC and FF to calculate a corrected power conversion efficiency that more accurately ref lects the true efficiencies of these devices . The hig her content of the triclinic phase in the PbPc film grown on top of P entacene, which leads to a stronger absorption in the near infrared, is one of the reasons for the enhancement of the EQE [Figure 4 1 B) ] of the templated device. In addition, the internal quantum efficiency (IQE) of the template device is enhanced in the entire PbPc absorption spectral region. This could be attributed to the improved morphology as we discussed above: the increased crystallinity and possibly a higher hole mobility of the Pb Pc grown on top of Pentacene can improve the charge collection efficiency. This is supported by a slight decrease in the saturation parameter S = J ( 1 V)/ J (0 V) from 1.21 in the non templated device to S = 1.18 in the device with Pentacene, meaning that a greater fraction of maximum amount of extractable charge carriers are collected at short circuit in the device with the Pentacene template . In order to relate OPV device performance to PbPc film properti es, t he effect of PbPc thickness on the device performance was also investigated. The short circuit EQE spectra of these devices are shown in Figure 4 2. The portion of the EQE spectra corresponding PbPc absorption ( > 550 nm) has two major peaks correspo nding to absorption of the monoclinic ( ~ 750 nm) and triclinic ( ~ 900nm) phases. For the devices grown on bare ITO [Figure 4 2 A) ], the device with 10 nm PbPc thickness has response dominated by the monoclinic phase. Contribution from the triclinic phas e increases continuously as the PbPc layer thickness increases. This can be understood by reconsidering the PbPc film absorption spectra [Figure 4 2 A) , inset]. Absorption by
107 the triclinic phase also increases continuously with PbPc thickness, so it follows that triclinic phase content is greater in the top sections of the PbPc layer closest to the exciton dissociating PbPc/C 60 interface. As a result, the 900 nm peak in spectral response increases with PbPc thickness. Figure 4 2. Short circuit EQE spectra of PbPc based devices with varying PbPc layer thickness. Films were grown on A) bare ITO and B) ITO/Pentacene. The PbPc thickness dependence is different for devices with a Pentacene interlayer , whose EQE and absorption spectra are shown in Figure 4 2 B) . Compared with PbPc layers grown on bare ITO, these films have a more constant monoclinic:triclinic character as a function of PbPc thickness. The exception to this rule occurs between 10 and 20 nm PbPc, where both the EQE and absorption spectra show a significant increase in contribution from the triclinic phase. While the 30 nm thick PbPc film absorbs significantly more light, the EQE spectra corresponding to 20 and 30 nm thick PbPc layers are v irtually identical. The exciton diffusion length in PbPc devices has been estimated to be ~5 10 nm  therefore, on average, excitons generated more than ~10 nm from the PbPc/C 60 interface will not generate photocurrent. Since the total film absorption is a superposition of the absorption of the entire thickness of the A B
108 PbPc film, additional abs orption by 30 and 40 nm thick PbPc layers does not necessarily result in an increase in EQE. The photovoltaic performances of these devices are summarized in Figure 4 3 . With or without a P entacene template, V OC is nearly constant except for the device wi th 10 nm PbPc on P entacene. In that case, the thin PbPc layer may not be sufficiently thick to cover the P entacene surface to form a continuous layer . In this case the V OC is limited by the P entacene/C 60 junction cell, which has a smaller V OC  . Overall, without the P entacene template, a maximum P was obtained with 30 nm PbPc film, whereas a 20 nm thick PbPc leads to the optimum performance of P = ( 2.2 Â± 0.1) % for devices with P entacene template. The J SC first increases as the thickness of the PbPc film increases to up to 30 nm. There is a fundamental tradeoff between the increase in optical absorption and decrease in the exciton diffusion efficiency due to the short exciton diffusion length [50,91] . In general, a further increase in PbPc thickness does not necessarily increase the short circuit current, but instead increases the series resistance of the device and reduces the fill factor, as shown for both cases. For the devices incorporating the Pentacene template, the triclinic phase is homogenously distributed and the NIR photons are mostly absorbed in th e bottom part of the film where the light first passes through, so the short circuit current rolls off right after the peak at 30 nm. On the other hand, without the templating layer, the excitons generated by the absorption of the triclinic phase are locat ed in the middle and upper part of the film closer to the PbPc/C 60 interface and therefore shifts the peak of short circuit current towards a greater PbPc thickness.
109 Figure 4 3. Comparison of the PbPc layer thickness dependencies of PbPc/C 60 solar cell performance characteristics for devices w i th (red) or without (black) a Pentacene template layer. We also notice that the fill factor of the non templated device decreases more slowly with increasing PbP c thickness than the P entacene templated device. The semi quantitative analysis of the optical absorbance spectra mentioned earlier suggests that in the top part of a thicker film, the content of the triclinic phase for the film on the bare ITO is higher than it is on the P entacene template. This can result in a higher hole mobility and thus better charge collection efficienc y for the non templated device when the donor layer is more than 20 nm thick.
110 4. 3 ZnPc based OPV Devices with a PEDOT:PSS/Tetracene Anode Interlayer Structure Following the discussion of Tetracene and PEDOT:PSS/Tetracene AIL structures on ZnPc film properties in Section 3.4 , this section will explore the effects of these AIL structures on the performance characteristics of ZnPc based OPV devices. The characteristics of planar heterojuncti on (PHJ) devices will be presented first in order to show that the Tetracene containing AIL structures function to circumvent the exciton diffusion bottleneck by increasing the internal quantum efficiency (IQE) corresponding to donor ZnPc absorption withou t sacrificing charge carrier transport. Next, the PEDOT:PSS/Tetracene AIL is inserted in a variety of ZnPc:C 60 device architectures to reveal the mechanism for this enhancement. The ZnPc used in these devices was synthesized in house by mixing zinc chlorid e with an excess of 1,2 dicyanobenzene. Further details regarding the synthesis and characterization of this ZnPc can be found elsewhere  . This synthetic route yi elds ZnPc with half the electrically active defect concen tration as ZnPc obtained from a commercial supplier (Sigma Aldrich), as well as a small amount of Cl in the resulting material (corresponding to half the atomic concentration of Zn), which is not obs erved in the ZnPc obtained from Sigma Aldrich. Ultimately, OPV devices based on this home synthesized have a higher fill factor and V OC than those based on commercially obtained ZnPc  . 4.3.1 Planar Heterojunction Devices Figure 4 4 shows A) the J V characteristics under 100 mW/cm 2 simulated AM 1.5G illumination and B) the short circuit EQE spectr a of PHJ OPV devices with different AIL structures and a 25 nm thick Z nPc donor layer. The photovoltaic
111 performance parameters of these devices are summarized in Table 4 2. The most immediate dif ference in device performance is the J SC : the addition of Tetracene has little effect on the J V characteristics while the PEDOT:PSS/Tetracene AIL increases J SC by 75%. The FF and V OC are minimally affected. While the J SC s of devices grown on bare ITO and ITO/Tetracene are similar (4.8 and 5.2 mA/cm 2 , respectively) the photocurrent corresponding to Zn Pc absorption ( > 550 nm) is significantly greater in the Tetracene templated device: the EQE at = 630 nm increases from 0.17 to 0.26, and more than doubles to 0.43 in the device with a PEDOT:PSS/Tetracene AIL structure. Figure 4 4. Photovoltaic performance characteristics of ZnPc based OPV devices. A) J V characteristics and B) short circuit EQE spectra of PHJ ZnPc devices grown on bare ITO (black), 10 nm Tetracene (red) and PEDOT:PSS/Tetracene (blue). All devices have C 60 (40 nm)/BC P (8 nm)/Al. Table 4 2. Photovoltaic performance parameters of PHJ ZnPc devices with various AIL structures under 100 mW/cm 2 simulated AM 1.5G illumination. A B
112 For a more detailed picture of photocurrent generation at short circuit conditions, we separate the EQE into the absorption efficiency, A , and the internal quantum efficiency, IQE , by measuring the reflectance through the device stack. In this case we assume that no light is transmitted through the device and calculate A = 1 R , where R is the fraction of light reflected. Figure 4 5 shows p lots of the A and IQE spectra for devices grown o n bare ITO or AIL covered ITO. The A corresponding to ZnPc absorption is most strongly affected by the PEDOT:PSS/Tetracene AIL, in agreement with the ZnPc film absorption results from Section 3 .4. Integrating A for > 600 nm yields a net 18% increase caused by the PEDOT:PSS/Tetracene AIL structure. While the Tetracene only AIL only increases A by 4% over this range, the spectral shape qualitatively features a higher absorption contribution fr om the aggregated ZnPc state, also in agreement with film absorption ex periments discussed in Section 3 .4. While significant, the increase in A pales in comparison with the increased IQE over the ZnPc > 600 nm, the ZnPc IQE is increased by net 41% and 124% for Tetracene and PEDOT:PSS/Tetracene AIL structures, respectively. Combined with small (< 5%) reductions in fill factor it is clear that these AIL structures solve the exciton bottleneck challenge, i ncreasing photocurrent generation without sacrificing carrier collection efficiency.
113 Figure 4 5. Two components of EQE spectra as a function of AIL structure. A) Device absorption efficiency ( A ) measured as 1 R through the device active area. B) Internal quantum efficiency (IQE) calculated as EQE( )/ A ( ). We can further characterize the enhancement mechanism by considering that the IQE is the product of the exciton diffusion ( ED ), charge transfer ( CT ), and charge collection ( CC ) efficiencies. Among these terms, the CT CC product has been shown to have extremely strong control over the fill factor in OPV devices [35,89,91] . In this set of devices, the presence of AILs has little effect on the f ill factor, so we attribute the increased photocurrent generation to more efficient exciton diffusion to a heterojunction. 4.3.2 Bulk and Planar Mixed Heterojunction Devices To test the hypothesis that enhancement is due to increased ED , devices with diff erent ZnPc:C 60 interface structures were fabricated in order to vary the nature of ED behavior. In PHJ devices, ED is limited because all excitons must migrate to the planar, exciton dissociating ZnPc/C 60 interface in order to generate photocurrent . On the other extreme, a properly designed co deposited donor:acceptor bulk heterojunction (BHJ) active layer configuration forms a morphology where exciton diffusion to a hetero interface is extremely efficient. Exciton diffusion efficiencies approaching 100% have been observed in similar blended donor:acceptor layers by photoluminescence A B
114 quenching experiments  and voltage bias dependent EQE measurements  . Finally, the PHJ and BHJ devices are also compared with planar mixe d heterojunction (P M HJ) devices where both neat and mixed ZnPc layers are present  . Excitons are generated in both neat and mixed donor layers in the P M HJ structure, so exciton diffusion efficiency falls somewhere between those of PHJ and BHJ devices. The BHJ device active layer structure consisted of ZnPc:C 60 ( 1:1 by weight , 30 nm)/C 60 (30 nm)/BCP (8 nm)/Al. This is not a BHJ in the most strict sense because a small portion of the ZnPc molecules do make contact with the neat C 60 layer. However since the only ZnPc present in the d evice is deposited in a mixed layer, it will be considered a BHJ. The P M HJ structure consisted of the layer structure ZnPc (20 nm)/ZnPc:C 60 ( 1:1 by weight , 20 nm)/C 60 (30 nm)/BCP (8 nm)/Al . The J V characteristics of BHJ and P M HJ devices are shown in Figures 4 6 A) and B) , respectively, and the photovoltaic performance parameters are summarized in Table 4 3. Figure 4 6. J V characteristics of ZnPc based OPV devices with a mixed ZnPc:C 60 layer. A) BHJ devices with no neat ZnPc layer and B) P M HJ devices with neat and mixed ZnPc layers. A B
115 For the BHJ device with no neat ZnPc layer [Figure 4 6 A) ] , the addition of a PEDOT: PSS /Tetracene AIL structure degrades device performance, particularly the fill f actor and V OC . In this structure, Tetracene makes direct contact with C 60 at the Tetracene/BHJ interface. OPV devices based on Tetracene/C 60 heterojunctions have been shown to have a lower V OC  than the ZnPc/C 60 device s in this work. Similar devic e s with two different donor materials present (one neat and one blended with acceptor C 60 ) have been shown to operate as two heterojunctions in parallel [105,106] , with charge transport (manifested in the fill factor) and the V OC limited by the lower performing don or/acceptor combination. In our case, this is the Tetracene/C 60 junction present in the BHJ device with PEDOT:PSS/Tetracene AIL. As a result, the V OC and fill factor are reduced from 0.60 V and 0.60 in the ITO/BHJ device to 0.54 V and 0.48, respec tively, in the AIL/BHJ device. However when the built in field sweeps charge carriers towards the el ectrodes, the J SC of both BHJ devices differ by only 0.3 mA/cm 2 , just on the edge of experimental error. The case of the P M HJ devices is quite different: while the fill factor and V OC are virtually unaffected, the addition of the PEDOT:PSS/Tetracene AIL increases J SC by 33% relative to the P M HJ device on bare ITO, compared with a 75% increase observed in PHJ devices. T his can be explained qualitatively by the contrasting spatial distributions of exciton generation: excitons are generated in both neat Z nPc and mixed ZnPc:C 60 layers in the P M HJ architecture. This leaves room for improvement in J SC via enhanced ED in the neat ZnPc layer, though not as much room for improvement as in the PHJ case where all excitons generated in ZnPc must diffuse through the neat ZnPc layer to reach a ZnPc/C 60 hetero interface.
116 Table 4 3. Photovoltaic performance parameters of BHJ and P M HJ ZnPc devices under 100 mW/cm 2 simulated AMA 1.5G illumination. Device structure BHJ BHJ P M HJ P M HJ AIL structure Bare ITO PEDOT:PSS/ Tetracene Bare ITO PEDOT:PSS/ Tetracene 8.8 Â± 0.2 9.1 Â± 0.1 8.1 Â± 0.3 10.8 Â± 0.4 0.60 0.48 0.58 0.57 0.60 0.54 0.63 0.65 3.2 Â± 0.1 2.4 3.0 Â± 0.1 4.0 Â± 0.2 The short circui t EQE spectra of these devices, shown in Figure 4 7 A) , further illustrate the dependence of photocurrent enhancement on heterojunction structure. For > 550 nm, P M H J spectral response increases across the entire ZnPc absorption region with the addition of the PEDOT:PSS/Tetracene AIL s tructure . While some of this increase can be attributed to increased absorption in the ordered neat ZnPc layer, the IQE spectra in Figure 4 7 B) show that the ED CT CC product also increases with the addition of the AIL structure. On the other hand, BHJ sp ectral response corresponding to ZnPc absorption is not affected by the insertion of the PEDOT:PSS/Tetracene AIL structure. The only difference in the two BHJ spectra occur over a small range from = 450 550 nm, corresponding to Tetracene absorption. Wh ile the BHJ device grown on bare ITO (closed red circles) shows a steep roll off towards = 550 nm consistent with the absorption spectra of ZnPc and C 60 , the increased EQE in the AIL/BHJ device (open red circles) in this region confirms photocurrent gene ration at Tetracene/C 60 interfaces present at the AIL/BHJ interface in this structure. The IQE spectra of the BHJ devices are unaffected by the AIL.
117 Figure 4 7. IQE spectra of ZnPc based devices with different active layer structures. A) IQE spectra of BHJ and P M HJ devices. B) Comparison of the AIL induced IQE enhancements of different device structures. To summarize the effect of the PEDOT:PSS/Tetracene AIL on photocurrent generation, the IQE of AIL containing devices are normalized t o those of devices on bare ITO and plotted in Figure 4 8 for the device structures considered here. The IQE is doubled in the PHJ device structure and gradually increases with increasing wavelength, while the BHJ and P M HJ show IQE ratios which are indepe ndent of incident wavelength. Since absorption by both neat ZnPc and ZnPc blended with C 60 is present in the P M HJ structure, the IQE enhancement falls in between those observed i n the PHJ and BHJ architectures, as predicted. A B
118 Figure 4 8. Relative IQE e nhancement in the ZnPc absorption region for the three ZnPc:C 60 active layer architectures. The possible reasons for increased ED include ( 1 ) increased exciton diffusivity and exciton diffusion length ( L D ), ( 2 ) exciton blocking at the AIL/Donor interface, and ( 3 ) exciton dissociation at the AIL/Donor interface. All three have been demonstrated in organic films and OPV devices [21,57,99,104 107,125] . Since ZnPc is not an efficient emitter, photoluminescence quenching is not a viable experimental option for determining whether excitons are blocked or dissociated at the Tetracene/ZnPc interface. The nature of the Tetracene/ZnPc heterojunction will be fu rther investigated in Chapter 5. As a final step, t he power conversion efficiency of the P M HJ device was maximized by inserting a cathode interlayer consisting of a 10 nm dihexyl perylene 3,4,9,10 bis(dicarboximide) (PTCDI)  layer between the C 60 and BCP layers. Cathode buffer layers have been shown to reduce exciton/polaron recombination in neat fullerene layers  and improve overall charge carrier collection dynamics in
119 OPV device s [108,109] . Here, t he optimized P M HJ device had an active lay er structure PEDOT:PSS/T etracene (10 nm) / ZnPc (15 nm) / ZnPc:C 60 ( 2:1 by volume, 20 nm) /C 60 (30 nm) /PTCDI (10 nm) /BCP (14 nm) . The peak EQE reached 0.69 at = 63 0 nm, contributing to a maximum J SC of (13.9 Â± 0.3) mA/cm 2 . Combined with a consistent V OC and high fill factor, a ZnPc:C 60 record P = (5.8 Â± 0.3) % was achieved  . Figure 4 9 . Photovoltaic performance characteristics of optimized P M HJ OPV device. A) J V characteristics and B) short circuit EQE spectra of optimized P M HJ ZnPc device with PEDOT:PSS/Tetracene AIL structure. While the ultimate P is greatest for the P M HJ device with a PEDOT:PSS/Tetracene AIL structure, comparing the PHJ device with t he mixed active layer cells on bare ITO confirms the utility of this simple AIL structure. Recalling the discussion of the exciton diffusion bottleneck in Section 2.3, the adoption of the BHJ structure has been the most successful method for achieving high photocurrent generation efficiency. This is confirmed in these ZnPc based devices: the J SC of the control BHJ and P M HJ devices on bare ITO are more than 70% greater than that of the control PHJ device. Here, we have shown that the addition of a simple A IL structure A B
120 to a simple PHJ device generates a similar J SC while maintaining a higher fill factor, leading to a greater P than the devices with mixed ZnPc:C 60 layers. 4. 4 DBP based OPV Devices with Exciton Dissociating and Exciton Blocking Anode Interlay er s The organic donor molecule Tetraphenyldibenzoperiflanthene (DBP) has attracted significant research interest in the last few years due to its favorable energy levels and strong optical absorption centered in the green yellow portion of the visible sola r spectrum  . When used in conjunction with a fullerene acceptor , DBP based OPV devices exhibit high quantum efficiencies and V OC > 0.9 V [20,21,98,128,129] , leading to a reported P = (8.1 Â± 0.4) % device with fullerene C 70 as acceptor  . In addition to embodying the current state of the art for evaporated organic solar cells, DBP is also well suited to function as a reference material for investigating the roles of different anode interlayer (AIL) species. Unlike the anisotropic molecular and domain orientations observed in many donor films, DBP films are rather amorphous due to the presence of four out of plane phenyl rings in the molecular structure (Figure 4 ron cores of neighboring molecules, which drive ordering and preferential stacking directions in other donor molecule films, such as phthalocyanines [ 35,61,104] . Whereas the effect of AIL structures on ZnPc devices can be assigned to modified ZnPc bulk properties in addition to the function of the AIL/donor interface, the addition of AIL structures to a DBP based device will only affect the donor int erfacial properties within the device stack. Therefore the convoluted effects on the donor film can be disentangled , leaving the interfacial structure as the lone variable.
121 As revealed by photoluminescence measurements in Section 3.5, DBP forms exciton q ue nching and exciton blocking interfaces with ITO and NPB, respectively. Here, we investigate the effects of these materials, as well as the relatively shallow E HOMO material Tetracene, as AIL species in DBP based OPV devices. By combining experimental resul ts with optical calculations and photocurrent modeling, the quenching, dissociating, and blocking behavior can be directly investigated in OPV device architectures. Furthermore, these results can be compared with ZnPc based OPV devices described in Section 4.3 to determine the blocking or dissociating ing nature of the Tetracene/ZnPc heterojunction. 4. 4.1 Planar Heterojunction Device Performance The J V characteristics of PHJ DBP based devices under 100 mW/cm 2 simulated AM 1.5G illumination are shown in Figu re 4 10 . All organic layers were deposited between 0.5 and 1 Ã…/sec while Al was deposited between 2 and 3 Ã…/sec. All devices have identical layers of ITO/DBP (17 nm)/C 60 (30 nm)/BCP (8 nm)/Al, with PEDOT:PSS (40 nm) /Tetracene (10 nm) and PEDOT:PSS (40 nm)/NPB (10 nm) AIL structures inserted, as shown schematically in the figure . The control device has a very high fill factor of (0.71 Â± 0.01) and V OC = 0.91 V, but a relatively low J SC , leading to a modest P = (3.0 Â± 0.1) %, in good agreement with p revious reports [20,21,80] . The addition of both Tetracene an d NPB containing AIL structures leads to significant increases in J SC : 32% and 40%, respectively. Both AIL structures cause a reduced fill factor, however in the case of the PEDOT:PSS/Tetracene AIL this reduction is much more significant. The reduced fill factor negates the increase in J SC , leading to a decreased P = 2.5 %. On the other hand, the device with PEDOT:PSS/NPB AIL structure maintains a relatively high fill factor of 0.66 and achieves P = (4.1 Â± 0.1) %, a
122 37% net enhancement over the control de vice. The photovoltaic performance parameters of these DBP based devices are summarized in Table 4 4. Figure 4 10 . J V characteristics of DBP based PHJ devices with different AIL structures under 100 mW/cm 2 simulated AM 1.5G illumination . Table 4 4. Photovoltaic pe rformance parameters of DBP (17 nm)/C 60 (30 nm) devices with various AIL structures under 100 mW/cm 2 simulated AM 1.5G illumination. AIL structure Bare ITO PEDOT:PSS/Tetracene PEDOT:PSS/NPB J SC (mA/cm 2 ) 4.7 Â± 0.1 6.3 Â± 0.1 6.6 Â± 0.1 FF 0.71 Â± 0.01 0.45 0.66 V OC (V) 0.91 0.90 Â± 0.01 0.94 P (%) 3.0 Â± 0.1 2.5 4.1 Â± 0.1 To investigate the spectral dependence of photocurrent generation, the short circuit EQE spectra of these three devices are plotted in Figure 4 1 1 alongside the absorption spectra of all organic materials present in these devices. All three devices
123 have similar local EQE maxima in the spectral region where C 60 is the primary While the absorption spectra of acceptor C 6 0 has some overlap with the AIL species (Tetracene, in particular), donor DBP is the only material behavior diverges: the control device EQE has a local maximum of 0. while those of Tetracene and NPB containing AIL devices peak at 0.56 (representing an increase of 47%) and 0.62 (63% increase), respectively. The peak EQE of 0.62 is among the highest reported in PHJ devices based on DBP and C 60 [20,21,127,128] . Figur e 4 11. Short circuit EQE and absorption coefficient spectra of DBP based PHJ devices and organic semiconductors present in said devices. Excitonic behavior near the anode(AIL)/DBP interface is illustrated on the right. Upon first glance these EQ E result s are inconsis tent with the different photocurrent generation mechanisms of quenching, dissociating, and blocking behavior at the anode (or AIL)/donor interface. In the case of the exciton quenching ITO/DBP
124 interface, excitons generated near the interface have a high probability of recombining on ITO and not contributing photocurrent. This is illustrated schematically in Figure 4 11. In the exciton dissociating cascade structure, excitons which would recombine at the ITO/DBP interface are instead dissociat ed into electrons and holes and generate additional photocurrent because there is no barrier to electron transport across the DBP/C 60 interface. Hence, photocurrent is effectively generated at two heterojunctions rather than one in the device with the casc ade PEDOT:PSS/Tetracene AIL structure. In the case of exciton blocking, excitons do not quench or dissociate at the AIL/DBP 60 interface. In this case the reflected excitons must diffuse, on average, a lo nger distance to reach the exciton dissociating DBP/C 60 heterojunction in order to generate photocurrent. However the equilibrium exciton distribution is affected such that photocurrent generation is increased. This will be studied more carefully in Sectio n 4.4.3. As a result, EQE corresponding to DBP absorption is expected to be greater in the cascade (PEDOT:PSS/Tetracene AIL) device than the blocking (PEDOT:PSS/NPB AIL) device. In this initial set of devices this trend does not hold, although both AIL con taining devices have greater DBP EQE than the control device on bare ITO. Recalling that photocurrent generation is a four step process, the absorption efficiency can be measured in order to calculate the internal quantum efficiency (IQE). The absorption e fficiency and short circuit IQE spectra of these DBP based devices are shown in Figure 4 12 . The absorption efficiency spectra in Figure 4 1 2 A) show
125 Tetracene layer accoun ts for the increased absorption ef optical effects induced by the addition of the 40 nm thick PEDOT:PSS between the substrate and active layers devices with AIL structures. However, all three absorption , nm corresponding to exc lusive DBP absorption. Peak A values differ by only 0.03 across the three device structures, confirming that the underlying AIL structures do not strongly affect the optical properties of the DBP films. Figure 4 12 . Spectral distributions of A) absorption and B) internal quantum efficiencies for DBP based OPV devices with different AIL structures. The IQE spectra qualitatively reproduce the trends observed in the EQE spectra , so we may conclude that enhanced DBP photon to electron conversion is largely driven by enhanced exciton to electron conversion, similar to the Tetracene/ZnPc devices. The in devices with the PEDOT:PSS/Tetracene AIL structure confirms that additional absorption by Tetracene does not correspond to efficient pho tocurrent genera tion. T he transparent NPB containing AIL structure eliminates this parasitic absorption, allowing for more absorption by C 60 and greater IQE in the blue A B
126 spectral confirms that the photoc urrent generation enhancement is again due to increased exciton diffusion efficiency in both AIL containing structures. M ore detailed analysis is required to fully understand this behavior, specifically addressing the different relative enhancements caused by the insertion of the cascade and exciton blocking AIL structures. This will be investigated more carefully in Section 4.4.3 and Chapter 5. 4. 4.2 Comparison of Planar and Bulk Heterojunction Device Characteristics As discussed in Section 1. 5 , blending the donor and acceptor species to form a bulk heterojunction (BHJ) is another way to improve photocurrent generation and P in OPV devices. We have fabricated P M HJ devices , consisting of mixed DBP:C 60 and neat C 60 active layers, with a range o f DBP:C 60 ra tios in order to increase photocurrent generation and to compare the devices with the relatively simple AIL/PHJ structure. The J V characteristics of P M HJ devices under 100 mW/cm 2 simulated AM 1.5G illumination with the structure ITO/DBP:C 60 (1:8, 1:2, and 1:1 by weight , 40 nm)/C 60 (10 nm)/BCP (8nm) /Al are shown in Figure 4 13, and the device performance characteristics are summarized in Table 4 5. The characteristics of the PEDOT:PSS/NPB/PHJ device described in Section 4.3.1 are also shown fo r comparison. Acceptor rich DBP:C 60 blends were found to have very good charge transport properties as evidenced by fill factors approaching 0.6, in agreement with other published studies of DBP:fullerene blends [98,129,137] . The optimal balance between carrier generation and charge transport occurs at a volume ratio of 1:2 DBP:C 60 , resulting in P = (3.6 Â± 0.1) %. This is greater than the control PHJ device grown on bare ITO but less than that of the PHJ device with a PEDOT: PSS/NPB AIL structure.
127 While the J SC of the optimized P M HJ device is slightly greater, the greater fill factor and V OC suggest that the AIL containing PHJ device more efficiently circumvents the exciton diffusion bottleneck. Figure 4 13. J V characteristics of DBP based P M HJ and PHJ devices under 100 mW/cm 2 simulated AM 1.5G illumination. Left: schematic P M HJ device stack. Table 4 5. Photovoltaic performance parameters of various DBP based OPV devices under 100 mW/cm 2 simulated AM 1.5G illumination. Device structure 1:8 P M HJ 1:2 P M HJ 1:1 P M HJ NPB/PHJ J SC (mA/cm 2 ) 5.9 Â± 0.1 7.1 Â± 0.2 7.1 Â± 0.1 6.6 Â± 0.1 FF 0.59 Â± 0.01 0.58 Â± 0.01 0.47 Â± 0.01 0.66 V OC (V) 0.93 0.88 0.87 0.94 P (%) 3.2 Â± 0.1 3.6 Â± 0.1 2.9 4.1 Â± 0.1
128 In the cases of the P M HJ device structures , significant photocurren t is expected to be generated by both donor and acceptor specie s due to the mixed nature of the DBP:C 60 layer. Therefore knowledge of the spectral distribution of photocurrent generation is necessary in order to focus on DBP driven behavior . The EQE and IQE spectra of P M HJ and NPB/PHJ devices are shown in Figure s 4 14 A) and B) , respectively. The EQE corresponding to DBP absorption incr eases with DBP content in DBP:C 60 ratio increases from 1:8 to 1:1. This is strongly dependent on the different amounts of DBP absorption in these films. Figure 4 14 B) sh ows that the IQE, which does not account for the different absorption eff iciencies, is relatively constant for all three P M HJ devices with the 1:2 structure having slightly higher peak values. EQE corresponding to C 60 absorption shows a similar trend, in creasing with decreasing DBP content in the mixed layer. nm in the P M HJ devices is attributed to absorption by the DBP C 60 charge transfer state  arising from the gre ater DBP:C 60 interface area in the devices with a mixed layer. While the optimized P M HJ device has a greater J SC , the PHJ device with PEDOT:PSS/NPB AIL has greater EQE in the spectral region corresponding to DBP absorption, with a peak value of 0.62. The PHJ device is due to lower C 60 IQE, a direct result of the C 60 being in a neat layer rather than a mixed layer. Now we can conclude that the addition of the PEDOT:PSS/NPB AIL structure to the PHJ device improves DB P photocurrent generation efficiency past that of an optimized device with a mixed DBP:C 60 active layer. The superior electrical properties of the neat DBP and C 60 layers over the DBP:C 60 blend also result in
129 improved fill factor and V OC . Altogether, the e xciton blocking structure functions as a robust and effective way to circumvent the exciton diffusion bottleneck, delivering greater peak EQE = 0.62 and fill factor = 0.66. However, the differences in photocurrent generation mechanisms in the quenching, bl ocking, and cascade structures remain to be more carefully addressed. Figure 4 14. Quantum efficiency spectra of DBP:C 60 P M HJ devices with three different DBP:C 60 blend ratios and a PHJ device with PEDOT:PSS/NPB AIL structure. A) EQE and B) IQE calculated as EQE/ A . 4.4.3 Simulation of Photocurrent Generation at Planar Heterojunction Interfaces To quantify the differences in photocurrent generation in PHJ devices with different AIL structures, we combine transfer matrix and exciton diffusion calculations to better understand the photocurrent generation mechanisms under these different scenarios. Specifically, we are interested in the relative contributions to photocurre nt generation of exciton dissociation and exciton blocking at the AIL/DBP interface. To examine the effect of AIL structures on carrier generation, we employ the framework developed by Pettersson et al.  , Peumans et al.  , and Burkhard et al.  for light propagatio n through a multilayer OPV device. Briefly, the reflection, transmission, A B
130 and absorption of light incident on the device stack are calculated from the complex refractive indices of all lay ers present in order to determin e the electric field distribution wi thin the structure. Next, the exciton generation rates within the photoactive layers of interest are calculated from the electric field modulus , the refractive indices, and the absorption coefficients of relevant layers. Once the exciton generation rate pr ofile G(x) is determined, we solve the steady state diffusion equation for excitons, ( 4 1) where e is the exciton concentration, x is position within the layer, L D is the exciton diffusion length, is the exciton lifetime, and G is the exciton generation rate. This equation is solved for two cases relevant to this investigation. In both cases, the DBP/C 60 interface is assumed to be perfectly quenching, i.e. the exciton concentration is zero at that location. For exciton quenching or dissociating behavior at the anode or AIL/DBP interface , the bou ndary condition e(x = 0) = 0 is enforced. For the case of a blocking AIL/DBP interface, we use the boundary condition e/ x (x = 0) = 0. We define x = 0 as the anode (or AIL)/donor interface and x DA as the donor/acceptor interface. These boundary conditions are in accordance with photoluminescence quenching experiments described in Section 3 .5. The general form of the solution to ( 4 1) is given by ( 4 2) Once e(x) is obtained under different sets of boundary conditions , the photocurrent generated by excitons arriving at an exciton dissociating interface is calculated according to
131 ( 4 3) where J ED is the excitonic photocurrent and x diss is the position of the exciton dissociating interface. For the co ntrol device on ITO, we assume a quenching anode/donor interface and a 100% efficient donor/acceptor interface so that the excitonic photocurrent is calculated by evaluating ( 4 3) at the DBP/C 60 interface. For the cascade device with PEDOT:PSS/Tetracene AIL, the equilibrium exciton distribution within the DBP layer is expected to be nearly identical to that in the control device. The o nly difference is that photocurrent generated from exciton dissociation at both the Tetracene/DBP and DBP/C 60 interfaces are summed together , i.e. (4 3) is calculated at x = 0 and x = x DA . For the blocking case , the distribution is calculated for one block ing and one quenching interface and the photocurrent is taken to be that at the DBP/C 60 interface only. Finally, EQE is calculated by normalizing the excitonic photocurrent to the incident optical power: ( 4 4) where q is the elementary charge, is the incident optical power, h constant, and is the photon frequency = . We note that the EQE calculated in this manner only accounts for photocurrent generated by exciton diffusion to an interface, hence the ED subscript. Further, in taking this figure as the EQE we assume that the CT CC product equals unity. Thus ( 4 3) and ( 4 4) function as upper limits to the heterojunction photocurrent and EQE, respectively.
132 The calculated equilibrium exciton distributions within the 1 7 nm thick DBP layer of a PHJ OPV device with exciton quenching (or dissociating) and exciton blocking interfaces at the anode (or AIL)/DBP in terface are shown in Figure 4 1 5 for = 615 nm incident light corresponding to DBP abso rption . Optical constants of all active layer materials were obtained from the literature  and the DBP/C 60 interface is assumed to be perfectly quenching . Figure 4 1 5 . Equilibrium exciton concentration profiles within the DBP layer of a PHJ OPV device excited by = 615 nm illumination . Device structures have an A) exciton quenching or dissociating interface or B) exciton blocking layer at the anode interf ace . . In the case of a quenching or dissociating AIL/DBP interface in Figure 4 15 A) , the exciton concentration falls to zero at the anode or AIL/DBP interface (x = 0). The total exciton population and the slope x at the DBP/C 60 interface both decrease with increasing L D because excitons have a higher probability of reaching a quenching or dissociating interface in a material with greater exciton diffusivity . This trend also holds for a blocking interface as shown in Figure 4 15 B) where the conc entration is greatest at the AIL/donor interface and decays to zero at the dissociating DBP/C 60 interface. In both A B
133 cases, the excitonic photocurrent will nonetheless increase with L D due to the L D 2 term in the numerator of ( 4 3). For a given L D value, both the total exciton co ncentration and the slope of the distribution are greater in the film with a blocking interface because excitons reaching the NPB/DBP interface are not removed from the population as they are in the case of a blocking or dissociating i nterface. Next, we calculate the excitonic photocurrent and EQE generated in a PHJ DBP based OPV device with active layer structure ITO/AIL/DBP (17 nm)/C 60 (30 nm)/BCP (8 nm)/Al (100 nm) by using ( 4 3) and ( 4 4) for = 615 nm incident light. The calculated EQE is plotted in Figure 4 1 6 as a function of L D for the cases of quenching (bare ITO), dissociating (PEDOT:PSS/Tetracene), and blocking (PEDOT:PSS/NPB ) interfaces . For all three structures, EQE increases with incr easing L D as expected. For a given value of L D , the calculated EQE is always greatest for the exciton dissociating device structure, followed by the device with a blocking interface, and the EQE of the device with a quenching interface is least. For values of L D significantly below the DBP layer thickness, the quenching and blocking EQE values are virtually identical. This is expected, since excitons generated far from the DBP/C 60 interface are unlikely to diffuse across the majority of the 17 nm thick DBP film in either case. When excitons from that same section of the DBP film can dissociate at the AIL/DBP interface , the required diff usion length is much less. Therefore, the EQE is expected to be greater for low L D values in the exciton dissociating cascad e structure. In addition to the calculations, the measured short circuit EQE values for each of the three DBP based PHJ devices are shown as horizontal lines in Figure 4 16.
134 Figure 4 16 . Calculated EQE ED as a function of L D in a DBP based PHJ device with various AIL structures under = 615 nm incident light . Horizontal lines correspond to measured short circuit EQE values for devices on bare ITO(solid, black), PEDOT:PSS/Tetracene (dashed, red), and PEDOT:PSS/NPB (dash + dot, blue) . The se simulation results present some significant deviations from experiment. First, the fabricated devices with an exciton dissociating Tetracene layer did not show greater EQE than the devices with the exciton blocking NPB layer as predicted by the calculations . The seco nd major discrepancy arises from comparison of measured and calculated values of EQE and EQE DA , which allows for an estimate of the L D parameter. For this purpose, the measured EQE values for all three device structures are shown in Figure 4 16 as horizont al lines. This procedure yields L D = 11.5 and 13 nm for the devices grown on bare ITO and PEDOT:PSS/NPB, respectively, and L D = 7 nm for the device with a PEDOT:PSS/Tetracene AIL structure. Processing conditions have previously been demonstrated to have a strong effect on donor film properties,
135 including L D [35,55,99,125] . However, the donor materials exhibiting thi s behavior also have highly anisotropic molecular properties and exhibit preferential arrangements within a solid film , in contrast with observations of DBP films . A change in the molecular arrangement would be inconsistent with the results described in Se ction 3 .5, where interlayer s did not affect the optical absorption spectra or the L D values extracted from photoluminescence experiments. The characteristics of DBP based OPV devices with different AIL structures will ne ed to be analyzed further in order to accurately determine the effect s of these structures . 4.5 Summary The variety of AIL and donor materials investigated in the previous two chapters present a wide framework for the study of donor film modification and OPV device performance. For all thre e donor materials chosen, the two phthalocyanines PbPc and ZnPc as well as the relatively new donor DBP, the P of OPV devices is significantly improved with the addition of an appropriate AIL structure. PbPc molecules arrange in an infrared absorbing tri clinic configuration at the very beginning stages of film growth on a substrate pre coated with a thin Pentacene PHJ devices have 38 % greater P than control devices grown on bare ITO substrates. ZnPc based PHJ devices showed significantly increased EQE when a Tetracene layer is introduced between the ITO anode and the ZnPc film. When a PEDOT:PSS layer is also incorporated to form a dual AIL structure, P is increased from (2.0 Â± 0.1) to (3.5 Â± 0.2) % due to the planarizing effect of PEDOT:PSS on the Tetracene film. By separating the absorption from the internal quantum efficiency and investigating the effects of the dual AIL on different active layer architectures, th e enhancement was
136 determined to be due to improved exciton diffusion efficiency. Whether the dual AIL functions as an exciton dissociating or blocking interface is not yet clear. This will be analyzed in more detail in the following chapter. Regardless, th e P M HJ structure was optimized to achieve P = (5.8 Â± 0.3) %, the highest reported value for the ZnPc:C 60 materials system. Finally, exciton dissociating and blocking AIL structures were compared directly in DBP based PHJ OPV devices. Inserting an excit on dissociating PEDOT:PSS/Tetracene AIL structure caused a precipitous reduction in fill factor from 0.71 to 0.45, negating the increase in J SC and ultimately reducing P from (3.0 Â± 0.1) to 2.5 %. The exciton blocking PEDOT:PSS/NPB AIL structure resulted in a 40% increase in J SC while retaining a relatively high fill factor of 0.66, increasing P to (4.1 Â± 0.1) %. Detailed calculations of the excitonic photocurrent for various device architectures yielded results which were inconsistent with measured EQE d ata. These devices will be investigated more thoroughly in the next chapter. Perhaps the most important conclusion from these device studies is that the insertion of a functional AIL structure into the PHJ device stack improves P past that of devices with mixed donor:acceptor BHJ structures. For devices on bare ITO, the P of ZnPc and DBP based devices is increased when the donor and acceptor species are blended together. In these cases balance between increased photocurrent gene ration efficiency and reduce d electrical properties of donor:acceptor blends has been managed in order to improve device performance. Alternatively, we have shown that inserting a functional AIL structure into a PHJ device can increase photocurrent generat ion while maintaining the superior electrical properties of neat donor and
137 acceptor films. As a result the P of PHJ devices with AIL structures is greater than devices with mixed active layers on ITO, and the exciton diffusion bottleneck is managed more e ffectively.
138 CHAPTER 5 COMPARISON OF ORGANIC PHOTOVOLTAIC DEVICES WITH EXCITON DISSOCIATING AND EXCITON BLOCKING INTERLAYER STRUCTURES 5.1 Introduction The results p resented in Chapter 4 conclude that the insertion of organic interlayer s between the anode and donor layer s can increase photocurrent generation and ultimately p ower conversion efficiency of OPV device s . In the case of the donor molecule Tetraphenyldibenzoperiflanthene ( DBP ) , the selection of a proper anode interlayer (AIL) material was shown t o be crucial: a Tetracene containing AIL with a more shallow E HOMO (a so improved photocurrent generation ( J SC ) but reduced the fill factor such that the P was decreased below that of the control device on bare ITO. On the other hand, the insertion of a wide bandgap NPB AIL species with E HOMO well aligned to that of DBP result ed in OPV devices with increased P because the increased J SC is not accompanied by a severely reduced fill factor. The interplay between photocurrent generation and charge carrier transport is a complex re lationship. By virtue of the different energy level alignments of AIL species Tetracene and NPB with donor DBP , this materials system presents a unique opportuni ty to directly investigate the effects of different AIL structures and functionalities on OPV device performance under well controlled conditions. A thorough investigation of DBP based planar heterojunction (PHJ) OPV devices will be presented in this chapter. T he effects of AIL structures on DBP device performance have been studied across a range of DBP thicknesses. By combining experimental measurements with excitonic photocurrent calculations, we gain insight into the enhancement mechanisms in each structure a nd reveal the reasons for different behavior. Finally, we
139 compare the characteristics of DBP based devices with ZnPc based devices in order to determine the nature of the Tetracene/ZnPc heterojunction. 5 .2 Review of Cascade based OPV D evices Cascade based OPV devices have been the subject of growing research in the last decade as a facile method of improving OPV device performance. To put the present work in context, a short review of the existing l iterature will be presented . A number of p revious studies o f the cascade architecture focused on the dynamics of exciton dissociation and charge separation at a heterojunction by manipulating the excited charge transfer state at the donor/acceptor interface with the inserti on a thin ( 5 nm) interlayer were shown to increase charge separation efficiency by reducing polaron polaron coupling after exciton dissociation  and increase the V OC by dir ectly replacing the species in contact with the acceptor C 60 [140 142] . The goal of the present work is to a void any modification of the Donor /C 60 interface and instead fo cu s on the function of the AIL/Donor structure. As such, previous those in the work of Barito et al.  , Cnops at el.  , and Schlenker et al.  are more appropriate for consideration here . In each of these studies, all cascade structured devices exhibit increased photocurrent generation but also have a decreased fill factor relative to control devices with a single exciton dissociatin g interface. This phenomenon was attributed to low electron mobility in the primary absorbing donor species in devices based on the donor boron subphthalocyanine chloride (SubPc) [105,106] .
140 As Figure 4 9 clearly shows, the cascade DBP device with PEDOT:PSS/Tetracene AIL exhibits a reduced fill factor best described by a sharp reduction in p hotocurrent beginning at V ~ 0.4 V and an S kink near the V OC . These features, in addition to being presented in aforementioned cascade structures [105,106] , have also been encountered in studies of hole transport layer (HTL)/donor interfaces in OPV device architectures [143 145] . H ole injection barr iers corresponding to a negative E HOMO offset between the HTL and donor materials have only a small effect on the J V characteris tic in the V < 0 region, in agreement with the similar reverse bias slopes shown in Figure 4 9. In the forward bias regime approaching the built in potential, a significant S kink has been observed in OPV devices with hole injection barriers. These phenome na have been confirmed with drift diffusion simulations [143,144] as well as transient photocurrent behavior  . Furthermore, Cheyns et al. have shown that the resistance at open circuit increases with injection barrier height  . These results are in good agreement with the J V behavior exhibited by the device with a PEDOT:PSS/Tetracene AIL in Figure 4 9 . Due to the similar voltage dependent characteristics observed in devices with a cascade structure and hole injection barriers at the HTL/Donor interface, we can conclude that these effects are inseparable. In the following sections, we will compare illumination intensity , voltage , and DBP thickness dependent photocurrent behavior in devices with quenching, cascade, and blocking anode/ AIL structures to further contrast these different device architectures. 5.3 Illumination Intensity Dependent Behavior Recalling the J V characteristics of DBP based PHJ devices in Section 4.4, both the c ontrol and NPB AIL containing devices have high fill factors > 0.6 and the
141 photocurrent shows very low slope under reverse voltage bias. The photocurrent saturation term S = J (V = 1 V)/ J SC is in the range of 1.04 1.05 in both devices, whereas S increases to 1.12 for the Tetracene containing cascade device. This suggests the presence of a strong er photocurrent loss mechanism in the cascade device . We can hypothesize that this loss mechanism is related to the different functions of the DBP layer in the diffe rent device structures. The devices grown on bare ITO and the PEDOT:PSS/NPB AIL are similar in that holes are the only charge carriers present in the DBP layer: excitons are dissociated into free carriers at one single location, the DBP/C 60 interface. In t he cascade structure with the PEDOT:PSS/Tetracene AIL, a second charge carrier generating interface is present at the Tetracene/DBP interface . This results in electrons also being present in the DBP layer in addition to holes present from exciton dissociat ion at the DBP/C 60 interface. The presence of both charge carriers within the same physical space makes this structure more akin to a bulk heterojunction (BHJ) active layer structure where hole and electron conducting species ar e blended together homogene ously . In these structures where electrons and holes are n ot spatially separated as in a planar heterojunction device , bimolecular recombination of charge carriers is a dominant photocurrent loss mechanism which result s in strong voltage dependent charge t ransport  and carrier collection [62,87,91] . Devices with bimolecular recombination exhibit a sub unity power law dependence between photocurrent ( J SC ) and illumination intensity ( P O ): as the carrier generation rate increases with illumination intensity, so does the carrier recombination rate. Bimolecular recombination becomes important when the recombination rate
142 increases more rapidly tha n the generation rate. Quantitatively, J SC can be related to the incident optical power intensity P 0 by (3 1) where the recombination parameter describes the recombination rate, with approaching 1 in the ideal case with no recombination. The J V behavior of DBP based PHJ devices with different AIL structures were measured under sim ulated AM 1.5G illumination across three orders of magnitude in intensity, from P O = 0.25 to 120 mW/cm 2 . The J SC values obtained from these measurements are plotted as a function of illumination intensity in Figure 5 1 A) . Figure 5 1. Incident power dependence of photocurrent generation for DBP based devices with different AIL structures. All devices had a DBP (12 nm)/C 60 (30 nm) donor/acceptor PHJ structure. A) Logarithmic J SC ( P O ) and B) s emi logarithmic R ( P O ) . Best fit lines are solid (ITO), dashed (PEDOT:PSS/Tetracene), and dotted (PEDOT:PSS/NPB). Both devices with a single exciton dissociating interface have = 0.99 1.00, indicating that bimolecular charge carrier recomb ination processes are negligible, as expected for PHJ device structures. In contrast, the cascade structure has a reduced A B
143 = 0.94, indicating that recombination processes become more important as more charge carriers are present in the device active laye rs. This is most clear in the low optical power region, where the J SC of the Tetracene containing device is noticeably greater than the NPB containing device. The J SC values then converge at higher light intensities. This behavior can also be visualized by normalizing the J SC to the incident power P O to give the responsivity R Â¸ which is plotted as a function of incident power density in Figure 5 1 B) . The most prominent feature of this data set is the significant negative slope present in the Tetracene conta ining cascade device structure. Considering monomolecular and bimolecular recombination mechanisms, we can write the following simple expression for the J SC : , (3 2) where k G , k M M , and k B M are the generation, monomolecular recombination, and bimolecular recombination rate constants, respectively. We have assumed that the carrier generation and monomolecular recombination rates have a first order dependence on incident power in accordance wi th previous reports and studies. Since the bimolecular recombination process requires charge carriers from two discrete molecules to recombine and negate one another, this process is second order in P O . After dividing though by P O , we have (3 3) where R O describes the rate at which excitons arrive at a dissociating interface and accounts for geminate self recombination processes . On a linear R ( P O ) plot, R O is the y intercept and the bimolecular recombination rate constant k B is given by the slope.
144 Looking back at Figure 5 1 B) , the cascade device has the greatest R O , followed by the blocking device , which are both greater than the control device on ITO. The pre sence of both electrons and holes in the DBP layer of the cascade device causes J SC to decrease over three times as rapidly with P O than the blocking device. The agreement between the voltage dependent ( S ) and carrier concentration dependent ( k BM ) photocurrent loss mechanisms is a strong indication of poor carrier collection efficiency CC in the cascade structure. Despite exhibiting the greatest photocurrent generation rate , the bimolecular loss processes present in the cascade structure severely l imit the charge transport characteristics, ultimately reducing the fill factor and P . On the other hand, the unipolar quality of the DBP layer imparted by the blocking configuration minimizes voltage and intensity dependent losses. 5.4 Spectrally Resolv ed, Voltage D ependent Photocurrent Behavior To this point, the spectral and voltage dependent performance characteristics of DBP based devices with different anode/AIL structures have strongly suggested that these AIL structures primarily affect device per formance by modifying donor behavior only . However, it is impossible to be sure without directly probing the donor layer within these OPV device structures. While unlikely, the voltage and intensity dependent behavior described in Sections 4.4 and 5.3 and assigned to the functionalities of the different AIL structures could possibly be affected by the unintentional modification of the DBP/C 60 interface or neat C 60 layer. As such, the behavior s of the various photoactive layers and interfaces need to be dis entangled. The behavior of the major absorbing layers in these structures can be individually assessed by taking advantage of the spec trally separated absorption wind ows of DBP and C 60 . Towards this end, we have directly measured the
145 photocurrent respons e to a mechanically chopped monochromatic light source via the lock in technique as described in Chapter 2 and in the work of Myers et al.  and Pandey et al.  . By varying the illumination wavelength and voltage bias, a pseudo J V curve can be constructed for absorption by the individual layers in the device provided they have spectrally separated absorption spectra. Referring back to Figure 4 11, we 60 and DBP a is not expected to be significant as the layer is one third the thickness of C 60 and has a lower absorption coefficient at that wavelength as well. The voltage depend illumination are shown in Figures 5 2 A) and B) , respectively. EQE is shown as a positive quantity in accordance with the conventional representation of EQE data. For both illumination wavelengths, the E QE of devices with a PEDOT:PSS/Tetracene AIL (red c ircles ) decays toward zero at a smaller forward bias voltage than devices with bare ITO or PEDOT:PSS/NPB AIL. This behavior is consistent regardless of whether excitons are generated in the donor or accept or layer and agrees well with the voltage dependence of photocurrent under broadband simulated solar illumination shown in Figure 4 10.
146 Figure 5 2 . Voltage dependent EQE corresponding to light absorption in the acceptor and donor layers. A) 45 0 nm and B) 615 nm incident light corresponding to C 60 and DBP absorption, respectively. Inset: EQE normalized to short circuit values. The hole and electron distributions within DBP and C 60 layers can be considered essentially identical whether excitons dissociate from the DBP side or the C 60 side of the heterojunction. As such, the voltage dependent photocurrent behavior under forward bias is consistent for exciton generation in DBP and C 60 layers, as is more clearly illustrated in the normalized EQE(V) curves (Figure 5 2 , inset s ). The photocurrent decay at smaller forward voltages characteristic of the cascade device is observed for exciton generation in both DBP and C 60 . In contrast to the PEDOT:PSS/Tetracene containing device, the EQE of the devices on bare ITO or with a PEDOT:PSS/NPB AIL structure remain high until a forward bias approaching the V OC is applied. This is consistent with the significantly greater fill factors in these device s: 0.66 and 0.71 for the blocking and quenching devices versus 0.45 for the cascade device. To explain this behavior we must consider the voltage dependence of each of the constituent terms of the EQE given by (1 10). Light absorption is not expected to sh ow A B
147 any voltage dependence, exciton diffusion is only affected by voltage under a narrow range of conditions, and charge transfer is assumed to be identical i n these devices due to sufficient energy level offsets. This leaves charge collection as the process which is most likely to cause the observed EQE voltage dependence. This is also supported by the increased bimolecular recombination rate observed only in t he cascade structure. While not always considered in the context of OPV devices, the diffusion component of photocurrent has been shown to play an important role in planar heterojunction photovoltaic device behavior  . In this discussion we consider only those holes and electrons which are only generated at the DBP/C 60 interface. Since Tetracene has a more shallow lying E HOMO than DBP, holes are injected into Tetracene at a smaller forward voltage than they would otherwise be injected into DBP (assuming similar interfacial e ffects among the AIL species). Once holes are injected from ITO/PEDOT:PSS into Tetracene, they wil l accumulate at the Tetracene/DBP interface and reduce the concentrati on gradient of holes in the direction away from the DBP/C 60 interface. Therefore hole extraction is inhibited by the reduced concentration gradient and electrostatic repulsion of accumul ated holes in Tetracene. As a result, the photocurrent rapidly decreases at a lower forward voltage than the cases of bare ITO and PEDOT:PSS/NPB. Owing to its deeper E HOMO , hole injection into NPB does not occur until a stronger forward bias is applied rel ative to Tetracene. Therefore the hole concentration gradient and diffusion current away from the DBP/C 60 interface is voltage bias closer to the DBP/C 60 V OC is reached.
148 Another key difference from the EQE(V) characteristics is the slope under reverse bias. For a more direct view of the voltage dependence of photocurrent generation, the EQE is normalized to the short circuit value in Figure 5 3 . For C 60 absorption [ = 450 nm, Figure 5 3 A) ] there is a linear increase in EQE with reverse bias for all devices considered. This finding is consistent with previous studies which assigned this behavior to polaron induced exciton quenching at the C 60 /BCP interface  and photoconductivity in neat C 60 layers [90,147] . Both of these mechanisms lead to a linear increase in photocurrent with reverse voltage bias. The similar (EQE)/ V slope s for all devices under = 450 nm illumination in Figure 5 3 A) confirms that the Figure 5 3 . Voltage dependent EQE normalized to short circuit values . EQE corresponds to A) 450 nm and B) 615 nm incident light. On the other hand, the reverse bias voltage dependence of EQE corresponding to = 615 nm illumination does exhibit a dependence on AIL structures. In Figure 5 4 B) , the EQE corresponding to DBP absorption of the device with a PEDOT:PSS/Tetracene AIL surpasses that of the PEDOT:PSS/NPB AIL under strong A B
149 reverse bias < 2 V. While the blocking device EQE only increases by 1% relative to its short circuit value, the cascade devic e EQE increases by 17% under a reverse bias of 4 V. This suggests that the cascade structure has ( 1 ) a greater charge generation rate and ( 2 ) a stronger voltage dependent charge collection mechanism than the other device structures. Both of these observat ions agree with the light intensity dependent behavior investigated in Section 5. 3 . The sharp increase in the EQE of the control ITO based device near 3 V is likely due to field induced exciton separation [50,89,90,147] since the applied voltage is dropped across a thinner organic active layer structure compared to the devices with additional 50 nm thick AIL structures, where this sudden increase in EQE is not observed. Whi le hole injection and diffusion current effects can explain the differences in photocurrent behavior under forward bias, and therefore the fill factor, a different explanation is required for reverse bias conditions. Here, holes are swept away from the DBP /C 60 interface (as well as the Tetracene/DBP interface, in the case of the cascade structure) and toward the electrodes with no barrier to extraction in any of the device architectures considered. Under strong reverse bias conditions, photocurrent losses a re minimized and the measured response approaches the photocurrent generation rate. We turn to thickness dependent device characteristics to elucidate these phenomena. 5.5 DBP Thickness Dependence In order to further compare the roles of quenching (ITO), exciton dissociating cascade (Tetracene), and exciton blocking (NP B) AIL structures, PHJ devices we re fabricated with a range of DBP thicknesses from 12 to 25 nm. The photovoltaic performance characteristics of devices with different AIL structures and DBP thick nesses are shown in Figure 5 4 . The devices grown on bare ITO exhibit a weak
150 dependence on DBP thickness: slight increases in J SC [Figure 5 4 A) ] are balanced by slight reductions in fill factor [Figure 5 4 B) ] with increasing DBP thickness, yielding P values between 2.8 and 3.0 % [Figure 5 4 D) ]. The devices with the exciton blocking PEDOT:PSS/NPB AIL structure exhibited a weak dependence on DBP thickness similar to the control ITO devices, with all P > 3.5 %. J SC and fill factor values vary little with DBP thickness, and all blocking devices exhibit S = J (V = 1 V)/ J SC < 1.05. On the other hand, the performance of the PEDOT:PSS/Tetracene devices exhibits a strong dependence on the DBP layer thickness. The device with the thinnest DBP layer achiev es the maximum cascade P = 3.2 %. When the DBP layer thickness is increased, the fill factor diminishes and S increases. Both of these suggest charge transport is hampered by the increasing carrier transit distance. In contrast to the results reported by Cnops et al. in a study on SubPc based cascade devices  , the cascade device fill factor and J SC both decrease with increasing DBP thickness. This can be explained by a significantly reduced Tetracene thickness in this work, different qualities of the Tetrace ne/DBP junction, or the possibility of reduced electron mobility in DBP versus SubPc. The DBP specific behavior will need to be disentangled from the behavior of the entire device to address this behavior.
151 Figure 5 4 . Photovoltaic performance summary of DBP based devices with different anode/AIL structures. All devices has the structure ITO/(AIL)/DBP/C 60 (30 nm)/BCP (8 nm)/Al. Towards this end, we measure the C 60 and DBP specific photocurrent as a function of volt age bias and DBP thickness . The normalized voltage dependent EQE 60 absorption, of devices with different AIL structures and DBP layer thicknesses are shown in Figure 5 5. We re define the photocurrent saturat ion S = EQE(V = 5 V)/EQE(0 V) to simplify the analysis and plot S as a function of DBP thickness for the three structures in Figure 5 5 D) . For C 60 absorption, S is very consistent for all AIL structures and DBP layer thicknesses. This A B C D
152 further affirms the stipulation that the DBP/C 60 heterojunction is not affected by the AIL structures and is not responsible for the changes in device performanc e. Therefore, the DBP thickness dependent device performance can be attributed to the DBP layer with even greater confidence. Figure 5 5. Voltage dependent EQE corresponding to C 60 absorption, normalized to the short circuit value. Devices consisted of ITO/(AIL)/DBP/C 60 (30 nm)/BCP (8 nm)/Al. AIL structures were A) bare ITO, B) PEDOT:PSS/Tetracene, and C) PEDOT:PSS/NPB. D) Photocurrent saturation term S as a function of AIL structure and DBP thickness. The thickness dependence of EQE corresponding to DBP absorption is different in several ways, as shown in Figure 5 6. First, th e absolute voltage dependence is less for all devices than in the case of absorption by C 60 . The maximum DBP S value A B C D
153 reaches just 1.32 for the thickest cascade device, which is less than any such value corresponding to C 60 absorption. S (V) is remarkably lo w for the quenching and blocking devices, with values only exceeding 1.05 for V < 3 V in the device on bare ITO. The sudden increase in EQE in Figure 5 6 A) , which increases slightly with reducing DBP thickness, is attributed to field induced exciton diss ociation  . The blocking device [Figure 5 6 C) ] exhibits excellent charge carrier collection: the short circuit EQE is more than 96% of the maximum value measured under strong reverse bias. The weak dependence of DBP EQE on voltage bias can explain the reasonable values of L D extracted from excitonic photocurrent calculations in Section 4.4.3: the calculations assumed a CT CC product of unity, which is borne out by the relatively constant DBP EQE values of the quenching and blocking devices. The DBP EQE of cascade devices with a PEDOT:PSS/Tetracene AIL structure show different voltage dependences for different DBP thicknesses. Here, S ( 5 V) increases from 1.10 to 1.32 as the DBP layer thicknesses increases from 12 to 25 nm. We can rationalize this by considering the distance electrons and holes must travel before escaping the ambipolar DBP layer: for a thicker DBP layer, electrons and holes generated a t the Tetracene/DBP and DBP/C 60 interfaces, respectively, must travel a greater distance before reaching a location without oppositely charged carriers. For a constant carrier mobility and voltage bias, this means that carriers in thicker DBP layers will h ave a lower drift velocity = and greater transit time within the DBP layer. This results in an increased probability of electron hole recombination before the carriers escape a thicker DBP layer. As a result, the reverse bias EQE is more sensitiv e
154 to voltage bias in devices with thicker DBP layers. This behavior agrees well with the reductions in J SC and fill factor with increasing DBP thickness in the cascade devices. Figure 5 6. Voltage dependent EQE corresponding to DBP absorption, normalized to the short circuit value. Devices consisted of ITO/(AIL)/DBP/C 60 (30 nm)/BCP (8 nm)/Al. AIL structures were A) bare ITO, B) PEDOT:PSS/Tetracene, and C) PEDOT:PSS/NPB. D) Photocurrent saturation term S as a function of AIL structure and DBP thickness. Another conclusion from Figure 5 6 B) also stems from the strong reverse voltage dependence of EQE observed across all DBP thickness es . The lack of EQE saturation near zero applied voltage bias means that the measured short circuit EQE does not correspond to perfectly efficient charge collection. For the cascade device with a 17 nm thick DBP layer, for instance, the maximum EQE is in reality at least 19 % greater than the short circuit value used to extract an L D estimate in Section 4.4.3. The A B C D
155 values extracted for quenching and blocking devices were in good agreement with one another (11.5 and 13 nm, respectively) while a relatively small value of 7 nm was determined for the DBP layer within the cascade stru cture. Since the calculated EQE values w ere based on the assumption of 100% charge collection, the reverse bias EQE corresponding to nearly perfect charge collection is much more relevant and should improve the L D estimate. 5 .6 Excitonic Photocurrent Calculations of DBP based Devices with Exciton Quenching, Blocking, and Dissociating AIL Structures To further verify the different magnitudes of charge carrier generation and recombination, the excitonic photocurrent calculations described in Section 4. 4.3 were applied to these devices with different DBP layer thicknesses. The exciton generation rates within the multilayer device stack structures were calculated by using the transfer matrix method [50,132,138] . The diffusion equat ions were then solved for exciton concentrations approaching zero at the anode(or AIL)/DBP interface in the case of perfect exciton blocking behavior . The EQE was calculated from (4 3) an d (4 4) over a range of L D values , again assuming that all carriers generated at heterojunction interfaces contribute photocurrent. Therefore the calculated EQE values represent upper limits to the EQE. The results of this treatment are plotted in Figure 5 7 for the cases of bare ITO, PEDOT:PSS/Tetracene, and PEDOT:PSS/NPB AIL structures. The experimentally measured maximum EQE value for each device structure is shown as a color coded solid horizontal line, and dotted drop downs represent the intersection o f the calculated and measured values.
156 Figure 5 7. Calculated maximum EQE corresponding to = 615 nm illumination as a function of exciton diffusion length L D for quenching, dissociating, and blocking anode/DBP interfaces across a range of DBP layer thicknesses. Measured EQE values are represented as solid horizontal lines, and dotted vertical lines show the intersection of calculated and measured values. From the plots, the extracted L D values are consistent across all AIL structures an d DBP layer thicknesses. The calculations and measurements yield L D values of (12.5 Â± 1.5) nm for the devices on bare ITO, (11.4 Â± 1.0) nm for the cascade structure, and (13.7 Â± 2.2) nm for the blocking AIL. These values are slightly higher than those obta ined from the photoluminescence measurements discussed in Section 3.5 and other values reported in literature [21,80,127] . However the differences are relativ ely A B C
157 small (< 5 nm) and are self consistent across the different device structures, so the conclusion of similar L D for DBP films grown on different underlayers is valid. The calculated EQE of the device with the exciton blocking AIL structure is less than the cascade device for a given L D and DBP thickness , yet is still significantly increased relative to the quenching device on bare ITO. A more direct view of this enhancement was shown in Figure 4 16 with the EQE values of quenching and blocking boundary c onditions plotted on the same axes. The trends observed there are consistent for the other three DBP thicknesses as well. However, the calculated EQE enhancement decreases with increasing DBP thickness. As the DBP layer thickness surpasses the DBP L D of 10 15 nm, excitons generated farthest from the dissociating DBP/C 60 interface become increasingly unlikely to reach the interface and contribute photocurrent before self recombining. Since all thicknesses studied were equal to or greater than the L D of DBP, exciton diffusion efficiency ( ED ) is expected to decrease with increasing DBP thickness. The EQE is then balanced by the greater absorption efficiency ( A ) in thicker DBP films, resulting in similar maximum measured EQE values for DBP thicknesses ranging from 12 to 21 nm. These calculations also validate the stipulation that photocurrent generation corresponding to DBP absorption increases in order of ITO < NPB < Tetracene as a direct consequence of exciton quenching, blocking, and dissociating behavior at the anode(AIL)/DBP interface. For all DBP thicknesses considered here, peak measured and calculated EQE values are greatest for the cascade structure becau se it is the only structure with two exciton dissociating interfaces. This conclusion nicely ties together the intensity dependent J V measurements and voltage dependent EQE measurements
158 presented in Sections 5.3 and 5.4, respectively. The increased photoc urrent generation rate under broadband solar illumination can now be directly assigned to the DBP layer within the cascade structure. This increase is met with a more rapid decrease of photocurrent with forward voltage bias in the cascade structure, ultima tely resulting in a relatively low fill factor which limits P of the cascade devices. 5. 7 Comparison of AIL/DBP Structures with Tetracene/ZnPc Devices Now that the different behaviors of cascade and exciton blocking AIL structures have been clearly differ entiated with DBP based OPV devices, we can revisit the ZnPc based devices with a PEDOT:PSS/Tetracene AIL structure discussed in Section 4.3 in order to clarify the nature of the Tetracene/ZnPc heterojunction. Both Tetracene and ZnPc have similar reported E HOMO values, so the exciton dissociating or blocking nature of the heterojunction is not clear. Further muddying the picture is the fact that the E HOMO of the partially chlorinated ZnPc used in the present work  has not been me asured in the solid state . Nevertheless, the characteristic s of DBP based devices with confirmed cascade and blocking structures can be used to clarify the picture by comparison with ZnPc ba sed devices. All results of ZnPc based devices in this section correspond to a PHJ structure consisting of ZnPc (20 nm)/C 60 (40 nm)/BCP(8 nm)/Al unless otherwise noted , while the AIL structure is PEDOT:PSS (40 nm)/Tetracene (10 nm). First, we have measured the J V characteristics of PHJ ZnPc devices on bare ITO and with a PEDOT:PSS/Tetracene AIL structure under a broad range of illumination intensities in Figure 5 8. The J SC values of these devices are plotted as a function of incident power P O in Figure 5 8 A) . These data reveal that the recombination parameter is 1.00 for the control device on ITO, signifying minimal photocurrent losses due to bimolecular recombination. However is reduced to (0.96 Â± 0.01) in the device with a
159 PEDOT:PSS/Tet racene AIL. This is similar to the cascade DBP device, which also had < 1. The responsivity of the PEDOT:PSS/Tetracene device R = J SC / P O also decreases with increasing carrier generation rate in Figure 5 8 B) , while the device on bare ITO exhibits consta nt R across all illumination intensities studied. These results suggest that the Tetracene/ZnPc interface is a cascade type, generating additional photocurrent by dissociating excitons at a second location in addition to the ZnPc/C 60 heterojunction . Figure 5 8. Illumination intensity dependent photovoltaic performance parameters of ZnPc based PHJ devices grown on bare ITO and ITO/PEDOT:PSS/Tetracene AIL structure. A) J SC and B) R = J SC / P O of these devices across three orders of magnitude light intensit y. Best fit lines are solid (Bare ITO) and dotted (PEDOT:PSS/Tetracene). The voltage dependent EQE corresponding to absorption by donor ZnPc 630 nm) was also measured and is plotted alo ngside that of various DBP based PHJ devices in Figure 5 9. It is immediately clear that it will be difficult to draw conclusions from the raw data shown i n Figure 5 9 A) because the ranges of EQE values are different for the two donor materia ls. The control DBP device exhibits significantly greater EQE than the ZnPc device , making it difficult to define a common baseline for comparison . The same can be said for devices with different AIL structures. A B
160 For this reason we have subtracted the short circuit EQE (V = 0) from the reverse bias EQE(V) in Figure 5 9 B) . Subtracting rather than normalizing to the short circuit values prevents over and under representation of reverse bias slope in sample s with a smaller and larger EQE(V =0) , respectively. Figure 5 9. Voltage dependent EQE corresponding to donor absorption for DBP and ZnPc based PHJ devices under = 615 and 630 nm illumination, respectively . A) Raw EQE data and B) EQE with short circuit values The excess EQE of the ZnPc device with PEDOT:PSS/Tetracene AIL structure diverges from that of the control ZnPc device at very small reverse bias voltages and has a greater negative slope than the control ZnPc device. Compared with the DBP devices, this behavior is analogous to the cascade structure. Whereas the excess EQE of the DBP based blocking device followed that of the control device very closely for small reverse vol tages, the cascade device showed greater excess EQE under all reverse bias conditions . The similarities betw een the EQE(V) behaviors of the confirmed DBP based cascade structure and the PEDOT:PSS/Tetracene/ZnPc device further A B
161 suggest that the Tetracene/ZnP c heterojunction has a n E HOMO offset which dissociates excitons, thereby forming a cascade structure. It should be noted that while t he magnitude of the excess EQE is less in the ZnPc based cascade device than the DBP based cascade structure , this does not change our conclusion . This difference can be explained by material dependent bimolecular recombination mechanisms present only in an ambipolar donor layer. The phenomena is also supported by the different values of the recombination parameter , calculat ed as (0.94 Â± 0.01) for the Tetracene/DBP/C 60 device and (0.9 6 Â± 0.01) for the Tetracene/ZnPc/C 60 device. The lower degree of bimolecular recombination in the ZnPc based device agrees with the lower observed excess EQE. Recombination of electrons and holes with in the donor layers most likely occurs at trapping centers energetically positioned within the band gap [148 150] . One can imagine these trap sites have different energetic positions, capture cross sections, and concentrations in the two different don or materials used in this study, thereby affecting, among other parameters, the voltage dependent EQE characteristics. The likelihood of Tetra cene forming an exciton blocking interface with ZnPc is low because both holes and electrons are required within the ZnPc layer for these traps to have this effect. As a final confirmation step, ZnPc based PHJ devices with ZnPc layer thicknesses spanning 8 nm to 51 n m were fabricated and characterized. For comparison, t he maximum EQE of these devices were calculated using (4 3) and (4 4) and ZnPc optical constants from the literature  . Figure 5 1 0 summarizes the experi mental and c alculation results for various ZnPc PHJ device structures. All devices had identical C 60 (40 nm)/BCP (8 nm)/Al layer structures. The measured EQE
162 1 1 A) . We have chosen short circu it values for devices on bare ITO and values measured under a 5 V voltage bias for the devices with a PEDOT:PSS/Tetracene AIL due to the significant reverse bias slope, making the reverse bias value a more accurate representation of the maximum EQE. Rather than increase initially with ZnPc thickness and decrease once the layer thickness exceeds the exciton diffusion length ( L D ), estimated to be in the 5 10 nm range in ZnPc [126,151 153] , the EQE of both sets of devices saturates past a thickness of 20 nm. For a constant L D , the predicted EQE values for the quenching case decrease once the layer thickness exceeds a critical value after which increased light abso rption does not increase photocurrent generation [Figure 5 1 0 B) ]. In order for the observed EQE to match the calculated values, L D must increase slightly with ZnPc thickness, from 6 nm to 13 nm when the ZnPc layer thickness is increased from 8 to 51 nm. Such a thickness dependent L D is possible in ordered film growth of planar phthalocyanines and has been observed in previous film and device studies. Vasseur et al. reported an increased L D from 5.5 nm to 12 nm when the Lead P hthalocyanine donor layer thic kness was increased from 20 to 40 or 60 nm. Rand et al. used photoluminescence quenching to measure ZnPc L D values of 15 nm and 25 nm for 150 nm thick ZnPc films with different molecular orientations. Taken together with multiple reports of ZnPc L D values in the 5 10 nm range extracted from devices with 20 40 nm thick ZnPc layers [126,151 153] , we conclude that the thickness dependent ZnPc L D obtained by combining these experimental measurements with theoretical calculations is physically feasible.
163 Figure 5 1 0 . EQE corresponding to = 630 nm illumination for PHJ ZnPc based devices with different ZnPc layer thicknesses and AIL structures. A) Measured values under short circuit conditions and 5 V reverse bias for devices on bare ITO and PEDOT:PSS/Tetracene, respectively. B) D) Calculated maximum EQE = A ED for boundary conditions approximating exciton dissociating, cascade, and blocking be havior at the AIL/ZnPc interface. Different color lines correspond to different assumed L D values, and black crossed squares correspond to the measured values for devices on bare ITO B) and PEDOT:PSS/Tetracene AIL (C,D). By comparing the EQE values calcul ated under dissociating and blocking boundary conditions with PEDOT:PSS/Tetracene/ZnPc device results, Figures 5 1 0 C) and D) , the blocking case can be eliminated as a possibility. In order for the EQE of devices with 40 and 51 nm thick ZnPc layers to achi eve the measured values under A B C D
164 exciton blocking conditions, L D would have to be approximately 25 nm and 35 nm, respectively. These are outside of reasonable limits and we thus conclude that the dissociating cascade structure is formed. One key difference be tween the DBP and ZnPc based cascade de vices remains to be addressed. While b oth control devices grown on bare ITO had high fill factors (0.71 Â± 0.01 and 0.66 for D BP and ZnPc, respectively), the DBP based cascade device exhibited a significantly reduced fill factor of 0.45 while the ZnPc based cascade device retained a high fill factor of 0.63. This can be explained by considering the energy level alignments and forward bias photocurrent behavior in the se devices, shown in Figure 5 12 . Whereas the reverse bias characteristics of the DBP and ZnPc based cascade devices were similar because hole transport towards the ITO anode is unimpeded, the forward bias behavior is very sensitive to the magnitude of th e hole injection barrier present at the Tetracene /Don or interface . The E HOMO of DBP is 5.5 eV below vacuum [20,21] and that of the chlorinated ZnPc used in this study , which has only been measured in solution  , is likely more shallow in the solid state than that of DBP . This stipulation is based on ZnPc devices with C 60 exhibiting a lower the lower V OC , which is strongly dependent on the donor E HOMO [66,69,70,146] . Therefore the hol e injection barrier HOMO between identical Tetracene layers and the two donor materials will be less in the ZnPc based cascade device. The effect of these different E HOMO offsets is apparent in the normalized forward bias EQE(V) characteristics shown in Figure 5 1 1 . For donor DBP, the insertion of a Tetracene layer results in a photocurrent decay at lower forward bias than the device without Tetracene . This ultimately leads to reduced P by way of a significant reduction
165 in fill factor. On the other hand , EQE of the ZnPc based cascade device drops at only a slightly lower voltage bias (< 0.1 V) than the ZnPc device on bare ITO. Rather than the precipitous drop observed in DBP devices, the fill factor only decreases by 5 % in the ZnPc cascade device as a result of the lower HOMO offset. Figure 5 1 1 . EQE(V) normalized to short circuit values for PHJ devices. DBP (blue) and ZnPc (red) based PHJ devices with (closed shapes) and without (open shapes) a PEDOT:PSS/Tetracene AIL structure. 5.8 Summary of Cascade and Blocking Structures In conclusion, the illumination intensity and voltage dependent loss mechanisms prohibi t cascade based devices with donor DBP from improving over control devices on bare ITO. Spectrally resolved, voltage dependent photocurrent measurements showed an increasing voltage dependent EQE with increasing DBP thickness, indicating inefficient charge carrier extraction within t he DBP layer . On the other hand, DBP devices with an NPB exciton blocking layer maintained photocurrent under forward bias approaching the V OC , high fill factor, and peak P = (4.1 Â± 0.1) %. Excitonic photocurrent calculations confirmed the increased photo current generation in both cascade and
166 exciton bl ocking structures. These conclusions were then used to diagnose Tetracene/ZnPc heterojunction in associated OPV devices to conclude that the Tetracene/ZnPc/C 60 device is a cascade structure. However, the for ward voltage characteristics of the ZnPc based devices are not as penal because the E HOMO offset between Tet racene and ZnPc is more readily overcome . From these studies it is clear that the electrical behavior of the AIL/Donor interface is highly material dependent and is likely to vary for other AIL/Donor material combinations . However, we can make some conclusive remark s regarding whether a cascade or blocking structure is best for a given donor material . A subtle but important difference between devices based on donors DBP and ZnPc is the different amounts of EQE enhancement induc ed by blocking and cascade structures. For a DBP based device with a 1 2 nm thick DBP layer (Figure 5 7), the difference between the EQE of the blocking (0.61) and cascade (0.72 ) structures is small . On the other hand, for a ZnPc based device with a 25 nm thick ZnPc layer (Figure 5 10) the calculated EQE of the device with an exciton blocking AIL (0.26) is only slightly more than half that of a device with a cascade structure (0.4 85% of incident light. The difference between these two donor materials is more clearly illustrated in Figure 5 12. Here, we hav e calculated the heterojunction(s) EQE corresponding t o donor absorption for quenching, blocking and cascade structures as a function of L D for PHJ devices. For DBP based devices, the EQE of the blocking device approaches that of the cascade device as L D increases to 20 nm. In the case of ZnPc absorption, the EQE of the blocking device is sig nificantly less than that of a cascade device for the
167 entire range of L D values considered . For a reasonable L D value of 10 nm for both donor materials , the EQE of a DBP based blocking device will be increased by 68% relative to a control device on ITO whi le the s ame L D in a ZnPc based blocking device only results in a 37% enhancement. Figure 5 12. EQE corresponding to donor absorption for devices with blocking and cascade AIL structures . Excitonic EQE for A) DBP and B) ZnPc based PHJ devices. C) , D) EQE of devices with AIL structures normalized to those of devices on bare ITO. The critical difference in these donor materials and devices is the relative mismatch between L D and the absorption length, L A = 1/ where is the peak absorption coefficient. While both donor materials have similar L D values, the peak DBP A B C D
168 is two to three times that of ZnPc. Therefore L A is two to three times less and closer matched to L D . In terms of OPV devices, this translates to a 25 nm thick ZnPc layer of a PHJ device absorbing the same fraction of incident light as a 12 nm thick DBP layer in a similar device structure. Excitons in ZnPc are not likely to travel the entire 25 nm film thickness before recombining, so the blocking interfac e is minimally helpful . In this case the cascade arch itecture removes this constraint by dissociating excitons at the AIL/Donor interface. For films with similar L D values such as ZnPc and DBP, the photocurrent and EQE generated by a thinner layer will be increased more by a blocking layer than that of a thicker film. So we conclude that donor materials with greater absorption coefficients are more likely to be compatible with exciton blocking AIL structures, while cascade AIL structures are more appropriat e for donor materials which require thicker films for efficient light absorption. In this case the critical parameter is the ratio between L D and the critical absorption length L A = 1/ . For DBP, L A / L D delivered more than 50% increased EQE. For ZnPc, L A /L D structure was needed to reach similar levels of EQE enhancement. In addition to assessing the electrical effects of the AIL/Donor interface, these bulk properties of the selected donor material can also provide insight into choosing an appropriate AIL structu re and functionality.
169 CHAPTER 6 TANDEM SMALL MO LECULE ORGANIC PHOTOVOLTAIC CELLS FOR BROAD SPECTRAL RESPONSE AND HIGH OPEN CIRCUIT VOLTAGE 6 .1 Introduction The incorporation of anode interlayer s in OPV devices in Chapters 4 and 5 successfully mitigated the exciton diffusion bottleneck by increasing exciton diffusion efficiency ( ED ) and internal quantum efficiency (IQE) without compromising charge transport properties of the donor materials. However the resulting devices still exhibited rather narrow spectral absorption , only harnessing a fraction of the broad solar illumination spectrum. In this chapter we investigate the combination of multiple donor materials with complimentary absorption profiles in a single tandem device. This chapter will be organized as follows. First, tandem cells will be introduced in terms of their potential, basics of operation, and strategies for maximizing their effectiveness. Next, candidate donor materials will be iden tified and incorporated into preliminary tandem devices. Methods for directly probing the individual subcells within a tandem device will then be described, and these methods will then be used to optimize tandem device performance. 6.2 Tandem Device Funda mentals OPV devices based on a single donor/acceptor (D/A) heterojunction (HJ) suffer from narrow optical absorption bandwidths and a trade off between low energy photon absorption and open circuit voltage ( V OC ). The reason for this first limitation is tha t most OPV cells employ a fullerene acceptor with relatively poor absorption in the visible range . Therefore the optical properties of OPV devices are strongly determined by the properties of the donor material. As we demonstrated in Chapter 5, incorporati ng multiple donors in a single cell (i.e. the cascade structure) often deteriorates the
170 electronic performance by introducing a second heterojunction between the two electrodes. While most donor layers can be thick enough to absorb > 90% of incident light within their spectral absorption range, performance is often limited by the narrowness of the absorption bandwidth . The absorption spectra of several donor materials are shown in Figure 6 1 A) . Furthermore, as described in Section 2.3, the required energy l evel offsets of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) at the D/A HJ necessitates that the absorption edge and maximum V OC are inversely proportional. This trade off is illustrated in Figure 6 1 B) . Figure 6 1. Trade offs in OPV devices. A) Donor absorption spectra are narrow relative to AM 1.5G solar illumination. B) Exciton dissociation required energy level offsets bring about a compromise between J SC (by way of NIR absorption) and V OC . One way to circumnavigate these limitations is to connect two D/A HJ based cells in series with a third electrode to form a multijunction tandem device [46,50,154] . By connecting two D/A HJ cells in series, the V OC of the tandem should approach the sums of the V OC s of the individual cells. The serial connection can be made by sandwiching a charge recombination zone (CRZ) between the two subcells. A CRZ A B
171 should consist of electron conducting and hole conducting layers which accept electrons from one subcell and holes from the other. Therefore holes (electrons) from one subcell will recombine with electrons (holes) from the other subcell in the CRZ, while the opposite carriers will be collected at the electrodes. In Figure 6 2 , two identical CuPc (15 nm) /C 60 (30 nm) devices are connected in series by a C RZ consisting of PTC B I (5 nm)/Ag (0.5 nm)/ MoOx (5 nm) . The V OC of the single D/A HJ device and tandem device are 0.48 V and 0.96 V, respectively, indicating that this CRZ structure effectively recombines holes and electrons. However, the J SC decreases from 3.5 to 2.0 mA/cm 2 and P is only improved from 1.0 % to 1.2 % . Figure 6 2. J V characteristics of a single device and a tandem device with CuPc (15 nm)/C 60 (30 nm) active layer structures. This significant reduction in J SC highlights the major challenge of tandem cell development: photocurrent generation effici ency must be maintained in the tandem configuration or else the increased V OC will be negated by a reduced J SC . To achieve this goal, subcells can be designed with different donor materials in order to harness photons from different regions of the solar sp ectrum. This strategy has been used to
172 demonstrate broad spectral response of two donor materials and high V OC resulting from the series connection between the two individual subcells [102,111,112] . In addition to materials selection, device structures can be engineered to max imize photocurrent generation. Since the thicknesses of the active layers in OPV devices are less than visible photon wavelengths, optical interference between incident photons and photons reflected by the metal cathode and back through the active layers i s an important consideration. Two waves will interfere constructively if they are in phase with one another, i.e. their wave fronts are displaced by an integer number of wavelengths: (6 1) H is the difference in wave front, m is any integer, is the wavelength of light, and n is the refractive index o f the organic layers. Considering the 180 degree phase shift between incident and reflected waves and the difference in path length 2d where d is the distance from the reflective cathode, (6 1) can be expressed as (6 2) Therefore, in order to maximize the photocurrent generation by a material absorbing at wavelength , the layer should be placed approximately x E,max = away from the cathode in order to maximize the optical electric field and exciton generation. An extreme example of this is illustrated in Figure 6 3. Here, we compare the A) J V characteristics and B) short circuit EQE spectra of two PbPc based devices wi th PbPc (20 nm)/C 60 (10 nm) /ET L/Al, where E T L denotes electron transport layer. The E T L of the control device is a 10 nm thick BPhen layer, while the second device has an optical spacer consisting of 10 nm BPhen/50 nm Cs 2 CO 3 doped BPhen (1:20 by volume).
173 Figure 6 3. PbPc/C 60 device performance with (red circles) and without (blue square s) a 50 nm thick transparent spacer layer. With the addition of the spacer, the J SC increases by more than 50% while fill factor and V OC are unaffected. The increased photocurrent is due to improved result illustrates why, in the tandem configuration, the short wavelength absorbing subcell is the back cell closer t o the reflective cathode and long wavelength absorbing subcell is the front cell. Similar transparent, conductive spacer layers have also been used to improve tandem device performance [155,156] but will not be used for tandem device incorporation in this work. 6.3 Donor Materials Selection and Preliminary Tandem D evices With the goal of pairing two donor materials with complimentary photovoltaic properties in a tandem device in order to capture both IR absorption and high V OC , we investigate the donor molecules Boron Subphthalocyanine chloride (SubPc), Indium Phtha locyanine chloride (In Pc), and Lead Phthalocyanine (Pb Pc). InPc and PbPc are chosen for low energy photon harvesting . InPc film growth can be engineered to  A B
174 900 nm when grown on a Pentacene template in Chapters 3 and 4. For blue and green absorption, we choose the donor SubPc. SubPc is an attractive OPV material due to its deep HOMO energy, which results in OPV devices having a very high V OC > 1 V when used with acceptor C 60 [48,95,130] . In addition to spectral absorption, the complementary nature of donors PbPc and SubPc extends to the c haracteristics of their respective OPV devices. The J V characteristics of devices based on these three donor materials under 100 mW/cm 2 simulated AM 1.5G illumination are shown in Figure 6 4 A) . In these devices, InPc and PbPc were grown on CuI and Pentace ne template layers, respectively, and the donor layers were 12 nm and 20 nm thick, respectively, while the rest of the PHJ structure consisted of C 60 (10 nm)/BPhen (10 nm)/BPhen:Cs 2 CO 3 (1:20 by volume, 50 nm)/Al. The Cs 2 CO 3 doped BPhen layer functioned as a transparent electron transport layer (ETL) to better approximate the position of the donor layers in the tandem configuration. A planar mixed ( P M ) HJ structure is used in SubPc based devices in order to increase photocurrent generation while maintaining a relatively high fill factor ( FF )  , with an optimized structure of ITO/MoO X (10 nm)/SubPc (5 nm)/SubPc:C 60 (1:4 by volume, 30 nm)/C 60 (15 nm)/BPhen (10 nm)/Al.
175 Figure 6 4. PV performance of single donor devices. A) J V characteristics and B) short circuit EQE spectra. Individually, these devices illustrate the drawbacks of devices based on a single D/A HJ. The templated PbPc/C 60 planar heterojunction (PHJ) device has significant NIR response and a relatively high J SC , but P is limited by a low V OC . Conversely, the SubP c:C 60 PMHJ device has a high V OC but P is limited by narrow spectral response and, consequently, a relatively low J SC . However, these characteristics present an opportunity to combine the complementary properties of these two donor materials in a tandem d evice having both NIR response and high V OC [157 159] . The photovoltaic performance of these de vices is summarized in Table 6 1 . Table 6 1. P hotovoltaic performance parameters of single D/A HJ devices with different donor species under 100 mW/cm 2 simulated AM 1.5G illumination. Donor material Pentacene/PbPc CuI/InPc SubPc P M HJ J SC (mA/cm 2 ) 7.7 Â± 0.5 5.1 Â± 0.1 6.1 Â± 0.1 FF 0.51 0.46 0.54 Â± 0.02 V OC (V) 0.51 0.73 1.07 Â± 0.01 P (%) 2.0 Â± 0.1 1.7 3.5 Â± 0.2 A B
176 6. 3 .1 Initial Tandem Device Structure For tandem OPV devices fabricated in this study, the long wavelength absorbing device is used as the front subcell, closer to the ITO anode, and the SubPc device is used as the back subcell, closer to the reflective Al cathode. This places the subcells clo ser to their respective first exciton generation maxima, approximated by x G, MAX , where x G , MAX is the distance from the cathode corresponding to maximum exciton generation, m is any integer (here, m = 0 for the first maximum), is the wav elength considered, and n is the refractive index of the photoactive layers. We use a Bathophenanthroline (BPhen, 10 nm)/Ag (0.5 nm)/MoO X (10 nm) trilayer as a CRZ for efficient recombination of electrons and holes generated in the front and back subcells, respectively. The J V characteristics of the initial PbPc/SubPc tandem device are shown in Fig. 6 5 A) beside a sc hematic tandem stack in Fig. 6 5 B) . By visual comparison it is immediately evident that the tandem device J SC of (4.3 Â±0.1) mA/cm 2 is significantly lower than that of either the PbPc or SubPc single device. This results in P = (3.3 Â± 0.2) %, which is less than that of the optimized SubPc:C 60 PMHJ single device. Photocurrent generation in a series connected tandem stack is limited by the subcell which generates the least photocurrent, according to J T (V) = J F (V) = J B (V), where the subscripts denote the tandem cell, front subcell, and back subcell. I t is clear that at least one of the subcells in the tandem stack is generating significa ntly less photocurrent than its respective single device near short circuit conditions. However, from this data alone it is speculative to distinguish between the possible reasons for reduced photocurrent generation in the tandem device: one current limiti ng subcell, insufficient photocurrent in both subcells, an inefficient CRZ, etc. A more detailed
177 analysis is required in order to ( 1 ) determine the underlying cause of reduced photocurrent generation and ( 2 ) increase tandem cell efficiency. Fig ure 6 5 . Photovoltaic response and layer structures of OPV devices. A) J V characteristics of single D/A HJ and tandem devices under 100 mW/cm 2 simulated AM 1.5G illumination . B) Schematic deve architecture of tandem device based on the optimized single D/A HJ device structures. 6.4 Individual Subcell Characterization W e use selective optical biasing to directly measure the EQE spectra of individual subcells in the tandem structur e  . While optical simulations of tandem structures [110,111,159] and the addition of transparent spacer layers to single devices  can also be used to elucidate subcell behavior within the tandem stack, we believe the procedure described herein provides additional complem entary information and is a more direct method of tandem subcell investigation. In order to measure the subcell spectral response in a series connected tandem , the subcell under test must be current limiting throughout the entire wavelength range considere d (i.e. the bias illumination must be more intense than the monochromatic probe illumination). We use high power A B
178 LEDs with emission centered at 590 nm and 850 nm to ensure this is the case. The 590 nm LED is used to selectively excite the SubPc subcell, wh ile the 850 nm LED excites the PbPc or InPc subcell. The spectra PbPc and SubPc device EQE is shown alongside the bias LED spectra in Figure 6 6. Figure 6 6. Spectra of device EQE and LED bias spectra for individual subcell characterization. While the SubPc device/subcell only responds to the 590 nm LED, the PbPc /InPc device/subcell has nonzero spectral response under both bias illumination conditions. In the case of the 590 nm bias LED, the SubPc device generates a factor of 5 10 greater photocurrent t han the PbPc device and ensur es that the PbPc subcell is current limiting for 590 nm optical biasing. We are able to achieve comparable photocurrent under narrow bandwidth illumination to that which is generated under 100 mW/cm 2 simulated broadband AM 1.5G illumination. This is demonstrated in Figure 6 7 .
179 Figure 6 7 . J V characteristics of OPV devices under various illumination sources. PbPc devices (red curves) and SubPc devices (blue curves) under simulated AM 1.5G illumination (closed shapes) and bia s LED illumination (open shapes). When the 850 nm bias LED is used, only the PbPc subcell responds and the SubPc subcell is current limiting during the spectral EQE measurement. Due to the voltage dependence of exciton dissociation and charge collection e fficiencies  , a corrective forward voltage bias is applied in order to reach short circuit conditions and avoid overestimation of EQE in the subcell under test  . We confirm this with voltage dependent photocurrent measurements of a SubPc device and subcell in Figure 6 8 A) . Since the CRZ structure is working properly as evidenced by the voltage addition in the tandem device, we expect the EQE of the SubPc subcell to have very similar voltage dependence to the SubPc standalone device. When no voltage correction is used, the SubPc EQE does not fall to zero until a forward bias > 1.5 V is applied. Therefore a corrective voltage must be applied to achieve short circuit conditions in the subcell of interest when the opposite subcell is optically biased by the DC LED
180 illumination. The magnitude of the voltage bias is the V OC of the biased subcell under LED illumination. Not accounting for this subtlety results in overestimation of subcell EQE, as depicted in Figure 6 8 B) . Figure 6 8 . Illustration of the need for voltag e bias in tandem subcell EQE measurements. A) Voltage dependent EQE under 590 nm illumination for a SubPc device (black curve) and subcell (red curves) . B) EQE spectra under no additional bias (closed circles) and a corrective forward voltage bias (open ci rcles). The short circuit EQE spectra of SubPc and PbPc subcells and single D/A HJ devices are shown in Figure 6 9 . By comparison with the EQE spectra of single devices, which have identical photoactive layer structures, it is clear that both subcells hav e significantly decreased EQE in the tandem configuration. Integration of the front and back subcell EQE spectra with the AM 1.5G spectrum gives J SC s of 4.2 and 4.3 mA/cm 2 , respectively. These represent reductions in short circuit photocurrent generation o f 49% and 31% in the PbPc and SubPc subcells, respectively, when compared to the single devices. Therefore we conclude that photocurrent generation in A B
181 circuit condit ions. Figure 6 9. Short circuit EQE spectra of single devices (closed shapes) and tandem device subcells (open shapes). The reduced J SC of the tandem cell relative to the single D/A HJ PbPc and SubPc based devices cannot be assigned to a single subcell: photocurrent generation must be increased in both front and back subcells in order to increase the P of the tandem device above the bes t performing single device. We also note that the FF of the tandem device falls between those of the single PbPc and SubPc devices, as expected in a tandem device with balanced photocurrent generation  . While subcell EQE measurements allow us to conclude that photocurrent generation must be increased in both subcells in order to increase P of the tandem device, these results alone do not offer a route to ward enhancement.
182 For a better understanding of tandem device behavior we perform transfer matrix calculations to generate the optical electric field and exciton generation profiles within the photoactive layers of the device structures used in this work [50,132,138] . Optical constants were obtained from literature [48,118] . Figure 6 7 plots the optical electric field as a function of position within the stack for wavelengths corresponding to characteristic SubPc, PbPc, and C 60 absorption in both tandem and single dev ice structures. From the plot, the optical field intensities for wavelengths corresponding to SubPc and C 60 absorption, 590 nm and 450 nm, respectively, are significantly reduced in the tandem configuration compared to the single D/A HJ SubPc device. This qualitatively agrees with the observed reduction in spectral response corresponding to SubPc:C 60 absorption when moving from the single D/A HJ to the tandem structure. Furthermore, the exciton dissociating PbPc/C 60 interface is located some 30 nm away fro m the 850 nm optical field maximum in the tandem configuration. According to the x G,MAX guideline, this interface is even farther from exciton generation maxima corresponding to shorter red and IR wavelengths where PbPc also absorbs strongly [Fi g. 1 C) ]. From subcell measurements and simulation results, it is clear that the initial tandem device structure is not optimized for maximum photocurrent generation in either subcell.
183 Figure 6 10 . Position dependence of the optical electric field intens ity calculated by the transfer matrix method within a single D/A HJ SubPc device ( filled shapes ) and a tandem device ( open shapes ). Here, x = 0 corresponds to the BPhen/cathode interface. 6. 5 Tandem Device Optimization The optical field profile in Figure 6 10 suggests that reducing the front cell C 60 thickness will increase PbPc response by shifting the exciton dissociating PbPc/C 60 interface closer to the = 850 nm optical field maximum. In addition, the SubPc:C 60 layer of the back subcell is 80% C 60 by volume and therefore heavily reliant on C 60 absorption for photocurrent production. Since incident light must first pass through the front cell C 60 layer, reducing the front cell C 60 and increase back cell absor ption of both C 60 and SubPc. Therefore, reducing the thickness of the front cell C 60 layer should serve two purposes, both of which can increase tandem device efficiency: (1) shift the PbPc/C 60 interface to a region of higher
184 optical electric field intensi ty corresponding to PbPc absorption, and (2) increase absorption and photocurrent generation in the SubPc:C 60 subcell. The latter part of this hypothesis is further supported by the exciton generation rate profiles for the mixed SubPc:C 60 layer calculate d under AM 1.5G illumination for various layer structures, plotted in Figure 6 11 . By comparing the single SubPc:C 60 PMHJ device (blue line) to the initial tandem configuration with 40 nm front cell C 60 thickness (filled blue squares) we note a significant decrease in the exciton generation rate. Figure 6 11 . Exciton generation rate within the mixed SubPc:C 60 layer of a SubPc device (solid line) and tandem devices with varying front cell C 60 layer thickness (lines broken by shapes) . Assuming that the SubPc single device and subcell have the same internal quantum efficiency (IQE), this corresponds to a 26% reduction in photocurrent contribution from the SubPc:C 60 layer. This is in good agreement with the results from Section 6.4, where a 31% reduction i n J SC was observed between the SubPc device
185 and subcell. These calculations also predict an increase in photocurrent generation in the SubPc:C 60 layer when the front subcell C 60 layer thickness is halved from 40 nm to 20 nm, and again from 20 nm to 10 nm. Again assuming the same IQE for SubPc:C 60 subcells, this corresponds to a 15% increase in photocurrent generation within the SubPc:C 60 layer in the optimized tandem device structure. The experimental short circuit EQE spectra of the front and back subcell s of tandem devices with different front cell C 60 thicknesses are shown in Figure 6 12. The IR response of the front subcell increases with decreasing C 60 thickness and eventually exceeds that of the single D/A HJ PbPc device. In the case of 10 nm front subcell C 60 thickness, integration of the EQE and AM1.5G solar irradiance spectra yields a J SC of 6.5 mA/cm 2 , which is greater than that of the single D/A H J SubPc device. Furthermore as the C 60 layer thickness is reduced and the PbPc/C 60 interface is shifted closer to the cathode in the tandem stack, EQE corresponding to monoclinic PbPc phase absorption ( = 740 nm) increases relative to triclinic phase abso rption ( = 890 nm). This trend experimentally confirms the effect of optical spacing in the tandem device, as blue shifted PbPc absorption and EQE should increase as the PbPc/C 60 interface is shifted nearer the cathode, as predicted by x G,MAX . Back subcell response increases with decreasing front cell C 60 thickness as well, to the extent that subcell EQE corresponding to peak SubPc absorption ( = 590 nm) matches that of the single D/A HJ SubPc device, as shown in Figure 6 12 B) . Integration of the highest performing SubPc subcell EQE and AM1.5G solar irradiance spectra yields J SC = 5.3 mA/cm 2 , showing that the SubPc subcell is current limiting under short circuit conditions. Despite an acceptor thickness of just 10 nm, the broad absorption
186 spect rum of the Pentacene templated PbPc front subcell ensures sufficient photocurrent generation to match that of the SubPc subcell. Figure 6 1 2 . Short circuit EQE spectra of A) PbPc and B) SubPc single D/A HJ devices (black lines) and tandem subcells (lines broken by shapes) with varying front cell C 60 layer thickness. In both subcells, spectral response increases with decreasing front subcell C 60 thickness. The J V characteristics of tandem devices with varying front cell C 60 layer thicknesses are shown in Figure 6 1 3 A) , while the performance characteristics are listed in Table 6 2 . The P of the PbPc/SubPc tandem device reaches a maximum of (4.5 Â± 0.2) %, represent ing a 36% increase over the initial tandem device. The V OC of the optimized tandem device reaches >97% of the sum of the individual cell V OC s as a result of increased photocurrent generation. While the J SC and V OC of tandem devices follow the predicted tre nd with decreasing front cell C 60 thickness, the evolution of FF is more subtle. Tandem device FF falls between those of the single D/A HJ PbPc and SubPc cells in the initial tandem cell with 40 nm front cell C 60 . When the front cell C 60 layer thickness is reduced to 20 or 10 nm the tandem device FF increases to 0.54 0.55, in good agreement with that of the SubPc:C 60 PMHJ device. Following the layer A B
187 structure optimization with PbPc, we have also substituted InPc as the front cell donor material. While the I nPc/SubPc tandem device achieves a higher V OC of 1.78 V, the low EQE and poor charge transport characteristics of the InPc subcell limit the J SC and FF this tandem. Figure 6 13 . J V characteristics of tandem cells with varying front cell C 60 layer thickness. Subcell specific EQE measurements also help explain the tandem cell FF behavior. As described in Section 6. 4.2, subcell photocurrent is balanced in the initial tandem structure and FF falls between those of the individual PbPc and SubPc devices. When the front cell C 60 thickness is reduced to 20 nm and 10 nm, the SubPc subcell becomes current limiting. Integration of the EQE and AM 1.5G solar spectra give J SC s of 5.0 and 5.3 mA/cm 2 in the SubPc subcells, compared with 5.9 and 6.5 mA/cm 2 i n the PbPc subcells in tandem devices with 20 and 10 nm front cell C 60 thickness, respectively. In the case of one current limiting subcell, the FF of the tandem device is expected to be determined by that subcell  . Indeed, the tandem device FF agrees A B
188 with that of the single D/A HJ SubPc device in tandem devices with 20 and 10 nm front cell C 60 layer thickness. Table 6 2. Photovoltaic performance parame ters of tandem devices under 100 mW/cm 2 simulated AM 1.5G illumination . Front subcell PbPc InPc 40 nm C 60 20 nm C 60 10 nm C 60 J SC (mA/cm 2 ) 4.3 Â± 0.1 5.1 Â± 0.1 5.5 Â± 0.1 4.3 Â± 0.1 FF 0.52 Â± 0.01 0.55 Â± 0.01 0.54 0.45 V OC (V) 1.47 Â± 0.01 1.50 Â± 0.02 1.53 1.78 P (%) 3.3 Â± 0.2 4.2 Â± 0.2 4.5 Â± 0.2 3.4 Â± 0.1 6. 6 Conclusions In summary, two donor materials with complementary photovoltaic device properties have been incorporated in a tandem cell. Simply placing a CRZ structure between the optimized single D/A HJ devices resulted in a decrease in P relative to the highest effic iency single device. Through a combination of subcell specific EQE measurements and optical simulations, we were able to determine the reason for insufficient photocurrent generation and construct a route toward improving tandem cell performance. By using these methods to modify the device structure, a tandem device with EQE > 30 % over the range 400 nm > > 900 nm, V OC > 1.50 V, and P = (4.5 Â± 0.2) % was achieved.
189 CHAPTER 7 CONCLUSIONS AND OUTLOOK 7 .1 Conclusions In summary, the work presented in this dissertation has investigated strateg ies for avoiding characteristic trade offs in OPV devices. In Chapter 3 , the insertion of different organic anode interlayer (AIL) structures was observed to directly modify the properties of several OPV donor materials . The absorptio n spectra of neat PbPc films were shown to exhibit a pronounced peak well in to the near infrared (NIR) when grown of a Pentacene template layer . While NIR absorption of PbPc films grown on ITO increases with PbPc thickness, PbPc films grown on Pentacene exhibit higher triclinic content at film thicknesses near the PbPc L D , making the templated PbPc advantageous for OPV device incorporation. Similarly, the absorption spectra of ZnPc films grown on a PEDOT:PSS/Tetracene AIL structure revealed c haracteristic shifts corresponding to stronger intermolecular aggregation than films grown on bare ITO . XRD analysis confirmed a change in the crystal structure within polycrystall ine domains from primarily edge on ZnPc conforma tions to a combination of ed ge on and face on molecular orientations. AFM topography showed that the insertion of a PEDOT:PSS layer planarizes the Tetracene layer, which in turn results in a more continuous ZnPc film surface . Finally, the amorphous donor material DBP was grown on AIL structures contain ing hole transporting material NPB , which was confirmed to function as an exciton blocking layer. From these three donor materials, we conclude that AIL structures can successfully modify the optical and electronic properties of bulk don or materials and AIL/donor interfaces, both of which are critical to OPV device performance.
190 In Chapter 4 , these AIL/donor materials systems were analyzed in the context of OPV device performance. The addition of a Pentacene layer to PbPc based planar hete rojunction ( PHJ ) devices improved the power conversion efficiency ( P ) from 1.5 % to 2.2 %, driven mostly by increased photocurrent generation in the NIR spectral region. The P of ZnPc based PHJ devices increased from 2.0 % to 3.5 % with the addition of a bilayer PEDOT:PSS/Tetracene AIL structure. By comparing the effect of the AIL structure on PHJ, BHJ, and P M HJ devices we concluded that the increased P was a direct result of increased exciton diffusion efficiency ( ED ) caused by A) improved crystallin ity of ZnPc films and orientation of ZnPc molecules and B) excitonic effects at the Tetracene/ZnPc interface. By optimizing the device architecture to maximize the combination of enhanced photocur rent in a neat ZnPc layer and high intrinsic photocurrent ge neration in a mixed ZnPc:C 60 BHJ layer, a maximum P = (5.8 Â± 0.3) % was achieved. For devices based on the donor DBP, bilayer PEDOT:PSS/Tetracene and PEDOT:PSS/ NPB AIL structures showed different effects of photovoltaic performance. While both structures increased photocurrent generation , the voltage dependent photocurrent behavior showed a strong dependence on AIL structure. In the case of the Tetracene/DBP cascade structure, this caused a reduced fill factor ( FF ) which negated the increase in short circuit current density ( J SC ). The device with an exciton blocking PEDOT:PSS/ NPB AIL maintained a relatively high FF , resulting in a 31% increase in P over the control device. Through careful analysis of the illumination intensity dependent photocurrent behavior and the voltage dependence of spectrally resolved photocurrent, we concluded in Chapter 5 that the critical difference between cascade and blocking
191 structures is the simultaneous presence of electrons and holes in the neat DBP layer of the cascade structure. We were then able to compare these characteristics to Tetracene templated ZnPc devices to determine that the Tetracene/ZnPc heterojunction forms a cascade str ucture. By combining the results from thin film and OPV device studies in Chapters 3 and 4 , respectively, we conclude that the introduction of AIL structures is a reliable and facile metho d to modify both bulk and interfacial donor properties for improved OPV device performance. For donors which can aggregate in polycrystalline phases such as PbPc and ZnPc, growth on polycrystalline organic AIL structures promot es bulk ordering more so than growth on weakly interacting oxide surfaces. For donors which form amorphous thin films, AIL structures can also improve OPV device performance by modifying the critical anode/ donor or AIL/donor interface. By introducing an additional exciton dissociating interface or eliminating exciton recombination at the edge of the d onor film in the cases of exciton dissociating and exciton blocking interfaces, respectively, photocurrent generation corresponding to donor absorption is enhanced by characteristi c of OPV devices: improved bulk donor film properties and AIL/donor interfaces allows more excitons to contribute photocurrent for a given donor film thickness without sacrificing electrical properties. As a result, the J SC of OPV devices is increased with minimal effect on the fill factor and improve the maximum generated power to a greater extent than was observed for devices with conventional donor:acceptor blend structures.
192 In Chapter 6 , tandem devices based on complimentary donor materials were fabrica ted in order to mitigate a second major drawback of OPV devices: the mutual exclusion of low energy photon harvesting and high voltage output . Single donor devices based on PbPc, InPc, and SubPc exhibited characteristics typical of OPV devices. We demonstr ated that combining these donors in a se ries connected tandem OPV cell harnesses the complementary properties of these donors in a single OPV device: spectral response s achieved in a device with V OC > 1.5 V. However, device engineering is also required in order to maximize tandem device performance. Sandwiching two optimized D/A HJ devices together initially resulted in a decrease in P relative to the best performing single device. By combining optical electr ic field modeling with subcell specific photocurrent measurements we were able to determine an optimal layer structure for maximum photocurrent generation, culminating with a maximum P = (4.5 Â± 0.2) %. In addition to the device performance, the direct mea surement of subcell behavior in the tandem stack architecture reveals critical information about subcell performance which is ignored in most of the tandem OPV literature. 7 .2 Future Work and Outlook Not long ago, the scientific community was skeptical that OPV devices would ever reach P = 10 %. With recent reports from private and public studies surpassing this goal within the last year, it seems that such skepticism can be disregarded and 15 % is th e next reasonable goal at the laboratory scale. Towards this end, a n exciting new trend in OPV research has been the development of non fullerene acceptor materials [18,82,115] . Despite being the acceptor material of ch o ice in the vast majority of OPV devices, fullerenes have inferior
193 optical properties compared with most donor materials. Relying on a single strong absorber, typically the donor material, puts significant strain on the single photoactive layer and plays in to the weakness of the D/A HJ structure . An 8.4 % efficient device with phthal o cyanine based acceptor layers was recently demonstrated by Cnops et al.  . With this impressive result as a n early bench mark for non fullerene acceptor based devices, the outlook for this approach is promising. A crucial next step for OPV research will be increased transparency with regard to the materials processing and characterization methods used in academic laboratories. There is currently a relatively wide spread in the reported P values of devices based on the same active layer materials, structures, and processing steps. Luber and Buriak present an excellent analysis of the variability in relatively simple P3HT:PCBM based polymer OPV devices  , but the same is true for other polymer and small molecule based devices. Accelerated progress will be possible if the academic community can do a better job of sharing detailed purification techniques, proc essing steps, and characteri zation procedures. OPVs are not likely to enter the global energy discussion because of the results obtained in one academic laboratory. Despite success in academia, commercial and practical implementation of OPV technology has been limited. To move this technology towards consumers, more effort must be spent on the more practical aspects of development such as device lifetime, stability, and scale up. More investigations of OPV device stability are appearing in the literature in cluding the key work by Peters et al.  demonstrating operating lifetimes approaching ten years . With the end goal being a panel with a lifetime on the order of 20 years, cell degradation, aging, and mitigation should be the focus of future research.
194 Many characteristics of the research presented in this dissertation are not compatible with the purported processing strengths of OPV technology: the ITO substrates rely on scarce indium, devices and films were fabricated u nder high vacuum conditions, and source material usage was very inefficient, to name a few. While the fundamental nature of these studies required careful control of experimental conditions as well as film and device structures, these studies did not embod y the strengths of high throughput, abundant source materials, and large flexible geometries which motivate the study of OPV technology. Toward this end, it is helpful to imagine the characteristics of the first commercially successful OPV module  . Now that OPV technology has matured in terms of performance, a shift of focus to large scale processing, environmentally friendly and abundant materials, and ambient conditions is appropri ate. There is a lot of ro om for research in these areas and progress in these practical fields will be necessary for commercial OPV implementation. The interlayers and tandem structures developed in this dissertation, as well as the large amount of similar work in this field, are useful for designing highly efficient structures. In the future, this sort of work can be used to determine target active layer structures to be processed at larger scales. OPV technology can only succeed if the balance between the se approaches is shifted towards the production end. While the materials and device engineering aspects have served many past graduate students well, future work should be more focused on practical considerations and development.
195 LIST OF REFERENCES  Tracking Clean Energy Progress, 2013.  International Energy Outlook 2013, 2013.  American Society for Testing and Materials, ASTM Standard G173 03, Standard Direct Normal and Hemispherical on 37 Degree Tilted Surface, 03 (2012) 1 21.  M. Green, K. Emery, Solar cell efficiency tables (version 43), Prog. Photovoltaics Res. Appl. 22 (2014) 1 9.  W. Shockley, H. Queisser, Detailed balance limit of efficienc y of p n junction solar cells, J. Appl. Phys. 32 (1961) 510 519.  S.R. Forrest, The path to ubiquitous and low cost organic electronic appliances on plastic, Nature. 428 (2004) 911 918.  J. Xue, Perspectives on Organic Photovoltaics, Polym. Rev. 50 ( 2010) 411 419.  D. Yue, P. Khatav, F. You, S.B. Darling, Deciphering the uncertainties in life cycle energy and environmental analysis of organic photovoltaics, Energy Environ. Sci. 5 (2012) 9163.  S.B. Darling, F. You, The case for organic photovolt aics, RSC Adv. 3 (2013) 17633.  B. Azzopardi, C.J.M. Emmott, A. Urbina, F.C. Krebs, J. Mutale, J. Nelson, Economic assessment of solar electricity production from organic based photovoltaic modules in a domestic environment, Energy Environ. Sci. 4 (201 1) 3741.  National Center for Photovoltaics, www.nrel.gov/ncpv/, (2013).  C.W. Tang, Two layer organic photovoltaic cell, Appl. Phys. Lett. 48 (1986) 183.  A. Pochettino, Photoelectric Behavior of Anthracene, Atti R. Accad. Lincei. 15 (1906) 17 1 179.  D. Kearns, M. Calvin, Photovoltaic Effect and Photoconductivity in Laminated Organic Systems, J. Chem. Phys. 29 (1958) 950.  I. Levin, C.E. White, The Photovoltaic Behavior of Organic Substances in Solution, J. Chem. Phys. 18 (1950) 417.
196  H. Kallmann, M. Pope, Photovoltaic Effect in Organic Crystals, J. Chem. Phys. 30 (1959) 585.  Y. Harima, K. Yamashita, H. Suzuki, Spectral sensitization in an organic p n junction photovoltaic cell, Appl. Phys. Lett. 45 (1984) 1144.  N. Beaumo nt, S.W. Cho, P. Sullivan, D. Newby, K.E. Smith, T.S. Jones, Boron Subphthalocyanine Chloride as an Electron Acceptor for High Voltage Fullerene Free Organic Photovoltaics, Adv. Funct. Mater. 22 (2012) 561 566.  S. Cho, L. Piper, Electronic structure o f C60/phthalocyanine/ITO interfaces studied using soft x ray spectroscopies, J. Phys. Chem. C. 114 (2010) 1928 1933.  D. Fujishima, H. Kanno, T. Kinoshita, E. Maruyama, M. Tanaka, M. Shirakawa, et al., Organic thin film solar cell employing a novel ele ctron donor material, Sol. Energy Mater. Sol. Cells. 93 (2009) 1029 1032.  M. Hirade, C. Adachi, Small molecular organic photovoltaic cells with exciton blocking layer at anode interface for improved device performance, Appl. Phys. Lett. 99 (2011) 1533 02.  A. Kahn, N. Koch, W. Gao, Electronic Structure and Electrical Properties of Interfaces between Metals and pi Conjugated Molecular Films, J. Polym. Sci. Part B Polym. Phys. 41 (2003) 2529 2548.  M.T. Greiner, M.G. Helander, W. M. Tang, Z. B. W ang, J. Qiu, Z. H. Lu, Universal energy level alignment of molecules on metal oxides., Nat. Mater. 11 (2012) 76 81.  A.J. Ikushima, T. Kanno, S. Yoshida, A. Maeda, Valence and conduction band edges of metal phthalocyanines behavior and carrier behavior , Thin Solid Films. 273 (1996) 35 38.  M.L. Tietze, W. Tress, S. PfÃ¼tzner, C. SchÃ¼nemann, L. Burtone, M. Riede, et al., Correlation of open circuit voltage and energy levels in zinc phthalocyanine: C60 bulk heterojunction solar cells with varied mixing ratio, Phys. Rev. B. 88 (2013) 085119.  W. Gao, A. Kahn, Controlled p doping of zinc phthalocyanine by coevaporation with tetrafluorotetracyanoquinodimethane: A direct and inverse photoemission study, Appl. Phys. Lett. 79 (2001) 4040.  R. Schlaf, B. a. Parkinson, P. a. Lee, K.W. Nebesny, N.R. Armstrong, HOMO/LUMO Alignment at PTCDA/ZnPc and PTCDA/ClInPc Heterointerfaces
197 Determined by Combined UPS and XPS Measurements, J. Phys. Chem. B. 103 (1999) 2984 2992.  Y. Hamada, H. Kanno, T. Tsujioka, H. Takahashi, T. Usuki, Red organic light emitting diodes using an emitting assist dopant, Appl. Phys. Lett. 75 (1999) 1682.  M. Pope, J. Burgos, J. Giachino, Charge Transfer Exciton State and Energy Levels in Tetracene Crystal, J. Chem. Phys. 43 (1965) 3367.  N. Koch, a. Kahn, J. Ghijsen, J. J. Pireaux, J. Schwartz, R.L. Johnson, et al., Conjugated organic molecules on metal versus polymer electrodes: Demonstration of a key energy level alignment mechanism, Appl. Phys. Lett. 82 (2003) 70.  P.G. S chroeder, C.B. France, J.B. Park, B. a. Parkinson, Energy level alignment and two dimensional structure of pentacene on Au(111) surfaces, J. Appl. Phys. 91 (2002) 3010.  M. Pope, C.E. Swenberg, Electronic Processes in Organic Crystals and Polymers, Sec ond, Oxford University Press, New York, 1999.  R. Fitzner, E. Reinold, A. Mishra, E. Mena Osteritz, H. Ziehlke, C. KÃ¶rner, et al., Dicyanovinyl Substituted Oligothiophenes: Structure Property Relationships and Application in Vacuum Processed Small Mole cule Organic Solar Cells, Adv. Funct. Mater. 21 (2011) 897 910.  J. Eichhorn, M. Stamm, K. Leo, M. Riede, Evaluation and control of the orientation of small molecules for strongly absorbing organic thin films, J. Phys. Ch em. C. 117 (2013) 11600 11609.  B.P. Rand, D. Cheyns, K. Vasseur, N.C. Giebink, S. Mothy, Y. Yi, et al., The Impact of Molecular Orientation on the Photovoltaic Properties of a Phthalocyanine/Fullerene Heterojunction, Adv. Funct. Mater. 22 (2012) 2987 2995.  M. Jones Jr., Organic Chemistry, 3rd ed., W. W. Norton & Company, New York, 2005.  S.O. Kasap, Principles of Electronic Materials and Devices, 3rd ed., McGraw Hill, New York, 2007.  G. Horowitz, Organic Field Effect Transistors, Adv. Mat er. 10 (1998) 365 377.  P.S. Abthagir, Y. G. Ha, E. A. You, S. H. Jeong, H. S. Seo, J. H. Choi, Studies of tetracene and pentacene based organic thin film transistors fabricated by the
198 neutral cluster beam deposition method., J. Phys. Chem. B. 109 (20 05) 23918 24.  F. Cicoira, C. Santato, F. Dinelli, M. Murgia, M. a. Loi, F. Biscarini, et al., Morphology and Field Effect Transistor Mobility in Tetracene Thin Films, Adv. Funct. Mater. 15 (2005) 375 380.  R. Pandey, A. a. Gunawan, K.A. Mkhoyan, R .J. Holmes, Efficient Organic Photovoltaic Cells Based on Nanocrystalline Mixtures of Boron Subphthalocyanine Chloride and C60, Adv. Funct. Mater. 22 (2012) 617 624.  B.P. Rand, J. Xue, S. Uchida, S.R. Forrest, Mixed donor acceptor molecular heterojunc tions for photovoltaic applications. I. Material properties, J. Appl. Phys. 98 (2005) 124902.  S. Tanida, K. Noda, H. Kawabata, K. Matsushige, N channel thin film transistors based on 1,4,5,8 naphthalene tetracarboxylic dianhydride with ultrathin polymer gate buffer layer, Thin Solid Films. 518 (2009) 571 574.  J.M. Mativetsky, H. Wang, S.S. Lee, L. Whittaker Brooks, Y. L. Loo, Face on stacking and enhanced out of plane hole mobility in graphene templated copper phthalocyanine., Chem. Commun. ( 2013) DOI: 10.1039/c3cc47516f.  B. Maennig, M. Pfeiffer, a. Nollau, X. Zhou, K. Leo, P. Simon, Controlled p type doping of polycrystalline and amorphous organic layers: Self consistent description of conductivity and field effect mobility by a microsco pic percolation model, Phys. Rev. B. 64 (2001) 195208.  J. Drechsel, B. MÃ¤nnig, F. Kozlowski, D. Gebeyehu, a. Werner, M. Koch, et al., High efficiency organic solar cells based on single or multiple PIN structures, Thin Solid Films. 451 452 (2004) 515 517.  T.B. Fleetham, N. Bakkan, J.P. Mudrick, J.D. Myers, V.D. Cassidy, J. Cui, et al., Enhanced open circuit voltage in organic photovoltaic cells with partially chlorinated zinc phthalocyanine, J. Mater. Sci. 48 (2013) 7104 7114.  H.H.P. Gomm ans, D. Cheyns, T. Aernouts, C. Girotto, J. Poortmans, P. Heremans, Electro Optical Study of Subphthalocyanine in a Bilayer Organic Solar Cell, Adv. Funct. Mater. 17 (2007) 2653 2658.  B. a. Gregg, S. G. Chen, R. a. Cormier, Coulomb Forces and Doping i n Organic Semiconductors, Chem. Mater. 16 (2004) 4586 4599.  P. Peumans, A. Yakimov, S.R. Forrest, Small molecular weight organic thin film photodetectors and solar cells, J. Appl. Phys. 93 (2003) 3693.
199  T. Forster, Transfer Mechanics of Electronic Excitation, Discuss. Faraday Soc. 27 (1959) 7 17.  S.M. Menke, W.A. Luhman, R.J. Holmes, Tailored exciton diffusion in organic photovoltaic cells for enhanced power conversion efficiency., Nat. Mater. 12 (2013) 152 157.  W.A. Luhman, R.J. Holmes, Investigation of Energy Transfer in Organic Photovoltaic Cells and Impact on Exciton Diffusion Length Measurements, Adv. Funct. Mater. 21 (2011) 764 771.  D.L. Dexter, A Theory of Sensitized Luminescence in Solids, J. Chem. Phys. 21 (1953) 836.  R. R. Lunt, N.C. Giebink, A. a. Belak, J.B. Benziger, S.R. Forrest, Exciton diffusion lengths of organic semiconductor thin films measured by spectrally resolved photoluminescence quenching, J. Appl. Phys. 105 (2009) 053711.  K.J. Bergemann, S.R. Forrest, Measurement of exciton diffusion lengths in optically thin organic films, Appl. Phys. Lett. 99 (2011) 243303.  J.D.A. Lin, O. V. Mikhnenko, J. Chen, Z. Masri, A. Ruseckas, A. Mikhailovsky, et al., Systematic study of exciton diffusion length in organi c semiconductors by six experimental methods, Mater. Horizons. (2014).  B. Verreet, P. Heremans, A. Stesmans, B.P. Rand, Microcrystalline organic thin film solar cells., Adv. Mater. 25 (2013) 5504 7.  M. Hiramoto, H. Fujiwara, M. Yokoyama, Three la yered organic solar cell with a photoactive interlayer of codeposited pigments, Appl. Phys. Lett. 58 (1991) 1062.  S. Uchida, J. Xue, B.P. Rand, S.R. Forrest, Organic small molecule solar cells with a homogeneously mixed copper phthalocyanine: C[sub 60 ] active layer, Appl. Phys. Lett. 84 (2004) 4218.  B.P. Rand, J. Xue, S. Uchida, S.R. Forrest, Mixed donor acceptor molecular heterojunctions for photovoltaic applications. I. Material properties, J. Appl. Phys. 98 (2005) 124902.  J. Xue, B.P. Rand , S. Uchida, S.R. Forrest, Mixed donor acceptor molecular heterojunctions for photovoltaic applications. II. Device performance, J. Appl. Phys. 98 (2005) 124903.  M. Shtein, P. Peumans, J.B. Benziger, S.R. Forrest, Micropatterning of small molecular we ight organic semiconductor thin films using organic vapor phase deposition, J. Appl. Phys. 93 (2003) 4005.
200  S.R. Forrest, Ultrathin Organic Films Grown by Organic Molecular Beam Deposition and Related Techniques., Chem. Rev. 97 (1997) 1793 1896.  J . Xue, S. Uchida, B.P. Rand, S.R. Forrest, 4.2% Efficient Organic Photovoltaic Cells With Low Series Resistances, Appl. Phys. Lett. 84 (2004) 3013.  N.C. Giebink, G.P. Wiederrecht, M.R. Wasielewski, S.R. Forrest, Ideal diode equation for organic hetero junctions. I. Derivation and application, Phys. Rev. B. 82 (2010) 155305.  N.C. Giebink, B.E. Lassiter, G.P. Wiederrecht, M.R. Wasielewski, S.R. Forrest, Ideal diode equation for organic heterojunctions. II. The role of polaron pair recombination, Phys . Rev. B. 82 (2010) 155306.  J.D. Myers, T. K. Tseng, J. Xue, Photocarrier behavior in organic heterojunction photovoltaic cells, Org. Electron. 10 (2009) 1182 1186.  B. Rand, D. Burk, S. Forrest, Offset energies at organic semiconductor heterojunc tions and their influence on the open circuit voltage of thin film solar cells, Phys. Rev. B. 75 (2007) 115327.  M.D. Perez, C. Borek, S.R. Forrest, M.E. Thompson, Molecular and Morphological Influences on the Open Circuit Voltages of Organic Photovolt aic Devices, (2009) 9281 9286.  W.J. Potscavage, A. Sharma, B. Kippelen, Critical interfaces in organic solar cells and their influence on the open circuit voltage., Acc. Chem. Res. 42 (2009) 1758 67.  American Society for Testing and Materials, ASTM Standard E973, Standard Test Method for Determination of the Spectral Mismatch Parameter Between a Photovoltaic Device and a Photovoltaic Reference Cell, (2010) 1 3.  V. Shrotriya, G. Li, Y. Yao, T. Moriarty, K. Emery, Y. Yang, Accurate Measuremen t and Characterization of Organic Solar Cells, Adv. Funct. Mater. 16 (2006) 2016 2023.  K.A. Emery, Solar simulators and i v measurement methods, Sol. Cells. 18 (1986) 251 260.  J. You, C. C. Chen, Z. Hong, K. Yoshimura, K. Ohya, R. Xu, et al., 10. 2% Power Conversion Efficiency Polymer Tandem Solar Cells Consisting of Two Identical Sub Cells., Adv. Mater. 25 (2013) 3973 8.
201  J. You, L. Dou, K. Yoshimura, T. Kato, K. Ohya, T. Moriarty, et al., A polymer tandem solar cell with 10.6% power conversio n efficiency., Nat. Commun. 4 (2013) 1446.  D.M. Chapin, C.S. Fuller, G.L. Pearson, A New Silicon p n Junction for Converting Solar Radiation into Electrical Power, J. Appl. Physics1. 25 (1954) 676.  H. Queisser, Forward characteristics and efficie ncies of silicon solar cells, Solid. State. Electron. 5 (1962) 1 10.  Z. He, C. Zhong, S. Su, M. Xu, H. Wu, Y. Cao, Enhanced power conversion efficiency in polymer solar cells using an inverted device structure, Nat. Photonics. 6 (2012) 591 595.  Y . Zhou, T. Taima, T. Kuwabara, K. Takahashi, Efficient Small Molecule Photovoltaic Cells Using a Crystalline Diindenoperylene Film as a Nanostructured Template., Adv. Mater. 25 (2013) 6069 6075.  B. Verreet, P.E. Malinowski, B. Niesen, D. Cheyns, P. He remans, A. Stesmans, et al., Improved cathode buffer layer to decrease exciton recombination in organic planar heterojunction solar cells, Appl. Phys. Lett. 102 (2013) 043301.  B. Verreet, K. Cnops, D. Cheyns, P. Heremans, A. Stesmans, G. Zango, et al. , Decreased Recombination Through the Use of a Non Fullerene Acceptor in a 6.4% Efficient Organic Planar Heterojunction Solar Cell, Adv. Energy Mater. (2014) DOI: 10.1002/aenm.201301413.  Textured Multicrystalline and 24.4% Monocrystalline Silicon Solar Cells, Appl. Phys. Lett. 73 (1998) 1991.  B. Kayes, H. Nie, R. Twist, 27.6% conversion efficiency, a new record for single junction solar cells under 1 sun illumination, Photovol t. Spec. Conf. (PVSC), 2011 37th IEEE. (2011) 4 8.  I. Repins, M.A. Contreras, B. Egaas, C. Dehart, J. Scharf, C.L. Perkins, 19.9 % efficient ZnO/CdS/CuInGaSe2 Solar Cell with 81.2 % Fill Factor, Prog. Photovoltaics Res. Appl. 16 (2008) 235 239.  S. Park, A. Roy, S. BeauprÃ©, S. Cho, Bulk heterojunction solar cells with internal quantum efficiency approaching 100%, Nat. Photonics. 3 (2009) 297 303.  R. Pandey, R.J. Holmes, Characterizing the charge collection efficiency in bulk heterojunction or ganic photovoltaic cells, Appl. Phys. Lett. 100 (2012) 083303.
202  J.J.M. Halls, C.A. Walsh, N.C. Greenham, E.A. Marseglia, R.H. Friend, S.C. Moratti, et al., Efficient photodiodes from interpenetrating polymer networks, Nature. 376 (1995) 498.  C.K. Renshaw, J.D. Zimmerman, B.E. Lassiter, S.R. Forrest, Photoconductivity in donor acceptor heterojunction organic photovoltaics, Phys. Rev. B. 86 (2012) 085324.  W. I. Jeong, Y.E. Lee, H. S. Shim, T. M. Kim, S. Y. Kim, J. J. Kim, Photoconductivity of C6 0 as an Origin of Bias Dependent Photocurrent in Organic Photovoltaics, Adv. Funct. Mater. 22 (2012) 3089 3094.  J. Xue, B.P. Rand, S. Uchida, S.R. Forrest, A Hybrid Planar Mixed Molecular Heterojunction Photovoltaic Cell, Adv. Mater. 17 (2005) 66 71.  B.P. Rand, J. Xue, F. Yang, S.R. Forrest, Organic solar cells with sensitivity extending into the near infrared, Appl. Phys. Lett. 87 (2005) 233508.  W. Zhao, J.P. Mudrick, Y. Zheng, W.T. Hammond, Y. Yang, J. Xue, Enhancing photovoltaic response o f organic solar cells using a crystalline molecular template, Org. Electron. 13 (2012) 129 135.  H. S. Shim, H.J. Kim, J.W. Kim, S. Y. Kim, W. I. Jeong, T. M. Kim, et al., Enhancement of near infrared absorption with high fill factor in lead phthalocya nine based organic solar cells, J. Mater. Chem. 22 (2012) 9077.  R. Pandey, R.J. Holmes, Graded donor acceptor heterojunctions for efficient organic photovoltaic cells., Adv. Mater. 22 (2010) 5301 5.  W. Tress, K. Leo, M. Riede, Effect of concentra tion gradients in ZnPc:C60 bulk heterojunction organic solar cells, Sol. Energy Mater. Sol. Cells. 95 (2011) 2981 2986.  M. Zhang, H. Wang, H. Tian, Y. Geng, C.W. Tang, Bulk heterojunction photovoltaic cells with low donor concentration., Adv. Mater. 2 3 (2011) 4960 4.  X. Xiao, J.D. Zimmerman, B.E. Lassiter, K.J. Bergemann, S.R. Forrest, A hybrid planar mixed tetraphenyldibenzoperiflanthene/C70 photovoltaic cell, Appl. Phys. Lett. 102 (2013) 073302.  G. Wei, X. Xiao, S. Wang, K. Sun, K.J. Bergemann, M.E. Thompson, et al., Functionalized squaraine donors for nanocrystalline organic photovoltaics., ACS Nano. 6 (2012) 972 8.
203  S. Wang, L. Hall, V. V Diev, R. Haiges, G. Wei, X. Xiao, et al., N , N Di aryl anilinosquaraines and Their Applica tion to Organic Photovoltaics, Chem. Mater. 23 (2011) 4789 4798.  J.D. Zimmerman, X. Xiao, C.K. Renshaw, S. Wang, V. V Diev, M.E. Thompson, et al., Independent control of bulk and interfacial morphologies of small molecular weight organic heterojuncti on solar cells., Nano Lett. 12 (2012) 4366 71.  J.D. Zimmerman, B.E. Lassiter, X. Xiao, K. Sun, A. Dolocan, R. Gearba, et al., Control of interface order by inverse quasi epitaxial growth of squaraine/fullerene thin film photovoltaics., ACS Nano. 7 (2 013) 9268 75.  W. Wang, D. Placencia, N.R. Armstrong, Planar and textured heterojunction organic photovoltaics based on chloroindium phthalocyanine (ClInPc) versus titanyl phthalocyanine (TiOPc) donor layers, Org. Electron. 12 (2011) 383 393.  B. E. Lassiter, R.R. Lunt, C.K. Renshaw, S.R. Forrest, Structural templating of multiple polycrystalline layers in organic photovoltaic cells., Opt. Express. 18 (2010) 444 450.  K. Cnops, B.P. Rand, D. Cheyns, P. Heremans, Enhanced photocurrent and open circuit voltage in a 3 layer cascade organic solar cell, Appl. Phys. Lett. 101 (2012) 143301.  A. Barito, M.E. Sykes, D. Bilby, J. Amonoo, Y. Jin, S.E. Morris, et al., Recovering lost excitons in organic photovoltaics using a transparent dissociation layer, J. Appl. Phys. 113 (2013) 203110.  C.W. Schlenker, V.S. Barlier, S.W. Chin, M.T. Whited, R.E. McAnally, S.R. Forrest, et al., Cascade Organic Solar Cells, Chem. Mater. 23 (2011) 4132 4140.  X. Xiao, K.J. Bergemann, J.D. Zimmerman, K. Lee, S.R. Forrest, Small Molecule Planar Mixed Heterojunction Photovoltaic Cells with Fullerene Based Electron Filtering Buffers, Adv. Energy Mater. (2013) DOI: 10.1002/aenm.201301557.  M.E. Thompson, A.N. Bartynski, C. Trinh, A. Panda, B. Lassiter, K. Ber gemann, et al., A Fullerene Based Organic Exciton Blocking Layer with High Electron Conductivity., Nano Lett. 13 (2013) 3315 3320.  J. Xue, S. Uchida, B.P. Rand, S.R. Forrest, Asymmetric tandem organic photovoltaic cells with hybrid planar mixed molec ular heterojunctions, Appl. Phys. Lett. 85 (2004) 5757.  D. Cheyns, B.P. Rand, P. Heremans, Organic tandem solar cells with complementary absorbing layers and a high open circuit voltage, Appl. Phys. Lett. 97 (2010) 033301.
204  B.E. Lassiter, J.D. Z immerman, A. Panda, X. Xiao, S.R. Forrest, Tandem organic photovoltaics using both solution and vacuum deposited small molecules, Appl. Phys. Lett. 101 (2012) 063303.  J. a Macko, R.R. Lunt, T.P. Osedach, P.R. Brown, M.C. Barr, K.K. Gleason, et al., M ultijunction organic photovoltaics with a broad spectral response., Phys. Chem. Chem. Phys. 14 (2012) 14548 53.  New world record for organic solar cells: Heliatek asserts itself with 12% cell efficiency as a technology leader, (2013).  K. Cnops, B.P. Rand, D. Cheyns, B. Verreet, M. a Empl, P. Heremans, 8.4% Efficient Fullerene Free Organic Solar Cells Exploiting Long Range Exciton Energy Transfer., Nat. Commun. 5 (2014) 3406.  A.K.K. Kyaw, D.H. Wang, D. Wynands, J. Zhang, T. Q. Nguyen, G.C. Bazan, et al., Improved light harvesting and improved efficiency by insertion of an optical spacer (ZnO) in solution processed small molecule solar cells., Nano Lett. 13 (2013) 3796 801.  K. Vasseur, K. Broch, A.L. Ayzner, B.P. Rand, D. Cheyns, C. Fra nk, et al., Controlling the texture and crystallinity of evaporated lead phthalocyanine thin films for near infrared sensitive solar cells., ACS Appl. Mater. Interfaces. 5 (2013) 8505 15.  K. Vasseur, B.P. Rand, D. Cheyns, L. Froyen, P. Heremans, Stru ctural Evolution of Evaporated Lead Phthalocyanine Thin Films for Near Infrared Sensitive Solar Cells, Chem. Mater. 23 (2011) 886 895.  S. Pfuetzner, C. Mickel, J. Jankowski, M. Hein, J. Meiss, C. Schuenemann, et al., The influence of substrate heatin g on morphology and layer growth in C60:ZnPc bulk heterojunction solar cells, Org. Electron. 12 (2011) 435 441.  A. Miyamoto, K. Nichogi, A. Taomoto, T. Nambu, M. Murakami, Structural control of evaporated lead phthalocyanine films, Thin Solid Films. 256 (1995) 64 67.  F.J. Meyer zu Heringdorf, M.C. Reuter, R.M. Tromp, Growth dynamics of pentacene thin films., Nature. 412 (2001) 517 20.  S. Ha, A. Kahn, Isolated molecular dopants in pentacene observed by scanning tunneling microscopy, Phys. R ev. B. 80 (2009) 195410.  C. SchÃ¼nemann, C. Elschner, a. a. Levin, M. Levichkova, K. Leo, M. Riede, Zinc phthalocyanine Influence of substrate temperature, film thickness, and kind of substrate on the morphology, Thin Solid Films. 519 (2011) 3939 39 45.
205  P. Sullivan, T.S. Jones, a. J. Ferguson, S. Heutz, Structural templating as a route to improved photovoltaic performance in copper phthalocyanine/fullerene C60 heterojunctions, Appl. Phys. Lett. 91 (2007) 233114.  R.R. Lunt, J.B. Benziger, S .R. Forrest, Relationship between crystalline order and exciton diffusion length in molecular organic semiconductors., Adv. Mater. 22 (2010) 1233 6.  Y. Terao, H. Sasabe, C. Adachi, Correlation of hole mobility, exciton diffusion length, and solar cel l characteristics in phthalocyanine/fullerene organic solar cells, Appl. Phys. Lett. 90 (2007) 103515.  D. Yokoyama, Z. Qiang Wang, Y. J. Pu, K. Kobayashi, J. Kido, Z. Hong, High efficiency simple planar heterojunction organic thin film photovoltaics with horizontally oriented amorphous donors, Sol. Energy Mater. Sol. Cells. 98 (2012) 472 475.  Y. Zhou, T. Taima, Y. Shibata, T. Miyadera, T. Yamanari, Y. Yoshida, Controlled growth of dibenzotetraphenylperiflanthene thin films by varying substrate t emperature for photovoltaic applications, Sol. Energy Mater. Sol. Cells. 95 (2011) 2861 2866.  Y. Zheng, W.J. Potscavage, T. Komino, M. Hirade, J. Adachi, C. Adachi, Highly efficient bulk heterojunction photovoltaic cells based on C70 and tetraphenyld ibenzoperiflanthene, Appl. Phys. Lett. 102 (2013) 143304.  C. F. Lin, V.M. Nichols, Y. C. Cheng, C.J. Bardeen, M. K. Wei, S. W. Liu, et al., Chloroboron subphthalocyanine/C60 planar heterojunction organic solar cell with N,N dicarbazolyl 3,5 benzene b locking layer, Sol. Energy Mater. Sol. Cells. 122 (2014) 264 270.  H. Gommans, B. Verreet, B.P. Rand, R. Muller, J. Poortmans, P. Heremans, et al., On the Role of Bathocuproine in Organic Photovoltaic Cells, Adv. Funct. Mater. 18 (2008) 3686 3691. [13 2] of photovoltaic devices based on organic thin films, J. Appl. Phys. 86 (1999) 487.  S. Yoo, B. Domercq, B. Kippelen, Efficient thin film organic solar cells based on p entacene/C[sub 60] heterojunctions, Appl. Phys. Lett. 85 (2004) 5427.  C. W. Chu, Y. Shao, V. Shrotriya, Y. Yang, Efficient photovoltaic energy conversion in tetracene C[sub 60] based heterojunctions, Appl. Phys. Lett. 86 (2005) 243506.
206  I. Kim, H.M. Haverinen, J. Li, G.E. Jabbour, Enhancement of device performance of organic solar cells by an interfacial perylene derivative layer., ACS Appl. Mater. Interfaces. 2 (2010) 1390 4.  T.B. Fleetham, J.P. Mudrick, W. Cao, K. Klimes, J. Xue, J. Li, E fficient Zinc Phthalocyanine/C60 Heterojunction Photovoltaic Devices Employing Tetracene Anode Interfacial Layers, ACS Appl. Mater. Interfaces. (2014) DOI: 10.1021/am500472m.  Z. Wang, D. Yokoyama, X. F. Wang, Z. Hong, Y. Yang, J. Kido, Highly efficient organic p i n photovoltaic cells based on tetraphenyldibenzoperiflanthene and fullerene C70, Energy Environ. Sci. 6 (2013) 249.  G.F. Burkhard, E.T. Hoke, M.D. McGehee, Accounting for interference, scattering, and electrode absorption to mak e accurate internal quantum efficiency measurements in organic and other thin solar cells., Adv. Mater. 22 (2010) 3293 7.  T.D. Heidel, D. Hochbaum, J.M. Sussman, V. Singh, M.E. Bahlke, I. Hiromi, et al., Reducing recombination losses in planar organi c photovoltaic cells using multiple step charge separation, J. Appl. Phys. 109 (2011) 104502.  J. Huang, J. Yu, W. Wang, Y. Jiang, Organic solar cells with a multicharge separation structure consisting of a thin rubrene fluorescent dye for open circui t voltage enhancement, Appl. Phys. Lett. 98 (2011) 023301.  S.L. Lai, M.F. Lo, M.Y. Chan, C.S. Lee, S.T. Lee, Impact of dye interlayer on the performance of organic photovoltaic devices, Appl. Phys. Lett. 95 (2009) 153303.  Y. Kinoshita, T. Hasob e, H. Murata, Control of open circuit voltage in organic photovoltaic cells by inserting an ultrathin metal phthalocyanine layer, Appl. Phys. Lett. 91 (2007) 083518.  W. Tress, K. Leo, M. Riede, Influence of Hole Transport Layers and Donor Materials o n Open Circuit Voltage and Shape of I V Curves of Organic Solar Cells, Adv. Funct. Mater. 21 (2011) 2140 2149.  W. Tress, O. InganÃ¤s, Simple experimental test to distinguish extraction and injection barriers at the electrodes of (organic) solar cells with S shaped current voltage characteristics, Sol. Energy Mater. Sol. Cells. 117 (2013) 599 603.  W. Tress, S. Corvers, K. Leo, M. Riede, Investigation of Driving Forces for Charge Extraction in Organic Solar Cells: Transient Photocurrent Measurement s on Solar Cells Showing S Shaped Current Voltage Characteristics, Adv. Energy Mater. 3 (2013) 873 880.
207  D. Cheyns, J. Poortmans, P. Heremans, C. Deibel, S. Verlaak, B. Rand, et al., Analytical model for the open circuit voltage and its associated res istance in organic planar heterojunction solar cells, Phys. Rev. B. 77 (2008) 165332.  W. Tress, K. Leo, M. Riede, Photoconductivity as loss mechanism in organic solar cells, Phys. Status Solidi Rapid Res. Lett. 7 (2013) 401 405.  R. a. Street, A. Krakaris, S.R. Cowan, Recombination Through Different Types of Localized States in Organic Solar Cells, Adv. Funct. Mater. 22 (2012) 4608 4619.  A. Sharma, S. Yadav, P. Kumar, S. Ray Chaudhuri, S. Ghosh, Defect states and their energetic position and distribution in organic molecular semiconductors, Appl. Phys. Lett. 102 (2013) 143301.  Y.S. Yang, S.H. Kim, J. I. Lee, H.Y. Chu, L. M. Do, H. Lee, et al., Deep level defect characteristics in pentacene organic thin films, Appl. Phys. Lett. 80 (20 02) 1595.  J. Yang, F. Zhu, B. Yu, H. Wang, D. Yan, Simultaneous enhancement of charge transport and exciton diffusion in single crystal like organic semiconductors, Appl. Phys. Lett. 100 (2012) 103305.  I. Kim, H.M. Haverinen, Z. Wang, S. Madaku ni, J. Li, G.E. Jabbour, Effect of molecular packing on interfacial recombination of organic solar cells based on palladium phthalocyanine and perylene derivatives, Appl. Phys. Lett. 95 (2009) 023305.  I. Kim, H.M. Haverinen, Z. Wang, S. Madakuni, Y. Kim, J. Li, et al., Efficient Organic Solar Cells Based on Planar Metallophthalocyanines, Chem. Mater. 21 (2009) 4256 4260.  A. Yakimov, S.R. Forrest, High photovoltage multiple heterojunction organic solar cells incorporating interfacial metallic nan oclusters, Appl. Phys. Lett. 80 (2002) 1667.  R. Schueppel, R. Timmreck, N. Allinger, T. Mueller, M. Furno, C. Uhrich, et al., Controlled current matching in small molecule organic tandem solar cells using doped spacer layers, J. Appl. Phys. 107 (2010) 044503.  M. Riede, C. Uhrich, J. Widmer, R. Timmreck, D. Wynands, G. Schwartz, et al., Efficient Organic Tandem Solar Cells based on Small Molecules, Adv. Funct. Mater. 21 (2011) 3019 3028.  J. Yang, W. Chen, B. Yu, H. Wang, D. Yan, Organi c tandem solar cell using active inter connecting layer, Org. Electron. 13 (2012) 1018 1022.
208  H. S. Shim, S. Y. Kim, J. Whan Kim, T. M. Kim, C. H. Lee, J. J. Kim, An efficient interconnection unit composed of electron transporting layer/metal/p doped hole transporting layer for tandem organic photovoltaics, Appl. Phys. Lett. 102 (2013) 203903.  Y. E. Lee, S. Y. Kim, W. I. Jeong, J. J. Kim, Microcavity tandem solar cells with a short circuit current higher than single cells, Sol. Energy Mater. Sol. Cells. 114 (2013) 59 64.  J. Gilot, M.M. Wienk, R.A.J. Janssen, Measuring the External Quantum Efficiency of Two Terminal Polymer Tandem Solar Cells, Adv. Funct. Mater. 20 (2010) 3904 3911.  B.E. Lassiter, C. Kyle Renshaw, S.R. Forrest, Understa nding tandem organic photovoltaic cell performance, J. Appl. Phys. 113 (2013) 214505.  E. Luber, J. Buriak, Reporting Performance in Organic Photovoltaic Devices, ACS Nano. (2013) 4708 4714.  C.H. Peters, I.T. Sachs Quintana, J.P. Kastrop, S. Bea uprÃ©, M. Leclerc, M.D. McGehee, High Efficiency Polymer Solar Cells with Long Operating Lifetimes, Adv. Energy Mater. 1 (2011) 491 494.  F.C. Krebs, M. JÃ¸rgensen, Polymer and organic solar cells viewed as thin film technologies: What it will take for them to become a success outside academia, Sol. Energy Mater. Sol. Cells. 119 (2013) 73 76.
209 BIOGRAPHICAL SKETCH John was born in Orlando, FL and grew up in Sykesville, MD. He attended Mount Saint Joseph High Schoo l where, after realizing his baseball career was only a dream, he found a passion for math and science. After earning his high school diploma in 2005, he began college studies at the University of Notre Dame, earning a Bachelor of Science in Chemical Engin eering in May 2009. John was accepted to the University of Florida Department of Materials Science and Engineering PhD program in February 2009 and began his study that August under the advisement of Dr. Jiangeng Xue. He received his Doctor of Philosophy in August 2014.